This report evaluates how HSV enters the brain to cause herpes simplex encephalitis following infection at a peripheral site. We demonstrate that encephalitis regularly occurred when BALB/c mice were infected with HSV and treated daily with 2-deoxy-d-glucose (2DG), which inhibits glucose use via the glycolysis pathway. The outcome of infection in the trigeminal ganglion (TG), the site to which the virus spreads, replicates, and establishes latency, showed marked differences in viral and cellular events between treated and untreated animals. In control-untreated mice, the replicating virus was present only during early time points, whereas in 2DG recipients, replicating virus remained for the 9-d observation period. This outcome correlated with significantly reduced numbers of innate inflammatory cells as well as T cells in 2DG-treated animals. Moreover, T cells in the TG of treated animals were less activated and contained a smaller fraction of expressed IFN-γ production compared with untreated controls. The breakdown of latency was accelerated when cultures of TG cells taken from mice with established HSV latency were cultured in the presence of 2DG. Taken together, the results of both in vivo and in vitro investigations demonstrate that the overall effects of 2DG therapy impaired the protective effects of one or more inflammatory cell types in the TG that normally function to control productive infection and prevent spread of virus to the brain.
Some of the most devastating consequences of a virus infection occur when they enter and infect the brain and cause encephalitis (1). The brain is normally well protected from infection, and few agents have the capacity to gain entrance. Access to the brain can occur in different ways, but the most common entry route is by crossing the blood–brain barrier (BBB) (2). The BBB is a multicellular organization that is composed of vascular endothelial cells tightly connected by specialized proteins, such as claudins and occludins, surrounded by pericytes, and supported by astrocyte cells on the CNS side (3). The normally functioning BBB allows the exchanges of nutrients and metabolites between the blood circulation and the brain but limits the passage of many soluble compounds as well as most pathogens. However, when the BBB is damaged physically or physiologically, its integrity is impaired, and the brain becomes more accessible (4). Although the BBB and its function have been the topic of many studies [reviewed in (5)], it is still not clear whether the host metabolic state could impact on the integrity of the BBB. Thus, changes in host metabolism could impair the barrier function of the BBB and permit viruses to cross and cause damage to CNS tissues. This might be the mechanism by which HSV infection causes encephalitis, an issue that is addressed in the present report.
Infection of the brain by HSV can result in notable damage to the brain that, if uncontrolled by antiviral drugs, may have major consequences on neuronal tissues (6). This herpes simplex encephalitis (HSE) syndrome can occur in a number of different circumstances. First, newborns born to HSV-seronegative mothers, if infected with HSV-2 around the time of birth, may develop encephalitis, which can have very destructive consequences, unless rapidly treated with effective antiviral drugs (7). Children born with one of several genetic defects that affect their ability to produce or respond to IFNs are prone to develop HSE and without modern therapies would rarely reach adulthood (6, 8). Healthy adults occasionally develop HSE, mostly in those who carry the virus in a latent from, perhaps for years, after primary infection (9, 10). Such persons, upon reactivation from latency that normally causes only local lesions, can result in virus spreading to the brain, resulting in HSE (11). Why this unusual event occurs and how virus passes to the brain remains unresolved, although a number of possible mechanisms have been suggested (12). These include direct passage to the brain via cranial nerves, such as the olfactory or trigeminal nerves, and crossing directly through the BBB. In addition, some agents can cross the BBB into the brain within leukocytes, a circumstance resembling the Trojan horse phenomenon (12, 13).
In past studies on experimental ocular infection of adult C57BL/6 mice with HSV, we had observed that if mice were given the drug 2-deoxy-d-glucose (2DG) soon after infection, which interferes with glucose metabolism, they regularly succumbed to HSE, whereas untreated animals did not (14). However, when therapy with 2DG was begun at a later time, when ocular inflammatory lesions were developing, the mice did not develop HSE. Furthermore, 2DG treatment decreased the severity of their ocular inflammatory lesions. We advocated that the latter effect was the consequence of suppressed inflammatory T cell responses that are mainly responsible for ocular lesions (14). The cause of encephalitis in animals treated from the onset of infection with 2DG was not defined, but one possibility could be damage to the metabolism of BBB cells, making the barrier more permeable to crossing by the virus. In this report, this issue is further evaluated, as are other ways the virus might access the CNS. We interpret our findings to indicate that, in the experimental mouse model we used, the most likely route of access to the brain was not via the BBB, made more permeable by 2DG therapy, but via the trigeminal nerve following continued and elevated viral replication in the trigeminal ganglion (TG) that occurred because 2DG treatment impaired the protective function of T cells in the TG.
Materials and Methods
Five to six-wk-old female BALB/C mice were purchased from Envigo (Indianapolis, IN) and were used for the ocular HSV-1 infection model. Animals were kept in an Animal Biosafety Level-2 facility, where water, bedding and instruments were regularly autoclaved. All mice were housed in the American Association of Laboratory Animal Science–approved facilities at the University of Tennessee (Knoxville, TN). All investigations were accomplished and followed the Institutional Animal Care and Use Committee guidelines and adhered to the Association for Research in Vision and Ophthalmology. Animals were visited by facility staff members and Office of Laboratory Animal Care (OLAC) veterinarians during the HSV-1 on infection by daily basis. Animals were supported with diet gel addition to their cages when they showed severe symptoms. Mice received 24-h watch if they presented a lethargic and hunched posture and were humanely euthanized by CO2 at the end of the 24-h watch.
