|Home | About | Journals | Submit | Contact Us | Français|
The major inducible 70-kDa heat shock protein (hsp70) protects against measles virus (MeV) neurovirulence in the mouse that is caused by a cell-associated noncytolytic neuronal infection. Protection is type I interferon (IFN) dependent, and we have established a novel axis of antiviral immunity in which hsp70 is released from virus-infected neurons to induce IFN-β in macrophages. The present work used vesicular stomatitis virus (VSV) to establish the relevance of hsp70-dependent antiviral immunity to fulminant cytopathic neuronal infections. In vitro, hsp70 that was constitutively expressed in mouse neuronal cells caused a modest increase in VSV replication. Infection induced an early extracellular release of hsp70 from viable cells, and the release was progressive, increasing with virus-induced apoptosis and cell lysis. The impact of this VSV-hsp70 interaction on neurovirulence was established in weanling male hsp70 transgenic and nontransgenic mice. Constitutive expression of hsp70 in neurons of transgenic mice enhanced viral clearance from brain and reduced mortality, and it was correlated with enhanced expression of type I IFN mRNA. Nontransgenic mice were also protected against neurovirulence and expressed increased type I IFN mRNA in brain when hsp70 was expressed by a recombinant VSV (rVSV-hsp70), indicating that hsp70 in the virus-infected cell is sufficient for host protection. In vitro data confirmed extracellular release of hsp70 from cells infected with rVSV-hsp70 and also showed that viral replication is not enhanced when hsp70 is expressed in this manner, suggesting that hsp70-mediated protection in vivo is not dependent on stimulatory effects of hsp70 on virus gene expression.
Viral encephalitides can be caused by either cytopathic or noncytopathic cell-associated (persistent) infection of neurons. Defining mechanisms underlying host protective immunity that are common to both types of virus-host relationships may enhance our ability to define basic determinants of neurovirulence. Noncytolytic infection of neurons of the brain is exhibited by members of multiple virus families, including Paramyxoviridae, Picornaviridae, Retroviridae, Arenaviridae, Coronaviridae, and Togaviridae. Spread within the central nervous system is mediated by neuroaxonal transport and direct cell-to-cell transmission in the absence of significant cell-free infectious viral progeny that may in part reflect restrictions in virus budding and the overall level of virus gene expression (reviewed in reference 1). Such a virus-host relationship poses a challenge to viral clearance that is dependent upon the local activation of virus-specific T cells that have been primed in the periphery. Activation of virus-specific T cells requires encounter with viral antigen-major histocompatibility complex class I (MHC-I) complexes and type I interferon (IFN) (2, 3), which is predominantly IFN-β in the brain (4). The challenge to clearance is that neurons are restricted in their expression of MHC-I (5, 6), and it is macrophages, not neurons, that are the primary source of type I IFN in the virus-infected brain (7). Therefore, effective clearance requires that virus-infected neurons activate uninfected brain macrophages (microglia) to cross-present viral antigen and to produce IFN-β.
Measles virus (MeV) exhibits a noncytopathic cell-associated neuronal infection. The mouse model was used to elucidate a novel axis of antiviral immunity whereby virus-infected neurons induce IFN-β in uninfected brain macrophages (i.e., microglia) through the major inducible 70-kDa heat shock protein (hsp70). hsp70 is recognized as an intracellular chaperone that may support virus replication (reviewed in reference 8), but hsp70 can also be released from cells to act as a damage-associated molecular pattern (DAMP) capable of activating innate immunity (9, 10). In vitro studies show that MeV infection causes an early release of hsp70 from viable neuronal cells, consistent with exosomal secretion or microvesicular shedding, and the extracellular hsp70 serves as a potent stimulus for IFN-β expression in mouse macrophages, including microglia (11). Induction of IFN-β by hsp70 is mediated by Toll-like receptors 2 and 4 (TLR2 and TLR4, respectively), which in the brain are expressed predominantly on macrophages (12, 13). In vivo, selective neuronal expression of hsp70 enhances the IFN-β response to intracranial MeV inoculation, reducing mortality in H-2d congenic C57BL/6 mice (11). The effectiveness of this hsp70-mediated antiviral immunity may help to explain the low incidence of MeV-induced neurological disease in humans. Constitutive expression of hsp70 is characteristic of humans and other primates, in contrast to the absence of significant constitutive expression in mice (14, 15). Furthermore, febrile responses define clinical cases of MeV infection in humans (16), with fever being a potent stimulus for hsp70 induction (17).
