Viral encephalitides are life-threatening diseases in neonates partly due to the irreversible damage inflammation causes to the CNS. This study explored the role of proinflammatory cytokines in the balance between controlling viral replication and eliciting pathologic immune responses in nonlytic viral encephalitis. We show that neonatal mice challenged with arenavirus Tacaribe (TCRV) develop a meningoencephalitis characterized by high IFN-γ and TNF-α levels and mild T cell infiltration. Neutralization of the TNF-α using mAb was associated with lower chemokine expression, reduced T cell infiltration, and lower levels of IFN-γ, and TNF-α in the CNS and led to 100% survival. Moreover, treatment with Abs to TNF-α improved mobility and increased survival even after the mice developed bilateral hind limb paralysis. Of note, animals treated with anti-TNF-α Abs alone did not clear the virus despite generating Abs to TCRV. Direct activation of the innate immune response using CpG oligodeoxynucleotides in combination with anti-TNF-α Abs resulted in 100% survival and complete viral clearance. To our knowledge, this is the first demonstration of the use of innate immune modulators plus Abs to TNF-α as therapeutics for a lethal neurotropic viral infection.
The CNS is regularly devoid of lymphocytes; however, astrocytes and microglia are known to express pattern recognition receptors and, upon infection, rapidly produce IFNs, proinflammatory cytokines, and chemokines that lead to local inflammation and cellular infiltration (1, 2). In the case of viruses that affect CNS neuronal integrity and function, this immune response can be beneficial in limiting the spread of the virus and in restricting virus-induced CNS damage. However, for noncytolytic viruses, the immune response generated may cause more damage to the host than the original viral insult.
New World arenaviruses (Tacaribe serocomplex) are a growing family of noncytolytic, enveloped, segmented, single-stranded RNA viruses of increasing medical importance as causative agents of South American hemorrhagic fevers (Junin, Machupo, Guaranito, and Sabia viruses) (3, 4, 5). Their clinical course is characterized by hemorrhagic disorders as well as by neurological signs including depression, ataxia, tremors, convulsions, and coma. Because of their high pathogenic potential (15–30% morbidity), human to human transmission, ease of growth in culture, plastic genomic structure, and lack of proven effective therapeutic approaches, these viruses have been included in the list of potential biological warfare agents (class A) by the National Institute of Allergy and Infectious Diseases (NIAID) and the Centers for Disease Control and Prevention (CDC) (5). The standard of care for patients with arenavirus-induced hemorrhagic fevers is supportive care and convalescent plasma (5, 6). The Tacaribe arenavirus (TCRV)4 is highly homologous to Junin, the agent of Argentine hemorrhagic fever (7), but has low pathogenic potential for humans. In mice, TCRV causes a lethal meningoencephalitis in neonatal mice that is characterized by diminished activity, weight loss, flaccid progressive paralysis, and death 1–2 wk after infection depending on the strain, dose, and route of inoculation (8, 9). Previous studies had shown increased survival in challenged mice that lacked T cells (8, 10, 11). More recently, our group showed that systemic administration of immunostimulatory oligonucleotides containing a CpG motif (CpG oligodeoxynucleotide (ODN)) at the time of infection accelerated Ab production to the virus and reduced levels of IFN-γ and TNF-α in the CNS, but only marginally improved survival (30–50%) (9).
Elevated levels of IFN-γ and TNF-α are present in the pathogenesis of multiple inflammatory processes of the CNS, including multiple sclerosis, stroke, and infectious diseases ranging from cerebral malaria and bacterial meningitis to HIV encephalopathy (12, 13, 14); however, their exact role in brain protection, repair, or pathology during viral infections remains controversial (15, 16, 17). In this study we identify TNF-α as a key mediator of the pathogenesis of Tacaribe-induced meningoencephalitis and show that activation of the innate immune system while blocking the deleterious effects of TNF-α leads to viral clearance and survival.