HSV-1 RE strain was obtained from Robert Hendricks, University of Pittsburgh (Pittsburgh, PA). The virus was propagated in Vero cell monolayers (American Type Culture Collection, Manassas, VA) to produce viral seed stocks. Virus propagations and quantification of viral titer were described previously (15, 16). A less virulent HSV-1 strain than RE was employed for latency experiments. The virus was constructed recombinantly to express the enhanced green fluorescein protein (EGFP) under the control of early viral promoter ICP0 and a monomeric red fluorescein protein (RFP) under the control of late viral glycoprotein C (gC) (17). HSV-1 KOS EGFP/RFP virus was kindly gifted by Paul R. Kinchington, University of Pittsburgh (Pittsburgh, PA). The virus was propagated in Vero cells, as described before (16).
For ocular infection, seven- to eight-week-old BALB/c mice (n = 4, with each experiment repeated three times) were infected under the injection of 1.25% solution of avertin anesthetic (2,2,2-Tribromoethanol; Acros Organics, Pittsburgh, PA). Ocular scratching was performed in the absence of tail and toe pinch reflexes. The epithelial surface of the anterior chamber on the cornea from each mouse was scarified 15 times per side in an X-shaped pattern using the tip of 27-gauge needle without damaging the globe or iris of the eye globe. A 3-μl drop of virus inoculum containing various virus doses (HSV-1 RE; 1 × 103, 5 × 103, and 1 × 104 PFU/3 μl) was applied to the eye after corneal scarification. The day of infection was counted as day 0, and the following days were mentioned as day(s) postinfection (dpi). A group of mice were also corneal scarified but mock-infected with PBS. For the latency experiments, seven- to eight-week-old BALB/c mice (n = 12) were subjected to ocular infection with HSV-1 KOS EGFP/RFP virus (1 × 106 PFU/eye).
Drug administration, sampling, and clinical scoring
2DG (Sigma‐Aldrich, St. Louis, MO) was dissolved in sterile PBS and injected into mice i.p. Mice had received 250 mg/kg of 2DG in 0.2-ml volume twice a day, in 10-h intervals, starting from the day of infection (day 0) up to day 10 postinfection (pi). The control group had received an equal volume of PBS. Mice were infected ocularly 5 h after receiving first injections for treatment, and a group of mice was also either mock-infected or left untreated with a drug at day 0. HSV-1 RE ocular-infected mice were regularly monitored and recorded for development of HSV-1–related lesions, including stromal keratitis (SK), angiogenesis, and clinical appearances. To evaluate viral replication in the eye, swap sampling was done daily, starting from day 1 up to day 8 pi. PBS-soaked swabs were applied on the surface of the eye, and swabs were collected individually. Swab specimens were stored in 2% FBS and antibiotic-containing PBS and kept at −80°C until performing virus quantification. The SK lesion severity and angiogenesis in the eyes of mice were examined through slit lamp biomicroscopy (Kowa Company, Nagoya, Japan) and recorded at 9 dpi along with a recording of some image examples from average score-showing mice. The severity of angiogenesis and the scoring of SK in the eyes were recorded as described previously (18, 19). SK and angiogenesis were documented in a blind manner. Development of lethargy and encephalitis with the sign of neurologic changes, including jumping when touched or circling itself when hanging by the tail, were also examined independently and recorded in percentage changes. Mice were euthanized if they did not improve after opening a clinical case with 24-h watch by OLAC veterinarians. Mice were recorded in survival analysis based on 24-h watch monitoring period results. To monitor the viral replication in the TGs from ocular-infected mice, mice were euthanized on days 2 and 3 pi and on 2-d intervals until the experiment's termination on day 9 pi. TGs were excised from the brain cavity, and TGs were processed individually. TGs were mechanically disrupted with tissue grinders (Pellet Pestle, Kontes). Collected and homogenized TGs were freeze–thawed three times to collect supernatants and assayed to quantify the HSV-1 titer on Vero cells.
