Targeting SARM1 as a novel neuroprotective therapy in neurotropic viral infections

Abstract

Viral encephalitis, resulting from neurotropic viral infections, leads to severe neurological impairment, inflammation, and exhibits high mortality rates with poor prognosis. Currently, there is a lack of effective targeted treatments for this disease, which poses a significant public health concern. SARM1 has been identified as the pivotal mediator of axonal degeneration and inflammation across various neuropathies, activated by an elevation in the NMN/NAD⁺ ratio. However, comprehensive in vivo investigations into the role of SARM1-mediated pathogenesis in viral encephalitis are still lacking. In this study, we established mouse models of viral encephalitis using Japanese encephalitis virus (JEV), herpes simplex virus-1 (HSV-1), and rabies virus (RABV) as representative pathogens. Our findings demonstrate that neurotropic virus infections elicit robust axonal degeneration, mitochondrial dysfunction, and profound neuropathological damage in cortical neurons via the activation of SARM1. In mouse models of viral encephalitis, deletion or inhibition of SARM1 effectively preserved axonal morphology and maintained mitochondrial homeostasis, while also attenuating the infiltration of CD45⁺ leukocytes in the cortex. Consequently, these interventions ameliorated neuropathological damage and enhanced survival outcomes in mice. Our findings suggest that SARM1-mediated axonal degeneration and brain inflammation exacerbate the pathological progression of viral encephalitis. Therapies targeting SARM1 emerge as viable and promising strategies for protecting neuronal function in the context of neurotropic viral infections.

Introduction

Viral encephalitis refers to an acute infection of the central nervous system caused by neurotropic viruses [1]. It has recently drawn more attention because of its high mortality and disability rates, along with its heavy economic burden [2]. The disease manifests as a syndrome characterized by neurological impairment leading to altered mental status and seizures, accompanied by acute fever and cerebrospinal fluid leukocytosis due to brain parenchymal inflammation [1,2,3,4,5]. While diagnostic methods and viral identification techniques have advanced significantly over the past two decades [6, 7], treatment options remain largely empirical and are established only for select viral infections, such as acyclovir for herpes simplex encephalitis [2, 3, 8]. This limited therapeutic arsenal, coupled with our incomplete understanding of the mechanisms underlying neurological impairment and inflammation in viral encephalitis, raises questions about whether a common pathological pathway exists that could be targeted for treatment.

Pathological axonal degeneration has emerged as a critical feature of neurological impairment across various neurological disorders, including traumatic injury, Alzheimer’s disease, Parkinson’s disease, multiple sclerosis, and amyotrophic lateral sclerosis [9,10,11,12]. This degenerative process, which typically precedes both clinical symptoms and neuronal cell death, involves the degeneration of axons or their branches. The resulting disruption of connections between neuronal cell bodies and their neural targets leads to neurological impairment and contributes to the progression of various neurodegenerative diseases [13,14,15].

SARM1 (sterile α and Toll/IL-1 receptor motif–containing 1) acts as a central executor in axonal degeneration, as its loss prevents axon degeneration in mouse models of traumatic injury and several neurological disorders [16,17,18,19,20]. This evolutionarily conserved, mitochondria-associated protein belongs to the toll/interleukin-1 receptor (TIR) domain family of NAD⁺ hydrolases [16]. Under physiological conditions, SARM1 exists as inhibitory octamers, with the N-terminal armadillo domain preventing TIR domain dimerization [21]. Acting as a metabolic sensor, SARM1 becomes activated when the nicotinamide mononucleotide (NMN)/NAD⁺ ratio increases. This activation triggers a cascade of events: hydrolysis of residual NAD⁺ into nicotinamide, ADPR, and cyclic ADPR (cADPR), followed by MAP kinase signaling activation, Ca²⁺ influx stimulation, and calpain activation, ultimately leading to axonal degeneration [16, 22,23,24,25]. Beyond its role in axonal degeneration, the TIR domain of SARM1 also participates in neuroinflammation mechanisms [16, 26]. It enhances innate immunity by regulating the transcriptional levels of cytokines and chemokines [19, 26]. These combined functions make SARM1 a critical mediator of neuronal cell death in response to various injuries, including oxygen and glucose deprivation, nerve transection, and vincristine treatment [27, 28].

Despite growing recognition of SARM1’s role in mediating axonal degeneration and inflammation in various neurological disorders, its contribution to viral encephalitis remains incompletely characterized [29, 30]. The role of SARM1 in viral neurological illnesses has been demonstrated by a number of research. It causes neuronal death during La Crosse virus (LACV) infection, restricts West Nile virus infection, and alters immune responses to vesicular stomatitis virus [18, 31, 32]. Recent in vitro studies have further demonstrated that both rabies virus (RABV) and Zika virus can trigger SARM1-mediated axonal degeneration, with Sarm1 gene knockout significantly delaying this process in cultured neurons [33, 34]. These findings led us to hypothesize that SARM1 activation occurs during viral encephalitis in vivo, contributing to axonal degeneration and subsequent neuropathology. Understanding SARM1’s role in this context, it would unveil novel therapeutic approaches to intervention. Otherwise, it is unclear which virus-induced encephalitis could benefit from targeting this pathway due to the varying pathogenicity of neurotropic viruses.

To investigate this hypothesis, we developed mouse models of viral encephalitis using three clinically relevant neurotropic viruses: Japanese encephalitis virus (JEV), herpes simplex virus-1 (HSV-1) and RABV. Our findings revealed an upregulation of SARM1 expression in cortical neurons during neurotropic virus infections, concomitant with a reduction in nicotinamide mononucleotide adenylate transferase 2 (NMNAT2) and NAD⁺ levels, as well as an elevation in cADPR levels. Furthermore, extensive axon degeneration was observed within the cortex of infected mice. To investigate the role of SARM1 in viral encephalitis, we utilized both pharmacological interventions using small molecule inhibitors and genetic approaches using Sarm1-knockout (KO) mouse [19, 35]. In mouse models of encephalitis caused by JEV and HSV-1, both pharmacological inhibition and genetic deletion of SARM1 resulted in delayed cortical neuron axon degeneration, preserved mitochondrial homeostasis, attenuated brain inflammation, ameliorated disease progression, and increased survival rates. Intriguingly, in RABV infection, while SARM1 manipulation delayed disease onset and extended survival time, it did not improve overall survival rates. These results point to SARM1 as a pivotal driver of axonal degeneration and brain inflammation during neurotropic virus infections, highlighting the interaction between SARM1 and neuronal mitochondrial dysfunction. Our results identify SARM1 inhibition as a promising therapeutic strategy for specific forms of viral encephalitis and potentially other neurodegenerative diseases characterized by axonal degeneration and brain inflammation.

Materials and methods

Biosafety and animal ethics statement

The packaging of the recombinant viruses, all cultures of viruses, the verification of virus inactivation, and the intracerebral injection of live viruses in suckling mice were completed in a Biosafety Level 2 Laboratory (BSL-2). All mice used in this study were handled strictly according to the recommendations described in the Chinese Ethical Guidelines for Laboratory Animal Welfare (GB 14925 − 2001). All mice were provided with adequate food and water. Animal experiments were approved by the Changchun Veterinary Institute of the Chinese Academy of Agricultural Sciences.

Mice

Wildtype C57BL/6J mice were purchased from the Guangdong Medical Laboratory Animal Center. SARM1 knockout (KO) mice (Strain no. S-KO-06922) on the C57BL/6J background were purchased from Cyagen Laboratories (Suzhou, China). SARM1 knockout mice were housed and time-mated in the same conditions as wildtype mice. Age-matched mixed groups of 8 to 10-weekold female and male mice were used during the performance of individual experiments. To confirm Sarm1 KO, Western blotting was performed using brain tissue. The SARM1 band was detected at 79 kDa in WT mice and was not present in the Sarm1^KO mice (Fig. 3C).

Cells

BHK-21, Vero, and Neuro-2a cell lines, were cultured in Dulbecco’s modified Eagle’s medium (DMEM, SH30243.FS, Hyclone) containing 10% fetal bovine serum (FBS, A5669801, Gibco) and 100 U/ml of penicillin and 100 mg/mL of streptomycin in humidified incubator at 37˚C with 5% CO₂. Primary cortical neuron cultures were generated from embryonic day 15 (E15) embryos from both wild-type and Sarm1^KO mice, based on previously published protocols [34]. Briefly, cortices from E15 embryos were separated under aseptic conditions, chopped into small pieces and digested with trypsin-EDTA solution (25200072, Gibco) with shaking at 300 rpm at 37˚C for 15 min. Following trypsinization, the tissues were subsequently treated with soybean trypsin inhibitor (17075029, Gibco) and DNase I (M0303L, NEB), and then gently homogenized to generate uniform cell suspensions. Cortical neurons were seeded on Poly-L-Ornithine/Laminin coated 6 well plates (354658, Corning) and cultured in neurobasal media with B27 supplement, glutamax (35050061, Gibco) and gentamicin (15710064, Gibco).

