cGAS-STING Signaling Pathway Mediates Brain Trauma-Induced Type I Interferon Response

28 Background : Inflammation is a key contributor of neuronal death and dysfunction following 29 traumatic brain injury (TBI). Recent evidence suggests that interferons may be a key regulator of 30 this response. Our studies evaluated the role of the Cyclic GMP-AMP Synthase-Stimulator of 31 Interferon Genes (cGAS-STING) signaling pathway a murine model of TBI. 32 Methods : Male, eight-week old wildtype, STING knockout ( -/- ), cGAS -/- , and NLRX1 -/- mice were 33 subjected to controlled cortical impact (CCI) or sham injury. Histopathological evaluation of tissue 34 damage was assessed using non-biased stereology, which was complemented by analysis at the 35 mRNA and protein level using qPCR and western blot analysis, respectively. 36 Results : We found that STING and Type I interferon-stimulated genes were upregulated after CCI 37 injury in a bi-phasic manner and that loss of cGAS or STING conferred neuroprotection 38 concomitant with a blunted inflammatory response at 24 hours post-injury. cGAS -/- animals 39 showed reduced motor deficit 4 days after injury (dpi), and amelioration of tissue damage was 40 seen in both groups of mice up to 14 dpi. Given that cGAS requires a cytosolic damage- or 41 pathogen- associated molecular pattern (DAMP/PAMP) to prompt downstream STING signaling, 42 we further show that mitochondrial DNA is present in the cytosol after TBI. Finally, our findings 43 demonstrate that NLRX1 may be an additional regulator that functions upstream to regulate cGAS- 44 STING pathway. 45 Conclusions : These findings suggest that the canonical cGAS-STING-mediated Type I interferon 46 signaling axis is a critical component of neural tissue damage following TBI and that mtDNA may be a possible trigger in this response. Taken together, these data confirm that STING-mediated IFN signaling is detrimental to TBI-induced tissue damage. We have shown that loss of cGAS or STING results in improved histological and functional measures up to 14 days after TBI. Additionally, we provide evidence that NLRX1 negatively regulates STING activation in the brain, offering an additional potential target for therapeutic intervention. Perhaps most significantly, this study is the first to investigate mtDNA as a possible trigger for STING-IFN signaling in neurotrauma. Overall, our findings indicate that the canonical cGAS-STING-mediated ISG response is an early neuroinflammatory event occurring after cortical trauma, which represents a novel therapeutic target for treatment. mouse an essential function of sting in the in interferon to monocytogenes and dinucleotides.


