Background
Rasmussen's encephalitis (RE) is a rare chronic neurological disorder mainly affecting children [
1]. The clinical manifestations of RE are characterized by unilateral drug-resistant epilepsy, progressive neurological, cognitive deterioration, and brain atrophy [
2,
3]. Hemispherectomy of the affected side still remains the only cure for controlling the seizures and cognitive deterioration caused by the disease [
4‐
6].
Although much progress has been made in the pathogenesis of RE, the fundamental cause remains unknown. Viral infections and autoimmunity are considered to be possible causes. Since 1958 when RE was first reported [
7,
8], several studies have reported the detection of a number of viruses including enteroviruses, Epstein-Barr virus (EBV), human cytomegalovirus (HCMV), human papillomavirus and herpes simplex virus (HSV) in the brains of RE patients. However, due to the variety of viruses detected, the lack of specificity, and the high infection rate of some viruses in the population, the causal relationship between a virus and RE has not yet been established.
In addition to viral infection, cytotoxic T lymphocyte (CTL) infiltration in the brain is a predominant feature of RE, and common T cell clones are shared among RE patients [
9], indicating that the CTL-mediated immune response plays an important role in the pathogenesis of RE [
10]. Autoimmunity was considered the cause of RE [
11]. However, anti-inflammatory therapy showed poor efficacy, and autoantibodies were detected in only a few patients, suggesting that the precise antigen triggering the T-cell response still requires careful identification [
12‐
15]. Recently, Anna et al. found that microglial nodules provides the environment for pathogenic T cells, accompanied with upregulation of interferon(IFN)-β [
16]. This finding indicates a possible association between CTL and virus infection in RE pathogenesis, and suggests that the interaction between viral infection and the immune response may lead to RE disease.
Human herpes viruses (HHVs) consist of a group of double-stranded DNA viruses and include HSV1, HSV2, varicella zoster virus (VZV), EBV, HCMV, HHV6, HHV7 and Kaposi's sarcoma associated herpes virus (KSHV) [
17]. Most of them are neurotropic and are likely to cause latent infection in the ganglion after acute infection [
18,
19]. Previously, we detected antigens of HCMV, EBV and HHV6 with positive rates ranging from 50% to 88.5% of RE cases [
20,
21]. The expression levels of HCMV, EBV and HHV6 antigens showed a positive correlation with brain atrophy, indicating the involvement of HHVs infection in this disease. However, HHV infection is common in humans and has a high seropositive rate in the population [
17]. In addition to their presence in RE, HHV antigens have also been detected in other epilepsy seizure diseases, such as temporal lobe epilepsy (TLE) [
22]. Therefore, the prevalence of HHV antigens in RE patients raises uncertainty about the effect of HHVs on pathogenesis of RE. Are HHVs true pathogenic factors or just an incidental coincidence?
To uncover the relationship between HHVs and RE, we collected more brain samples in this study, not only from RE patients, but also from TLE patients and from people suffering traumatic brain injuries (TBI). By comparing HHV antigen expression, CTL-mediated immune response and innate immune response in these samples, we found that RE was correlated with HHVs infection and characterized by relatively lower level of innate immune response and the release of GZMB by cytotoxic T cell.
Materials and methods
Patients
30 patients (16 males and 14 females) who met the 2005 European diagnostic criteria for RE [
23] and underwent operation in Beijing Sanbo Brain Hospital from 2008 to 2018 were enrolled in this study. The scoring scheme of brain atrophy degrees of RE patients for the MRI was followed a previous report [
24]. 20 patients (13 males and 7 females) who were diagnosed as temporal lobe epilepsy (TLE) and underwent operation due to intractable epilepsy, as well as 16 patients (10 males and 6 females) who underwent surgical treatment for traumatic brain injuries (TBI), were included as controls. RE patients had a mean age of seizure onset of 5.6 years and a mean age at surgery of 6.8 years. All TLE and TBI cases matched for age at the time of surgery.
Brain tissue preparation
For RE and TLE cases, at least two blocks of brain tissues from different areas were collected. For TBI cases, brain tissue was collected at the edge of resected contusion area. The tissues were fixed, embedded, sliced, and then subjected to histopathological analysis. For the staining of each antigen, at least three sections from a single patient were used and each of the three sections was selected from ten successive sections so that the three sections were apart from each other at least 40 µm.
