Loss of NLRX1 Exacerbates Neural Tissue Damage and NF-κB Signaling following Brain Injury

Traumatic brain injury (TBI) is a complex neurologic condition that has emerged as an important cause of morbidity and mortality in the adolescent and young adult populations. It is defined differently throughout the literature but is generally accepted as any external force that causes injury to the brain. These events may or may not involve injury to the skull or overlying tissues. Conversely, nontraumatic brain injuries have a wide range of causes but are not directly associated with physical trauma. Examples of nontraumatic brain injury can include brain tumors, meningitis, hypoxic/anoxic brain injury, stroke, or aneurysm. In traumatic and nontraumatic brain injury, the resulting morbidity and mortality seen clinically are not typically due to the actual primary injury itself, but rather the secondary changes that occur in the brain as a result of the injury. These secondary changes are associated with the activation of the innate and adaptive immune system and include inflammation, infiltration of immune cells, release of excitatory neurotransmitters, cerebral edema, vasospasm, ischemia, hypoxia, free-radical damage, and others (1–3). A great deal of research has looked into the role of the adaptive immune system in brain injury and how it can modulate these various secondary processes. However, only relatively recently have researchers begun to focus on the role of the innate immune system (1). The majority of studies have been focused on the role of microglial cells in the progression of the injury, because they are the predominant innate immune cell type in the brain (2). Somewhat similar to macrophages, microglial cells express a diverse variety of pattern recognition receptors (PRRs) that modulate their response to injury and drive many of the critical secondary changes seen in the brain following injury (2).

PRRs are proteins associated with plasma and endosomal membranes, as well as the cytosol itself. These receptors and sensors are responsible for recognition of various foreign and host molecular motifs (known as damage-associated molecular patterns or pathogen-associated molecular patterns). Once these proteins bind or sense their respective ligands, they are responsible for initiating a variety of cellular responses, including the activation of key inflammatory signaling pathways, such as the NF-κB signaling cascade. NF-κB signaling has previously been shown to be important in the pathogenesis of brain injury. For example, Lian et al. (4) showed that mice lacking IκBɑ, an NF-κB inhibitory protein, showed significantly increased neuroinflammation when assessed in a model of TBI. Moreover, in another TBI model, suppressing the NF-κB signaling pathway through exogenous VEGI treatment attenuated brain injury (5). Thus, these studies illustrate that unrestricted NF-κB signaling is an important component in the pathophysiology of brain injury.

NOD-like receptors (NLRs) are intracellular PRRs that are generally classified as either inflammasome-forming NLRs or regulatory NLRs. The inflammasome-forming NLRs are well studied and are characterized by their capacity to initiate the formation of the multiprotein inflammasome complex. Inflammasome formation ultimately facilitates the maturation and activation of the proinflammatory cytokines IL-18 and IL-1β. Several inflammasome-forming NLRs have been evaluated in the context of brain injury. For example, multiple studies have suggested that NLRP1 and NLRP3 may play critical regulatory roles in TBI in humans and rodents (6–8). Moreover, NLRC4 and AIM2 have also been implicated in contributing to nontraumatic brain injury in stroke models (9). However, although the inflammasome-forming NLRs have been well studied in multiple types of brain injury, there is a paucity of data pertaining to the contribution of the regulatory NLRs. This subgroup includes positive and negative regulatory NLRs that function through the direct regulation of inflammatory signaling pathways, such as NF-κB, AKT, and IFN signaling (10–12). Three NLRs have been shown to function as negative regulators, including NLRP12, NLRC3, and NLRX1. NLRX1 is of particular interest in the context of brain injury because it has been shown to play a role in a variety of cellular processes important to injury pathogenesis in a diverse range of cell types and tissues. For example, NLRX1 negatively regulates NF-κB signaling, type I IFN signaling, and reactive oxygen species (ROS) production, as well as acts as a positive regulator of autophagy in macrophages and fibroblasts (11, 13–15). NLRX1 has been best characterized in the context of pathogen recognition (13, 15, 16). However, recent studies have extended the function of NLRX1 beyond this initial role in modulating host–pathogen interactions and identified contributions to cancer, chronic obstructive pulmonary disease, inflammatory bowel disease, and the modulation of cell death (11, 17–19).

The purpose of this study was to investigate the role of NLRX1 in the pathogenesis of brain injury. We hypothesized that loss of NLRX1 would exacerbate NF-κB signaling and tissue damage following TBI. To test this, we used an Nlrx1^−/− mouse in a model of controlled cortical impact (CCI) and evaluated quantitative and qualitative measures of brain injury. Consistent with our hypothesis, mice lacking NLRX1 demonstrated increased pathophysiological features consistent with increased brain injury compared with wild-type control mice. Increased lesion volume in Nlrx1^−/− mice was associated with increased NF-κB signaling and influx of CD11b⁺ microglia and/or macrophage populations in the lesion site. These findings correlate with NLRX1-dependent effects identified in a relevant neuroblast cell line in which we performed cell death assays. Gene-expression analysis from the cortex of CCI-injured Nlrx1^−/− mice show increased changes in mRNA expression levels of genes involved in NF-κB signaling. Similar correlations were found in gene-expression data from human patients following brain injury associated with ruptured aneurysms. Collectively, these studies provide evidence of NLRX1’s role beyond host–pathogen interactions and further our knowledge regarding the underlying mechanisms involved in brain injury.

