Involvement of RIG-I Pathway in Neurotropic Virus-Induced Acute Flaccid Paralysis and Subsequent Spinal Motor Neuron Death

ABSTRACT Poliomyelitis-like illness is a common clinical manifestation of neurotropic viral infections. Functional loss and death of motor neurons often lead to reduced muscle tone and paralysis, causing persistent motor sequelae among disease survivors. Despite several reports demonstrating the molecular basis of encephalopathy, the pathogenesis behind virus-induced flaccid paralysis remained largely unknown. The present study for the first time aims to elucidate the mechanism responsible for limb paralysis by studying clinical isolates of Japanese encephalitis virus (JEV) and Chandipura virus (CHPV) responsible for causing acute flaccid paralysis (AFP) in vast regions of Southeast Asia and the Indian subcontinent. An experimental model for studying virus-induced AFP was generated by intraperitoneal injection of 10-day-old BALB/c mice. Progressive decline in motor performance of infected animals was observed, with paralysis being correlated with death of motor neurons (MNs). Furthermore, we demonstrated that upon infection, MNs undergo an extrinsic apoptotic pathway in a RIG-I-dependent fashion via transcription factors pIRF-3 and pIRF-7. Both gene-silencing experiments using specific RIG-I-short interfering RNA and in vivo morpholino abrogated cellular apoptosis, validating the important role of pattern recognition receptor (PRR) RIG-I in MN death. Hence, from our experimental observations, we hypothesize that host innate response plays a significant role in deterioration of motor functioning upon neurotropic virus infections.

understand the etiology of virus-induced AFP further limits the development of potential therapies.
Like polio, many emerging viral infections, such as enterovirus, West Nile virus (WNV), and Japanese encephalitis virus (JEV), are causing AFP outbreaks annually due to endemic circulation of virus among populations (17)(18)(19)(20)(21)(22). JEV and CHPV are two such viruses, endemic to Southeast Asia and the Indian subcontinent, which are mainly responsible for causing AFP and encephalitis in young children (23). Patients suffering from JEV and CHPV principally develop flu-like symptoms and characteristic features of lower motor neuron (MN) disease, such as lower limb wastage and flaccid weakness in affected populations (24,25). Historically, the first clinical case of JEV-induced AFP was identified in 1945 on Okinawa Island, where the native population presented features of flaccid paralysis with sudden onset of fever and headache. When observed microscopically, autopsy samples of JE-positive patients displayed damaged anterior horn cells that were histologically similar to the lesions observed in acute poliomyelitis disease (26). Extensive electrophysiological and radiological studies performed on conscious JE-infected patients further suggested a role of anterior horn cells in JEVinduced muscle wasting (27). Additionally, magnetic resonance imaging (MRI) scans showed abnormal signal intensity on T2-weighted images of spine, indicating anterior horn cell injury in convalescent patients (28). Consistent with the clinical case reports, the anatomical basis of virus-induced AFP was first demonstrated in mice after neuroadapted Sindbis virus infection (29). Later, various studies conducted showed evidence of apoptotic injury in mouse central nervous system (CNS) upon EV71, WNV, reovirus, and Sindbis virus infection (30)(31)(32)(33)(34); however, the precise mechanism behind MN death and virus-induced AFP remained elusive.
Research in our laboratory previously defined a role of oxidative stress and neuroinflammation in brain, which answered some of the basic questions of virus-induced encephalitis (35)(36)(37)(38)(39)(40). Therefore, in this study, we developed an experimental model of virus-induced AFP that replicates all features of clinical AFP. Interestingly, we found that viral replication in lumbar cord correlates with behavioral impairments and paralysis in diseased mice, with RIG-I signaling playing a central role in activating the extrinsic apoptotic pathway. Thus, overall, we addressed two important questions of viral AFP. First, we described that sustained activation of RIG-I/pIRF3-7 signaling upon infection triggers the host endogenous apoptotic pathway, which, being a part of the antiviral response system, leads to MN death. Second, our investigations further highlighted that RIG-I-induced MN death during neurotropic virus infection is independent and distinct from well-described type 1 interferon (IFN)-induced apoptosis.