The role that hsp70 plays in the outcome of cytopathic and productive neuronal infections of brain is unknown. In mice, cytopathic and productive neuronal infections include encephalitis caused by vesicular stomatitis virus (VSV) (18, 19), herpes simplex virus (HSV) (20), and H5N1 influenza virus (21). These infections are characterized by the formation of cell-free infectious viral progeny that may enhance both direct cell-to-cell transmission during neural spread and dissemination through extracellular (e.g., cerebrospinal fluid) pathways. Cytopathic infections may also alter the release of hsp70 that serves as an induction stimulus for IFN-β. For VSV and HSV, neuronal apoptosis is considered an important mediator of neuropathogenesis (22, 23). In tumor cell lines, hsp70 release has been associated with necrosis but not apoptosis (24–26). Release of hsp70 has not been examined in the context of virus-induced apoptosis. These considerations may influence the balance between hsp70-mediated enhancement of virus replication within the cell and hsp70-mediated type I IFN-dependent antiviral immune responses in cytopathic and productive neuronal infection. A weakened immune response combined with hsp70-mediated stimulation of virus gene expression within cells can result in enhanced neurovirulence (27).
The present work defined the effectiveness of hsp70-mediated type I IFN induction and protection against VSV neurovirulence in the mouse. In this system, the level of virus replication in brain following intranasal inoculation determines the level of cytopathic effect and mortality (28), and neurovirulence is regulated by the expression of IFN-β (29). For these infections, virus spreads from the olfactory mucosa through the olfactory nerves to infect the olfactory bulbs of the brain and is followed by widespread dissemination throughout the brain that results in death. Studies using the mouse N2a neuronal cell line showed a modest enhancement of VSV replication in cells that overexpress hsp70. However, in mice, overexpression of hsp70 accelerated VSV clearance and decreased mortality. This appeared to be due to enhanced innate immune responses in the brains of infected mice as assayed by IFN-β and TLR4 mRNA levels.
The Institutional Animal Care and Use Committee for The Ohio State University provided supervision of animal care for all aspects of the study, including review and approval of the experimental protocols, in accordance with the Public Health Service Act of the United States of America. Animal care facilities are AAALAC accredited.
Construction and recovery of wild-type (Indiana strain) recombinant VSV (rVSV) and rVSV-VP1 were performed as described previously (30). Recombinant rVSV-VP1 harbors the VP1 gene of human norovirus at the VSV genome junction between the G and L genes. Also generated was rVSV containing hsp70 (product of the HSPA1A gene) at the VSV G/L gene junction (rVSV-hsp70). A plasmid containing the genome cDNA, pVSV1(+) GxxL, was kindly provided by Sean Whelan. The hsp70 gene was amplified by high-fidelity PCR and cloned into pVSV(+)GxxL at SmaI and XhoI sites, which resulted in the construction of plasmid pVSV1(+)-hsp70. All genome cDNA constructs contained the VSV gene start and gene end sequences, and fidelity of the subcloning was confirmed by sequencing. To recover infectious virus from the genome cDNAs, BSRT7 cells were infected with a recombinant vaccinia virus (vTF7-3) expressing T7 RNA polymerase, followed by cotransfection with a plasmid encoding the VSV genome [pVSV1(+)-HSP70] and plasmids pN, pP, and pL. At 96 h posttransfection, cell culture supernatants were filtered through a 0.2-μm filter, and the recombinant VSV was further amplified in BSRT7 cells. Subsequently, individual plaques were isolated and seed stocks were amplified in BSRT7 cells as described previously (31).
