Type-1 angiotensin receptor signaling in central nervous system myeloid cells is pathogenic during fatal alphavirus encephalitis in mice

Inbred C57BL/6 mice, mice heterozygous for deficiencies of the individual Nox subunits, p47^phox or gp91^phox, both bred on a C57BL/6 background, mice expressing green fluorescent protein (GFP) in one allele of the CX3C chemokine receptor-1 (CX3CR1 or fractalkine receptor) bred on a C57BL/6 background (hereafter referred to as CX3CR1^GFP/+ mice) and mice deficient in AT1R bred on a C57BL/6 background (hereafter referred to as AT1R knockout (KO) mice) were all obtained from Jackson Laboratories (Bar Harbor, ME). Double KO gp91^phox/p47^phox mice (hereafter referred to as gp91/p47 double knockout (DKO) mice) were generated de novo and bred on site. Animals were housed under specific pathogen-free, barrier facility conditions on a 10-h/14-h light/dark cycle with food and water available ad libitum. All animal procedures were completed using isoflurane anesthesia (Abbott Laboratories, Chicago, IL). All animal experiments, including those incorporating paralysis or death as study endpoints, were conducted in strict accordance with protocols approved by the University of Michigan Committee on Use and Care of Animals as well as with federal guidelines.

NSV viral stocks were grown and assayed for plaque formation on BHK-21 cells. Stock titers of 10⁷ plaque-forming units (pfu)/ml were stored at −80 °C until use. To induce encephalomyelitis, mice were injected with 1000 pfu of NSV or NSV-GFP suspended in 20 μl of phosphate-buffered saline (PBS) directly into the right cerebral hemisphere. For those experiments where tissue samples were not being collected for ex vivo analyses, each infected animal was scored by a blinded examiner into one of the following groups: (1) normal or minimally affected, (2) mild paralysis (some weakness of one or both hind limbs), (3) moderate paralysis (weakness of one hind limb, paralysis of the other hind limb), (4) severe paralysis (complete paralysis of both hind limbs), or (5) moribund/dead. Moribund mice were euthanized immediately and were considered to have died the following day for all statistical comparisons, as prior studies demonstrated that all animals reaching this disease stage never survived more than another 24 h [7, 10, 13]. Animals also received intraperitoneal (i.p.) injections of telmisartan (Sigma-Aldrich, St. Louis, MO) or a vehicle control twice daily in a volume of 200 μl of 0.9 % sodium chloride solution, pH 9.0. Such an alkalinized diluent was both necessary for and capable of maintaining full drug solubility. Doses up to 100 mg/kg/day were administered starting 12 h after viral challenge. Other mice received twice daily i.p. injections of captopril (10 mg/kg/day, Sigma-Aldrich) suspended in 200 μl of PBS or a PBS vehicle control. All of these experiments were conducted in strict accordance with protocol # PRO00005049 approved by the University of Michigan Committee on Use and Care of Animals.

At predetermined disease stages, some mice were euthanized via transcardiac perfusion with PBS under anesthesia. For animals where in vivo labeling of ROS production was desired, mice were first injected i.p. with 50 mg/kg dihydroethidium (DHE, Sigma-Aldrich) 1 h prior to perfusion. For all tissue enzyme immunoassays (EIA), bead-based multi-analyte cytokine and chemokine assays, Western blots, flow cytometry-based sorting of immune cell subsets, virus titration assays, and biochemical assays of tissue superoxide production, brains and spinal cords were quickly dissected for further processing. For immunohistochemical and immunofluorescence studies, animals were perfused a second time with chilled PBS containing 4 % paraformaldehyde (PFA) before dissection.

The Cba 87 strain of western equine encephalitis virus (WEEV) was generated from the complementary DNA (cDNA) clone pWE2000 as previously described [30, 31]. A low passage stock of infectious virus generated in Vero cells was expanded twice in C6/36 mosquito cells to produce viral stocks containing 10⁷ pfu/ml. All virus aliquots were stored at −80 °C until use. Experiments using infectious WEEV were conducted in an animal-compatible, federally certified biosafety level 3 (BSL3) containment facility in strict accordance with standard operating procedures and protocols approved by both the University of Michigan Institutional Biosafety Committee and the University of Michigan Committee on Use and Care of Animals (animal protocol # PRO00005204). Mice were housed in our animal BSL3 facility on a 10-h/14-h light/dark cycle with food and water available ad libitum.

