Neuronal Ablation of Alpha/Beta Interferon (IFN-α/β) Signaling Exacerbates Central Nervous System Viral Dissemination and Impairs IFN-γ Responsiveness in Microglia/Macrophages

IFN-α/β induction limits CNS viral spread by establishing an antiviral state, but also promotes blood brain barrier integrity, adaptive immunity, and activation of microglia/macrophages. However, the extent to which glial or neuronal signaling contributes to these diverse IFN-α/β functions is poorly understood. Using a neurotropic mouse hepatitis virus encephalomyelitis model, this study demonstrated an essential role of IFN-α/β receptor 1 (IFNAR1) specifically in neurons to control virus spread, regulate IFN-γ signaling, and prevent acute mortality. The results support the notion that effective neuronal IFNAR1 signaling compensates for their low basal expression of genes in the IFN-α/β pathway compared to glia. The data further highlight the importance of tightly regulated communication between the IFN-α/β and IFN-γ signaling pathways to optimize antiviral IFN-γ activity.

ABSTRACT

Alpha/beta interferon (IFN-α/β) signaling through the IFN-α/β receptor (IFNAR) is essential to limit virus dissemination throughout the central nervous system (CNS) following many neurotropic virus infections. However, the distinct expression patterns of factors associated with the IFN-α/β pathway in different CNS resident cell populations implicate complex cooperative pathways in IFN-α/β induction and responsiveness. Here we show that mice devoid of IFNAR1 signaling in calcium/calmodulin-dependent protein kinase II alpha (CaMKIIα) expressing neurons (CaMKIIcre:IFNAR^fl/fl mice) infected with a mildly pathogenic neurotropic coronavirus (mouse hepatitis virus A59 strain [MHV-A59]) developed severe encephalomyelitis with hind-limb paralysis and succumbed within 7 days. Increased virus spread in CaMKIIcre:IFNAR^fl/fl mice compared to IFNAR^fl/fl mice affected neurons not only in the forebrain but also in the mid-hind brain and spinal cords but excluded the cerebellum. Infection was also increased in glia. The lack of viral control in CaMKIIcre:IFNAR^fl/fl relative to control mice coincided with sustained Cxcl1 and Ccl2 mRNAs but a decrease in mRNA levels of IFNα/β pathway genes as well as Il6, Tnf, and Il1β between days 4 and 6 postinfection (p.i.). T cell accumulation and IFN-γ production, an essential component of virus control, were not altered. However, IFN-γ responsiveness was impaired in microglia/macrophages irrespective of similar pSTAT1 nuclear translocation as in infected controls. The results reveal how perturbation of IFN-α/β signaling in neurons can worsen disease course and disrupt complex interactions between the IFN-α/β and IFN-γ pathways in achieving optimal antiviral responses.

IMPORTANCE IFN-α/β induction limits CNS viral spread by establishing an antiviral state, but also promotes blood brain barrier integrity, adaptive immunity, and activation of microglia/macrophages. However, the extent to which glial or neuronal signaling contributes to these diverse IFN-α/β functions is poorly understood. Using a neurotropic mouse hepatitis virus encephalomyelitis model, this study demonstrated an essential role of IFN-α/β receptor 1 (IFNAR1) specifically in neurons to control virus spread, regulate IFN-γ signaling, and prevent acute mortality. The results support the notion that effective neuronal IFNAR1 signaling compensates for their low basal expression of genes in the IFN-α/β pathway compared to glia. The data further highlight the importance of tightly regulated communication between the IFN-α/β and IFN-γ signaling pathways to optimize antiviral IFN-γ activity.

INTRODUCTION

Alpha/beta interferon (IFN-α/β) induction and signaling through the IFN-α/β receptor (IFNAR) are fundamental components of innate immunity to virus infections. IFN-α/β binding to the IFNAR induces a signaling cascade to activate numerous IFN-stimulated genes (ISGs), which encode pattern recognition receptors, antiviral mediators, immunomodulatory factors, transcription factors, and members of the IFN-α family. This amplifies the IFN-α/β response to establish an antiviral state (1, 2). IFN-α/β signaling also shapes the transition from innate to adaptive immune responses by promoting major histocompatibility complex (MHC) class I presentation and T cell activation and survival, as well as B cell responses (2–5). While IFN-α/β signaling limits viral spread, it is insufficient to reduce virus load in the absence of adaptive immune responses (6–8). Moreover, IFNAR signaling is tightly balanced by induction of ISGs encoding negative regulators to limit cytotoxic effects (9). The finding that different cell types are equipped with distinct homeostatic and inducible expression patterns of genes in the IFN-α/β pathway (10–14) has invigorated investigation into the interplay between initial IFN-α/β inducer and responder cells in various tissues.

Understanding how tissue resident cells contribute to protective IFN-α/β responses in the central nervous system (CNS) is especially important due to its composition of fully differentiated, nonregenerating postmitotic neurons and oligodendrocytes, as well as microglia and astrocytes dedicated to providing trophic and metabolic support. Although acute viral encephalitis is rarely fatal, resolution is associated with long-term neurological disabilities (15). While all CNS resident cells have the capacity to induce and respond to IFN-α/β (6, 7, 11, 16), differential regulation of IFN-α/β-related genes in distinct CNS cell types may determine infection outcome (13, 14). Microglia are early participants in innate immunity to many viral infections, but astrocytes have also emerged as potent IFN-β producers during primarily neuronotropic infections (6, 8, 17–20). Nonrenewable neurons comprise the most vulnerable target for many neurotropic viruses, including West Nile virus (WNV) and rabies virus (21, 22). Despite overall lower basal expression levels of genes in the IFN-α/β pathway in neurons relative to glia (14), they can be sufficient for early control of some viral infections (11). However, the few studies showing neuronal IFN-α/β production and signaling following viral and prion infections indicate that these responses are highly limited (23–27). Nevertheless, by responding to IFN-α/β, neurons may be primed to better sense infection and activate antiviral or proinflammatory chemokines (6). Region-specific innate responses in neurons may further determine susceptibility to viral infection (27). We thus sought to investigate the contribution of IFNAR1 signaling in controlling encephalomyelitis induced by the neurotropic mouse hepatitis virus A59 strain (MHV-A59) using mice with specific IFNAR1 abrogation in CaMKIIα-expressing neurons (CaMKIIcre:IFNAR^fl/fl mice). CaMKIIα is heavily expressed throughout the forebrain, with robust expression in excitatory neurons and in the superficial (II and III) and deep (V and VI) layers of the neocortex (28). Critical to this study, various mouse lines expressing distinct reporter proteins under the control of the CaMKIIα promoter revealed that CaMKII-expressing neurons are most abundant in the forebrain, including the cerebral cortex (CC) and hippocampus (HC), but very low in the brain stem (BS) and cerebellum (CB) (29, 30). An additional advantage is the developmentally regulated expression of CaMKIIα in postmitotic neurons (31). CaMKIIα transcripts do not emerge until day 1 postnatum and are robustly expressed only at the adult stage (31), consistent with no role in early development or morphological changes in CaMKIIα knockout (KO) mice (31, 32). IFNAR ablation restricted to postmitotic neurons in CaMKIIcre:IFNAR^fl/fl mice thereby limits alterations in IFN-α/β-dependent genes in developing mice under homeostatic conditions. This is important, as total IFNAR-deficient mice express up to 10-fold-lower basal levels of IFN-α/β-associated genes in brains and neonatal cultured neurons relative to IFNAR-sufficient mice (13).