N2a cells have been stably transfected to constitutively express hsp70 (N2a-HSP) together with vector-transfected controls (N2a-V) (32). Triplicate cultures of N2a-HSP or N2a-V cells were infected at a multiplicity of infection (MOI) of 0.01 or 10 with rVSV, rVSV-hsp70, and rVSV-VP1. Tissue culture medium (Dulbecco's modified Eagle medium [DMEM] containing 5% fetal calf serum) was harvested at defined intervals, clarified by centrifugation at 1,000 × g for 15 min, and analyzed for infectious viral progeny, hsp70 release, and cell lysis. The virus titer was determined by a plaque assay performed in Vero cells. Cell lysis was based upon lactate dehydrogenase (LDH) release, as described previously, using a commercially available assay (Biovision) (11). Maximal LDH release was defined by cells treated with 1% Triton X-100 for 30 min. hsp70 concentration in culture supernatants was determined using a commercially available enzyme-linked immunosorbent assay (ELISA) (hsp70 high-sensitivity enzyme immunometric assay; Enzo Life Sciences) as described previously (11). The standard curve was established using serial dilutions of purified human recombinant hsp70-1 (product of the HSPA1A gene). The primary antibody used for detection is a rabbit polyclonal antibody recognizing human, rat, and mouse hsp70 and does not cross-react with other hsp70 family members (e.g., hsc70).
Levels of virus gene expression and induction of apoptosis was based upon Western blot analysis of total cell protein. Detached and adherent cells (harvested by scraping) were pelleted at 1,000 × g for 15 min at 4°C and lysed with radioimmunoprecipitation assay (RIPA) buffer (Santa Cruz Biotechnology). Twenty-five μg of total protein was resolved by electrophoresis on 10% bis-Tris polyacrylamide gels (Life Technologies) and transferred to nitrocellulose membranes. Membranes were probed with mouse monoclonal antibodies specific to VSV G (Sigma) and hsp70 (Enzo Lifescience), rabbit polyclonal antibody against casapase-3 (Cell Signaling Technology), and rabbit monoclonal antibody against β-actin (Cell Signaling Technology). Primary antibodies were detected with horseradish peroxidase-labeled anti-mouse (BD Biosciences) or anti-rabbit (Cell Signaling Technology) IgG. Signal was detected by chemiluminescence (Superwest pico chemiluminescence detection kit; Life Technologies). Protein band intensity was quantified by scanning and analysis with KODAK molecular imaging software.
H-2d congenic C57BL/6 mice that constitutively overexpress human hsp70 (from the HSPA1A gene) under the control of the neuron-specific enolase promoter have been previously described and are referred to here as hsp70-transgenic (TG) mice (27, 32). Male TG and nontransgenic (NT) mice, 4 to 6 weeks of age, were inoculated intranasally with 1 × 106 PFU of rVSV in 1× phosphate-buffered saline (PBS) for a total volume of 30 μl. The challenge dose was based upon preliminary studies in which the 50% lethal dose (LD50) for rVSV was shown to be 1 × 105 PFU. All animal experiments included control mock-infected mice that were inoculated with 1× PBS. Data from these controls are presented in the IFN analysis but not the mortality analysis or analysis of brain viral burden. For survival analyses, weight change and clinical symptoms were monitored until 15 days postinfection. Removal criteria included a weight loss of >20% and were typically associated with dehydration, social withdrawal, and paralysis. Kaplan-Meier survival curves were constructed, and Wilcoxon and log-rank tests were used to define statistically significant differences in survival between treatment groups. The advantage of this approach over comparisons based on LD50 measurements include the ability to compare survival times in addition to overall mortality rates and the more judicious use of animals.