The generation and characterization of Nlrx1^−/− mice have been previously described (15). All mice were maintained on the C57BL/6J background. All animals used in experiments were male and between 2 and 4 mo of age. C57BL/6J and Nlrx1^−/− mice were maintained as separate colonies. All studies involving mice were repeated at least three independent times, with three to seven mice per genotype and treatment group. All experiments were conducted in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals and were conducted under the approval of the Virginia Tech Institutional Animal Care and Use Committee and the Virginia–Maryland College of Veterinary Medicine.

All mouse neuroblastoma (N2A) cell lines were generated as previously described (20). Briefly, we used OriGene TrueORF cDNA Clones with TurboFectin to generate Nlrx1 stable knock-in cells and OriGene short hairpin RNA (shRNA) plasmid against Nlrx1 to generate the knockdown cells. Cell death was evaluated by flow cytometry using annexin V/propidium iodine (PI) staining. H₂O₂ was used to stimulate neuron cell death. All experiments were performed four independent times, and each experiment included all of the experimental groups. We evaluated 10,000 events, excluding cellular debris and cellular clumps, based on side and forward scatter. The analysis was performed using automatic assignment of the set gates to all samples within the FlowJo workspace.

Mice were anesthetized using i.p. injection of ketamine and xylazine. Following the loss of consciousness, mice were positioned in a stereotaxic frame (21, 22). Body temperature was maintained at 37°C and monitored via rectal probe throughout the surgery. During surgery, a skin incision was made, followed by a 5-mm craniotomy using a portable drill over the right parietal-temporal cortex (−2.5 mm anterior/posterior and 2.0 mm lateral from bregma). Subsequently, CCI was induced by application of the eCCI-6.3 device (Custom Design & Fabrication; 4-mm impounder) at a velocity of 3.5 m/s, depth of 0.5 mm, and 150 ms impact duration (21, 22). Animals designated as sham controls were placed under anesthesia, as described above, and following loss of consciousness, received a skin incision and closure only. Skin incisions were closed using Vetbond Tissue Adhesive (3M). Following surgery, all animals were placed in a heated cage to maintain body temperature for 1 h. At 1, 3, or 14 d post–CCI injury, mice were euthanized, and brain tissue was removed following decapitation. Fresh frozen tissue was embedded in OCT and coronally sectioned at 30 μm thickness. Serial sections were taken 300 μm apart and stained for Nissl substance (22). Rotarod behavior assessment was performed as previously described (22, 23).

Contusion volume was assessed by a blinded investigator using a Cavalieri Estimator from Stereo Investigator (MBF Bioscience) and an Olympus BX51TRF motorized microscope (Olympus America). Contusion volume (cubic millimeters) was determined as previously described (22). Briefly, volume analysis was performed by estimating the area of tissue loss in the ipsilateral cortical hemisphere for five coronal serial sections at or around the epicenter (−1.1 to −2.6 mm posterior from bregma) of injury. Nissl-stained serial sections were viewed under brightfield illumination (original magnification ×4). A random sampling scheme was used that estimates every tenth section from rostral to caudal, yielding five total sections to be analyzed. A randomly placed grid, with points spaced 100 μm apart, was placed over the ipsilateral hemisphere, and the area of contusion was marked within each grid. Contusion boundaries were identified by loss of Nissl staining, pyknotic neurons, and tissue hemorrhage. The contoured area, using grid spacing, was used to estimate total tissue volume based on section thickness, section interval, and total number of sections within the Cavalieri program. Data are presented as volume of tissue loss or contusion volume (cubic millimeters) for wild-type and Nlrx1^−/− mice. A Stereo Investigator optical fractionator was used on serial coronal brain sections to estimate the total number of CD11b⁺ cells within 1500 μm (−0.2 to −2.5 anterior/posterior) of injured cortical tissue, as previously described (22, 24).

Total RNA was collected from brain specimens following mechanical homogenization, lysis, and extraction using TRIzol and the manufacturers’ protocols. The purified RNA was quantified, and 1 μg was pooled from three randomly chosen brains prior to the cDNA reaction. Expression profiles were assessed using the RT2 Profiler PCR Array Platform PAMM-025Z (QIAGEN), following the manufacturer’s protocols. Ingenuity Pathways Analysis (IPA) software was used to assess the array data. In addition to the profiling studies, RNA samples (5 μg) were archived using a cDNA Archive Kit (ABI) and evaluated via RT-PCR using specifically targeted commercially available primer/probe sets (ABI).

Human NLRX1 expression was evaluated using a publicly accessible microarray meta-analysis search engine (http://www.nextbio.com/b/search/ba.nb), as previously described (25). Gene-expression and pathway analyses in human subjects and rodents were conducted using the following array data series (available through the National Center for Biotechnology Information: https://www.ncbi.nlm.nih.gov/): GSE11686, GSE3307, GSE58294, GSE36233, GSE1767, GSE21079, GSE66573, GSE43591, and GSE20141.

Data were analyzed using GraphPad Prism, version 6 (GraphPad, San Diego, CA). A Student two-tailed t test was used for comparison of two experimental groups. Multiple comparisons were done using one-way and two-way ANOVA where appropriate, followed by the Tukey posttest for multiple pairwise examinations. Correlation was also computed using GraphPad Prism. Changes were identified as statistically significant at p <0.05. Mean values are reported with SEM.