MHV-A59 replicates in both glia and neurons following intracranial (i.c.) infection and causes benign encephalomyelitis in immunocompetent adult mice (33–35). While IFN-α/β is essential in limiting viral spread through the CNS, IFN-γ signaling is required for T cell-mediated control of infectious virus to undetectable levels (35–37). While microglia/macrophages initially produce IFN-α/β (20), astrocytes are delayed IFN-α/β producers in vivo (12). Here, we demonstrate that abrogation of IFNAR1 in CaMKIIα-expressing neurons results in acute mortality coincident with the inability to control viral replication in selected brain regions. Ineffective virus control could not be attributed to deficits in early global IFN-α/β responses, nuclear localization of phosphorylated signal transducer and activator of transcription 1 (pSTAT1), T cell trafficking, or IFN-γ production, but responsiveness to IFN-γ was impaired. Our findings demonstrate a crucial role for neuronal IFNAR1 in restricting viral replication in both neurons and glia, as well as in regulating communication between neurons and glia to optimize IFN-γ responses.

RESULTS

IFNAR signaling in neurons is required for viral control and host survival.

Based on MHV-A59 tropism for neurons, but inability to induce IFN-α/β in neurons in vitro (20), we ablated IFNAR1 in CaMKIIα-expressing neurons to assess the importance of IFNAR1 signaling in this cell type to overall viral control. MHV-A59 infection of CaMKIIcre:IFNAR^fl/fl and IFNAR^fl/fl control mice elicited overall similar disease onset, including hunched back and ruffled fur by day 5 postinfection (p.i.); however, while IFNAR^fl/fl mice continued to show mild clinical symptoms, CaMKIIcre:IFNAR^fl/fl mice exhibited acute disease progression as manifested by acute hind-limb paralysis in 65% of mice and severe hunching and wasting (Fig. 1A), resulting in 100% mortality by day 7 p.i. (Fig. 1B).

FIG 1.

IFNAR signaling in neurons mediates protection against viral encephalomyelitis. IFNAR^fl/fl (n = 27) and CaMKIIcre:IFNAR^fl/fl (n = 26) mice infected with MHV-A59 were monitored for disease progression (A) and survival rate (B) as described in Materials and Methods. Virus titers (C) in brain supernatants were quantified by plaque assay (n = 3/group at day 4 p.i. and n = 6 or 7/group at day 6 p.i.). Viral replication (D) was determined by viral N mRNA encoding viral nucleocapsid protein (viral-N) using reverse transcription-PCR (RT-PCR) (n = 6/group at day 4 p.i. and n = 11 or 12/group at day 6 p.i.). Data are the means ± SEM from two or three independent experiments and were analyzed by unpaired two-tailed Student’s t test and two-way ANOVA. “*” indicates significance between IFNAR^fl/fl and CaMKIIcre:IFNAR^fl/fl mice, whereas “#” indicates significance between days 4 and 6 p.i. in the same group. * and #, P < 0.05; **, P < 0.01; *** and ###, P < 0.001; ****, P < 0.0001.

To evaluate whether rapid disease progression in CaMKIIcre:IFNAR^fl/fl mice was associated with impaired virus control, we measured infectious virus and mRNA encoding the viral nucleocapsid (N) protein in the brain. Virus titers were similar in both groups of mice at day 4 p.i. and decreased by day 6 p.i. in IFNAR^fl/fl mice, but they remained high in CaMKIIcre:IFNAR^fl/fl mice (Fig. 1C). Viral N mRNA levels were significantly higher in CaMKIIcre:IFNAR^fl/fl mice than in controls at day 4 p.i., suggesting that viral replication was already increased at early times. Consistent with the difference in viral titers at day 6 p.i., IFNAR^fl/fl mice showed a reduction in viral N mRNA, whereas viral replication was sustained at high levels in CaMKIIcre:IFNAR^fl/fl mice (Fig. 1D).

Loss of IFNAR1 signaling in neurons increases virus dissemination.

To assess whether increased viral replication in CaMKIIcre:IFNAR^fl/fl mice correlated with virus dissemination throughout the CNS, we analyzed the distribution of viral N mRNA in distinct brain structures. Evaluation of olfactory bulb (OB), cerebral cortex (CC), hippocampus (HC), diencephalon (DC), cerebellum (CB) and brain stem (BS) showed comparable viral N mRNA levels in CaMKIIcre:IFNAR^fl/fl and IFNAR^fl/fl mice at day 4 p.i., with the highest expression in the OB and lowest viral replication in the CB (Fig. 2). While viral N mRNA levels were significantly reduced in all regions of infected IFNAR^fl/fl mice by day 6 p.i., only the OB and HC showed viral control in CaMKIIcre:IFNAR^fl/fl mice. Overall, viral N mRNA levels remained significantly higher in all regions except in the CB of CaMKIIcre:IFNAR^fl/fl compared to IFNAR^fl/fl mice. The apparent discrepancies of viral N mRNA levels between whole brain (Fig. 1D) and the subregions (Fig. 2) may reside in the distinct sizes of different subregions and exclusion of some regions not dissected. Sustained high viral N mRNA levels of the CC, DC, and BS likely dilute the significant reduction of viral N mRNA levels in the smaller OB and HC, making direct comparison unfeasible.

FIG 2.

IFNAR abrogation in neurons results in increased viral replication throughout the brain. Viral replication in different brain regions was analyzed by viral N mRNA levels using RT-PCR. Data are the means ± SD of brain regions (n = 4 or 5 mice/time point) and were analyzed by unpaired two-tailed Student’s t test and two-way ANOVA. “*” indicates significance between IFNAR^fl/fl and CaMKIIcre:IFNAR^fl/fl mice, whereas “#” indicates significance between days 4 and 6 p.i. in the same group. *, P < 0.05; ** and ##, P < 001; ###, P < 0.001. OB, olfactory bulb; CC, cerebral cortex; HC, hippocampus; DC, diencephalon; CB, cerebellum; BS, brain stem.

Viral antigen (Ag) detected with monoclonal antibody (MAb) J3.3 specific for viral N protein was more abundant throughout the brains of CaMKIIcre:IFNAR^fl/fl mice than of IFNAR^fl/fl mice and largely excluded the cerebellum (Fig. 3A), consistent with viral N mRNA levels. Unlike the expression pattern of CaMKIIα-driven reporter proteins (30), viral dissemination was not limited to neurons in the forebrain but also affected mid- and hindbrain areas in CaMKIIcre:IFNAR^fl/fl mice (Fig. 3A and B). The staining pattern of viral Ag-positive cells was consistent with neuronal morphology but also included cells with glial morphology in the CC (Fig. 3B.a and b) and midbrain (Fig. 3B.d) and a more mixed pattern in other regions (Fig. 3B.c, e, and f). While virus-infected cells with neuronal morphology were present and distributed sporadically, there was an overall tendency of preferential infection of cells with glial morphology in BS regions, including the pons (Fig. 3B.e) and medulla (Fig. 3B.f). Intense viral Ag-positive foci were also found in the motor cortex region (Fig. 3A.a) and tracks, including brain stem (Fig. 3A and B.d to f), providing connectivity to the spinal cord. Viral Ag was indeed readily detected in spinal cords of CaMKIIcre:IFNAR^fl/fl mice compared to control mice (Fig. 3C), with infected loci prominent in the gray matter, including ventral horns harboring motor neurons (38). This robust infection may potentially underlie the unusual hind-limb paralysis within 6 days of infection.

FIG 3.