Materials and Methods
Mouse mAbs to TCRV were provided by Dr. M. J. Buchmeier of the Scripps Research Foundation (La Jolla, CA) and have been previously described (18). Low endotoxin, azide-free, purified rat anti-mouse TNF-α neutralizing mAb (XT-122) and purified rat IgG1 were obtained from BioLegend. Neutralizing hamster mAb (clone 2E2) to mouse TNF-α/β was obtained from the National Cancer Institute’s Biological Resources Branch Preclinical Repository (Fisher BioServices) at Rockville, MD, purified control hamster IgG1 was obtained from BD Biosciences, and phosphorothioate CpG ODN 1555 (GCTAGACGTTAGCGT; the unmethylated CpG motif is underlined) was synthesized at the CBER core facility. All ODN had <0.1 endotoxin unit of endotoxin per milligram of ODN as assessed by a Limulus amebocyte lysate assay (QCL-1000). Escherichia coli LPS, DNase, poly-l-lysine, and laminin were obtained from Sigma-Aldrich. Rabbit anti-glial fibrillary acidic protein (GFAP) and carbocyanin (Cy)3-, Cy5-, or FITC-conjugated donkey anti-mouse and anti-rabbit IgG secondary Abs were obtained from Chemicon. RPMI 1640, DMEM with high glucose (4,500 mg/L), DMEM/F12 (1:1) nutritional supplemented medium, heat-inactivated horse serum, HEPES, HBSS, l-glutamine solution, penicillin-streptomycin solution (50 U and 50 μg per milliliter, respectively), trypsin (0.25%)-EDTA (1 mM), and trypan blue were purchased from Invitrogen. FBS and normal goat serum were obtained from HyClone.
Virus and growth conditions
TCRV (VR-114, strain TRVL 11573; American Type Culture Collection) was obtained as a suckling mouse brain desiccate and resuspended and expanded as described (9). TCRV levels in Vero culture and brain extracts were determined using the 50% tissue culture-infective dose (TCID50) method as described (9).
Mouse cortical mixed glia cultures
Neonatal cerebral cortices from BALB/c or C57BL/6 (B6) mice were collected on days 1–3 of life and placed in cold DMEM. After removing the meninges, the cortices were minced and dissociated with trypsin-EDTA at 37°C for 15 min followed by passing through fire-polished Pasteur pipettes with DNase (3 μg/ml) and washing. The cell suspension was plated onto poly-l-lysine (0.05 mg/ml) and laminin (0.1 mg/ml)-coated culture plates or tissue culture cover slips (Fisherbrand; Fisher Scientific) Cells were plated at a density of 100,000 cells/cm2 in DMEM/F12 medium and grown in a humidified 5% CO2 incubator at 37°C for 14–21 days until reaching confluence, at which time they were infected. Medium was changed every 3 or 4 days. At the time of infection the cultures had >80% astroglia, 5–10% microglia, 1–5% neurons, and occasional oligodendrocytes (<1%).
In vitro infections and immunocytochemistry
Primary mixed cortical neuron-glia cultures were infected with 2000 TCID50 of TCRV for 1–2 h. After removal of the virus, fresh medium was added and cells were cultured for 1–4 days. Controls included uninfected cultures, cultures mock-infected with heat-killed virus (inactivated at 100°C for 10 min), or cultures stimulated with LPS (1 μg/ml). For staining, cells were fixed in acetone/methanol (1:1), blocked in 5% normal goat serum in PBS for 4 h at 4°C, and then incubated overnight at 4°C using mouse anti-TCRV 2-7-2 (1/500) and anti-GFAP Abs (1/1000). Following a 1-h incubation at room temperature with anti-mouse Cy3 (1/200), anti-rabbit Cy5 (1/400) cells were viewed using a LSM 5 PASCAL confocal microscope (Carl Zeiss) and then processed using Zeiss LSM (version 2.8) software.
Animals, infections, and treatments
BALB/c and B6 wild-type (WT) mice were obtained from the National Cancer Institute. B6 IFN-γ−/− and RAG1−/− mice were purchased from The Jackson Laboratory. All experiments were performed with genetically engineered mice backcrossed >8 generations and the controls consist of littermates or the parental WT strain. Mice were housed in sterile microisolator cages in the specific pathogen-free animal facility of the U.S. Food and Drug Administration’s Center for Biologics Evaluation and Research (Rockville, MD) and were bred at 6–12 wk of age. All experiments were approved by the U.S. Food and Drug Administration’s Animal Care and Use Committee (Rockville, MD).
Neonatal mice were infected with 2000 TCID50 of TCRV i.p. or intracranially(10 μl) 1–3 days after birth. Mice that died within 48 h after inoculation were excluded from the study. Uninfected mice that received saline were used as controls. For survival studies the mice were monitored daily, but infections were allowed to proceed to their natural outcome. Tissues and sera were collected under sterile conditions.
Infected mice were treated with the anti-TNF Ab 2E2 (50 μg/mouse i.p. in a 10-μl volume) starting on day 3 or day 7 postinfection (p.i.) every 3 days until weaning or with equal amounts of hamster IgG1. Where indicated (two studies), the treatment was delayed until hind leg paralysis was evident. In some studies, infected mice were treated with LEAF-purified rat anti-mouse TNF-α (XT122; BioLegend) or purified control rat IgG1. The therapeutic effect of both anti-TNF Abs was similar (100%). In selected experiments the animals received CpG ODN 1555 (50 μg/mouse i.p.) once on day 3 p.i. alone or together with the above-described anti-TNF-α treatment.
