The neuropathology of fatal encephalomyelitis in human Borna virus infection

The classical Borna disease virus 1 (BoDV-1; species Mammalian 1 orthobornavirus) can lead to lethal infections in a broad range of mammals. The virus is endemically present in white-toothed bicolored shrews (Crocidura leucodon) [1, 7, 16], while horses and sheep are main accidental dead-end hosts [9]. Historically, a severe neurological disorder (Borna disease) due to an encephalomyelitis lymphocytaria non-purulenta with loss of neurons had been described for accidental hosts [18]. It is assumed that the virus enters the organism via the olfactory network and spreads through the limbic system to nearly all cortical areas, brainstem, cerebellum, and spinal cord [9]. In contrast to dead-end hosts, reservoir hosts feature a disseminated virus distribution to nearly every organ with shedding of infectious virus [24]. Other animals, including rats and bank voles, can be infected experimentally via different routes [10, 19]. Until recently, no report on confirmed human bornavirus infection existed. In 2015, the first human infections with another orthobornavirus (variegated squirrel 1 bornavirus [VSBV-1]; species Mammalian 2 orthobornavirus) were described, which resulted in lethal encephalitis of three breeders of variegated squirrels (Sciurus variegatoides). All three patients showed similar central nervous system (CNS) symptoms with lesions in cerebral cortex, basal nuclei, and brainstem, but without involvement of the spinal cord [17]. Subsequently, two more VSBV-1-induced human encephalitis cases occurred, whereof one was lethal [32, 33]. The first evidence for human infections with the “classical” BoDV-1 was published by two simultaneous reports just recently in 2018. A group of three recipients of solid organ transplants from the same donor developed severe encephalitis of preliminary unknown origin, which was later convincingly demonstrated to be BoDV-1-associated [28]. Furthermore, BoDV-1 infection led to the death of a previously healthy 25-year-old man [21]. Shortly afterwards, a report on a sporadic BoDV-1 infection with initial symptoms mimicking a Guillain–Barré syndrome was reported [5]. All five patients suffered a massive non-purulent encephalitis leading to death in four of the cases [5, 21, 28]. One of the deceased immunocompromised transplant recipients was autopsied and provided full insight into CNS and peripheral nerve pathology (patient 1). This incident triggered a broad research on post mortem cases with yet non-classified encephalitis. Our screening revealed autoptic material of five additional cases of fatal human BoDV-1. All patients had appeared completely healthy prior to the neurologic disease.

With these six cases, our study provides the first comprehensive description of the neuropathology of human BoDV-1 infection with focus on inflammatory lesions and virus distribution patterns at a systemic and cellular level.