Introduction 51
Traumatic brain injury (TBI) is a complex neurological condition that is a leading cause of 52 death and disability in children and adults [1]. Injury occurs in two phases: an initial, acute 53 mechanical injury resulting from the external force, and secondary injury/cell death due to 54 complications such as hypoxia, ischemia, and inflammation [2,3]. While the use of improved 55 safety measures has helped minimize the severity of the initial impact, little progress has been 56 made in understanding or treating secondary injuries. 57 Neuroinflammation is a key mediator of secondary brain injury; however, anti-58 inflammatory pharmacological approaches largely fail in clinical trials [4]. Interferons (IFNs) are 59 elevated in post-mortem humans TBI samples (IFN-γ) [5,6] and in experimental TBI murine 60 models (IFN-α, IFN−β, IFN−γ) [6,7], but their functional role has been understudied in TBI. 61 Interferons are produced in response to detection of pathogen associated molecular patterns 62 (PAMPs) by pattern recognition receptors (PRRs) [8]. Upon detection of pathogenic nucleic acids, 63 Loss of STING [18], IFNAR [6], or IFNβ [20,21] function has been shown to be beneficial 121 in TBI outcome; however, the mechanism regulating their induction remains unclear. The 122 canonical STING-cGAS pathway is activated by binding of viral nucleic acids found in the 123 cytoplasm [14], resulting in production of the second messenger cGAMP which binds and 124 activates STING [11,12]. In addition, mitochondrial DNA (mtDNA) can activate STING in 125 models where mtDNA packaging proteins and mitochondrial permeability proteins are disrupted 126 genetically [13,27] and it is present in cerebral spinal fluid and serum following TBI [28,29]. 127 To determine whether mtDNA is present in the cytoplasm, we isolated the cytoplasmic 128 fraction of cells isolated from the ipsilateral cortex. We used primers that targeted two different 129 locations on the mitochondrial genome corresponding to the coding region for COX1 and ND1 130 To determine whether cytoplasmic nuclear DNA was also present, we performed western 137 blotting on cytoplasmic extracts at the 2hrs to evaluate the expression of the nuclear protein high 138 mobility group box protein 1 (HMGB1), whose expression is increased when nuclear DNA is 139 present in the cytosol [30,31]. cGAS also is more easily bound to and activated by HMGB1 coated 140 nuclear DNA than in its free form [32]. We found HMGB1 was present in cytosolic fractions 141 isolated from both contralateral and ipsilateral hemispheres ( Figure 2f); however, ipsilateral 142 cytoplasmic HMGB1 expression was not increased compared to contralateral (Figure 2g). This suggests that mtDNA is more likely to drive cGAS activation in the damaged cortex after CCI 144 injury. showed the number of apoptotic neurons was significantly reduced in the ipsilateral cortex of 158 STING -/mice after injury and trending toward a significant reduction in cGAS -/mice ( Figure 3f). 159 Although cGAS/STING deficiency is neuroprotective, no difference was observed on blood-brain 160 barrier function as seen by quantifying Evans Blue infiltration in the damaged cortex compared to 161 contralateral ( Figure 3g). Our results suggest that the cGAS-STING pathway contributes to the 162 neurotoxic effects induced by CCI injury. 163 Behavioral impairments have been previously assessed in IFNβ −/− mice after TBI [20], 164 therefore we sought to provide further confirmation that canonical cGAS-STING signaling is 165 critical in TBI outcome. Using rotarod assessment, we found no difference in motor function 166 between sham-injured cGAS -/and WT mice (Supplemental Figure 2a). However, cGAS -/mice 167 showed a significant reduction in motor deficit at 4dpi compared to WT (Supplemental Figure 2b) 168 but no difference at 7 and 14dpi (Supplemental Figure 2b). cGAS -/mice also showed a significant 169 reduction in lesion volume at 14dpi relative to WT (Supplemental Figure 2c-d), despite their 170 comparable motor performance (Supplemental Figure 2b). Similarly, STING -/mice also showed 171 reduced lesion volume at 14dpi (Supplemental Figure 2c-d). We also assessed mRNA levels of 172 IFNA4, IFNB1, and IL-6 at 14 days post-injury. Interestingly, all three genes were downregulated 173 at this chronic timepoint relative to WT sham animals (Supplemental Figure 4e). 174

Loss of cGAS-STING ameliorates pro-inflammatory gene expression after CCI injury 175
In addition to histological and functional changes, we profiled changes in gene expression 176 in the cortex at 24hrs post-injury in WT, STING -/-, and cGAS -/mice. We found no difference in 177 the contralateral cortex when compared to sham (Supplemental Figure 3), therefore we used 178 contralateral tissue when performing our relative analysis. Both STING -/-, and cGAS -/mice showed in the ipsilateral cortex when compared to WT. To provide further insight into the transcriptional 181 changes, we assessed the complete panel of genes described in Figure 1. We found all genes tested 182 were significantly altered in STING -/ mice compared to WT (Supplemental Figure 4) genes was used to verify purity of the isolated cell populations (Figure 5a). We observed that 193 microglia showed the greatest enrichment of transcripts for both cGAS and STING, when compared 194 to all other cell types, (Figure 5b-c). This suggests that microglia may represent the main cell 195 source influencing the type I interferon response via cGAS-STING pathway in TBI. 196

