Neuron-derived mitochondrial DNA (mtDNA) activates microglia via the Z-DNA binding protein 1 (ZBP1)-mediated pathway in mild traumatic brain injury

Significance

Mild traumatic brain injury (mTBI) is the most common form of brain injury, but the biological signals linking early neuronal damage to brain inflammation remain poorly understood. We show that mitochondrial DNA (mtDNA) released from injured neurons acts as a danger signal that activates specifically microglia, the resident immune cells of the brain. After mTBI, damaged mtDNA is packaged into extracellular vesicles and triggers inflammatory signaling through the cytoplasmic sensor protein called Z-DNA binding protein 1 (ZBP1). Depleting ZBP1 reduces early inflammation but worsens later memory-related cognitive deficits, confirming the importance of early protective immune responses. Together, these findings reveal a pathway connecting neuronal injury to immune activation and suggest potential therapeutic targets for improving recovery after mTBI.

Abstract

Traumatic brain injury (TBI) is a leading cause of morbidity and mortality, with closed-head mild TBI (mTBI) accounting for nearly 90% of all cases. Early pathological events include microglial activation and neuronal mitochondrial dysfunction; however, their interconnection in mTBI remains poorly understood. Using a clinically relevant closed-head weight-drop mouse model, we identified mitochondrial DNA (mtDNA)-specific damage and increased expression of innate inflammatory markers (IL-1α/β, IL-6, TNFα, and CXCL1) in the cerebral cortex during the acute mTBI phase. Mechanistically, neurons subjected to in vitro injury model of mTBI exhibited early mtDNA-specific damage followed by mtDNA release via extracellular vesicles (EVs) together with the neuronal and exosomal markers. The released neuronal mtDNA induced a robust microglial activation mediated by binding to the cytoplasmic DNA/RNA sensor Z-DNA–binding protein 1 (ZBP1), triggering activation of the ZBP1–TBK1–IRF3 pathway resulted IL-6 and TNFα expression. An early, enhanced amounts of mtDNA, neuronal and exosomal markers were measured in EVs circulating in the blood of mice subjected to mTBI. ZBP1 knockout (KO) mice displayed suppressed microglial—but not astrocytic—activation in the cortex during the acute mTBI phase. We also measured accumulation of mtDNA-specific damage in the hippocampus during the postacute mTBI phase. The absence of microglial activation in ZBP1 KO mice exacerbated hippocampal-related memory deficits in the postacute mTBI phase. Collectively, our findings identify mtDNA–ZBP1 signaling as a key mechanism regulating microglial activation in mTBI.

Traumatic brain injury (TBI) is a major healthcare concern with no effective treatment options. Persistent neuroinflammation is a major contributor to cognitive and motor impairments. Microglia, the resident immune cells of the central nervous system (CNS), are essential for the neuroinflammatory response induced by TBI, which differs significantly between mild and moderate-to-severe forms. In mild TBI (mTBI), microglial activation is typically transient and spatially restricted, leading to an early but controlled immune response that supports debris clearance and synaptic remodeling (1, 2). By contrast, in moderate to severe TBI, microglial responses are more widespread and prolonged, often characterized by sustained production of proinflammatory cytokines and chronic neuroinflammation, which contribute to extensive neuronal loss and white matter damage (1, 2). Thus, when properly regulated, initial microglial activation is neuroprotective; however, persistent activation can shift activated microglia toward a maladaptive, neurotoxic profile that contributes to long-term neurological deficits.

An increasing number of reports describe chronic microglial activation beyond the acute phase in mTBI (3–8). Also, the importance of bidirectional crosstalk between neurons and microglia has been proposed (9–11). Dysfunctional neurons release signals that perpetuate microglial reactivity, while activated microglia alter neuronal excitability and synaptic plasticity through cytokine signaling, complement cascade activation, and aberrant synaptic pruning (9, 12). This maladaptive feedback loop is thought to contribute to the persistent cognitive and behavioral deficits observed after mTBI and may underlie the increased susceptibility to later neurodegenerative processes (13). Still, the molecular mechanisms underlying these processes remain poorly understood.

Mitochondrial dysfunction, particularly in neurons, is an early hallmark of TBI. Due to their high metabolic demand, neuronal mitochondria are especially susceptible to oxidative stress, which can lead to mitochondrial DNA (mtDNA) damage (14, 15). While mtDNA is highly sensitive to oxidative injury, its role in mTBI remains poorly understood and is often considered a downstream byproduct of TBI-induced inflammation and oxidative stress. Although the dynamic interplay between microglial activation and neuronal dysfunction is increasingly recognized as a mechanism linking the acute immune response in mTBI to chronic neurological outcomes (9–11), the molecular mechanisms of neuron–microglia crosstalk remain incompletely elucidated.

Closed-head mTBI, which accounts for approximately 90% of all TBI cases, is primarily caused by falls, motor vehicle accidents, and contact sports (16). The common injury mechanism involves impact-induced acceleration and deceleration forces that damage brain parenchyma (17). However, frequently used rodent TBI models, such as controlled cortical impact (CCI) and fluid percussion injury (FPI), do not fully replicate these biomechanical features (18). In this study, we employed a previously validated, clinically relevant, closed-head weight-drop mouse model of mTBI to investigate the development of neuroinflammation (19). Our findings demonstrate that mTBI induces mtDNA-specific damage, leading to its release from neurons via extracellular vesicles (EVs), which subsequently activate microglia. We identified Z-DNA–binding protein 1 (ZBP1) as a key intracellular receptor that binds neuronal mtDNA and triggers the expression of proinflammatory markers in microglia. ZBP1 knockout (KO) mice showed an absence of microglial activation in mTBI. Moreover, while mTBI induced cognitive deficits in WT mice, these were exacerbated in ZBP1 KO mice, further supporting a key role of ZBP1 in microglial activation. We propose that mtDNA–ZBP1 axis is a mechanism of microglial activation in mTBI.

Results

mTBI Induces Transient mtDNA-Specific Damage and Expression of Innate Inflammatory Markers in the Cerebral Cortex During the Acute mTBI Phase.

To replicate the most frequent forms of human mTBI in mice, we used a modified version of a previously developed and validated TBI model (20), incorporating an impactor that delivers a controlled mechanical force to induce a single impact with consistent injury site and severity (19). This mTBI model induces an impact-driven 180° rotation of the mouse’s head and body, replicating the acceleration–deceleration forces typically observed in human mTBI (SI Appendix, Fig. S1A). Injury severity was assessed using the Neurological Severity Score (NSS) performed at multiple time points (19, 21). Based on NSS, we determined that dropping a 150 g weight from 1 m induced mTBI, whereas dropping 150 g weight from 1.5 m resulted in moderate/severe TBI (SI Appendix, Fig. S1B). To independently confirm injury severity, we measured the righting reflex in mice subjected to mTBI, which was less than 7 min (SI Appendix, Fig. S1C), further validating mild injury based on prior studies (22). We also assessed the release of high-mobility group box 1 (HMGB1), a well-known damage-associated molecular pattern (DAMP) marker (23), in plasma samples at 3 hpi. HMGB1 was detected only in the plasma of mice with moderate/severe TBI, but not in those with mTBI (SI Appendix, Fig. S1D). Together, these findings confirm that our clinically relevant mouse model replicates mTBI, which we used to investigate the link between mitochondrial dysfunction and neuroinflammation.