Viruses

JEV SA14-14-2 strain (GenBank accession no. U14163) were provided by Professor Bo Zhang (Wuhan Institute of Virology, Chinese Academy of Sciences, Wuhan, China) and was propagated in BHK-21 cells by infectious clones as previously described [36]. HSV-1 strain 17 syn⁺ (a wild type strain) (GenBank accession no. NC_001806) was obtained from the Wuhan Institute of Virology, Chinese Academy of Science and was propagated at a low MOI in Vero cells [37]. The RABV strain challenge virus standard (CVS-11) was kindly provided by Changchun Veterinary Institute of the Chinese Academy of Agricultural Sciences (Changchun, China) and was amplified in BHK-21 cells.

Mouse infections and survival studies

Adult WT and Sarm1^KO mice (8 to 10 weeks old) of both sexes were anesthetized and were inoculated intracranially (i.c.) with 2 × 10⁶ PFUs of JEV or HSV-1 in a volume of 0.5ul, or the same volume of RABV (2 × 10⁶ TCID₅₀). Correspondingly, mice from the mock group were administered endotoxin-free saline by the same route. These viruses or saline were stereotactically injected into the brain at coordinates X = 1.00 mm, Y = 0.80 mm, and Z = 1.5 mm (i.e., the right ventricle of the mouse) using a stereotaxic apparatus (Reward, Shenzhen, China). For survival studies, groups of mice infected with JEV, HSV-1, or RABV were monitored daily for up to 12 days post-intracranial inoculation to assess mortality as well as the onset of clinical signs and symptoms indicative of acute encephalitis and generalized illness.

Viral encephalitis clinical sign scoring

Symptom scores for mice infected with JEV, HSV-1, or RABV were assessed based on previously established criteria. The scores given for the different clinical symptoms were as follows: score 1, piloerection; score 2, weight loss and hunched posture; score 3, restriction in movement, seizures, mild body stiffening, slight hindlimb extension but no hindlimb paralysis; score 4, hindlimb paralysis and occasional tremor; score 5, body stiffening, hindlimb paralysis and tremor [38,39,40].

Pharmacological interference with CD38 inhibitor 1, FK866 or SIC4

C57BL/6J mice infected with JEV, HSV-1, or RABV were orally administered CD38 inhibitor 1 (15 mg/kg) twice a day from 1 dpi until the end of the experiment. FK866 or SIC4 (20 mg/kg) was intraperitoneally injected twice daily. All compounds were solubilized and diluted using PEG-300/PBS/DMSO. The CD38 inhibitor 1 and FK866 were procured from Selleck (catalog numbers T14913 and T2644, respectively). The isothiazole derivative SIC4, recently discovered by Raul Krauss et al., has been found to exhibit inhibitory effects on the activity of SARM1 [35]. Its chemical name is 3-oxo-5-((2-(trifluoromethyl)phenyl)amino)-2,3-dihydroisothiazole-4-carbonitrile (CAS no. 287196-91-2), and it can be readily obtained from Accela ChemBio Co., Ltd (Shanghai, China). Meanwhile, vehicle-treated animals received injections of the same amount of PEG-300/PBS/DMSO twice daily.

Western blotting

For western blotting, cerebral cortex or neurons lysates were prepared as described before [34, 41]. The protein extracts were prepared by lysing neurons or homogenizing cerebral cortex of mice with RIPA lysis buffer (P0013B, Beyotime) containing a mixture of protease and phosphatase inhibitors (P1046, Beyotime). The following antibodies were used: rabbit anti-SARM1 (1:2,000; ab226930, Abcam); mouse anti-GAPDH (1:10,000; 60004-1-Ig, Proteintech); rabbit anti-Cox IV (1:5,000; 11242-1-AP, Proteintech); rabbit anti-NMNAT2 (1:1,000; 27698-1-AP, Proteintech); rabbit anti-Nf-M (1:2,000; NB300-133, Novus); mouse anti-JEV-NS1 (1:2,000; GTX633820, Genetex); mouse anti-HSV-1-gD (1:1,000; sc-21719, Santa Cruz); Mouse anti-RABV-G (1:1,000; MABF1967, Merck); HRP-conjugated anti-rabbit (1:10,000; 31460, Thermo); and HRP-conjugated anti-mouse (1:10,000; 31430, Thermo). The positive signal was generated by ECL detection kit (P10300, NCM Biotech) using Amersham ImageQuant 800 system (Cytiva, China).

Hematoxylin and eosin (H&E) staining and immunohistochemistry

Whole brain tissues were collected from different groups of mice at indicated days following intracerebral inoculation with JEV, HSV-1, or RABV and subsequently analyzed and compared histopathologically. Briefly, the mice were perfused transcardially with phosphate-buffered saline and post-fixed with 4% paraformaldehyde (PFA). Mouse brain tissue was harvested and fixed in 4% PFA for 24 h at room temperature. The fixed tissues were dehydrated overnight, embedded in paraffin, and sectioned coronally. The sections were rehydrated and stained with H&E (C0105S, Beyotime) according to the manufacturer’s instructions. H&E-stained brain sections from different mice groups were then scored in a blinded fashion by a pathologist using a scoring system (0–5) with increasing severity for quantification of viral encephalitis-associated pathologic markers that included the appearance of gliosis, neuronal loss, neuronophagia, hemorrhage, perivascular cuffing, and degree of meningeal mononuclear cell infiltration [42].

For immunohistochemistry, tissue sections from mouse brains were placed on silane-coated slides, deparaffinized, and rehydrated using decreasing concentrations of ethanol. Antigen retrieval was performed by using citrate buffer (pH 6.0) (P0083, Beyotime) at 100 °C for 15 min. The sections were then treated with S-vision immunohistochemical polymeric antibody (goat anti-rabbit) kit reagents (G1302, Servicebio) or ImmPRESS-AP horse anti-mouse IgG polymer (alkaline phosphatase) detection kit (MP-540215, Vectorlabs), according to the manufacturer’s instructions. The following primary antibodies were used at the respective concentrations: rabbit anti-SARM1 (1:300, ab226930, Abcam), rabbit anti-beta III Tubulin (1:1000, ab18207, Abcam), rabbit anti-Iba1 (1:500, 019-19741, FUJIFILM), mouse anti-GFAP (1:500, MAB360, Merck), and mouse anti-NeuN (clone A60) (1:200, MAB377, Merck). Images were obtained using a Nikon ECLIPSE Si microscope and analyzed with the Image J software.

The axon degeneration index was determined using particle analyzer plugin (http://rsb.info.nih.gov/ij/download.html), as described previously [22, 43]. The β III tubulin-stained axon images were subjected to binarization, subtraction, and measurement of the total axon area. The degeneration index was calculated from the same field of view captured at 5 days post JEV or HSV-1 infection, or at 8 days post RABV infection, in comparison with the control group. Subsequently, the axon degeneration index was determined as the ratio of fragmented axon area to total axon area. At least three biological replicates were performed for each condition.

Nissl’s staining

Nissl staining was performed as previously described [44]. The paraffin Sect. (5 μm thick) were rehydrated and soaked in Nissl staining solution (C0117, Beyotime) for 6 min at room temperature, then washed with double distilled water for 30s, twice, dehydrated with 95% ethanol and 100% ethanol, respectively, and finally soaked in xylene for 10 min and fixed with neutral resin. The images were collected by a Nikon ECLIPSE Si microscope and analyzed quantitatively by Image J.

Immunofluorescence

Mice were perfused with sterile cold PBS and subsequently with 4% PFA. The brains were removed and kept in 4% PFA for 48 h at 4 °C, followed by two incubations in a 30% sucrose solution. Brains were taken out and embedded in Tissue-Tek OCT compound (4583, Saint Louis) and frozen at -20 °C. Fixed tissue was processed to make 10 μm thick cryosections by Leica CM1860 cryostat and mounted on slides. The antigen retrieval process was performed at 100 °C using citrate buffer (P0083, Beyotime). After permeabilization and blocking, the slides were incubated with primary antibodies against Nf-M (1:1000, NB300-133, Novus), JEV-NS1 (1:500; GTX633820, Genetex), HSV-1-gD (1:100; sc-21719, Santa Cruz), RABV-G (1:200, 800 − 092, FUJIREBIO), P2RY12 (1:1000, NBP2-33870, Novus), or GFAP (1:500, MAB360, Merck), overnight at 4 °C. Slides were then washed and incubated with species appropriate secondary antibodies for 2 h at room temperature. Images were obtained using a Nikon ECLIPSE Si microscope and analyzed with the Image J software.

To quantify axonal degeneration by neurofilament loss in mouse cerebral cortex neurons after neurotropic virus infection, the integrated density of neurofilament immunofluorescence in cortical neuronal axons were quantified from the binarized images after normalizing threshold to mock-infected cortical neuron images.

To determine the extent of P2RY12⁺ and GFAP⁺ glial cells coverage, 3 to 4 images were selected for each cerebral cortex at a magnification of 20X. These images underwent a similar thresholding process and subsequent particle analysis using the particle analyzer plugin command in Fiji (Image J). The resulting fraction of area occupied by glial cells was then normalized to mock-infected controls. Quantifications were performed in a blinded manner.

