Detection of Astrovirus in Historical Cases of European Sporadic Bovine Encephalitis, Switzerland 1958–1976

European sporadic bovine encephalitis is a frequent diagnosis in neurologically diseased cattle, but its etiology remained unresolved. Using in situ hybridization, we have detected a recently discovered neurotropic bovine astrovirus in historical tissues in a high proportion of brain samples of affected cattle. Our results suggest that astroviruses were already involved in the pathogenesis of the disease several decades ago, but have gone undetected.


Tissue samples
We selected formalin-fixed paraffin-embedded (FFPE) brain tissues of histologically confirmed cases of ESBE (n = 14, Table 1), of bovines without histological evidence of encephalitis (n = 8), and of encephalitis cases of other etiology than viral (n = 8) from our archives. Tissues included sagittal sections of the brainstem, cerebellum, cerebrum, hippocampus, and additional CNS regions, if available. As the paraffin was friable, tissues were reembedded in fresh paraffin to be suitable for microtome cutting. All cases showed lesions consistent with the diagnosis of ESBE, which comprise neuronal necrosis, gliosis, and lymphohistiocytic perivascular cuffs. Details on the ESBE cases included in this study are presented in Table 1.

histopathology and In Situ hybridization
Serial sections of 4 μm were cut from FFPE blocks and mounted on SuperFrost Ultra plus glass slides (R. Langenbrinck) for hematoxylin and eosin staining and in situ hybridization (ISH). Pathological lesions were evaluated semi-quantitatively as mild, moderate, or severe inflammation. We applied a non-radioactive digoxigenin-labeled ISH using two distinct probes (A and B) for the detection of BoAstV CH13/ NeuroS1 RNA, as described previously (8). ISH-probes A and B were generated by in vitro transcription from cloned cDNA with the DIG RNA Labeling Kit (Roche Applied Science). Probe A is complementary to a sequence at the 5′ end of open-reading frame (ORF) 2, whereas probe B is complementary to a sequence in the center of ORF2 of the viral genome. Briefly, tissue sections were deparaffinized, rehydrated, and treated with 0.2M HCl. Proteinase K (Roche) was added to a buffer containing 1M Tris-HCl and 0.1M CaCl2, incubated on the slides at 37°C for 15 min and deactivated by 4% (w/v) paraformaldehyde. Next, the slides were treated with a hybridization mix containing 50% (v/v) deionized formamide, 0.05% Ficoll (w/v), 0.05% polyvinylpyrrolidone (w/v), 0.05% BSA (w/v), 4× SSC, and 0.25% yeast RNA (w/v). After 2 h of prehybridization at 50°C, slides were incubated with either of the two ISH probes [1 ng/ ml in hybridization mix, supplemented with 0.5% (w/v) yeast RNA and 10% (w/v) dextran sulfate] for 15 h at 50°C. Intense washing and RNAse treatment were performed to enhance binding specificity. Binding of probes was finally detected using anti-digoxigenin-AP-Fab fragments (Roche) and NBT/BCIP substrate (Roche).

rna extraction from FFPe Tissue Blocks
Formalin-fixed paraffin-embedded tissues were cut to 20 μm sections using fresh blades for each sample. The microtome was

Detection of BoastV ch13/neuros1
Twelve ESBE cases revealed a positive deep purple staining in at least one of the tested brain regions with ISH probe A. However, none of the cases showed positive labeling with probe B. Negative control tissue sections analyzed in parallel in each run (Figure 1), as well as the brain tissues of bovines without histological lesions (n = 8) and with encephalitis of other etiologies than viral (n = 8), remained unlabeled with both probes. Collectively, these data confirm the presence of astrovirus RNA in the brains of 85% (12/14) of the historical cases of ESBE. In these animals, the astrovirus RNA could be detected in different regions of the central nervous system, but the distribution pattern was variable between cases ( Table 3). The brainstem was astrovirus positive in the majority of the cases, but mostly single neurons were labeled, whereas in the hippocampus, the infection tended to be more severe, affecting almost all cells of the pyramidal and molecular layer. However, the limited number of cases and the limited availability of brain regions for some cases of non-suppurative encephalitis do not allow further conclusions on the virus distribution.