IFNAR abrogation in neurons results in virus dissemination throughout the CNS. (A) Representative images of viral N protein distribution in brains from both mouse groups at day 6 p.i. detected by immunoperoxidase staining using J3.3 MAb (red chromogen; hematoxylin counterstain). Scale bars = 1 mm. (B) Insets from panel A show infected area or cells at higher magnifications. Scale bars = 50 μm. Insets of panel B show infected cells indicating typical neuronal morphology at higher magnifications. (C) Representative images of viral N protein distribution in spinal cord (lumbar spine) from both groups at day 6 p.i. detected by J3.3 Mab. Scale bars = 0.5 mm.

Expanded neuronal infection in brains was verified by dual staining for viral N protein in combination with Ab specific for neurons (NeuN). Focused analysis on the CC (Fig. 3A and B.a), where infection clustered in cells with neuronal morphology, revealed elevated neuronal infection in CaMKIIcre:IFNAR^fl/fl mice compared to IFNAR^fl/fl mice at both days 4 and 6 p.i. (Fig. 4A). Increased infection of glial cells at day 6 p.i. was tested by costaining for viral N protein in combination with Abs specific for microglia/macrophages (IBA1) and astrocytes (GFAP). Although IBA1⁺ cells were proximal to viral N⁺ areas, viral N⁺ IBA1⁺ cells were sparse in control mice (Fig. 4B). CaMKIIcre:IFNAR^fl/fl mice exhibited more clusters of viral N⁺ IBA1⁺ cells with viral N localized in the cytoplasm or processes of IBA1⁺ cells; IBA1⁺ cells were also found around viral N⁺ areas (Fig. 4B). Although the relative proportion of the dually viral N⁺ and IBA1⁺ area per total viral N⁺ area showed a trend toward increased infection of IBA1⁺ cells in CaMKIIcre:IFNAR^fl/fl mice, it did not reach significance (Fig. 4C). In contrast, colocalization of viral Ag with anti-GFAP Ab was significantly increased in CaMKIIcre:IFNAR^fl/fl compared to IFNAR^fl/fl mice at day 6 p.i. (Fig. 4D and E). Astrocyte infection may even be underestimated, as anti-GFAP Ab prominently stains astrocyte main stem branches but not cytoplasm or fine processes (12, 39, 40), thereby excluding viral Ag localized proximally but not overlapping with GFAP. Irrespectively, these results imply that loss of IFNAR1 on CaMKIIα-expressing neurons leads to not only more disseminated infection of neurons but also glial populations in most brain regions.

FIG 4.

IFNAR abrogation in neurons results in increased virus infection in neurons but also increased astrocyte and, to a lesser extent, microglia/macrophage infection. Shown is representative staining of viral N protein of the CC region using anti-J3.3 MAb in combination with anti-NeuN MAb (A), anti-IBA1 (B), and anti-GFAP (C) as markers for neurons, microglia, and astrocytes, respectively, and DAPI to identify nuclei. Scale bars: 50 μm (A, B, and D). Inset scale bars: 20 μm (A) and 10 μm (B and D). The proportion of viral N⁺ IBA1⁺(C) or viral N⁺ GFAP⁺ (E) per total viral N⁺ area in each field, respectively, is quantified in scatterplots. Each circle represents one field analyzed from 2 to 4 sections and 3 mice/group; the SEM is indicated by crossbar. Data were analyzed by unpaired two-tailed Student’s t test and two-way ANOVA. “*” indicates significance between IFNAR^fl/fl and CaMKIIcre:IFNAR^fl/fl mice. **, P < 001.

Abrogated IFNAR signaling in neurons does not impair overall early IFN-α/β responses, but elevated virus replication does not sustain IFN-α/β.

To assess whether overall sustained virus replication in CaMKIIcre:IFNAR^fl/fl mice was attributable to impaired overall or regionally different innate immune responses, we analyzed transcripts for selected IFN-α/β and ISGs, namely, Ifnβ1, Ifnα4, Ifnα5, and Ifit mRNAs (Fig. 5). The levels of all five mRNAs in total brains were higher at day 4 than day 6 p.i. in both mouse groups (Fig. 5A) irrespective of sustained viral replication in CaMKIIcre:IFNAR^fl/fl mouse brains. Only Ifnα5 and Ifit2 mRNAs were significantly higher in CaMKIIcre:IFNAR^fl/fl mice than in control mice at both days 4 and 6 p.i. We also selected three brain regions for similar analysis, namely, the OB, which showed efficient viral control, the CC, which revealed poor viral control, and the BS, which showed no virus control in CaMKIIcre:IFNAR^fl/fl mice (Fig. 2). The overall transcription profiles mirrored those in the whole brains (Fig. 5B to F), with the exception of Ifit2 mRNA levels in the CC, which increased in control mice and were even more pronounced in CaMKIIcre:IFNAR^fl/fl mouse brains at day 6 p.i. (Fig. 5F). Notably, higher IFN-α/β and ISGs transcripts in the CC of CaMKIIcre:IFNAR^fl/fl mice at day 6 p.i. coincided with poor control of viral replication, while the BS region with no viral control retained levels of these transcripts similar to those of control mice. These data indicated that failed viral control is not due to impaired overall or region-specific early IFN-α/β and ISG responses.

FIG 5.

Altered CNS expression of IFN-α/β genes and ISGs in the absence of IFNAR on neurons. Total brains and selected brain regions, namely, the OB, CC, and BS from MHV-A59-infected IFNAR^fl/fl and CaMKIIcre:IFNAR^fl/fl mice, were analyzed for Ifnα4, Ifnα5, Ifnβ1, and ISG (Ifit1 and Ifit2) mRNA levels by quantitative RT-PCR. (A) Expression of indicated mRNAs in whole brain-derived RNA. (B to F) OB, CC, and BS expression of Ifnβ1 (B), Ifnα4 (C), Ifnα5 (D), Ifit1 (E), and Ifit2 (F) mRNA. Data represent the means ± SEMs from whole brain (A) from two independent experiments (n = 6/group at day 4 p.i. and n = 11 or 12/group at day 6 p.i.) and the means ± SDs of brain regions (B to F) (n = 4 or 5 mice/time point). Data were analyzed by unpaired two-tailed Student’s t test and two-way ANOVA. “*” indicates significance between IFNAR^fl/fl and CaMKIIcre:IFNAR^fl/fl mice, whereas “#” indicates significance between days 4 and 6 p.i. in the same group. * and #, P < 0.05; ** and ##, P < 0.01; ###, P < 0.001; **** and ####, P < 0.0001.

Abrogated IFNAR1 signaling in neurons does not impair early expression of proinflammatory mediators.

CNS cells also induce chemokines and cytokines following infection, which facilitate recruitment of innate and adaptive immune cells to infected sites (41, 42). The proinflammatory cytokines tumor necrosis factor (TNF), interleukin 1β (IL-1β), and IL-6 contribute to both protective innate and adaptive immune responses but also promote blood-brain barrier (BBB) disruption (43, 44). Prolonged elevated expression of these factors by uncontrolled virus replication can be associated with toxicity and immune damage in the CNS (12, 15, 21). We therefore assessed whether dysregulated expression of these factors potentially contributed to disease. The patterns of Tnf, Il1β, and Il 6 mRNA expression were overall similar in whole brains of the two groups, declining between days 4 and 6 p.i. (Fig. 6A). Analysis of the distinct brain regions showed more diverse patterns for Tnf mRNA, with lower levels in the OB but higher levels in the CC and BS of CaMKIIcre:IFNAR^fl/fl mice than in their wild-type (WT) counterparts at day 6 p.i. (Fig. 6B). Surprisingly, Il1β mRNA in the OB, CC, and BS of IFNAR^fl/fl mice were upregulated between days 4 and 6 p.i., while no changes were evident in those regions in CaMKIIcre:IFNAR^fl/fl mice (Fig. 6C). Finally, similar to the case with whole brains, Il6 mRNA expression in all three brain regions had declined by day 6 compared to day 4 p.i. (Fig. 6D). These data suggest that increased viral replication in CaMKIIcre:IFNAR^fl/fl mice is not associated with elevated levels of proinflammatory mediators. Nevertheless, analysis of whole brains versus selected regions indicates that more nuanced regional responses may contribute to pathogenesis.

