Immune-Mediated Protection from Measles Virus-Induced Central Nervous System Disease Is Noncytolytic and Gamma Interferon Dependent

Neurons of the mammalian central nervous system (CNS) are anessential and largely nonrenewable cell population. Thus, virusinfections that result in neuronal depletion, either by virus-mediatedcell death or by induction of the cytolytic immune response,could cause permanent neurological impairment of the host. Ina transgenic mouse model of measles virus (MV) infection ofneurons, we have previously shown that the host T-cell responsewas required for resolution of infection in susceptible adultmice. In this report, we show that this protective responsedid not result in neuronal death, even during the peak of T-cellinfiltration into the brain parenchyma. When susceptible micewere intercrossed with specific immune knockout mice, a criticalrole for gamma interferon (IFN- ${gamma}$ ) was identified in protectionagainst MV infection and CNS disease. Moreover, the additionof previously activated splenocytes or recombinant murine IFN- ${gamma}$ to MV-infected primary neurons resulted in the inhibition ofviral replication in the absence of neuronal death. Together,these data support the hypothesis that the host immune responsecan promote viral clearance without concomitant neuronal loss,a process that appears to be mediated by cytokines.

INTRODUCTION

Top

Introduction

CD8⁺ cytotoxic T lymphocytes (CTL) mediate their antiviral effectsby recognizing viral peptides presented by class I major histocompatibilitycomplex (MHC) molecules on infected target cells. Engagementof the T-cell receptor with the peptide-loaded MHC leads toperforation of the target cell membrane, delivery of cytolyticgranules, and induction of apoptotic death of the infected targetcell (37, 56). Thus, CTL serve a crucial role in restrictingviral spread by selectively eliminating infected cells. In mosttissues, the loss of infected cells is accompanied by enhanceduninfected cell division to replenish the cell population. However,immune-mediated lysis of infected cells may not be an optimalstrategy for clearance of all virus infections, especially thosethat involve tissues with little capacity for renewal, suchas the central nervous system (CNS) (38, 42).

To limit immune cell entry into the CNS and to restrict T-cell-neuroninteractions, multiple anatomical and biochemical barriers existwithin the brain, including the presence of the blood-brainbarrier, limited lymphatic drainage from the CNS, and the paucityof class I MHC molecules on resident brain cells (38, 49). Theseproperties collectively contribute to an "immune-privileged"environment under normal conditions. Nevertheless, activatedT lymphocytes patrol the CNS parenchyma (17, 53) and are recruitedinto the brain after infection by many neurotropic viruses (2,19, 30, 41). In some circumstances, including herpes simplexvirus encephalitis, such inflammatory responses are associatedwith marked CNS damage (19). In others, however, the immuneresponse appears able to resolve an infection without apparentdamage to the host, although the basis of such virus clearanceremains poorly understood.

Previous work from our laboratory and others suggested thatthe host immune response may mediate viral clearance via noncytolyticstrategies. For example, immune-mediated clearance of a persistentlymphocytic choriomeningitis virus (LCMV) infection within theCNS is noncytopathic (33, 42, 50) and requires the presenceof T-cell-elaborated cytokines such as gamma interferon (IFN- ${gamma}$ )(51). A similar dependence on noncytolytic immune clearancehas also been shown for Sindbis virus (2), mouse hepatitis virus(26, 34, 48), and vesicular stomatitis virus (22-24). Noncytolyticviral clearance is not restricted to the CNS: in a transgenicmouse model of hepatitis B virus infection of the liver, cytokineswere also shown to be crucial for reduction of viral load (14,15). However, in each of these model systems, either the immuneresponse must be adoptively transferred into persistently infectedor otherwise tolerant mice, or the infection occurs in multiplecell types in which different clearance mechanisms may be operative(5, 12, 27).

To define the role of the host antiviral immune response inthe resolution of a neuron-restricted virus infection, we employeda transgenic mouse model in which a human measles virus (MV)receptor, CD46, was expressed specifically in CNS neurons underthe transcriptional control of the rat neuron-specific enolase(NSE) promoter (39). We have previously shown that these micesupport infection by MV, Edmonston strain (MV-Ed), but thatthe pathogenic outcome of infection is age dependent (39). NeonatalNSE-CD46⁺ mice develop an unrestricted neuronal infection andfatal CNS disease, whereas adult mice mount a rapid and robustT-cell response which protects them from neurological impairmentand death (25). We show here, by using primary neuron culturesand knockout mice deficient in components of the host response,that the adult immune response noncytolytically protects againstMV infection and CNS disease. This process appears to be dependenton cytokines such as IFN- ${gamma}$ .

