Acute Murine H5N1 Influenza A Encephalitis

Abstract

Avian influenza A virus H5N1 has the proven capacity to infect humans through cross‐species transmission, but to date, efficient human‐to‐human transmission is limited. In natural avian hosts, animal models and sporadic human outbreaks, H5N1 infection has been associated with neurological disease. We infected BALB/c mice intranasally with H5N1 influenza A/Viet Nam/1203/2004 to study the immune response during acute encephalitis. Using immunohistochemistry and in situ hybridization, we compared the time course of viral infection with activation of immunity. By 5 days postinfection (DPI), mice had lost substantial body weight and required sacrifice by 7 DPI. H5N1 influenza was detected in the lung as early as 1 DPI, whereas infected neurons were not observed until 4 DPI. H5N1 infection of BALB/c mice developed into severe acute panencephalitis. Infected neurons lacked evidence of a perineuronal net and exhibited signs of apoptosis. Whereas lung influenza infection was associated with an early type I interferon (IFN) response followed by a reduction in viral burden concordant with appearance of IFN‐γ, the central nervous system environment exhibited a blunted type I IFN response.

INTRODUCTION

In the recent years, H5N1 avian influenza A virus has spread from migrating water birds to domestic poultry where it has become endemic in Asia, the Middle East, Europe and Africa (2). This has led to concern that a H5N1 virus could genetically adapt or reassort with a human‐transmissible virus to efficiently transmit between human populations, permitting potential emergence of a highly pathogenic pandemic virus. Although H5N1 remains an avian virus, as of late 2010, there have been 512 confirmed human cases of H5N1 reported to the World Health Organization, 304 of which were fatal (36). These cases have primarily arisen in individuals with close contact with affected poultry.

While largely known for respiratory infections, influenza outbreaks have long been associated with neurological sequelae. Influenzal encephalopathies, encephalitis, encephalitis lethargica and Reye's syndrome are rare but serious central nervous system (CNS) diseases that manifest with influenza infection, especially in young children (34). H5N1‐infected cells can be detected in the brains of acutely infected birds, mice, ferrets and humans 16, 17, 18, 28, 32, 35. Recently, Jang et al have reported that C57B6 mice infected with A/Viet Nam/1203/2004 (VN/04) showed evidence of chronic microglial activation and loss of dopaminergic neurons approximately 50 days following viral clearance (25). This suggests that influenza A infection of the brain can lead to long‐lasting effects that might contribute to the pathological processes observed in human proteinopathies such as Parkinson's disease.

In the current study, we evaluated CNS and lung immune responses during acute infection after intranasal inoculation of BALB/c mice with H5N1 influenza VN/04. This H5N1 variant, isolated from a pharyngeal swab of a boy who died in Vietnam, is able to replicate in mice without genetic manipulation (28). We examined the cellular distribution of viral replication in the lung and brain along with the time course of the intra‐CNS immune response.

MATERIALS AND METHODS

Animals and infections

Female BALB/c mice (Harlan Sprague Dawley, Indianapolis, IN, USA) were infected with influenza virus A/Viet Nam/1203/2004 (H5N1) (VN/04) at 6–8 weeks of age. After anesthetization with ketamine HCl (100 mg/mL) and xylazine (20 mg/mL), the mice were inoculated intranasally with 50 µL of phosphate‐buffered saline (PBS) containing 5 × 10³ plaque‐forming units (PFUs) of VN/04. Mice were monitored daily for weight loss and clinical illness (ie, inactivity, ruffled fur and hunched back). Mice were housed and maintained in microisolators according to standards of the National Research Council Guide for the Care and Use of Laboratory Animals, the Animal Welfare Act and the Centers for Disease Control and Prevention/National Institutes of Health Biosafety in Microbiological and Biomedical Laboratories. The University of Pittsburgh Institutional Animal Care and Use Committee approved all the experiments.