FIG 6.

Altered CNS mRNA expression of chemokines and proinflammatory cytokines in the absence of IFNAR on neurons. Total brains (A and E) and selected brain regions, namely, the OB, CC, and BS (B to D and F to H), from MHV-A59-infected IFNAR^fl/fl and CaMKIIcre:IFNAR^fl/fl mice were analyzed for Cxcl1, Ccl2, Ccl5, Tnf, Il1b, and Il6 mRNA levels as indicated. Data represent the means ± SEMs of whole brains of (A and E) from two independent experiments (n = 6/group at day 4 p.i. and n = 11 or 12/group at day 6 p.i.) and the means ± SDs of brain regions (B to D and F to H) (n = 4 or 5 mice/time point). Data were analyzed by unpaired two-tailed Student’s t test and two-way ANOVA. “*” indicates significance between IFNAR^fl/fl and CaMKIIcre:IFNAR^fl/fl mice, whereas “#” indicates significance between days 4 and 6 p.i. in the same group. * and #, P < 0.05; ##, P < 0.01; *** and ###, P < 0.001; **** and ####, P < 0.0001.

We also assessed mRNA levels of CXCL1, a prominent neutrophil chemoattractant, and CCL2, a monocyte and NK1.1 cell-recruiting chemokine. Cxcl1 and Ccl2 mRNAs were elevated and sustained in CaMKIIcre:IFNAR^fl/fl mouse brains compared to control brains at day 6 p.i. (Fig. 6E). This trend was similar in OB, CC, and BS (Fig. 6F and G). However, Cxcl1 and especially Ccl2 mRNA levels were only significantly higher in CaMKIIcre:IFNAR^fl/fl mouse brain regions relative to controls at day 6 p.i. (Fig. 6F and G). Assessment of transcript encoding CCL5, a T cell chemoattractant prominently produced by T cells during MHV infection (45, 46), revealed overall increased Ccl5 mRNA levels in whole brains over time in both groups (Fig. 6E); however, only the OB of IFNAR^fl/fl mice showed an increase between days 4 and 6 p.i. (Fig. 6H). Sustained Cxcl1 and Ccl2 mRNA levels thus implied that enhanced recruitment of myeloid cells to the CNS of CaMKIIcre:IFNAR^fl/fl mice may contribute to morbidity, but similar Ccl5 mRNA levels supported no major deficits in T cell recruitment.

Loss of IFNAR1 on neurons increases neutrophil accumulation but does not impair T cell recruitment.

Consistent with elevated Cxcl1 mRNA, neutrophils were increased 3- and 10-fold in CaMKIIcre:IFNAR^fl/fl mouse brains compared to IFNAR^fl/fl mouse brains at days 4 and 6 p.i., respectively (Fig. 7A). However, increased Ccl2 mRNA levels in CaMKIIcre:IFNAR^fl/fl mice did not correlate with significantly higher numbers of macrophages or NK1.1 cells relative to those in IFNAR^fl/fl mice (Fig. 7A). As both CD4 and CD8 T cells are essential to reduce infectious MHV, with IFN-γ playing a prominent antiviral role (41, 46), we further assessed if lack of virus control is due to impaired T cell recruitment or IFN-γ production within the CNS. Neither CD4 or CD8 T cell recruitment was impaired in CaMKIIcre:IFNAR^fl/fl mice (Fig. 7B), and IFN-γ reached even higher levels than in IFNAR^fl/fl mice (Fig. 7C). As CD8 T cells are primary mediators reducing infectious MHV within the CNS, we further assessed the virus-specific CD8 T cell fraction responding to the dominant MHV-A59 spike protein-derived S598 peptide via IFN-γ production (47). The frequency of peptide-stimulated IFN-γ-producing CD8 T cells was almost twice as high in CaMKIIcre:IFNAR^fl/fl mouse brains as in IFNAR^fl/fl mouse brains (Fig. 7D). The higher fraction of IFN-γ-producing cells in unstimulated CaMKIIcre:IFNAR^fl/fl mouse CD8 T cells versus IFNAR^fl/fl mouse CD8 T cells (15.1% versus 3.7%) implicated elevated responsiveness to endogenous viral Ag in the cell preparation, consistent with elevated viral replication (Fig. 1). Ifnγ mRNA levels in whole brain were also substantially increased in both mouse groups at day 6 relative to day 4 p.i. (Fig. 7E), reflecting T cell infiltration (Fig. 6E). Interestingly, the patterns of Ifnγ mRNA expression were brain region specific (Fig. 7F), with extensive upregulation of Ifnγ mRNA expression in the OB and BS in both groups but in only the CC of CaMKIIcre:IFNAR^fl/fl mice (Fig. 7F). These data supported the notion that elevated and prolonged IFN-γ production in CaMKIIcre:IFNAR^fl/fl mice at day 6 p.i. is driven by increased Ag presentation (48, 49). Overall poor control of infectious virus in CaMKIIcre:IFNAR^fl/fl mice despite evident T cell function thus suggested limited access of T cells to selected infected areas or limited responsiveness of infected cells to IFN-γ. Histological analysis showed that the distributions of CD3⁺ T cells in the perivascular space versus parenchyma were similar in the two mouse groups, with the majority localizing to the parenchyma of the cerebral cortex by day 6 p.i. (Fig. 7G). Taken together, these results implied that rapid virus dissemination outpaced the T cell response or that a subset of infected cells, specifically neurons, were unresponsive to CD8 T cell function.

FIG 7.

Abrogated IFNAR signaling in neurons increases neutrophil infiltration. Numbers of brain infiltrated neutrophils, macrophages, and NK1.1 cells (A) and CD4 and CD8 T cells (B) in infected IFNAR^fl/fl and CaMKIIcre:IFNAR^fl/fl mice determined by flow cytometry at the indicated time points. (C) IFN-γ protein in brain supernatants measured by ELISA. (D) Cells derived from brain at day 6 p.i. were stimulated with S598 peptide (1 μM) for 5 to 6 h and analyzed for IFN-γ production. Ifnγ mRNA levels in whole brains (E) and regions of the OB, CC, and BS (F) were determined by RT-PCR. Data represent the means ± SEMs of whole brains (A to C and E) from two independent experiments (n = 6/group at day 4 p.i. and n = 11 or 12/group at day 6 p.i.) and the means ± SDs of brain regions (F) (n = 4 or 5 mice/time point). Data were analyzed by unpaired two-tailed Student’s t test and two-way ANOVA. “*” indicates significance between IFNAR^fl/fl and CaMKIIcre:IFNAR^fl/fl mice, whereas “#” indicates significance between days 4 and 6 p.i. in the same group. * and #, P < 0.05; ** and ##, P < 0.01; ###, P < 0.001; ****, P < 0.0001. (G) Representative staining of T cell distribution in the CC was determined using anti-CD3 MAb in combination with anti-laminin Ab. Scale bar = 50 μm.