NLRX1 functions as a negative regulator of inflammation and modulates a variety of pathways associated with brain injury (10, 12, 26). Thus, we hypothesized that Nlrx1^−/− mice would demonstrate significantly increased tissue damage following TBI. To evaluate this hypothesis, we used previously characterized Nlrx1^−/− mice in a CCI model (21, 22). Wild-type and Nlrx1^−/− animals were subjected to a moderate CCI using a 4-mm impounder at a velocity of 3.5 m/s, depth of 0.5 mm, and 150 ms impact duration, as previously described (6, 21, 22). TBI progression was evaluated 3 d postinjury, which was empirically determined to be the peak time point for acute neural tissue damage and pathophysiology in prior studies using wild-type C57BL/6J animals in our CCI model (6, 21, 22). Upon necropsy, Nlrx1^−/− mice were observed to have significantly larger gross lesions and hemorrhage compared with the wild-type animals (data not shown). Serial sections were generated from the whole brain and subjected to Nissl staining. Contusion boundaries were identified based on the loss of Nissl stain, pyknotic neurons, and tissue hemorrhage. Consistent with the gross lesion observations, Nlrx1^−/− mice were found to have more expansive contusion boundaries, which radiated out significantly further from the impact epicenter compared with the injuries observed in the wild-type animals (Fig. 1A–F). Using the Cavalieri Estimator, we found significant differences in contusion volume between Nlrx1^−/− mice (10.28 ± 1.15 mm³) and wild-type animals (4.55 ± 0.47 mm³) post–CCI injury (Fig. 1G). These data support a role for NLRX1 in limiting neural tissue loss and TBI pathogenesis in the cortex following CCI injury.

We next sought to determine whether the increased damage following TBI in Nlrx1^−/− mice resulted in any broad behavioral effects. We used Rotarod behavioral analysis to determine whether motor deficit and recovery were affected following CCI injury. Mice were trained on the Rotarod 4 d prior to CCI injury, and the animals were subjected to motor assessments at 3, 7, and 14 d post–CCI injury or post–sham operation. The time to fall was determined and then normalized to the average baseline time for each animal. No significant differences were observed between Nlrx1^−/− and wild-type sham-operated mice (Nlrx1^−/− 110.70 ± 11.42%; wild-type 115.37 ± 5.24%) (data not shown). However, following CCI injury, Nlrx1^−/− mice performed significantly worse on the Rotarod (Fig. 1H). Wild-type and Nlrx1^−/− mice demonstrated motor deficits following injury. However, at 3 d post–CCI injury, Nlrx1^−/− mice had a significant increase in motor deficits on the Rotarod compared with wild-type animals (Nlrx1^−/−, 32.76 ± 1.0%; wild-type, 59.17 ± 5.45%) (Fig. 1H). This trend continued through day 7 (Nlrx1^−/−, 51.36 ± 7.6%; wild-type, 79.35 ± 7.8%) and day 14 (Nlrx1^−/−, 55.8 ± 16.8%; wild-type, 81.73 ± 5.21%) (Fig. 1H). Together, these data indicate that Nlrx1^−/− mice experience increased motor impairments following CCI injury, and recovery is attenuated compared with wild-type animals. The Rotarod data correlate with the contusion volume estimates and confirm that loss of NLRX1 significantly impacts neural tissue damage and subsequent motor function recovery following CCI injury.

NLRX1 is broadly expressed in the brain and in a variety of cells known to influence TBI pathogenesis, including neurons, microglia, and astrocytes (Supplemental Fig. 1). To better characterize the increased CCI injury observed in Nlrx1^−/− mice, we next sought to determine the effects of NLRX1 loss on microglia and astrocyte populations in the lesion following TBI. Frozen sections were prepared for immunohistochemistry (IHC), and specimens were evaluated for CD11b and glial fibrillary acid protein (GFAP) expression at 3 d post–CCI injury (Fig. 2A–D). In the intact brain, the microglia should be the only CD11b⁺ cells. However, following injury, disruption to the blood–brain barrier can allow an increased influx of macrophages and neutrophils into the lesion that will also contribute to the CD11b⁺ cell population. Increased CD11b⁺ cells were observed in all of the mice following CCI injury compared with the sham-operated animals. However, an increase in CD11b⁺ cells was observed in Nlrx1^−/− mice compared with wild-type animals following CCI injury (Fig. 2A, 2B). In addition, astrocytes can also significantly impact TBI pathogenesis (reviewed in Ref. 27). We used the prototypical astrocyte marker GFAP to evaluate astrogliosis in the peri-lesion cortex after CCI injury. No significant differences were observed in density or distribution of GFAP⁺ cells between Nlrx1^−/− mice and wild-type animals following CCI injury (Fig. 2C, 2D). Together, these data suggest that NLRX1 functions directly or indirectly to modulate microglia proliferation and/or macrophage influx, with no effects on astrocyte populations following CCI injury.