Tissue

The upper and middle right lobes of the lung and the left hemisphere of the brain (n = 5 mice per time point) were harvested daily on days 0–7 postinfection. Tissues were weighed and snap frozen for viral analysis. The upper left lobe of the lung and the right hemisphere of the brain were fixed in 10% buffered formalin. After 7 days of fixation, tissue was paraffin embedded and 6‐µm sections were prepared for histopathological analysis. For immunohistochemistry, in situ hybridization (ISH) and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL), single sagittal sections of the right hemisphere were examined for each mouse at all time points.

Plaque assay

Lung and left forebrain (including olfactory bulbs) viral titers were determined by plaque assays using Madin Darby canine kidney (MDCK) cells (ATCC, Manassas, VA, USA). Monolayers of MDCK cells were grown in six‐well plates to 90% confluency. Frozen tissue was thawed and resuspended at a concentration of 0.1 g/mL by forcing tissue through a 70‐µm cell strainer (BD Biosciences, Bedford, MA, USA). Single‐cell suspensions were cleared of cell debris by centrifugation at 2000 RPM for 10 minutes. After four washes with Dulbecco's modified Eagle's medium (DMEM), 0.1 mL of lung or brain supernatant serial dilutions were added to MDCK cells for 1 h at room temperature, with shaking every 15 minutes. Unabsorbed virus was washed away with DMEM, and cells were overlayed with 1.6% w/v agar (SeaKem^® GTG^®, Lonza, Rockland, ME, USA) mixed 1:1 with L‐15 media (Cambrex, East Rutherford, NJ, USA) and 0.6 µg/mL trypsin (Sigma, St. Louis, MO, USA) for 48 h at 37°C, 5% CO2. Agarose was removed, and cells were fixed with 2 mL of 10% buffered formalin. To visualize plaques, formalin was removed and cells were stained with crystal violet. Plates were washed with water and allowed to dry overnight. Plaques were counted to determine PFU/g tissue using the formula (number of plaques × dilution factor)/0.1 mL × 1 mL/0.1 g. Samples were run in duplicate.

Immunohistochemistry and lectin histochemistry

Immunostaining was performed as described before (9). Formalin‐fixed paraffin‐embedded (FFPE) sections containing lung and brain were stained using antibodies against Influenza A virus (Maine Biotechnology Services, Portland, ME, USA), beta‐III tubulin (N‐term; Epitomics, Burlingame, CA, USA), microtubule‐associated protein 2 (MAP‐2; SMI 52; Covance, Princeton, NJ, USA), neurofilament (Invitrogen, Carlsbad, CA, USA), neuronal nuclei (NeuN; Millipore, Billerica, MA, USA), ionized calcium binding adaptor molecule 1 (Iba‐1; Wako Pure Chemical Industries, Osaka, Japan), glial fibrillary acidic protein (GFAP; Dako, Carpinteria, CA, USA), activated capase‐3 (Cell Signaling, Technology, Danvers, MA, USA), CD3 (Dako); granzyme B (Abcam, Cambridge, MA, USA) or TIA‐1 (Immunotech, Marseille, France). For lectin histochemistry, sections were incubated with biotin‐conjugated Wisteria floribunda agglutinin (WFA) (Sigma) instead of antibodies.

In situ hybridization

Vectors containing full‐length clones of murine interferon (IFN)‐α, ‐β, and ‐γ cDNAs (Open Biosystems, Huntsville, AL, USA) and 760 bp of influenza A/California/04/2009 matrix protein (MP) were linearized to generate sense and antisense templates. ³⁵S‐labeled riboprobes were synthesized using MAXIscript in vitro transcription kit (Ambion, Austin, TX, USA).