Viral quantification

For quantification of virus in infected mice, after dissecting the cortex from the JEV or HSV-1-infected mouse brains, a 100 mg sample was precisely weighed and subsequently homogenized by adding 600 µl of DMEM. After three repeated freeze-thaw cycles, the homogenates were pelleted by centrifugation at 10,000 g for 10 min, and the supernatants were used for a plaque assay on monolayers of BHK-21/Vero cells seeded in 12-well tissue culture plates. The titers of JEV and HSV-1 were measured by PFU as previously described [38, 42].

Viral titers of RABV in the mouse cerebral cortex were determined by direct fluorescent antibody staining in BHK-21 cells. The cerebral cortex was homogenized as described above. Virus suspension was serially diluted 10-fold to a concentration of 10^− 9 in serum-free medium. Then, 100 µl of each dilution was added in octuplicate into a 96-well plate containing BHK-21 cells with 70% confluence and incubated for viral absorption for 2 h. After that, the supernatant containing virus was replaced with fresh medium containing 2% FBS and incubated at 37˚C with 5% CO₂ for 5 days. The plates were fixed with 4% PFA for 10 min at room temperature, after permeabilization and blocking, stained with FITC conjugated anti-rabies monoclonal globulin (1:200, 800 − 092, FUJIREBIO) in 0.5% BSA/PBS with 0.005% Evans blue. Finally, the plates were read using Nikon ECLIPSE Si microscope to determine median tissue culture infectious dose (TCID₅₀) [34].

Measurements of NAD⁺ and cADPR levels

To determine NAD⁺ levels, cerebral cortex lysates should be prepared following the previously described Western blotting methodology. Supernatant samples were prepared for NAD⁺ and cADPR quantification using the NAD/NADH assay kit (ab65348, Abcam) and fluorimetric cADP-Ribose assay kit (MAK553, Merck), respectively, following the manufacturer’s instructions [33].

ELISA

To measure neurofilament light chain measurements in plasma, blood samples were collected from different groups of mice at indicated days after intracerebral inoculation with JEV, HSV-1, or RABV. The samples were immediately placed on ice and centrifuged at 10,000 g for 5 min at 4 °C. Subsequently, a volume of 15 µl plasma was transferred into round bottom plates (96-well) for the determination of neurofilament light chain (Nf-L). Plasma Nf-L levels were quantified using the mouse Nf-L ELISA kit (NBP2-80299, Novus), following the manufacturer’s instructions. Cytokine levels in mouse cerebral cortex lysates were measured by ELISA using mouse IFN-γ (EMC101g.96, Neobioscience), TNF-α (EMC102a.96, Neobioscience), and IL-6 (EMC004.96, Neobioscience) ELISA kits according to the manufacturer’s instructions.

Flow cytometry

On day 5 or 8 post-infection, mice were anaesthetized and transcardially perfused with ice-cold PBS until the effluent was clear to facilitate the removal of intravascular leukocytes. Cerebral cortex from individual animals were gently cut into small pieces in cold Hank’s balanced salt solution (HBSS) (H6648, Sigma-Aldrich) with a disposable surgical scalpel and incubated with collagenase D (05401020001, Roche) and DNase I (M0303L, NEB) for 30 min in a shaking incubator at 37^oC. Digested tissues were filtered through a 70 μm pore filter (352350, Falcon) to remove cell debris. The homogenate was resuspended in 10 ml of 30% Percoll (P1644, Sigma-Aldrich) in RPMI 1640 medium (11875119, Gibco) which was gradually overlaid on top of 2 ml of 70% isotonic Percoll. Then a gradient was centrifuged at 500 g for 30 min at room temperature. The myelin debris was removed from the top of the gradient and discarded; the resulting layer of leukocytes was collected from the 30%/70% interface and washed with 1 × HBSS. The total number of viable leukocytes was determined by the trypan blue exclusion test. Cell pellets were resuspended in FACS buffer (PBS + 0.5% FBS), blocked with purified rat anti-mouse CD16/32 (101302, 1:500, BioLegend), and incubated with PE anti-mouse CD45 (103105, 1:100, BioLegend) for 30 min on ice in the dark. Cells were then washed with FACS buffer, and analyzed using Beckman CytoFLEX instrument (Beckman, USA) and FlowJo software.

To analysis of mitochondrial depolarization, all steps were carried out on ice except where noted. Neuro-2a cells were seeded in 6-well plates and treated with SIC4 (10 µM) or vehicle, 2 h post JEV infection. After JEV infection (MOI = 1) for 36 h, the cell monolayer was disrupted by gently pipetting the supernatant up and down. Cells were centrifuged at 2,000 g for 5 min and washed once with PBS. A staining solution of 100 nM TMRM (T5428, Sigma-Aldrich) and 100 nM MitoTracker Green (M7514, Invitrogen) in pre-warmed Opti-MEM (31985070, Gibco) was used to resuspend cell pellets. Samples were incubated at 37℃ for 30 min in the dark. Cells were centrifuged, washed three times in PBS and resuspended in 400 µl PBS/2% BSA. 50 µM FCCP (T6834, Targetmol) was added to sample just before analysis mitochondrial depolarization. Cells were acquired using Beckman CytoFLEX S flow cytometer using PE channel for TMRM and FITC channel for MitoTracker Green.

To compare calcium ion concentration in SIC4- and vehicle-treated Neuro-2a cells during JEV infection for 36 h, cells processed as described above and treated with FCCP (50 µM) for 1 h as the positive control for Ca²⁺ influx from the mitochondria to the cytosol. A staining solution of Opti-MEM containing a final concentration of 5 µM Fluo-8 AM (ab142773, Abcam) was prewarmed to 37℃ for 30 min in the dark. Samples were then centrifuged and washed three times with Opti-MEM. Samples were processed on a Beckman CytoFLEX S instrument (FITC channel) and analyzed with FlowJo software.

Analysis of SARM1 clustering by confocal microscopy

Neuro-2a cells were plated onto 15-mm coverglass-bottom petri-dishes (801002, Nest) and then infected with JEV, HSV-1 or RABV (MOI = 1) for 36 h. As a positive control, cells were stimulated with FCCP (50 µM) for 45 min. Immunofluorescence staining and co-localization analysis of cells were performed as described previously [45]. Antibodies used were rabbit anti-SARM1 (1:300, ab226930, Abcam) and mouse anti-TOM20 (1:500, sc-17764, Santa Cruz). Fluorescent images of fixed cells were taken with a confocal laser scanning microscope (TCS-SP8, LEICA, Germany).

Mitochondrial isolation

Primary neurons were infected with JEV, HSV-1 or RABV (MOI = 1) for 24 h. Mitochondria extracts were obtained by cell mitochondria isolation kit (C3601, Beyotime) according to the manufacturer’s instructions. Briefly, the neurons were scraped, washed twice in cold PBS and resuspended in cytoplasmic lysis buffer on ice for 10 min, and then homogenized using homogenizer. The lysates underwent centrifugation at 700 g for 10 min at 4℃. The resulting supernatants were removed and subjected to further centrifugation at 3,000 g for 15 min at 4℃ to separate the heavy mitochondrial fraction. The isolated mitochondrial pellets were washed 3 times with PBS. The removed supernatants were spun at 12,000 g for a duration of 20 min to eliminate peroxisomes and lysosomes, and the cytosolic fractions were collected. Loading buffer containing DTT was added to the mitochondrial and cytosolic fractions. The samples were sonicated, boiled, and resolved on SDS-PAGE. Immunoblotting analysis was conducted utilizing Cox IV (1:5,000; 11242-1-AP, Proteintech) as an indicator of mitochondria, for GAPDH (1:10,000; 60004-1-Ig, Proteintech) as a marker for the cytosol.

Transmission electron microscopy (TEM) and cristae density of mitochondria

JEV- or mock-infected mice were anesthetized using isoflurane and transcardially perfused with prewarmed PBS at 37℃ until the flow-through became clear. Subsequently, prefixation was performed sequentially using Karnofsky’s fixative (E672005, Sangon biotech) and 2.5% glutaraldehyde (G5882, Merck). Following brains removal, the M1 and M2 motor cortices were meticulously dissected under a stereomicroscope. Subsequently, they were cut into 1mm³ sections and immersed in a 2.5% glutaraldehyde solution at 4 °C for a duration of 72 h. The tissues were then fixed in 1% osmium tetroxide, dehydrated stepwise with graded ethanol (50%, 70%, 80%, 90%, and 100%), cleared in acetone, and embedded in EMbed 812 epoxy resin (GE14120, Emcn). Ultrathin sections were cut to 300–400 nm thickness, and then stained with 2.0% uranyl acetate solution and 2.0% lead citrate solution. Images were taken with an H-7650 transmission electron microscope (Hitachi Ltd., Japan). The detailed morphology of mitochondria was analyzed using ImageJ. The mitochondrial cristae density, representing the surface area of mitochondrial cristae per unit volume [46], was quantified using the formula: S_V = (4/π)B_A, where B_A denotes the boundary length density estimated by counting intersections on test lines (I_L) multiplied by π/2 [47].