CaMKIIcre:IFNAR^fl/fl mice exhibit impaired IFN-γ signaling in microglia/macrophages.

Our previous studies showed that uncontrolled MHV replication and elevated IFN-α/β in mice with IFNAR1 ablation in astrocytes coincided with loss of responsiveness of microglia to IFN-γ signaling (12). Importantly, staggered in vitro stimulation of macrophages with poly(I·C) to mimic viral infection and IFN-γ demonstrated impaired IFN-γ responsiveness in poly(I·C)-pretreated cells (12), supporting direct interference of IFN-α/β with IFN-γ pathways. Although elevated IFN-α/β in CaMKIIcre:IFNAR^fl/fl versus control mice was not as robust as in mGFAPcre:IFNAR^fl/fl mice (12), we nevertheless tested whether the efficacy of IFN-γ signaling is also affected in CaMKIIcre:IFNAR^fl/fl mice. Shared downstream signaling factors, including STAT1, following receptor engagement by IFN-α/β and IFN-γ (1, 50) make it difficult to distinguish IFN-γ receptor (IFNGR) from IFNAR1-specific responses in vivo (51). One prominent indicator of IFN-γ signaling in microglia/macrophages is induction of the transcription factor class II transactivator (CIITA), a master regulator of MHC class II expression (52, 53). Despite higher CNS IFN-γ levels, Ciita mRNA expression was barely induced in whole brains or selected brain regions of CaMKIIcre:IFNAR^fl/fl mice, contrasting robust upregulation by day 6 p.i. in control mice (Fig. 8A and B). Moreover, both microglia and monocyte-derived macrophages from CaMKIIcre:IFNAR^fl/fl mice were significantly impaired in MHC class II expression at day 6 p.i. compared to those from IFNAR^fl/fl mice (Fig. 8C). Both the relative percentage (Fig. 8D) of MHC class II expressing cells and magnitude per cell reflected by mean fluorescence intensity (MFI; data not shown) were significantly lower.

FIG 8.

Abrogated IFNAR signaling in neurons results in impaired IFN-γ signaling in microglia/macrophages. Ciita mRNA levels in brains (A) and indicated anatomical regions (B) were determined by quantitative RT-PCR at indicated time points. (C and D) Histogram and bar graphs show MHC class II expression on microglia/macrophages from brains of both mouse groups determined by flow cytometry at day 6 p.i. Cxcl9 mRNA levels in whole brains (E) and indicated anatomical regions (F) were measured by quantitative RT-PCR. Data represent the means ± SEMs of whole brains (A, D, and E) from two independent experiments (n = 6/group at day 4 p.i. and n = 11 or 12/group at day 6 p.i.) and the means ± SD of brain regions (B and F) (n = 4 or 5 mice/time point). Data were analyzed by unpaired two-tailed Student’s t test and two-way ANOVA. “*” indicates significance between IFNAR^fl/fl and CaMKIIcre:IFNAR^fl/fl mice, whereas “#” indicates significance between days 4 and 6 p.i. in the same group. *, P < 0.05; ** and ##, P < 0.01; ***, P < 0.001; **** and ####, P < 0.0001.

Another factor specifically upregulated by IFN-γ is the chemokine CXCL9, which is mostly expressed by endothelial cells, dendritic cells, and microglia/macrophages (54, 55). Cxcl9 mRNA in whole brain was upregulated in control mice but not in CaMKIIcre:IFNAR^fl/fl mice at day 6 p.i. (Fig. 8E). In contrast, Cxcl9 mRNA levels were significantly increased in the OB, CC, and BS of both groups, with only the OB revealing lower levels in CaMKIIcre:IFNAR^fl/fl mice than in WT mice (Fig. 8F). Taken together, these data indicate that loss of IFNAR1 signaling in neurons impaired IFN-γ-mediated Ciita mRNA expression in myeloid cells throughout the CNS, similar to our results for mice with INAR ablation in astrocytes (12). Region-specific differences in Cxcl9 mRNA expression possibly reflect CXCL9 expression in nonmyeloid cells.

To assess if nuclear translocation of pSTAT1, mediating both IFNAR and IFNIγR signaling (1), was altered in microglia, brain sections were stained for pSTAT1 in combination with IBA1 or the cytoplasmic neuronal marker HUR. Naive mice showed no pSTAT1 reactivity (data not shown). Analysis of CC as a prominently infected area revealed that pSTAT1⁺ nuclei in the immediate vicinity or overlapping with IBAI reactivity were overall sparse in both groups at days 4 (data not shown) and 6 (Fig. 9A) p.i. Moreover, pSTAT1⁺ IBA⁺ reactivity was largely found in areas of highly activated, clustered foci of microglia/macrophages (Fig. 9A). Contrasting the few pSTAT1⁺ IBA⁺ cells, pSTAT1 was abundant in CC neurons of both mouse groups at days 4 and 6 p.i. (Fig. 9B). No evidence for reduced pSTAT1 reactivity in CC neurons of CaMKIIcre:IFNAR^fl/fl mice relative to control mice suggested that early IFN-γ signaling may be responsible for pSTAT1 nuclear translocation at day 4 p.i. Sustained and even enhanced pSTAT1 reactivity in neurons of CaMKIIcre:IFNAR^fl/fl mice relative to control mice at day 6 p.i., when IFN-γ levels were high, supported this notion. Unfortunately, derivation of both CaMKII and pSTAT1 Abs from the same host preempted specific analysis of CaMKII⁺ neurons. Interestingly, clusters of neurons, possibly neuronal aggregates, were frequently observed in the CC and other brain regions at day 6 p.i., with higher prevalence in CaMKIIcre:IFANR^fl/fl mouse brains. pSTAT1⁺ HuR⁺ neurons were detected both outside and within clusters (Fig. 9B). While we failed to confirm reduced pSTAT1 in microglia of CaMKIIcre:IFNAR^fl/fl mouse brains relative to control mice as a possible mechanism underlying reduced MHC class II expression during peak IFN-γ levels, these findings emphasize the complexities in studying partially overlapping IFN-α/β and IFN-γ responses in vivo.

FIG 9.

Phosphorylation of STAT1 is more prominent in neurons relative to microglia/macrophages. Representative staining of nuclear pSTAT1 in the CC region was determined using anti-pSTAT1 MAb in combination with anti-IBA1 (A) and HuR (B) Abs as markers for microglia/macrophages and neurons and DAPI to identify nuclei (representative images from 5 to 7 sections and 3 mice/group/time point). Arrows in panels A and B indicate pSTAT1⁺IBA1⁺ and pSTAT1⁺HuR⁺ cells, respectively. Scale bars = 50 μm.

DISCUSSION

Protective IFN-α/β responses depend on integration of signals mounted by initial IFN-α/β producer cells and responder cells to efficiently block viral spread while limiting IFN-α/β-mediated toxicity (16, 56). However, in vivo studies addressing the relevance of IFNAR1 signaling in distinct CNS cell types to innate protection are limited, especially for neurons (11, 16). Neurons have limited regenerative capacity but are susceptible to many neurotropic viruses (21, 57). The importance to preserve neuronal function is reflected by the relative resistance of mature neurons to both direct virus- and immune-mediated cytolysis (21, 58). Further, primary pathways controlling viral replication in neurons tend to involve noncytolytic mechanisms, e.g., IFNs and antibodies (11, 58). The participation of neurons in innate immune responses is supported by their capacity to produce as well as respond to IFN-α/β in vitro but to a lesser extent in vivo (57, 59, 60). The present report highlights an essential role of IFNAR1 signaling specifically in CaMKIIα⁺ neurons in restricting virus spread within the CNS and preventing mortality.