RESULTS
Characterization of neurotropic virus-induced AFP. In this study, we screened two clinical isolates of neurotropic virus, JEV (strain GP-78) and CHPV (strain 1653514), for their ability to induce flaccid paralysis in BALB/c mice (Fig. 1A). Both viral strains induced AFP by day 5 postinfection (pi) for JEV and day 2 pi for CHPV in mice. The degree of virus-induced paralysis was accurately determined by performing series of behavioral tests. Animals were scored on a scale of 0 to 3 to study general locomotion in mice (GLB score) (Fig. 1B), ranging from the ability to walk freely to no movement at all. Postinfection, both experimental groups showed significant differences in GLB score compared to respective age-matched controls. For JEV, 3/5 animals studied displayed hind leg inhibition with severe limp at day 5 of infection, while 2/5 animals exhibited features of tiptoe or knuckle walking. With disease progression (7 days pi [dpi]), 4/5 animals in the JEV group exhibited complete paralysis of hindlimbs (GLB score = 3) with zero movement, while 1/5 displayed severe limping. Akin to the JEV group, mice infected with CHPV also showed significantly higher GLB scores than mock-infected animals. For CHPV, 3/5 mice developed hind leg inhibition with severe limp shortly after infection (2 dpi), while 4/5 mice showed complete paralysis of lower limbs by day 3 pi. To further assess MN degeneration, hindlimb clasping score (HLC) ( Fig. 1C) was calculated, wherein the infected group again displayed significantly higher scores than their control littermates. Interestingly, from both the JEV and CHPV experimental groups, 2/5 animals developed partial retraction of hindlimbs toward midline when suspended freely in air, while with progression of infection, 4/5 animals in the JEV and 3/5 animals in the CHPV group developed complete retraction of hindlimbs (HLC score, 3). In a further attempt to analyze gait abnormalities, we did footprint analysis, wherein significant decrease in stride length, sway length, and stance length were recorded compared with mock-infected animals (Fig. 1D).
To further correlate clinical observations of AFP with motor neuron pathology, we performed Nissl staining using spinal cord (SC) sections. Selective loss of motor neurons was observed exclusively in anterior horn of the lumbar region compared with cervical and thoracic SC sections of the same animal (data not shown). Both JEV-and CHPV-infected lumbar sections displayed features of chromatolysis, like eccentric nucleus and cellular swelling of MN ( Fig. 2B and D), compared to healthy MNs observed in mock-infected mice ( Fig. 2A and C). Thus, collectively these findings suggest that virus infection not only impaired overall motor coordination but also induced clinical AFP with degenerating motor neurons.
Virus titer in lumbar SC corresponds to disease progression. To determine whether active replication of virus is associated with disease severity, we next performed plaque  hindlimb paralysis appeared. Pups were sacrificed and lumbar cord tissue was harvested at days 1, 3, 5, and 7 postinfection (pi) for JEV and days 1, 2, and 3 pi for CHPV. Blue bar charts represent JEV infection and orange bar charts represent CHPV infection. (B) Characteristic changes in locomotion of infected animals with representative posture of duck foot motion and hindlimb paralysis were scored quantitatively and analyzed as GLB (general locomotory behavior) score. Data are represented as means 6 standard deviations (SD) for 5 animals per group where P values were determined (*, P , 0.05; **, P , 0.01; ***, P , 0.001) using one-way ANOVA followed by post hoc Bonferroni test. (C) Altered hindlimb clasping phenotype with typical posture of hindlimbs retracted toward the midline or away from midline was scored as HLC (hindlimb clasping) score where data are represented as means 6 SD for 5 animals per group, with statistical significance being determined (*, P , 0.05; **, P , 0.01; ***, P , 0.001) using one-way ANOVA followed by post hoc Bonferroni correction. (D) Severity in gait of the animals was quantified by footprinting analysis where error bars are represented as mean 6 SD for 5 animals, with P values determined (*, P , 0.05; **, P , 0.01; ***, P , 0.001) using two-tailed paired t test. assays using lumbar cord homogenates at various dpi. To our interest, infectious viral particles were detected only at day 5 and 7 pi for JEV and day 2 and 3 pi for CHPV, suggesting inability of virus to penetrate the blood-brain barrier (BBB) at early dpi ( Fig. 3A and C). Additionally, to confirm in vivo viral replication, two-step quantitative PCRs targeting glycoprotein gene of JEV and P, N gene of CHPV were performed to detect viral RNA extracted from lumbar cords at various time points pi. Significant upregulation of GP-78 ( Fig. 3B) at days 5 and 7 pi for JEV and P, N gene ( Fig. 3D) at days 2 and 3 pi for CHPV were detected. Overall, results from the quantitative real-time PCR (qRT-PCR) analysis and plaque assay paralleled the behavioral deficits observed in mice, indicating an association between viral replication and paralysis onset.