Brain pathology was assessed at 5 and 10 days p.i. as previously described (9). Briefly, BALB/c mice were euthanized and perfused intracardially with cold PBS followed by 4% paraformaldehyde. Serial sagittal sections (15 μm thick) were then collected at ∼400-μm intervals spanning an entire hemisphere and labeled with Abs to TCRV and GFAP, followed by anti-mouse Cy3 or anti-rabbit Cy5 as described above. Controls (not shown) included slides unstained and stained with isotype- and species-matched Abs or with secondary Abs alone. Slides were coded and analyzed by a “blinded” reader. Tissues were examined with a Zeiss LSM PASCAL laser confocal microscope. Viral Ag colocalization analysis was confirmed by acquiring Z stacks of selected areas with the ×63 objective. Where possible, brain sections immediately adjacent to those showing strong viral Ag expression were stained using H&E stains for assessing cellular infiltrates and changes to the tissue architecture.
Viral Ag-specific IgG Ab levels in sera were determined as previously described (9). Convalescent pooled sera from CpG ODN-treated, TCRV-infected survivors were used as a standard.
Nested PCR and real-time PCR
TCRV RNA was assessed by nested PCR analysis using primers located in the glycoprotein/nucleoprotein region and GAPDH as an internal control (see Table I). First, RT-PCR was performed with a Platinum PCR SuperMix kit (Invitrogen) and 200 nM each TCRV/GAPDH primer (Table I) as per the manufacturer’s instructions. The thermocycler conditions for the RT-PCR were as follows: 50°C for 30 min followed by 94°C for 2 min and 25 cycles of denaturation at 94°C for 20 s, annealing at 55°C for 30 s, and extension at 68°C for 30 s. For the nested PCR, 1 μl (1/30) of the RT-PCR product was added to 39 μl of PCR mixture consisting of 1× buffer II without MgCl2 (Applied Biosystems), 4 mM MgCl2, 1.2 mM each dNTP, 300 nM each TCRV-nested primer (Table I), and 1.25 U of AmpliTaq Gold DNA polymerase. The amplification conditions were 94°C for 5 min and 25 cycles of denaturation at 94°C for 15 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s. Amplification products were analyzed by 2% agarose gel electrophoresis in 1× TAE buffer (Tris-acetate and EDTA). Relative amount of the nested PCR amplicon was normalized using the housekeeping gene as reference upon scanning of the gel with a Bio-Rad gel scanner and using Quantity One 4.4.1 (Bio-Rad) and Image Gauge 4.1 software (Fuji Film).
|Gene .||Accession No.a .||Region .||Amplicon .||Primer Type .||Sequence (5′ to 3′) .|
|TCRV glycoprotein||M20304||618–862||245 nt||Forward||TTCAAGAGCTGATGGCAATG|
|TCRV glycoprotein||M20304||695–850||156 nt||Nested-forward||GAACCCTGTTTTGAGGGTGA|
|Gene .||Accession No.a .||Region .||Amplicon .||Primer Type .||Sequence (5′ to 3′) .|
|TCRV glycoprotein||M20304||618–862||245 nt||Forward||TTCAAGAGCTGATGGCAATG|
|TCRV glycoprotein||M20304||695–850||156 nt||Nested-forward||GAACCCTGTTTTGAGGGTGA|
From the European Molecular Biology Laboratory-European Bioinformatics Institute (EMBL-EBI) database.
Total RNA from murine cerebellum, whole brain homogenates, or primary glial cell cultures was extracted by using TRIzol reagent and reverse transcribed into cDNA using a First-Strand cDNA synthesis kit (GE Healthcare) or a High Capacity cDNA reverse transcription kit (Applied Biosystems) as per the manufacturers’ instructions. Relative mRNA levels for RANTES/CCL5, CD3ε, IFN-γ, MHC class II, perforin, and TNF-α genes were analyzed using the corresponding TaqMan gene expression assay by real-time PCR (Applied Biosystems), whereas IFN-α levels were determined using the SYBR Green PCR master mix kit (Applied Biosystems) and primers provided by Dr. K. Ozato (National Institutes of Health, Bethesda MD). Values for each target gene were normalized using 18S RNA or GAPDH. Expression values were calculated using the 2−ΔCt cycle threshold method (19) and expressed relative to the unstimulated control.
