Of Murines and Humans: Modeling Persistent Powassan Disease in C57BL/6 Mice

ABSTRACT Powassan infection is caused by two closely related, tick-transmitted viruses of the genus Flavivirus (family Flaviviridae): Powassan virus lineage I (POWV) and lineage II (known as deer tick virus [DTV]). Infection is typically asymptomatic or mild but can progress to neuroinvasive disease. Approximately 10% of neuroinvasive cases are fatal, and half of the survivors experience long-term neurological sequelae. Understanding how these viruses cause long-term symptoms as well as the possible role of viral persistence is important for developing therapies. We intraperitoneally inoculated 6-week-old C57BL/6 mice (50% female) with 103 focus-forming units (FFU) DTV and assayed for infectious virus, viral RNA, and inflammation during acute infection and 21, 56, and 84 days postinfection (dpi). Although most mice (86%) were viremic 3 dpi, only 21% of the mice were symptomatic and 83% recovered. Infectious virus was detected only in the brains of mice sampled during the acute infection. Viral RNA was detected in the brain until 84 dpi, but the magnitude decreased over time. Meningitis and encephalitis were visible in acute mice and from mice sampled at 21 dpi. Inflammation was observed until 56 dpi in the brain and 84 dpi in the spinal cord, albeit at low levels. These results suggest that the long-term neurological symptoms associated with Powassan disease are likely caused by lingering viral RNA and chronic inflammation in the central nervous system rather than by a persistent, active viral infection. The C57BL/6 model of persistent Powassan mimics illness in humans and can be used to study the mechanisms of chronic disease.

DTV-infected mice, but viremia did not differ between symptomatic mice euthanized during the acute phase and those that survived the acute phase ( Fig. 1A and B). Based on previous mouse studies, about 60% mortality was expected during the first 14 dpi (20,21). In the current study, mice infected with DTV experienced 17% mortality (n = 6) ( Fig. 2A). Symptomatic mice were euthanized between 8 and 10 dpi, at an average of 8.3 (standard error [SE], 0.3) dpi. The most common signs of illness were weight loss and a hunched back (Fig. 2B). Two mice developed weight loss and hunched backs but recovered before euthanasia criteria were met. Difficulty ambulating, from either paralysis of a single rear leg or an unusual gait, was noted in three mice. Lethargy or moribundity was documented for 67% of the mice euthanized during the acute phase, and all acutely ill mice were dehydrated. Here, we refer to mice euthanized during the first 2 weeks of infection as acute mice.
As seen in Fig. 2C, the average weight of acute mice decreased by 10% at 8 dpi compared to weight on the day of infection (day 0), whereas the mice that survived the infection steadily gained weight. On average, the weight of acute mice at euthanasia was 16.9% lower than the initial average weight. The average weight at euthanasia of mice sampled 21, 56, and 84 dpi was greater than their average starting weight. The average weight of acute mice was significantly less than that of the mock-infected mice 7 dpi (see Table S1 in the supplemental material). The average weight of the mice that survived the infection was significantly less than that of the mock-infected mice 21 dpi; however, all the mice had weights equal to or more than their starting weight ( Fig. 2C; Table S1). At 56 and 84 dpi, the average percent weight change of the infected mice did not differ from that of the mock-infected mice (Fig. 2C). In summary, early viremia was present in the majority of DTV-infected mice, but only a minority of the mice experienced symptoms or weight loss that most often resulted in euthanasia.
Infectious DTV was detected only in the brains of acute mice. At the time of euthanasia, only acute mice had infectious virus present in the brain and brain stem ( Fig. 3A and B). The average viral titer in the brain and brain stem of acute mice was 3.5 log 10 FFU/g (SE, 0.8) and 1.8 log 10 FFU/g (SE, 0.7), respectively. Infectious virus was not detected in any other tissue regardless of euthanasia group (limit of detection, 0.2 FFU/g).
