Autoimmune encephalitis in humans: how closely does it reflect multiple sclerosis ?

For detailed pathological analysis of the HAE case, one hemispheric paraffin embedded block through the mid thalamus was available (Fig. 1). Hemispheric sections were stained with hematoxylin eosin, Luxol fast blue myelin stain, Bielschowsky silver impregnation for axons, and with the modified Turnbull reaction for iron. Immunocytochemistry was in part performed on the entire hemispheric section or, alternatively, in dissected hemispheric sections, containing the lesions, the normal appearing white matter and the cerebral cortex.

Immunocytochemistry was performed on de-paraffinized sections using the primary antibodies listed in Table 1. This table also contains information on methods of antigen retrieval. Visualization of bound primary antibodies was done with a biotin avidin technique ([13], Table 1). In situ hybridization for proteolipid protein mRNA was performed according to Breitschopf et al. [8] and TUNEL staining according to Gold et al. [19]. To determine the proliferation rate of T-cells and B-cells, the sections were first stained for proliferating cell nuclear antigen (PCNA) using an alkaline phosphatase technique followed by the respective markers CD3 and CD20, visualized with a biotin-avidin-peroxidase method. In situ hybridization for Epstein Barr Virus (EBER) was performed using the EBER pNA detection kit (DAKO Y5200), including an EBV⁺ cerebral lymphoma as a positive control.

Quantitative analysis of axonal density was performed in the microscope at a magnification of 20x in sections stained for phosphorylated neurofilaments. The number of axons, crossing a line of 2 mm were determined by manual counting at three sites of the lesions and the normal appearing white matter. A similar approach was also used to determine the number of axons with disturbance of axonal transport, reflected by the positivity for amyloid precursor protein, and for the presence of axons with terminal end bulbs.

Analysis of the entire brain hemispheric sections showed large periventricular inflammatory demyelinating lesions with partial preservation of axons (Fig. 1). Peri-venous finger like extensions of the lesions into the adjacent white matter were prominent (Dawson Fingers, [11]). In addition, numerous small peri-venous areas of demyelination were seen in the entire cerebral white matter. Demyelination was also present in the cortex and in deep grey matter nuclei as previously described in MS. [30] In the cortex small intra-cortical and larger subpial lesions were visible. Grey matter demyelination was particularly pronounced in the hippocampus (Fig. 1a and b). The vast majority of the lesions showed active demyelination at the edges, characterized by a dense rim of activated macrophages and the presence of early myelin degradation products within macrophages and microglia Fig. 1c, d, Fig. 2a). Demyelination was associated with complete loss of oligodendrocytes (Fig. 1d) and remyelination was absent. TUNEL staining revealed some cells with DNA fragmentation with a morphological appearance of necrosis at the edge and in the center of the lesions. Nuclear accumulation of apoptosis inducing factor, a hallmark of cell degeneration in Pattern III lesions of multiple sclerosis [56] was not found. Profound microglia activation was also seen in the peri-plaque white matter (Fig. 2a) and clusters of activated microglia (so called microglia nodules, [41]) were abundant (Fig. 2b). Staining for neurofilament revealed only a moderate reduction of axons within the demyelinated lesions (Figs. 1e, 2e) in comparison to the adjacent normal appearing white matter (Fig. 2d). The reduction of axonal density was 23 % at the lesion edges and 34 % in the lesion center. However, axonal spheroids and end bulbs (Fig. 2e) reactive for amyloid precursor protein (APP, [13]) were numerous in particular at the active edge of the lesions (Fig. 2f). At the lesion edge 32 % of the axons showed accumulation of APP, but only 12 % revealed the presence of axonal end bulbs as an indicator for axonal transection. Highly activated “protoplasmic” astrocytes were dispersed at sites of active demyelination and in the peri-plaque white matter (Fig. 2g,h), and some of them contained multiple nuclei or nuclear fragments (Creutzfeldt Peters cells; Fig. 2h insert). Fibrillary astrocytic gliosis was seen in the inactive lesion center (Fig. 2i). Active demyelination was associated with deposition of activated complement (C9neo antigen; Fig. 2c) in all lesions with initial stages of myelin destruction. This was associated with dressing of myelinated fibers by immunoglobulin (IgM; Fig. 3l).