Mitochondrial dysfunction is a hallmark of TBI, but the kinetics and extent of DNA damage—particularly mtDNA damage, given its high susceptibility to injury—have not been examined in mTBI. We used PCR-based amplification of long DNA fragments (LAqPCR), a method that allows simultaneous assessment of mtDNA and nuclear (nu)DNA integrity (24). Analysis of multiple brain regions revealed a transient increase in mtDNA-specific damage in the cerebral cortex immediately after injury (1 and 3 hpi; Fig. 1A). In contrast, mtDNA damage in the hippocampus gradually increased during the postacute mTBI phase (3 and 10 dpi; Fig. 1B). No significant nuDNA damage was detected in either the cerebral cortex or hippocampus up to 10 dpi (Fig. 1 A and B). Furthermore, we did not detect significant changes in mtDNA or nuDNA integrity in the thalamus, cerebellum, liver, or heart (SI Appendix, Fig. S2). These findings demonstrate spatiotemporal mtDNA-specific damage in mTBI, with acute, transient cortical damage and gradual, postacute accumulation in the hippocampus. Moreover, the absence of nuDNA damage indirectly validates the mild nature of the injury.

Fig. 1.

mTBI induces concurrent mtDNA-specific damage and expression of innate inflammatory markers in the murine cerebral cortex acutely postinjury. mtDNA and nuDNA damage were measured in the (A) cerebral cortex and (B) hippocampus of mice at several timepoints (1 h to 10 d) postinjury (LA-qPCR; N = 15 for sham, N = 10 for mTBI groups). (C) Heat map of the inflammatory response (Innate and Adaptive Array, N = 10 for sham and mTBI groups), and (D) validation of expression of selected inflammatory markers in the cerebral cortex of mice subjected to mTBI (RT-qPCR, N = 10 for sham and mTBI groups/group). The average expression of inflammatory markers in sham animals were set as 1. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, calculated using a Kruskal–Wallis one-way ANOVA test. IL, interleukin; TNFα, tumor necrosis factor α; KC (CXCL1), chemokine (C-X-C motif) ligand 1.

Next, we investigated longitudinal changes in cortical inflammatory marker expression following mTBI using an innate and adaptive immune response array. Analysis of 84 distinct targets revealed a progression from acute innate inflammation → interferon-driven antiviral state → adaptive immune activation → chronic mixed inflammatory/adaptive response over 10 dpi (SI Appendix, Fig. S3 A and B). This indicates that the initial innate inflammatory response did not resolve during the acute phase but instead progressed into a chronic neuroinflammatory response. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses identified cytokine, type I interferon, and chemokine signaling as the predominant biological processes in the acute mTBI phase (SI Appendix, Fig. S3C). The increased expression of several proinflammatory cytokines, IL-1α, IL-1β, IL-6, TNFα, and KC (CXCL1), in the cortex during the acute mTBI phase (1 to 3 hpi) was validated by RT-qPCR (Fig. 1 C and D). We have previously identified, in cultured epithelial cells, a direct link between mtDNA damage and inflammatory responses (25, 26). Here, using an animal model, we measured early cortical mtDNA damage associated with acute innate inflammatory responses in mTBI. Together, our findings reveal previously unrecognized concurrent transient mtDNA-specific damage and proinflammatory cytokine expression in the mouse cerebral cortex during the acute phase of mTBI. These data suggest a potential spatiotemporal link between mtDNA damage and inflammatory responses but whether or how they are interconnected is currently unknown.

Neurons Subjected to In Vitro Injury Model of mTBI Release Damaged mtDNA Via EVs.

Highly metabolic neurons are particularly susceptible to injury, and TBI has been shown to increase intramitochondrial oxidative stress due to Ca²⁺ overload, with mtDNA being especially vulnerable to oxidative damage due to limited repair mechanisms and lack of protective histones (14, 27). To directly test the effect of TBI on DNA damage in neurons, we used rapid stretch injury (RSI), to induce injury similar to TBI to cells, effectively serving as an in vitro TBI model (28). Using neurons differentiated from human neural stem cells (hNSCs) that closely resemble primary neurons, we detected transient, mtDNA-specific damage immediately after RSI (0.5 to 1 hpi; Fig. 2A), without evidence of nuDNA damage (Fig. 2B). These findings confirm the high sensitivity of neuronal mtDNA to injury, while the absence of nuDNA damage indicates mild injury induced by RSI. Additionally, we measured a transient, acute decrease in neuronal mtDNA content at 0.5 to 1 hpi post-RSI (Fig. 2C). Thus, we tested for the presence of DNA in EVs derived from hNSCs-derived neurons using primers specific to two distinct mtDNA genes, COXIII and NAD1 (SI Appendix, Fig. S4A). At 1 h post-RSI, we measured a 4-fold increase in mtDNA in neuron-derived EVs (Fig. 2D), whereas no changes in nuDNA content were detected at any time point (Fig. 2E). A quantitative comparative analysis of mtDNA and nuDNA in EVs at 1 h post-RSI revealed over 250-fold more mtDNA than nuDNA (Fig. 2F), indicating the specificity of neuronal mtDNA release in response to RSI. Moreover, RSI-induced release of neuronal mtDNA occurred predominantly within EVs, as the “free-floating” fraction (media after EV isolation) contained negligible amounts of mtDNA (Fig. 2G). Nanoparticle tracking analysis (NTA) of neuron-derived EVs showed a marked increase in both total EVs number and mean size at 1 h post-RSI (Fig. 2H), suggesting the release of a specific EVs subpopulation in response to injury. Western blot analysis confirmed the presence of exosomal (CD63) and neuronal (NFL) markers in EVs released at the earliest time points post-RSI (Fig. 2I). To independently validate these findings, we subjected differentiated rat neuroblastoma B35 cells to RSI. Similarly, as we detected in neurons differentiated from hNSCs, we measured early mtDNA-specific damage, decreased mtDNA content, and selective release of mtDNA via EVs in response to RSI (SI Appendix, Fig. S4 B–G).

Fig. 2.

RSI leads to acute mtDNA damage and its release by EVs from hNSCsdifferentiated neurons. (A) mtDNA damage, (B) nuDNA damage, and (C) mtDNA content in hNSCs differentiated neurons subjected to RSI (LA-qPCR and qPCR, respectively, N = 3). (D) Amount of mtDNA and (E) nuDNA within EVs isolated from medium of cultured hNSCs-differentiated neurons (qPCR, N = 3). (F) Comparison of mtDNA and nuDNA amounts in EVs isolated from medium at 1 h post-RSI from hNSCs-differentiated neurons (qPCR, N = 3). (G) Quantification of mtDNA in EVs and culture media after EV isolation (qPCR, N = 9). (H) Number and size of EVs released from hNSCs-differentiated neurons at 1 and 3hpi (NTA, N = 3). (I) The presence of neuronal (NFL) and exosomal (CD63) markers in EVs isolated from media of control and RSI hNSCs− differentiated neurons (Western blotting, representative images on N = 3 is shown). *P < 0.05, **P < 0.01, and ****P < 0.0001, calculated with Kruskal–Wallis one-way ANOVA or t test for two-group analysis. CD63, tetraspanin CD63; NFL, neurofilament light.