Prior studies evaluating NLRX1 in cancer have found a role for NLRX1 in the regulation of macrophage proliferation (11). Specifically, macrophages harvested from Nlrx1^−/− mice demonstrated a significant increase in proliferation under specific physiological conditions (11). Following TBI, microglia in the lesion area undergo proliferation in response to injury and play vital roles in the modulation of secondary injury and recovery (27, 28). Although microglia are essential for proper subacute resolution of damage in the CNS, acutely these cells can also negatively impact pathogenesis by propagating inflammation to tissues adjacent to the site of injury and instigating significant collateral damage if dysregulated (29). Thus, we next sought to determine whether the increase in CD11b⁺ cells was associated with increased microglia proliferation or macrophage influx following CCI injury. Frozen sections were prepared for IHC, and specimens were evaluated for CD11b and Ki67 reactivity 3 d post–CCI injury. Interestingly, we observed a significant increase in CD11b⁺ and Ki67⁺ cells in the injured cortex of Nlrx1^−/− and wild-type mice (Fig. 2E–L). However, we observed a greater density of these cells in Nlrx1^−/− mice compared with wild-type animals (Fig. 2E–L). Indeed, in wild-type mice, very few double-positive cells were observed (Fig. 2E–H). Quantitative assessment of CD11b⁺ cells using an optical fractionator showed an increase in the number of CD11b⁺ cells present in the injured cortex of Nlrx1^−/− mice (22,133 ± 3,458 cells) compared with wild-type mice (7,966 ± 1,270 cells) at 3 d post–CCI injury (Fig. 2M). Consistent with the increased CD11b immunoreactivity, these Ki67 findings suggest that NLRX1 regulates microglia proliferation following TBI. It is also possible that infiltrating macrophages contribute to this cell population. However, in either case, these findings are consistent with increased TBI pathogenesis and could contribute directly or indirectly to enhanced injury progression and motor deficits.

NLRX1 has been previously shown to negatively regulate NF-κB signaling and this attenuation is correlated with increased proliferation and tumorigenesis in cancer models (11, 30). Thus, we next profiled gene expression to evaluate NF-κB signaling following TBI in the Nlrx1^−/− and wild-type brain following CCI injury. Fresh 4 × 4 mm cortical tissue was dissected from the ipsilateral and contralateral hemispheres of each animal at 3 d postinjury. Total RNA was extracted from three randomly chosen wild-type or Nlrx1^−/− brains and the RNA from each genotype was pooled in equal concentrations for cDNA synthesis (Supplemental Fig. 2). Changes in gene expression were evaluated using a panel of superarrays (QIAGEN) chosen to evaluate pathways associated with inflammatory signaling, cell death, and proliferation, as previously described (11, 15). Gene expression was determined using the ΔΔ cycle threshold method to calculate fold change, as described by the manufacturer. A panel of eight housekeeping genes was used to normalize the expression of each gene on the array, and the fold change in gene expression between the respective CCI-injured and uninjured Nlrx1^−/− and wild-type brains was determined (Supplemental Fig. 3). Gene-expression data were further analyzed using IPA to better define and visualize the signaling pathways influenced by the loss of NLRX1 (Fig. 3).

The gene-expression studies and pathway analysis revealed a significant increase in NF-κB signaling in the cortical brain tissue from Nlrx1^−/− and wild-type mice following injury. However, canonical NF-κB signaling was significantly upregulated in Nlrx1^−/− lesions compared with those in wild-type animals (Fig. 3, Supplemental Fig. 3). This evaluation revealed that 47 genes associated with inflammation and NF-κB signaling were significantly upregulated, defined as ≥2-fold increase in gene expression over wild-type (Supplemental Fig. 3). The genes with the greatest increase in expression in Nlrx1^−/− brains included Card10, Relb, and Il10, all of which were upregulated >7-fold compared with wild-type. Conversely, the following three genes were significantly downregulated in Nlrx1^−/− brains: Slc20a1, Tnf, and Ccl2 (Supplemental Fig. 3). IPA of the global changes in gene expression revealed a significant increase in NF-κB signaling in Nlrx1^−/− lesions (Fig. 3). This increase was observed in genes categorized across each functional group in the NF-κB signaling cascade, including ligands and receptors, downstream signaling mediators, kinases, cytoplasmic sequestering and release factors, and four of the five NF-κB transcription factors (Fig. 3). Significant differences were also observed in NF-κB–responsive genes associated with the immune response. Although, generally, these NF-κB–responsive genes were significantly upregulated, notable exceptions included Tnf and Ccl2, which were significantly downregulated following CCI injury in Nlrx1^−/− mice (Fig. 3, Supplemental Fig. 3). Finally, several genes in the NF-κB signaling cascade and associated with apoptosis were also found to be significantly upregulated in lesions from Nlrx1^−/− brains compared with wild-type lesions (Fig. 3, Supplemental Fig. 3).

NLRX1 is broadly expressed in the brain, and NF-κB is a master regulator of gene transcription in cells throughout the CNS (Supplemental Fig. 1) (31, 32). Thus, we next sought to identify the cell types associated with the increased NF-κB signaling in the CCI-injured cortical tissue of Nlrx1^−/− mice. To identify these cells, we used IHC targeting p-p65, which is a direct assessment of the activation of the canonical NF-κB signaling cascade. Likewise, p65 was found to be significantly upregulated in the gene-expression and pathway analysis studies in Nlrx1^−/− mice (Fig. 3). IHC analysis revealed increased immunoreactivity for p-p65 in the cortical lesions following CCI injury in Nlrx1^−/− and wild-type animals (Fig. 4). In both genotypes, the p-p65 staining was localized to CD11b⁺ cells (Fig. 4A, 4D). Indeed, despite their lower numbers compared with Nlrx1^−/− mice, the overwhelming majority of p-p65⁺ cells in wild-type animals were also double positive for CD11b (Fig. 4A–C). Consistent with the gene-expression findings, we observed significantly increased numbers of CD11b⁺ p-p65⁺ double-positive cells in Nlrx1^−/− mice compared with wild-type animals (Fig. 4D–F). These data suggest that the CD11b⁺ cell population is a major source of NF-κB signaling, which may be suppressed by NLRX1 in the cortex following CCI injury. Together, these findings are consistent with NLRX1’s role as a negative regulator of canonical NF-κB signaling observed in other diseases and model systems and may underlie the increased proliferation, recruitment, and function of these cells in the brain following trauma (11, 15).