FFPE tissue sections containing lung and sagittal sections of brain were deparaffinized with Histoclear and microwaved in 1× citrate buffer pH 6.0 (Invitrogen). Hybridization was carried out as described (15) in hybridization mix [20 mM Hepes pH 7.2, 1 mM ethylenediaminetetraacetic acid (EDTA), 1× Denhart's reagent, 0.1 mg/mL Poly A, 0.6 M NaCl, 20% dextran sulfate, 50 µg/mL yeast tRNA, 50% formamide, and 0.1 M dithiothreitol (DTT)] containing riboprobes at either 12 500 or 50 000 cpm/µL at 50°C overnight. Hybridized sections underwent a series of washes: 5× saline‐sodium citrate (SSC); 5× SSC/10 mM DTT, 42°C, 30 minutes; 2× SSC/50% formamide/10 mM DTT, 60°C, 20 minutes; riboprobe wash solution (0.1 M Tris, pH 7.5; 50 mM EDTA; 0.4 M NaCl), 37°C, 10 minutes, 2×; riboprobe wash solution containing 25 U/mL RNase T1 and 25 µg/mL RNase A, 37°C, 30 minutes; riboprobe wash solution, 37°C, 10 minutes; 2× SSC, 37°C, 10 minutes, 2×. Sections were dehydrated in graded ethanol series containing 0.3 M ammonium acetate, coated with emulsion (Eastman Kodak company, Rochester, NY, USA) for 10 days, 4°C. Slides were developed with D19 (Sigma) and fixed in Rapid Fix (Sigma) for 4 minutes. Control riboprobes for nonviral RNA did not hybridize to lung or brain tissue at any time point postinfection. Noninfected tissue did not show hybridization with the viral probes.

TUNEL labeling

Detection of DNA strand breaks in brain sections was performed using the In situ Cell Death Detection Kit, Fluorescien (Roche Applied Science, Indianapolis, IN, USA) according to the manufacturer's instructions with one exception. The concentration of proteinase K was increased 2×.

RESULTS

Intranasal VN/04 infection of mice is lethal

BALB/c mice were infected intranasally with 5 × 10³ PFU VN/04. As a measure of morbidity, all mice showed weight loss by 3 days postinfection (DPI) that progressed and necessitated humane sacrifice by 7 DPI (Figure 1A). Mice displayed ruffled fur, hunched back and reduced activity by 4 DPI. Virus replication was detected in the lungs as early as 1 DPI (6.7 × 10⁴ PFU/g tissue) and peaked at 3 DPI (5.5 × 10⁸ PFU/g tissue), remaining high until 7 DPI (Figure 1B). To visualize the location of viral replication in the lung, ISH for influenza A matrix protein (MP) was performed on fixed lung tissue at each time point (Figure 2A, left column). At 2 DPI, viral‐infected cells were detected in both bronchial and alveolar spaces. As infection progressed, the alveolar compartment showed greater involvement compared with bronchial cells.

Figure 1.

Weight loss and viral replication of influenza VN/04 in BALB/c mice following intranasal challenge. A. Weight loss was assessed daily following viral challenge. Mice that lost substantial amounts of their original body weight were euthanized. Viral burdens in the lungs (B) and the brain (C) were determined by plaque assays using Madin Darby canine kidney (MDCK) cells (n = 5/time point). Dashed line indicates limit of detection. Abbreviation: PFU = plaque‐forming unit.

Figure 2.

Influenza in situ hybridization (ISH) shows infected cells in lung at 1 day postinfection (DPI) and in brain at 4 DPI. A. Representative ISH time course for influenza A MP is depicted (1–7 DPI) for lung (left column) and brain (right column). B. Severity of influenza ISH foci was accessed daily in sagittal sections of brains (n = 5/time point). Scoring: 0 = no definitive signal, 1 = occasional focus, 2 = focus in most fields, 3 = more than one focus per field.

VN/04 is detected in the brain at 4 DPI

Viral replication is first detected in the brains of most VN/04 infected mice at 4 DPI (1.5 × 10³ PFU/g tissue) and continues to increase until euthanization at 7 DPI (5.3 × 10⁴ PFU/g tissue) (Figure 1C). These results mirrored analyses of influenza A MP ISH of sagittal sections of brain (Figure 2A, right column and 2B). VN/04‐infected cells were first detected primarily in the brain stem and olfactory cortex. By 7 DPI, viral replication had spread into other cortical areas, largely sparing the striatum. Infection was not uniform in these regions and tended to occur in small clusters throughout the brain.