JEV and CHPV infect motor neuron both in vivo and in vitro. To determine whether virus-induced AFP is linked to active infection of MNs, we performed immunohistochemistry (IHC) with lumbar sections using anti-JEV NS3 and anti-CHPV P-protein antibody. Motor neurons were identified using anti-SMI32, a specific marker for anterior horn cells, which appeared intact with no immunoreactivity toward viral antigen in mock-infected samples. Both JEV-and CHPV-infected pups displayed clear colocalization of SMI32 and NS3 and P-protein at days 5 and 7 pi for JEV and days 2 and 3 pi for CHPV ( Fig. 4A and D), with quantification graphs attached ( Fig. 4C and F). Interestingly, viral protein kinetics (blots) showed significant upregulation of NS3 at day 7 pi for JEV and P-protein at day 3 pi for CHPV ( Fig. 4B and E), which is in line with IHC data and further corroborates the association of MN infection with AFP.
Having shown that viral replication in MN is associated with onset of paralysis, we next studied permissiveness of virus in NSC34 cells to further investigate molecular events contributing to AFP pathogenesis. For this, we infected NSC34 cells with JEV and CHPV at a multiplicity of infection (MOI) of 1 and 0.1, respectively, and harvested intraperitoneally. Viral titers were determined from infected lumbar cord tissue in a daywise manner (days 1, 3, 5, and 7 after JEV infection and days 1, 2, and 3 after CHPV infection) for studying viral kinetics with disease progression. Blue bar charts represent JEV-infected data and orange bar charts represent CHPV-infected data. (A and C) Lumbar tissue homogenates prepared and pooled from 2 mice for each time point of infection were analyzed for infectious virus count using plaque assay. Dashed lines represent the limit of detection of plaque assays for all experiments performed. (B and D) Total RNA was extracted from lumbar SC tissue for analyzing relative abundance of viral mRNA using qRT-PCR, where GAPDH was used as the loading control. Data represent mean 6 standard deviation (SD) from a minimum of 3 independent experiments, where P values were determined (*, P , 0.05; **, P , 0.01; ***, P , 0.001) using one-way ANOVA followed by post hoc Bonferroni test.
Acute Flaccid Paralysis and Spinal Motor Neuron Death ® samples at various hours postinfection (hpi) to study time-dependent kinetics of viral replication. Phase-contrast images were captured for monitoring cellular health through brief periods of infection, where NSC34 cells displayed cytopathic changes characterized by rounding up of cells by 12 and 24 hpi for JEV and 6 and 12 hpi for CHPV (see Fig. S1 in the supplemental material). To confirm active replication of virus in NSC34 cells, we performed immunostaining with virus-specific antibodies. No signal was detected in mock-infected cells, while with progression of infection, increased viral staining was observed both in JEV (Fig. 5A, v)-and CHPV (Fig. 5B, v)-infected NSC34 cells. In addition to this, terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL)-positive cells were also observed at 24 hpi for JEV (Fig. 5A, vi) and 12 hpi for CHPV (Fig. 5B, vi), suggesting a correlation between apoptotic death of MN and flaccid paralysis.
Next, to study time-dependent kinetics of viral replication, we performed qRT-PCR analysis at various hpi. Significant upregulation of viral copies was evident at 6, 12, and 24 hpi (Fig. 5A, ii), an observation in parallel with time-dependent reduction of viability from 100% in control to 63.08% in 24-h JEV-infected NSC34 cells (Fig. 5A, i). In contrast, infection with CHPV reduced NSC34 viability from 100% to 56% (Fig. 5B, ii) within 12 FIG 4 Virus-induced AFP correlates with spread of viral antigen in lumbar motor neurons. Ten-day-old BALB/c mice were either mock infected or inoculated intraperitoneally with 3 Â 10 4 PFU of JEV or 1 Â 10 4 PFU of CHPV. Infected cord tissues were collected in a daywise manner as indicated for studying paralysis onset and disease progression. Blue bar charts represent JEV infected data, and orange bar charts represent CHPV infected data. (A and D) Lumbar cord sections were double stained for motor neuron marker SMI-32 and viral antigen NS3 for JEV and P-protein for CHPV infection. Representative epifluorescence images show colocalization of viral antigen with MN marker and DAPI, indicated by white arrows in respective panels. Scale bar denotes 50 mm. (B and E) Lumbar cord lysates from both PBS-injected and virus-infected pups were analyzed using Western blotting to determine relative abundance of viral proteins (NS3 for JEV and P-protein for CHPV). (C and F) Representative graphs showing percentage of JEV-and CHPV-infected cells along with double-positive cells for SMI32 (motor neuron marker) and viral protein markers in respective lumbar cord sections. Data here represent mean 6 SD from a minimum of 3 independent experiments, where statistical significance (*, P , 0.05; **, P , 0.01; ***, P , 0.001; NS, nonsignificant) was calculated using one-way ANOVA followed by post hoc Bonferroni correction.