Statistical analysis

The experiments were conducted with a minimum of three independent replicates, yielding consistent results. Representative image data were selected from at least three randomly chosen fields. Statistical significance was assessed using Student’s t-test or one-way analysis of variance (ANOVA). Survival data were analyzed by comparing Kaplan-Meier survival curves with a log-rank (Mantel-Cox) test. GraphPad Prism v8.0 was used to present means through histograms, accompanied by error bars representing the standard error of the mean [48] in all graphs. A P value less than 0.05 was considered statistically significant. Additional details regarding statistical tests can be found in the respective figure legends.

Results

SARM1 expression was upregulated and activated in cortical neurons of mice with viral encephalitis

Viral encephalitis is exquisitely characterized by neurological impairment with neuroinflammation [2]. Given the role of SARM1 in mediating cell death, cytokine production, and microglia activation following different viral infections, we postulated that SARM1 may be activated in viral encephalitis [18, 31, 32]. To examine the involvement of SARM1 in this disease, it is essential to develop an accurate mouse model, which recapitalizes the neurological phenotypes observed in human patients. Since there are several neurotropic viruses that cause encephalitis, we chose JEV, HSV-1, and RABV to model this disease based on their genus and pathogenicity. Mice were intracranially inoculated with JEV, HSV-1, or RABV to assess disease severity and progression over time. Daily monitoring included evaluation for clinical symptoms of encephalitis (hunched posture, tremors, seizures, and hind limb paralysis) as well as signs of general illness such as lethargy, body weight loss, and reduced mobility. In the mock-infected group, mice remained healthy throughout the experiment without experiencing any weight loss or mortality (Fig. 1A-C). However, infection with JEV, HSV-1 or RABV resulted in severe neurological manifestations and significant body weight loss in mice within 4 to 10 days post-infection (dpi), ultimately leading to their demise (Fig. 1A-C). To characterize the virus-induced neuropathological changes, we examined coronal brain sections from infected and control mice. Mock-infected brain tissue exhibited normal cytoarchitecture, characterized by large round nuclei and abundant cytoplasm (Figure S1A-C, Supporting Information). JEV and HSV-1 infections induced severe focal inflammation in cortex and thalamus, marked by distinctive features including perivascular cuffs with mononuclear cells or microglia, hemorrhage, necrosis, and extensive disruption of the normal cytoarchitecture (Figure S1A-C, Supporting Information). RABV infection, however, produced milder inflammatory changes without obvious neuronal cytopathology compared to mock infection (Figure S1A-C, Supporting Information). These results suggest that mice serve as a reliable model for studying viral encephalitis.

Fig. 1

Neurotropic virus infections of mouse cortical neurons induce SARM1 activation and increased production of cADPR. (A-C) Survival curves, weight loss and clinical scores of infected C57BL/6 mice were measured from day 0 until the end of the infection. Intracranial infections with JEV, HSV-1, RABV or mock (PBS) was performed in C57BL/6 mice (n = 6 for each mock-infected; n = 18 for each neurotropic virus infection group), respectively. Neurological disease evaluation included mortality (A), mean daily body weight loss (B), and clinical scores (C). The mock infection mice did not have any symptoms. The scores of disease severity or 30% loss of initial body weight were used as termination criteria. (D-F) The immunoblot analysis of SARM1, NMNAT2, and viral proteins in whole-cell lysates from the mouse cortex was conducted at the indicated time points following mock, JEV (D), HSV-1 (E), and RABV (F) infections using GAPDH as an internal control. (G) Representative micrographs of immunohistochemistry of brain sections stained for SARM1 from mock-, JEV (5dpi)-, HSV-1 (5dpi)-, or RABV (8dpi)-infected mice. (H and I) Mice were mock-infected or infected with neurotropic viruses at the indicated time points, respectively; the NAD⁺ (H) and cADPR (I) levels were tested in the mouse cortex, quantified as the ratio of (NAD⁺ or cADPR concentration) / (NAD⁺ or cADPR concentration of mock infected group) × 100%. (J) The day before neurotropic virus infection, mice were orally given vehicle or CD38 inhibitor 1 (twice daily). Compared with the mock-infected group, the levels of NAD⁺ and cADPR in the cortex of mice were detected after JEV (5dpi), HSV-1 (5dpi), and RABV (8dpi) infection, respectively. Individual data points are shown (n = 12 for each group). Data are expressed as means ± SEM, Student’s t test, * P<0.05, ** P<0.01,*** P<0.001

SARM1 exhibits preferential expression in neuronal populations within the brain [26]. To investigate the expression pattern of SARM1 in mouse models of viral encephalitis, we performed Western blot analysis. Our findings revealed increased SARM1 levels in the cortex following viral infection, concurrent with decreased NMNAT2 expression (Fig. 1D-F). Immunohistochemistry staining further demonstrated that SARM1 was predominantly localized to pyramidal neurons and significantly elevated upon JEV, HSV-1, or RABV infection compared to mock-infected controls (Fig. 1G). Since NMNAT2 catalyzes NAD⁺ production from NMN and ATP [49], we hypothesized that its virus-induced degradation in cerebral cortex might lead to reduced NAD⁺ levels and NMN accumulation. As an increase in the NMN/NAD⁺ ratio activates SARM1, causing catastrophic NAD⁺ depletion and cADPR production [22], we investigated the impact of viral infections on these metabolic markers. Indeed, NAD⁺ levels in the cerebral cortex were significantly reduced in JEV, HSV-1, or RABV-infected mice compared to mock-infected controls (Fig. 1H), accompanied by increased cADPR levels (Fig. 1I), a validated biomarker of SARM1 activation [50]. Considering CD38 as another NADase [51], in order to rule out its effect on the decline of NAD⁺ levels after viral infections, we administered the CD38 inhibitor 1 to mice. CD38 inhibition increased NAD⁺ levels only in mock-infected group but showed no significant effect in virus-infected groups (Fig. 1J). Notably, rather than inhibiting cADPR production, CD38 inhibitor 1 treatment showed enhanced cADPR elevation in the cerebral cortex when mice infected with these viruses (Fig. 1J). Collectively, these findings indicate that neurotropic virus-induced encephalitis triggers SARM1 activation and cADPR production in the mouse cerebral cortex.

Axonal degeneration of cortical neurons occurs in the mouse models of viral encephalitis

Nerve damage is a key pathological manifestation of viral encephalitis [2]. Previous studies have reported that infection of primary neurons with ZIKV and RABV can induce axonal degeneration, in vitro [33, 34]. To comprehensively characterize virus-induced neuronal pathogenesis in vivo, we established mouse models of viral encephalitis using multiple neurotropic viruses. We initially tracked temporal changes in axonal integrity by monitoring neurofilament medium chain (Nf-M) expression in the cerebral cortex following JEV infection. Axonal degeneration became evident as early as 3 dpi, progressing to almost complete loss of Nf-M staining by 5 dpi, indicating extensive axonal destruction (Figure S2A-C, Supporting Information). Similar patterns of neurofilament-positive axonal loss were observed with HSV-1 and RABV infections, manifesting at 5 dpi and 8 dpi, respectively (Fig. 2A-B). While axonal structural proteins underwent specific degradation, neuronal cell bodies and nuclei remained largely intact during HSV-1 and RABV infections (Fig. 2C-D). To further quantify axonal degeneration, we examined beta III tubulin expression as an additional axonal marker. Immunohistochemistry revealed granular disintegration of axons in JEV, HSV-1, and RABV-infected cortical neurons, corroborating our Nf-M findings (Fig. 2C-D). These findings demonstrate substantial axonal degeneration in the mouse cerebral cortex following neurotropic viral infections. Furthermore, Western blot analysis of cortical lysates confirmed the specific degradation of Nf-M in all three viral infections (Fig. 2E). Given that neurofilament light chain (Nf-L) serves as a clinically relevant biomarker of axonal degeneration in various neurodegenerative disorders [50], we measured plasma Nf-L levels in infected mice. While mock-infected mice maintained stable baseline plasma Nf-L levels (214.13 pg/ml), infected mice showed significant elevations coinciding with the onset of neurological symptoms: at 3 dpi for JEV and HSV-1, and 6 dpi for RABV (Fig. 2F). Our observations reveal extensive axonal degeneration in the mouse cerebral cortex following neurotropic virus infections. The strong correlation between plasma Nf-L levels and disease progression points to its potential as a biomarker for viral encephalitis induced neuronal damage.