We previously reported elevated levels of mtDNA within EVs isolated from the blood of human TBI survivors, proportional to injury severity (29), as well as in the plasma of mice subjected to moderate/severe TBI (30). To extend these findings, we analyzed the DNA content of EVs isolated from the plasma of mice subjected to mTBI. A significant increase in mtDNA, but not nuDNA, was detected in plasma-derived EVs at 1 hpi (SI Appendix, Fig. S5A). Western blot analysis revealed increased levels of neuronal (neural cell adhesion molecule, NCAM) and exosomal (CD63) markers in plasma EVs at 1 hpi (SI Appendix, Fig. S5B). These findings parallel our in vitro results, demonstrating the early release of mtDNA from neurons via CD63⁺ EVs to the circulation. Together, these results identify an active process of mtDNA release in mTBI, mediated by CD63⁺ neuronal EVs, as an early response to injury.

Neuronal mtDNA Is a Potent Activator of Microglia Via the Z-DNA–Binding Protein 1 (ZBP1) Signaling Pathway.

Microglial activation is an early event in TBI to quickly mitigate initial injury, serve for neuroprotection, and maintain homeostasis. To test whether mTBI directly activates microglia, we subjected cultured HMC3 microglia to RSI. No changes in the expression of two prominent proinflammatory markers, IL-6 and TNFα, were detected at 24 h post-RSI (Fig. 3A). In addition, RSI did not induce the release of mtDNA or nuDNA via EVs from microglia at 24 h post-RSI (SI Appendix, Fig. S5C). These findings are consistent with previous reports showing that primary (stretch) injury of TBI affects microglial cytoskeletal reorganization rather than inflammatory marker expression (31).

Fig. 3.

Neuronal mtDNA activates microglia by binding to ZBP1. (A) RSI does not increase IL-6 and TNFα expression in cultured microglia (RT-qPCR, N = 9). Increased secretion of IL-6 and TNFα from microglia treated with (B) RSI-induced neuronal EVs, (C) lipofectamine encapsulated isolated neuronal mtDNA, measured at 24 h after treatment (ELISA, N = 4). (D) Neuronal mtDNA binds to ZBP1 but not to other DNA receptors at 5 h posttreatment (PLA). (E) Activation of ZBP1 pathway by neuronal mtDNA demonstrated through interactions of ZBP1 with TBK1 and IRF3 (PLA). In vitro data are based on three independent experiments. Representative PLA images from three independent experiments are shown, with PLA signal quantification from five independent images. **P < 0.01, calculated with Kruskal–Wallis one-way ANOVA. ZBP1, Z-DNA binding protein 1; cGAS, cyclic GMP-AMP synthase; NLRP3, NOD-, LRR- and pyrin domain-containing protein 3; TLR9, Toll-like receptor 9; AIM2, absent in melanoma 2; TBK1, TANK-binding kinase 1; IRF3, interferon regulatory factor 3.

We next tested the effects of neuron-derived EVs on microglia and found robust production of TNFα and IL-6 when microglia were exposed to EVs derived from RSI-injured neurons (1 hpi), but not from control neurons (Fig. 3B). Because we had detected a high amount of mtDNA in EVs derived from RSI-injured human neurons (Fig. 2 D–G), we treated naïve microglia with equal amounts of isolated mtDNA or nuDNA encapsulated with lipofectamine to mimic EV-mediated delivery. Increased TNFα and IL-6 production was observed only in microglia treated with mtDNA (Fig. 3C). In addition, EVs isolated from mouse plasma at 1 hpi that contained elevated amounts of mtDNA induced TNFα and IL-6 production in cultured microglia compared to EVs from sham animals (SI Appendix, Fig. S5D). In addition, we treated cultured astrocytes with an equal number of EVs isolated from CTR and RSI neurons and measured GFAP expression as a hallmark of reactive gliosis. The lack of increased GFAP expression (SI Appendix, Fig. S5E) further supports the specificity of neuronal EVs in activating microglia, but not astrocytes. These results identify neuron-derived mtDNA as a potent and specific activator of microglia.

We then aimed to identify the DNA receptor within microglia that recognizes neuronal mtDNA. A key feature of EVs is their ability to deliver cargo into the cytoplasm of target cells, and several intracellular nucleic acid sensors involved in immune responses have been identified in mammalian cells (32). Using a proximity ligation assay (PLA), which allows detection of binary interactions between DNA and DNA receptors, we tested all known intracellular nucleic acid receptors that could potentially interact with neuronal mtDNA. Because PLA is highly dependent on antibody specificity, we used two approaches: Microglia were treated with a) isolated neuronal mtDNA and b) neuronal mtDNA amplified by PCR in the presence of bromodeoxyuridine (BrdU), both encapsulated with lipofectamine. Using DNA- and BrdU-specific antibodies, PLA signals showed interactions only with Z-DNA–binding protein 1 (ZBP1), but not with cyclic GMP-AMP synthase (cGAS), NOD-, LRR-, and pyrin domain-containing protein 3 (NLRP3), Toll-like receptor 9 (TLR9), or Absent in Melanoma 2 (AIM2) (Fig. 3D and SI Appendix, Fig. S6A). We further validated the interaction between neuronal mtDNA and ZBP1 using an alternative set of antibodies (SI Appendix, Fig. S6B). Activation of the ZBP1 pathway by neuronal mtDNA was also confirmed by testing interactions with downstream mediators of ZBP1 (33): TANK-binding kinase 1 (TBK1) and interferon regulatory factor 3 (IRF3) (Fig. 3E and SI Appendix, Fig. S7A). Although activation of the cGAS–STING pathway in TBI has been reported only in aged animals (34), we tested for STING phosphorylation and its interaction with TBK1 and IRF3 in microglia transfected with neuronal mtDNA. No evidence of cGAS–STING activation was detected in mtDNA-treated microglia (SI Appendix, Fig. S7B), ruling out its direct involvement in mtDNA recognition, at least during the early phase. Together, these in vitro findings identify ZBP1 as the sole primary receptor for neuronal mtDNA in driving microglial activation.

Lack of Microglial Activation in the Cerebral Cortex of ZBP1 KO Mice Following mTBI.