Immunohistochemical staining (IHC)
In brief, the endogenous peroxidase activity and non-specific antibody binding sites were blocked with 3% hydrogen peroxide and 1% bovine serum albumin (BSA), respectively. After antigen retrieval, the sections were incubated at 4 °C overnight with the primary antibodies. The detailed information of the primary antibody was shown in Additional file
1. Then incubation with the secondary antibody and DAB was completed routinely. Counterstained with hematoxylin was used to visualize the outline of the cells. The counterstained nucleus served as the control, which is conducive to obtaining a clearer cell morphology. All sections were observed recorded, and photographed under a light microscope (Olympus X-cite 120, Olympus, Japan).
Scoring methodology of IHC
The IHC results were scored as previously described [
20]. Cells showing yellow or brown particles in the cytoplasm or nucleus were considered positive and were counted and analyzed using image analysis software (Image-Pro Plus 6.0; Media Cybernetics Inc., Bethesda, USA). The semi-quantitative results were expressed as the percentage of positive cells combined with a subjective assessment of staining intensity. The staining intensity was scored as 0 (colorless), 1 (light yellow), 2 (yellow or brown), and 3 (dark brown); the percentages of positive cells were denoted as 0 (< 5%), 1 (5–25%), 2 (26–50%), 3 (51%–75%), and 4 (> 75%). The multiplication of the score for staining intensity and that for percentages of positive cells was used to evaluate the immunostaining results as follows: overall scores of 0, 1− ≤ 3, > 3–6, and > 6 were defined as negative, weakly positive, moderately positive and strongly positive, respectively.
Double immune-fluorescence staining
The sections were treated for antigen retrieval and then were simultaneously incubated with two primary or secondary antibodies. The information for all antibodies were shown in Additional file
1: Table S1. The images were taken using a laser scanning confocal microscope (TCS SP8 Leica, Germany).
DNA Dot blot assay
Total DNA was extracted using a DNA extraction kit (DNeasy Blood & Tissue Kit, Qiagen, Germany), and DNA dot blot assay was conducted according to Cai’s protocol [
25]. Briefly, denatured DNA was added to a nylon membrane to generate a dot array, and then DNA was fixed with UV-light for 7 kJ/10 min, followed by incubation with the preheat DIG Easy Hyb buffer in at 37 °C for 30 min. Denature DIG labeled DNA was treaded by boiling water and cooling in ice, and then was incubated with DIG Easy Hyb buffer at 42 °C overnight, and followed by washing with Blocking buffer. Then, after incubation with 5 mL DIG antibody solution at 25 °C for 1 h, Chemiluminescent substrate in alkaline phosphatase detection buffer was added to the membrane.
TUNEL staining
The apoptosis was analyzed using an In Situ Cell Death Detection Kit, Fluorescein (11684795910, Roche) according to the manufacturer’s instructions (520 nm) and counterstained with DAPI (460 mm) for nuclei in paraffin sections. The images were taken using a confocal laser microscope (TCS SP8 Leica, Germany). Five none-overlapping views of lesion area in each section were chosen randomly by an investigator blinded to the experiment design.
Multiplex flow cytometry
The cytokine levels in brain tissue and cerebrospinal fluid (CSF) samples of patients were quantified by AimPlex Human Multiple Immunoassays kits (Beijing Quantobio Biotechnology Co., China). Briefly, antigen was incubated with capture antibody conjugated beads, then with biotinylated detection antibodies, and finally with streptavidin-PE successively. Fluorescence signals of the sample were acquired by a flow cytometer (NovoCyte D1040) and levels of cytokines were analyzed with FCAP Array 3.0.
Statistical analysis
For the scoring data obtained from IHC image, Mann–Whitney U test and Spearman's rank correlation coefficient was used to compare the difference and determine the correlation respectively. For analysis involving cumulative viral score or MRI score, Student’s t-test and Spearman's rank correlation coefficient was used to compare the difference and determine the correlation respectively. For data from multiplex flow cytometry, Student’s t-test and Pearson correlation coefficient was used to compare the difference and determine the correlation respectively. Statistical analysis was performed using SPSS 23.0 software. P-values < 0.05 were considered statistically significant.