Genomic DTV RNA was detected in the brain and spinal cord of DTV-infected mice until 21 dpi. We also performed strand-specific in situ hybridization (ISH) to look for DTV RNA in the infected mice. In acute mice and mice sampled 21 dpi, we assayed for positive-sense (genomic) DTV RNA visualized in the brain, spinal cord, vertebral bone marrow, inguinal lymph node, mammary gland, liver, pancreas, spleen, and kidney. We looked for replicative-form RNA by performing ISH for the complementary strand of DTV RNA in the brain, spinal cord, and vertebral bone marrow. For mice sampled at 56 and 84 dpi, ISH was performed to detect genomic and replicative form DTV RNA only in the brain, spinal cord, and vertebral bone marrow.
In the cerebellum, genomic DTV RNA was detected in all acute mice as well as 60% of mice sampled 21 dpi (Fig. 4A and B). The percentage of tissue affected was greater and the distribution of the signal was more widespread in the acute mice than in mice sampled 21 dpi; however, the difference was not significant using Dunn's multiplecomparison test (Fig. 4A). The mean ISH score for acute mice was 3 (SE, 0.4), and that for mice sampled 21 dpi was 0.6 (SE, 0.2), indicating that in acute-mouse brain samples, 26 to 50% of the sample had detectable genomic DTV RNA multifocally distributed, in  (50) contrast to the brain samples from mice sampled 21 dpi, where less than 10% of the sample had detectable single foci of genomic DTV RNA. Widespread genomic DTV was also detected in the cortex, dentate gyrus and hippocampus, and thalamus of acute mice (Fig. 5). Only two mice sampled 21 dpi had single cellular foci of genomic RNA in the cortex (n = 1) and dentate gyrus (n = 1). Replicative DTV RNA was detected in 2 acute mice. Low levels of genomic DTV RNA were detected above the threshold 2.3 log 10 copies/g in the spinal cord of 83% acute mice and 50% of mice sampled 21 dpi (Fig. 4C and D). As for the brain, the difference between ISH score in acute mice and mice sampled 21 dpi was not significant using Dunn's multiple-comparison test (Fig. 4C). The mean ISH score for acute mice was 1.5 (SE, 0.3) and 0.5 (SE, 0.2) for mice sampled 21 dpi. These scores indicate that less than 25% of the acute mouse samples were affected with multifocal signal while less than 10% of the samples from mice sampled 21 dpi were affected with a singular focus of signal.
We quantified DTV RNA in the brain via quantitative reverse transcription-PCR (qRT-PCR). Viral RNA levels decreased with sampling time (Fig. 6). Acute mice had 3 to 6 log more RNA in the brain than mice at all three other time points. While the overall levels of RNA were low after the acute phase, mice sampled at 21 dpi also had significantly more RNA (3.9 log 10 RNA copies/g) in the brain than mice sampled at 56 and 84 dpi (2.1 and 1. log 10 RNA copies/g, respectively), but RNA levels in the latter two were not different from each other.
One liver cell of an acute mouse was positive for genomic DTV RNA (Fig. S1). No infected mouse had detectable genomic DTV RNA in the inguinal lymph node, mammary gland, pancreas, spleen, or kidney.
In summary, DTV RNA was present in the brains and spinal cords of DTV-infected mice after the acute phase. In the brain, RNA was detected at low levels until 84 dpi, and the ISH results indicate that viral RNA was located primarily in the cerebellum. RNA was not quantified in the spinal cord, but RNA was detected via ISH at 21 dpi.
