CSF Brain-Reactive Autoantibodies are Elevated in Patients with Viral Encephalitis

Dear Editor,

Brain-reactive autoantibodies are thought to play an important role in mediating disorders of the central nervous system (CNS). These antibodies direct the processes underlying several diseases, such as multiple sclerosis (MS), neuromyelitis optica (NMO), and neuropsychiatric systemic lupus erythematosus [1–3]. In infectious diseases of the CNS, the pathogen itself is regarded to play a major role in the pathogenesis [4]. In recent years, post-viral autoimmune reactions have been identified to be critical factors that influence the pathogenesis [5, 6]. The best understood example is anti-NMDAR (N-methyl-d-aspartate receptor) encephalitis [7], which is commonly caused by viral infections, especially by herpes simplex virus (HSV) infection.

The outcome of a viral insult to the CNS is determined by the neurovirulence of the infecting virus [8]. While patients with virus infection may recover from the neurological dysfunction after virus clearance, some may suffer from post-viral autoimmune encephalitis, which is reported to occur a few weeks or months after the infection [9]. In many cases, brain-reactive autoantibodies have been detected in parallel with the disease onset, suggesting that autoimmunity is an early event in viral encephalitis (VE) [10]. Indeed, the major role of the virus itself in the pathogenesis of the disease in the acute phase of infection is evident. However, it is not yet clear whether the virus-associated autoimmune reactions have pathological effects in the early stage of the disease, rather than only in the post-viral stages.

In the present study, we aimed to investigate the levels of brain-reactive autoantibodies in the cerebral spinal fluid (CSF) and their correlation with disease severity. We also tested whether these antibodies have pathogenic effects on neurons in vitro. The types of brain-reactive autoantibodies generated after viral infections are related to the type of virus. Furthermore, one specific virus may trigger different types of autoantibodies [11–13]. Therefore, we regard the brain-reactive antibodies as an integral pool in which to investigate their levels and functions.

Here, we recruited 48 VE patients and 40 normal controls (NCs). There were no significant differences in age, gender, and CSF glucose levels between the NC and VE participants (P > 0.05). The CSF white blood cell (WBC, P < 0.0001) and CSF protein levels (P < 0.0001) significantly differed between the two groups. Moreover, the major CSF viruses identified were Epstein-Barr virus and HSV (Table S1).

To confirm that brain-reactive autoantibodies exist in human CSF, we first conducted immunofluorescence staining. Human brain sections were co-stained with human CSF and anti-NeuN antibodies. All sections contained double-positive cells (CSF and anti-NeuN antibody) in both the NC and VE groups, indicating that brain-reactive autoantibodies are ubiquitous in human CSF. Notably, immunofluorescence staining showed that most of the CSF brain-reactive autoantibodies preferentially bound to neuronal membranes (Fig. 1A, B). Moreover, the number of double-positive cells, indicating the relative level of brain-reactive autoantibodies, was significantly higher in VE patients than in NCs (Fig. 1D). These results indicated that brain-reactive autoantibodies are increased in the CSF of VE patients. MRI abnormalities indicated parenchymal damage or meningopathy, so we further conducted subgroup analysis to investigate the CSF levels of brain-reactive autoantibodies in patients with and without MRI abnormalities. We found that the relative CSF autoantibody levels were significantly higher in patients with MRI abnormalities than those without them (Fig. 1C, E).

Fig. 1.

CSF levels of brain-reactive autoantibodies in VE patients as indicated by immunofluorescence staining. A, B Representative immunofluorescence staining images of CSF (red) and anti-NeuN antibodies (green), and DAPI (blue) in human brain. C Representative MRI and immunofluorescence staining images from VE patients with or without parenchymal damage. D, E Quantification of double-positive cells in human brain. Data are expressed as the mean ± SEM (n = 40 for NC group, n = 48 for VE group; n = 40 for no-parenchymal damage group, n = 5 for parenchymal damage group). Scale bar = 50 µm.

We used ELISA assays to explore the changes of inflammatory factors in VE patients, and found that the levels of CSF tumor necrosis factor alpha (TNF-α) and interleukin (IL)-6 were significantly higher in VE patients than in NCs (both P < 0.001). However, we found no significant difference in CSF IL-1β levels between NCs and VE patients (Fig. S1). Analysis by Spearman correlations between indicators of disease severity and the relative levels of brain-reactive autoantibodies as indicated by immunofluorescence staining showed that none of the inflammatory factors was correlated with the relative level of autoantibodies (Fig. S2A–C). However, we found that the CSF WBC and protein levels were positively correlated with the relative level of CSF brain-reactive autoantibodies in VE patients (Fig. S2D, E), indicating that these autoantibodies might participate in the pathogenesis of the disease and might be potential biomarkers of disease severity. No significant correlation was found between the CSF levels of brain-reactive autoantibodies and glucose (Fig. S2F).