Ex vivo latency and reactivation
A latency experiment was described previously and applied to this study with some modifications (20). Latently infected TGs (5 wk after ocular infection) were excised from HSV-1 KOS EGFP/RFP–infected animals (1 × 106 PFU/eye). TGs were enzymatically dissociated with collagenase type I (Sigma-Aldrich, St. Louis, MO) at a 3 mg/ml concentration for 45 min at 37°C and mechanically disrupted with a tissue grinder. Contaminating erythrocytes were removed using erythrocyte lysis buffer. Homogenized and pooled TGs were dispersed in wells of 24-well culture plates containing the 2 TGs per well in total. To measure the reactivating virus titer in the presence of 2DG, TGs maintained for 8 d in the presence of complete DMEM or in glucose-free media (BRL Life Technologies, Grand Island, NY) containing 10 U/ml recombinant murine IL-2 (PeproTech, Rocky Hill, NJ). A group of wells was supplemented with 2DG (0.5 mM), and supernatants were replaced with fresh culture media (1:20) containing the same concentration of 2DG by 2-d intervals. Control wells were replaced with complete media. Preliminary experiments were performed with T cells to establish that this dose successfully inhibited glycolysis and was nontoxic to the cells. Collected supernatants were tested on Vero cells, and a plaque assay was performed, followed by crystal violet staining as described before (15). Some of the wells were examined under the ×4 objective lens of the EVOS FLc Imaging System (Thermo Fisher Scientific, Waltham, MA) at day 4, and images were acquired for the expression of early (EGFP–ICP0) and late (RFP–gC) viral proteins. In a separate experiment to measure the endogenous T cell–mediated latency maintenance, T cells were depleted with the addition of 150 μg/ml anti-CD4 (clone GK 1.5; Bio X Cell) and anti-CD8 (clone 2.43; Bio X Cell) mixture to latently infected and homogenized TG culture media, as described before (20). In a comparison of the effect of 2DG on T cell activation during the reactivation, a group of wells were either supplemented with 2DG (0.5 mM) or left untreated. Reactivation was terminated at 3 d of explant culture. The supernatant was tested on the Vero cell to measure reactivated virus titer in a crystal violet staining experiment.
In vivo BBB permeability and extravasation of Evans blue
To investigate the BBB leakage, Evans blue dye (Sigma-Aldrich, St Louis, MO) was administered to mice as described previously with some modifications (21). Mice (4 dpi or 8 dpi) received 2% of Evans blue dye (4 ml/kg) that was dissolved in PBS and injected to mice via an i.p. route and left for 16 h. A cardiac PBS perfusion was performed on mice under the effect of a lethal dose of avertin (50 mg/kg). The brain was excised from the cranial cavity and kept on ice. To quantify leaked Evans blue from BBB, a TCA (Sigma-Aldrich, St Louis, MO)-based dye extravasation assay was applied as described (22). To quantify the dye leakage, supernatants were diluted with 95% ethanol (1:3; v/v), and Evans blue fluorescein intensity was measured at 620/680 nm wavelength (Synergy HTX; BioTek) to obtain absorbance values. Absorbance values were measured with Evans blue dye standards and converted to µg/g of the brain.
H&E histological staining
For corneal histological examination, corneas were excised from mice and kept in 10% buffered formalin solution. Tissues underwent routine histologic processing, and staining was performed with H&E as performed previously (23).
Harvesting tissue samples for single-cell analysis and flow cytometry
On day 9 pi, individual corneas and TG were excised, suspended in complete RPMI 1640 (10% FBS in RPMI 1640), and digested with Liberase at 1.5 mg/ml (Roche Diagnostics, Indianapolis, IN) for 45 min at 37°C in a humidified atmosphere of 5% CO2. In a separate experiment, the study was terminated at 5 dpi, and tissues were processed to obtain single cells. Corneas and TG were disrupted by crushing with a 1-ml syringe plunger on a 40-micron cell strainer, and a single-cell suspension was kept in complete RPMI 1640. All samples were divided or counted for the analysis of both innate and T cell responses.
Flow cytometry analysis
To analyze T cell responses, single-cell suspensions were isolated from corneal and TG tissues and divided in half. Half of the suspensions were stimulated in a mixture of PMA (50 ng), ionomycin (500 ng), and brefeldin A (10 µg/ml) for 4 h for T cell analysis, and the other half was left unstimulated for the investigation of innate cell responses. All staining applications were performed at 4°C. For the flow cytometry analysis of T cells, cells were incubated with anti-CD16/CD32 (Fc block) for 15 min at room temperature along with the reagent for the LIVE/DEAD staining kit. Cell surface staining was done with respective surface fluorochrome-labeled Abs in flow cytometry buffer for 30 min. For intracellular IFN-γ staining, cells were permeabilized with an intracellular staining kit (BD Biosciences, San Jose, CA) for 30 min. Afterward, cells were stained with anti–IFN-γ Abs for 30 min. Finally, the cells were washed thrice with flow cytometry analysis buffer and resuspended in 1% paraformaldehyde. The fluorochrome-stained cells were examined to compute the cell with different phenotypes using an LSR II (BD Biosciences, San Jose, CA), and the data were analyzed using FlowJo software (Tree Star, Ashland, OR). All doublet cells were gated out during the gating of cells for flow analysis, followed by gating of the live-cell population. Cells were gated from live populations and defined as CD4+, CD8+, CD69+ and IFN-γ+ T cells. Proinflammatory T cells were gated as CD4+IFN-γ; activated T cells were gated either CD4+CD69+IFN-γ+ T cells or CD8+CD69+IFN-γ+ T cells. To study innate cell responses by flow cytometry, cells were blocked and stained in mixture of LIVE/DEAD Staining Kit and anti-CD16/CD32. Cells were stained with a multistaining Abs mixture containing anti-CD45, CD11b, Ly-6G, and F4/80. To analyze the neutrophil phenotype cells, CD45+ CD11b+ Ly-6G+ gating was done, and for the analysis of macrophages, CD45+ CD11b+ F4/80+ gating was performed.