Neurons are the predominant infected cell in the brain

By morphology, the predominant infected cell type in the brain appeared to be the neuron. However, to determine the types of cells infected by VN/04, double‐label immunofluorescent staining was performed using an antibody against influenza A and markers for neurons, astrocytes or microglia. Neither the microglial marker Iba‐1 nor the astrocytic marker GFAP colocalized with influenza staining at 6 or 7 DPI (Figure 3B,C). However, some cases had infected choroid epithelial cells at 5–7 DPI (Figure 3D). Antibodies against either NeuN, neurofilament, beta‐III tubulin or MAP‐2 were used in conjunction with anti‐influenza A to determine whether infected cells expressed neuronal markers. The majority of infected cells showed clear expression of influenza A proteins in conjuction with beta‐III tubulin or MAP‐2‐labeled neurons (Figure 3A) at 5–7 DPI; however, there were numerous infected cells with clear neuronal morphology that did not colabel with neuronal markers.

Figure 3.

Influenza staining is observed in neurons lacking perineuronal nets. A–C. Double‐label immunofluorescent stains of sagittal sections from 6 or 7 days postinfection show that neurons, not microglia or astrocytes, were the predominant infected cell. Influenza A staining is shown in the middle column while staining for microtubule‐associated protein 2 (MAP‐2) (A), ionized calcium binding adaptor molecule 1 (Iba‐1) (B) and glial fibrillary acidic protein (GFAP) (C) is shown in the left column. The merged image is shown in the right column. D. Some brains show infected choroid epithelial cells (green) shown between yellow lines. E. Nearly all infected neurons (green) did not have evidence of a perineuronal net [Wisteria floribunda agglutinin (WFA), red]. F. Neurons coated with a perineuronal net (red) did not show evidence of infection (green).

Perineuronal nets (PNNs) are not detected on infected neurons

The soma and proximal dendrites of specific neurons are surrounded by highly condensed matrices of proteoglycans termed PNNs. PNNs are part of the extracellular matrix (ECM) and are thought to protect fast‐spiking neurons from excitotoxic damage 20, 31. Disruption of the ECM evidenced by PNN loss has been observed in a number of neurodegenerative conditions such as Alzheimer's disease 12, 14, 21 and Creutzfeldt–Jakob disease 6, 19 and during CNS infections such as neurosyphilis (13) and lentiviral encephalitis 5, 29. Visualization of PNN can be achieved by staining with the lectin WFA, which preferentially binds terminal N‐acetylgalactosamines. Using this lectin histochemistry with anti‐influenza staining, we explored the integrity of the ECM during H5N1 encephalitis. The majority of influenza‐infected neurons did not have PNN (Figure 3E). Similarly, most neurons with intact PNN were not infected (Figure 3F). Rarely, infected cells were surrounded by WFA staining, but the staining was dimmer than noninfected neurons. Some regions containing abundant neuronal infection showed diffuse absence of WFA staining, even in noninfected neurons. Interestingly, WFA staining was present around the majority of cells in the lung, and similar to infected neurons, infected cells in the lung did not stain with WFA, consistent with a universal loss of ECM around many types of infected cells.

Appearance of CNS CD3 T cells coincides with the appearance of virus

Coinciding with the detection of H5N1 in the brain, perivascular and parenchymal CD3+ T cells were observed in infected areas of some mice at 4 DPI (Figure 4B). CNS T cells were more abundant in all mice at later stages of infection (Figure 4A,B). While some of the CD3+ T cells in the lung exhibited a cytotoxic profile (ie, colabeled with granzyme B or TIA‐1, two markers of cytotoxic granules), the majority of CNS CD3+ T cells showed little to no colocalization with granzyme B (Supporting Information Figure S3).

Figure 4.