Low-level inflammation was detected until 56 dpi in the brain and 84 dpi in the spinal cords of DTV-infected mice. The brains and spinal cords of acute mice demonstrated meningitis, encephalitis, myelitis, gliosis, and neuronal necrosis. Inflammatory cells were predominantly lymphocytes and macrophages but also included neutrophils, which expanded the meninges, occupied Virchow-Robin spaces, and infiltrated the adjacent white and gray matter. At 21 dpi, lesions consisted predominantly of meningitis and glial nodules with rare neuronal satellitosis. Extremely rare foci of lymphocytes were identified in the meninges, perivascular spaces, and white matter in four of 10 spinal cord sections from the 56-dpi mice and two of 10 spinal cord sections from the 84-dpi mice, whereas none were identified in the mock-infected sections (Fig. S2A). Two of the 10 56-dpi mice had similar findings of lymphocytes in the meninges of the  Table S1. Gray triangles, mock-infected mice; purple diamonds, mice sampled during the acute phase; pink circles, mice that survived the acute phase.
brain, but none of the 10 84-dpi mice or mock-infected mice had any abnormal findings (Fig. S2B). Severity of meningitis and encephalitis in the brains of acute mice and mice sampled 21 dpi did not differ when analyzed with Dunn's multiple-comparison test but was greater than that in mice sampled at later time points ( Fig. 7A and B) (P , 0.05). Gliosis was noted in 50% of the mice sampled at 21 dpi. The severity of meningomyelitis in the spinal cord was greatest in acute mice and mice sampled at 21 dpi, but meningomyelitis was also detected at low levels in mice sampled 56 and 84 dpi (Fig. 7C). However, the mean ISH severity scores were significantly greater than those for the mock-infected mice only for acute mice and mice sampled on 21 dpi, not those sampled on 56 or 84 dpi (Dunn's multiple-comparison test P , 0.05).
Neuronal necrosis, characterized by cytoplasmic hypereosinophilia, nuclear pyknosis, and cellular debris, was present in the cerebral cortex and dentate gyrus 6 dpi (Fig. S3). Also present at 6 dpi was neuronal degeneration, characterized by enlarged or swollen cell bodies and foamy-appearing cytoplasm, a generalized reduction or loss of neurons, and the aforementioned inflammation. Similar lesions were not documented in the mock-infected mice or mice sampled at 56 or 84 dpi (Fig. S3).
No pathological findings were detected in the inguinal lymph node, mammary gland, liver, pancreas, spleen, kidney, or vertebral bone marrow of acute mice or mice sampled on day 21 or later. To summarize, severe CNS inflammation was present in the symptomatic mice euthanized during the acute phase, while mice that survived the acute phase had low levels of inflammation until 56 dpi in the brain and 84 dpi in the spinal cord.
Viremia at 3 dpi did not differ for mice that died during the acute infection, recovered, or experienced persistent disease. Mice were classified as "persistent" if they survived the acute infection and had detectable viral RNA, infectious virus, and/or  inflammation, "recovered" if they survived the acute infection but did not have detectable viral RNA, infectious virus, or inflammation, or "acute" if they were euthanized during the acute infection. While the mean viral titer in persistent mice was lowest (1.8 log 10 FFU/mL; SE, 0.2) compared to acute and recovered mice, there was not a significant difference between viremia levels 3 dpi by persistence status (Fig. 8A). Persistence status was not influenced by gender (Fig. S4).
Low levels of viral RNA in the brain were present in mice that experienced persistent disease. While early viremia did not differ by persistence status, level of viral RNA at the time of euthanasia did differ (Fig. 8B). Acute mice experienced the highest level of viral RNA in the brain (7.0 log 10 RNA copies per g brain; SE, 0.4), while viral RNA in recovered mice was below the level of detection. Low levels of viral RNA, above the limit of detection of 2.3 log 10 copies/g, were detected in almost all persistently infected mice (average, 3.0 log 10 RNA copies per g brain; SE, 0.3).