Discussion
Recently, increasing evidence has revealed the crucial role of the immune process in RE pathogenesis, but the primary cause that initiates immunity remains to be elucidated. In this study, we demonstrated that HHV antigens were prevalent in both RE and TLE patients. However, these two diseases displayed distinct patterns with respect to the status of infiltrated cytotoxic T cells, the production of cytokines, the polarization of microglia and the activation of DNA sensors, indicating that different immune responses to HHV infection occurred in RE and TLE cases. In particular, the STING and IFI16 at protein level, as well as the production of IFN-β were lower in RE patients, but the activation of CD8+ T lymphocytes was stronger. This implied that the innate immune response in RE cases was relatively less activated and that the compensating activated cytotoxic T cells inclined to eliminate HHV-infected neural cells. In contrast, a balance between host immunity and HHVs maintained a steady state in the brains of TLE cases. Thus, the brains of RE was continually damaged and developed into a progressive neurological disorder characterized by brain atrophy.
Viral infection has long been proposed as the etiology of RE [
7,
8]. In our previous research, antigens of various HHVs including HCMV, EBV and HHV6 were detected in RE brain tissues [
20,
21]. Nevertheless, HHV infection is very common in humans [
17], so that the prevalence of HHV antigens in RE patients did not necessarily mean that HHVs were pathogenic factors. In the current work, by comparing viral antigens in RE, TLE and TBI samples, we found that HHV antigens were dominant in RE and TLE brains but not in TBI cases. On the one hand, the presence of HHVs in both RE and TLE samples indicated that HHVs were not specifically associated with RE, so that HHV infection would not directly lead to RE. On the other hand, the detection of HHVs in RE and TLE were much more common than in TBI brains so that their presence in some situations was likely to trigger the pathogeneses of brain disorders such as RE and TLE.
Notably, the observation of herpes viruses in RE patients’ brains does not mean the identification of a specific pathogen for the development of RE. In our research, more than one type of herpes virus was found in RE brain, and they were all able to trigger an abnormal immune response in RE patients. Based on these finding, we hypothesized that RE was caused by an imbalanced immune response triggered by some pathogen infections, rather than a specific pathogen. Moreover, it cannot be excluded that viruses other than herpes virus may perform similar effect as herpes virus although we had no direct evidence yet.
The mechanisms underlying the prevalence of herpes viruses in brains of RE patients are unknown. Some herpes viruses usually cause infection in unilateral body such as cold sores, genital herpes and herpes zoster caused by HSV-1, HSV2 and VZV infections, respectively. In general, after the primary infection, herpes viruses are transported along the nerve fibers into brain and persist there. When resistance of the host is low, the latent herpes viruses are activated and cause infection. The repeated latency and activation may link to RE pathogenesis that often occurs in unihemisphere. Although most herpes viruses are neurotropism virus and prevalent in population, their presence in health people is much lower than RE patients and TLE patients. The abnormalities in immune response, somatic mutations, somatic mosaicism, and the up regulation of regulators of T-cell function in chronic viral infections might be the possible factors for the susceptibility of RE patients to herpes virus infection. The above questions deserved to further investigation in the future.
Besides HHV infection, we also investigated the infiltration of CD8+ T cells in the brain. Although their role in the development of RE is supported by a number of studies [
31] as well as our own results, the antigen triggering the CD8+ T-cell response is still unknown. According to our results, both RE and TLE patients had CD8+ T cell infiltration in the brain, indicating that CD8+ T cells were likely to be recruited by HHV infection. Moreover, our study showed that neurons were the dominant cell types infected by HHVs, the infiltrated CD8+ T cells in the RE brain were often closely adjacent to HHV antigen-positive neuronal cells and neurons were co-stained with anti-GZMB antibody. These results suggest that CD8+ T cells primarily targeted HHV-infected neurons and that the infiltration of CD8+ T cells was the consequence of HHV infection. It has been demonstrated that HSV-1-specific memory CD8+ T cells are selectively activated and retained in sensory ganglia to block HSV-1 reactivation from latency [
32,
33]. As latent HHV infection is difficult to eliminate, the CD8+ T cell-mediated immune response will coexist with the latent infection for a long time and thus lead to chronic but persistent damage to the brain, which might directly promote the occurrence and progression of RE. Furthermore, because all RE patients were infected with more than one type of HHV, the CD8+ T cells in the brain were possibly a divergent population that further deteriorated the immune-mediated injury. It must also be mentioned that a substantial number of CD8+ T cells appeared in the TLE brain, but their status seemed to be different. The coexistence of HHVs and quiescent CD8+ T cells has been reported in sensory ganglia, where HHV-specific CD8+ T cells are retained to prohibit HHV reactivation from latency [
33]. Based on this observation, a possible explanation was that CD8+ T cells in RE cases were activated to kill the infected cells, while those in TLE cases were just monitoring their target cells. Notably, this is just a hypothesis and needs more data to support in the future.