Brains were harvested at defined intervals for viral titration. The left half of the brain was homogenized at 5,000 rpm for 15 s at 4°C using a Precellys bead beating homogenizer (Cayman chemical) and 1.4-mm ceramic beads. Virus titers in homogenates were determined by a plaque assay on Vero cells.
Real-time reverse transcription-quantitative PCR (RT-qPCR) analysis of total brain RNA was used to quantify IFN-β and TLR4 transcripts at 1 day p.i. in the brain, using 4 animals per group. For this purpose, approximately 100 mg of brain rostral to bregma 1.32 and including the olfactory bulbs was harvested at 1 day p.i. and processed for total RNA isolation using the RNeasy lipid tissue minikit (Qiagen), followed by DNase I treatment (Turbo DNA-free; Ambion). Five μg of total RNA was denatured at 70°C for 5 min and annealed to 0.4 μg of oligo(dT) primers. Primer-annealed RNA was added to a reverse transcriptase reaction cocktail that included 25 nmol of deoxynucleoside triphosphate (dNTP) and 1 μl of AffinityScript multiple-temperature reverse transcriptase (Stratagene) in a final volume of 20 μl. The cDNA reaction mixtures were incubated for 90 min at 48°C. The level of IFN-β transcript cDNA was quantified by SYBR green real-time qPCR using a Roche 480 LightCycler as described previously (11). Levels of TLR4 transcript cDNA were analyzed using 5′-CGCTCTGGCATCATCTTCATTG-3′ sense and 5′-CCTCAGGTCCAAGTTGCCGTTTC-3′ antisense primers (33). The qPCR was performed using the LightCycler 480 SYBR green I master mix (Roche). The reaction program included preincubation at 95°C for 10 min and then 45 cycles of amplification at 95°C for 10 s, 60°C for 12 s, and 72°C for 15 s. Total SYBR green fluorescence was measured at 72°C. Quantifications were based upon fit-point analysis with arithmetic baseline adjustment. Melting peak and melting curve analyses were performed using the polynomial calculation method.
Statistical analysis of in vivo and in vitro studies included two-tailed unpaired t test (t test) and one-way analysis of variance (ANOVA). Bonferroni's multiple-comparison test was used to compare pairs of groups as the posttest of ANOVA. A P value of <0.05 was considered statistically significant. All experimental results were reproducible, and representative results are illustrated.
To determine the effect of hsp70 on viral replication, we used established mouse neuroblastoma cell lines that either lack constitutive hsp70 (N2a-V) or constitutively express hsp70 (N2a-HSP) (11, 32). N2a-HSP and N2a-V cells were infected with rVSV at a low multiplicity of infection (MOI of 0.01). Infectious viral progeny release and viral G protein expression were measured at 8, 16, and 24 h p.i., with a goal of correlating infection parameters to patterns of extracellular hsp70 release (see below). Significant increases in viral progeny release were observed in N2a-HSP relative to N2a-V cells at all three time points (Fig. 1A). Western blot analysis of total cell protein showed that VSV G was detectable at 16 and 24 h p.i. in both N2a-V and N2a-HSP cells, although the level was increased in N2a-HSP cells (Fig. 1B). The β-actin-adjusted levels of G were statistically significantly different at 24 h p.i. (P < 0.05 by t test).
Infections were repeated at an MOI of 10 (i.e., single-cycle conditions). Expression of VSV G protein was detected at 8, 16, and 24 h p.i. Small but statistically significant increases in G protein levels were detected in N2a-HSP relative to Na-V cells at each time point (Fig. 2B). Differences in infectious progeny release were not observed under these conditions. Collectively, results indicate a modest stimulatory effect of hsp70 on VSV replication, with increases in viral G expression consistently observed in both multistep and single-cycle infections of N2a cells.