A common denominator between pathogenesis in MHV-A59-infected CaMKIIcre:IFNAR^fl/fl and mGFAPcre:IFNAR^fl/fl mice (12) was the inability to control virus and mortality by day 7 p.i. However, rapid onset of hind limb paralysis was unique to CaMKIIcre:IFNAR^fl/fl mice and likely attributable to elevated infection of neurons in the locomotor circuit (38). The lethal outcome, despite lower viral loads in CaMKIIcre:IFNAR^fl/fl mice than in mGFAPcre:IFNAR^fl/fl mice, may be due to elevated infection of the brain stem. Increased neutrophil infiltration may further contribute to tissue damage and rapid lethality to CaMKIICre:IFNAR^fl/fl mice, reminiscent of the case with mGFAPcre:IFNAR^fl/fl mice (12).

MHV-A59-infected CaMKIIcre:IFNAR^fl/fl mice exhibited increased viral replication at day 4 p.i., coincident with increased Ifnα and Ifit2 mRNA levels compared to those in controls. However, by day 6 p.i., Ifnα/β mRNAs were significantly reduced, despite sustained viral replication. This outcome contrasted that for mGFAPcre:IFNAR^fl/fl mice, in which vastly increased viral loads at day 6 relative to day 4 p.i. coincided with sustained or even increased mRNA levels of IFN-β, IFN-α4, and IFIT1/2 (12). Together, these data highlight that blockade of IFN-α/β signaling in astrocytes versus CaMKIIα⁺ neurons, while both are lethal, has unique effects on viral control and overall regulation of genes in the IFN-α/β pathway. The finding that infection of primary neuronal cultures with MHV-A59 elicits poor IFN-β induction relative to glia (20, 61) further suggests that neurons depend on initial extrinsic IFN-α/β sources to establish an antiviral state. Notably, compared to astrocytes and microglia, naive cortical neurons from adult mice express relatively low mRNA levels encoding IFN-α/β-associated factors (14). In another study, comparison of primary hippocampal neurons to fibroblasts revealed that neurons express higher homeostatic Ifnα/β mRNA levels but lower levels of selected ISGs (11). Despite lower basal levels of STAT1 and STAT2 proteins in neurons, IFN-β treatment induced rapid and robust STAT1/2 activation. Although IFN-β treatment did provide protection from lymphocytic choriomeningitis virus (LCMV) or measles virus infection in vitro, IFNAR1-deficient neuronal cultures as well as mice exhibited higher measles virus replication (11). The different IFN-α/β signatures in region-specific neurons warrant more studies to assess IFN-α/β pathway responses to infection in vivo. A strictly regulated neuronal response in vivo is indeed demonstrated following Theiler’s encephalomyelitis virus (TMEV) and La Crosse virus infection, in which only ∼3% of infected neurons induced IFN-α/β (26). A subsequent report revealed that astrocytes and microglia, although not productively infected by La Crosse virus, are more prominent IFN-β producers than productively infected neurons (60). Similar results were found in other experimental neuronal infections, including vesicular stomatitis virus (VSV), TMEV, and rabies virus (19), supporting the notion that neurons rely on paracrine IFN-α/β to limit viral spread in vivo. IFNAR1 signaling on neurons was indeed required for restricting VSV spread from the olfactory bulb to the brain parenchyma (16, 25). Interestingly, protective mechanisms involved indirect activation of microglia/macrophages in the OB to block VSV CNS dissemination (16).

Varied IFN-α/β responses in distinct neuronal populations may further contribute to region-specific innate responses. For example, higher mRNA levels of genes associated with innate host defenses in primary granule cell neurons of the cerebellum than cortical neurons were associated with enhanced resistance to neurotropic virus infections, including MHV-A59, and higher responsiveness to IFN-β (27). Very low viral N mRNA levels in the cerebellum of both CaMKIIcre:IFNAR^fl/fl and IFNAR^fl/fl mice compared to all other brain regions supported a notable resistance of this area and are consistent with the observation that CaMKII-driven gene activity does not affect neurons in the cerebellum (27). Interestingly, similar viral N mRNA loads comparing the same brain regions of CaMKIIcre:IFNAR^fl/fl and control mice at day 4 p.i. suggested that IFNAR1 signaling in neurons does not exert effects early, although IFN-α/β transcripts are already upregulated by day 3 p.i. in control mice (12, 35). Although expression patterns of IFN-α/β pathway transcripts were slightly different between the groups in distinct regions, they were overall reduced by day 6 p.i. and there was no obvious association with distinct viral control. Downregulated Ifnα/β mRNAs at day 6 p.i. despite elevated virus load suggested that IFNAR1-intact microglia and astrocytes were desensitized by negative regulators, such as SOCS1, or dysregulation of signaling components IRF9, STAT1, and STAT2 (62–64). It remains to be assessed to what extent differential control in distinct regions is attributable to specific innate or adaptive immunity.

IFN-γ, mainly derived from T cells during MHV infection (65, 66), is a critical antiviral mediator (41), which peaks slightly delayed relative to IFN-α/β (12). Higher Ifnγ mRNA levels in CaMKIIcre:IFNAR^fl/fl mouse brains than in control mouse brains possibly reflect increased viral load and T cell engagement. Subtle region-specific regulation of T cell activity was indicated by similar Ifnγ mRNA levels at day 6 p.i. in the OB, which showed virus control, but higher levels in the CC and BS of CaMKIIcre:IFNAR^fl/fl brains, which showed poor or no virus control. Transcript levels of CCL5, also prevalently expressed by T cells (46), were also increased at day 6 p.i.; however, no differences in the OB, CC or BS suggested distinct region-specific regulation of CCL5 and IFN-γ transcription. Overall, there was no evidence for defective T cell infiltration or activation. In this context, it is of interest that Ilβ mRNA levels were slightly decreased by day 6 p.i. in total brains of CaMKIIcre:IFNAR^fl/fl mice compared to control mice. Remarkably, specific brain regions of CaMKIIcre:IFNAR^fl/fl mice showed no induction of Il1β mRNA by day 6 p.i. Of note, IL-1β has been shown to synergize with IFN-α/β to control WNV infection in cortical neurons (67), and IL-1β signaling enhances IFN-β and ISGs in bone marrow-derived dendritic cells (68). Unlike a critical role of local IL-1β signaling in reactivating CD4 T cells in the CNS during WNV infection (69), no deficits in regional Ifnγ mRNA, total IFN-γ protein, or CD8 T cell-specific IFN-γ production ex vivo suggested that reduced Ifnβ mRNA did not impact T cell stimulation in the MHV model.

The inability to effectively control virus could thus not readily be attributed to defective T cell activation. Although we cannot exclude that neurons in CaMKIIcre:IFNAR^fl/fl mouse brains are not susceptible to IFN-γ, IFN-γ is crucial for viral control in neurons during Sindbis or measles virus infection (11, 70–72). There is also evidence for interactions of latently herpes simplex virus 1 (HSV-1)-infected neurons with CD8 T cells (73–75). As both IFN-α/β and IFN-γ can upregulate the class I antigen presentation machinery, IFNAR1 deficiency in CaMKIIα⁺ neurons may dysregulate IFN-γ responsiveness and/or MHC class I antigen presentation, although efficient pSTAT1 nuclear localization argues against this. However, the defect in MHC class II expression on microglia/macrophages from CaMKIIcre:IFNAR^fl/fl mouse brains supported suppression of IFN-γ responses, and a recent report suggests that viral clearance from neurons is mediated by microglia (76). Irrespectively, robust MHC class II expression on microglia in the presence of high viral loads and IFN-γ levels in MHV-infected IFNAR-deficient mice suggest that impaired IFN-γ responsiveness is not a result of viral components disrupting IFN-γ signaling (36).