In addition to regulating NF-κB signaling and CD11b⁺ cell proliferation and recruitment, NLRX1 has been shown to modulate cell death in neurons (20). Neuronal death is a critical component of the lesion outcome observed in the CCI injury model, which was exacerbated in Nlrx1^−/− mice. Likewise, we observed significantly increased expression of several genes associated with cell death and apoptosis in Nlrx1^−/− lesions (i.e., Fadd, Fasl, and Agt) compared with those in wild-type animals (Fig. 3, Supplemental Fig. 3). To evaluate the contribution of NLRX1 to neuronal cell death, we next used an in vitro model based on mouse N2A cells (20). N2A cells are a murine neuroblastoma cell line that was shown in previous studies to derive multiple neuronal cell types (33, 34). NLRX1 was either knocked down or overexpressed, and cell death was evaluated following H₂O₂ treatment. ROS has been suggested to be an important mediator of acute brain injury following TBI. In this study, we demonstrate that treatment of N2A cells with 100 mM H₂O₂ for 18 h resulted in a significant decrease in cell viability in all cell lines with the exception of N2A^Nlrx1 cells, which overexpress Nlrx1 (Fig. 5). Increased NLRX1 in N2A cells led to a significant reduction in H₂O₂-induced cell death compared with control cells or cells with shRNA knockdown of Nlrx1 (N2A^sh), which conversely display enhanced cytotoxicity (Fig. 5B, 5C). Annexin V/PI staining allows for the quantification of cells in the early stages of apoptosis, when phosphatidylserine is flipped and faces outside of the cell, increasing its availability for annexin V binding. To exclude necrotic cells and cells in the late stages of apoptosis, we used PI stain, which penetrates all dead cells. Using this technique, we observed a significant increase in apoptosis after treatment with H₂O₂ in all cell lines with the exception of N2A^Nlrx1 (Fig. 5A). Our data indicate that apoptosis is significantly reduced in N2A^Nlrx1 cells and significantly increased in N2A^sh cells (Fig. 5A). Thus, we can conclude that NLRX1 plays a key antiapoptotic role in neuroblasts under oxidative stress like that occurring acutely following TBI, and our data suggest that NLRX1 may function to attenuate neuron loss in vivo. However, it should be noted that future in vivo and ex vivo studies using primary neurons are needed to provide additional insights necessary to confirm this hypothesis.

Our findings that NF-κB signaling is significantly upregulated in Nlrx1^−/− mice following TBI are consistent with prior studies linking NLRX1 as a negative regulator of inflammation with its protective role in attenuating neuronal cell death (20). In human patients, increased NF-κB signaling has been shown to promote neurodegenerative mechanisms during Parkinson’s disease and Alzheimer’s disease (35–39). However, there have been few human studies directly evaluating NF-κB signaling following trauma and/or TBI. Likewise, there is a paucity of data associated with noninflammasome-forming NLRs, such as NLRX1, in human patients following brain injury. To address these shortcomings and evaluate the relevance of our findings in human patients, we conducted a retrospective analysis of gene-expression data archived as National Institutes of Health Gene Expression Omnibus datasets from human patients following ruptured brain aneurysms (GSE26969 and GSE54083) (25). Ruptured brain aneurysms are considered nontraumatic brain injuries that have several clinical features in common with traumatic contusion injury and our CCI injury model, including massive hemorrhage, infiltration of peripheral-derived immune cells, neuronal and glial apoptosis, reduced cerebral blood flow, and blood–brain barrier permeability (40–42). Also, there are several well-characterized and controlled human brain aneurysm data sets publicly available to evaluate our hypotheses. We mined data from several independent studies evaluating gene-expression changes in specimens collected from the aneurysmal dome following ruptured, superficial, or unruptured intracranial aneurysms in up to 10 human subjects per condition (43, 44). Each study evaluated gene expression on >41,000 transcripts from 3 to 10 patients in each group. Expression data were normalized to GAPDH, and the fold change in expression was compared against the superficial aneurysm data.

As mentioned above, several studies have evaluated the contribution of inflammasome-forming NLRs in TBI in human and rodent models (6–9, 45). However, data pertaining to noninflammasome-forming regulatory NLR family members have yet to be evaluated. This group of NLRs includes NLRX1, which has functional characteristics similar to two additional family members: NLRP12 and NLRC3. All three of these NLRs have been suggested to exert negative-regulatory pressure on signaling pathways associated with the activation of other families of PRRs, such as TLRs and Rig-I–like helicase receptors (12, 15, 46, 47). We evaluated the expression of these three NLRs in the human aneurysm patient subsets and found that NLRX1 (0.33 ± 0.16 aneurysm versus 1.00 ± 0.03 superficial) and NLRP12 (0.26 ± 0.13 aneurysm versus 1.01 ± 0.05 superficial) were significantly downregulated in these human patient populations compared with NLRC3 (1.46 ± 0.90 aneurysm versus 0.72 ± 0.53 superficial) (Fig. 6A). In addition to these three negative-regulatory NLRs, we evaluated expression patterns for the remaining 19 NLR family members (Supplemental Fig. 4). We found significant dysregulation among the majority of NLRs evaluated. Among the pyrin family NLRs, we found that NLRP1, NLRP2, NLRP3, NLRP4, and NLRP13 were significantly downregulated following aneurysm, whereas NLRP6, NLRP7, NLRP10, and NLRP11 were significantly upregulated in these human patients (Supplemental Fig. 4A). Likewise, among the card family NLRs, NOD1 and NLRC5 were significantly upregulated, whereas NLRC4 was significantly downregulated (Supplemental Fig. 4B). Finally, assessment of CIITA and NAIP revealed that both of these NLRs were significantly downregulated following aneurysm (Supplemental Fig. 4C). Together, these data show broad dysregulation among the NLR family members in the context of brain injury in human subjects. However, consistent with the findings from Nlrx1^−/− mice in the CCI injury model, it appears that loss or attenuation of NLRX1 is associated with increased brain injury and pathogenesis in both human and rodent models.