Release of hsp70 into culture supernatants was measured by ELISA in N2a-HSP and N2a-V cells at 8, 16, and 24 h p.i. with rVSV (MOI of 0.01). Increases in extracellular hsp70 levels were demonstrated in N2a-HSP and N2a-V cells at all three time points, with the level of release being significantly greater in the infected N2a-HSP than N2a-V cells (P < 0.05 by ANOVA) (Fig. 3A). By 24 h p.i., the extracellular hsp70 concentration for infected N2a-HSP cells approximated that observed when N2a-HSP cells were lysed with detergent. The source of hsp70 in the N2a-V cells may reflect virus-induced endogenous hsp70 (34), which was previously reported for Edmonston MeV in this cell line (35). These low levels of virus-induced endogenous hsp70 are not apparent on Western blots used to show the abundant hsp70 that is present in N2-HSP cells (Fig. 1B).
Release of hsp70 was compared to the onset of cell lysis based upon the release of LDH into culture supernatants and the onset of apoptosis based on detection of cleaved caspase 3 in total cell protein by Western blotting. There was no rise in extracellular LDH levels above the background release from uninfected cells at the 8-h time point, indicating that hsp70 release preceded cell lysis. Floating and dead cells were detected at 16 and 24 h p.i. in both cell lines by light microscopy, accompanied by significant increases in LDH levels in culture supernatants and evidence of caspase 3 activation, consistent with the cytopathic nature of VSV infection (Fig. 3B and andC).C). There were no significant differences in the level of LDH release between infected N2a-hsp70 and N2a-V cells at 16 and 24 h p.i. The level of cleaved caspase-3 was significantly greater in infected N2a-HSP cells than infected N2a-V cells at both 16 and 24 h p.i.
To determine whether hsp70 has a protective effect against VSV neurovirulence, weanling male NT (n = 11) and TG (n = 10) mice were inoculated intranasally with 1 × 106 PFU of rVSV. For survival analyses, animals were monitored daily over a 15-day time course. Previous studies show that virus-induced mortality is not observed after 15 days p.i. (30). A body weight loss of >20% relative to that at day 0 was the best indicator for early removal and was associated with dehydration, social withdrawal, and paralysis. The first rVSV-infected NT animals had to be euthanized at 7 days p.i. and the last at 9 days p.i., with an overall mortality of 91%. In contrast, TG mice showed only 50% mortality that occurred between 7 to 8 days p.i., a significant difference between groups (P < 0.05) (Fig. 4A).
Mortality was inversely correlated with viral clearance at 7 to 8 days p.i. based upon a temporal analysis of viral titer in brain homogenates. The left half of the brain was harvested from animals during the early phase of infection at 3 days p.i., the period of peak brain viral titers at 6 to 7 days p.i., the period of viral clearance at 8 to 9 days p.i., and in animals surviving to 15 days p.i. Both NT and TG mice had readily detectable levels of virus in brain at 3 days p.i. (Fig. 4B). Levels were one-half log higher in TG mice, although the difference from NT mice was not statistically significant. Mean brain viral titers remained relatively constant in TG mice from 3 to 7 days p.i., whereas titers continued to rise in NT mice, reaching a peak at 7 days p.i. that was statistically significantly different from the brain viral titer of TG mice (P < 0.05 by t test). Reductions in titer ensued for animals surviving beyond 7 days p.i., with mean titers being significantly less in TG mice. Clearance was detected in 3 of 4 TG mice that survived to 8 days p.i. compared to 0 of 4 NT mice. Infectious virus was not detected in mice surviving to 9 and 15 days p.i. Results show that hsp70 does not suppress virus replication at early times of infection. Instead, the hsp70 protective effect reflects enhanced viral clearance.