During acute CNS infections in vivo, IFN-α/β often precedes IFN-γ signaling, but how the overlap of respective responses influences each is not well understood. IFN-α/β and IFN-γ activate signaling cascades through their specific receptors, but share downstream signaling components such as JAKs and STATs. While there is clear overlap in the target genes they induce, there are also uniquely activated and downregulated genes (77). The complex relationship between the IFN-α/β and IFN-γ pathways is evidenced by both positive and negative cross talk (9, 78–80). For example, attenuated STAT1 expression in IFNAR^−/− cells diminishes IFN-γ and cytokine signaling (81) and IFNAR1 activation can optimize IFN-γ responses by facilitating assembly of IFN-γ-activated transcription factors (82, 83). However, in the measles virus model, IFNAR deficiency did not impede IFN-γ-mediated protection, despite initially higher virus loads (11). Other studies, including MHV infection of mGFAPcre:IFNAR^fl/fl mice, revealed that elevated and/or prolonged IFN-α/β signaling can also impair IFN-γ responses in myeloid cells in vitro and in vivo (12, 84–86). Several possibilities may explain impaired microglia/macrophage IFN-γ responsiveness in infected CaMKIIcre:IFNAR^fl/fl mice. Enhanced responsiveness of IFNAR-competent glia may increase expression of negative regulatory ISGs, such as SOCSs or USP18, which inhibit JAK/STAT signaling (9). Consumption or sequestration of IFNGR signaling components may also be affected, potentially due to STAT2-mediated inhibition of STAT1 (63). While we cannot exclude downregulation of IFNGR surface expression (85), this would more likely result in partial, rather than robust, impairment in CIITA-dependent MHC class II expression. In this context, it is of interest that pSTAT1 was overall sparse in microglia/macrophages compared to neurons, irrespective of the abrogation of IFNAR. While initially unexpected, these results are consistent with the notion that IFN-γ signaling in neurons may be more potent in the absence of prior IFN-α/β signaling, as negative-feedback mediators elicited through the IFNAR pathway have not been activated. Nevertheless, higher numbers of pSTAT1⁺ neurons in CaMKIIcre:IFNAR^fl/fl mouse brains coincident with peak IFN-γ production imply that neuronal responses to IFN-γ are not effective in preventing viral spread. Further investigation is clearly required to determine cross regulation of IFN-α/β and IFN-γ signaling cascades in distinct CNS cell types.

In summary, our results highlight the essential role of IFNAR1 signaling specifically for CaMKIIα⁺ neurons in protecting against increased viral spread within neurons but also glia. Further, the inability of neurons to respond to IFN-α/β impaired IFN-γ mediated MHC class II expression on CNS microglia/macrophages. These results thus add to increasing evidence that loss of protection not only is mediated by enhanced susceptibility of an IFNAR1-impaired cell type to infection but also involves other functions, including dysregulation of BBB integrity, impaired microglia barrier formation, and impaired IFN-γ signaling. Further studies are clearly required to unravel the numerous potential mechanisms underlying impaired IFN-γ signaling in microglia/macrophages. Analysis of both direct microglia intrinsic and indirect neuronal signals to microglia will be necessary to better understand the communication between the IFN-α/β and IFN-γ signaling pathways to optimize antiviral activity.

MATERIALS AND METHODS

Mice, virus, and infection.

IFNAR^fl/fl mice engineered to contain loxP sites flanking exon10 of the IFNAR1 gene were originally produced on the 129/Ola background in U. Kalinke’s laboratory (Paul-Ehrlich-Institut, Langen, Germany) as described previously (87) and backcrossed onto the C57BL/6 background in R. Schreiber’s laboratory (Washington University School of Medicine, St. Louis, MO) as described previously (88). CaMKIIαcre^+/− mice were kindly provided by M. Diamond (Washington University School of Medicine). IFNAR^fl/fl mice were crossed in-house with CaMKIIαcre^+/− mice to generate CaMKIIαcre^+/− IFNAR^fl/fl (referred to as CaMKIIcre:IFNAR^fl/fl) mice and subsequently crossed with IFNAR^fl/fl mice for experimentation. Offspring were genotyped for the presence of the Cre recombinase gene and flanking loxP sites by PCR using the following primers: for Cre, forward (F), 5′-GTCCAATTTACTGACCGTACACC-3′, and reverse (R), 5′-GTTATTCGGATCATCAGCTACACC-3′, and for IFNAR flox, F, 5′-TGCTTTGAGGAGCGTCTGGA-3′, and R, 5′-CATGCACTACCACACCAGGCTTC-3′. Cre-negative IFNAR^fl/fl littermates were used as WT controls (IFNAR^fl/fl). All mice were housed under pathogen-free conditions at an accredited facility at the Cleveland Clinic Lerner Research Institute. Mice of both sexes were infected intracranially (i.c.) at 6 to 7 weeks of age with 2,000 PFU of the hepato- and neurotropic coronavirus MHV-A59 clone encoding enhanced green fluorescent protein (EGFP), kindly provided by Volker Thiel (Institute of Virology and Immunology [IVI], Bern, Switzerland) (89). MHV-A59 was propagated on delayed brain tumor (DBT) astrocytoma monolayers, and titers were determined as described previously (90). Virus in the CNS of individual mice was measured by plaque assay on DBT cells using clarified supernatants from brain homogenates prepared as described previously (12). Clinical disease severity was graded daily using the following scale: 0, no disease symptoms; 1, ruffled fur; 2, hunched back or inability to turn upright; 3, severe hunching/wasting or hind-limb paralysis; 4, moribund condition or death (35). All animal procedures were approved by the Institutional Animal Care and Use Committee of the Cleveland Clinic (PHS assurance number A3047-01) and conducted in compliance with the Guide for the Care and Use of Laboratory Animals from the National Research Council (91).

Isolation of CNS cells and flow cytometric analysis.

Brains from mice perfused with cold phosphate-buffered saline (PBS) were homogenized in Dulbecco’s PBS (DPBS; pH 7.4) using Ten Broeck tissue homogenizers as described previously (12). Homogenates were centrifuged at 450 × g for 10 min at 4°C. Cells were resuspended in RPMI medium containing 25 mM HEPES (pH 7.2), adjusted to 30% Percoll (Pharmacia, Uppsala, Sweden), underlaid with 1 ml of 70% Percoll and centrifuged at 850 × g for 30 min at 4°C. Cells were collected from the 30%/70% interface, washed with RPMI medium, counted, and suspended in fluorescent activated cell sorting (FACS) buffer (0.1% bovine serum albumin in DPBS). Fcγ receptors were blocked with 1% mouse serum and rat anti-mouse CD16/32 MAb (clone 2.4G2: BD Biosciences, San Diego, CA) for 20 min on ice prior to staining with fluorescein isothiocyanate (FITC)-, phycoerythrin (PE)-, peridinin chlorophyll protein (PerCP)-, or allophycocyanin (APC)-conjugated MAbs specific for CD45 (clone 30-F11), CD8 (clone 53-6.7), CD4 (clone GK1.5), Ly6G (clone 1A8), CD11b (clone M1/70), MHC class II (clone M5/114.15.2) (all from BD Bioscience, Mountain View, CA), and F4/80 (Serotec, Raleigh, NC) in FACS buffer. For virus-specific IFN-γ production by CD8 T cells, brain-derived cells after day 6 p.i. were cultured with 1 μM spike protein-derived S598 peptide (47) for 5 to 6 h with 1 μl of Golgi Stop (BD Bioscience)/ml. After stimulation, cells were stained for CD45 and CD8 surface expression, fixed, and permeabilized to detect intracellular IFN-γ (clone 4S.B3; eBioscience). Samples were analyzed on a BD Accuri C6 Plus (BD Biosciences). Forward- and side-scatter signals obtained in linear mode were used to establish a gate containing live cells while excluding dead cells and tissue debris. Data were analyzed using FlowJo v10 software (Tree Star Inc., Ashland, OR).