MATERIALS AND METHODS

Top

Materials and Methods

Cells and virus. Vero fibroblasts were maintained in Dulbecco modified Eaglemedium (Life Technologies, Grand Island, N.Y.), supplementedwith 5% fetal calf serum, 2 mM L-glutamine, 100 U of penicillinper ml, and 100 ng of streptomycin per ml. Primary hippocampalneurons were prepared from day 16 embryonic mice, as describedpreviously (1, 35, 40), except that the cells were maintainedin Neurobasal Medium (Life Technologies) containing 4 µgof glutamate/ml in the absence of an astrocyte feeder layer.These cultures are routinely >95% MAP-2⁺ (data not shown).All cells were maintained at 37°C in a humidified incubatorwith 5% CO₂.

MV-Ed was purchased from American Type Culture Collection (Manassas,Va.) and passaged and titered in Vero fibroblasts. LCMV Armstrong(LCMV-Arm; a gift from Michael Oldstone, The Scripps ResearchInstitute) was passaged on BHK-21 fibroblasts and plaque purified,and titers were determined on Vero fibroblasts.

Mice. Inbred C57BL/6 (H-2^b) and homozygous NSE-CD46 transgenic mice(line 18; H-2^b) (39) were maintained in the closed mouse breedingcolony of The Fox Chase Cancer Center. All experimental protocolswere reviewed and approved by the Institutional Animal Careand Use Committee. Homozygous NSE-CD46⁺ and haplotype-matchedhomozygous immune knockout (KO) mice with targeted gene deletionswere intercrossed for three generations to obtain CD46⁺ miceon the desired KO background and maintained by backcrossingto the appropriate KO mice. Genotypes of all mice used in theseexperiments were confirmed prior to infection. CD4 KO (21),ß-2-microglobulin KO (ß-2-mic KO [55]),and RAG-2 KO (44) progeny were identified by flow cytometryon peripheral blood lymphocytes immunostained with fluorophore-conjugatedantibodies to mouse CD4 and CD8 antigens. Genotypes of IFN- ${gamma}$ KO (6) and perforin KO (20) mice were determined by PCR analysisof DNA isolated from tail biopsy samples.

Mice were infected with 10⁴ PFU of MV-Ed. The inoculum was dilutedin phosphate-buffered saline (PBS) and delivered intracerebrallyto metofane-anesthetized mice along the midline, in a volumeof 20 µl, by using a sterile 27-gauge needle. Additionalmice were challenged with 2 x 10⁵ PFU of LCMV-Arm, deliveredintraperitoneally in a volume of 200 µl.

RNA hybridization. Brains were removed from infected mice at 1, 3, 5, 7, and 13days postinfection (three mice/group), quick-frozen in liquidnitrogen, and homogenized in Tri-Reagent (Sigma Chemical Co.,St. Louis, Mo.) by using a tissue homogenizer (Virtis, Gardiner,N.Y.). Total RNA was isolated according to the manufacturer'sinstructions and resuspended in diethyl pyrocarbonate (DEPC)-treatedwater. Then, 5-µg RNA samples were mixed to a final concentrationof 48% formamide and 6.6% formaldehyde and incubated at 68°Cfor 15 min. The samples were then chilled on ice and dilutedtwofold with 20x SSC (1x SSC is 0.15 M NaCl plus 0.015 M sodiumcitrate). The RNA samples were applied to a nylon membrane byusing a slot blot apparatus; the membrane was UV cross-linkedand hybridized with a radioactive probe at 68°C for 1 h(QuikHyb; Stratagene, La Jolla, Calif.). DNA probes were labeledby random priming (Prime-It II; Stratagene) by using [³²P]dCTP(Dupont-NEN, Boston, Mass.). A DNA fragment from peF1 (4) wasused as a template for hybridization to F as a representativeMV RNA. The signal intensity of radioactive slots was quantifiedby phosphorimager analysis (Fuji).

Immunohistochemical analysis of mouse tissues. Brains from 4 to 5 mice/time point postinfection were removed,immersed in tissue embedding compound (Fisher Scientific, Pittsburgh,Pa.), snap-frozen in a dry ice-isopentane bath, and stored at-70°C. Horizontal cryosections (10 µm) were air driedand stored at -70°C. On the day of staining, sections werefixed in ice-cold 95% ethanol, rehydrated in PBS, and blockedfor 20 min with 0.1% bovine serum albumin-PBS (Sigma), followedby an avidin and biotin block (Vector Laboratories, Burlingame,Calif.). To detect MV-infected cells, a human immune serum (agift from Michael Oldstone) was used at a dilution of 1:2,000,followed by a biotinylated anti-human secondary antibody (1:300;Vector). For LCMV immunostaining, a polyclonal mouse anti-LCMVantibody (American Type Culture Collection) was used at a dilutionof 1:750, followed by a biotinylated, anti-mouse immunoglobulinG secondary antibody. Rat anti-mouse CD4 (clone RM4-5; 1:100;Pharmingen, San Diego, Calif.) or rat anti-mouse CD8 ${alpha}$ /CD8ßantibodies (clones 53-6.7 and 53-5.8, respectively; 1:100 each;Pharmingen) were used to identify CD4⁺ and CD8⁺ T lymphocytes.To detect B cells, rat anti-mouse B220 (Pharmingen) was usedat a dilution of 1:1,000. These sections were then incubatedfor 1 h at room temperature with a biotinylated anti-rat immunoglobulinG secondary antibody at a 1:200 dilution (Vector). Sectionslabeled with biotinylated antibodies were treated for 30 minwith a streptavidin-peroxidase conjugate (ABC Elite; Vector),followed by visualization with diaminobenzidine (0.7 mg/ml in60 mM Tris) and urea-H₂O₂ (0.2 mg/ml), purchased as preweighedtablets (Sigma). All cells were counterstained with hematoxylin(Sigma) and mounted with an aqueous mounting medium (CrystalMount; Fisher). Uninfected tissues or omission of the primaryantibody served as negative controls. For all histological analyses,at least three sections per brain were examined from three differenthorizontal levels, and at least four mice per experimental groupwere assessed.

