1. Introduction
The first interest in viral infections of the central nervous system (CNS) dates back to antiquity [
1], but neurovirology has only recently become a growing field with the discovery of less invasive techniques of surgery and diagnostic sampling in the late 20th century. Since the invention of new molecular tools and methods, like PCR and next generation sequencing (NGS), the spectrum of potential infectious agents has expanded, both in animals and in humans, providing new directions in investigating etiologically unresolved cases or diseases.
Astroviruses are small, non-enveloped, single-stranded RNA viruses with a star-like shape when viewed in the electron microscope. Since the discovery of the first astrovirus in a human stool sample in 1975 [
2], astroviruses have been described in several mammalian and avian species and are known as a cause of diarrhea and gastroenteritis in children [
3]. Recently, growing evidence of neurotropic astrovirus species has arisen in cases of encephalitis and meningitis in humans and animals [
4,
5,
6,
7,
8,
9,
10,
11]. Non-suppurative encephalitis is an inflammatory pattern observed in the context of post-mortem histopathological brain examinations and is indicative of viral infections. It is characterized by perivascular cuffs composed of mononuclear cells, by gliosis, and by neuronal necrosis. In Switzerland, non-suppurative encephalitis has been diagnosed in approximately 15% of neurologically-diseased cattle, but in many cases the etiology could not be determined and remained unknown [
12,
13,
14,
15]. These cases occurred sporadically and in single animals and were designated as “European sporadic bovine encephalitis” (ESBE). However, it has yet to be determined whether these cases of non-suppurative encephalitis were indeed caused by a single virus, or instead by several different viruses.
Using NGS, BoAstV has been found in the brains of cattle with non-suppurative encephalitis, termed BoAstV CH13 or BoAstV NeuroS1 [
6,
7]. As both virus isolates represent the same genotype species, it is hereafter referred to BoAstV Ch13/NeuroS1 [
16]. In small-scale retrospective studies, BoAstV CH13/NeuroS1 was found in the brains of 5/22 and 3/32 cattle with unresolved non-suppurative encephalitis in Switzerland and in the USA, respectively. These results have provided a first hint towards the possible association of the virus with the disease, but did not allow for robust conclusions.
The postulates of Koch are still considered the gold standard to prove causal relationships of microbes and diseases. These postulates define the following requirements for proof of causation: (i) microorganisms are associated with a disease and its lesions, (ii) microorganisms can be isolated from the host and purely cultured, and (iii) the disease is reproducible by inoculation of healthy hosts and that the same microorganism can then be isolated from that host [
17]. While these criteria remain an ideal to strive for, the conditions cannot always be met. Not all viruses readily replicate in cell culture systems and brain samples of diseased individuals are often not available. In consequence, many cases of viral infection of the CNS remain unresolved, leaving us largely unaware of potential threats for animal and public health. In 2015, Lipkin and Anthony suggested a refined classification for the assessment of causation, which is adapted to the situation in virus discovery by NGS [
18]. Criteria are defined to determine (i) a possible causal relationship (level 1) or (ii) a probable causal relationship (level 2), even in the absence of a culture system for virus propagation, and they include the effect of vaccines and specific drugs into the final level (level 3) of confirmed causation (
Figure 1). While a causal relationship is regarded as possible if a virus has been discovered in one or several diseased individuals, the relationship becomes probable with a statistical association of the virus with the disease, and when the virus is found at high concentration at the site of pathology.
For BoAstV CH13/NeuroS1, the postulates of Koch have not yet been fulfilled, because the virus cannot be isolated in cell culture and a reverse genetics system is not yet available. In the present study, we aimed at taking the first step towards assessing causality, by following the suggestions of Lipkin and Anthony. We performed a large-scale retrospective investigation of unresolved cases of non-suppurative encephalitis in order to assess the statistical association of BoAstV CH13/NeuroS1 infection and neurological disease and its neuroanatomical correlation to the pathological lesions in cattle. The results provide important insights on the frequency of BoAstV CH13/NeuroS1 infection in the Swiss cattle population and further support it being a major cause of encephalitis in cattle.