Bhaskar et al. . Infectious viral particles were detected readily in supernatants obtained upon infection from both viruses studied ( Fig. 5A and B, iv) with an increased expression of viral protein NS3 in JEV-and P in CHPV-infected cells ( Fig. 5A and B, iii). Additionally, cytokine bead array (CBA) analysis of culture supernatant obtained upon infection further substantiated that seeded NSC34 cells were dying due to active replication of virus and not because of neuroinflammation (Fig. S2). JEV and CHPV infection induces caspase-dependent extrinsic apoptosis in MN. Having shown that viral replication leads to apoptosis in vitro, we next wanted to determine which apoptotic pathway is activated in dying MNs. For this, we examined expression levels of various caspases, including caspase-8, caspase-9, and caspase-3, with their associated proteins BAX, BCL2, and c-PARP postinfection. Both in vivo and in vitro studies revealed significant overexpression of cleaved (c)-casp-8, c-casp-3, and c-PARP protein with no active form of caspase-9 being detected. Distinct upregulation of c-casp8 was observed in NSC34 cells at 12 hpi for JEV and 6 hpi for CHPV, which later declined significantly at 24 hpi for JEV and 12 hpi for CHPV (Fig. 6A, ii). Interestingly, a similar pattern of c-casp8 was observed in vivo at day 5 pi for JEV and day 2 pi for CHPV. In addition to this, the active form of casp-3 was detected at 24 hpi in vitro and day 7 pi in vivo upon JEV and 12 hpi in vitro and day 3 pi in vivo upon CHPV treatment (Fig. 6A). Further, to validate that MNs were dying via caspase-dependent pathways only, we performed experiments with Z-VAD-FMK, a pancaspase inhibitor. Pretreatment with Z-VAD-FMK attenuated expression levels of cleaved caspases and significantly reduced apoptosis in NSC34 cells compared with respective virus-infected sample ( Fig. 6B and C, i). Cytosolic casp3/7 activity was greatly reduced in Z-VAD-treated cells compared with untreated samples ( Fig. 6B and C, iv). Plaque assays further confirmed that reduced c-casp-3 in Z-VAD-FMK-treated cells was purely due to inhibition of apoptosis and not because of compromised viral replication ( Fig. 6B and C, ii and iii).
JEV and CHPV infection activates RIG-I pathway. Next, we studied expression level of RIG-I, a well-described PRR involved in initiating antiviral signaling upon infection, both in infected lumbar tissue and NSC34 cells. Immunoblot analysis showed significant upregulation of RIG-I at days 5 and 7 pi for JEV and days 2 and 3 pi for CHPV in mice, with similar trends in infected NSC34 cells, where marked upregulation of RIG-I was evident by 6 hpi for JEV and 3 hpi for CHPV (Fig. 7A, i and ii). A transient upregulation of phosphorylated IRF3 and pIRF7, critical markers of the RIG-I/interferon signaling axis, was detected at 5 dpi for JEV and 2 dpi for CHPV that eventually declined at later days of infection. Consistent with our in vivo data, infected NSC34 cells showed similar trends of pIRF3 and pIRF7 at 12 hpi for JEV and 6 hpi for CHPV that declined over time. Activation of type 1 interferon signaling was confirmed by qRT-PCR analysis of IFN-a and IFN-b, both in infected lumbar tissue and NSC34 cells by day 5 pi in vivo and 6 hpi in vitro upon JEV and day 2 pi in vivo and 3 hpi in vitro upon CHPV (Fig. 7B, i and ii).