RNase protection assay
The mCK-5c multiprobe template set (BD Biosciences) was used to simultaneously detect transcripts of the mouse chemokine genes lymphotactin, RANTES, MIP-1α, MIP-1β, MIP-2, IP-10, MCP-1, and eotaxin, as well as the housekeeping genes L32 and GAPDH, which allow normalization. In vitro transcription was conducted using a BD Biosciences kit as per the manufacturer’s instructions with minor modifications. To facilitate precipitation of the labeled probes, 20 μg of oyster glycogen (Roche Molecular Biochemicals) was used. The probe set was labeled with [32P]UTP (3000 Ci/mmol; 10 μCi/μl; GE Healthcare). Target RNA (15 μg) was dried under a vacuum and resuspended in 15 μl of hybridization buffer supplemented with the probe set (4.8–7.6 × 105 cpm). An RNase protection assay (RPA) was performed as previously described (20). Dried RPA gels were exposed to Molecular Dynamics (GE Healthcare) PhosphorImager plates and plates were scanned with a Typhoon 9200 bioimaging analyzer (Molecular Dynamics/GE Healthcare). Photo-stimulated luminescence was analyzed with ImageQuant 5.2 software (Molecular Dynamics). Background noise was subtracted before normalization to L32 housekeeping gene expression. Values are expressed as arbitrary units.
Fold increases in mRNA expression were tested by ANOVA followed by either Tukey’s or Dunn’s multiple comparison posttest. Differences in survival were tested using the Kaplan-Meier method to determine survival fractions and the Mantel-Haenszel log rank test to determine p values. Only statistically significant differences are indicated in figures (*, p < 0.05; **, p < 0.01; ***, p < 0.001).
TCRV induces up-regulation of neuro-inflammatory cytokines and chemokines
WT neonatal mice infected with TCRV develop hind limb paralysis and seizures and die 12–18 days after challenge (Fig. 1,A). Amplification of viral RNA in the CNS using TCRV-specific primers (Table I) showed that the virus reached the cerebellum within 3 days of the challenge and then spread to the cerebrum, peaking 4–10 days later (Fig. 1,B). Natural history of the disease and viral load were similar in BALB/c or B6 mice (Fig. 1, A and C).
The infection led to the up-regulation of proinflammatory cytokines in the CNS (9). As shown in Fig. 1 D, mRNA for TNF-α, IFN-α, and RANTES became evident 5 days p.i., whereas up-regulation of IFN-γ mRNA was delayed until day 10 p.i. The increase in IFN-α was accompanied by significant up-regulation of several type I IFN-inducible genes including oas1a, pkr, and irf-7 (9) and IFN-inducible protein 1 with tetratricopeptides 1 and 3 (data not shown). Similarly, the up-regulation of IFN-γ and was associated with the up-regulation of IFN-γ-inducible genes including ifi30 and il-12rβ2, indicating that functional IFN-γ protein was being produced (data not shown). In contrast, no significant increase in IL-4, IL-15, or IL-17 was evident (data not shown). Mirroring viral progression, cytokine/chemokine mRNA up-regulation in whole brain started on day 7. Based on these data, we chose day 10, when viral loads were maximal but mortality was still low, as the optimal time point for comparing chemokine and cytokine expression in subsequent studies.
TCRV induces CNS cells to produce RANTES and IFN-α but not IFN-γ and TNF-α in vitro
T cells are known to cross the blood-brain barrier (BBB) during viral infections and pioneering studies on TCRV had shown that thymectomized mice survive an infectious challenge, suggesting that T cells play an important role in disease pathogenesis (8, 11). Despite this, H&E staining of cerebellum sections showed no significant cellular infiltration even in areas that had high levels of viral Ag (Fig. 1, F and G, bottom panels). Similarly, no significant increase in infiltrating CD4+ or CD8+ T cells could be identified by immunohistochemistry or flow cytometry on days 5 or 10 p.i. (data not shown). To determine whether T cells infiltrated the CNS in infected mice, mRNA levels for CD3ε were assessed 5, 7, and 10 days p.i. As shown in Fig. 1,E, no significant increase in CD3ε was observed before day 10 p.i., a time when most infected mice already show overt clinical signs of disease. Concurrent with increase CD3ε transcripts, there was increased expression of MHC class II (Fig. 1 E) but no evidence of increased expression of CD19 (B cell marker) or CD49b/DX5 (pan-NK cell marker) (data not shown), suggesting that by day 10 there is some T cell infiltration but no significant disruption in the permeability of the BBB.