Actively demyelinating lesions were also seen in the cortex (Fig. 1a). Fig. 1g shows an active subpial lesion in the insular cortex. Active myelin destruction was associated with activated microglia (Fig. 1h), containing myelin degradation products (Fig. 1l) and expressing the phagocytosis associated marker CD68 (Fig. 1i) and NADPH oxidase (Fig. 1m). Active cortical demyelination was topographically related to meningeal inflammation, consistent of CD3 positive T-cells (Fig. 1j) and CD20 positive B-cells (Fig. 1k). However, no deposition of activated complement was seen in active cortical lesions at the site of active demyelination, as described before in MS. [9].

Overall these data show, that the lesions in HAE reflect all pathological hallmarks seen in classical MS, following a pattern of demyelination, which is associated with antibody and complement deposition (Pattern II, [36]). Pattern III demyelination, defined by loss of myelin associated glycoprotein, distal oligodendrogliopathy, oligodendrocyte apoptosis and a concentric type of demyelination [1] were not encountered in this patient.

The demyelinating pathology, described above occurred on a background of a very severe inflammatory reaction (Fig. 3). Numerous very large inflammatory infiltrates were present within the lesions, while inflammation was sparse in the normal appearing white matter and virtually absent in the cortex. However, inflammation was present in the leptomeninges in close proximity to cortical demyelinated lesions (Fig. 1a, j, k). The inflammatory infiltrates contained high numbers of CD3⁺ T-cells (Fig. 3 b) and a very similar number of CD8⁺ cells (Fig. 3d). In contrast, CD4⁺ T-cells were exceptionally rare. As an example, in the very large inflammatory infiltrate, depicted in Fig. 3c, only one single CD4⁺ T-cell was found (asterisk and insert). By far, however, the most numerous components of the inflammatory infiltrates were CD20⁺ B-cells, which outnumbered T-cells by a factor of 3 to 4 (Fig. 3e). In addition, CD138 positive plasma cells were seen in separated regions of the infiltrates (Fig. 3f), which mainly contained IgM (Fig. 3i) but only occasionally IgG (Fig. 3g) or IgA (Fig. 3h). The large inflammatory infiltrates in the perivascular spaces and meninges showed clear separation of T-cell, B-cell and plasma cell dominated regions in a pattern, similar to that described in lymph-follicle like inflammatory aggregates in MS. [47] Diffuse inflammatory infiltrates within the lesion parenchyma mainly consisted of CD8⁺ T-cells (Fig. 3d) and IgM⁺ plasma cells (Fig. 3j). At the lesions edges, dressing of myelin sheaths with IgM was noted (Fig. 3k). Proliferation of leukocytes was determined by the expression of the proliferation marker PCNA in T-cells and B-cells. Double stained cells were CD8⁺ cells in perivascular cuffs and in diffuse infiltrates in the parenchyma (Fig. 3l, m) and CD20⁺ perivascular cells (Fig. 3n). Quantitative data on the percentage of different leukocyte subsets in perivascular cuffs are provided in Table 2.

The unusually high B-cell component within the lesions may indicate Epstein Barr Virus infection of these cells, as suggested before to occur in MS lesions [48]. To address this question we performed in situ hybridization for EBV/EBER using an autopsy case with a cerebral EBV related lymphoma as a positive control, which showed prominent EBV EBER signal in the nucleus of malignant B-cells (Fig. 4k). In contrast, no single EBER positive cell was found in the HAE case (Fig. 4l). It is unlikely that the absence of EBER reactivity within our case of HAE is due to technical reasons, since EBER in situ hybridization provides consistent results in human autopsy brain tissue [32] and the quality of tissue preservation, including mRNA preservation, in our case of HAE was better compared to that seen in other routine autopsy tissues (see in situ hybridization for proteolipid protein mRNA shown above).