2. Materials and Methods
A total of 97 cases of histologically-confirmed bovine non-suppurative encephalitis of unresolved etiology were collected from the archives of the Vetsuisse Faculty in Bern and Zurich, Switzerland. All animals diagnosed were presenting with neurological symptoms, mainly gait abnormalities and behavioral changes. As negative controls, we analyzed samples of different brain regions of healthy animals without brain lesions (n = 52), and of animals with neurological disorders but with lesions indicating a bacterial origin (n = 6), under the same conditions. All cases were submitted to the neuropathological diagnostic services between 1985 and 2015. Tissue samples included formalin-fixed paraffin-embedded (FFPE) tissue of the CNS, mainly the brainstem, cerebellum, cerebrum and hippocampus. All tissues were analyzed by in situ hybridization (ISH) for the presence of BoAstV CH13/NeuroS1 RNA, and positive results were confirmed by immunohistochemistry (IHC). Tissues of cases with a known BoAstV CH13/NeuroS1 positive and negative status served as positive and negative controls.
A non-radioactive digoxygenin-labeled ISH was done according to the protocol of Bouzalas et al. [
7], using two ISH-probes (A and B) that are complementary to the 5’ end of open-reading frame (ORF) 2 and the center of ORF2 of the BoAstV CH13/NeuroS1 genome. Briefly, after deparaffinization, rehydration and 0.2M hydrogen chloride (HCL) treatment, we added Proteinase K (Roche, Mannheim, Germany), deactivated it with 4% paraformaldehyde, and incubated the tissue sections in a prehybridization mix before the hybridization-step and the application of the ISH-probes A and B on serial sections overnight. Probe binding was detected by anti-digoxygenin-AP-Fab fragments (Roche) and nitro blue tetrazolium (NBT)/5-bromo-4-chloro-3-indolyl-phosphate (BCIP) substrate (Roche).
IHC was performed according to the validated ORF2-con protocol [
19] and each run included negative and positive control sections. The hyperimmune serum ORF2-con detects viral epitopes on the conserved N-terminus of the capsid protein of BoAstV CH13/Neuro-S1. After deparaffinization, dehydration of the FFPE-tissue and peroxidase blocking in 3% H
2O
2 we used microwave heating (20 min at 95 °C) in Dako Target Retrieval solution, pH 9 (Agilent Technologies, Santa Clara, California, USA) as antigen retrieval method. Blocking was performed with 10% normal goat serum in phosphate-buffered saline (PBS)-Tween solution, after which the ORF2-con immunoserum (1:100) was added overnight at 4 °C. Antibody binding was detected by the Dako REAL Detection System according to the instructions of the manufacturer (Agilent Technologies). IHC and ISH grading was performed according to the number of stained neurons. Serial sections of all slides were submitted to hematoxylin-eosin staining for histopathological assessment of lesions.
ISH and IHC results were assessed using a semi-quantitative grading: (−) = no positive cells; (+) = 1–10 positive cells; (++) = 10–50 positive cells, and (+++) = >50 positive cells. Histopathological lesions were similarly graded as on haematoxylin and eosin (HE) stained slides: (−) = no lesions, (+) = mild lesions, (++) = moderate lesions, and (+++) = severe lesions. The assessment of both the ISH/ IHC and the histopathological grading was not performed on specific microscopic fields but on complete sections. The topographical correlation of lesions with BoAstV CH13/NeuroS1 detection was judged as good correlation, partial correlation, and the absence of correlation. A good correlation is seen in sections where the virus is located in the same microscopic field, affecting the same neuroanatomical area that shows the highest grade of lesions. A partial correlation does show an overlap in the presence of lesions and the virus, while not precisely the same, but a close-by neuroanatomical area is affected, or when major lesions comprise glial nodes or perivascular cuffs, but not neurons. Absence of correlation reflects a very large discrepancy between lesions and the presence of viral RNA (vRNA) or proteins [
20].
Statistical analysis was done by Fisher’s exact test using the Prism software (Version 5; GraphPad, La Jolla, California, USA).
4. Discussion
Using different methods of virus detection in FFPE brain tissues, we have performed a retrospective study on the frequency and neuroanatomical distribution of the neurotropic astrovirus BoAstV CH13/NeuroS1 in 97 cases of cattle diseased of non-suppurative encephalitis of unknown etiology and 58 negative controls, so as to further assess the causal relationship of virus infection and disease. The BoAstV was detectable in 34% of analyzed cases, but none in healthy animals, indicating a significant correlation to the disease. The histopathological findings in hematoxylin-eosin stained serial sections occur mainly in the very same areas where vRNA or proteins were detected by ISH and IHC, and in a majority of sections they are of a comparable intensity, which further substantiated the probable causal relation.