To confirm the role for lymphocytes in viral pathogenesis, we infected RAG1−/− mice, which are genetically deficient in T and B cells, with TCRV. Unlike their infected B6 WT counterparts (or μMT mice lacking mature B cells; Ref. 9), RAG1−/− mice showed 100% survival (Fig. 2,A). This increased resistance was not due to differences in viral titers (Fig. 2,B), as both strains showed similar viral loads in brain by day 10 postinfection. In contrast, RAG1−/− mice had significantly lower levels of IFN-γ and TNF-α transcripts (175- and 3-fold, respectively, p < 0.05; Fig. 2 C), suggesting that the increased survival was secondary to a reduced inflammatory response to the virus.
To further discern the role of local and infiltrating cells in the elicited immune response, an ex vivo culture system of primary brain cells was used. These cultures consisted primarily of astrocytes and microglia but contained neurons and some oligodendrocytes. The virus replicated in these cultures, reaching levels of 107 TCID50/ml after 4 days (Fig. 3, A and B). The infection of the culture was followed by the up-regulation of mRNA for RANTES and IFN-α (Fig. 3,C). In contrast, no mRNA for TNF-α or IFN-γ (Fig. 3,C and not shown) were evident in these cultures, suggesting that these cytokines are primarily produced by peripheral or infiltrating cells. LPS stimulation of these same cultures rapidly induced high levels of TNF-α mRNA (Fig. 3 C) and protein (not shown), confirming that the cultures are able to produce this cytokine.
Improved survival in TCRV-infected mice lacking IFN-γ
Because mice that lack T (and B) cells survive the infection and have fewer brain transcripts for IFN-γ and TNF-α, we next investigated the role of these cytokines in the pathogenesis of TCRV meningoencephalitis. IFN-γ, a pleiotropic cytokine secreted by activated T (CD8+ and CD4+ Th1 cells) and NK cells, is known to play important roles in noncytopathic CNS virus infections because it regulates inducible NO synthase (iNOS), TLR expression, MHC, perforin, and Fas expression on target cells and T cell homeostasis (21, 22, 23). As shown in Fig. 4, mice lacking IFN-γ (IFN-γ−/−) challenged with TCRV showed weaker up-regulation of RANTES and CD3ε and no increase in perforin transcripts (data not shown) but higher levels of IFN-α than WT controls. Despite a tendency to have lower viral loads in CNS, (Fig. 4, A and B), only 36% of infected mice survived the challenge, indicating that the pathogenic response to the virus can proceed in the absence of IFN-γ or perforin.
Role of TNF-α in TCRV-induced meningoencephalitis
TNF-α is a multifunctional proinflammatory cytokine pivotal in the regulation of the host response during infection and inflammation (24, 25). Studies of West Nile virus encephalitis had suggested that the early induction of TNF-α plays an important role in increasing the permeability of BBB, allowing for viral entry into the CNS (26). Because TCRV-infected animals have increased TNF-α levels starting early in the disease, we next explored its role in establishing CNS viral infection and subsequent disease progression. Infected mice were treated i.p. with neutralizing mAb to TNF-α or control Ab every 3 days, starting at day 3 or day 7 p.i. until the mice were weaned (day 21). These time points were chosen to represent the stage when the virus first enters the CNS (day 3) or a time when the virus is well established in the CNS and symptoms of encephalitis first become evident (day 7) (Fig. 1,B). As shown in Fig. 5,A, all mice treated with anti-TNF-α mAb starting at day 3 survived the infection (as did 90% of those treated starting on day 7). In comparison, mice that received control Ab perished (Fig. 5,A). Of note, the increased survival did not result from improved viral clearance or reduced infection of the CNS, as treated mice tended to have higher viral loads than the untreated ones (Fig. 5,B). Furthermore, the administration of anti-TNF-α Abs significantly prolonged survival in mice challenged intracerebrally with TCRV (Fig. 5,C). This suggested that, in this model, the mechanism by which TNF-α contributes to TCRV pathogenesis is not by increasing BBB permeability and allowing for viral infection of the CNS as has been suggested for West Nile virus (26). Similarly improved survival was evident in B6 WT and B6 IFN-γ−/− mice (Fig. 5,D) lending further support to the key role of TNF-α in TCRV-induced meningoencephalitis. Importantly, improved clinical outcome was evident even when the anti-TNF-α Ab treatment was started after the mice had overt clinical signs of meningoencephalitis (day 13). Indeed, anti-TNF-α Ab treatment in mice with complete paralysis of the hind legs resulted in significantly improved motility and function within days (Fig. 6 and supplemental videos).5
Anti-TNF-α treatment reduces Tacaribe-induced inflammatory responses in the CNS
The mechanism by which anti-TNF-α Abs protect mice from lethal meningoencephalitis was next examined. RPA analysis showed that CNS of treated mice had significantly reduced transcript levels for RANTES/CCL5 (p < 0.05), lymphotactin (p < 0.05), IP-10/CXCL-10 (p < 0.05), and MIP-2 (p < 0.01) at day 10 p.i. as compared with uninfected controls. MIP-1α, MIP-1β, and MCP-1 were also impacted but to a lesser extent (Fig. 7,A). Consistent with this reduction in chemokine expression, treated mice had reduced numbers of infiltrating T cell in CNS (p < 0.001) and tended to have lower MHC class II expression (p = 0.05) (Fig. 7,B). The studies of TCRV-infected primary CNS cultures had suggested that the virus triggers RANTES/CCL5 and IFN-α production by local cells while the increase in TNF-α and IFN-γ levels required infiltrating peripheral cells. Consistent with this, anti-TNF-α-treated mice had significantly reduced expression of IFN-γ mRNA (p < 0.001) and reduced expression of TNF-α as compared with untreated mice (Fig. 7 C), whereas no impact on transcripts levels for IFN-α or perforin were evident. Together, these data demonstrate that TNF-α plays a key role in the regulation of the inflammatory response to the virus.