Protection of mice against VSV neurovirulence is dependent upon type I IFN (36), and VSV is a poor inducer of type I IFN in vitro and in the central nervous system of susceptible mice (29, 37). Expression of IFN-β at 1 day p.i. was directly measured by real-time RT-qPCR of total RNA isolated from rostral portions of brain, representing infected and uninfected TG and NT mice. Significant induction of IFN-β transcripts was observed only in infected TG mice relative to uninfected controls (P < 0.05 by ANOVA) (Fig. 5). Induction in infected relative to uninfected NT mice was not observed. hsp70 induces IFN-β through TLRs, particularly TLR4 (38). TLR4 is primarily expressed on cells of myeloid origin in the mouse brain (i.e., microglia), is upregulated in response to TLR4 ligands (12), and is a specific marker for macrophage activation in the brain in response to infection (39, 40). Therefore, we examined TLR4 transcript expression as evidence of hsp70-mediated TLR signaling and macrophage activation. Expression of TLR4 transcripts were measured by real-time RT-qPCR. TLR4 transcripts were significantly elevated in brains of infected TG mice relative to uninfected TG or infected NT mice at 1 day p.i. (P < 0.05 by ANOVA) (Fig. 5).
The model for hsp70-mediated protection against viral neurovirulence is centered on hsp70 that is expressed within (and released from) the virus-infected cell. Data from hsp70 TG mice do not, however, rule out the potential contribution of hsp70 that is expressed from uninfected neurons (e.g., the potential of hsp70 from uninfected neurons to contribute to immune modulation or to mediate protection from cytotoxic inflammatory responses). In order to restrict our analysis to hsp70 that is expressed within the virus-infected cell, an rVSV was designed to express hsp70 from the genome (rVSV-hsp70) (Fig. 6A). Protective effects of the virus-expressed hsp70 were measured in NT mice, using as a control an rVSV containing an unrelated gene in the same genome position (i.e., a human norovirus capsid gene, VP1) (30). The size of the hsp70 gene (1.9 kb) is comparable to that of the VP1 gene (1.7 kb).
First, effects of the gene insertions on viral replication in N2a-V cells and the extracellular release of hsp70 were characterized. N2a-V cells were infected with rVSV-hsp70 or rVSV-VP1 at an MOI of 0.01 and monitored for cell-free infectious viral progeny release and for G protein expression at 8, 16, and 24 h p.i. hsp70 was readily expressed from the genome of rVSV-hsp70, being detected at 16 and 24 h p.i. by Western blotting of total cell protein (Fig. 6C). Expression of G protein followed the same time course as hsp70. hsp70 expressed in this manner did not alter cell-free infectious viral progeny release. Significant differences in viral progeny release were not observed between rVSV-hsp70- and rVSV-VP1-infected N2a-V cells (Fig. 6B). This was correlated with the lack of any significant difference in the level of G protein expression between N2a-V cells infected with rVSV-hsp70 and rVSV-VP1 (Fig. 6C).
Extracellular release of hsp70 and virus-induced cytopathic effects were measured in rVSV-hsp70- and rVSV-VP1-infected N2a-V cells (MOI of 0.01). Increased levels of hsp70 in culture supernatants were detected at 8 h p.i. for both rVSV-hsp70- and rVSV-VP1-infected N2a-V cells relative to uninfected controls (Fig. 7A). These low levels may reflect virus-induced endogenous hsp70 that is not apparent on Western blot analysis of total cell protein, similar to the pattern described for rVSV infection of N2a-V cells (Fig. 1C). Levels of extracellular hsp70 showed progressive elevations at 16 and 24 h p.i. for rVSV-hsp70- but not rVSV-VP1-infected cells. The significant increase in extracellular hsp70 levels in rVSV-hsp70- relative to rVSV-VP1-infected cells was correlated with the ability to detect hsp70 in the cell by Western blotting of total protein. Release of hsp70 was not dependent upon cell lysis. Significant increases in LDH release were detected at 24 h p.i. for both rVSV-hsp70- and rVSV-VP1-infected cells (Fig. 7B). Release of hsp70 occurred during a period of virus-induced apoptosis, which was first evident at 16 h p.i. based upon evidence of caspase 3 activation by Western blotting (Fig. 7C). There was no difference in the levels of caspase 3 activation in rVSV-hsp70- and rVSV-VP1-infected cells.