DISCUSSION
While acute infection with POWV or DTV is rare, the incidence is increasing, and it is likely that the incidence of chronic disease will increase as well (5). For patients suffering with chronic Powassan disease, prolonged neurological illness can have a major impact on their quality of life (1,3). Elucidating the etiology and pathogenesis of chronic Powassan disease will better prepare medical professionals and public health officials to diagnose, treat, and prevent this serious disorder. In the current study, we developed an animal model with C57BL/6 mice to study chronic Powassan disease. We predicted that mice that survived the acute infection would display signs of persistent disease, possibly with detectable infectious virus, viral RNA, and/or inflammation. Our data suggest that persistent viral RNA and virus-induced chronic inflammation may play roles in the long-term neurological disease but infectious virus does not. This finding is in agreement with other studies of chronic disease after acute viral infection (23,24). By detecting viral RNA with ISH staining and qRT-PCR, we found that viral RNA after the acute phase is limited in quantity and spread across central nervous system (CNS) tissues, indicating a restricted persistent infection or residual RNA. We hypothesized that low initial viremia would contribute to the establishment of persistent disease. However, we did not find a difference in viremia 3 dpi, likely refuting this hypothesis. Further research on host, viral, and vector factors is needed to determine the mechanisms of persistent Powassan disease.
Chronic signs of Powassan disease in humans can include hemiplegia, headache, wasting, memory problems, muscle weakness, ataxia, tremors, and respiratory failure and potentially results in death (1, 3). We were unable to evaluate mice with severe acute symptoms for progression to chronic disease, as they were euthanized during the acute phase. The surviving mice in the current study did not display obvious gross neurological signs after the acute infection. However, we found inflammation of the meninges in mice sampled 21 and 56 dpi, while meningomyelitis was observed in mice sampled 21, 56, and 84 dpi. Additionally, gliosis was documented in mice sampled 21 dpi. These findings Mouse Model of Persistent Powassan Disease mBio resemble those observed in a human patient 39 days after symptom onset (10). In that case, although no viral particles were detected via electron microscopy, chronic inflammation of the meninges and perivascular spaces was documented at autopsy, characterized by focal infiltration of the medial temporal lobes, ventral midbrain, and basal ganglia. Gliosis of the cerebral gray matter was also noted. Our ability to detect infectious virus and viral RNA in tissues other than the CNS was limited due to a low level of systemic infection. This was particularly true for the ISH detection of replicative RNA, which is typically present a ratio of 1:100 to genomic RNA (25)(26)(27). In a single acute mouse, we identified one cell in the liver, likely a Kupffer cell, that was positive for genomic DTV RNA; perhaps this reflected phagocytosis of debris from DTV-infected cells. Hermance et al. (22) detected low levels of DTV viral RNA via qRT-PCR in the popliteal lymph node and kidneys of DTV-infected BALB/c mice that survived to 24 dpi (,10 2.26 and 10 1.87 FFU equivalents/mg RNA, respectively) of DTV RNA after footpad inoculation but did not detect RNA in the spleen or testes. Future studies with different inoculation dosages and routes may enhance the ability to detect infectious virus and/or RNA in CNS and non-CNS tissues using the C57BL/6 mouse model, particularly at later time points postinfection.
We used the i.p. route of infection for this study. Although numerous studies of tick-borne and mosquito-borne flavivirus have used this route (18,20,28,29), it does not replicate the natural route of transmission by vector bite. When a tick takes a blood meal from a host, numerous tick proteins are also injected into the host along with the virus (30-32). These salivary gland proteins help the tick evade the host immune response to ensure successful feeding, but they also benefit the transmission and pathogenesis of viral and bacterial pathogens (30)(31)(32). Specifically, POWV transmission and pathogenesis in BALB/c mice after footpad inoculation with 10 3 FFU POWV and homogenized I. scapularis salivary glands was enhanced (32). Future studies using the C57BL/6 mouse model of persistent Powassan disease will incorporate alternate routes of inoculation, the inclusion of salivary gland extracts, and/or the use of live infected ticks to model the natural route of transmission and evaluate its impacts on the development of chronic disease.