For WEEV infections, recipient mice were inoculated subcutaneously with 10⁴ pfu of Cba 87 suspended in a total volume of 50 μl of PBS. Cohorts of infected mice were weighed and scored daily using a clinical rating scale previously established for WEEV infection: 0, normal; 1, slightly ruffled fur, no visible signs of infection; 2, very ruffled fur, definite signs of infection with reduced cage activity; 3, very ruffled fur, hunched posture, reduced mobility; 4, very ruffled fur, hunched posture, little to no mobility, rapid breathing (moribund); and 5, dead [32]. All animals also received i.p. injections of telmisartan (Sigma-Aldrich) or a vehicle control twice daily in a total volume of 200 μl of 0.9 % sodium chloride solution, pH 9.0, at a dose of 100 mg/kg/day starting 12 h after viral challenge. Animals reaching a disease score of 4 were euthanized immediately and were considered to have died the following day for all statistical comparisons, as prior disease progression studies demonstrated that moribund mice never survived more than an additional 24 h [32].

Control and NSV-infected mice were perfused with PBS, and brains and spinal cords extracted, snap-frozen on dry ice, and stored at −80 °C until use. Thawed tissues were minced and homogenized in 0.5 ml of PBS containing a protease inhibitor cocktail (Roche Life Science, Indianapolis, IN). Homogenates were centrifuged to pellet all remaining tissue debris. The total protein content in each tissue supernatant was measured using the Pierce Coomassie Protein Assay Reagent (Thermo Fisher Scientific, Rockford, IL). Ang II levels were measured in each sample using a commercially available EIA kit (Cayman Chemical Company, Ann Arbor, MI). The results presented reflect pg Ang II/mg of total protein extracted from tissue of four to six animals at each time point examined. The Milliplex mouse cytokine detection system (EMD Millipore, Billerica, MA) was used according to the manufacturer’s instructions to quantify cytokine and chemokine levels directly in tissue extracts. Plates were read on a Luminex 200 instrument (EMD Millipore), and cytokine and chemokine concentrations (pg/ml) calculated by the BioPlex manager software (Bio-Rad, Hercules, CA) using standard curves. Results presented reflect the mean ± standard error of the mean (SEM) of cytokine or chemokine quantities per milliliter of plasma or per milligram of total extracted tissue protein from five animals at each time point. The lower limit of detection for these assays was 1.6 pg/ml.

Brains and spinal cords were homogenized in a tissue lysis buffer (10 mM Tris, 1 % sodium dodecyl sulfate (SDS), 1 mM sodium orthovandate, pH 7.6) supplemented with a commercial protease inhibitor cocktail (Roche Life Science). Lysates were centrifuged to remove undigested tissue debris and the total protein concentration of each supernatant was determined using the Pierce Coomassie Protein Assay Reagent (Thermo Fisher Scientific). Samples were boiled in 4× protein sample buffer and 20 μg/well run on SDS-polyacrylamide gels. Proteins were transferred to PVDF membranes and blocked overnight at 4 °C in a 5 % non-fat skim milk solution in Tris-buffered saline (TBS) containing 0.5 % Tween 20. Membranes were then incubated with one of the following primary antibodies: polyclonal goat anti-AT1R (1:500, Santa Cruz Biotechnology, Santa Cruz, CA), polyclonal rabbit anti-gp91 (1:1000, Abcam, Cambridge, MA), or polyclonal rabbit anti-p47 (1:500, EMD Millipore) for 1 h at room temperature. All primary antibodies were raised against synthetic peptides derived from the corresponding human gene sequence, and both cross-reactivity and specificity with the corresponding mouse protein previously established [33‐35]. Following five washes, membranes were then incubated with either rabbit anti-goat horseradish peroxidase (HRP)-conjugated secondary antibody or goat anti-rabbit HRP-conjugated secondary antibody (both used at 1:10,000, EMD Millipore) for 1 h at room temperature. Membranes were washed five times again, and the HRP signal detected using ECL Western blotting detection reagent (GE Healthcare Life Sciences, Pittsburgh, PA) on X-ray film. Membranes were then stripped using Western blot stripping buffer (Thermo Fisher Scientific) and relabeled with β-actin loading control antibody (1:5000, Thermo Fisher Scientific) using the same steps described above. Once all the β-actin signals were obtained, all band densities were quantified using the ImageJ software package (NIH, Bethesda, MD). The band density for each protein was first normalized to the β-actin signal detected in the same lane, and the mean of the uninfected samples were then set to an arbitrary expression level of 1.0. Relative protein expression across the full course of acute NSV infection was determined compared to uninfected controls, and relative expression in five samples at each disease stage was analyzed for statistical significance.