NLRX1 negatively regulates cGAS-STING activation after CCI injury 198
We recently show that loss of NLRX1 exacerbates tissue damage after CCI injury, in part, 199 by increasing NF-κB activity in microglial and/or peripheral-derived immune cell [45]. It is also 200 well-established that NLRX1 may sequester STING to prevent the interferon response [46], 201 however, this association has not been evaluated in the brain. To test whether NLRX1 represents 202 a novel upstream regulator of STING in the cortex after injury, we evaluated activated STING 203 expression and the ISG response. Interestingly, NLRX1 -/mice showed a significant increase in 204 activated (phosphorylated) p-STING (S365) compared to WT at 3dpi (Figure 5a The present work demonstrates that STING is upregulated in the ipsilateral cortex of CCI-223 injured mice, which correlates with a biphasic increase in a variety of cytokines, including IFNA4 224 and IFNB1. While previous work has shown that loss of endogenous STING reduces lesion size 225 following TBI [18], recent evidence suggests that STING may be able to function independently 226 of its canonical upstream mediator, cGAS [22,35,36,53]. Therefore, this study sought to determine 227 the effects of cGAS deficiency and to identify a potential DAMP that may influence the induction 228 of the canonical cGAS-STING pathway in CCI injury. Our data shows that cGAS and STING are 229 highly expressed in microglia and that cGAS -/mice display significant neuroprotection and a 230 blunted ISG response, similar to STING -/mice. This correlates with the observation of cytoplasmic 231 mtDNA in the damaged cortex and suggests mtDNA is a possible DAMP that induces cGAS-232 STING pathway in microglia leading to type I interferon-induced tissue damage in TBI. 233 Moreover, we demonstrate that NLRX1 is a novel upstream regulator of STING in this response. Our study selected a panel of genes associated with the Type I interferon response, 235 including pro-inflammatory (IL-6), anti-inflammatory (IL-10), and pro-immune migratory (MCP-236 1) cytokines, as well as transcription factors (STAT1, STAT2, and IRF7), interferons (IFNA4 and 237 IFNB1), and ISGs (CXCL10, IFIT1, IFIT3, and IFIH1). We determined that loss of cGAS or 238 STING resulted in a broadly blunted immune response 24hrs after injury. Recent work has 239 suggested that STING simultaneously stimulates the production of pro-and anti-inflammatory 240 cytokines to facilitate maintenance of gut homeostasis [55], and studies in mouse models of 241 systemic lupus erythematosus (SLE) have indicated STING signaling can be pro-or anti-242 inflammatory depending on the model [56-58]. Still, the autoimmune syndrome SAVI that results 243 from gain-of-function mutations in STING results in excessive inflammation, indicating a 244 primarily pro-inflammatory role for STING [59]. Our data shows altered mRNA expression of 245 both pro-and anti-inflammatory cytokines in cGAS -/and STING -/mice, suggesting that the effects 246 of cGAS-STING signaling is highly complex and likely context-dependent. Further, the 247 unselective upregulation of mRNAs for proteins with predominantly antiviral roles, such as IFIT1 248 and IFIT3, suggests that this innate immune pathway is activated aberrantly after injury, unlike its 249 normal role in viral or bacterial clearance. Further investigation is needed to clarify how the 250 balance of pro-and anti-inflammatory cytokines is disrupted or skewed by alterations in cGAS-251