CD3 infiltrate and Interferon (IFN)‐α,‐β,‐γ transcripts in the central nervous system (CNS) of VN/04‐infected mice. A. CD3+ cells (green) are observed in the vicinity of infected neurons (red). B. CD3 T cells appear in the CNS at 4 days postinfection (DPI). Time course of CD3 T cell infiltrate into the CNS following VN/04 infection. Sagittal sections of brain were immunofluorescently stained for CD3 and scored for severity of infiltrate. Scoring: 0 = no definitive signal, 1 = occasional CD3+ cells (>20), 2 = CD3+ cells observed in most fields, 3 = abundant numbers of CD3+ cells observed in most fields. C. Representative image of IFN‐αin situ hybridization (ISH) in the CNS of a VN/04‐infected mouse at 6 DPI. D. IFN‐α transcripts were not abundant in the CNS of VN/04‐infected mice. IFN‐α ISH was performed on sagittal sections of brain and the number of foci were counted for each mouse. E. Representative image of IFN‐β ISH in the CNS of a VN/04‐infected mouse at 7 DPI. F. Evaluation of IFN‐β transcripts during the course of VN/04 infection. G. Representative images of IFN‐γ ISH in the CNS of VN/04 infected mice at 5, 6 and 7 DPI. H. Frequent IFN‐γ transcripts were detected in the CNS at 7 DPI.

IFN‐γ dominates the IFN response in the CNS

IFNs are well known for their antiviral/anti‐influenza activity, although some strains of H5N1 have been reported to be resistant to the antiviral effects of IFN (33). Interestingly, cells expressing IFN‐α transcripts were rare in the CNS throughout the course of infection (Figure 4C,D), while in the lungs, IFN‐α was expressed primarily on 2 DPI (Supporting Information Figure S2A,B). IFN‐β transcripts were more abundant in both the lung and brain (Supporting Information Figure S2C,D and Figure 4E,F), but whereas there was prominent signal for IFN‐β in the lung from 2 to 7 DPI, the brain showed substantially lower presence of IFN‐β transcripts and not until 6 and 7 DPI. By far, IFN‐γ transcripts dominated the IFN response in the lungs and CNS (Supporting Information Figure S2E,F and Figure 4G,H). In the lung, IFN‐γ transcripts were detected beginning at 4 DPI and peaked by 6 DPI, while in the brain, a few cells expressing IFN‐γ transcripts were observed at 5 DPI with modest increase through 7 DPI. We hypothesized that the IFN‐γ signal was associated with CD3+ T cells, so IFN‐γ hybridization and CD3 immunohistochemistry were combined to learn the identity of the IFN‐γ‐expressing cells. The IFN‐γ signal intensely labeled mononuclear infiltrating cells, masking the proposed CD3 signal. These results suggest that suppression of virus in the lung was associated with robust IFN‐γ expression, whereas the in the brain, the IFN‐γ response was more muted.

TUNEL+ cells are observed in areas of infection at 7 DPI

TUNEL and activated caspase‐3 immunohistochemistry were performed to begin to elucidate whether infected neurons undergo apoptosis. At 7 DPI, three of five mice showed abundant TUNEL staining in regions associated with early influenza infection including the brain stem, olfactory bulb and cortex (Figure 5A,B). Influenza immunohistochemistry in conjunction with TUNEL staining was attempted to verify that H5N1‐infected neurons were undergoing apoptosis; however, the TUNEL staining procedure prevented successful anti‐flu immunohistochemistry or in situ hybridization. Examining H5N1‐infected mice brains stained for activated caspase‐3 showed fewer apoptotic cells than the TUNEL slides. Although some activated caspase‐3‐stained cells had the morphological appearance of T cells, there were few CD3‐positive T cells that colocalized with activated caspase‐3 staining by 7 DPI (Figure 5C). MAP‐2 immunohistochemistry showed that some neurons in regions of infection colocalized with activated caspase‐3; however, the MAP‐2 staining was generally weak, suggesting neuronal degradation (Figure 5D).

Figure 5.

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL)+and activated caspase‐3 cells are observed in areas of infection. DNA strand breaks were labeled in sagittal sections of VN/04 infected brains using TUNEL technology. A. Quantitation of TUNEL+ cells. B. Image of TUNEL+ cells in the brain stem at 7 days postinfection (DPI). C. Representative image of CD3 (green) and activated caspase‐3 (act‐casp‐3; red) in the brain stem at 7 DPI. D. Image showing four cells with activated caspase‐3 staining in an area of infection at 7 DPI. Three of the activated caspase‐3+ (red) cells colabeled with microtubule‐associated protein 2 (green).