Unless otherwise specified, all tissues used for histological studies were post-fixed for 6 h in 4 % PFA in PBS, cryopreserved overnight in 30 % sucrose in PBS, and snap-frozen in CRYO-OCT Compound (Thermo Fisher Scientific). Eight micron frozen sections were cut, collected on SuperFrost Plus slides (Thermo Fisher Scientific) and stored at −20 °C until staining.

Six days after NSV challenge, ten anesthetized C57BL/6 mice underwent transcardiac perfusion with chilled PBS. A parallel cohort of five mice were infected and also treated with 100 mg/kg/day of telmisartan for 6 days followed by transcardiac perfusion. Brains and spinal cords were collected from each mouse and homogenized into small fragments. Tissue was suspended in Hank’s balanced salt solution (HBSS) containing 28 U/ml DNase (Sigma-Aldrich) for 30 min at 37 °C. Infiltrating mononuclear cells (MNC) were isolated from tissue digests of five telmisartan-treated and five untreated mice by centrifugation over 30 %/70 % Percoll gradients (GE Healthcare Life Sciences), counted, washed extensively with PBS containing 2 % fetal bovine serum (FBS), and stained with fluorescently conjugated anti-CD45, anti-CD11b, anti-Ly6C, anti-Ly6G, and anti-CD11c antibodies for myeloid cell phenotyping, as well as anti-CD40 and anti-CD86 to assess myeloid cell activation (all from eBiosciences). The following staining patterns defined each myeloid cell subset: CD45^low/CD11b + (microglia), CD45^high/CD11b+/CD11c-/Ly6C+/Ly6G- (macrophages), CD45^high/CD11b+/CD11c-/Ly6C-/Ly6G- (monocytes), CD45^high/CD11b+/CD11c-/Ly6C-/Ly6G+ (neutrophils), and CD45^high/CD11b+/CD11c + (dendritic cells). In the remaining five NSV-infected mice, CNS mononuclear cells were pooled and the endogenous and recruited myeloid cell subsets were physically separated from one another into CD45^low/CD11b + (microglia) and CD45^high/CD11b + (all infiltrating myeloid cells) populations using a BD FACSAria high-speed cell sorter (BD Biosciences, San Jose, CA). Cells were stored in PrepProtect RNA stabilization solution (Miltenyi Biotec, Auburn, CA) at −20 °C until RNA isolation could be performed.

Flow sorted cell subsets were thawed, pelleted, and carefully removed from the PrepProtect solution. Total RNA was isolated from each cell population and cDNA generated using a high-capacity cDNA reverse transcription kit according to the manufacturer’s instructions (Thermo Fisher Scientific). Quantitative PCR (qPCR) was performed to measure agtr1a and β-actin mRNA transcripts using the MyiQ Single Color Real-Time PCR Detection System and a Bio-Rad iQ5 cycler (Bio-Rad, Hercules, CA). TaqMan® gene expression assays for both agtr1a and β-actin were obtained from Thermo Fisher Scientific. Levels of agtr1a transcripts were calculated relative to β-actin using the following formula: 2^[Ct (β-actin) − Ct (target gene)] × 1000, where Ct is the threshold cycle at which the fluorescent signal became significantly higher than background. Results presented reflect relative agtr1a mRNA expression in each cell population done in three experimental replicates.