In our previous work with POWV in C57BL/6 mice (20), we observed 40% survival. In the current DTV study, overall survival was higher than expected, at 83% (20,21). This level of survival more closely reflects that of humans with symptomatic disease (1,5) and enabled us to study more mice during the postacute phase of infection. The variation in survival rate observed between the two studies may relate to the limited number of mice used in the previous study, as five animals may have been too few to provide an accurate estimate of survival. Alternatively, a direct comparison between the two Powassan virus lineages using the current study design may yield interesting differences with respect to survival and chronic disease. According to VanBlargan et al. (21), infection of C57BL/6 mice with POWV was more lethal than infection with DTV (93% versus 57% mortality). However, the relative lethality of the two virus lineages in chronic disease has not been examined. In addition to between-lineage comparisons, the model should be replicated with other strains of DTV to determine if within-lineage differences exist, which would impact the utility of the model. POWV and DTV are closely related to tick-borne encephalitis virus (TBEV) the etiologic agent of tick-borne encephalitis (TBE) (13). During the acute infection, symptoms of TBE range from headache and fever to cognitive impairment and paralysis, with about 50% of patients reporting encephalitic symptoms (33). The infection can also progress to a chronic condition and result in a number of CNS conditions, such as headache, memory loss, and Kozhevnikov epilepsy (13,33). Like persistent Powassan disease, TBE progresses to a chronic condition in 31% of patients, with a case fatality rate of 1.2% on average (33)(34)(35)(36)(37)(38). TBEV causes 300 times more symptomatic cases annually than POWV/DTV (5,39) and is therefore more heavily studied. Animal models for acute (40,41) and chronic (42-44) TBE have been developed, but until the current study, animal models of only acute Powassan disease had been developed (20)(21)(22). Pogodina et al. (42,43) found infectious TBEV in rhesus macaques for at least 738 dpi. Morozova et al. (44) experimentally infected wild-caught Northern red-backed voles (Myodes rutilus) and striped field mice (Apodemus agrarius) with suspensions of TBEV infected ticks and found TBEV RNA, presumably in the serum, until at least 120 dpi in 50% and 35% of animals, respectively. Infectious virus was detected during the first 2 weeks of infection, and TBEV protein envelope antigen was found in less than 30% of animals between 16 and 120 dpi. Like our results, these data suggest that TBEV RNA can persist long after the initial infectious period. However, the viruses need to be evaluated using the same animal model to make direct comparisons between TBEV, POWV, and DTV and generate hypotheses about chronic disease in humans.
A reliable animal model of chronic Powassan disease will permit research on the etiology, pathogenesis, and treatment of chronic disease. Multiple mechanisms of flavivirus persistence have been proposed, such as defective interfering particles (45,46), protection from apoptosis by NS4A induced phosphatidylinositol 3-kinase dependent autophagy (47), and evasion of the host antiviral response by NS5 inhibition of interferonmediated JAK-STAT signaling pathway (48) (reviewed by Mlera et al. [49]). Additionally, HEK293T cells persistently infected with Langat virus (LGTV), another tick-borne flavivirus, demonstrated upregulation of the prosurvival oncogenes encoding AKT2 and ERBB2 and downregulation of the proapoptotic genes encoding Bad and beta interferon 1 compared to acutely infected cells (50). Transcriptomic analyses using animal models such as C57BL/ 6 mice will further our understanding of the mechanisms of viral RNA persistence.