Thereafter, we employed short interfering RNA (siRNA)-mediated silencing to knock down RIG-I expression to validate its role in virus-induced caspase-dependent death of MNs. Gene silencing using specific RIG-I (Ddx58) esiRNA significantly downregulated charts represent CHPV infection data. (A and B, i) Percent cell viability of NSC34 cells after JEV and CHPV infection quantified at various time points using WTS assay. (ii) Total RNA was extracted from both mock and experimental NSC34 cells following JEV and CHPV infection at the indicated time points. Relative abundance of viral transcripts was determined using qRT-PCR analysis as fold change, where GAPDH was used as the loading control. (iii) Cell lysates prepared from both mock-infected and virus-infected NSC34 cells were subjected to Western blot analysis at the indicated time points for studying viral protein kinetics. Densitometric analysis was performed with b-actin as a loading control to determine fold change expression of indicated proteins. (iv) Viral titer in NSC34 culture supernatant was determined using plaque assay at indicated time points after JEV and CHPV infection. Dashed lines represent the limit of detection of plaque assays for all experiments conducted. (v) Representative immunostaining of NSC34 cells with antibodies against viral protein NS3 (JEV) and P-protein (CHPV) with nuclear marker DAPI. White arrows show populations of infected NSC34 cells. Scale bar denotes 100 mm. (vi) Representative TUNEL staining of JEV and CHPV at late time points of infection with DAPI stained in blue. Scale bar denotes 100 mm. Here, data represent mean 6 SD from a minimum of 3 independent experiments, where statistical significance (*, P , 0.05; **, P , 0.01; ***, P , 0.001; NS, nonsignificant) was calculated using one-way ANOVA followed by post hoc Bonferroni correction. cleaved caspases and rescued apoptosis at 12 and 24 hpi for JEV and 6 and 12 hpi for CHPV ( Fig. 8A and B) with survival of live-NSC34 cells (Fig. S5). However, no changes in viral protein expression ( Fig. 8A and B, ii) were detected, indicating that downregulation of apoptotic effector proteins upon gene silencing were not due to impaired viral replication. In addition to this, we performed qRT-PCR analysis and plaque assays to further substantiate our findings, which, in parallel to our observations, showed no significant differences in viral replication ( Fig. 8A and B, iii and iv). There was significant decrease in pIRF3 and pIRF7 expression at 12 hpi for JEV and 6 hpi for CHPV in RIG-I-transfected NSC34 compared with the respective negative control. In line with immunoblot analysis, siRNA-mediated RIG-I silencing downregulated type 1 interferon production in both JEV-and CHPV-infected cells (Fig. S3), indicating that upon infection both viruses activate the RIG-I/pIRF3-7/interferon axis to initiate virus-induced apoptosis.
RIG-I-mediated apoptosis is independent of type 1 interferon pathway. Having shown that siRNA-mediated RIG-I silencing downregulated IFN expression and rescued MN death, we next determined the direct role of IFN signaling in virus-induced apoptosis. To address this, we performed two experiments, first with interferon-responsive cell line NSC34, where the receptor antibody neutralization assay was carried out to block any downstream IFN signaling, and second with the interferon-unresponsive cell line Vero. Surprisingly, in both experiments, JEV and CHPV led to apoptosis of NSC34 and Vero cells with distinct upregulation of c-casp3 and c-PARP proteins ( Fig. 9A to C). NSC34 cells upon infection displayed no significant changes in the active form of casp-3 and viral protein expression for all three conditions studied ( Fig. 9A and B, ii). Similar to NSC34 cells, Vero cells displayed cytopathic changes like rounding up of cells and refraction upon infection (data not shown), suggesting that both viruses induce apoptosis independent of host IFN signaling.
RIG-I deficiency suppresses viral replication in vivo. To explore whether RIG-I knockdown in mouse SC altered the behavioral outcome of flaccid paralysis and whether such changes were associated with reduction in virus-induced apoptosis, we infected RIG-I morpholino-treated (RIG-I-Mo) BALB/c mice with virus (Fig. 10A). Immunoblot analysis from RIG-I-Mo mice showed effective RIG-I silencing compared with negative controltreated morpholino (NC-Mo) and only virus-infected animals (Fig. 10B). Significant upregulation of c-casp3 and c-PARP was detected in both only virus-infected (A3, A4) and infected NC-Mo mice (A5, A6), while a lumbar isolate of infected RIG-I-Mo (A7, A8) animals showed drastic reduction in c-casp3 and c-PARP expression. Notably, both JEV and CHPVinfected RIG-I-Mo mice displayed remarkable reduction in viral protein expression compared with respective NC-Mo and only infected animals, suggesting an important role of the RIG-I/pIRF3-7 axis in viral apoptosis and replication in vivo.