The activation of CD8+ T cells in RE but not TLE cases was further supported by the cytokine analysis. During peripheral infection in ganglia, the CXCL10/CXCR3 signaling pathway is indispensable for recruitment of HHV-specific CD8+ T cells [
29] and IL-15 plays an essential role in selectively activating these cells [
30]. It has been demonstrated that reactivation of HSV-1 by UV-B significantly increases both the number and function of HSV-specific memory CXCR3+ CD8+ T cells in trigeminal ganglia latently infected with HSV-1. Moreover, the delivery of exogenous CXCL10 boosts the number and function of HSV-1 specific CD8+ T cells and improves the protection against recurrent herpesvirus infection and disease in mice [
29]. In another experiment, incubation with IL-15 induced CD38/HLA-DR expression in HCMV- and EBV-specific CD8+ T cells while other inflammatory cytokines such as IL-2, IL-7, IFN-γ, IFN-α and TNF-α did not activate EBV and HCMV specific CD8+ T cells [
34]. In our study, CXCL10 and IL-15 in brain tissues were the most significantly upregulated in RE than in TLE cases, consistent with the results that the infiltration of CD8+ T cells was prevalent in RE cases. All these results provided more evidence supporting the hypothesis that CD8+ T cells were activated in RE patients.
What caught our attention was the different clinical outcome between RE and TLE cases despite the large number of HHVs similarly detected in them. In subsequent experiments, we compared the innate immune response against HHVs between RE and TLE cases. During the infection process, HHV DNA can be recognized by DNA sensors TLR3/9, STING and IFI16, which are present in the endosome, cytoplasm and nucleus, respectively. STING and IFI16 activation has been shown to stimulate the production of type I interferon [
14]. Moreover, IFI16 is also crucial for the latent infection of HHVs. Megan Orzalli reported that HSV-1 infection is sensed by nuclear IFI16 [
35], and Gina further proved that IFI16 is required for the maintenance of EBV latency in vitro [
36]. In our study, lower STING and IFI16 protein levels were observed in RE than in TLE cases, which consequently led to the insufficient production of IFN-β. IFN-β is a crucial cytokine of innate immunity and plays a key role in antiviral immunity. The insufficient IFN-β production would undoubtedly impair the defense against HHV infection, and make the HHVs easier to reactivate. This could explain why HHV antigens and genomes were slightly higher in RE samples. The mechanisms leading to the reduced expression of STING and IFI16 are not clear. An in-depth investigation at the level of biochemistry and genetics is required in the future.
Conclusion
Based on these results, we propose the following model to explain the pathogenesis of RE. In people who suffer latent HHV infection in the CNS, innate immunity (IFN-β) and adaptive immunity (CD8+ T cells), collaborate to restrict the reactivation of HHVs. The balance between immunity and HHV latency cannot protect CNS in patients with functional disorders such as TLE, but does avoid the neuronal damage and loss. In RE cases, however, IFN-β production is not sufficient because of the lack of STING and IFI16 activation for some unknown reasons. Relatively lower lever of IFN-β production impairs the defense against HHVs and makes them easier to reactivate. As a result, the brain tissue increased the production of CXCL10 and IL-15 and consequently activated the HHV-specific CD8+ T cells. HHVs usually cause infection in the unilateral body, which may be linked to that RE pathogenesis also often occurs in unihemisphere. The activation of CD8+ T cells compensates for the insufficiently activated IFN-β and controls the spread of HHVs by inducing apoptosis in infected cells; however, it inevitably results in the loss of large number of neurons and finally leads to chronic unihemispheric brain atrophy. Although the proposed model explains many features of RE, it still needs further verification by in vitro studies and animal experiments. Moreover, the mechanisms leading to the insufficient expression of STING and IFI16 also need further investigation. Genetic variation of the host or mutation of the viruses are potential factors for the alternative signaling activation in RE. Although the detailed investigation into these possibilities is beyond the scope of this work, we acknowledge that they are important to ultimately understand the occurrence of RE, and efforts are currently underway in our lab to identify such factors.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.