This retrospective study was approved by the local ethical committee of the Technical University Munich. The families of patients 1, 2, and 6 gave informed consent to complete autopsy, whereas the families of patients 3, 4, and 5 agreed to brain autopsy only. For patients 1, 2, and 6, macroscopic evaluation of the thoracic and abdominal cavity, including evaluation of all organs, was conducted by two pathologists. For microscopic analysis, representative samples from peripheral organs, including heart, lung, liver, kidney, trachea, esophagus, stomach, intestine, and others were obtained. Brain dissection was carried out by two neuropathologists after routine formalin-fixation of the undissected brain. For microscopic analysis, representative samples from multiple selected regions of the brain and spinal cord were retrieved from following areas: frontal, parietal and occipital cortex, hippocampus, basal ganglia, mesencephalon, pons, medulla oblongata, cerebellum, as well as cervical, thoracic and lumbar parts of the spinal cord. From patients 1 and 6, peripheral nerves including the sciatic nerve and brachial plexus from patient 1, and sural and femoral nerve as well as brachial and sacral plexus from patient 6 were sampled in addition to the autonomic nerve branches attached to visceral organs and tissues. All tissue samples were further fixed in 10% buffered formalin and, after paraffin embedding, cut into 2 µm-thick sections. Subsequently, different stains were performed, including hematoxylin and eosin (H&E) stain, Grocott’s methenamine silver stain, and periodic acid-Schiff (PAS) stain. We conducted an established avidin–biotin complex technique for immunohistochemistry (IHC) using the monoclonal Bo18 antibody, which is directed against the nucleoprotein (N) of BoDV-1 [11], as described previously in animals [15, 35]. Furthermore, BoDV-1 RNA in situ hybridization (V-BoDV1-G targeting 2-969 of NC_001607-1:2236-3747, RNAscope^®, Advanced Cell Diagnostics, Inc., USA) was performed as described previously [34]. For patients 1, 2, 3, and 5, infiltrating inflammatory cells were characterized by IHC using antibodies directed against CD3 (Cell Marque, The Netherlands; monoclonal, rabbit, dilution 1:500), CD20 (Dako Denmark A/S, Denmark; monoclonal, mouse, dilution 1:500), CD4 (Ventana Roche, Switzerland; monoclonal, rabbit, ready-to-use), CD8 (Ventana Roche, Switzerland; monoclonal, rabbit, ready-to-use), CD68 (Dako Denmark A/S, Denmark; monoclonal, mouse, dilution 1:15,000), CD138 (Cell Marque, The Netherlands; monoclonal, mouse, dilution 1:50), and Iba1 (abcam, USA; polyclonal, rabbit, dilution 1:500). These stainings were performed using a fully-automated staining system (Ventana BenchMark ULTRA; Ventana Medical Systems; Tucson; USA). Briefly, the slides were exposed to heat-induced epitope uncovering in pH 8.4 buffer at 95 °C for 40 min. After incubating with H₂O₂, the slides were incubated with the primary antibodies indicated above, followed by visualization using the 3,3′-diaminobenzidine-(DAB-)based OptiView DAB IHC Detection Kit (Ventana Medical Systems).

To characterize the cellular distribution of the infection, immunolabelling with Bo18 antibody was combined with cell markers Iba1 (abcam, USA; polyclonal rabbit, dilution 1:500), GFAP (Dako Denmark A/S, Denmark; rabbit polyclonal, dilution 1:500), S-100 (Dako Denmark A/S, Denmark; rabbit polyclonal, 1:500) and synaptophysin (Synaptic Systems, Germany; rabbit polyclonal, dilution 1:200) using an antibody removal agent (LinBlock, Linaris, Germany) and a green second chromogen (Histogreen^®, Linaris, Germany).

In the same vein, RNA in situ hybridization of BoDV-1 RNA (V-BoDV1-G targeting 2-969 of NC_001607-1:2236-3747, RNAscope^®, Advanced Cell Diagnostics, Inc., USA) was performed, followed by immunohistochemical counterstaining with glial fibrillary acidic protein (GFAP; Dako Denmark A/S, Denmark; monoclonal, mouse, dilution 1:500), S-100 (Agilent, USA; polyclonal, rabbit, dilution 1:6000), myelin basic protein (MBP; Cell Marque Corporation, USA; polyclonal, rabbit, ready-to-use, dilution 1:2), neurofilament (NFpan; Dako Denmark A/S, Denmark; monoclonal, mouse, dilution 1:200), Iba1 (abcam, USA; polyclonal, rabbit, dilution 1:500), CD68 (Dako Denmark A/S, Denmark; monoclonal, mouse, dilution 1:15.000) and CD3 (Cell Marque, The Netherlands; monoclonal, rabbit, dilution 1:500). BoDV-1-positive horse brain sections were used as controls. Lesions and immunohistochemical patterns were recorded and mapped for the individual patients.

For ultrastructural analyses, primary formalin-fixed brain samples from patient 6 were postfixed with Karnovsky-fixative (0.1 M cacodylate-buffer with 2.5% glutaraldehyde and 2% paraformaldehyde), followed by fixation in 1% osmium tetroxide at pH 7.3 for 2 h. Afterwards, they were dehydrated in graded ethanols, and embedded in the EMbed-812 epoxy resin (all reagents from Science Services, Germany; automated LYNX-tissue processor Leica, Germany). After 48 h of heat polymerisation at 60° C, semithin (0.8 µm) sections were cut and stained with toluidine blue and basic fuchsin. Light microscopy was used for selection of representative cells and ultrathin (80 nm) sections were cut with a diamond knife on a Reichert Ultracut-S ultramicrotome (Leica, Germany) and double stained with aqueous 2% uranyl acetate and lead citrate solutions for 10 min each. The sections were examined in a LEO912AB electron microscope (Zeiss, Germany) operating at 100 kV, equipped with a side-mounted CCD-camera (TRS, Lagerlechfeld / Germany) capable to record images with 2kx2k pixels. Documentation was done with the iTEM software (Olympus Soft Imaging Solutions GmbH, Germany).