DISCUSSION

As influenza outbreaks were first associated with neurological complications, it has long been speculated that influenza infection might mediate an encephalitis. However, direct evidence in humans of influenza‐infected neuronal or CNS support cells and accompanying inflammation is sparse 17, 18. Mouse and ferret models examining highly pathogenic strains of H5N1 have noted brain infection 16, 25, 28, 35; however, the pathogenesis of CNS influenza infection is poorly understood. The data presented in this report show that influenza RNA, protein and virions were detected in the brain at 4 DPI. Encephalitis was observed in approximately 75% of VN/04‐infected mice with frequency and severity increasing by 7 DPI. As reported before 16, 25, 28, 35, this confirms that intranasal wild‐type H5N1 infection leads to encephalitis in mammalian hosts.

VN/04 infection was lethal to BALB/c mice, requiring sacrifice by 7 DPI because of severe weight loss. It is not known whether some mice might have survived infection after this time point, but it is interesting to speculate the outcome of CNS infection after 7 DPI. It is possible that neuronal infection would continue to spread throughout the brain leading to death from neurological complications. Alternatively, inherent CNS immune responses or infiltrating innate and adaptive immune responses might clear neuronal infection or limit the spread of infection. There is no known mechanisms for persistence of influenza viral replication in the CNS, but some reports suggest persistence of viral antigen 3, 4. Intriguingly, viral‐induced death of neurons, the immune response to clear virus and/or chronic immune activation in the CNS might lead to long‐term neurological deficits or disease. Jang et al have described one such instance where apparent murine survivors of VN/04 infection demonstrated evidence of chronic microglial activation and dopaminergic neuron loss (26). If mice or ferrets do indeed survive acute H5N1 encephalitis infection, these models will be important to dissect long‐lasting effects of influenza on human CNS function.

Neurons were the predominantly infected cell in H5N1 encephalitis. It is known that macrophages (30) can be infected by influenza, and there are reports suggesting that microglia are infected (25); however, we did not observe VN/04‐infected microglia using Iba‐1 or F4/80 immunohistochemistry. Likewise, astrocytes did not demonstrate influenza staining. To determine if only neurons were infected, a number of neuronal markers were used, such as MAP‐2, beta‐III tubulin, NeuN and neurofilament. Interestingly, there were numerous examples of influenza+/neuronal marker+ cells but there were also a large number of infected cells that did not colocalize with neuronal markers (with the exception of beta‐III tubulin). These cells had the morphological appearance of neurons, suggesting that infected neurons might lose expression of neuronal transcription factors and structural markers.

Similarly, infected neurons demonstrated decreased or no WFA staining for ECM. Two possible hypotheses arise from these observations: the PNN or ECM is degraded or not maintained after influenza infection, or alternatively, influenza only infects cells without PNN or N‐acetylgalactosamines. Of these two possibilites, we would favor the former as the distribution of infected cells includes regions known to have PNN in noninfected mice. As PNN modulate synaptic plasticity in mature neurons (23) and protect fast‐spiking neurons from excitotoxic damage 20, 31, disruption of the ECM during influenza encephalitis could contribute to neurodegeneration by loss of trophic interactions, exacerbating abnormal protein accumulation, excitotoxicity or oxidative stress 1, 10, 11, 22.

By 7 DPI, some mice had frequent TUNEL+ cells suggesting that at least some influenza‐infected neurons and, perhaps, some infiltrating T cells were undergoing apoptosis. Activated caspase‐3 staining showed a slight increase by 7 DPI, but there were fewer activated caspase‐3+ cells than TUNEL+ cells. Surprisingly, the majority of activated caspase‐3+ cells did not colabel with infected cells or CD3+ T cells, but some colocalized with MAP‐2. The MAP‐2 signal was weak and some of the activated caspase‐3+ neurons appeared shrunken. This suggests that affected neurons at this acute time point are undergoing caspase‐independent cell death. However, it is possible that detection of activated caspase‐3 could continue to increase after 7 DPI (25).