STING activity. 252
Recent findings show that mRNA expression of STING and key ISGs are elevated up to 253 60 days after experimental TBI [20], indicating that STING activity may also contribute to chronic 254 neuroinflammation. Consistently, we found that cGAS -/mice showed reduced motor deficits 255 compared at 4dpi and reduced lesion volume up to 14 days post-injury. Interestingly, we found 256 that the type I interferon ISG response was significantly reduced by 4dpi (data not shown), and entirely resolved at 14 days. These data suggest that while the cGAS-STING signaling axis is 258 acutely activated after injury, additional subsequent mechanisms may further contribute to the 259 chronic progression on injury after trauma [60]. Further work is needed to define the temporal 260 dynamics of cGAS-STING signaling after TBI. 261 Conflicting evidence exists regarding whether NF-κΒ signaling is a major pathway 262 activated downstream of STING [61][62][63][64]. However, recent work in mice with a point mutation in 263 STING (S365A) that interfered with IRF3 binding elucidated that the switch between NF-κΒ 264 signaling and Type I interferon signaling was context-dependent [65]. With the generation of these 265 STING point mutation mouse models, future work could further define the contribution of 266 different downstream effects of STING during TBI. Yet, Type-1 IFN receptor (IFNAR1) knockout 267 mice are protected from TBI injury [6] indicating that the interferon pathway is still a major 268 contributor to neuroinflammation in TBI. However, future work is needed to elucidate cell-type 269 specific effects mediating the IFN response to TBI. Taken together, these data confirm that STING-mediated IFN signaling is detrimental to 281 TBI-induced tissue damage. We have shown that loss of cGAS or STING results in improved 282 histological and functional measures up to 14 days after TBI. Additionally, we provide evidence 283 that NLRX1 negatively regulates STING activation in the brain, offering an additional potential 284 target for therapeutic intervention. Perhaps most significantly, this study is the first to investigate 285 mtDNA as a possible trigger for STING-IFN signaling in neurotrauma. Overall, our findings 286 indicate that the canonical cGAS-STING-mediated ISG response is an early neuroinflammatory 287 event occurring after cortical trauma, which represents a novel therapeutic target for treatment.
Body temperature was continually monitored via rectal probe and maintained at 37°C with an 304 autoregulated heating pad. A 4 mm craniotomy was made with a portable drill over the right 305 parietal-temporal cortex (-2.5 mm A/P and 2.0 mm lateral from bregma). Moderate CCI was 306 induced with an eCCI-6.3 device (Custom Design and Fabrication, Richmond, VA, USA) using a 307   C  IFIT3  ATC ATG ATG GAG GTC AAC CG  TTG CAC ACC CTG TCT TCC AT  IFNA4  CTT TCC TCA TGA TCC TGG TAA TGA T  AAT CCA AAA TCC TTC CTG TCC TCC  IFNB1  AAC TCC ACC AGC AGA CAG TG  GGT ACC TTT GCA CCC TCC AG  RIG-I  GAG TAC CAC TTA AAG CCA GAG  AAT CCA TTT CTT CAG AGC ATC C  IFIH1  CGG AAG TTG GAG TCA AAG C  TTT GTT CAG TCT GAG TCA TGG  IL-10  AGA CCAAGGTGTCTACAAGGC  TCA TCA TGT ATG CTT CTA TGC

Evans Blue 384
Twenty-four hours after CCI injury, animals received an intravenous injection of 300 µL 385 Evans blue. After 3 hours, animals were sacrificed, and ipsilateral and contralateral hemispheres 386 were collected. Distribution of Evans Blue was verified by opening the thoracic and abdominal 387 cavities. Tissue was incubated in 500 µL 10% formamide at 55°C for 24 hours, then centrifuged 388 for 4 minutes at 210 x g to pellet the tissue. Absorbance for each hemisphere was measured in 389 triplicate at 610 nm. animal's performance was compared to its baseline measurement, and average performance for all 398 animals was reported. After the final day of testing, animals were euthanized for histology, qPCR, 399 or western blotting as described above. 400

Cytosolic Fraction 401
The cytosolic fraction was extracted as previously reported

Murine cells were isolated using the Worthington Dissociation Kit (Worthington Biochemical 412
Corporation, Lakewood, NJ, USA) and slight modifications to published protocols [43,78]. 413 Briefly, WT animals were deeply anesthetized with a ketamine (500 mg/kg)/xylazine (10 mg/kg) 414 cocktail and hand perfused with cold PBS to remove blood. The brain was removed, cortices 415 dissected, and finely minced in warmed papain with DNase. Tissue was digested in papain at 37°C

Statistical analysis 431
Data were analyzed with GraphPad Prism 9 (GraphPad, San Diego, CA, USA). A student's 432 two-tailed t-test was used for comparison of two experimental groups. One-way or two-way experimental groups as appropriate. Differences were considered statistically significant at p <

Competing Interests 467
The authors declare that they have no competing interests. 468