TUNEL (terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling) assays. Cryosections were fixed in 10% neutral buffered formalin, postfixedin ethanol-acetic acid (2:1), and quenched for endogenous peroxidaseactivity with 3% hydrogen peroxide in PBS. Sections were labeledwith a terminal deoxynucleotidyltransferase (TdT)-digoxigenin-deoxynucleosidetriphosphate mixture (Intergen, Purchase, N.Y.). In order toend label, slides were equilibrated, covered with working-strengthTdT in reaction buffer, and incubated in a humidified chamberat 37°C for 1 h. Reactions were washed in stop buffer. Ananti-digoxigenin peroxidase conjugated-antibody was used todetect end-labeled DNA fragments. This step was followed byvisualization with diaminobenzidine as previously described.All tissues were counterstained with methyl green (Sigma) andmounted in Crystal Mount (Fisher).

Sickness assays. Mice were weighed throughout infection by using a digital balance,and the weights were recorded in grams to the first decimalplace. In addition, mice were assessed by using two subjectivemeasures of health. Balance was assayed by placing the infectedmouse on a three-quarter-inch wooden dowel rod and scored asfollows: 0, mouse balanced on rotating rod for >20 s; 1,mouse balanced on rod for 5 to 20 s; 2, mouse balanced on rodfor <5 s; and 3, mouse demonstrated no ability to grip rod.Movement was also documented following placement of infectedmice into an 8-in. circle imprinted in the base of a 2-in. by2-in. opaque plastic container. Scores were assigned as follows:0, movement out of the circle occurred in <5 s; 1, movementout of the circle occurred between 5 to 20 s; 2, movement outof the circle took >20 s and/or evidence of seizures or tremorswere noted during a 1-min observation period; and 3, hindlimbparalysis.

Ex vivo viral clearance assays. Primary hippocampal neurons were infected with either MV-Edor LCMV-Arm at a multiplicity of infection (MOI) of 3. Eitherimmediately after infection or 24 h postinfection (hpi), lymphocytesor recombinant IFN- ${gamma}$ was added to the cultures. Red blood cell-depleted,bulk splenocytes were isolated from both perforin-deficientmice and C57BL/6 mice that had been challenged with LCMV 7 dayspreviously or that were uninfected. Splenocytes were added toMV- or LCMV-infected primary hippocampal neurons at effector/targetratios of 25:1 for cell death assays and 10:1 for viral RNAclearance assays. In separate studies, recombinant mouse IFN- ${gamma}$ ,heat-inactivated mouse IFN- ${gamma}$ (65°C, 30 min), or PBS-0.2%bovine serum albumin was also added to MV- or LCMV-infectedneurons. After an extensive washing, neurons were harvested24 h later by scraping them in Tri-Reagent (Sigma). Total RNAwas isolated according to the manufacturer's instructions andresuspended in DEPC-treated water. The RNA samples were thenanalyzed by slot blot as described above. LCMV RNA was hybridizedwith a radiolabeled probe derived from the cloned LCMV nucleoproteingene. To determine the proportion of infected cells in someexperiments, cells on coverslips included in the cultures werefixed at the time of harvesting and immunostained for the presenceof viral antigen, and the immunopositive cells were counted.

RESULTS

Top

Results

Kinetics of immune-mediated resolution of neuronal MV infection. We previously demonstrated (25) that the adaptive immune responseof adult NSE-CD46⁺ transgenic mice was essential for protectionagainst MV infection, since CD46⁺ mice backcrossed to T- andB-cell-deficient RAG-2 KO mice developed unrestricted neuronalinfection and CNS disease after MV inoculation. To explore thebasis of this protection, we performed an extensive time courseanalysis of brain tissues collected from immunocompetent, NSE-CD46⁺adults (6 to 8 weeks of age) after MV challenge.