Our in vitro studies identified ZBP1 as a specific receptor for neuron-derived mtDNA in microglia. To confirm this in vivo, we subjected wild-type (WT) and ZBP1 KO mice (SI Appendix, Fig. S8) to mTBI and compared acute inflammatory marker expression in the cerebral cortex. Compared to WT mice, ZBP1 KO mice showed reduced expression of IL-6 and KC (CXCL1) at 1 and 24 hpi (Fig. 4A). Analysis of Iba1-stained brain sections revealed suppressed microglial activation in the cortex of ZBP1 KO mice at 2 and 24 hpi, quantified based on morphometry, with the area and perimeter of Iba1⁺ cells at 24 hpi (Fig. 4 B and C). Interestingly, astrocyte activation, assessed by GFAP staining, was similar in WT and ZBP1 KO mice following mTBI supported by quantification of GFAP area to identify hypertrophic astrocytes (Fig. 4 D and E), indicating a specific role for ZBP1 in microglial activation.

Fig. 4.

Inhibition of microglia activation in the cerebral cortex of ZBP1 KO mice in mTBI. (A) Inhibition of IL-6 and KC (CXCL1) expression measured at 1 and 24 hpi in the cerebral cortex of WT and ZBP1 KO mice (RT-qPCR, N = 10/group). Microglia activation induced by mTBI in WT but not in ZBP1 KO determined by (B) Iba1 staining of cerebral cortex sections at 2 and 24 hpi showing larger soma and shortened/thickened processes, morphometric features associated with activation, and (C) quantification of increased microglia area and perimeter of Iba1+ cells (IF, N = 5/group). (D) mTBI causes similar astrocyte activation in both WT and ZBP1 KO mice at 2 and 24 hpi shown by hypertrophic astrocytes identified by GFAP staining of brain sections and (E) quantification of astrocyte area as a measure of the hypertrophy (IF, N = 5/group). (F) Primary microglia isolated from the cerebral cortex are free of astrocyte and neuronal contamination determined by Iba1, GFAP, and NeuN markers for microglia, astrocytes, and neurons, respectively (Western blotting, representative images on N = 3 is shown). (G) Isolated primary microglia from WT but not from ZBP1 KO mice at 1 hpi showed increased expression of TNFα indicating inhibition of microglia activation by mTBI in ZBP1 KO mice (RT-qPCR, N = 5/group). (H) Activation of naïve primary microglia isolated from WT but not from ZBP1 by mtDNA measured by the IL-6 expression (RT-qPCR, N = 5/group). **P < 0.01 and ***P < 0.001, calculated with Kruskal–Wallis one-way ANOVA or t test for two-group analysis. (Scale bar, 100 µm.) ZBP1, Z-DNA binding protein 1; IL, interleukin; KC (CXCL1), chemokine (C-X-C motif) ligand 1; Iba1, ionized calcium-binding adapter molecule 1; GFAP, glial fibrillary acidic protein; NeuN, neuronal nuclear protein; TNFα, tumor necrosis factor α.

To further validate the essential role of ZBP1 in microglial activation, we subjected WT and ZBP1 KO mice to mTBI, followed by primary microglial isolation from the cerebral cortex using magnetic-activated cell sorting (MACS). The purity of MACS-isolated primary microglia was confirmed by Western blot analysis using Iba1, GFAP, and NeuN antibodies as markers for microglia, astrocytes, and neurons, respectively (Fig. 4F). Inhibition of TNFα expression was measured in primary microglia isolated from ZBP1 KO mice at 1 hpi but not in WT microglia (Fig. 4G). Furthermore, primary microglia isolated from naïve WT and ZBP1 KO mice were treated with lipofectamine-encapsulated neuronal mtDNA. An inhibition of IL-6 expression was detected in microglia isolated from ZBP1 KO mice but not from WT mice (Fig. 4H). Together, these findings identify the critical role of ZBP1 as the primary receptor for neuron-derived mtDNA in driving microglial activation during the acute phase of mTBI.

Lack of ZBP1 Worsens Neurological Deficits Induced by mTBI.

Microglial activation plays a crucial role in the earliest phase postinjury, contributing to neuroprotection and repair (35). Our data revealed a lack of microglial activation in ZBP1 KO mice during the acute mTBI phase that as we proposed is due to inhibition of mtDNA/ZBP1 signaling pathway. However, we also detected accumulation of mtDNA-specific damage in postacute mTBI phase in the hippocampus (Fig. 1B). Thus, we investigated the long-term effects of mTBI in WT and ZBP1 KO mice. We used the novel object recognition (NOR) test to evaluate hippocampal-linked nonspatial memory impairments. The NOR test was quantified using the discrimination index (DI), measured at 2 h and 24 h posttraining to assess short- and long-term memory, respectively (36). The majority of WT and ZBP1 KO sham mice had DI values ranging between 25 and 50, indicating a preference for the novel object, intact short- and long-term memory, and no preinjury differences between WT and ZBP1 KO mice (Fig. 5A). Interestingly, while mTBI impaired both short- and long-term memory in WT mice assessed at 5 dpi and 21 dpi, the deficits were more severe in ZBP1 KO mice (Fig. 5A). In addition, we used the open field (OF) test to assess gross locomotor activity that reflect deficiency mainly in including cortex and hippocampus (37). In WT mice, the OF test revealed hyperlocomotor activity at 3 dpi, which normalized by 21 dpi (Fig. 5B), consistent with previous findings (38, 39). In contrast, ZBP1 KO mice exhibited persistent hyperactivity even at 21 dpi (Fig. 5B). We conclude that in our preclinical mouse model of mTBI induces acute locomotor impairments and chronic memory deficits in WT animals and that these deficits are exacerbated in ZBP1 KO mice. We speculate that the absence of ZBP1 prevents microglial activation, which is critical for injury resolution in the acute phase, and thus contribute to exacerbation of the injury also in the postacute phase.

Fig. 5.

mTBI worsens neurological deficits in ZBP1 KO mice. (A) The NOR test showed deficits in short and long memory impairment induced by mTBI at 5 and 21 dpi in WT animals that worsen in ZBP1 KO mice (N = 10/group). (B) The open field test showed transient (at 3 dpi) increase of gross motor activity in WT but persistent (at 3 and 21 dpi) increase of gross motor activity in ZBP1 KO mice induced by mTBI (N = 10/group). *P < 0.05, **P < 0.01, and ***P < 0.001, calculated with Kruskal–Wallis one-way ANOVA. (C) Model of microglia activation in mTBI by mtDNA–ZBP1 pathway. Neuron-specific mtDNA damage induced by mTBI results in the release of damaged mtDNA from neurons via EVs that activates microglia through by binding and activating the ZBP1 signaling pathway.

In summary, using both in vitro and clinically relevant mouse models of mTBI, we identified a pathway of microglial activation. We found that mTBI induces mtDNA-specific damage and promotes its release from neurons via EVs. Moreover, we observed spatiotemporal differences in mtDNA-specific damage: acute and transient in the cerebral cortex, but sustained in the hippocampus during the postacute phase of mTBI. We identified neuronal mtDNA as an activator of microglia through binding to the intracellular receptor ZBP1, leading to the expression of inflammatory markers (Fig. 5C). In addition, the absence of ZBP1 resulted in exacerbation neurological deficits in mice subjected to mTBI. Overall, our findings highlight the critical role of the mtDNA–ZBP1 axis in microglial activation that contribute neurological outcomes following mTBI.