Anti-TNF-α-treated mice develop a chronic viral infection
Most immune responses to viruses result in virus elimination; however, incomplete responses can result in chronic viral infection. Early in the disease, anti-TNF-α treated mice have viral loads at least as high as their untreated counterparts. To explore whether mice that survive the infection are able to eventually clear the virus, the viral load of infected mice was assessed at the end of the anti-TNF-α treatment (day 25) and 1 mo later (day 55). As shown in Fig. 8 A, treated mice did not clear the infection regardless of whether they were treated starting on day 3 or day 7 of infection (107–109 TCID50 at weaning). Of note, viable virus was detectable in treated mice as late as 14 mo p.i. (not shown), suggesting that these animals become chronically infected.
The innate immune system has the inherent capacity to sense a wide spectrum of infectious agents through a series of germline-encoded proteins known as TLRs. Synthetic CpG ODNs act on TLR9 to trigger an immunostimulatory cascade that is characterized by B cell proliferation, dendritic cell maturation, NK cell activation, and the secretion of a variety of cytokines, chemokines, and polyreactive Igs. Previous work from our laboratory had shown that treatment with 1 dose of CpG ODN up to 3 days after a TCRV challenge improved survival rates (30–50%). The improved survival was associated with the accelerated development of TCRV-specific Abs, and all surviving animals completely cleared the virus by time of weaning (9). To test whether the addition of a TLR ligand would improve viral clearance in anti-TNF-α-treated mice, infected mice were treated with a single dose of CpG ODN on day 3 p.i. plus a regimen of Abs to TNF-α as described above. As shown in Fig. 8,B, the addition of a CpG ODN dose (50 μg/mouse) to the anti-TNF-α therapy was not detrimental to survival (100%, as compared with 30% survival in mice treated with CpG ODN alone) or to impact on the IFN-γ, TNF-α, and CD3ε transcript levels in brain of anti-TNF-α-treated mice (Fig. 8,D) but did accelerate the development of Abs to TCRV (Fig. 8,C; p < 0.05). The accelerated development of anti-TCRV Abs was associated with viral clearance (Fig. 8 E; p < 0.01). Thus, combining anti-TNF-α Abs with an innate immune response inducer resulted in 100% survival and improved control of viral load in CNS. This suggests that an optimal therapeutic regimen may need to combine blocking the inflammatory response to the virus with improved innate immune activation and accelerated Ab development.
The central role of the immune system in the CNS is to contain and control the spread of infection. Paradoxically, however, this response may be more pathogenic than the offending pathogen. We report that modulating the inflammatory response that follows a challenge with neurotropic TCRV by administering neutralizing Abs to TNF-α plus CpG ODN protects 100% of mice from lethal meningoencephalitis. Protection was evident even in mice with overt signs of neurologic damage. The improved survival was not secondary to delayed or reduced viral tropism to the CNS or to improved control of viral replication. Instead, blocking TNF-α reduced the local production of chemokines, and subsequent T cell infiltration and was associated with lowering the CNS levels of IFN-γ and TNF-α transcripts. Together, these studies show that TNF-α plays a key role in amplifying the neuro-inflammatory response to TCRV that is necessary for viral clearance and that blocking TNF-α results in improved survival but persistent infection. Of note, when the Abs to TNF-α were administered to animals that received an innate immune response stimulant (CpG ODN) early in infection, both, survival and viral clearance were significantly improved. These data provide evidence for the first successful therapeutic approach to late interventions for a New World arenavirus.