In addition to the NLR genes, we analyzed the expression of the genes associated with inflammation and NF-κB signaling in the mouse CCI injury model (shown in Supplemental Fig. 3) in the human aneurysm datasets. Similar to the mouse findings, the NF-κB signaling cascade was significantly upregulated in the human subjects. Indeed, there was a positive correlation between increased NF-κB gene expression and aneurysm severity (superficial versus unruptured: r = +0.405, p = 0.002; superficial versus ruptured: r = +0.439, p = 0.0008). The 10 genes with the largest differences in expression between the two aneurysm groups are shown in Fig. 6B. Similar to the mouse TBI model, the majority of genes associated with NF-κB signaling were either significantly upregulated or unchanged in the human subjects (Fig. 6B, data not shown). However, we did observe a few notable differences in the human subject data. Specifically, in human subjects, CCL2 expression is significantly upregulated following ruptured aneurysm (2.80-fold increase), even though NLRX1 is significantly downregulated (Fig. 6A, 6B). Conversely, in the mouse, when NLRX1 is ablated, Ccl2 was significantly downregulated following CCI (8.61-fold decrease) (Fig. 3). In total, seven genes were found to be significantly downregulated in the human subjects. These included TNF expression, which was significantly downregulated in Nlrx1^−/− mice following CCI and in the human aneurysm subjects (Figs. 3, 6C). Conversely, expression of IL-10, EGFR, TNFSF14, CD40, and TLR2 was significantly downregulated in the human subjects but was significantly upregulated in the mouse CCI model in the absence of NLRX1 (Figs. 3, 6C). However, despite these relatively few discrepancies, the sum of these data reveal that NF-κB signaling is significantly upregulated in mouse and human and identifies a diverse range of mediators associated with this signaling cascade that impact the pathogenesis of brain injury.

In this article, we provide evidence that NLRX1 negatively regulates NF-κB signaling, apoptosis and infiltration of CD11b⁺ cells following TBI. Genetic ablation of Nlrx1 in mice results in significantly increased brain pathophysiology following CCI. Mechanistically, our findings suggest that NLRX1 attenuates brain injury through the regulation of NF-κB signaling, the recruitment and expansion of CD11b⁺ cell populations in the lesion area, and potentially by limiting apoptosis in neuronal cell populations. In human patients, the downregulation of NLRX1 expression and concomitant upregulation of NF-κB signaling are correlated with injury severity following aneurysm. Thus, the findings from the mouse and human models appear consistent and further support the importance of NLRX1 and the pathways regulated by this unique protein in the brain following injury.

There is a growing appreciation that NLR family members have diverse functions in diseases beyond those directly associated with pathogens. Indeed, studies using human clinical samples and/or mouse models have shown a role for these unique proteins in the modulation of neuroinflammation (48, 49). The vast majority of studies have focused on inflammasome-forming NLRs in mice and their contribution to the pathophysiology of neurodegenerative diseases, including Alzheimer’s disease, Parkinson’s disease, and multiple sclerosis (49–51). Beyond neurodegenerative diseases, inflammasome-forming NLRs have also been found to significantly modulate various pathologic features of traumatic and nontraumatic brain injury. Of the inflammasome-forming NLRs in brain injury, the NLRP1 and NLRP3 inflammasomes are the best characterized in mice, and the majority of studies have focused on mechanisms associated with regulating IL-1β/IL-18 and pyroptosis (6–8, 45, 50). However, aside from the inflammasome, there is a paucity of data pertaining to the role of noninflammasome-forming regulatory NLRs in brain injury. The regulatory NLR subgroups can be divided into positive regulators that promote inflammation (i.e., NOD1/NLRC1 and NOD2/NLRC2) and negative regulators (i.e., NLRP12, NLRC3, and NLRX1) (52). NOD2 is the only regulatory NLR that has been evaluated in the context of brain injury (53–55). In models of ischemia-reperfusion injury, NOD2 was found to be significantly upregulated in microglia and astrocyte populations following focal cerebral ischemia-reperfusion injury; pretreatment with NOD2 agonists significantly increased infarct volume and neurologic dysfunction in wild-type mice and rats (53–55). Furthermore, genetic ablation of Nod2 significantly improved stroke outcomes and attenuated neuroinflammation (53). Nod2^−/− mice exhibited significantly reduced levels of IL-1β, IL-6, and TNF associated with attenuated NF-κB, MAPK, and JNK signaling (53). All of these findings are consistent with the role of NOD2 as a positive regulator of inflammation signaling.