Beyond elucidating the molecular mechanisms behind chronic Powassan disease, it will be interesting to examine the full impact that Powassan infection has on behavioral and memory outcomes in animal models. Such studies may shed light on longterm human symptoms and treatments. Mice that survived infection with West Nile virus, a mosquito-borne flavivirus, exhibited impairments in spatial learning when evaluated 46 to 50 dpi (51). Cornelius et al. (28) demonstrated that infection of C57BL/6 mice with LGTV was asymptomatic; however, anxiety, memory, and hippocampal neuron morphology were affected until at least 28 dpi, when the experiment concluded. This study indicates that asymptomatic infection with a tick-borne flavivirus can impact animal behavior. This finding is particularly worrisome, as between 50% to 98% of human flavivirus cases are asymptomatic (13,(52)(53)(54)(55)(56). Further research is needed to study the full impacts of clinical and subclinical neurotropic flavivirus infection on long-term health and well-being. In the context of Powassan disease, future studies utilizing the C57BL/6 model should be conducted to study the mechanistic underpinnings of chronic Powassan disease and its impact on behavior, memory, learning, and motor skills. Ultimately, understanding the causes and effects of chronic Powassan disease at an organismal level will support disease mitigation efforts.

MATERIALS AND METHODS
Ethics statement. All procedures with DTV in cell culture or in vivo were conducted in biosafety level 3 (BSL-3) or animal BSL-3 (ABSL-3) facilities at Rocky Mountain Laboratories (RML). Operating protocols were approved by the RML Institutional Biosafety Committee. The animal study was conducted under protocol 2020-074-E, which had been approved by the RML Animal Care and Use Committee. Mice were housed in ventilated HEPA-filtered cage systems with no more than 3 mice per cage. Food and water were provided ad libitum. The animal holding room was maintained at 72°F and 50% humidity with 12h/12-h light-dark cycles. All animal procedures were conducted by trained and certified personnel in compliance with the Association for the Assessment and Accreditation of Laboratory Animal Care guidelines.
Cells and virus. Vero cells (African green monkey kidney epithelial cells) were obtained from ATCC (Bethesda, MD) and maintained at 37°C with 5% CO 2 in Dulbecco's minimum essential medium (DMEM; Gibco, Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, Thermo Fisher Scientific) and 0.05 mg/mL gentamicin (Gibco, Thermo Fisher Scientific). DTV (Powassan virus lineage II, Spooner strain) was provided by Gregory Ebel at Colorado State University. A viral stock was generated by infecting Vero cells at a multiplicity of infection (MOI) of 0.005 and titrated as previously described (57,58).
Experimental DTV infection in mice. Six-week-old C57BL/6J mice (Jackson Laboratory) were infected via i.p. inoculation using 25-gauge needles with 70 mL of 10 3 FFU of DTV diluted in DMEM (n = 36, 44% female) or mock infected with DMEM (n = 6, 50% female). At 3 dpi, 50 to 100 mL of blood was collected via cheek bleed with a lancet. Blood was centrifuged in microtubes with serum gel (Sarstedt) for 5 min at 3,380 Â g to collect serum, which was then stored at 280°C. Mice were scored daily for the first 2 weeks and then once a week for signs of acute illness, and animals exhibiting 20% weight loss, seizures, and/or limb paralysis were humanely euthanized in accordance with the American Veterinary Medical Association's guidelines for the euthanasia of animals. Terminal blood was collected along with cerebrum, cerebellum, brain stem, spinal cord, liver, spleen, kidney, mammary tissue, cervical lymph node, and reproductive organs (ovary or testes). Organs and tissues were divided into three samples. One was placed into a tissue cassette and fixed in neutral buffered formalin for histology and ISH. The remaining two samples were stored in separate cryotubes at 280°C. Infectious virus was assayed by an immunofocus assay. Prior to the assay, each tissue sample was weighed before the addition of 500 mL DMEM. Samples were homogenized for 30 s at 4.0 m/s using a MP Fast Prep-24 (MP Bio) and then assayed immediately for infectious virus.
Mice that survived the acute phase of the infection were randomly assigned to three endpoint groups and euthanized 28 dpi, 56 dpi, or 84 dpi in equal female-to-male ratios for each time point (Table 1). Samples were collected as noted for the acutely ill mice.