For all six patients, the presence of BoDV-1 RNA in the brain was confirmed by reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) and complete BoDV-1 genome sequencing approaches as described earlier [28].

Numerous descriptions of the neuropathological findings of BoDV-1 encephalitis in naturally infected animals have been published since 1911 when Joest and Degen gave the first report on Borna disease in horses [18]. Nevertheless, a detailed neuropathological depiction of BoDV-1 infection in humans is missing so far, as human bornavirus infections were confirmed only recently in VSBV-1-infected breeders and animal care takers of exotic squirrels [17, 32, 33] and in BoDV-1-infected patients, including three recipients of organ transplants from the same donor [5, 21, 28]. All patients had developed encephalitis and neurologic disease resembling Borna disease in horses and sheep, which had been fatal in most cases [5, 17, 21, 28, 32, 33]. One organ recipient was autopsied at our institute and gave for the first time deeper insights into human BoDV-1 encephalitis [28].

Through BoDV-1 screening of current post mortem cases with yet non-classified encephalitis, autoptic material of five additional sporadic cases of fatal BoDV-1 infection was detected. With a series of six cases, we are now able to give a first detailed and systematic description of the neuropathological findings in human BoDV-1 infection.

All brains featured similarities in their histopathological appearance, which can be summarized as lymphocytic sclerosing panencephalomyelitis with strong formation of microglial nodules. With co-staining of BoDV-1 RNA in situ hybridization and cellular markers, infected cell types including neurons, astrocytes, and oligodendrocytes were identified. In contrast, microglial cells as well as macrophages and lymphocytes were not infected. In neurons, BoDV-1 was detected in cell bodies as well as neuritic and dendritic processes. Furthermore, in all patients, the so-called Joest–Degen inclusion bodies were found. Due to their morphology, these intranuclear inclusions are accounted among Cowdry-type B inclusions [6, 18, 26].

Even if the histopathological aspects were consistent in all patients, they are not specific for bornavirus encephalitis, though, as other virus encephalitides exhibit similar appearances [4, 8]. Perivascularly accentuated lymphocytic infiltrates, together with glial and microglial activation is often found in arbovirus infections including tick-borne encephalitis [8]. Nevertheless, nuclear inclusion bodies as seen in all BoDV-1-infected patients have not been described for tick-borne encephalitis and might be a helpful distinctive criterion. It should be noted that Southern Germany, the home area of all patients of this study, is an endemic region for both, tick-borne encephalitis virus and BoDV-1 [14, 20]. Furthermore, the strong sclerosing aspect due to glial activation together with perivascular accentuated inflammatory infiltrates as seen in bornavirus encephalitis might resemble subacute sclerosing panencephalitis (SSPE) as secondary manifestation of measles virus infection [4]. Although the clinical aspects of SSPE usually differ from the described courses of BoDV-1 infection, bornaviruses should be considered as possible differential diagnoses.