Mice were inoculated intracerebrally with 10⁴ PFU of MV-Ed,and brains were collected at specific time points throughoutinfection for RNA and immunohistological analysis. Virus-infectedneurons were first detectable at 3 to 4 days postinoculation(dpi) by both methods (Fig. 1; Table 1), reaching maximal infectionat 7 to 10 dpi (Fig. 2A). The appearance of virus-infected neuronsin the CNS correlated with the entry of both CD4⁺ and CD8⁺ Tlymphocytes (Fig. 2B and C). As previously reported (25), B220⁺cells were rarely observed in the infected CNS, representingmaximally 3% of the immune cell infiltrate. After 13 dpi, viruswas no longer detectable, and this was accompanied by a decreasednumber of intraparenchymal T cells, although the number of Tcells present within the brains of recovering mice remainedsubstantially higher than that observed in uninfected controls(Table 1; Fig. 2F and G).

View larger version (15K):
[in this window]
[in a new window]
FIG. 1. MV RNA levels in brains of infected NSE-CD46⁺ adult mice. Immunocompetent NSE-CD46⁺ mice were inoculated intracerebrally with MV-Ed, and brains were harvested at 1, 3, 5, 7, and 13 dpi. Tissues were homogenized, and total RNA was hybridized with a radiolabeled probe specific for the MV fusion gene and quantified with a Fuji densitometer as described in Materials and Methods. The extent of MV-F RNA expression is shown as adjusted pixel values relative to a GAPDH (glyceraldehyde-3-phosphate dehydrogenase) internal control and compared to baseline (inoculation of mice with PBS). At least four mice/group were evaluated; the standard error is indicated.

View this table:
[in this window]
[in a new window]
TABLE 1. Profile of lymphocyte infiltration and apoptosis in brains of MV-infected, NSE-CD46⁺ immunocompetent adults^a

View larger version (82K):
[in this window]
[in a new window]
FIG.2. Lymphocytic infiltration and cell death in brains of MV-infected NSE-CD46⁺ mice. Immunocompetent NSE-CD46⁺ mice were inoculated intracerebrally with MV-Ed, and brains were harvested throughout infection. Frozen tissues were sectioned and immunostained for the presence of MV antigens (A and E), CD4⁺ T cells (B and F), CD8⁺ T cells (C and G), and DNA ends by TUNEL assay (D and H; arrows indicate stained nuclei). Magnification, x84. Higher-magnification images (x277) of T cells are shown in panels B2 and C2. (A to D) Serial sections taken from a representative mouse at 10 dpi; (E to H) serial sections taken from a mouse at 21 dpi. A dense region of infiltration was selected to emphasize TUNEL-positive cells at this time point.

While the majority (>95%) of infected mice remained healthythroughout the time course, a small number of immunocompetentadults consistently showed signs of sickness after viral challenge,including tremors, ataxia, and a hunched posture, as seen withone mouse in Table 1 at the 16-dpi time point and in two micein Table 2, experiment 2. When examined histologically, no MV-immunopositiveneurons were observed in brains of these mice, but all had highlevels of T-cell infiltration and cell death. For example, thesick mouse in Table 1 had three to four times more CD4⁺ andCD8⁺ T cells within the parenchyma than matched healthy mice(648 versus 196 CD4⁺ T cells and 766 versus 182 CD8⁺ T cells)and the highest levels of TUNEL-positive nuclei observed inthe time course. These data suggest that, in a small percentageof immunocompetent hosts, an overly aggressive immune responsemay be associated with neuropathology.

View this table:
[in this window]
[in a new window]
TABLE 2. Consequence of MV infection in CD46⁺ transgenic mice on various immunodeficient backgrounds^a

Throughout the time course, the extent of cell death withininfected brains was uniformly low, a finding consistent withthe lack of illness observed in most infected mice. However,some cell death above the baseline level was seen: first observedat the peak of viral infection and immune cell infiltration(day 10; Fig. 2D; Table 1) and remaining elevated at later timespostinfection (day 21; Fig. 2H) when no virus-infected cellswere observed (Fig. 2E). Given the persistence of a small numberof TUNEL-positive cells at late time points (when T cells wereabundant but no MV-positive neurons were detectable) and theabsence of cell death at early times postinfection (when MV-immunopositiveneurons and T cells were present in the infected parenchyma),this low level of cell death likely reflects activation-inducedcell death of infiltrating lymphocytes rather than immune-mediateddeath of MV-infected neurons.