New world arenaviruses of the Tacaribe serocomplex (which includes Junin, Machupo, Guaranito, and Sabia viruses) primarily cause devastating and often lethal hemorrhagic fevers. However, virus has been also isolated from cerebrospinal fluid and neurological signs, including depression, ataxia, tremors, convulsions, paresis, and muscle atrophy, have been described during acute infection and in convalescent patients (27, 28, 29). The mechanism by which arenaviruses cause neurological symptoms is not well understood but could be secondary to coagulation disorders or to direct or Ab-mediated neurotoxicity, or, as suggested by the animal models, could result from the cytokine cascade elicited by the virus. Indeed, elevated levels of TNF-α and IFN-α were reported in patients with Argentine hemorrhagic fever (30).
In mice, New World arenavirus-induced encephalitides are characterized by mild gliosis, little cellular infiltration, and delayed disruption of the cerebellar architecture, which do not seem to justify, per se, their lethality (Fig. 1, F and G) (31). Despite the paucity of infiltrating cells, early observations that thymectomized and nudenu/nu mice survive TCRV infection suggested a role for T cells (8, 10, 11) that was confirmed in this study using RAG−/− mice (Fig. 2). The absence of apparent cellular infiltrates and low CD3ε transcript levels suggest that the T cells might localize to the perivascular space rather than penetrate the brain parenchyma and would consequently be lost during perfusion (hence not evident by immunohistochemistry). The precise mechanism by which T cells cause tissue damage and death is not fully understood; however, as shown above, T cell infiltration occurs after the onset of clinical disease and coincides with a massive up-regulation of chemokines and cytokines that amplify the inflammatory response. Also unknown at this time is whether CD8+ T cells are critical to the pathogenesis of TCRV as was suggested by studies with the arenavirus lymphocytic choriomeningitis virus (32). Indeed, our preliminary studies suggest that the presence of either CD4+ or CD8+ T cells is sufficient to cause pathology and death (J. A. Pedras-Vasconcelos, and D. Verthelyi, manuscript in preparation).
IFN-γ and TNF-α are considered as critical regulators of T cell-mediated inflammation. Both appear to play key roles in the development of encephalitis, as mice deficient in either have improved survival (Fig. 4 and 5 and data not shown). In contrast, mice lacking IL-12, IL-10, or IL-4 showed no change in their clinical course (data not shown). Studies with the lymphocytic choriomeningitis virus show that IFN-γ-producing CD8+ CTLs mediate cell death in a perforin-dependent mechanism, whereas TNF-α has a minor impact on disease (12, 32, 33). In contrast, in Tacaribe meningoencephalitis IFN-γ-induced perforin plays a secondary role in pathology because infected IFN-γ−/− mice, which do not up-regulate perforin upon infection, showed a minor improvement in survival while those receiving neutralizing Abs to TNF-α survived 100% despite no significant reduction in perforin levels (Fig. 7 C). This suggests that the pathogenic roles of IFN-γ and TNF-α are distinct for Old and New World arenaviruses.
TNF-α levels in sera and CSF of patients with viral encephalitis such as herpes simplex and Japanese encephalitis virus are often elevated (15, 16, 17, 34). Similarly, increased TNF-α levels are evident in West Nile- and Sindbis-infected mice. Previous reports show that astrocytes and microglia are capable of producing TNF-α (13, 35); however our studies show low TNF-α levels before cellular infiltration and no TNF-α produced by primary cultures infected with TCRV (Fig. 3 C). This suggests that the primary sources of TNF-α early in the disease are likely APCs lining the choroidal plexus, the meninges, or the perivascular space whereas later, infiltrating cells such as macrophages, granulocytes, NK, and/or CD4+ T cells play a role in producing TNF-α (36).
The role of TNF-α in encephalitis is controversial. In herpes simplex, blocking TNF-α was shown to increase viral titers and mortality (16, 17). In other viral encephalitides TNF-α was shown to play a role in neuronal injury, promote astrocyte proliferation and gliosis, and increase the BBB permeability by inducing iNOS and the generation of higher reactive oxygen species (9, 15, 37, 38). In TCRV-infected mice, TNF-α appears to mediate increased expression of chemokines and proinflammatory cytokines and also cellular infiltration (Fig. 7 and data not shown). Blocking TNF-α reduced the expression of iNOS in CNS; however, preliminary studies show that anti-TNF-α Ab treatment improves the survival of iNOS −/− mice (data not shown), suggesting that regulating iNOS expression is not the mechanism by which anti-TNF-α Abs improve outcome. Moreover, the protective mechanism of anti-TNF-α appears to be to reduce neuronal cytotoxicity and any subsequent exacerbation of the neuro-inflammatory response without affecting the tropism of the virus, because anti-TNF-α treatment prolonged the survival of mice infected intracranially. Further, although higher levels of CD3ε transcripts are present in TCRV-infected mice, the following observations were made: 1) disruption of the BBB was not noted (by using the Evans blue method) in infected animals until after they showed hind limb paralysis (not shown); and 2) treatment with anti-TNF-α improved the outcome very late in the disease. This suggests that the protective effect of anti-TNF-α Abs is not limited to reducing cellular infiltration. Lastly, administration of anti-TNF-α did not appear to impact the early stages of the innate anti-viral response of the resident cells, as IFN-α levels were similar in treated and untreated animals (Fig. 7 C). Of note, although IFN-α is known to limit viral growth, in TCRV encephalitis the response induced was not sufficient to control viral replication because mice treated with anti-TNF-α became chronically infected with titers of 107–109 TCID50 per brain for months after infection (data not shown).