correlation of BoastV ch13/neuros1 rna and histopathological Brain lesions
To further investigate whether astrovirus positive cells were topographically related to histopathological lesions, ISH labeling was assessed semi-quantitatively based on the number of ISH labeled cells. The ISH labeling was in most cases confined to neurons, as judged by the morphological appearance of the cells. Interestingly, in case 11930, we detected a positive staining in mononuclear cells of a perivascular cuff in addition to the neurons (Figures 2A,B), and in case 3628, cells in a glial node were reactive with the ISH probe as well (Figures 2C,D). In all other cases, astrovirus RNA was found predominantly in neurons, but without a clear association to neuronal necrosis, gliosis, or perivascular infiltrations. In three animals, labeled neurons were found in proximity to histological lesions (Figures 2E,F). However, in some animals, astrovirus RNA was detected in neurons without any association to lesions, and vice versa, in many areas with lesions, the cells were devoid from astrovirus RNA (Figures 2G,H).     The discrepancy in the reactivity of ISH probe A and B in ESBE cases was unexpected, since more recent BoAstV CH13/NeuroS1 positive tissues (8), as well as the positive control tissues used in the present study, were invariably reactive with both probes (Figures 1C,D). This raised the possibility that the historical cases had been infected with an astrovirus different to BoAstV CH13/NeuroS1. To address this point, we extracted total RNA from FFPE tissue and performed two different RT-PCR protocols.
Despite several attempts, RNA yields for most FFPE tissues were below the detection limit (as assessed by Fragment Analyzer CE12), and RT-PCRs did not yield amplicons. Yet in one animal (case 3466), we were able to obtain amplicons by both RT-PCR protocols. Amplicon sequences were >98% identical to those of BoAstV CH13/NeuroS1, confirming that the target region for ISH probe B shows no significant difference in sequence analysis and that this animal was very likely to have been infected with BoAstV CH13/NeuroS1 (Figure 3).

DiscUssiOn
We found astrovirus RNA in the brains of a high proportion of historical ESBE cases, but not in control animals, which underpins the association with the disease. The astrovirus is present in different brain regions, affecting not only neurons but also inflammatory cells in perivascular cuffs and glial nodes, without revealing a clear neuroanatomical distribution pattern. Generally, the microtopographical localization of the viral RNA did not match that of the histopathological lesions in the affected brains. This could be due to specific disease pathomechanisms: the immune response can lead to rapid clearance of the virus in inflammatory lesions, as known for tick-borne encephalitis (10). However, it could also indicate that the astrovirus is neuroinvasive, but with limited neurovirulence, and that other yet unidentified viruses could play a role as cofactors. Indeed, there has been evidence for a paramyxovirus in brain tissues of cases of ESBE (6,11), and very recently, we have reported the second neurotropic astrovirus (BoAstV CH15) in two cows with nonsuppurative encephalitis in Switzerland (12). More comprehensive studies aiming at identifying disease-associated pathogens beyond BoAstV CH13/NeuroS1 in cases of bovine encephalitis have identified additional candidate viruses, such as parainfluenzavirus 5, bovine polyomavirus 2, and bovine herpesvirus 6, which might be involved in the etiology of non-suppurative encephalitis in cattle as well (13). At the moment, we have not yet established the tools to detect these viruses in historical FFPE tissues; however, it will be interesting to investigate such cases for potential coinfections with BoAstV-CH/NeuroS1.
Long-term storage of tissues, re-embedding in fresh paraffin, and loss of RNA integrity may account for negative RT-PCR results and for the failure in detecting viral RNA by ISH probe B. Still, the virus likely corresponds to BoAstV CH13/NeuroS1, because ISH probe A was specifically designed for this virus, and we recovered BoAstV CH13/NeuroS1 sequences in one of these animals by RT-PCR. However, we cannot exclude the possibility that ISH-positive cases with unconfirmed virus sequences might be infected with a different neurotropic astrovirus with sequence similarities to BoAstV-CH13/NeuroS1 in the binding region of ISH probe A. This may, in particular, be applicable in the two cases in which the IHC labeling indicated virus infection in cells other than neurons.
Viral encephalitis of unknown origin has been a frequent diagnosis in cattle with neurological disease ever since neuropathologists such as Frauchiger and Fankhauser started to investigate such cases. Now, we are increasingly gaining knowledge on possible etiologies, among which neurotropic astroviruses are to be found. Our results provide evidence that astroviruses were involved in the pathogenesis of bovine poliencephalitis decades ago and serve as a prerequisite for understanding the pathogenesis of so far unresolved viral neuroinfectious diseases.