IFN- ${gamma}$ is crucial for the resolution of neuronal infection. To more precisely define components of the host response necessaryfor clearance, homozygous CD46⁺ mice possessing genetic immunodeficiencieswere infected and monitored. To determine the contribution ofindividual T-cell populations, CD46⁺ mice were intercrossedwith both CD4 KO mice (21) and ß-2-mic KO mice thatlack CD8⁺ T cells and class I MHC expression (55). In addition,CD46⁺ mice were also intercrossed with mice lacking either thepore-forming protein, perforin (20), or the cytokine IFN- ${gamma}$ (6),a Th1 cytokine produced by CD4⁺ T cells, CD8⁺ T cells, and NKcells. In addition to its well-established role in immune cellrecruitment and activation, IFN- ${gamma}$ has also been implicated inthe noncytolytic clearance of neurotropic virus infections (22,34, 51).

As shown in Table 2, MV-infected adults lacking either CD4⁺T cells or CD8⁺ T cells were more susceptible to MV infectionand CNS disease than immunocompetent adults (59 and 41% morbidityversus 2% morbidity in immunocompetent CD46⁺ mice). However,compared to immunodeficient CD46⁺/RAG-2 KO mice, both CD4 KOand ß-2-mic KO mice showed reduced morbidity (Table2, experiment 1) and reduced extent of CNS infection (Fig. 3).This was also correlated with a more extended interval betweeninoculation and illness ( $~$ 19 days versus $~$ 14 days). These datasuggested that, while the CD4⁺ and CD8⁺ T-cell compartmentsare each involved in the anti-MV response within the brain,both are required for complete protection. In immunohistochemicalassays of MV-infected CD46⁺/CD4 KO or CD46⁺/ß-2-micKO mice, the absence of one T-cell subset did not appreciablyaffect the degree of infiltration of the other subset into thebrain parenchyma (data not shown).

View larger version (147K):
[in this window]
[in a new window]
FIG. 3. Both T-cell subsets partially protect against MV infection in CD46⁺ mice. CD46⁺ mice on the indicated KO background were inoculated intracerebrally with MV-Ed, and brains were harvested at 20 dpi. Brains were sectioned and stained for the presence of MV antigens, and representative images are shown. (A) CD46⁺; (B) CD46⁺/RAG-2 KO; (C) CD46⁺/CD4 KO; (D) CD46⁺/ß-2-mic KO. Magnification, x100.

Consistent with our previous observation that immune-mediatedresolution of MV from susceptible mice was associated with minimalcell lysis, the deletion of the perforin gene had a modest impacton the ability of mice to resolve an MV challenge (2% in CD46⁺mice versus 18% in CD46⁺/perforin KO; Table 2, experiment 1).In sharp contrast, deletion of IFN- ${gamma}$ resulted in morbidity in $~$ 70% of infected mice compared to 94 to 100% in CD46⁺/RAG-2 KOmice. When immunocompetent CD46⁺ adults, CD46⁺/RAG-2 KO adults,and CD46⁺/IFN- ${gamma}$ KO adults were compared in sickness assays thatassessed movement, coordination, and weight loss, MV-infectedCD46⁺/IFN- ${gamma}$ KO mice showed marked motor problems that paralleledthose seen in infected CD46⁺/RAG-2 KO with respect to kinetics,although the severity of CNS disease was less in infected CD46⁺/IFN- ${gamma}$ KO mice (Fig. 4). Moreover, the average interval to CNS diseasewas more similar to infected RAG-2 KO mice than previously observedwith CD4 and ß-2-mic KO mice.

View larger version (14K):
[in this window]
[in a new window]
FIG. 4. CD46⁺/IFN- ${gamma}$ KO mice show progressive illness and weight loss after MV inoculation. Mice were subjected to three clinical analyses as described in Materials and Methods. (A) Balance and movement scores were combined and averaged from results obtained at 1 to 3 weeks postinoculation. wt, CD46⁺ immunocompetent mice; IFN KO, CD46⁺/IFN- ${gamma}$ KO; RAG KO, CD46⁺/RAG-2 KO. (B) Percent body weight loss plotted for CD46⁺ immunocompetent mice, CD46⁺/IFN- ${gamma}$ KO mice, and CD46⁺/RAG-2 KO mice. The maximum weight loss during the time course of infection is shown. For both panels, the mean values are represented by horizontal bars, and individual scores are represented as black diamonds. All of the infected mice from two experiments are included in this analysis.

One primary role of IFN- ${gamma}$ is the recruitment and activation ofthe adaptive immune response. It was therefore possible thatthe deletion of IFN- ${gamma}$ from these mice resulted in reduced T-cellinfiltration, leading to infection and disease. To test thishypothesis, time-matched samples of immunocompetent CD46⁺ miceand CD46⁺/IFN- ${gamma}$ KO mice were collected throughout infection andstained for the presence of T cells within the brain. At notime point postinfection were substantial differences in intraparenchymalT-cell values noted between the groups (data not shown), indicatingthat the morbidity observed in infected CD46⁺/IFN- ${gamma}$ KO mice wasnot due to reduced T-cell infiltration.