DISCUSSION
In this study, we propose an AFP mouse model with detailed characterization of molecular pathways that lead to paralysis and spinal MN death. Here, we have used clinical isolates of two different neurotropic viruses belonging to two different families,  namely, Flaviviridae and Rhabdoviridae, for inducing paralysis in mice. When injected intraperitoneally, both viruses successfully induced AFP with consistent features of abnormal posture, hindlimb clasping, and gait abnormalities. This observation is in line with retrospective studies performed with JE-infected patients who presented prodromal features of nonspecific illness, which rapidly progressed to flaccid paralysis both in JE conscious and convalescent patients (41,42).
Since lower-limb paralysis is a more commonly reported symptom of neurotropic virus infections, we mainly restricted our focus to the lumbar segment of SC. Both JEVand CHPV-infected lumbar sections displayed features of chromatolysis, with increased viral replication in infected tissue with time. Consistent with these findings, paralyzed mice showed direct colocalization of SMI32 with virus-specific antigens, suggesting a strong association between paralysis onset and virus replication. Additional support came from studies with JE-conscious patients who displayed damaged anterior horn  (27,43). Since it is difficult to delineate complex signaling pathways leading to MN death in animal models, we planned to perform experiments employing a reductionist model comprised of infection of motor neuron cell line NSC34 with JEV and CHPV. Together, findings from our in vivo and in vitro experiments led us to conclude that direct viral injury is responsible for paralytic disease and MN death upon infection. Numerous reports previously demonstrated the role of apoptosis in pathogenesis of lethal encephalitis (30,31,34,44); however, no report has yet illustrated the role of spinal MN death in virus-induced AFP. In this study, we reported conclusively that upon infection MN undergoes extrinsic apoptotic signaling via activation of the RIG-I/ pIRF3-7 axis. Essentially, we demonstrated that RIG-I-transfected-NSC34 cells abrogated apoptosis, with no significant activation of other viral PAMP expression (see Fig. S4 in the supplemental material), suggesting that RIG-I, not the TLR system, plays a critical role in MN apoptosis. Interestingly, RIG-I ablation in NSC34 cells reduced interferon expression upon infection, reinforcing the notion that IFN signaling is activating virusinduced apoptosis of MNs. However, receptor-blocking experiments with IFNAR antibody clearly suggested that virus-induced apoptosis in NSC34 cells is independent of interferon signaling. It is possible that RIG-I activation stimulates the production of pIRF3-induced IFN-independent genes (likely ISG54 and ISG56) that eventually lead to infection-elicited apoptosis (45,46). Nonetheless, it needs to be highlighted that it is still uncertain how IFN-independent genes orchestrate the induction of apoptosis during infection in MNs.
Surprisingly, with our in vivo model, we made contrasting observations where CNS ablation of RIG-I significantly enhanced resistance to infection, clearly visible by reduced expression of viral proteins and apoptotic markers rather than increased susceptibility to infection, as reported previously (46,47). Notably, in our paradigm, transient ablation of RIG-I in CNS using vivo-morpholino attenuated both antiviral and inflammatory arms of innate immune response, which may have interfered with replication of virus in SC compared to the control-treated morpholino animals, which displayed no change in virus protein expression. A plausible explanation for our interesting in vivo observation could be the decreased permeability of the BBB for systemic virus due to localized ablation of RIG-I in brain and SC. Although our in vivo data do not strongly support whether attenuated cellular death in RIG-I-knockdown animals is mediated directly via type 1 interferon response and/or indirectly via host cytokine response upon infection, our results provide one clear explanation as to how transient silencing of RIG-I in motor neuron NSC34 cells attenuated apoptosis by suppressing pIRF3/7 signaling effectively by blocking p-IRF3-induced IFN-independent genes (likely ISG54 and ISG56) and not any other IFN-dependent pathways (interferon-b production and IFNAR signaling).
In conclusion, the central finding of this paper is that upon infection, MNs induce RIG-I-dependent activation of apoptosis and type 1 interferon pathways, although upon infection the host cell remains resistant to antiviral effects of IFNs and undergoes apoptosis by following IFN-independent pathways.