Reagents and fluorochrome-conjugated Abs for flow cytometry analysis
The LIVE/DEAD Fixable Near-IR Staining Kit was from Thermo Fisher Scientific. Rat anti-CD4 (RM4-5), anti-CD8 (53-6.7), anti-CD69 (H1.2F3), anti–IFN-γ (XMG1.2), anti-CD45 (30-F11), anti-CD11b (M1/70), anti–Ly-6G (1A8), protein transport inhibitor (brefeldin A), and unconjugated anti-CD16/32 (2.4G2) Abs were obtained from BD Biosciences (San Jose, CA). A Fixation/Permeabilization Solution Kit was also obtained from BD Biosciences (San Jose, CA). Anti-F4/80 (BM8) was obtained from eBioscience (Invitrogen, Carlsbad, CA). Phorbol 12-myristate 13-acetate-PMA and ionomycin were purchased from Sigma-Aldrich (St. Louis, MO).
In vitro BBB endothelial cell
To study the effect of 2DG on in vitro BBB and mRNA expressions of permeability-maintaining proteins of primary mouse brain microvascular endothelial cells (PMBVEC) (BALB-5023; Cell Biologics, Chicago, IL) was purchased and seeded in a 24-well plate at 1 × 105/ml with a Complete Mouse Endothelial Cell Medium Kit (M1168; Cell Biologics, Chicago, IL). The 24-well plates were precoated with gelatin-based coating solution (6950; Cell Biologics, Chicago, IL).
RT-PCR and mRNA amplification
To quantify mRNA expressions of occludin, zonula occludens-1/tight junction protein-1 (ZO-1/Tjp-1) and claudin-5 from PMBVEC cells were maintained in 24-well plates, and total RNAs were isolated from the wells by using an RNA Isolation Kit (QIAGEN) on a daily basis up to day 4 after the seeding of the cells. Cells were lysed directly in the wells after the removal of the supernatant. All procedures followed the kit manufacturer’s standard protocol. cDNA was converted from 500 ng of isolated RNA with the help of GoScript Reverse Transcription System (Promega). TaqMan FAM-labeled anti-mouse specified primers were obtained and purchased from Thermo Fisher Scientific under the Applied Biosystems trademark, and quantitative PCR (qPCR) assays were performed by using a 7900 Fast Real-time PCR System (Applied Biosystems). TaqMan gene expression assay probes directed to target for Ocln (Mm00500912_m1), ZO-1/Tjp-1 (Mm00493699_m1), and Cldn5 (Mm01169675_s1) gene regions were employed during the quantitative RT-PCR amplification processes and mixed with TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific, Waltham, MA). A β-actin (Mm00607939_s1) was included as an internal amplification control. The relative expression levels of different molecules were normalized to β-actin using a cycle threshold calculation. The relative expression between experimental groups or isolated total RNAs was calculated using the 2−ΔΔCT × 1000 formula. Results presented with relative mRNA expressions. The 2DG dose concentrations and effective dose inhibition responses evaluated in the presence of various 2DG concentrations in culture medium was as described previously (14, 24).
GraphPad Prism 7 software (GraphPad Software, La Jolla, CA) was used for statistical analysis and presentations of graphs in most figures. Kaplan–Meier survival estimation analysis was used to define percentage survivability, and log-rank (Mantel–Cox) tests were performed, and significance was estimated. A Sidak multiple comparisons test was applied with two-way ANOVA to calculate significance of importance between 2DG-treated and -untreated groups in the data analysis of viral load on eye swabs, whereas one-way ANOVA with multiple comparison statistical analysis was performed, followed by Tukey test in the Evans dye quantification analysis. A two-tailed nonparametric Student t test analysis was performed to estimate significant differences between 2DG-treated and -untreated groups. The level of significance for some experiments was revealed as follows: ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05. Some results indicate p > 0.05 and are expressed not significant (ns). SD of each median was considered for some analysis, and it was considered as SEM during data analysis of flow cytometry analysis results.
Effect of 2DG therapy on outcome of ocular infection with HSV
Initial experiments were done to determine the susceptibility of young adult BALB/c mice to ocular infection with different doses of HSV-1 RE. All animals survived infection at the lowest dose tested (1 × 103 PFU) and showed no clinical signs of encephalitis, although a few animals developed mild ocular lesions. However, with an infection dose of 1 × 104 PFU, 45% of mice either died or had to be terminated because of advanced signs typical of encephalitis. These included ruffled fur, hunched posture, excitability to touch and eventual lethargy. In all cases of encephalitis, virus could be cultured from brain stem samples. An intermediate dose was chosen for subsequent studies (5 × 103 PFU), which, on average from three separate experiments, caused encephalitis necessitating termination in around 12.5% of mice. Most of the animals infected at this dose level (including those that developed encephalitis) had ocular lesions (angiogenesis and inflammatory reactions in the corneal stroma) by the time experiments were terminated on day 9 or 10.