Virus quantification from blood and tissues. Immunofocus assays on Vero cells were used to quantify infectious DTV as previously described (57). Briefly, Vero cells were seeded 1 day prior to infection in 12-or 24-well plates at 1 Â 10 5 or 5 Â 10 4 cells per well, respectively. Serum or homogenized tissue was serially diluted 1:10 in DMEM. Medium was removed from the Vero cells and replaced with diluted sample. Plates were incubated at 37°C with gentle rocking for 1 h before inoculum was aspirated. Cells were washed once with 1Â phosphate-buffered saline (PBS), and an overlay of 0.8% methylcellulose in DMEM was added. After 6 days incubation at 37°C, the cells were fixed with 100% methanol. Cells were incubated with Powassan hyperimmune ascites fluid (ATCC) diluted 1:1,000 in Opti-MEM and anti-mouse horseradish peroxidase-conjugated secondary antibody (Dako) diluted 1:1,000 in Opti-MEM for 1 h each; then, plaques were developed using diaminobenzidine (Sigma) and 0.0135% H 2 O 2 (Sigma). Foci were counted and used to calculate viral titer (log 10 FFU) per milliliter of serum or per gram of tissue.
Viral RNA isolation and qRT-PCR. RNA was isolated from brain tissue with TRIzol (Invitrogen) followed by the RNeasy minikit (Qiagen) per manufacturers' protocols. Briefly, 30-mg brain samples were homogenized in 1 mL of TRIzol for 30 s at 4.0 m/s using a MP Fast Prep-24 (MP Bio) followed by the addition of 200 mL chloroform (Sigma). The phenol-chloroform phase was separated from the aqueous phase via centrifugation for 15 min at 12,000 Â g at 4°C. The RNA was then precipitated, washed, and eluted using the RNeasy minikit.
qRT-PCR was performed using the TaqMan Fast Virus 1-Step master mix kit (Applied Biosystems) according to the manufacturer's protocol on the QuantStudio 6 Flex real-time PCR system (Applied Biosystems). DTV-specific primers targeting NS5 (forward, 59-GATCATGAGAGCGGTGAGTGACT-39, and reverse, 59-GGATCTCACCTTTGCTATGAATTCA-39) and probe (59-6-carboxyfluoresceing [FAM]-TGAGCA CCTTCACAGCCGAGCCAG-MGBNFQ) were utilized (59). A standard curve was generated with 10-fold dilutions of homogenized mouse brain samples spiked with DTV stock. Numbers of genome copies per milliliter were calculated using a slope of 23.62 and y intercept of 34.53 from the standard curve (R 2 of 0.99).
Histology and ISH. Tissues were fixed in 10% neutral buffered formalin for a minimum of 7 days according to IBC-approved standard operating procedures (SOP). Tissues were processed with a Sakura VIP-6 Tissue Tek processor, on a 12-h automated schedule, using a graded series of ethanol, xylene, and PureAffin. Embedded tissues were sectioned at 5 mm and dried overnight at 42°C prior to staining with hematoxylin and eosin (H&E). The inflammation severity in H&E sections was scored by a blinded researcher as 0 (none), 1 (minimal), 2 (mild), or 3 (moderate).
Statistics. Change in body weight was analyzed by first transforming each weight value to percentage of the initial weight on day 0 for each mouse; then, groups were compared with repeated-measures mixed-effects models. Tukey-Kramer or Fisher post hoc tests were used to identify differences between the groups. Histological and ISH severity scores were compared using the Kruskal-Wallis nonparametric test followed by Dunn's multiple-comparison test. Viral titers for tissues and serum were compared for mice infected with 10 3 FFU only using analysis of variance (ANOVA) or Kruskal-Wallis tests depending on normality followed by Tukey's or Dunn's multiple-comparison tests, respectively. Gender and persistence status was analyzed using Fisher's exact tests. Statistical analyses were performed using GraphPad Prism version 8.2 for MacOS (GraphPad Software).
Data availability. All data from this study can be found in Data Set S1 in the supplemental material.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. DATA SET S1, XLSX file, 0.02 MB.