To assess immune response during infection we stained for activated microglia and natural killer (NK) cell and T cell infiltrates. Reactive microglia were observed in regions of infected neurons. Likewise, CD3+ T cell infiltrates were observed at 4 DPI, coinciding with the appearance of virus, and continued to increase in prominence throughout the remainder of infection. Contrary to the lung, these T cells were negative or exhibited weak staining, only visible at high powers for granzyme B and TIA‐1, suggesting they lacked cytotoxic potential. This is consistent with other viral encephalitides where T cell proliferation is inhibited 24, 26, cytotoxicity is diminished 7, 24, and IL‐2 production is inhibited (24). As no good antibodies are available for immunohistochemistry, NK cells were not detected with immunohistochemical methods, but there were granzyme B‐positive cells that did not colabel with CD3, suggesting a small presence of NK cells. It is possible that the brain milieu actively inhibits expression of cytotoxic molecules while promoting less damaging antiviral responses such as IFNs in an effort to protect neurons. Alternatively, it may too early in the course of brain infection to detect cytotoxic T cells. Again, if mice are able survive H5N1 infection, it can be determined whether a T cell response with diminished cytotoxicity is effective in clearing neuronal infection or preventing the spread of infection to other neurons.

As CNS IFN‐γ transcripts mirror the increase of T cells, CD8 T cells are likely the cellular source of IFN‐γ in the CNS. IFN‐γ is known to be important in controlling neuronal viral infection 8, 27. Type I IFN‐α and β transcripts were rarely detected in the CNS and lagged behind appearance in the lung; whereas IFN‐γ transcripts followed the pattern observed in the lung. In the lungs, there was a concomitant decrease in viral replication with appearance of IFN‐γ producing cells, whereas no decrease in viral replication in neurons was observed upon the presence of IFN‐γ‐producing cells. This acute disease provides an excellent model to dissect innate and adaptive immune responses that may protect the brain from lethal encephalitis.

Supporting information

Figure S1. Influenza in situ hybridization (ISH) shows early bronchial involvement followed by alveolar involvement. Severity of influenza ISH foci was accessed daily in sections of lung bronchi (A) and alveoli (B) (n = 5/time point). Scoring: 0 = no definitive signal, 1 = occasional focus, 2 = focus in most fields, 3 = more than one foci per field.

Figure S2. Type I interferon (IFN) transcripts increase in lung shortly after VN/04 infection, while type II IFN transcripts appear later. (A) Representative image of IFN‐α in situ hybridization (ISH) in the lungs of a VN/04‐infected mouse at 2 days postinfection (DPI). (B) IFN‐α transcripts were primarily detected at 2 DPI in VN/04 infected mice. IFN‐α ISH was performed on sections of lung and the number of foci were counted for each mouse. (C) Representative image of IFN‐β ISH in the lungs of a VN/04‐infected mouse at 2 DPI. (D) Evaluation of IFN‐β transcripts during the course of VN/04 infection. (E) Representative images of IFN‐γ ISH in the lungs of a VN/04‐infected mouse at 6 DPI. (H) Frequent IFN‐γ transcripts were detected in the lungs at 6 and 7 DPI.

Figure S3. CD3 T cells in the lung showed colocalization with granzyme B, while the majority of CD3 T cells in the brain showed little to no granzyme B colocalization. (A) Representative image of granzyme B (green) and CD3 (red) in the lungs of a 7 days postinfection (DPI) VN/04‐infected mouse. (B) Image of a CD3‐labeled (green) T cell showing polarization of granzyme B (red) in the lungs of a 7 DPI VN/04‐infected mouse. (C) Representative image of granzyme B (green) and CD3 (red) in the brain of a 7 DPI VN/04‐infected mouse. (D) A rare T cell (green) in the brain of a 7 DPI mouse showing colocalization with granzyme B (red arrow).

Supporting info item

Supporting info item

Supporting info item

Supporting info item