Discussion

In this study, we show a direct and active role of neuron-derived mtDNA in microglial activation through a ZBP1-mediated signaling pathway. Although mtDNA is highly vulnerable to damage due to its limited repair mechanisms, lack of protective histones, and proximity to the mitochondrial respiratory chain (40), currently damage to mtDNA is viewed as a passive outcome of oxidative or inflammatory stress. In particular, neuronal mtDNA is susceptible to oxidative damage because of the high energy demands of neurons (15).

Mechanical stretch, a key mechanism of primary injury in TBI, rapidly depolarizes the mitochondrial membrane and induces mtROS production in neurons (41, 42). Our in vitro and in vivo data demonstrated mtDNA-specific damage as early as 30 min post-RSI in cultured neurons and at 1 hpi in the cerebral cortex of injured animals, leading to the release of damaged mtDNA into the extracellular space—likely as part of mitochondrial quality control processes. We speculate that this mechanism reflects the bacterial origins of mitochondria, as bacteria extrude their DNA under stress, often via membrane vesicles, which function as signaling molecules (43–47). Cytoplasmic and extracellular mtDNA exerts a strong inflammatory effect. It is small circular DNA with bacterial-like methylation patterns and each cell contains hundreds to thousands of mtDNA copies. Moreover, during replication and stress conditions, mtDNA can form long single-stranded regions, three-stranded D-loop structures, oxidized lesions, and RNA–DNA hybrids—all of which are highly immunogenic. Because of its evolutionary bacterial origin, mtDNA is recognized by the immune system as “foreign” rather than “self” DNA triggering inflammatory responses similar to those elicited by pathogens (48). Our findings emphasize that even modest and transient mtDNA damage results in the release of hundreds to thousands of mtDNA copies into the extracellular space, thereby contributing to mTBI-induced neuropathology by triggering inflammatory responses.

Our data also highlight the underexplored role of EVs in TBI-induced pathologies. We identified a critical function of neuron-derived EVs, enriched with mtDNA after TBI, in activating microglia immediately after injury. While various mechanisms of glial cell activation have been proposed, the role of mitochondrial signals remains less understood. Mitochondrial DAMPs—including cytochrome c, cardiolipin, mitochondrial transcription factor A (TFAM), and also mtDNA—have recently emerged as important amplifiers of neuroinflammation in neurodegenerative diseases (49, 50). Still, their role in TBI has not been extensively studied (51) and our data shed light on the role of mtDAMPs in particular mtDNA. In general, DAMPs released from necrotic or damaged cells typically activate receptors on the surface of immune cells. Indeed, in moderate to severe TBI, DAMPs from injured cells or tissues activate microglia through various surface receptors, including TLR4, cytokine receptors, tyrosine kinase receptors, TREM2, and purinergic receptors (P2X/P2Y) (52). However, necrosis-derived DAMPs are susceptible to degradation by proteases and nucleases in the extracellular environment, including CSF and blood, which limits their impact over time. In contrast, EVs protect their molecular cargo, can cross cellular membranes and BBB, and activate intracellular receptors upon fusion with recipient cells. Therefore, the mechanism of microglial activation identified in this study—mediated by EV-encapsulated mtDNA—represents a fundamentally distinct and potentially more sustained pathway compared with activation via “classical” necrotic DAMPs. Notably, while mTBI does not cause immediate tissue damage or DAMP generation, microglial activation still occurs (53). Our study provides evidence that EV-mediated transfer of neuronal mtDNA causes microglial activation in mTBI.

Although this study focuses primarily on the acute phase of mTBI, we also detected accumulation of mtDNA damage in the hippocampus during the postacute phase (3 and 10 dpi), suggesting that microglial activation by neuron-derived mtDNA extends beyond the acute phase and is not restricted to the cerebral cortex. Initial activation and recruitment of microglia to the injury site are critical for resolving injury (1, 54). Thus, failure of microglial activation during the acute phase of mTBI would be expected to worsen neurological outcomes. Indeed, it has been shown that chronic, but not acute, depletion of microglia worsens neurological deficits in animals subjected to TBI (55). Our comparative assessment of motor and cognitive deficits in WT and ZBP1 KO mice following mTBI further supports this view, as ZBP1 KO mice developed more severe neurological deficits. However, our data cannot exclude the possibility that while inhibition of the mtDNA/ZBP1 pathway in the acute phase is detrimental, its role in the postacute phase may differ, as chronic microglial activation has been shown to exacerbate TBI outcomes (56). The development of inducible, microglia-specific ZBP1 KO models is warranted to fully elucidate the role of the mtDNA/ZBP1 pathway in both acute and postacute phases.

Another key finding of our study is the critical role of ZBP1 in microglial activation in mTBI. We demonstrated that neuron-derived mtDNA binds to and activates the ZBP1 signaling pathway in microglia and identified ZBP1 as the primary and sole DNA sensor during microglial activation. Recent reports have shown enhanced expression of ZBP1 particularly in microglia (57, 58). Interestingly, we showed that neuronal mtDNA does not directly interact with other known DNA/RNA sensors, such as cGAS, NLRP3, or TLR9, in the acute phase. While NLRP3 inflammasome and cGAS/STING pathway activation have been implicated in neurodegenerative diseases (59–62), their roles in TBI-related neuroinflammation remain less defined. cGAS/STING pathway activation in TBI has been observed only in aged animals (34). NLRP3 activation has been reported in TBI (63, 64), but it requires a priming to induce expression of key inflammasome components (65), with protein levels typically peaking 3 to 7 dpi in the rat cortex (66). Thus, we do not exclude the potential involvement of other DNA receptors in neuroinflammation, particularly during chronic mTBI phases. It is possible that early ZBP1 activation initiates the expression of inflammatory ligands such as TNFα, which may be required to prime the NLRP3 inflammasome. Similarly, although we did not observe cGAS/STING pathway activation in our models, emerging evidence suggests potential cooperation between ZBP1 and cGAS (67). ZBP1’s role is well established in pathogen responses, where it detects viral nucleic acids and initiates inflammation and programmed cell death to control infection (68). An increasing number of reports show a role for ZBP1 in sterile inflammation, including skin and bowel inflammation (69, 70), doxorubicin-induced cardiotoxicity (67), bone marrow failure (71), and now in mTBI. At this point, we propose that mtDNA released from injured neurons via EVs act as a DAMP that triggers innate immune signaling in microglia. When internalized, cytosolic mtDNA may bind to ZBP1 and initiate downstream inflammatory signaling pathways, including activation of NF-κB and inflammasome-associated responses, leading to increased production of proinflammatory cytokines such as TNFα and IL-6. Consequently, mtDNA–ZBP1 interactions promote microglial activation and contribute to neuroinflammatory propagation following neuronal injury. Still, this potential mechanism needs further research to clarify downstream signaling pathway and potential cooperation between ZBP1 and other DNA receptors in TBI.