To evaluate the effect of inhibiting glycolysis on the outcome of HSV ocular infection, mice were infected with 5 × 103 HSV RE, and one group received twice daily systemic therapy with 250 mg/kg 2DG from the day of infection. A third group was treated in the same way with 2DG, but these animals were not infected with HSV. Experiments proceeded for 10 d with animals being terminated earlier if showing advanced signs of encephalitis. The combined results of three separate experiments of the same design are shown in (Fig. 1A, 1B. They indicate that uninfected mice treated with 2DG had no detectable clinical consequences. However, the clinical outcome of infected mice was altered following treatment with 2DG. Significantly more infected mice developed encephalitis when treated with 2DG. The experiments revealed that, whereas around 12.5% of control animals developed encephalitis (confirmed as virus-positive after euthanasia), >85% of 2DG recipients developed encephalitis. Curiously, the time of onset of neurologic signs was usually delayed by 1 d in 2DG-treated animals compared with infected control mice (Fig. 1A), but their disease progressed more rapidly once it became evident (Fig. 1A, 1B).
Although 2DG therapy made animals more susceptible to developing encephalitis, these animals had diminished ocular inflammatory reactions. Eye swab samples were taken from animals at different times pi to quantify levels of virus present. Peak responses were present in 48 h samples and were, on average, 1 log higher in animals treated with 2DG. Animals in the drug-treated group also stayed positive for a longer period than untreated controls (see (Fig. 2). With respect to ocular inflammatory lesions, in one experiment, whereas 16 out of 16 eyes in the untreated group showed lesions by day 9, with a mean score of 3.3, in the 2DG-treated mice, only 11 of 16 eyes developed detectable ocular SK and lesions and were of a milder mean score of 1. The eyes of 2DG-treated mice had markedly fewer inflammatory lesions and less angiogenesis, especially at the later time periods (Fig. 3A, 3B). Examples of average responses of each group, recorded by histopathology at day 9 pi are shown in (Fig. 3C, 3D, which shows that the inflammatory reaction in the corneal stroma was markedly reduced in 2DG-treated animals.
Some corneas were also removed for Liberase digestion to enumerate and characterize the cell types present in the stroma of treated and control-infected mice. Samples collected on day 9 revealed the presence of CD45+ innate inflammatory cells, but there were 19-fold fewer recovered cells in the 2DG recipients (Fig. 4A, 4B). Moreover, when comparing the numbers of macrophages (CD45+CD11b+F4/80+) and neutrophils (CD45+CD11b+Ly-6G+) in day 9 samples, the experiment showed that macrophage numbers were reduced 19-fold in 2DG (Fig. 4C, 4D) animals, and neutrophil numbers were reduced by 29-fold (Fig. 4E, 4F). The number of total CD4+ T cells, as well as those that were IFN-γ producers (Fig. 5A–D), the main orchestrators of SK lesions (25, 26), were also compared. Total CD4+ T cells and IFN-γ–expressing CD4+ T cells present in samples from 2DG recipients were reduced 3-fold and 9.9-fold, respectively, compared with untreated infected controls (Fig. 5A–D).
Taken together, these results demonstrate that inhibition of glucose metabolism in mice infected with HSV can result in encephalitis, but the approach reduced the inflammatory damaging reaction in the eye to HSV infection.
Responses in the TG in 2DG-treated and -untreated infected mice
Acute infection experiments
Experiments were also done to compare the TG response following ocular infection of 2DG-treated and -untreated mice. The TG is the site where HSV rapidly locates after ocular infection and where it establishes latency (27). In these experiments, BALB/c mice were infected with HSV RE (5 × 103 PFU/eye), and TGs were collected at different times from infected control and 2DG-treated animals. The level of infectious virus was quantified after the TGs were minced and freeze–thawed three times. At the first collection time (day 3), replicating virus was detectable in samples from both groups, and titers were similar in magnitude (ns, p > 0.05) (Fig. 6). However, in samples collected on days 7 and 9, replicating virus was absent in the TGs from control-infected animals but was still present in readily detectable quantities (> 1 × 103 PFU) in the TG from 2DG-treated animals (Fig. 6). These data indicate that, after the initial period, all virus in the untreated animals was in the latent form, but virus was still actively replicating throughout the observation period in the TG of 2DG recipients. This might be occurring because the inflammatory cells present in the TG, which are thought to control infection and favor latency maintenance (28–30), were too few in number and perhaps functionally impaired in mice that received the metabolic inhibitor.
To gain further insight on the effects of 2DG therapy on inflammatory cells in the TG, experiments were done in which control-untreated and 2DG-treated BALB/c mice were ocularly infected with HSV RE, and single-cell suspensions were prepared from individual TGs on day 9 to record the numbers of inflammatory cells present. In samples from treated animals, the number of innate cells (CD45+) compared with untreated controls was reduced, on average 2-fold (Fig. 7A, 7B), which applied to both macrophages (CD45+, CD11b+, F4/80+) (Fig. 7C, 7D) and neutrophils (CD45+, CD11b+, Ly-6G+) (Fig. 7E, 7F). The numbers of CD4 and CD8 T cells were also reduced in 2DG-treated samples. The CD4 cells were reduced by, on average 3-fold (Fig. 8A, 8B), and CD8 cells were reduced by 4.2-fold (Fig. 8B, 8C). Some TG cultures were stimulated with PMA/ionomycin, and the number of CD4+ IFN-γ–producing cells (Th1) were enumerated. These were, on average, 3-fold reduced in TG from 2DG-treated animals (Fig. 8D, 8E).