IFN- ${gamma}$ exhibits a direct, antiviral role in MV-infected primary neurons. The requirement for IFN- ${gamma}$ but not perforin in MV clearance fromthe infected CNS suggested that T-cell-elaborated cytokinesmay play a direct role in viral clearance, without necessitatingthe lysis of infected neurons. If true, the addition of eitherperforin-deficient, activated T cells or recombinant cytokinesto infected neurons in culture should result in inhibition ofviral replication.

To test these hypotheses, virus-infected primary hippocampalneurons cultured from CD46⁺ embryos were incubated with splenocytesobtained from C57BL/6 mice or perforin KO mice. The donor micewere either naive or infected with LCMV 7 days previously. LCMVwas chosen as an immunogen because of the robust CD8⁺ T-cellresponse generated in vivo after intraperitoneal inoculation.Total red blood cell-depleted splenocytes were isolated andimmediately cultured with LCMV- or MV-infected primary neurons.In trypan blue cell viability studies, no neuronal loss wasobserved after a 6-h incubation with any of the effectors (datanot shown), a finding consistent with the reported lack of classI MHC antigens on the neuronal cell surface (38). Interestingly,however, when RNA was isolated from infected neurons after a16-h incubation with the effector splenocytes, loss of viralRNA was observed from both LCMV- and MV-infected neurons (Fig.5). In the case of LCMV-infected neurons, addition of unprimedC57BL/6 or perforin KO splenocytes had no effect on viral RNAlevels, whereas LCMV-primed lymphocytes resulted in a 30 to50% loss of viral RNA. This effect was perforin independent.Interestingly, a similar loss of viral RNA was observed fromMV-infected neurons, although in these cultures, the incubationof infected neurons with either primed or unprimed lymphocytesresulted in reduced viral RNA levels. Together, these data indicatethat activated splenocytes can result in noncytolytic inhibitionof neuronal viral replication; from these observations we hypothesizedthat cytokines elaborated by LCMV-specific splenocytes may suppressMV replication in infected primary neurons.

View larger version (23K):
[in this window]
[in a new window]
FIG. 5. Splenocytes can inhibit viral replication ex vivo in the absence of neuronal loss. Primary hippocampal neurons obtained from embryonic day 16 NSE-CD46⁺ mice were infected with either LCMV-Arm or MV-Ed (MOI = 3) for 2 days prior to the assay. Wild-type, haplotype-matched (H-2^b) splenocytes were obtained either from C57BL/6 mice or from perforin KO mice that were uninfected or that had been challenged with LCMV-Arm 7 days previously. Splenocytes were incubated with infected neurons at a ratio of 10:1 overnight. After incubation, total viral RNA loads were quantified as described in Materials and Methods.

Finally, to establish if purified cytokines could inhibit viralreplication in infected neurons, cultured CD46⁺ hippocampalneurons were infected with either MV-Ed or LCMV-Arm (MOI = 3),and recombinant murine IFN- ${gamma}$ (100 U/ml) was added either immediatelyafter infection or 24 hpi. At 2 dpi, infected cells were harvestedfor immunochemistry and RNA analysis. As controls, PBS or heat-inactivatedIFN- ${gamma}$ (65°C; 30 min) was also added to the infected cultures.As shown in Fig. 6, the addition of IFN- ${gamma}$ immediately after infectionresulted in $~$ 50% inhibition in MV RNA load, compared to no effectseen with either PBS-treated or heat-inactivated, IFN- ${gamma}$ -treatedcultures. Importantly, the addition of IFN- ${gamma}$ at 24 hpi (thatis, after the establishment of an infection) also resulted inreproducibly reduced MV RNA levels. The addition of IFN- ${gamma}$ appearedto block subsequent viral spread: as shown in Fig. 7, the extentof MV infection in primary cultures treated with IFN- ${gamma}$ for 24h and collected 48 hpi (Fig. 7B) paralleled that seen in culturescollected at 24 hpi (Fig. 7A). In contrast, untreated neuronscollected at 48 hpi (Fig. 7C) showed extensive viral spread.Together, these data suggest that T-cell-elaborated cytokines,such as IFN- ${gamma}$ , play a direct role in the resolution of certainviral infections.

View larger version (21K):
[in this window]
[in a new window]
FIG. 6. The effect of IFN- ${gamma}$ on MV infection in susceptible primary hippocampal neuron cultures. Primary hippocampal neurons were infected with MV-Ed at an MOI of 3. IFN- ${gamma}$ was added immediately, at 24 hpi, or as a heat-inactivated negative control (100 U/ml). Neurons were also infected with LCMV at an MOI of 1 and treated with IFN- ${gamma}$ . The percent RNA remaining at 48 hpi is shown compared to untreated samples, and the standard error is indicated.