Oxidative injury has recently been identified as a major tissue alteration associated with demyelination and neurodegeneration in MS lesions [14] and it was suggested that oxidative injury may be amplified in the presence of iron in the aging human brain and its liberation in active MS lesions [22]. In line with the age of the patient profound iron accumulation was found in the brain (Fig. 1f), which was most prominent in the basal ganglia and the substantia nigra, followed by the cortico/subcortical border zone. Iron reactivity was mainly seen in myelin and oligodendrocytes as described before ([20]; Fig. 4a, c). Active areas of the white matter lesions contained a low iron load, but iron accumulated in macrophages in the lesion center (Fig. 4a). In contrast, massive iron load was present in lesions, which affected the basal ganglia (Fig. 4b). As described before in MS, iron was lost from myelin and oligodendrocytes and accumulated in macrophages and microglia in these lesions (Fig. 4d). In addition, granular iron reactivity was present in the extracellular space, as described before in MS lesions [22]. Oxidative injury, reflected by the cytoplasmic accumulation of oxidized lipids in white matter lesions and the cortex was low in comparison to that seen in acute MS cases with Pattern III pathology or in the lesions of patients with progressive MS. [14, 21] However, in the lesions affecting areas with high iron content in the basal ganglia accumulation of oxidized phospholipids was seen in neurons, dystrophic axons and glia (Fig. 4f-j).

Consistent with the clinical presentation, moderate changes of Parkinson’s disease were seen in the substantia nigra, characterized by a mild reduction of pigmented cells and the presence of few cells with Lewy bodies, reactive for α-synuclein (Fig. 1e, insert). No Lewy bodies were seen in the hippocampus or cortex (Braak&Braak stage III of Parkinson´s disease; brainstem predominant Lewy body type pathology according to McKeith criteria). Despite the presence of neurodegeneration the substantia nigra was not targeted by the inflammatory or demyelinating process.

Similarities and differences between the case of HAE and MS or animals models of EAE are summarized in Table 3. Most pathological hallmarks, considered to be specific for MS, were shared between HAE and MS. One major difference was that none of the lesions showed a demyelinating pattern, characterized by distal oligodendrogliopathy and oligodendroglia apoptosis (pattern III; [34]). This pattern of demyelination seems to represent an extreme example of tissue injury mediated by oxidative and mitochondrial injury [14, 38] and its absence in this case is in line with the relatively mild oxidative injury, which was largely restricted to areas with high iron content and liberation into the lesions.

A major quantitative difference was also seen in the composition of inflammatory infiltrates. In the case described here the infiltrates largely contained CD20⁺ B-cells and CD8⁺ T-cells, while CD4⁺ lymphocytes were virtually absent. In addition plasma cells were found, which dominantly contained IgM. In MS inflammation is dominated by CD8⁺ T-cells and the number of CD20⁺ B-cells is much lower [16]. Furthermore, plasma cells in MS lesions mainly produce IgG and in much lower incidence IgA and IgM [12]. Also in MS CD4⁺ T-cells are sparse.

For the comparison with animal models we focused on the two extreme examples of the EAE spectrum, MOG peptide induced chronic EAE in C57Bl6 mice, which is a purely CD4⁺ T-cell mediated disease, which is most easy to induce and most reproducible, and is, thus, preferentially used in immunological studies of MS. [18] The other is the chronic EAE model induced by recombinant MOG in incomplete Freund’s adjuvant in marmoset monkeys, which appears to be most close regarding genetic background and method of sensitization to the case described in this study, but is heterogeneous between animals regarding clinical disease and severity of the pathological alterations [27]. The pathology of marmoset EAE revealed an inflammatory demyelinating disease in many essential criteria similar to MS and HAE, following a pattern of demyelination, that suggest involvement of demyelinating antibodies (Pattern II; [36]). There were, however some differences in astrocyte and microglia reaction and the inflammatory reaction in marmosets is dominated by T-cells, while the B-cell component is less pronounced and IgM producing plasma cells are sparse. Pathological alterations, related to iron deposition in the central nervous system and to oxidative injury were sparse or absent (Table 3).

In contrast to EAE in humans and primates, the lesions in chronic mouse models after active sensitization with MOG_35–55, are fundamentally different in many respects. Primary demyelination is sparse and larger focal lesions are mainly due to inflammatory neurodegeneration with secondary demyelination. Inflammation is dominated by CD4⁺ T-cells and the contribution of other leukocyte subsets is sparse. Furthermore, macrophage and microglia reaction are profoundly different between this model and the human condition [44]. Finally, the extensive oxidative injury seen in the brain of patients with progressive MS, which is associated with slow expansion of pre-existing lesions and diffuse damage of the grey and white matter was neither reproduced in marmoset or mouse EAE models, nor in our case of HAE.

However, comparing EAE models with MS one has to consider that numerous different EAE models have been described, and the pathological alterations in the brain and spinal cord cover a broad spectrum, which resides between the two extremes, elaborated above in this study [34, 40, 42, 49, 52].