Gene expression analysis.

RNA from whole brains or distinct anatomical brain regions was extracted using TRIzol reagent according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA). Following treatment with DNase I using a DNA-free kit (Ambion, Austin, TX), cDNA was synthesized using Moloney murine leukemia virus reverse transcriptase (Invitrogen) in buffer containing 10 mM deoxynucleoside triphosphate mix, 250 ng of random hexamer primers, and oligo(dT) (1:1 ratio) (Invitrogen). RNA expression was assessed using either SYBR green master mix (Applied Biosystems, Foster city, CA) or TaqMan fast master mix (Applied Biosystems) as described previously (35). The following primers were used with SYBR green master mix: Gapdh, F, 5′-CATGGCCTTCCGTGTTCCTA-3′, and R, 5′-ATGCCTGCTTCACCACCTTCT-3′; viral nucleocapsid (N) gene, F, 5′-GCCAAATAATCGCGCTAGAA-3′, and R, 5′-CCGAGCTTAGCCAAAACAAG-3′; Cxcl9, F, 5′-TGCACGATGCTCCTGCA-3′, and R, 5′-AGGTCTTTGAGGGATTTGTAGTGG-3′; and Il6, F, 5′-ACACATGTTCTCTGGGAAATCGT-3′, and R, 5′-AAGTGCATCATCGTTGTTCATACA-3′. TaqMan fast master mix and TaqMan primers/probes were used to assess Gapdh, Ifnα4, Ifnα5, Ifnβ1, Ifit1, Ifit2, Cxcl1, Ccl2, Ccl5, Il1β, Ifnγ, and Ciita mRNA levels using the 7500 Fast real-time PCR system (Applied Biosystems). Gene expression in total tissue mRNA was normalized to respective Gapdh mRNA expression in each mRNA preparation and converted to a linearized value using the formula 2e(Ct^Gapdh − Ct^gene)Í1,000.

Immunohistochemistry and immunofluorescence.

Brains from PBS-perfused mice were fixed with 10% neutral zinc-buffered formalin and embedded in paraffin for viral N protein analysis. Distribution of viral antigen (Ag) was determined in 10-μm-thick sagittal sections using the J3.3 MAb specific for the carboxyl terminus of the viral N protein as the primary MAb, biotinylated horse anti-mouse IgG as the secondary Ab, and streptavidin-conjugated horseradish peroxidase and 3,3′-diaminobenzidine substrate (Vectastain-ABC kit; Vector Laboratories, Burlingame, CA) as described previously (12, 35). High-resolution whole-slide scanning was performed using the Leica Scanner Aperio AT2 slide scanner (Leica Microsystem, Wetzlar, Germany) with a 20× objective. Sections in each experimental group were evaluated blindly, and representative fields were identified.

For immunofluorescence analysis, half brains were embedded in OCT compound (Scigen Scientific, Gardena, CA), flash-frozen in liquid nitrogen, and stored at −70°C. Blocks were cut into 12- to 14-μm sections using a cryostat at –17°C. Frozen sections were fixed with 4% paraformaldehyde for 15 min at room temperature and permeabilized in 0.3% Triton X-100 in PBS for 15 min. Nonspecific Ab binding was blocked using 1% bovine serum albumin, 10% goat or donkey serum, and goat anti-mouse IgG (H+L) (1:300 dilution, Jackson ImmunoResearch, West Grove, PA). Virus N protein was identified with J3.3 MAb (1:100; Cleveland Clinic Hybridoma Core) in combination with rabbit anti-NeuN (1:1,000; Millipore), rabbit anti-IBA1 (1:500; Wako), or rabbit anti-glial fibrillary acidic protein (1:5,000, GFAP; Dako) Abs marking neurons, microglia/macrophages, or astrocytes, respectively (12, 35). To assess CNS T cell infiltration, sections were stained with rabbit anti-laminin Ab (1:2,000; Cedarlane Laboratories) and rat anti-mouse CD3 MAb (1:100; eBioscience). For pSTAT1 staining, tissue sections were permeabilized with ice-cold methanol and then stained with rabbit anti-pSTAT1 (1:400; Cell Signaling) in combination with mouse anti-HuR (1:300; Santa Cruz Biotechnology) or mouse anti-IBA1 (1:250; Cleveland Clinic Hybridoma Core) to identify neurons and microglia/macrophages, respectively. Secondary Abs were goat anti-mouse Alexa Fluor 488- or 594 and goat anti-rabbit Alexa Fluor 594- or 488-conjugated IgG (1:1,000; Invitrogen). Sections were mounted with Vectashield antifade mounting medium with 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories, Burlingame, CA). Images were acquired using an inverted Leica SP8 confocal microscope (Leica Microsystems, GmbH, Wetzlar, Germany) and Keyence (Osaka, Japan) BZ-X710 digital microscope. All images were analyzed using Fiji v1.0. software. Quantification of viral N protein staining areas and the relative proportion colocalizing with GFAP or IBA1 reactivity were calculated per field of interest using Image-Pro Plus v7.0. For immunohistological analysis, representative data are presented from 4 to 7 separate fields per mouse with 3 mice per group per time point.

IFN-γ ELISA.

IFN-γ in supernatants of brain homogenates described above was measured by enzyme-linked immunosorbent assay (ELISA) as described previously (35). Briefly, 96-well plates were coated with 100 μl of 1-μg/ml purified rat anti-mouse IFN-γ MAb (R4-6A2; BD Bioscience) in 0.1 M disodium hydrogen phosphate, pH 9.5, at 4°C overnight. Following blocking with 10% fetal calf serum (FCS) in PBS for 1 h, samples or recombinant IFN-γ standard (BD Biosciences) was added and plates were incubated at 4°C overnight. Bound IFN-γ was detected using biotinylated rat anti-mouse IFN-γ MAb (XMG1.2; BD Biosciences) and avidin peroxidase followed by 3,3′,5,5′-tetramethylbenzidine (TMB reagent set; BD Biosciences) 30 min later. Optical densities were read at 450 nm with a Bio-Rad model 680 microplate reader and analyzed using Microplate Manager 5.2 software (Bio-Rad Laboratories, Hercules, CA).

Statistical analysis.

Statistics were determined using unpaired two-tailed Student’s t test and two-way analysis of variance (ANOVA) with Bonferroni posttest; respective analysis is indicated in the figure legends. A P value of less than 0.05 was considered statistically significant. Graphs were plotted using GraphPad Prism 7.0a (GraphPad Software, Inc., La Jolla, CA).

Data availability.

The study did not generate any plasmids/mouse lines/new/unique reagents. No new software codes were developed during the study. Further information and request for existing reagents should be directed to and will be fulfilled by the corresponding author, Cornelia C. Bergmann (bergmac@ccf.org).