NT mice (n = 10 per group) were inoculated intranasally with 1 × 106 PFU of rVSV-hsp70 or rVSV-VP1. Mice were monitored daily over a 15-day time course. NT mice infected with rVSV-hsp70 showed no mortality (100% survival), whereas the first death in NT mice infected with rVSV-VP1 occurred at 7 days p.i., with an overall mortality rate of 40% (P < 0.05) (Fig. 8A). The increased survival in rVSV-VP1 relative to parent rVSV-infected NT mice (Fig. 4) and the increased survival in rVSV-hsp70- relative to rVSV-infected TG mice reflects attenuation associated with the genome insertions. The attenuation is illustrated by comparing in vitro infection parameters for rVSV-VP1 and rVSV-hsp70 in N2a-V cells (Fig. 6 and and7)7) to those of rVSV in N2a-V and N2a-HSP cells (Fig. 1 and and3).3). The challenge dose of rVSV-hsp70 was subsequently increased to 1 × 107 PFU, and again there was no mortality induced in NT mice (Fig. 8B). Increasing the challenge dose to 1 × 108 PFU resulted in 33% mortality (Fig. 8B), such that a 100-fold increase in the rVSV-hsp70 challenge dose was required to achieve a mortality rate comparable to that achieved with 1 × 106 PFU rVSV-VP1.
Increased survival for rVSV-hsp70-infected mice (1 × 106 PFU challenge dose) was correlated with enhanced IFN-β mRNA expression in the brain. Total RNA was isolated from rostral brain sections harvested at 1 day p.i. to measure IFN-β transcripts using real-time RT-qPCR. The transcript levels were significantly increased in rVSV-hsp70-infected mice relative to rVSV-VP1-infected and uninfected controls and were also increased relative to rVSV-infected NT mice (P < 0.05 by ANOVA) (Fig. 9).
Results show that hsp70 protects against VSV neurovirulence, and that protection is correlated with enhanced innate immune responses, as measured by levels of IFN-β and TLR4 transcripts in brain. For virus entering a cell that constitutively expresses hsp70, any stimulatory effects of hsp70 on virus replication are offset by the ability of hsp70 to mediate a host protective response. Findings with rVSV-hsp70 suggest that enhanced innate immune induction does not rely on hsp70-mediated increases in virus gene expression, and that it is hsp70 within the virus-infected cell that mediates the IFN response and protection against neurovirulence. The probable link between hsp70 in the virus-infected cell and IFN induction is the ability of VSV to induce extracellular release of hsp70. hsp70 is a ligand for TLR2 and TLR4, which are primarily expressed on macrophages in the brain (12, 13). Macrophage responses to TLR2 and TLR4 ligands include production of IFN-β (41, 42), and the ability of hsp70 to directly induce IFN-β in a TLR2/4-dependent manner has recently been shown in both mouse microglial cell lines and in primary cultures of mouse bone marrow-derived macrophages (11). Current results show a progressive increase in the release of hsp70 during the course of VSV-induced apoptosis, and while cell lysis might contribute to release, lysis is not required. Cell lysis has been shown to be a mechanism for hsp70 release in parvovirus-infected tumor cells in vitro (25). hsp70 release from viable cells can occur via exosomal secretion or microvesicular shedding (43), whereas the mechanism of hsp70 release from cells undergoing apoptosis is unknown.