This study has several limitations. We used only male mice to minimize variability in TBI outcomes, although sex differences in TBI responses have been reported. Future studies should include female mice to address potential sex-specific effects. We also employed only one injury severity and a single TBI model. Expanding future investigations to include moderate and severe TBI, as well as alternative models such as CCI and FPI, will enhance potential generalizability of our findings. This study also focused exclusively on young adult animals. Given the increased TBI risk in older individuals, future work should incorporate middle-aged and aged mice to better reflect age-related susceptibility and responses.

In summary, using both in vitro and clinically relevant mouse models of mTBI, we identified a pathway of microglial activation. We show that mTBI causes mtDNA-specific damage in neurons, triggering its release via EVs. This damage exhibits distinct spatiotemporal patterns—acute and transient in the cerebral cortex, but sustained in the hippocampus during the postacute phase. Released neuronal mtDNA activates microglia by binding to the intracellular receptor ZBP1, driving inflammatory gene expression. Importantly, loss of ZBP1 exacerbates neurological impairments following mTBI. These results underscore the pivotal role of the mtDNA–ZBP1 axis in microglial activation in mTBI.

Materials and Methods

Animal and Experimental Procedures.

Wild-type (WT) C57BL/6J mice (strain #000664) were purchased from The Jackson Laboratory. C57BL/6J ZBP1 KO mice were a generous gift from Dr. Balachandran (Fox Chase Cancer Center, Philadelphia, PA). Animals were housed under a 12 h light/dark cycle at 21 to 23 °C with free access to a standard chow diet and water. All procedures were approved by the Institutional Animal Care and Use Committee at the University of Texas Medical Branch (Protocol number: 1911088B), following the US NIH guidelines.

Closed-skull weight-drop mouse mTBI model.

To induce mTBI, we used a nonpenetrating, closed-skull weight-drop model that we adapted from Kane et al. (20) that we recently improved by introducing an impactor to reduce variability in injury site and severity, and allow for equal distribution of the damage across both hemispheres (19). Briefly, 10- to 12-wk-old male C57BL/6J WT and ZBP1 KO mice underwent general health assessments and handling for 1 min per day for 5 d to minimize handling-related stress prior to injury. To induce injury, mice were anesthetized with 4% isoflurane until the righting and paw pinch reflexes were lost (1 to 1.5 min) and placed in a prone position on a slit and scored aluminum foil. The head was positioned directly under a polycarbonate guiding tube with the impactor located at the end. A brass 150 g weight was dropped from a height of 1 m to generate mTBI or from 1.5 m to generate moderate/severe TBI, causing the impactor to strike the dorsal side of the mouse’s head. This impact resulted in the mouse breaking through the aluminum foil barrier and undergoing a 180-degree flip before falling 10 cm onto a foam cushion. Postinjury, mice were immediately placed on a warming pad until they regained consciousness. The righting reflex, defined as the time taken to transition from a supine to a prone/standing position, was recorded. Sham animals underwent anesthesia and placement on aluminum foil without injury. At various time points postinjury, mice were killed, and tissue and blood samples were collected.

Evaluation of neurological deficits.

The NSS was used to assess injury severity. This test is considered an animal analog to the Glasgow Coma Scale used in clinical settings and primarily evaluates deficits related to the sensorimotor cortex. The NSS includes assessments of exit circle performance, seeking behavior, monoparesis/hemiparesis, paw grip strength, straight walking, startle reflex, beam walking (3 cm, 2 cm, and 1 cm widths), and round stick balancing (19, 21).

To assess hippocampal-linked nonspatial memory deficits, we performed the novel object recognition (NOR) test over 4 d, including habituation, training, and testing at 2 h and 24 h to evaluate short- and long-term memory, respectively (36). Briefly, on the first day, mice were introduced to an open chamber without objects to familiarize them with the environment for 10 min. On the second day, the habituation session was repeated for 10 min. On the third day, two identical objects were placed in the chamber, and the mouse was given 10 min to explore them. Two and twenty-four hours later, one of the objects was replaced with a novel object that the mouse had not encountered before. The AnyMaze tracking system recorded the time spent exploring each object. We used the DI to analyzed NOR defined as (time exploring the novel object − time exploring the familiar)/(time exploring novel + familiar) * 100 (36). A DI of 0 indicates equal preference for familiar and novel objects, a positive DI indicates a preference for the novel object, and a negative DI indicates a preference for the familiar object.

We used the open field (OF) test to assess gross locomotor activity. Briefly, mice were individually placed in a large, open, and square (40 cm x 40 cm) arena for 10 min. The AnyMaze tracking system recorded the total distance traveled and time spent in the center vs. periphery.

Cell Lines.

All cell lines used inM this study were maintained in a humidified incubator at 37 °C with 5% CO₂. The K048 line of hNSC was cultured as described in (72). Briefly, cells were grown as neurospheres in DMEM/F12 basic media supplemented with 20 ng/mL of epidermal growth factor, 20 ng/mL of fibroblast growth factor 2, 10 ng/mL of leukemia inhibitor factor, and N2 and passaged every 10 d. For neural differentiation, hNSCs were seeded at a density of 5.2 × 10⁴ cm⁻² in 24-well BIOFLEX® culture plates (Flexcell International Co.) precoated with 0.01% poly‐D‐lysine and 1 μg cm⁻² laminin (Invitrogen). Cells were incubated with media containing 20 ng/mL of EGF, 10 ng/mL of leukemia inhibitor factor, and 1 μg/mL of laminin for 4 d, and then switched to DMEM/F12 medium with WW1 for 10 d with two-thirds of the medium replaced every 3 to 4 d. Human HMC3 microglia (ATCC CRL-3304) were cultured in EMEM (ATCC #30-2020) supplemented with 10% heat-inactivated FBS (GIBCO #10082147) and 100 U/mL of penicillin–streptomycin (GIBCO #15140148). In experiments that require EVs isolation/characterization, we used EV-depleted FBS (FBS was ultracentrifuged 150,000×g for 15 h).

Rapid Stretch Injury.

RSI was performed using the FX-500 T™ Tension System (Flexcell International Corporation) as described earlier (28). Briefly, hNSCs were cultured on BIOFLEX® culture plates prior to RSI as mentioned above. To test the effect of RSI on microglia, 100,000 HMC3 microglia were seeded on BIOFLEX® culture plates precoated with 0.01% poly-D-lysine and 1 μg cm⁻² laminin the day before RSI. The cells were stretched at a frequency of 20 cycles, with a minimum vacuum pressure of −80 kPa, over 45 s (0.67 Hz) at room temperature. After RSI, hNSCs and HMC3 microglia were returned to the 37 °C cell culture incubators.

Mitochondrial and Nuclear DNA Damage.