To measure the amount of infectious virus present in CNS tissues, animals were perfused extensively with chilled PBS and brains and spinal cords were extracted, weighed, snap-frozen on dry ice, and stored at −80 °C until virus titrations assays were performed. At that time, 20 % (w/v) homogenates of each sample were prepared in Dulbecco’s modified Eagle’s medium (DMEM, Sigma-Aldrich), and serial tenfold dilutions of each homogenate were assayed for plaque formation on monolayers of BHK-21 cells. Results presented are the mean ± SEM of the log₁₀ of viral pfu per gram of tissue derived from a minimum of three animals at each time point.

The capacity of CNS tissues to generate ROS was measured directly ex vivo using the well-established cytochrome c reduction assay on fresh tissue membrane extracts, as described [37, 38]. Briefly, animals were perfused with chilled PBS, and brains and spinal cords were extracted, weighed, and homogenized to a final concentration of 1 mg/ml total tissue protein in DMEM. One hundred-eighty microliters of each homogenate was then added to 96-well, flat-bottom plates in duplicate. Purified superoxide dismutase (SOD, Sigma-Aldrich) was then added to half the wells at a final concentration of 200 U/ml, while DMEM was added to the remaining wells as a control. All wells then received 500 μmol/l purified cytochrome c (Sigma-Aldrich) and 100 μmol/l purified NADPH (Sigma-Aldrich) as a specific electron donor, and plates were incubated at 37 °C for 30 min. Immediately thereafter, the absorbance of each well was read at 540, 550, and 560 nm using a PowerWave HT spectrophotometer (BioTek, Winooski, VT). Tissue O₂ ⁻ production by each sample was calculated by first subtracting the average optical density (OD) readings at 540 and 560 nm from the average OD reading at 550 nm (ΔOD). The difference in ΔOD value in the absence and presence of SOD was then divided by the extinction coefficient of cytochrome c using the following formula: (ΔOD without SOD – ΔOD with SOD)/21.1. Results are presented as nmol O₂ ⁻ produced/mg tissue protein.

The Prism 5.0 software package (GraphPad Software, La Jolla, CA) was used for all statistical analyses. Student’s t test was applied when comparing a single group under two experimental conditions, a one-way analysis of variance (ANOVA) with a post hoc Bonferroni’s multiple comparison test was used to investigate the significance of a single group’s change over time, while a two-way ANOVA with a post hoc Bonferroni’s multiple comparison test was utilized to compare experimental findings between two groups over time. Differences in outcome among individual cohorts of infected mice were determined using a log-rank (Mantel-Cox) test. In all cases, differences at a p < 0.05 level were considered significant.

To explore whether Ang II-AT1R signaling influences disease outcome in mice with NSV encephalomyelitis, we measured Ang II concentrations directly in CNS tissues over time. These assays showed that Ang II levels increased in both the brains and spinal cords of mice over the first few days following NSV infection (Fig. 1a, b). Furthermore, local Ang II production was suppressed via systemic administration of captopril (Fig. 1c), an angiotensin converting enzyme (ACE) inhibitor that prevents Ang II cleavage from its precursor, angiotensinogen (AGT). Importantly, mice treated with this drug showed enhanced survival compared to animals receiving a vehicle control (Fig. 1d). These data suggest that local Ang II production could be an important early event that triggers CNS damage and leads to fatal disease in mice with NSV infection.

We next characterized CNS AT1R expression, demonstrating that tissue receptor levels increased steadily in both the brains and spinal cords of NSV-infected animals (Fig. 2a, b). A representative Western blot is shown (Additional file 1: Figure S1). Using CX3CR1^GFP/+ mice with constitutively green microglia [39], AT1R expression was identified by histochemical staining on a subset of these cells in the brains of naïve animals (Fig. 2c). Because CX3CR1-positive cells could also represent GFP-positive monocytes infiltrating from the periphery [39], these two cell populations were separated from the brains of NSV-infected animals by flow sorting for ex vivo analysis. Using this approach, anti-AT1R antibodies proved unreliable to detect surface expression by flow cytometry, but agtr1a transcripts were found at equivalent levels in CD45^low/CD11b + microglia and CD45^high/CD11b + infiltrating myeloid cells at peak disease (Fig. 2d). No AT1R expression was detected by histochemical staining on CX3CR1-negative cells in either control or NSV-infected tissues (Fig. 2c and data not shown). Finally, AT1R KO mice challenged with NSV demonstrated improved survival compared to wild-type (WT) controls (Fig. 2e). These results confirm that AT1R signaling contributes to NSV pathogenesis. Furthermore, local pharmacological blockade of AT1R in the CNS, if achieved, would act mainly on the endogenous and recruited myeloid cell populations in this disease setting.