View larger version (59K):
[in this window]
[in a new window]
FIG. 7. Inhibition of MV spread in primary neurons after IFN- ${gamma}$ addition. MV-infected, primary hippocampal neurons were treated with IFN- ${gamma}$ or PBS and were subsequently fixed and immunostained as described in Materials and Methods. (A) Collected at 24 hpi. (B) IFN- ${gamma}$ added at 24 hpi; collected at 48 hpi. (C) PBS added at 24 hpi; collected at 48 hpi. Magnification, x87.

DISCUSSION

Top

Discussion

In this study, a genetic approach was taken to determine thebasis of immune-mediated clearance of MV from the CNS of infected,CD46⁺ transgenic mice. Our data indicate that clearance occurredwithout neuronal loss and that both CD4⁺ and CD8⁺ T lymphocytescontributed to this noncytolytic protection. While deletionof the perforin gene did not appreciably impact the abilityof the immune response to resolve the infection, deletion ofthe IFN- ${gamma}$ gene had a significant effect which could not be attributedto reduced infiltration of T cells into the infected CNS parenchyma.In retrospect, if cytokines such as IFN- ${gamma}$ play a crucial rolein viral clearance, the partial protective effect observed inCD46⁺ CD4 KO and ß-2-mic KO mice is reasonable, sinceIFN- ${gamma}$ is produced by CD4⁺ T cells, CD8⁺ T cells, and NK cells.Thus, even on a RAG-2 KO background, a small number of micecan survive a persisting viral challenge, which may be due tothe intact innate response in these animals. While the innateresponse per se may not be crucial for protection against MV-mediatedneuropathology in this model (since CD46+/RAG-2 KO mice havea high morbidity rate after MV inoculation), the potential roleof the innate response in the recruitment and/or education ofthe subsequent adaptive response is being considered in ongoingstudies.

The ability of activated T cells to restrict viral replicationin the absence of neuronal loss--and the importance of IFN- ${gamma}$ in this process--was further confirmed in ex vivo experimentswith primary neuron cultures. Collectively, these data supportand extend work from our laboratory and others that describesa novel, noncytolytic role of the antiviral T-cell responsein the brain. The merits of such a strategy are especially clearin tissues in which cell renewal does not occur, as in the CNS.

It is likely that our ex vivo studies underestimate the impactof activated T cells and IFN- ${gamma}$ on MV replication, since the primaryneuron cultures were obtained from a neuron-rich CNS structure(hippocampus at embryonic day 16) and were cultured in the absenceof supporting glia (these cultures are routinely >95% MAP-2⁺).Thus, if MV infection of neurons induced elevated cytokine synthesisby adjacent glia or if glial contact is required for recognitionof infected cells in vivo, highly purified neuronal culturesmay not contain all of the appropriate cellular participantsneeded for maximal clearance. Nevertheless, the relatively modesteffect of both activated T cells and recombinant cytokines onneuronal viral loads ex vivo and the fact that not all CD46⁺/IFN- ${gamma}$ KO mice succumb to CNS disease support the conclusion that,while IFN- ${gamma}$ plays a central role in protection, complete protectionis multifactorial. Cytokines are known to act in concert (18);it is therefore likely that other Th1 cytokines, such as interleukin-2and tumor necrosis factor alpha, also participate in the resolutionof a neuron-restricted MV infection in vivo.

How cytokines such as IFN- ${gamma}$ mediate their protective effectsin the resolution of established virus infections is not clear.At least two hypotheses exist. First, cytokines elaborated byinfiltrating immune cells may exert their well-established effectsby preventing viral spread, restricting infection to a smallnumber of cells. If "classic" CTL-mediated cytolysis then occurs,it is unlikely that much damage would ensue, given the relativelysmall number of cells that would be targets for lysis. Our datado not contradict this hypothesis, since we did observe a modestdegree of apoptosis in infected brains in vivo, and treatmentof infected primary neurons with IFN- ${gamma}$ appeared to restrict viralspread (Fig. 7). However, (i) TUNEL-positive cells were detectablein the absence of virus-immunopositive neurons and (ii) at earlystages of infection in vivo (when T cells were abundant in MVantigen-positive brains) no cell death was detected. We thereforehypothesize that the observed cell death was due to the lossof activated lymphocytes by activation-induced cell death atthe later stages of the immune response (18) and not due todirect neuronal lysis. The absence of a correlation betweenneuronal infection and death therefore suggests that cytokinesmay play a more direct role in viral clearance than solely byinhibiting viral spread.