To assess mitochondrial and nuclear DNA damage, we used a gene-specific semiquantitative PCR assay followed by quantification of the DNA damage as described earlier (24). Briefly, total DNA was isolated from brain regions, EVs isolated from blood, or cultured cells using the DNeasy Blood & Tissue Kit (QIAGEN #69506) following the manufacturer’s recommendations. Mitochondrial DNA (mtDNA) damage was evaluated using two primer pairs, generating amplicons of 117 bp and 10 kb for mouse mtDNA, 211 bp and 8.9 kb for human NSC-derived neurons and 235 bp and 3.5 kb for rat B35 neuroblastoma cells. The short mtDNA amplicon served as a normalization factor for mtDNA copies. Nuclear DNA (nuDNA) integrity was assessed using a 10 kb amplicon for human and mouse and 12 kb for rat DNA. Taq DNA Polymerase and LongAmp Taq DNA Polymerase (New England BioLabs #M0273X and #M0323L) were used to amplify short and long DNA targets, respectively. The number of cycles was determined individually for each cell line, blood, or tissue to ensure logarithmic PCR amplification. Each PCR was performed in three technical replicates. Primer sequences are shown in Table 1.

Table 1.

Primers used in this study

Name	Forward (5′–3′)	Reverse (5′–3′)
Primers for measurement of mitochondrial and nuclear DNA damage
m-mt117bp	CCCAGCTACTACCATCATTCAAGT	GATGGTTTGGGAGATTGGTTGATGT
m-mt10kb	GCCAGCCTGACCCATAGCCATAATAT	GAGAGATTTTATGGGTGTAATGCGG
m-nuDNA10kb	TATCTCTCTTCCTCTTCACTTCTCCCCTGG	CGTGATGCCGCCGTTGAGGGTCTCCTG
h-mt221bp	CCCCACAAACCCCATTACTAAACCCA	TTTCATCATGCGGAGATGTTGGATGG
h-mt8.9 kb	TCTAAGCCTCCTTATTCGAGCCGA	TTTCATCATGCGGAGATGTTGGATGG
h-nuDNA10kb	TGGGATTACACGTGTGAACCAACC	GCTCTACCCTGTCCTCTACCGTCC
Primers for expression of inflammatory markers
m-IL1α	AGTAGCAACCAACGGGAAGG	CTTCCTCTGAGTCATTGGCGA
m-IL1β	TGGACCTTCCAGGATGAGGACA	GTTCATCTCGGAGCCTGTAGTG
m-IL6	GGAGTCACAGAAGGAGTGGC	AGGTTTGCCGAGTAGATCTCAA
m-KC	AGACCATGGCTGGGATTCAC	CGCGACCATTCTTGAGTGTG
m-IL4	CCCCCAGCTAGTTGTCATCC	TGACCTCGTTCAAAATGCCG
m-CCL5	CCTCACCATATGGCTCGGAC	GCACTTGCTGCTGGTGTAGA
m-TNFα	GGTGCCTATGTCTCAGCCTCTT	GCCATAGAACTGATGAGAGGGAG
Primers to quantify the amount of mtDNA and nuDNA
h-mtCOXIII	TGACCCACCAATCACATGC	ATCACATGGCTAGGCCGGAG
h-nuACTB	CATGTACGTTGCTATCCAGGC	CTCCTTAATGTCACGCACGAT
m-mt117bp	CCCAGCTACTACCATCATTCAAGT	GATGGTTTGGGAGATTGGTTGATGT
m-nuDNA	TTTGCTCCTGGGCCTCCAAGTT	AGCCCGTGACTGCCACAAATCA

Innate and Adaptive Immune Response Array.

Total RNA was isolated from the cerebral cortex of sham and TBI mice using the TRIzol Reagent (Invitrogen) and the RNeasy Mini Kit (Qiagen #74104), following the manufacturer’s instructions, including an on-column DNase treatment. A total of 300 to 1,000 ng of high-quality total RNA was reverse-transcribed using the RT2 First Strand Synthesis Kit (Qiagen #330404) and subsequently used for the RT² Profiler™ PCR Array Mouse Innate & Adaptive Immune Responses Array (Qiagen #330231) to evaluate the expression patterns of 84 genes involved in the mouse immune response. Qiagen’s online web analysis tool was used to generate comparative heat maps, and fold change was calculated by determining the ratio of mRNA levels to control values using the Δ threshold cycle (Ct) method (2^−ΔΔCt). All data were normalized to the average of five housekeeping genes: Gusb, Hprt, Hsp90ab1, Gapdh, and Actb.

RNA Extraction and RT-qPCR.

Total RNA was extracted from brain tissue or cells using the RNeasy Mini Kit (QIAGEN #74104), and cDNA was synthesized using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems #4368814) according to the manufacturer’s recommendations. cDNA from 0.5 to 1 μg of RNA was used for qPCR analysis with the Maxima SYBR Green/ROX qPCR Master Mix (Thermo Scientific #K0221) and the CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad) under the following cycling conditions: 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. The expression of inflammatory markers was normalized to ACTB. The ΔΔCt method was used for qRT-PCR data analysis. Each qPCR was performed in two technical replicates. Primer sequences are shown in Table 1.

EVs Isolation and Characterization with NTA.

EVs released into the culture medium of hNSCs or microglia and present in mouse plasma were isolated using ultracentrifugation, as earlier described (30). This method is commonly used to isolate “small EVs”. Briefly, three sequential centrifugation steps were performed to separate cellular debris from EVs (to preclear the medium/plasma): 1,000×g for 5 min, 5,000×g for 10 min, and 10,000×g for 30 min, followed by EVs isolation with 100,000×g for 2 h; all steps were performed at 4 °C. The EVs pellet was resuspended in PBS for subsequent assays. The number and size of EVs were measured in PBS-diluted, precleared medium/plasma using NTA with the NanoSight NS300 (Malvern Panalytical). One ml of precleared medium/plasma was diluted 10× with PBS. Samples were then recorded by NTA (three capture intervals of 60 s per sample, with camera level 7 and threshold 2). Each readout was performed in three technical replicates.

Real-Time qPCR.

To quantify the amount of mtDNA and nuDNA in EVs and determine mtDNA content in cultured cells, total DNA was isolated from EVs and cells using the DNeasy Blood & Tissue Kit (QIAGEN #69506) according to the manufacturer’s recommendations. qPCR was performed using the CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad) and the Maxima SYBR Green/ROX qPCR Master Mix (Thermo Scientific #K0221). The thermal cycling conditions were as follows: 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. To determine mtDNA content in cultured cells, the expression of the mtDNA-specific genes was compared to the expression of the nuclear gene (ACTB), which was used as a control/normalization factor. Each qPCR was performed in technical duplicates. Primer sequences are shown in Table 1.

Microglia Treatment with Neuronal EVs.

Briefly, 300,000 microglia (HMC3 cells) were seeded per well in 12-well plates and treated with 10⁷ EVs isolated from control (CTR) and 1 h post-RSI hNSCs. HMC3 were also treated with isolated mtDNA (100 ng) and nuDNA (100 ng) from hNSCs, encapsulated within Lipofectamine 2000 (Thermo Fisher Scientific #11668019) to ensure its delivery to microglia’s cytoplasm, as well as EVs isolated from the plasma of sham and TBI mice. After 24 h treatment, the amount of IL-6 and TNFα in the culture medium was measured using DuoSet Kits (R&D Systems #DY206-05) according to the manufacturer’s recommendations.

Proximity Ligation Assay.