We could also show using flow cytometry analysis that the frequency of cells showing activation and the functional marker IFN-γ were diminished in 2DG recipients. Thus, in mice treated with 2DG, the frequency of CD69-expressing CD8 T cells was decreased by 12% (Fig. 9A, 9B), and the frequency of CD4+CD69+ was decreased by 10% (Fig. 9C, 9D). CD8+ and CD4+ T cells that expressed IFN-γ were diminished by 16% (Fig. 9E, 9F) and 7%, respectively (Fig. 9G, 9H). Taken together, these data support the notion that 2DG therapy inhibits the participation of inflammatory cells in their function of maintaining latency.
Ex vivo reactivation experiments
Experiments were also done to measure the effects of 2DG on the efficiency by which HSV reactivates from latency. For this purpose, BALB/c mice were ocularly infected with an HSV KOS strain of virus that expressed both the EGFP and RFP markers. This virus was less pathogenic than HSV RE, with no animals succumbing to encephalitis, even when given 1 × 106 PFU/eye. Animals developed mild ocular lesions that did not cause major irritation or face scratching, and the lesions resolved. Consequently, mice could be maintained for 5 wk without needing to be culled for ill health. The TGs were collected and pooled at 5 wk pi and after enzymatic dissociation, single-cell suspensions were prepared, and multiple cultures were established in 24-well plates. Half of the cultures received normal media (DMEM, 10% FBS) and the others the same media to which 0.5 mM 2DG was added (14). The cultures were maintained for up to 8 d, with the culture fluids removed and replaced daily. The amount of virus present in the removed fluids was quantified using Vero cell monolayers. These experiments revealed that cultures that contained 2DG became virus-positive one day earlier, on day 3, compared those without 2DG, which became positive on day 4 (Fig. 10A). Comparisons of viral levels at day 4 were significantly higher in 2DG-containing cultures, as also could be shown by confocal microscopy of some cultures (Fig. 10B). Peak levels of virus were present on day 6, and these were similar in magnitude in both drug-treated and -untreated cultures. Some virus was still detectable in all cultures on day 8 (Fig. 10A).
These experiments revealed that the presence of 2DG accelerated the breakdown of latency possible because of inhibitory effects on lymphoid cells in the TG cultures that were helping to maintain latency. Some additional support for this notion was obtained in experiments in which Abs to both CD4 and CD8 T cells were added to some TG cultures at a concentration previously shown to inhibit T cell activity; the effect on time of viral reactivation was evaluated and compared with cultures with 2DG-treated and -untreated controls (20). The results showed that, by day 3, the cultures that contained the anti–T cell Abs, as well as those treated with 2DG, showed viral reactivation by day 3 (Fig. 10C). In contrast, in the control cultures, virus did not reactivate until day 4 (Fig. 10A). Curiously, reactivation did not occur in the absence of glucose (Fig 10C). These experiments indicate that 2DG likely acts by affecting the latency control activity of TG T cells. Conceivably, the effect of 2DG also could be directed at the function of latently infected neuronal cells, but experiments comparing the reactivation kinetics of TG cultures from which T cells were removed showed no differences when cells were cultured in normal media or media with added 2DG (data not shown).
Does HSV access the brain via the BBB?
Although the previously described experiments indicate that the facilitated entrance of HSV to the CNS that occurs when glucose metabolism is inhibited could be the consequence of infection control in the TG, permeability alterations in the BBB may also be involved. Experiments were performed to determine whether daily 2DG therapy would damage the BBB, making it more permeable for passage by a virus into the CNS. Initial experiments were done to compare leakage of the dye Evans blue from the vasculature into the brain in both uninfected controls and uninfected 2DG-treated animals as assessed at 5 and 9 d after dye injection. No significant differences were observed between treated and untreated animals (data not shown). However, in virus-infected controls and infected 2DG-treated animals, levels of dye in brain samples examined at 9 d were significantly elevated over that observed in uninfected mice (Fig. 11A, 11B). In addition, almost 2-fold more dye was seen in the brains of daily 2DG-treated infected mice. We considered the possibility that the increased BBB leakage caused by HSV and 2DG therapy might be the consequence of hematogenous HSV affecting the BBB; however, we failed to detect virus in blood at any stage pi (data not shown).