A second hypothesis is that T-cell-elaborated cytokines, suchas IFN- ${gamma}$ and tumor necrosis factor alpha, participate directlyin viral clearance. IFN- ${gamma}$ is crucial for the noncytopathic suppressionof herpes simplex virus type 1 infections in trigeminal ganglia(3) and photoreceptor cells in the eye (10) and in the neuroninfections caused by LCMV (51), Borna virus (16), mouse hepatitisvirus (36), and Theiler's virus (7). Indeed, neurons expressIFN- ${gamma}$ receptors (43, 52), providing support for the idea thatneurons respond to this cytokine. For MV, the importance ofIFN- ${gamma}$ has been noted in both humans (9) and mice (using a neuroadaptedstrain of MV) (8), although whether IFN- ${gamma}$ acts directly or indirectly(e.g., by activating T cells) to suppress viral replicationcould not be concluded from these studies. Both IFN- ${alpha}$ /ßand IFN- ${gamma}$ act by binding receptors on the cell surface, triggeringrecruitment of the JAK kinases and activation of STAT proteins(47). This results in a cascade of intracellular events thatleads to the transcription of IFN response genes, includingthe 2'-5' OAS system, the Mx proteins, inducible nitric oxidesynthase, and the double-stranded RNA-PKR pathway. Collectively,activation of these genes contributes to the antiviral state(47). Whether such proteins are associated with viral clearancein established infections remains to be determined.

One surprising aspect of our work was the persistence of T cellsin the brains of infected mice long after virus was no longerdetectable by immunochemical and RNA hybridization methods (Table1; Fig. 1). While we do not yet fully understand the basis forthis T-cell retention, we speculate that the absence of CTL-mediatedlysis of infected neurons may result in a "smoldering" infection;cytokine-dependent, noncytolytic clearance may take longer toresult in complete virus loss. Our present efforts to test thishypothesis include use of more sensitive detection strategiesto determine whether viral RNA persists in brain tissue afterviral antigens are no longer detectable by immunohistochemistry.

The ability of T lymphocytes to eliminate virus that is restrictedto CNS neurons is pertinent to an ongoing debate in neuroimmunology.Neurons are generally thought not to express either class Ior class II MHC molecules (38); thus, how either CD4⁺ or CD8⁺T cells mediate their protective effects is obscure. There area number of possible explanations for this. (i) CD4⁺ and CD8⁺T cells, educated in the periphery, may constitutively secretecytokines; recruitment of these cells into the MV-infected parenchymamay, therefore, result in elevated localized cytokine concentrations.However, Whitton and coworkers (46) have shown that cytokinesynthesis occurs only when T cells are engaged with their target;upon disengagement, cytokine secretion is rapidly turned off.These data suggest that infiltrating T cells in the NSE-CD46model are not constitutively producing antiviral cytokines withinthe CNS and that this scenario is unlikely. (ii) Previous reportshave shown that class I MHC can be induced on CNS neurons inresponse to injury (28, 29, 31, 32, 54) or virus infection,including MV (11, 13). Thus, while class I and II MHC expressionmay be below detection limits in the uninfected brain, MV infectionmay induce expression, allowing for MHC-T-cell receptor contactto occur. However, our inability to detect upregulated MHC onMV-infected neurons or in infected brains (unpublished observations),the ability of LCMV-specific splenocytes to restrict MV replicationex vivo (Fig. 5), and the survival of a substantial proportionof class I MHC-deficient ß-2-mic KO mice after MVinfection (Table 2) collectively indicate that T cells do notrecognize virus-specific peptides in the context of class IMHC on neurons. (iii) A hypothesis that we favor is that, invivo, infiltrating T cells interact with adjacent microgliathat, though uninfected, may still present viral epitopes viaclass II MHC presentation to CD4⁺ T cells or via cross-presentationto CD8⁺ T cells (45). Elaboration of cytokines as a result ofthis interaction may then indirectly allow for cytokine-mediated,noncytolytic viral clearance from adjacent neurons to occur.A major focus of our present work is to determine whether suchinteractions occur, to establish the mechanism of viral clearancewithout concomitant neuronal loss, and to test whether thismode of viral clearance is applicable to other neurotropic pathogens.

ACKNOWLEDGMENTS

We thank Melinda Vaughn, Donna Mscisz, Joan Cole, and Alec Belmanfor assistance with the experiments. We are also appreciativeof the gifts of LCMV-Arm and human MV-reactive serum from MichaelB. A. Oldstone of The Scripps Research Institute in La Jolla,Calif. Finally, we are grateful to Bill Mason and Luis Sigalfor their invaluable comments regarding the manuscript.

G.F.R. was supported by a grant from the National Instituteof Mental Health (MH-56951) and a grant from the F. M. KirbyFoundation. C.E.P. was supported by an individual NRSA award(NS-11100).

FOOTNOTES

* Corresponding author. Mailing address: The Fox Chase Cancer Center, 7701 Burholme Ave., Philadelphia, PA 19111. Phone: (215) 728-3617. Fax: (215) 728-2412. E-mail: gf_rall{at}fccc.edu.

Present address: Laboratory of Molecular Medicine and Neuroscience,National Institute of Neurological Disorders and Stroke, NationalInstitutes of Health, Bethesda, Md.

REFERENCES

Top