To investigate the functions of brain-reactive autoantibodies in the CSF, we isolated them from the CSF of NCs and patients with VE. After co-culture of the autoantibodies and SH-SY5Y cells for 24 h, phase-contrast microscopy showed that autoantibody stimulation decreased the number of adhering cells and shortened the neurites (Fig. 2A–C). Moreover, brain-reactive autoantibodies from the VE group were more toxic to SH-SY5Y cells than those from the NC group, as reflected by shorter neurite length (Fig. 2B, C). To confirm this, we performed CCK8 analysis and found that brain-reactive autoantibodies decreased the viability of SH-SY5Y cells (Fig. 2J). In addition, brain-reactive autoantibodies from the VE group were more toxic to SH-SY5Y cells, as reflected by lower cell viability than those from the NC group (Fig. 2J). We also analyzed the apoptosis of SH-SY5Y cells with TUNEL staining and demonstrated that treatment with brain-reactive autoantibodies from both the VE and NC groups increased the ratio of TUNEL-positive SH-SY5Y cells compared to controls (Fig. 2D–F). The ratio of TUNEL-positive cells in the VE group approximately doubled that in the NC group (Fig. 2K). We then assessed neurite outgrowth in primary neuronal cultures from C57 mice. Tuj1 staining indicated that primary neurons cultured with brain-reactive autoantibodies from the VE group had shorter neurites than those cultured with the autoantibodies from the NC group or vehicle controls (Fig. 2G–I, L).

Fig. 2.

Effects of human CSF brain-reactive autoantibodies on the viability of SH-SY5Y cells and neurite outgrowth of primary neurons. A–C Representative light-microscopic images of SH-SY5Y cells exposed to CSF brain-reactive autoantibodies. D–F Representative images of TUNEL-stained SH-SY5Y cells (arrows) exposed to CSF brain-reactive autoantibodies. G–H Representative Tuj1-staining of primary neurons exposed to CSF brain-reactive autoantibodies. J–L Quantification of viability of SH-SY5Y cells (J), TUNEL-positive SH-SY5Y cells (K), and neurite length of primary neurons (L). Data are expressed as the mean ± SEM (n = 5/group). Scale bar = 50 µm.

The elevation of brain-reactive antibodies in the CSF of VE patients may be explained by several possible mechanisms (reviewed in [14]). First, the release of neuronal antigens after neuronal degeneration. Specific antigens are transported to local lymph nodes and presented to specific B cells, which then mature into plasma cells. Co-stimulatory signals might stem from necrotic tissue or the virus directly, including pathogen-associated molecular patterns. Alternatively, the inflammatory milieu subsequent to viral infection might lead to unspecific stimulation of brain plasma cells. This could explain why the autoantibodies are not restricted to target NMDARs, but also to dopamine D2 receptors [15], α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors, and gamma aminobutyric acid receptors [9]. Such polyspecific B-cell activation is seen in other types of CNS autoimmune disorder, such as MS. Furthermore, autoimmunity may result from molecular mimicry, where epitopes on the viral surface that are similar to neuronal antigens may trigger specific autoantibodies that cross-react with neurons [16]. In the present study, the VE patients were infected by various viruses, which could have generated different autoantibodies. Therefore, we regard the brain-reactive autoantibodies as an integral pool in which to investigate their levels and functions.

Brain-reactive autoantibodies have been suggested to participate in different pathological processes [17, 18]. Previously, the recognition of autoantibodies in CNS diseases has mainly been for the autoimmune diseases, such as MS and NMO. It has been suggested that brain-reactive autoantibodies can also be triggered by viruses [19]. However, post-viral autoimmune reactions are thought to occur in the late stage of the disease or after clearance of the viruses [9]. In the early stage of VE, the neurological dysfunction may be directly related to the neurovirulence of the infecting virus. Currently, few studies have focused on the role of autoantibodies in the pathogenesis of VE at the early stage. In the present study, we found that brain-reactive autoantibodies are elevated in the CSF of VE patients. The pathological importance of such an elevation is not yet clear. Brain-reactive autoantibodies, which may cross-bind with neurons, are generally regarded to be neurotoxic in many diseases [20, 21]. We found that brain-reactive autoantibodies were correlated with diverse disease-related factors (CSF protein and CSF WBC), indicating that this pool of autoantibodies may participate in the pathogenesis of the disease.

We further co-cultured SH-SY5Y cells with isolated autoantibodies, and found that they were toxic in vitro, as reflected by decreased cell viability and shortened neurites, indicating that pathogenic autoantibodies exist among those isolated from the CSF of VE patients. Furthermore, the viability of SH-SY5Y cells was lower and the average neurite length was shorter after exposure to autoantibodies from VE patients than to those from NCs. Moreover, the autoantibodies decreased the neurite growth of primary neurons. This finding raised the possibility that the proportion of pathogenic autoantibodies is higher in the CSF of VE patients. However, we did not isolate specific autoantibodies, so further studies are needed focusing on the role of specific autoantibodies in the pathogenesis of VE at the early stage.

In summary, we found that human CSF contains pathogenic brain-reactive autoantibodies, which are increased in VE patients at the early stage. We propose that brain-reactive autoantibodies participate in the pathogenesis of VE at the early stage. Studies with a larger sample size are needed to verify the findings of the present study.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Conflict of interest

The authors declare that they have no competing interests.