Although our experiments provided no indication that inhibiting glucose metabolism with 2DG caused the BBB of uninfected mice to become more permeable to leakage of Evans blue dye, we also performed experiments in vitro using PMBVEC to measure the effect of 2DG on the production of molecules known to control the permeability of the BBB. These were the proteins occludin, ZO-1/Tjp-1, and claudin-5. Cultures were terminated at different time points, and the levels of mRNA for the three proteins as well as β-actin as a control were quantified by qPCR. Initial experiments were also performed with the PMBVEC to measure any toxic effects of different concentrations of 2DG during a 72 h period. The test experiments were done using a range of 2DG concentrations (up to a 50% toxic level), and the concentrations of mRNA for the three permeability proteins and β-actin were quantified at different time periods (data not shown). The results in (Fig. 12A–C show that the presence of 2DG at nontoxic levels resulted in significant inhibition of permeability protein mRNA levels, with the greatest effect on claudin-5. Curiously, some reports have indicated that claudin-5 can play the most relevant part in controlling the permeability of the BBB (31–33).
Few viruses succeed in accessing and replicating in the brain, but when they do, the outcome can be devastating and is often lethal. HSV is an occasional cause of encephalitis in adult humans, and without rapid antiviral therapy this disease has debilitating consequences (10). Heterologous hosts infected with HSV, such as some primates and laboratory rodent models, are more prone to develop herpes encephalitis, as is the homologous human host if immune functions are compromised (8, 34). Thus, as our own previous studies revealed, mice given 2DG, which inhibits their ability to use glucose, from the onset of infection were highly prone to develop encephalitis (14). In this report, we further explore this outcome and provide reasons to explain their heightened susceptibility to encephalitis. We showed, using a model of BALB/c mice that were infected in the cornea with the RE strain of HSV-1, that inhibiting glucose use using 2DG administration resulted in most mice developing encephalitis. In contrast, the infected 2DG-treated mice developed far less severe ocular lesions, a response known to be mediated by the host immune response to the infection (25, 26). Indeed, our studies showed markedly diminished inflammatory reactions in the eyes of 2DG-treated mice. In comparing the outcome of infection in the TG, the site to which virus spreads, replicates, and establishes latency (35, 36), marked differences in viral and cellular events were observed between treated and untreated animals. In control-untreated mice, replicating virus was present only during early time points, whereas in 2DG-treated mice, replicating virus was present for the entire nine-day observation period. This outcome was likely the result of the 2DG impairing the virus-controlling function of inflammatory cells, particularly T cells, in the TG. In fact, comparisons of the inflammatory cell contents in the TG from untreated and 2DG-treated animals revealed that 2DG-treated animals had significantly decreased numbers of innate cells as well as T cells. Moreover, we showed that T cells in the TG of treated animals were less activated, and a lesser fraction expressed the protective IFN-γ–producing function compared with untreated controls. Other studies revealed that inhibiting glucose metabolism in vitro TG cultures from previously infected mice with established latency resulted in accelerated breakdown of latency and increased the amount of virus produced by TG, especially in the early stages. In addition, the effects of inhibiting T cell activity in TG cultures were similar to 2DG treatment, further indicating that 2DG acted by compromising T cell activity in the TG. Taken together, the results of both in vivo and in vitro investigations demonstrate that 2DG therapy impairs the protective effects of one or more inflammatory cell types in the TG that normally function to control productive infection. This local inhibitory effect could allow more prolonged virus production and facilitate anterograde passage of the virus along the nerve axons to the brain, but this outcome was not formally demonstrated (37). What still needs further study is the determination by which the mechanisms of the 2DG therapy mediated its effect and whether, indeed, it was directed only at T cell function or had additional off-target activities noted by others, such as antiviral effects (38, 39). In our system, the latter mechanism was unlikely because the use of 2DG resulted in higher levels of virus production. We did not ascertain which metabolic pathways were primarily affected by 2DG in the system we studied, but investigations are currently underway to provide this information.
We also evaluated the alternative explanation that by inhibiting glucose metabolism, 2DG made mice more susceptible to herpes encephalitis as a result of altered function of cells of the BBB making the barrier more permeable to viral passage into the brain. However, we failed to find evidence that the regimen used for 2DG treatment in normal untreated mice caused any significant changes to BBB permeability. Nevertheless, the permeability of the BBB was increased in HSV-infected mice, and this increase was even greater in HSV-infected 2DG recipients. We suspect that an explanation for these findings could be that HSV, once present in the brain from access via the neural route, could be damaging the BBB from the brain side; this possibility needs to be further explored. Accordingly, once HSV has entered the CNS, it is known to infect and change the function of astrocytes that form part of the BBB (40–42). It was also curious to observe that in vitro studies using brain-derived epithelial cells revealed that the presence of 2DG in cultures resulted in the diminished production of occludin, ZO-1/Tjp-1, and claudin-5, a group of molecules involved in maintaining the restrictive permeability of the BBB (31). Thus, changes in BBB permeability during HSV infection, especially if treated with inhibitors of glucose metabolism, was observed, but whether viral passage can occur via the BBB route requires further study.