Because of these early studies associated with inflammasome-forming NLRs and NOD2 modulation of ischemia–reperfusion injury and stroke pathophysiology, we sought to expand upon the clinical relevance of these findings and evaluate the expression patterns of all 22 NLR family members in human patients (Fig. 6, Supplemental Fig. 4). We took advantage of the robust publicly available gene-expression data sets for ruptured, unruptured, or superficial intracranial aneurysms in human subjects and used bioinformatics-based approaches to mine these data (43, 44). We chose aneurysm because of our joint interests in traumatic and nontraumatic (ruptured aneurysm or therapeutic anticoagulation) progressive hemorrhagic brain injury. Aneurysms have similar pathophysiological features that are relevant to TBI and the respective animal models, including the CCI injury model used in the current study and previously to evaluate NLR function in mice (6, 9). As in severe cases of traumatic intracranial hemorrhage following TBI, intercerebral hemorrhage in patients with ruptured aneurysms results in increased cranial pressure, hypoxia/ischemia, necrosis and infiltration of the immune-derived compartment, blood–brain barrier disruption, and oxidative stress. These common pathological features are also well established in animal models of brain contusion injury (oxidative stress, brain edema, blood–brain barrier permeability, and autonomic dysfunction from TBI). Our data demonstrate that a large number of NLRs, 15 of the 22 identified in humans, are significantly dysregulated in human aneurysm specimens, with the majority (9 of 15) upregulated with increasing aneurysm severity (Fig. 6, Supplemental Fig. 4). Interestingly, we did not observe any significant differences in NOD2 expression in the aneurysm specimens (Supplemental Fig. 4B). This is counter to our anticipated results based on the previous preclinical mouse model data for Nod2^−/− animals (53–55). However, we did observe a significant increase in NOD1 gene expression in the aneurysm specimens (Supplemental Fig. 4B). NOD1 and NOD2 are functionally related and modulate similar biochemical pathways. Thus, it is possible that NOD1 may have a more prevalent role in human brain injury, and future studies using Nod1^−/− mice may shed insights into this intriguing observation.

Although NOD1 and NOD2 are regulatory NLRs that function to promote inflammation, NLRP12, NLRC3, and NLRX1 are regulatory NLRs that negatively regulate inflammatory signaling cascades activated by other classes of PRRs. Interestingly, the gene-expression data analysis from the human aneurysm studies revealed that NLRX1 and NLRP12 are significantly downregulated in ruptured aneurysms compared with the superficial aneurysms, whereas we did not observe a significant difference in NLRC3 expression (Fig. 6A). NLRX1 and NLRC3 negatively regulate canonical NF-κB signaling, whereas NLRP12 attenuates canonical and noncanonical NF-κB signaling pathways (15, 46, 47, 56). Thus, consistent with the findings that NLRX1 and NLRP12 were downregulated, our expanded analysis of genes associated with inflammation revealed that a large number of genes associated with NF-κB signaling were significantly upregulated in the human aneurysm specimens, and this upregulation was correlated with injury severity (Fig. 6B). The role of NF-κB signaling in ruptured aneurysm and TBI is quite complex. In the CNS, NF-κB signaling regulates inflammation and apoptotic cell death following nerve injury and damage. This pathway has also been found to contribute to infarction and cell death in various stroke models and patients (57, 58). However, NF-κB signaling is also a critical component for neuronal survival and the attenuation of neurodegeneration, and activation of this cascade also facilitates recovery postinjury (59). Ultimately, the impact of increased NF-κB signaling in brain injury likely depends on which NF-κB factors are activated, where the injury occurs, and what cell types are involved. Likewise, it is reasonable to speculate that the temporal dynamics of pathway activation is also a significant factor. The current data demonstrate that NF-κB signaling is a common feature that is dysregulated following ruptured aneurysm in humans and brain contusion injury in the rodent, suggesting a potential common target based on known similarities in the pathological progression of these disorders.

NLRX1 has been predominantly studied in the context of inflammation associated with virus or bacteria exposure (52, 60, 61). However, recent studies have emerged that reveal a more dynamic role for NLRX1 in modulating pathological conditions that extend beyond specific host–pathogen interactions (11, 18, 20). There have been very few studies evaluating NLRX1 function in the CNS. To our knowledge, the only prior in vivo study of NLRX1 in the CNS characterized its role in modulating experimental autoimmune encephalomyelitis (EAE) (62). EAE is a common autoimmune mouse model of multiple sclerosis and is associated with the loss of immunological tolerance to CNS-derived Ags. NLRX1 was found to play a robust protective role in the EAE model (62). Nlrx1^−/− mice were shown to have significantly increased clinical parameters associated with disease progression and corresponding increases in tissue damage during EAE (62). Nlrx1^−/− animals were also more susceptible to myelin-reactive T cells following adoptive transfer (62). The mechanism associated with NLRX1 protection in the EAE model was further associated with the attenuation of macrophage and microglia activation (62). Indeed, microglia and macrophages cultured from Nlrx1^−/− mice have been previously shown to generate excessive amounts of various proinflammatory cytokines, including IL-6 and CCL2, following pathogen-associated molecular pattern stimulation (10, 11, 15, 62). This excessive activation of the microglia and macrophages in Nlrx1^−/− mice resulted in chronic inflammatory signaling and, ultimately, enhanced neurodegeneration.