Fig. 2

Neurotropic virus infections induce axonal degeneration in mouse cortical neurons. (A) Representative immunofluorescence staining images of viral proteins (green), and neurofilament medium chain (Nf-M) (red) in cerebral cortex tissue sections of mice at 5 days post JEV or HSV-1 infection, and 8 days after RABV infection. (B) Quantification of neurofilament immunostaining intensity in the axonal fluorescence images shown in A, indicated by Nf-M antibody. IntDen: Integrated Density. (C) Representative immunohistochemistry staining of beta III tubulin in brain tissue Sect. 5 days post JEV or HSV-1 infection, and 8 days post RABV infection. (D) Degree of axonal degeneration after neurotropic virus infections, quantified as a degeneration index (DI), where a DI of 0.35 or above represents degenerated axons, indicated by a horizontal dotted line. (E) Immunoblot analysis of Nf-M and SARM1 levels in mouse cerebral cortex after JEV, HSV-1, or RABV infection at indicated time points. GAPDH is used as an internal control. (F) Mice were infected by neurotropic viruses JEV, HSV-1 or RABV at the indicated time points, and neurofilament light chain (Nf-L) levels were measured in plasma. All three neurotropic viruses induced increases in plasma Nf-L in a time-dependent manner. Data correspond to means from replicate experiments, and error bars indicate mean ± SEM of data from 6 mice in each group. Student’s t test, ns, P > 0.05; ***P < 0.001

SARM1 is required for the execution of neurotropic viruses induced axonal degeneration

Having established SARM1 activation and axonal degeneration as consequences of neurotropic virus infection, we investigated whether SARM1 plays a causative role in virus-induced axonal degeneration. We employed both pharmacological inhibition and genetic knockout approaches to test this hypothesis. Initially, we evaluated the impact of inhibiting the SARM1 pathway on delaying axonal degeneration post-infection with these neurotropic viruses. The irreversible inhibitor of SARM1, denoted as SIC4, belongs to the class of isothiazole derivatives. It directly binds to activated SARM1, suppresses its NADase activity, and preserves axon integrity in paclitaxel-induced peripheral neuropathy [35]. Wild-type mice infected with JEV showed significant loss of Nf-M positive axons in the cortex by 5 dpi when treated with vehicle (Fig. 3A-B, S3A, Supporting Information). In contrast, mice receiving intraperitoneal injections of SIC4 (20 mg/kg, twice daily) starting one day post-infection showed substantial protection of cortical neuronal axons. Similarly, SARM1 knockout mice exhibited minimal loss of Nf-M staining, with most axons remaining intact (Fig. 3A-B, S3A, Supporting Information). Quantitative analysis confirmed significantly higher Nf-M fluorescence intensity in both SIC4-treated and SARM1 knockout mice compared to vehicle-treated controls. This protective effect extended to other neurotropic viruses, with both SIC4 treatment and SARM1 knockout delaying axonal degeneration following HSV-1 (5 dpi) or RABV (8 dpi) infection (Fig. 3A-B, S3A, Supporting Information). To ensure that this protection did not result from reduced viral infection, we examined viral protein expression and viral titers in cortical neurons. Neither SIC4 treatment nor SARM1 knockout significantly affected viral protein levels (Fig. 3A-B) or viral titers (Figure S3B-C, Supporting Information). Western blot analysis of cerebral cortex lysates confirmed significantly higher Nf-M levels in both SIC4-treated and SARM1 knockout mice compared to wild-type controls following infection with JEV (5 dpi), HSV-1 (5 dpi), or RABV (8 dpi) (Fig. 3C). Moreover, plasma levels of Nf-L, a biomarker of axonal degeneration were significantly reduced in both SIC4-treated and SARM1 knockout mice compared to their respective controls across all viral infections (Fig. 3D). Collectively, these findings demonstrate that SARM1 is essential for neurotropic virus-induced axonal degeneration and that its inhibition provides substantial axonal protection without affecting viral replication or protein expression in neurons.

Fig. 3

Inhibition or deletion of SARM1 significantly delays axonal degeneration in cerebral cortex neurons infected with neurotropic viruses. (A) Representative double immunostaining was performed on coronal cerebral cortex sections from 8-week-old wild type (WT) and Sarm1^KO mice, 5 days after infection with JEV or HSV-1, or 8 days after RABV infection. The sections were stained for JEV-NS1, RABV-G or HSV-1-gD (green, virus marker) and Nf-M (red, axonal markers), and mounted with Vectashield containing DAPI (blue; nuclear marker). (B) Quantification of neurofilament immunostaining intensity in the axonal fluorescence images from experiments shown in panel A, indicated by Nf-M antibody. IntDen: Integrated Density. (C) Immunoblot analysis of the Nf-M and SARM1 levels in cerebral cortex with JEV, HSV-1 or RABV-infected WT or Sarm1^KO mice. (D) Plasma neurofilament light chain (Nf-L) levels were quantified from WT and Sarm1^KOmice and mice receiving JEV, or HSV-1 infection (5 days) or RABV infection for 8 days. (E and F) Cerebral cortex NAD⁺ (E) and cADPR (F) levels in WT or Sarm1^KOmice after JEV, HSV-1, or RABV infection. Data represent mean ± SEM; individual data points are shown. Statistical analysis was performed with one-way ANOVA and Holm-Bonferroni multiple comparison. ns, P > 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001

The activation of SARM1 catalyzes the breakdown of NAD⁺ and NADP, leading to the production of cADPR and ADPR [41]. We investigated how SARM1 inhibition or deletion affects these metabolites during neurotropic virus infections. We measured metabolite levels in the cerebral cortex of vehicle-treated, SIC4-treated, and SARM1 knockout mice following infection with neurotropic viruses. Under baseline conditions, SARM1 knockout mice showed modestly elevated NAD⁺ levels and slightly reduced cADPR levels compared to wild-type controls, although these differences were not statistically significant (Fig. 3E-F). Following infections with JEV, HSV-1, or RABV, a decrease in NAD⁺ levels was observed in the cerebral cortex of both SIC4-treated wild-type and SARM1 knockout mice compared to mock-infected counterparts; however, there was a significant increase in NAD⁺ levels compared to vehicle-treated wild-type mice (Fig. 3E). The virus-induced elevation of cADPR in the cerebral cortex was attenuated by SIC4 treatment, with the magnitude of reduction varying among different viral infections (Fig. 3F). As expected, this viral infection-dependent increase in cADPR from the cerebral cortex was effectively inhibited in SARM1 knockout mice (Fig. 3F). Nevertheless, even in the absence of SARM1 activity, NAD⁺ levels still decreased by approximately 30% in virus-infected cortices compared to mock-infected controls (Fig. 3E). This outcome suggests that NAD⁺ biosynthesis may have been impacted by neurotropic virus infections or that other NAD⁺-related hydrolases aside from SARM1 are activated. Thus, these results show that SARM1 is essential for driving axonal degeneration and facilitating cADPR production in cortical neurons infected by neurotropic viruses.

Neurotropic virus infections cause SARM1 localization in the mitochondria, which is related to mitochondrial damage

The aforementioned study indicated that SARM1 plays a critical role in the axonal degeneration associated with neurotropic virus infections. We subsequently explored the underlying mechanisms of SARM1 activation and axonal degeneration during these infections. Considering that the N-terminal domain of SARM1 contains a mitochondrial localization signal sequence [52], we first examined whether neurotropic virus infections induce SARM1 localization to the mitochondria in Neuro-2a cells. Confocal microscopy analysis demonstrated a significant overlap between SARM1 and the mitochondrial marker TOM20 in Neuro-2a cells infected with JEV, HSV-1, or RABV, but not in mock-infected controls (Fig. 4A). Fluorescence intensity profile analysis further confirmed this striking colocalization of SARM1 with TOM20 (Fig. 4A). To validate these microscopy findings biochemically, we analyzed cortical neurons infected with JEV, HSV-1, or RABV and found increased levels of SARM1 within the mitochondrial fraction relative to mock-infected controls, while the cytosolic fraction showed no such increase (Fig. 4B). Despite mitochondrial localization of SARM1, the mechanism by which it induces axonal degeneration during neurotropic virus infections remains unclear. Since previous studies showed that injury-induced loss of mitochondrial membrane potential is SARM1-dependent [53], we examined whether SARM1 mediates similar effects during viral infection. We focused on JEV infection as a representative neurotropic virus for these functional studies. To accurately compare mitochondrial membrane potential in mock- and JEV-infected Neuro-2a cells, we gated for Mito Tracker-Green-positive cells and assessed the loss of TMRM (tetramethylrhodamine methyl ester) fluorescence in intact mitochondria. This analysis revealed that vehicle-treated JEV-infected cells exhibited increased mitochondrial depolarization compared to SIC4-treated counterparts, while mock-infected cells treated with vehicle did not show such changes (Fig. 4C-D). In addition to mitochondrial depolarization, calcium homeostasis is disrupted in injured axons [41]. To determine whether SARM1 activation during viral infection affects calcium homeostasis, we measured cytoplasmic Ca²⁺ levels using a calcium-dependent fluorescent probe. We measured Ca²⁺ levels in Neuro-2a cells infected with JEV for 36 h or treatment them with FCCP (50 µM) for 1 h as a positive control for Ca²⁺ influx from mitochondria to the cytosol. Both JEV infection and FCCP treatment resulted in a significant increase in cellular fluorescence intensity compared to mock-infected controls (Fig. 4E-F). Notably, pretreatment with the SARM1 inhibitor SIC4 strongly attenuated the elevation of cytosolic Ca²⁺ triggered by JEV infection (Fig. 4E-F).