We next investigated the effects of telmisartan on the course of NSV encephalomyelitis, a drug known to effectively penetrate the BBB and cause sustained AT1R blockade without affecting AT2R signaling [28]. Cohorts of mice were treated twice daily with the drug beginning 12 h after viral challenge. Escalating doses up to 100 mg/kg/day were used, below those previously shown to lower systemic arterial blood pressure in rodents [40]. Analyses of disease outcomes in these cohorts showed that drug treatment attenuated the development of severe hind limb paralysis and prevented death in a dose-dependent manner compared to animals given a vehicle control (Fig. 3). These data further support a role for AT1R signaling in NSV pathogenesis.

Virus titrations were performed on tissues derived from vehicle- and telmisartan-treated animals to investigate whether the drug had any effect on CNS virus replication or spread. These assays showed that virus replication was completely unaltered in both the brains and spinal cords of NSV-infected mice treated with the highest protective dose of telmisartan examined (Fig. 4). Furthermore, infection with NSV-GFP demonstrated that drug treatment did not change virus tropism for neurons or reduce the number of infected cells in heavily targeted CNS regions such as the hippocampus or the lumbar spinal cord (data not shown). These results show that telmisartan produces benefits in mice with NSV encephalomyelitis through a mechanism other than one involving direct antiviral activity.

To investigate whether telmisartan exerted any anti-inflammatory effects in NSV-infected animals, we measured total MNC infiltration into the CNS, the recruitment of individual myeloid cell subsets into the brain, the cellular activation status of these CNS myeloid cells, and circulating and CNS levels of myeloid cell-related immune factors after 6 days of treatment. Ex vivo analyses of CNS isolates from infected animals showed that total tissue MNC were not reduced in telmisartan- compared to vehicle-treated mice (Fig. 5a). The proportions of five distinct myeloid cell subsets were equivalent between the two groups (Fig. 5b), and telmisartan had no effect on expression levels of the activation markers, CD40 and CD86, on any of these cell types (Fig. 5c, d). When a panel of myeloid factors was measured in both plasma and brain tissue from NSV-infected animals, telmisartan had minimal effects on the production of any of these mediators in either tissue compartment (Fig. 6). We conclude that telmisartan does not exert a global anti-inflammatory effect in either the periphery or the CNS during NSV infection.

To determine the broader relevance of telmisartan as a treatment for alphavirus encephalitis, separate cohorts of mice were challenged with the known human pathogen, WEEV. Much like what was observed in NSV-infected animals, systemic administration of telmisartan at a similar dose caused reduced disease severity in WEEV-infected mice compared to vehicle-treated controls (Fig. 7a). Half of the treated animals survived infection compared to what was uniformly lethal without the drug (Fig. 7b). These data confirm that telmisartan can protect a significant proportion of mice using a lethal challenge model of a clinically relevant alphavirus. As ARB drugs are in widespread human use, and as antiviral drugs effective against the alphaviruses still remain a long way from human application [41, 42], blockade of this signaling pathway might be considered for testing in both equine and human cases of acute alphavirus encephalitis.