Previous studies had shown that treatment with CpG ODN improves the response to multiple pathogens, including TCRV (9, 39). Although CpG ODN alone is not fully protective, combination therapy using CpG ODN plus anti-TNF-α Abs was effective in improving survival (Fig. 8,B) and accelerated the induction of virus-specific Ab production (Fig. 8,A) and viral clearance, therefore avoiding chronic infection (Fig. 8 E). These data are in line with our previous observation in which CpG ODN appeared to both accelerate the rate and modify the quality of the anti-TCRV Abs (9). In those studies, μMt mice (lacking mature B cells) treated with sera derived from CpG ODN-treated mice had a higher survival rate as compared with those treated with sera from convalescent mice that were not treated with CpG ODN. Although the precise role of anti-TCRV Abs in viral clearance is not yet clear, development of the Abs early in infection appears to be critical, as empirical data show that convalescent sera are effective in patients with hemorrhagic fevers only when administered during the first days of infection.
The last few decades have shown an increase in the number of highly pathogenic RNA viruses that cause encephalitis, including a significant number of emerging or reemerging zoonotic viruses that are also considered potential biowarfare agents (40, 41). There are no effective therapeutics for containing and treating diseases caused by any of these viruses, and there is limited, if any, vaccine available for most (41, 42). The availability of antiviral drugs directed at these viruses would provide treatment and a strong deterrent against their use as biowarfare agents. Early administration of convalescent sera, ribavirin, and CpG ODN treatments have shown partial efficacy against some arenavirus infections (9, 29, 43). Given their severe morbidity and high mortality, compounds that could be used for containment and treatment need to be identified. Several viral encephalitides are characterized by increased TNF-α levels in CNS. TNF-α inhibitors, including mAbs to TNF-α and soluble TNF-α receptors have been licensed in the last few years for the treatment of autoimmune diseases (44). As shown above, the use of neutralizing anti-TNF-α Abs results in the survival of TCRV-infected mice even when administered late in disease after the mice develop hind leg paralysis. Although it remains to be seen whether anti-TNF-α therapy will ameliorate infection with lethal Junin, Machupo, Guanarito, or Sabia viruses and its impact on the hemorrhagic component of the disease, these findings suggest that this strategy may hold promise as a therapeutic tool for the neurologic effects of pathogenic arenavirus and other virus that induce high levels of pathogenic TNF-α in the CNS. Of note, although human to human transmission of this rodent-born virus is rare, the persistent viral infections would constitute a major health concern in an outbreak. Further, several studies have suggested that persistent viral infections may manifest in adults as neurodegenerative and cognitive disorders (45, 46). This suggests that a combination of anti-TNF-α Abs with other antiviral agents or immunomodulatory drugs such as CpG ODN may be required for optimal protection of infected subjects.
We thank the personnel of the animal facility for care of the mice and Dr. M. J. Buchmeier (of the Scripps Research Foundation, La Jolla, CA) for providing the Abs to TCRV. We thank Vivian Wang and Lucja Grajkowska for technical assistance. We also thank Drs. Amy Rosenberg, Felix Yarovinsky, and Lewis Markhoff for careful review of the manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
All experiments were approved by the FDA Animal Care and Use Committee. The assertions herein are the private ones of the authors and are not to be construed as official or as reflecting the views of the Food and Drug Administration.
Abbreviations used in this paper: TCRV, Tacaribe arenavirus; BBB, blood-brain barrier; B6, C57BL/6; Cy, carbocyanin; GFAP, glial fibrillary acidic protein; iNOS, inducible NO synthase; ODN, oligodeoxynucleotide; p.i., post-infection; RPA, RNase protection assay; TCID50, 50% tissue culture-infective dose; WT, wild type.
The online version of this article contains supplemental material.