Fig. 4

Neurotropic virus infections induce mitochondrial localization of SARM1 and mitochondrial damage. (A) Immunofluorescence of cells labeled with SARM1 (green) and TOM20 (red) antibodies. Neuro-2a cells were mock-infected or infected with JEV, HSV-1 or RABV (MOI = 1) and fixed at 36 h post-infection. As a positive control, cells were stimulated with FCCP (50 µM) for 45 min, and fixed. Nucleic acids were stained with DAPI (blue). White boxes indicate zoomed areas. Colocalization quantifications were done using the Plot Profile plugin in ImageJ. (B) Immunoblot analysis of SARM1 in the cytosol or mitochondrial fraction of cortical neurons. The cortical neurons were mock-, or JEV-, or HSV-1- or RABV-infected for 36 h. Cox IV and GAPDH were used as loading controls. (C) Neuro-2a cells were mock, or infected with JEV (MOI = 1) for 36 h and either treated with vehicle or 10µM SIC4. FCCP was used as the positive control. Cells were stained with TMRM and MitoTracker Green and analyzed using CytoFLEX S. Flow cytometry plots are representative of three independent experiments. (D) Bar chart is mean ± SEM (n = 3) showing percentage of cells relative to JEV infection vehicle treatment that are MitoTracker Green positive and TMRM negative. (E) The concentration of Neuro-2a cytosolic Ca²⁺ was determined by measuring fluo-8 fluorescence using flow cytometry (loaded with 1 mg/ml fluo-8-AM for 30 min) during JEV infection for 36 h either treated with vehicle or 10 µM SIC4. FCCP as the Ca²⁺ mobilization positive control. (F) Relative quantification of the mean fluorescence intensity of fluo-8 for the experiments in panel E. (G) Electron microscope images of cortical nerves from 8-week-old WT and Sarm1^KO mice showing representative axonal mitochondria. Morphological changes in the mitochondria associated with JEV infection. TEM images were acquired at 5 dpi from mock- and JEV infected WT and Sarm1^KO mice. Data are representative of 5–6 fields for each group. (H) Quantification of axons displaying abnormal mitochondria at 5 dpi from mock- and JEV-infected WT and Sarm1^KO mice. (I) Quantification of cristae density of mitochondria in axons. **P < 0.01, ***P < 0.001, ns, no significant difference

To further investigate the effects of SARM1 on neuronal mitochondrial damage during JEV infection, we performed ultrastructural analysis of mitochondria in cortical neurons wild-type and Sarm1 knockout (Sarm1^KO) mice using electron microscopy. In wild-type mice, JEV infection resulted in severe mitochondrial abnormalities in both neuronal cell bodies and axons, characterized by swollen morphology and either reduced cristae density or complete loss of discernible cristae, compared to mock-infected controls (Fig. 4G-I; Figure S4A-C, Supporting Information). Treatment with the SARM1 inhibitor SIC4 substantially protected mitochondrial integrity during JEV infection, with only minor cristae density defects observed compared to mock-infected controls (Fig. 4G-I; Figure S4A-C, Supporting Information). Furthermore, neurons from Sarm1^KO mice maintained normal mitochondrial morphology and cristae density even after JEV infection (Fig. 4G-I; Figure S4A-C, Supporting Information). In cell bodies, vesicular structures indicated by red arrows represent replication and transcription complexes that emerge following JEV infection, implying that neither SARM1 inhibition nor deletion affected JEV replication (Figure S4A, Supporting Information). Thus, we propose that activation of SARM1 is essential for mitochondrial damage during neurotropic virus infections. This SARM1-dependent mitochondrial damage pathway may represent a key mechanism underlying virus-induced axonal degeneration.

Deletion of Sarm1 mitigates neuropathological alterations and brain inflammation associated with neurotropic virus infections in vivo

Based on our findings that SARM1-mediated axonal degeneration and mitochondrial damage in cortical neurons, we next investigated its contribution to cerebral cortex pathology during neurotropic virus infections in vivo. We intracranially infected WT and Sarm1^KO mice with 10³ plaque-forming units (PFU) of JEV, a dosage that induces neurological disease with 100% incidence in 8-week-old WT mice. Sarm1^KO mice exhibited significantly delayed disease onset and milder neurological symptoms compared to WT counterparts (Figure S5A, Supporting Information). Furthermore, while all WT mice succumbed to death 7 days post-JEV infection, approximately 25% of Sarm1^KO mice survived beyond this period (Fig. 5A). Similar protective effects of Sarm1 deletion were observed following intracranial infection with HSV-1 or RABV, although in the case of RABV, Sarm1 deletion only delayed rather than prevented mortality (Fig. 5B-C; Figure S5B-C, Supporting Information). Histopathological examination further revealed striking differences in virus-induced tissue damage between genotypes (Fig. 5D). JEV and HSV-1-infected WT mice manifested moderate to severe disease characterized by pronounced gliosis, neuronal loss, neuronophagia, hemorrhage, perivascular cuffing, and prominent meningeal mononuclear cell infiltration (Fig. 5D). Conversely, cerebral cortex sections from JEV- or HSV-1-infected Sarm1^KO mice exhibited minimal pathological changes with only sparse foci of gliosis, negligible perivascular cuffing, and relatively mild meningeal mononuclear cell infiltration (Fig. 5D). Blinded evaluation of these pathological markers by an expert pathologist confirmed significantly reduced neuropathology scores in Sarm1^KO mice (Fig. 5E). Notably, RABV infection induced only minimal to mild cortical damage, with no significant pathological differences between Sarm1^KO and WT mice (Fig. 5D-E). Additionally, we performed immunohistochemistry on cerebral cortex sections using NeuN antibodies (a neuronal marker) to evaluate the effect of SARM1 on neuronal loss following neurotropic virus infections. The NeuN staining revealed significantly reduced neuronal loss in the cerebral cortex of Sarm1^KO mice at 5 days post-infection with JEV or HSV-1 compared to WT mice (Fig. 5F; Figure S5E, Supporting Information). Consistent with the histopathology findings, RABV infection did not cause significant neuronal loss in either genotype compared to mock-infected mice (Fig. 5F; Figure S5E, Supporting Information). Subsequent Nissl staining results indicated that neurotropic virus infection led to central Nissl body lysis, neuronal cell body swelling, and nuclear deviation in cortical neurons relative to mock group (Figure S5F, Supporting Information). Importantly, these cellular pathologies were largely prevented in Sarm1^KO mice (Figure S5F, Supporting Information).

Fig. 5

Sarm1 KO restores neuropathological alterations and reduces the brain inflammation induced by neurotropic viruses in mice. Eight-week-old mouse pups of wild type (WT) or Sarm1 knockout (KO) were intracranially infected with JEV, HSV-1, RABV or mock-infected. (A-C) Survival plot showing percent survival of WT and Sarm1 KO mice after JEV (A), HSV1 (B), and RABV (C) infection (n = 6 for each mock-infected; n = 18 for each neurotropic virus infection group). (D) Representative micrographs of the cerebral cortex of WT and Sarm1 KO mice infected with JEV (5 dpi), HSV-1 (5 dpi), RABV (8 dpi) or mock-infected. Red triangle: meningitis; red arrow: hemorrhage and hyperemia or perivascular cuffing; Hollow triangle: necrotic loci or neuronophagia. (E) Pathological scores of the cerebral cortex sections as shown in (D). (F) Coronal sections of the brains as shown in (D) were subjected to immunohistochemistry with NeuN antibody. Quantitative analysis of the number of NeuN⁺ cells from WT or Sarm1 KO mice. (G-I) WT and Sarm1 KO mice were infected with JEV, HSV-1, RABV or mock-infected (n = 6 for each group). At the indicated time points, levels of IFN-γ (G), TNF-α (H) and IL-6 (I) in cerebral cortex homogenates were measured by ELISA. (J) Representative micrographs of immunohistochemistry of brain sections stained for Iba-1 and GFAP from WT and Sarm1 KO surviving mice at 35 days after HSV-1 infection. (K) IFN-γ, TNF-α and IL-6 were measured with ELISA in cerebral cortex samples of WT and Sarm1 KO surviving mice with HSV-1 infection (35 dpi). All data are presented as mean ± SEM, and statistical significance was determined using Student’s t-test, one-way ANOVA was used with Tukey’s or multiple-comparison test or Log-rank (Mantel-Cox) test. *P < 0.05, **P < 0.01, ***P < 0.001. ns, not significant. ND, not detected

To assess whether SARM1’s role in neuronal damage extends to neuroinflammatory responses, we measured proinflammatory cytokine levels in the cerebral cortex following viral infection. Sarm1 depletion significantly reduced the expression of IFN-γ, TNF-α, and IL-6 in mice infected with JEV or HSV-1 compared to WT counterparts (Fig. 5G-I). Notably, while RABV infection showed a distinct pattern where Sarm1 deletion only reduced IL-6 levels without affecting IFN-γ or TNF-α expression (Fig. 5G-I), consistent with the milder neuropathology observed in RABV-infected mice. (Fig. 5D-E). In addition to its role in orchestrating the acute inflammatory response during neurotropic virus infections, SARM1 also influenced chronic neuroinflammatory responses. In WT mice that survived HSV-1 infection, we observed persistent activation of microglia and astrocytes at 35 days post-infection. In contrast, these glial cells returned to a quiescent state in Sarm1^KO mice (Fig. 5K). This difference in chronic glial activation was accompanied by significantly lower levels of IFN-γ and IL-6 in the cerebral cortex of Sarm1^KO mice compared to WT survivors (Fig. 5L). Collectively, these results demonstrate that SARM1 contributes to both acute and chronic neuroinflammatory responses during neurotropic virus infections, with its deletion providing protection against both immediate tissue damage and persistent inflammation.