MATERIALS AND METHODS
Viruses. The GP78 strain of JEV was propagated in suckling BALB/c mice (48), while CHPV strain 1653514, a kind gift by Dhrubajyoti Chattopadhyay (Amity University, Kolkata, India), was propagated in the Vero E6 cell line (35). Both viral preparations were analyzed briefly for determining number of PFU using plaque assay as described earlier (49). Mouse experiments and behavioral scoring. All studies performed with animals were approved by the Animal Ethics Committee of National Brain Research Centre (approval no. NBRC/IAEC/2017/130) and were in accordance with the guidelines of the Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA), Ministry of Environment and Forestry, Government of India. BALB/c mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and were housed at the pathogenfree and climate-controlled animal facility of the NBRC. Viral infections were performed on 10-day-old litters that were always housed with their mothers for milk feeding. Pups of either sex were inoculated intraperitoneally with 3 Â 10 4 PFU of JEV or 10 4 PFU of CHPV in a 25-ml volume. Mice were examined twice Densitometric analysis was performed using FIJI to determine fold change in expression of viral proteins and cleaved-casp-3 in mock-infected or virus-infected NSC34 cells with or without treatment with IFNAR or IgG control antibody. The data are represented as mean 6 SD from a minimum of 3 independent experiments, where P values were calculated (*, P , 0.05; **, P , 0.01; ***, P , 0.001) using one-way ANOVA followed by post hoc Bonferroni correction.
daily, and behavioral experiments were performed once a day until clinical features of paralysis appeared (Table 1).
Tissue processing and IHC. For lumbar cord isolation, pups were deeply anaesthetized with ketamine and perfused transcardially using phosphate-buffered saline (PBS). Mice were sacrificed at days 1, 3, 5, and 7 of JEV and days 1, 2, and 3 of CHPV infection. Tissue samples obtained were used for RNA and protein isolation; however, for immunohistochemistry (IHC) and Nissl staining, pups were transcardially fixed with paraformaldehyde (PFA) solution. Isolated lumbar cords were postfixed in PFA overnight, followed by cryopreservation in 30% sucrose until saturated. Serial sections of 20-mm thickness  a For evaluating GLB score, experimental pups were allowed to move freely in an open box having dimensions of 35 by 15 cm and were scored 0, 1, 2, and 3, representing none, moderate, marked, and extreme magnitude of locomotion deficits. b For evaluating HLC score, mice were grasped from the base of the tail and suspended briefly in the air for 15 s, and pups were scored 0, 1, 2, and 3 on the basis of the position of hindlimbs. c For the footprinting test, hind paws of mice were painted in nontoxic dye and pups were motivated to walk on narrow lanes lined with white paper. For analysis and quantification, the first few and last few footmarks were excluded and only the middle portion of each run was evaluated.
Nissl staining. Lumbar sections were briefly rehydrated in Milli-Q and were subjected to cresyl violet staining. Sections were next immersed in differentiating solution (1:1 absolute ethanol and dioxane), followed by dehydration in dioxane (Merck-Millipore, Sigma) and tissue clearance in xylene (Merck-Millipore, Sigma). Finally, slides were mounted with DPX (dibutylpthalate polystere xylene) (Sigma) and were dried for 24 h in dark boxes. Images were captured on a Leica DMRXA2 microscope at Â5 and Â40 oil magnifications. Cell viability assay. Viability of cultured cells was determined using cell proliferation reagent WST1 (Roche, Sigma) per the manufacturer's instructions. NSC34 cells were seeded at a density of 2 Â 10 4 cells per well in separate 96-well plates, followed by infection at said MOIs. Twenty microliters of WST1 solution was added to each well after 6, 12, and 24 hpi for JEV and 3, 6, and 12 hpi for CHPV. Absorbance was recorded at 490 nm using a Multiskan Sky spectrophotometer (Thermo Fischer), reflecting the formation of formazan product by metabolically active cells. Results were expressed as relative percentages of virus-infected cells to that of mock-infected cells.
Transfection procedures and siRNA knockdown. Transient transfection of NSC34 cells was performed using Lipofectamine RNAiMAX reagent (Invitrogen, USA) per the manufacturer's instructions. Briefly, cells were cultured in DMEM until they attained a confluence of 60 to 70%, after which cells were transfected with 30 pmol siRNA specific to RIG-I (Ddx58) and a negative-control enhanced green fluorescent protein (eGFP). Transfection was performed in Opti-MEM for 8 h, after which cells were maintained in 5% DMEM for 16 h. Cells were then serum starved for an additional 2 h before being treated with virus as stated earlier. Postinfection, samples were harvested and supernatants were collected at different time points for various biochemical assays.