In general, the findings from prior EAE studies are consistent with our findings for TBI. Similar to EAE, Nlrx1^−/− mice have significantly enhanced injury progression, tissue damage, and behavioral abnormalities consistent with increased brain lesion volume (Fig. 1). Also consistent with the prior EAE findings, we observed increased CD11b⁺ cells in the lesion that is consistent with increased microglia proliferation and/or macrophage influx following damage (Fig. 2). These findings are consistent with the increased cell recruitment observed in the EAE studies, as well as prior observations that CD11b⁺ cells from Nlrx1^−/− mice are hyperproliferative under controlled ex vivo conditions (11). CD11b⁺ proliferative cells appear to be a major contributing source of the increased NF-κB signaling at the 3-d time point following TBI (Fig. 4). The significant increase in gene expression associated with NF-κB signaling in the TBI mouse model corresponds with the findings from the human ruptured aneurysm specimens and the prior data associated with EAE, with a few notable exceptions. For example, in the mouse EAE model, IL-6 and CCL2 expression was significantly upregulated and correlated with disease severity in Nlrx1^−/− mice (62). However, 3 d following TBI, CCL2 is significantly downregulated in Nlrx1^−/− animals, and expression of Il-6 was not significantly changed in NLRX1-deficient animals compared with wild-type (Fig. 3, data not shown). Similarly, in the EAE model, no differences were observed in TNF generation, whereas TNF was significantly downregulated in the TBI model (Fig. 3). These few discrepancies are likely associated with the fundamental differences between the two models. Specifically, the EAE model is considered a model of “sterile inflammation” and is inherently driven by immune system hyperactivation (62), whereas in the TBI model, the resultant inflammation is driven purely by acute mechanical damage recognition in the CNS and the resultant injury response and resolution. Despite the few discrepancies in specific cytokine levels, the overall consistency between the phenotypes and proposed mechanisms support the conclusion that NLRX1 limits inflammation in the CNS, and its downregulation following CCI injury may contribute to cellular apoptosis and tissue loss. Furthermore, the two studies suggest that the attenuation in pathology is associated through the modulation of inflammatory signaling cascades in CD11b⁺ cell populations.

Beyond inflammatory signaling cascades and CD11b⁺ cells, it is critical to consider the impact of NLRX1 on neuronal cell death following TBI. In addition to dysregulated inflammation, neuronal cell death is a hallmark feature associated with the severity of injury. As our data indicate, we observed significantly increased cortical tissue loss in Nlrx1^−/− mice compared with wild-type animals following TBI (Fig. 1). The increase in apoptotic gene expression in Nlrx1^−/− mice partially correlates with increased NF-κB signaling in these animals (Fig. 3). A variety of genes associated with cell death and apoptosis were significantly upregulated in our brain-profiling studies (Fig. 3, Supplemental Fig. 3). These observations are consistent with our previous work evaluating NLRX1 in N2A cell death using an in vitro model system (20). In these prior studies, NLRX1 was knocked down or overexpressed in N2A cell lines, and mechanisms associated with cell death were evaluated. Using this system, NLRX1 was found to modulate the balance between necrosis and apoptosis in N2A cells following rotenone treatment (20). Attenuation of NLRX1 was found to promote apoptosis, rather than necrosis, through the enhancement of DRP1 phosphorylation and regulation of mitochondrial fission (20). Although these data support our in vivo findings, we sought to expand upon these prior studies using an in vitro model with greater relevance to TBI. Thus, we used the previously described N2A cells and exposed them to H₂O₂. Similar to the previous rotenone findings, overexpression of NLRX1 attenuated cell death, whereas knockdown of NLRX1 enhanced cell death following H₂O₂ exposure (Fig. 5). These findings support the previous observations with rotenone and provide additional mechanistic insight into the in vivo loss of neurons observed in Nlrx1^−/− mice following TBI. Several prior studies have evaluated the role of NLRX1 in cell death and proliferation in monocytes, fibroblasts, and epithelial cells (11, 19, 30, 63). Likewise, NLRX1 has also been shown to modulate ROS signaling and autophagy in these diverse cells following activation via pathogen-associated molecular pattern stimulation and/or pathogen exposure (14, 16, 30, 64, 65). Together, our findings and the results of these prior studies reveal a relationship among NLRX1, mitochondria, and cell death. However, the exact mechanism of NLRX1 in these processes remains elusive. Our findings associated with NLRX1 in N2A cells, and likely neurons, are consistent with many of these prior studies and provide additional mechanistic insights associated with the enhanced injury progression observed in Nlrx1^−/− mice. Because this is an area of intense research focus, future studies will likely reveal additional mechanistic insights associated with NLRX1 in neuronal cell death.

NLRX1 is a unique member of the NLR family and has diverse regulatory functions that are associated with a myriad of human diseases beyond pathogen infections. The contribution of NLRX1 and other noninflammasome-forming regulatory NLRs in the modulation of brain pathophysiology is an emerging area of research interest with tremendous potential. In this article, we have shown that NLRX1 protects against brain injury in mouse models and human subjects, in part through the negative regulation of NF-κB signaling. Our findings are consistent with the limited other studies conducted thus far with NLRX1 in the CNS. We anticipate that future studies will better define the role of NLRX1 in other human neurologic disorders. Likewise, our gene-expression and profiling studies have uncovered a broad range of genes and pathways that have not previously been associated with brain injury in humans or mice. Thus, studies of these genes should reveal novel targets and pathways for future therapeutic strategies to treat brain injury.

We thank Dr. Jenny P.Y. Ting (University of North Carolina Chapel Hill) for kindly providing the Nlrx1^−/− mice used in this study. We also would like to acknowledge Dr. Andrea Bertke and Angela Ives (Virginia–Maryland College of Veterinary Medicine) for technical assistance and support.

The authors have no financial conflicts of interest.