The α-herpes virus infection, in its homologous hosts rarely gains access to the CNS. This is fortunate because severe neuropathology usually results (43). In humans infected with HSV, lesions are usually confined to peripheral sites, and the virus travels retrograde in peripheral nerves to the local ganglion, where it is maintained in a nonreplicating latent form, likely for life (44). When the virus reactivates from latency, a common occurrence, it almost invariably disseminates from the ganglion to the peripheral site but almost never ascends in axons that enter the brain (45). However, HSV infection of the brain does become a reality under some circumstances that include inadequate control by the immune system. This is most commonly seen when newborns from seronegative mothers are infected with HSV (7). However, occasionally, HSE does occur in immunocompetent adults, with the syndrome usually happening in persons already latently infected with HSV and who may have experienced previous HSV lesions confined to the periphery (46). The reason HSV is usually excluded from entry to the CNS is unclear, but this report raises the possibility that concurrent metabolic imbalances at the time of infection or virus reactivation may play a role. Accordingly, metabolic imbalances could compromise the protective function of one or more components of the immune system that normally control infection and preclude the passage of virus in the axons that access the CNS. In support of this possibility, we showed in a model system that mice infected with HSV that additionally received 2DG, which inhibits glucose metabolism in the cells mainly responsible for antiviral control, were highly likely to develop lethal HSE (14). Others have also advocated that the nature of the metabolic environment at the peripheral site at the time and place where HSV reactivates from latency in humans could explain why such reactivations are sometimes subclinical but at other times symptomatic and tissue damaging (47).
Our results in a mouse model system provide evidence that viral dissemination to the brain was most likely the consequence of the 2DG therapy causing a reduction in the number and function of inflammatory cells in the TG that under normal circumstances act to suppress viral replication and favor the maintenance of latency. The idea that HSV replication is controlled and that latency is established and maintained in the TG by the action of inflammatory cells, particularly CD8 T cells, was initially advocated by the Hendricks group (20, 48). However, the hypothesis remains controversial, and the topic of how HSV latency is established, maintained, and terminated excites passionate debate among herpes virologists. Our studies add support for the inflammatory cell control notion because we could show that in animals with inhibited glucose metabolism, the numbers and function of inflammatory cells present in the TGs were significantly diminished in 2DG recipients, and viral replication was increased. We assume that the consequence of the lack of control by the 2DG-inhibited T cells also could result in virus spilling into the trigeminal root axons, with virus gaining access to the brain to cause encephalitis. However, confirmation of this pathogenesis will require further studies to detect and quantify the presence within and follow the passage of virus in nerve axons and entry sites into the brain, a challenging study for which we would need to involve talented collaborators.
A second series of experiments, designed to reveal the effects of metabolic manipulation on the pathogenesis of herpes encephalitis, was performed using explanted TG from animals with well-established latency. In these experiments, individual TG were collected five weeks after ocular infection, a time when replicating virus was no longer present and animals were viable, healthy, and any mild eye lesions that initially resulted from the ocular infection had fully resolved. Such studies revealed that the inclusion of 2DG in the media during ex vivo culture accelerated the time when virus reactivated from latency by on average one day. The effect on 2DG was assumed to be the consequence of inhibiting the function of inflammatory cells, particularly T cells, still present in ganglia at that stage of infection because the effect caused by inhibiting glycolysis could be duplicated if the T cells were inhibited with specific Abs. The observations that 2DG impairs the expansion and function of proinflammatory cells is well known from in vitro studies that include those done previously in our own laboratory (14).
Many virus infections that do access the brain accomplish the effect by crossing the BBB, which in many instances needs to be made more permeable for the viral passage to occur (2). Consequently, we expected that the HSE observed in the HSV-infected mice that received 2DG might be explained by the metabolic inhibitor causing damage to the permeability function of the BBB. However, we failed to find evidence in vivo that treating uninfected mice with 2DG caused their BBB to become more permeable, at least as measured by the intravascular leakage into the brain of Evans blue dye. However, there could have been subtle effects on the function of the BBB that could not be detected because in an in vitro model system of brain endothelial cells we could show that the presence of 2DG in the culture media under conditions that inhibited glycolysis did significantly reduce the expression of three proteins involved in controlling the permeability function of the BBB. Moreover, the inhibitory effects of 2DG were more apparent against the tight junction protein claudin-5, the protein that some studies have advocated is mostly responsible for maintaining impermeability of the BBB (32, 49). Currently, we are exploring the use of improved in vitro approaches to evaluate the influence of metabolic inhibitors on BBB function, but at the present stage of our investigations, we favor the hypothesis that encephalitis in the mouse model of encephalitis is more likely to be the consequence of virus passing to the brain via innervating nerve tracts rather than passage via the BBB.
We thank Manikannan Mathayan, Sujata Agarwal, and Siddheshvar Bhela for contribution to the study. We thank Paul R. Kinchington for providing a stock of HSV-1 KOS-ICP0-EGFP/gC-RFP. We also thank University of Tennessee, Knoxville, TN Animal Biosafety Level-2 facility staff members, OLAC technicians, and veterinarians for visits and clinical inspections.
This work was supported by Foundation for the National Institutes of Health Grants R21AI142862 and R01EY005093.
Abbreviations used in this article
enhanced green fluorescein protein
herpes simplex encephalitis
Office of Laboratory Animal Care
primary mouse brain microvascular endothelial cell
red fluorescein protein
zonula occludens-1/tight junction protein 1
The authors have no financial conflicts of interest.