The level of VSV-induced extracellular release of hsp70 was proportionate to intracellular levels of hsp70. This was best illustrated by in vitro infections of N2a cells with rVSV-hsp70 (Fig. 7). hsp70 levels exhibited by hsp70 TG mice and those encountered in rVSV-hsp70-infected cells more closely approximate the levels observed in other species, including humans and other primates. Quantification of hsp70 by ELISA in brain total protein extracts reveals no detectable (basal) hsp70 expression in NT mice, in contrast to 1 ng/mg in hsp70 TG mice and 130 ng/mg in humans (44). Differences in the level of hsp70 between hsp70 TG mice and humans reflects in part the fact that hsp70 is constitutively expressed only in neurons of the TG mice compared to all cells of the human brain. When hsp70 is viewed as an IFN-β induction stimulus, the lack of constitutive expression of hsp70 in mice may help to explain the unique susceptibility of mice to VSV neurovirulence that is linked to deficient IFN-β production (45). VSV is capable of inducing endogenous hsp70 (34), and mouse and human hsp70 are highly homologous proteins (46, 47). However, Western blot analysis of VSV-infected N2a cell total protein indicates that such virus-induced levels are very low relative to that achieved with stable transfection or expression from the VSV genome. Detection of significant extracellular hsp70 release was always correlated with our ability to detect hsp70 within cells by Western blotting, this being observed only in the N2a-HSP cell line or N2a-V cells infected with rVSV-hsp70. As such, failure of VSV to induce significant levels of hsp70 is consistent with failure to induce a host protective IFN-β response.
The causal relationship between enhanced type I IFN production and hsp70-mediated protection against VSV neurovirulence remains to be shown. For the MeV model of brain infection, this relationship was established using mice that lack a functional type I IFN receptor gene (IFNAR−/−). In this model, virus infection is restricted to the brain as a result of both the intracranial route of inoculation and the poor degree of permissiveness of other tissues to virus replication. Since IFNAR−/− mice cannot be protected by hsp70-induced IFN-β, the level of mortality is comparable between TG IFNAR−/− and NT IFNAR−/− mice (11). Similar results have been obtained with VSV (M. Oglesbee, unpublished observations). Disruption of the type I IFN receptor resulted in 100% mortality within 3 days p.i., with no statistically significant difference in survival between the TG IFNAR−/− and NT IFNAR−/− groups. In this approach, however, interpretation of the result is complicated by the fact that loss of type I IFN signaling enhances infection of multiple organ systems following intranasal VSV challenge, making it unclear if the cause of mortality is virus brain infection (36). Although we know that type I IFN induction in the brain is required to suppress VSV neurovirulence and that hsp70 enhances induction of IFN-β transcripts following infection, we lack proof that hsp70-mediated protection is entirely dependent upon type I IFN. Therefore, we must also consider additional contributions of extracellular hsp70 to host defense, including the ability to directly mediate microglial activation (48) and cross-presentation of viral antigen (49).
Kinetic analysis of brain titers of VSV suggests that the predominant effect of hsp70 is to reduce total viral titers and enhance clearance during the late stages of virus infection. Significant differences in brain viral burden between infected hsp70 TG and NT mice is not observed until 7 days p.i., at which point viral clearance is accelerated in the hsp70 TG mice, and this is associated with reduced mortality that occurs in the 7- to 9-day p.i. interval. Similar findings were made in the mouse model of MeV brain infection, where hyperthermic induction of hsp70 resulted in enhanced viral clearance during late stages of infection that was correlated with enhanced virus-specific T cell responses (14). That type I IFN might contribute to viral clearance at these late time points is consistent with the role of IFN-β in stimulating macrophage activation in support of antigen presentation and cross-presentation (50) and the virus-specific T cell activation that is responsible for viral clearance (2, 3). The potential of extracellular hsp70 to enhance cross-presentation of viral antigen would also be relevant to clearance at these late stages of infection.
Collectively, the relationship between elevated hsp70 expression and VSV and MeV neurovirulence in the mouse supports a model for a basic mechanism of host defense with potential broad applicability. In both the cytopathic and productive VSV infection and the noncytopathic and nonproductive infection for MeV, virus-infected neurons release hsp70 into the extracellular environment, where it functions as a damage-associated molecular pattern (DAMP), stimulating innate immune responses that ultimately support viral clearance.
This work was funded through a Pilot Research Grant Program, Public Health Preparedness in Infectious Disease, The Ohio State University (J.L. and M.O.).
Published ahead of print 24 July 2013