FK866 or SIC4 treatment alleviates neuropathology and brain inflammation in mice infected with neurotropic viruses

The important role of SARM1 in mediating neuropathology, prompted us to explore its potential as a therapeutic target for neurotropic virus infections. In addition to direct inhibition of SARM1 NADase by SIC4, FK866 was employed in this study. It acts as a feedback inhibitor to impede SARM1 activation. FK866 has been administered in mice following spinal cord injury and demonstrated favorable pharmacological tolerance [19]. To evaluate these compounds’ therapeutic potential in viral encephalitis, we intracranially infected 8-week-old mice with JEV, HSV-1, or RABV in 0.5 µl volume, while also including mock-infected controls. Beginning one day post-infection, mice received intraperitoneal injections of either FK866 (20 mg/kg) or SIC4 (20 mg/kg) twice daily, with vehicle (PEG-300/PBS/DMSO) serving as a control. Following JEV infection, clinical symptoms manifested one day later in both FK866 and SIC4 treatment groups compared to the vehicle group; neurological signs such as hunch posture and tremors began to resolve eight days after JEV infection (Figure S6A, Supporting Information). Survival analysis showed that treatment with FK866 or SIC4 resulted in survival rates of 11% and 16%, respectively, while all vehicle-treated mice succumbed to death between days 5 and 7 post-infection (Fig. 6A). In HSV-1 or RABV-infected mice, both FK866 and SIC4 treatments alleviated clinical symptoms and improved survival rates of HSV-1 infected mice (Fig. 6B-C; Figure S6B-C, Supporting Information). Further histopathological examination revealed extensive infiltration of inflammatory cells and necrotic loci in the cerebral cortex of vehicle-treated mice infected with JEV or HSV-1; however, these pathological changes were significantly attenuated following treatment with FK866 or SIC4 (Fig. 6D; Figure S6D-E, Supporting Information). At 8 days post-RABV infection, SARM1 inhibitors reduced cortical edema in comparison to their vehicle-treated counterparts (Fig. 6D; Figure S6D-E, Supporting Information). Notably, no apparent pathological changes were observed in mice treated with control vehicles or in those administered FK866 or SIC4 without prior viral infections (Fig. 6D; Figure S6D-E, Supporting Information). Moreover, during JEV or HSV-1 infection, Nissl staining and immunostaining demonstrated that FK866 or SIC4 treatment significantly preserved both Nissl bodies and NeuN⁺ cells within the cerebral cortex compared to vehicle-treated controls, further substantiating the neuroprotective effect of SARM1 inhibition in viral encephalitis models (Fig. 6E-F). In addition, RABV infection showed no difference in the density of Nissl bodies and NeuN⁺ cells when compared to the mock-infected group (Fig. 6E-F).

Fig. 6

Inhibition of SARM1 by FK866 or SIC4 alleviates neuropathology and brain inflammation in neurotropic viruses-infected mice. Eight-week-old mouse pups were intracranially infected with 2 × 10⁶ PFU of JEV or HSV-1, or 2 × 10⁶ TCID₅₀ of RABV in 0.5 µl, or mock-infected, and subsequently treated twice a day with FK866 (20 mg/kg, i.p.), SIC4 (20 mg/kg, i.p.), or vehicle (PEG-300/PBS/DMSO). (A-C) Survival curves of JEV (A), HSV-1 (B), RABV (C)- and mock-infected mice treated with FK866, SIC4, or control vehicle (n = 6 for each mock-infected; n = 18 for each neurotropic virus infection group). (D) Pathological scores of the cerebral cortex sections. (E) Quantitative analysis of the number of NeuN⁺ cells of JEV (5 dpi), HSV-1 (5 dpi), RABV (8 dpi) or mock-infected mice treated with FK866, SIC4 or vehicle. (F) Quantitative analysis of density of Nissl bodies in the mouse cerebral cortex. Mice were infected and treated as above described. (G) Double immunostaining P2RY12 (red) and GFAP (green) in coronal cerebral cortex slices of JEV (5 dpi), HSV-1 (5 dpi), RABV (8 dpi) or mock-infected mice treated with FK866, SIC4 or vehicle (n = 5 per group). (H and I) Area fraction occupied by P2RY12⁺ (H) and GFAP⁺ (I) cells and the average density in the same area; data shown as fold change (FC) relative to mock-infected mice, with one dot being one mouse. (J) Flow cytometric analysis of cerebral cortex infiltration of CD45⁺ cells 5 days after JEV or mock infection of mouse brain tissues. (K) Bar graphs show the number of leukocytes expressing CD45 in the cerebral cortex of JEV (5 dpi), HSV-1 (5 dpi), RABV (8 dpi) or mock-infected mice treated with FK866, SIC4 or vehicle (n = 4 per group). All data are presented as mean ± SEM, and statistical significance was determined using one-way ANOVA test. Kaplan-Meier survival curves of animals infected with these viruses using a log-rank test are shown. ns, not significant. *P < 0.05, **P < 0.01, ***P < 0.001

Brain inflammation serves as both a biomarker of viral encephalitis and an aggravating pathogenic factor. Therefore, we investigated whether SARM1 inhibition modulates neuroinflammatory responses during neurotropic virus infections. Double immunostaining showed that even with treatment using SARM1 inhibitors, the densities of P2RY12⁺ microglia and GFAP⁺ astrocytes in mock-infected mice did not significantly differ from those in vehicle-treated controls (Fig. 6G-I). However, infection with JEV, HSV-1, or RABV triggered marked increases in both P2RY12⁺ and GFAP⁺ cell densities, accompanied by hypertrophic morphological changes characteristic of activation (Fig. 6G-I). Notably, treatment with either FK866 or SIC4 significantly reduced both the density and activation state of GFAP⁺ astrocytes in the infected cerebral cortex, while P2RY12⁺ microglial responses remained unaffected (Fig. 6G-I). This selective effect suggests that SARM1 specifically regulates astrocyte activation during viral encephalitis. Furthermore, we sought to determine whether SARM1 inhibition might influence the recruitment of inflammatory cells into the brain during the onset of viral encephalitis. As illustrated in Figure S6D-E, prior HE staining indicated a significant decrease in infiltrating cell density within FK866- or SIC4-treated mice following JEV, HSV-1, or RABV infection compared to vehicle-treated controls. To quantify this effect more precisely, we subsequently performed flow cytometric analysis of CD45⁺ (a marker of common leukocyte antigen) cells isolated from the cerebral cortex of FK866-, SIC4-, or vehicle-treated mice at 5 days post-intracranial inoculation with JEV or HSV-1, as well as at 8 days following RABV inoculation. No significant differences were observed in the density of CD45⁺ cells among mock-infected mice. However, a notable reduction in the recruitment of CD45⁺ cells was evident in the cerebral cortex of FK866- or SIC4-treated viral encephalitis mice compared to their vehicle-treated counterparts (Fig. 6J-K; Figure S6F-G), indicating that inhibition of SARM1 activity alleviates the infiltration of inflammatory cells in these models. These findings demonstrate that SARM1 inhibition produces multiple beneficial effects during viral encephalitis: protection against axonal degeneration (as shown earlier), selective suppression of astrocyte activation, and reduced inflammatory cell infiltrations. Collectively, this combination of neuroprotective and anti-inflammatory effects suggests that SARM1 inhibitors may offer therapeutic potential for treating neurotropic virus infections.

Discussion

Here, we identify SARM1 as a critical mediator of neuropathology in murine models of neurotropic viral infection. Viral challenge triggers SARM1-dependent mitochondrial dysfunction through loss of membrane potential and pathological Ca²⁺ release, driving axonal degeneration and brain inflammation. Genetic ablation or pharmacological inhibition of SARM1 (via FK866 or SIC4) preserved neuronal integrity, attenuated neuroinflammatory responses, and improved survival outcomes. Our findings mechanistically link SARM1 activation to viral encephalitis pathogenesis and establish its therapeutic targeting as a strategy to mitigate currently untreatable virus-induced neuronal injury.

Viral encephalitis remains therapeutically intractable beyond acyclovir for HSV, with most interventions limited to supportive care due to the absence of validated molecular targets [2, 3]. To address this therapeutic gap, we established intracranial infection models using three neurotropic viruses representing distinct genomic classes: JEV (flavivirus), HSV-1 (alpha-herpesvirus), and RABV (lyssavirus). These models successfully recapitulated hallmark neuropathological features and disease progression documented in previous studies [39, 40, 54], with no sex-specific differences in outcomes. Consistent with prior reports [26], SARM1 exhibited predominant neuronal localization in our models. Notably, neurotropic viral infection induced neuronal SARM1 upregulation, contrasting with its reported decline in aging APP/PS1 mice and post-injury elevation in spinal cord trauma [17, 19], suggesting context-dependent regulatory mechanisms. Mechanistically, aligned with previous NMNAT2 depletion investigations [55], late-stage infection coincided with reduced cortical NMNAT2, a labile NAD⁺ synthase transported axonally, potentially elevating the NMN/NAD⁺ ratio and thereby activating SARM1. Corroborating SARM1 activation, viral encephalitis induced decreased cortical NAD⁺ levels concurrent with elevated cADPR, a direct enzymatic product of SARM1 [50]. While CD38 inhibition (via CD38 inhibitor 1) failed to rescue NAD⁺ levels, it amplified cADPR accumulation. Given that CD38 functions as both an NADase and a cADPR hydrolase [25], we propose that cADPR overproduction, stemming from virus-induced SARM1 activation, accumulates further under CD38 inhibition.