In contrast to the constancy of the histopathological changes, interindividual differences regarding the distribution patterns of viral antigens and inflammatory infiltrates were observes. Two patients showed accentuation of lymphocytes in the basal nuclei, whereas in one patient, the hippocampal area and in another the upper brain stem were primarily affected. Two patients, however, showed a more uniform distribution including the grey as well as the white matter without clear local accentuation. It should be noted, however, that the described distribution patterns reflect the end-stage pathology after terminal disease. Nevertheless, accentuations might indicate primarily affected sites that may be targeted by diagnostic biopsy testing. All five previously healthy patients developed central neurological deficits early in the course of disease, whereas peripheral symptoms were not predominant. In contrast, the immunocompromised patient 1 developed initial symptoms mimicking Guillain–Barré syndrome, which is in line with BoDV-1 infection and inflammation observed in the PNS of this patient. Similar symptomology had been described for the second kidney recipient [28] and a recently published previously healthy BoDV-1-infected patient [5]. Furthermore, small nerves of peripheral organs of patients 1 and 2, including trachea and esophagus, were BoDV-1-positive. Infectious neuropathies have been described for other viruses, including hepatitis C virus and several herpesviruses, but are very rare incidents and occur only in a rather late course of the virus infections [2]. For patient 1, though, the neuritis was an early event, given the development of peripheral symptoms before any signs of CNS affection. Infection of the PNS has already been observed in experimentally BoDV-1-infected animals [19]. For human VSBV-1 infection, no PNS infection was described; the spinal cord showed no affection either [17, 32, 33]. Naturally BoDV-1-infected accidental animal hosts, including horses and sheep, rarely show signs of PNS infection [9]. In contrast, reservoir hosts, such as BoDV-1-infected bicolored white-toothed shrews or VSBV-1-infected squirrels, harbor the virus not only in central and peripheral nervous tissue, but also in non-neuronal cells including mesenchymal and epithelial cells of nearly every organ system, e.g., the respiratory and urinary tracts. However, bornavirus-induced inflammation is not observed in these hosts [7, 24, 27, 29]. Infectious virus can be detected in several excretions of BoDV-1-infected shrews, including saliva, urine, and feces [24], which is assumed to infect horses and possibly other accidental hosts via an olfactory route [23], followed by a spread to the limbic system [9]. This may explain the primary affection of these phylogenetically older areas of the CNS in horses [10, 22]. For humans, the route of entry has not been clarified, yet. Only for patient 1, the infected kidney transplant was confirmed as the source of infection, since retrospectively, the donor was retrospectively proven to be seropositive for BoDV-1, all three organ recipients suffered from bornavirus encephalitis, and identical viral sequences were found in the brains of both kidney recipients [28]. A significant accentuation of the limbic system indicating an olfactory route, as described for horses [10, 22] was observed only in patient 5, whereas patients 3 and 4 revealed a strong accentuation of the basal nuclei.

Interestingly, the course of disease was prolongated in patient 1 compared to the other patients (14 weeks compared to 2–6 weeks). The cytotoxicity in BoDV-1 infection is known to be mediated by BoDV-1-specific CD8-positive T lymphocytes [3, 13, 30]. Thus, therapeutic immunosuppression after kidney transplantation might have decelerated the course of disease in patient 1 and might give lead to therapeutic possibilities. T-lymphocyte depletion or suppression starting before or early after experimental BoDV-1 infection was demonstrated to prevent immunopathology in rodent models [12, 25, 31].

The strong interindividual differences lead to several intriguing questions including infections routes, dose of infectious virus, the impact of the duration of infection, susceptibility, and potential risk factors. Here, we present the first comprehensive description on the neuropathology of human victims of bornavirus encephalitis. Our report is based on only six autopsies since the first confirmed human BoDV-1 infections appeared just recently [5, 21, 28]. Our data aim to initiate a better understanding of the circumstances and pathobiology of human Borna disease that are clearly needed to identify so far undiagnosed cases and develop an epidemiological basis.

On the basis of six brain autopsies, human BoDV-1 infection was shown to result in lymphocytic sclerosing panencephalomyelitis with detection of BoDV-1-typical eosinophilic, spherical, intranuclear Joest–Degen inclusion bodies. Bornavirus infection should be considered in cases matching these characteristics. While the composition of histopathological changes appears consistent, the lesion distribution pattern varies interindividually, which might indicate different routes of entry or individual immune status. Infection of the PNS may occur and results in clinical presentation resembling Guillain–Barré syndrome in some cases. The common and overlapping neuropathological signs we here describe will likely stimulate the identification of additional cases. These could then be used to gain a clearer depiction of locoregional manifestation patterns and to the identification of favorable locations for diagnostic biopsies.