Live/dead assay. NSC34 cells were briefly seeded in 12-well plates and transfected upon reaching confluence. Posttransfection, cells were treated with virus at said MOIs for 12 and 24 hpi for JEV and 6 and 12 hpi for CHPV. NSC34 cells were then incubated with LIVE/DEAD reagent (Invitrogen, Thermo Fischer) at a concentration of 4 mM ethidium dimer-D1 and 2 mM Calcein-AM in the dark. Images were captured and analyzed using a Zeiss Axio Vert.1 microscope at Â10 magnification.
Immunocytochemistry. Briefly, NSC34 cells were grown in 8-well chamber slides, followed by infection with JEV and CHPV at said MOIs. Postinfection, cells were fixed with 4% paraformaldehyde solution and were blocked using bovine serum albumin (Sigma). Cells were then incubated overnight with primary antibodies NS3 (1:500) and CHPV P protein (1:500) at 4°C, followed by extensive washing. Finally, cells were incubated with appropriate Alexa fluor-tagged secondary antibody and were mounted with DAPI. Images were captured on a Zeiss Apotome microscope at Â20 magnification.
RNA isolation and qRT-PCR analysis. Total cellular RNA was isolated from lumbar cord tissue and NSC34 cells using Tri reagent (Sigma-Aldrich, USA). Briefly, cDNA was synthesized using random hexamer primers (Verso cDNA synthesis kit), and quantitative real-time PCRs (qRT-PCRs) were performed using Power SYBR green (Applied Biosystems, USA) with gene-specific primers ( Table 2). Relative gene abundance for each reaction was determined using the delta threshold cycle method with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a loading control in a ViiA 7 real-time PCR system (Applied Biosystems). Results were expressed as fold differences between mock-and virus-infected samples.
TUNEL staining. TUNEL assay was carried out using Promega's DeadEnd fluorometric TUNEL system per the manufacturer's protocol. Briefly, NSC34 cells were seeded at a density of 2 Â 10 5 cells per well in 8-well chamber slides and infected with virus as described before. Postinfection, cells were fixed with PFA, followed by incubation with recombinant terminal deoxynucleotidyl transferase (rTdT) enzyme mix. Later, cells were washed extensively and mounted with Vectashield mounting medium and DAPI. Images were captured on a Zeiss Apotome at Â20 magnification.
Antibody neutralization assay. For antibody-based neutralization assay, NSC34 cells were seeded at a density of 5 Â 10 4 cells per well on 24-well plates. Prior to infection, cells were treated with 10 mg/ ml isotype IgG1 R (clone MOPC1; BioLegend) or purified anti-mouse IFNAR-1 (clone MARI-5A3; BioLegend) blocking antibody for 1 h. Cells were then washed with PBS once and infected with virus at said MOIs. Postinfection, cells were maintained in 5% DMEM containing 10 mg/ml neutralizing antibodies and samples were harvested at the indicated time points for immunoblotting.
RIG-I gene knockdown and mouse infection. For vivo-morpholino experiments, 10-day-old BALB/c pups were randomly assigned to four groups: group 1, mock infection; group 2, virus infection only; group 3, virus infection and negative-control vivo-morpholino; and group 4, virus infection and RIG-I vivo-morpholino. Starting a day prior to infection, mice belonging to groups 3 and 4 were injected intracranially with a single dose of RIG-I-specific vivo-morpholino (18.5 mg/kg of body weight; Gene Tools, USA) and negative control (18.5 mg/kg), respectively. After 24 h, all animals were infected intraperitoneally with either PBS or virus at day 0 of infection. Animals were then sacrificed at day 7 pi for JEV and day 3 pi for CHPV, and lumbar cords were harvested for immunoblot analysis.
Statistical analysis. Data were represented as means 6 standard deviations (SD) from a minimum of three independent experiments unless otherwise stated. GraphPad Prism 5 and KyPlot2.0 were used for data analysis and preparation of graphs. Time-course studies were analyzed using one-way analysis of variance (ANOVA) with Bonferroni post hoc test, whereas differences between two groups were evaluated using paired two-tailed Student's t test with 95% confidence. Any P value less than 0.05 was considered statistically significant.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.