Given the known capacity of ARBs to suppress Nox-mediated ROS production in both vascular smooth muscle and renal podocytes [22, 23, 29], we next sought to determine whether telmisartan suppresses tissue ROS production and oxidative neuronal injury as a potential mechanism through which it protected NSV-infected animals. To investigate Nox activity in the CNS during disease, tissue subunit expression and tissue enzyme activity were measured ex vivo. Via Western blot, multiple Nox subunits were induced in both the brains and spinal cords of NSV-infected animals over time (Fig. 8a, b). A representative blot is shown (Additional file 2: Figure S2). Using an established method to quantify superoxide production in tissue extracts via ex vivo reduction of cytochrome c and exogenous NADPH as an electron donor [37, 38], we found clear biochemical evidence that enzymes capable of generating ROS were induced in both the brains and spinal cords of animals 72 h following NSV challenge (Fig. 8c, d). Furthermore, this enzyme activity was potently suppressed in tissues derived from infected mice treated with telmisartan, and it was ablated in samples collected from infected gp91/p47 DKO animals not treated with the drug (Fig. 8c, d). On the other hand, telmisartan had no effect on the induction of Nox subunits in the CNS by Western blot (data not shown). We conclude that Nox is the principal enzymatic source of tissue ROS in this disease setting, and that telmisartan suppresses Nox activity in the CNS without inhibiting expression of the enzyme complex itself.

To investigate the spatial expression of ROS directly in CNS tissues, animals were infused with DHE immediately prior to sacrifice. This cell-permeable dye gets oxidized to the fluorescent compound, ethidium, by intracellular ROS and trapped within cells [43]. We found that DHE labeling was increased in the spinal cord 96 h after viral challenge when compared to labeled tissue derived from uninfected animals. This ROS signal remained largely confined to gray matter regions where infected cells were also found (Fig. 9a, b). Systemic treatment with telmisartan at protective doses largely abolished this DHE staining (Fig. 9c). Similar patterns of DHE labeling suppressed by telmisartan treatment were also observed in the brain, and no DHE labeling above background was found in tissues from NSV-infected but otherwise untreated gp91/p47 DKO mice (data not shown). Since drug treatment did not inhibit virus replication in either the brain or spinal cord (Fig. 4), we conclude that ROS do not exert any antiviral effect during alphavirus encephalitis.

ROS generated as part of an inflammatory response can damage cellular lipids, proteins, and nucleic acids. A lipid peroxidation product detectable in cell membranes, 4-HNE, is both a validated marker for, as well as a direct mediator of, oxidative injury to neurons [44, 45]. We found that CNS tissues derived from NSV-infected animals showed prominent 4-HNE immunoreactivity in spinal gray matter within 72 h of viral challenge compared to uninfected controls (Fig. 10a, b). At higher magnification, heavy labeling of neuronal cell bodies, with moderate staining of gray matter neuropil, was observed (Fig. 10c). Treatment with telmisartan visibly reduced 4-HNE staining observed in infected tissue samples (Fig. 10d). Similar patterns of 4-HNE immunoreactivity were also identified in the hippocampus (data not shown).

Based on this distribution of staining, the effects of telmisartan on the survival of hippocampal and lumbar ventral spinal motor neurons were quantified following NSV infection. Using Fluoro-Jade C labeling, an established histochemical marker of irreversible neuronal injury [46], damaged hippocampal neurons were easily identified on day 7 post-infection compared to naïve controls (Fig. 11a), and quantification of these cells showed that the drug exerted a measurable neuroprotective effect compared to vehicle-treated controls (Fig. 11b). Likewise, silver staining of sections through the lumbar spinal columns of these mice allowed the identification of lumbar ventral nerve roots carrying axons from lumbar ventral motor neurons (Fig. 11c); quantification of these axons showed that survival of these cells was also enhanced by telmisartan (Fig. 11d). Together, these findings demonstrate that a systemically administered ARB can penetrate the CNS to inhibit Nox activity and local ROS production, thereby limiting oxidative neuronal damage during NSV encephalomyelitis.

Although multiple enzymes could theoretically be generating ROS in the CNS following NSV infection, Nox appears to be their primary source given the results of both our biochemical assays (Figs. 8c, d) as well as the absence of DHE staining in infected gp91/p47 DKO mice (data not shown). When gp91/p47 DKO mice were challenged with NSV, these hosts survived infection significantly better than WT controls (Fig. 12a). Furthermore, when telmisartan was given to cohorts of NSV-infected gp91/p47 DKO hosts, the survival benefit conferred by the drug was abolished compared to DKO animals treated with a vehicle control (Fig. 12b). These data confirm that Nox activity is pathogenic during NSV infection, and that Nox is the main therapeutic target of telmisartan in this alphavirus encephalitis model.