However, we observed exceptions in single sections of the brain where pathological lesions are present, but no virus was detectable, or vice versa. Since animals under investigation were at different stages of the disease and the intensity of lesions was variable between individuals, we propose that different CNS infection stages may be accountable for these results.
The BoAstV CH13/NeuroS1 was detectable in all parts of the CNS, but the intensity of infection was variable across cases. As it is known that bovine non-suppurative encephalitis shows a variation of intensity of perivascular cuffs, gliosis, and neuronal necrosis throughout the brain and brainstem [
12], our findings by ISH and IHC were to be expected, particularly since the presence of the virus mostly correlates to the pathology. Still, we cannot define a clear neuropathological phenotype of infection and suggest testing different brain regions in suspect cases to not miss a BoAstV CH13/NeuroS1 positive case. Although the brainstem does seem like a favorable region for testing since the majority of cases reveal a BoAstV CH13/NeuroS1 positive signal in this region, there are exceptions that urge for additional tests of at least one more brain region, preferably the cortex or hippocampus.
In all cases in this study, the BoAstV CH13/NeuroS1 shows a cell tropism for neurons only. vRNA or proteins could not be detected in microglia or astrocytes in any of the investigated cases, indicating that these cells might not be the major target. This result is in line with one of our previous studies, where there is evidence of BoAstV CH13/NeuroS1-positive glial nodes only in one case [
20].
It is known that enteric astrovirus infection in humans mainly occurs in winter [
21], but our study provides evidence that neurotropic isolates might also cause more clinical infections from the onset of winter to the period of spring. Nevertheless, this result is to be interpreted with caution, because the number of cases of non-suppurative bovine encephalitis normally drops during summer months. It is important to note that cattle in Switzerland are usually kept in the Alps during the summer and neurological or behavioral changes might not be detected or reported in a timely manner by the farmers, which could have an impact on studying seasonal effects of virus infection. However, it is reasonable to conclude that affected cattle are mainly of a younger age (3–5 years), and preliminary results of another ongoing clinical study show that in most cases, astrovirus-infected cattle suffer from gait abnormalities and changes in behavior (data not shown). The full clinical picture is still unresolved and therefore we would strongly urge for an enhancement to the system of reporting neurologically-diseased cattle and for considering the BoAstV CH13/NeuroS1 infection already at the stage of primary veterinary consultations in the field.
Although a high proportion of cases of non-suppurative encephalitis show BoAstV CH13/NeuroS1 infection in the CNS, approximately two thirds of the cases still remain unresolved. In previous studies, we have identified other viruses in such cases, among them another astrovirus, BoAstV CH15, and a report from Germany reported a very similar virus in the brain of a diseased cow [
9,
10,
22]. It is possible that more divergent, and unknown astroviruses could also cause viral encephalitis, but have gone undetected so far.
In conclusion, our results provide further evidence for a probable causal relationship between BoAstV CH13/NeuroS1 and neurological disease. The proof of causation is still needed and depends largely on successful virus isolation in cell culture and experimental induction of the disease. First attempts to propagate the virus in colorectal adenocarcinoma cells (Caco-2), which are frequently used for human enteric astroviruses [
23], and in fetal bovine brain cells (FBBC-1) were inconclusive, but more comprehensive studies are under way in our laboratory. In addition to this, it is necessary to study the route of infection in diseased animals and include other organ systems, outside the CNS, in the analysis. Only in two of the BoAstV CH13/NeuroS1 positive cases did we have sections of lung, liver, and intestines available, but none of these samples was positive for vRNA or proteins. A systematical investigation of other organs in a significant number of BoAstV CH13/NeuroS1 positive cases would be necessary to give an insight into the broader pathology of this disease.
Studies aiming at understanding other possible viruses causing bovine non-suppurative encephalitis, at developing in vitro systems, as well as tools for serological diagnostics and research into disease epidemiology, should be encouraged to further expand our understanding of the causes and significance of neuroinfectious diseases, which potentially threaten animal and public health.