Pathological axonal degeneration represents a hallmark feature across various neurological diseases, including traumatic brain injury, Alzheimer’s disease, Parkinson’s disease and multiple sclerosis. While previous studies demonstrated RABV- and ZIKV-induced NAD⁺ depletion and axonal degeneration in vitro [33, 34], our findings expand this knowledge by revealing that JEV, HSV-1, and RABV similarly induce cortical axonal degeneration in vivo. Moreover, corroborating previous findings [56], we observed elevated plasma levels of Nf-L, an axonal degeneration marker, and highlighting its potential utility as a diagnostic biomarker for viral encephalitis induced neuronal damage. SARM1 inhibition or deletion preserved cortical axonal integrity in viral encephalitis without altering viral replication, establishing SARM1-dependent degeneration as a conserved mechanism across neurotropic infections and neurodegenerative conditions [17, 19, 57, 58]. As previously mentioned, this mechanism may serve as a neuroprotective response to restrict viral transmission [18, 34]. Neurotropic viruses, including Theiler virus, HSV, poliovirus, RABV, and West Nile virus, utilize microtubule-dependent axonal transport to reach their target sites [59]. Given that neurotropic viruses can induce peripheral neuronal damage and potentially influence infection efficiency via peripheral routes [60], we established a viral encephalitis model through intracranial injection. This approach enabled more precise evaluation of SARM1’s role in disease pathogenesis.

While SARM1 is recognized as a central mediator of virus-induced axonal degeneration, the upstream viral factors that trigger its activation remain incompletely understood. Emerging evidence suggests that viral envelope components, such as the RABV matrix protein, may initiate SARM1 activation by disrupting mitochondrial energy metabolism [20]. This is exemplified by the interaction between the RABV matrix protein and mitochondrial Slc25a4, which induces ATP and NAD⁺ depletion, activates calcium-dependent proteases, and ultimately drives axonal fragmentation, a pathological cascade absents in furious form of RABV strains [22, 24, 41, 61,62,63]. This differential pathogenicity highlights the need to further investigate how various neurotropic viruses contribute to SARM1-mediated axonal degeneration, NAD⁺ depletion, and the broader mechanisms underlying viral encephalitis. Our study demonstrates that JEV, HSV-1, and RABV infections promote SARM1 accumulation in mitochondria, inducing depolarization, calcium dyshomeostasis, and cristae loss in cortical neurons. However, these effects were reversed by SARM1 inhibition or knockout, consistent with observations in LACV infection and NLRP3 inflammasome activator stimulation [18, 64], suggesting SARM1-mediated mitochondrial dysfunction represents a conserved stress response. Mechanistically, SARM1 activation depletes cytosolic NAD⁺, a crucial molecule for mitochondrial function. Although mitochondria possess the machinery to synthesize their own NAD⁺ through NMNAT3, the active import of NAD⁺ from the cytoplasm via the transporter SLC25A51 remains essential [65]. This dual dependency indicates that cytosolic NAD⁺ depletion directly impacts the mitochondrial NAD⁺ pool, thereby disrupting NAD⁺-dependent mitochondrial functions, particularly oxidative phosphorylation and ATP production. Furthermore, SARM1 activation can rapidly deplete ATP by inhibiting glycolysis, which requires sufficient NAD⁺ stores [66]. Notably, glycolysis serves as the primary ATP source for fast axonal transport [66, 67]. Thus, ATP depletion impairs molecular motors, leading to compromised mitochondrial migration and RABV-induced axonal mitochondrial aggregation [68]. These findings underscore the need to elucidate SARM1 mitochondrial targeting mechanisms across various neuronal injury contexts.

SARM1 regulates both neurological symptoms and brain inflammation beyond its canonical role in axonal degeneration [18, 20, 69,70,71]. Consistent with this broader functionality, SARM1 deletion significantly improved both survival outcomes and neurological manifestations in mice infected with various neurotropic viruses, including reduced JEV-induced tremors, attenuated HSV-1-associated seizures, and improved motor function in RABV infection (CVS11 strain). Sarm1 deletion significantly delayed the progression of hindlimb paralysis, with mutant mice retaining partial mobility (e.g., standing attempts) at late infection stages. However, the absence of coordinated hindlimb movements indicates that Sarm1 deletion confers substantial but incomplete protection against virus-induced neural injury. Regarding brain inflammation, our findings corroborate previous research linking mitochondrial SARM1-MAVS interaction to brain inflammation and neuronal death in LACV infection [18]. Extending this mechanism, we demonstrate that SARM1-driven brain inflammation critically contributes to JEV- and HSV-1-induced neuropathology independent of viral replication. Previous studies have demonstrated that neuronal SARM1 induces inflammatory cytokines and chemokines production, potentially orchestrating glial cell-mediated brain inflammatory responses [26, 72]. Our observations of decreased proinflammatory cytokines levels (IFN-γ, TNF-α, and IL-6) in Sarm1-deficient viral encephalitis models further support this mechanism. The observed anti-inflammatory effects may stem from direct suppression of glial activation, secondary mitigation of axonal degeneration, or both mechanisms simultaneously, as SARM1 could concurrently drive axonopathy and initiate inflammation via mitochondrial dysfunction (e.g., MAVS signaling). Elucidating this mechanistic duality requires further investigation into spatiotemporal regulation of SARM1-dependent glial activation and axonal pathology across viral models.

Recent advances have established SARM1 as a compelling therapeutic target in neurological disorders, given its critical role in axonal degeneration and brain inflammation [19, 72,73,74]. Our study demonstrates that two mechanistically distinct SARM1 inhibitors, FK866 (a NAMPT-mediated feedback inhibitor) and SIC4 (a direct NADase blocker), effectively suppress virus-induced axonal degeneration, brain inflammation, and cortical damage [35, 75]. Additionally, both inhibitors reduced cerebral infiltration of peripheral CD45⁺ immune cells and attenuated astrocyte proliferation/activation, while showing no direct impact on microglial activity. Notably, astrocytes, which critically regulate neuroimmune responses by supporting neuronal function and orchestrating immunity during injury or infection, may be modulated by SARM1 to indirectly influence brain inflammation and antiviral defense [19, 31, 72]. This observation aligns with previous findings demonstrating that Sarm1 depletion in traumatic brain injury models attenuates astrocytes activation and enhances functional recovery [76]. While SARM1 is absent in microglia [19, 26], its expression in neurons and astrocytes suggests that FK866 and SIC4 treatment-mediated neuroprotection may indirectly impact microglia through cellular cross-talk mechanisms. Hence, targeting SARM1 with FK866 or SIC4 offers a promising strategy to mitigate brain inflammation and neuronal damage, with astrocyte-mediated immune regulation emerging as a pivotal pathway. Further studies should clarify how SARM1 modulates astrocyte-microglia interactions in viral infection and/or neuronal injury contexts. Currently, research on the specific role of SARM1 in astrocytes remains in its early stages. Future experiments could utilize SARM1 astrocyte-conditional knockout or overexpression mice, combined with viral infection or brain inflammation models, to provide critical insights into its functional mechanisms. Integration of advanced techniques including single-cell sequencing, spatial transcriptomics, and proteomics in these models could further clarify SARM1’s contributions to astrocyte-driven neuroimmune regulation. Such studies would not only illuminate the potential role of SARM1 in astrocytes but potentially identify novel targets for treating neurological disorders induced by brain inflammation or viral infection.

In summary, our findings demonstrate that both genetic deletion and pharmacological inhibition of SARM1 provide substantial protection against neurotropic virus-induced neuropathology through multiple mechanisms. Upon viral invasion of the CNS, SARM1 drives axonal degeneration, mitochondrial dysfunction, and inflammatory cell infiltration into the cortex. The marked survival benefits observed in SARM1-inhibited or -deleted viral encephalitis models highlight its pivotal role in disease pathogenesis. Importantly, the conservation of SARM1-dependent pathways across various neurological disorders characterized by mitochondrial dysfunction suggests broader therapeutic applications beyond viral encephalitis. Our validation of SARM1 inhibitors, specifically FK866 and SIC4, represents a significant therapeutic advance, offering promising treatment strategies for both viral encephalitis and other neuronal injury diseases.

Conclusion

In conclusion, our study identified neurotropic virus infections result in a reduction of NAD⁺ levels in mouse brain tissue and trigger the activation of SARM1. The activated SARM1 leads to mitochondrial depolarization and axonal degeneration. Deletion or inhibition of SARM1 maintains mitochondrial homeostasis and axonal morphology, alleviates neuropathological damage, and improves survival outcomes in mice with viral encephalitis. This work might have therapeutic implications for neurotropic viral infections.