HMC3 microglia were seeded at a density of 20,000 cells per well in Lab-Tek Chamber Slides with Cover (Thermo Scientific #12-565-110 N). After 48 h at 37 °C with 5% CO₂, cells were treated with isolated mtDNA or bromodeoxyuridine (BrDU, Molecular Probes #B23151)-labeled mtDNA isolated from hNSCs, using Lipofectamine 2000 (Thermo Fisher Scientific #11668019) as the transfection reagent. After 5 h of treatment, cells were fixed and permeabilized with 4% paraformaldehyde and 0.2% Tween-20 in PBS for 20 min. Following three washes with PBS, PLA was performed according to the manufacturer’s instructions using the Duolink In Situ Kit (SIGMA #DUO92002-100RXN) with the following pairs of primary antibodies: dsDNA (Abcam #ab27156), BrDU (Sigma #B8434), ZBP1 (Protein Tech #13285-1-AP), ZBP1 (Novus #NBP1-76854), TLR9 (Cell Signaling #5845S), NLRP3 (Cell Signaling #13158S), cGAS (Cell Signaling #15102S), AIM2 (Cell Signaling #8055), STING (Cell Signaling #13647), IRF3 (Abcam #ab124399), TBK1 (Thermo Scientific #108A429), and P-Ser (SIGMA #P3430). Images were captured using a Nikon Eclipse 80i inverted fluorescence microscope equipped with a Photometrics CoolSNAP HQ2 camera and NIS-Elements BR 3.10 software. Protein interactions were visualized as red fluorescent signals. The number of interactions per cell was quantified by dividing the number of PLA signals by the number of DAPI signals visible in the field, using three separate fields from three independent experiments.

Isolation of Primary Microglia from Mouse Brain Tissue.

Primary microglia were isolated from the mouse cerebral cortex using MACS technology and cultured according to the manufacturer’s recommendations (Miltenyi Biotec). Briefly, we used the Adult Brain Dissociation Kit (#130-107-677) and the gentleMACS Octo Dissociator with Heaters to generate a single-cell suspension. Magnetic anti-mouse CD11b microbeads (#130-093-634) and MC columns (#130-042-201) were used to separate microglia from the rest of the brain cells. The entire mouse cortex was used for each microglia isolation. Isolated primary microglia were cultured on poly-L-Lysine (0.01%) coated 24-well plates using DMEM supplemented with L-glutamine, FBS, and penicillin/streptomycin according to Miltenyi’s protocol.

Western Analysis.

Total proteins of EVs isolated from 1 mL of cultured medium (Fig. 2I) or 100 µL of mouse plasma (SI Appendix, Fig. S7B), 100 µL of mouse plasma, 40 µg of total cell extract of the cerebral cortex (SI Appendix, Fig. S8B), or primary microglia isolated from the mouse brain cortex (Fig. 4E) were precipitated with acetone and lysed with NP-40 lysis buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% NP-40, and 5 mM EDTA). Proteins were separated using NuPAGE 4 to 12% BisTris Mini gels (Thermo Fisher Scientific #NP0329BOX) and transferred to a nitrocellulose membrane. Membranes were probed with primary antibodies: HMGB1 (Cell Signaling #3935), NCAM (Abcam #ab220360), CD63 (Invitrogen #PA5-92370), Iba1 (Abcam #ab283319), GFAP (Abcam #ab7260), NFL (Abcam #9035), ZBP1 (AdipoGen #AG-20B-0010), β-actin (Santa Cruz #sc-47778) and NeuN (Abcam #ab177487) at 4 °C overnight, followed by HRP-linked secondary anti-mouse/rabbit antibodies (Cell Signaling #7076S/#7074S) for 1 h at room temperature. HRP signals were detected using the ECL Chemiluminescence Substrate Reagent Kit (Thermo Fisher Scientific #WP20005) and the G:Box (Syngene) with GeneSnap software.

Immunofluorescence Analysis of Brain Sections.

Brain section analysis was performed in a blinded fashion. For ex vivo studies, right atrial perfusion was performed using cold PBS and 4% paraformaldehyde (PFA). Extracted brains were then fixed overnight in 4% PFA at 4 °C. Fixed whole brains were hemisected and cryoprotected in 30% sucrose for 3 d at 4 °C in 15 mL conical tubes. For cryosectioning, tissues were embedded in Optimal Cutting Temperature Compound (#4585, Fisher Scientific) and hardened in a −80 °C freezer for about 3 h. Thirty-micron sagittal brain sections were cut using a cryotome. Sections were blocked in 5% goat serum (diluted from 10% serum; #500,622, Life Technologies) in PBS with Triton-X100 for 1 h at room temperature. The primary antibodies (anti-Iba-1 #019-19741, Fujifilm Labchem Wako, and anti-GFAP #ab7260, Abcam) were incubated overnight at 4 °C. Brain sections were washed three times with PBS/Triton-X100 (PBST) for 15 min each with gentle agitation. The secondary antibody (ThermoFisher Scientific Goat Anti-Rabbit Alexa 568, 1:500, #A-11011) was incubated overnight with gentle agitation at room temperature. Sections were then washed with PBS and mounted on glass slides with mounting media (0.5% DMSO, 50% TDE, and 49.5% PBS).

Microglia activation was quantified by changes in morphology, including loss of thin, ramified/elongated processes, increased soma size, and a shift toward hypertrophic and amoeboid shapes. Iba1-stained sections were contrast-enhanced and segmented by threshold for microglia using ImageJ (NIH, Version 2.14.0). A semiautomated macro was developed in-house to streamline image processing and subsequent analysis for transformation index analysis. The process includes Z projection, background subtraction, thresholding, and conversion to a binary mask to isolate microglial morphology. All microglia within the field of view were selected, while those cut off by the image edges were excluded from the analysis. The number of microglia analyzed per image ranged from 10 to 30, depending on the field. Metrics collected were cell area and perimeter to assess activation through transformation index as described previously (73, 74).

Astrocyte activation, characterized by cellular hypertrophy, was quantified based on changes in total GFAP-positive area, encompassing both the soma and cellular processes. To assess morphological changes, confocal z-stacks were collapsed into maximum-intensity Z projections using ImageJ (NIH). Images were subjected to background subtraction, thresholding, and conversion to binary masks to isolate GFAP-positive astrocytes for quantification of area. All astrocytes within the field of view (N = 5/group) were analyzed, excluding cells intersecting the image borders, with an average value identified per field. Total GFAP-positive area per object was used as the primary metric to identify reactive astrocytes.

Statistical Analysis.

All statistical analyses were performed using GraphPad Prism software. Unless otherwise stated, the presented in vitro data are based on at least three biological replicates (cells from different passage numbers or differentiated at different passage numbers) and presented as mean ± SD. The number of animals used in each experiment is provided in the figure legends, and graphs are presented as mean ± SEM. Additionally, most techniques have technical replicates (qPCR, RT-qPCR, LA-qPCR, and ELISA). After performing a normality test on the generated data, t-tests were used to determine statistical significance between two groups. For the analysis of more than two groups, a Kruskal–Wallis one-way ANOVA test was used after performing normality tests. Significant differences are denoted as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.