LCMV infection induces T-cell mediated immunopathology

Following LCMV infection, we analysed the infected status of the brain by immunohistochemistry and could identify the meninges and the ependymal cells of the choroid plexus and the ventricles as the primary sites of LCMV replication with little to no infection of the brain parenchyma that also have been observed by others (Figure 13) (Kang and McGavern 2008).

LCMV represents a non-cytophathic virus and the outcome of a severe form of encephalomyelitis is the immunopathologic result of infiltrating, cytotoxic CD8⁺ T cells

Immunoproteasome assembly in the brain of LCMV-infected mice Discussion

(Baenziger et al. 1986) (Dixon et al. 1987). In accordance, we observed a strong infiltration of the meninges, and the CSF-containing ventricles with CD3⁺ T-cells following intracranial LCMV infection (Figure 14).

4.4 (Immuno)proteasome expression in the brain

In spite of the tremendous progress made in proteasome research, the characterization of proteasome expression, proteasome activity and proteasome function in the central nervous system is poorly defined.

Constitutive proteasomes are ubiquitously, although not homogenously, expressed in the CNS: In the naïve rat brain most 20S proteasome is expressed in the nuclei, whereas pyramidal cortical neurons and motor neurons in the ventral horn of the spinal cord also showed cytosolic expression (Mengual et al. 1996) (Okada et al. 1991). Similar findings were obtained for yeast, where 80% of the total proteasome was expressed in the nucleus (Russell et al. 1999). In humans, proteasome expression is distributed in the cytoplasm, nuclei, dentrites, axons and the synaptic buttons of a variety of cells including both neuronal and glial cell populations (Ding and Keller 2001). Furthermore, a down-regulation in the proteasome expression and activity has been shown to be associated with many neurodegenerative diseases such as Alzheimer´s disease, Parkinson´s disease or Huntington´s disease (Piccinini 2008).

Concerning the expression of immunoproteasome, a low basal expression has been reported in the human brain suggesting that both types of proteasome are existing in the human brain under non-inflammatory conditions (Piccinini et al. 2003). An up-regulation further could be observed during neuronal differentiation, age-progression and neurologicqal disorders such as Huntington´s disease (Hernandez et al. 2003; Diaz-Hernandez et al. 2004) suggesting the involvement of chronic inflammation in normal age progression as well as in neurological disorders.

Similar to humans, a low basal expression has also been reported in the central nervous system of mice (Ferrington et al. 2008). We could not see this basal expression of immunoproteasome in the naïve mouse brain analysed by western blot analysis, 2d-gel analysis, and immunohistology. However, following real time PCR analysis, we obtained CT (cycle threshold) values of approximately 23 for MECL-1, 27 for LMP2

Immunoproteasome assembly in the brain of LCMV-infected mice Discussion

H20 control, indicating that there is some basal expression of immunoproteasome in the uninflamed brain. This expression, however, seems to be quite low, or at least too low to be measured with the herein applied techniques.

To get a deeper insight into immunoproteasome expression in the inflamed brain we established an immunohistochemical protocol for the detection of immunoproteasome-producing cells using antibodies directed against the outer regions of mMECL-1 and mLMP2. These antibodies were a kind gift from Benôit van den Eynde, Brusseles, Belgium, and they turned out to be excellently suited for the histological detection of immunoproteasome in brain cryosections, probably due to their binding at the outer, accessible regions of the proteasome. Moreover, they also were able to detect the respective pre-cursors pre-MECL-1 and pre-LMP2. We obtained specific stainings for the respective subunits in brain cryosections of intracranially LCMV-infected mice that were absent in naïve mice and in the sections of the respective knockout mice underlining the reliability of the performed stainings (Figure 15 + 16). Such important negative controls have not been applied in earlier immunohistochemical studies of immunoproteasome expression (Frentzel et al. 1993) (Ferrington et al. 2008) (Egerer et al. 2006) strongly questioning the validity of these studies.

Co-stainings with cell markers specific for neurons (NeuN) and oligodendrocytes (CNPase) showed no to low co-localization with the respective immunoproteasomal subunit (Figure 28 +29) indicating that these cells do no not seem to express immunoproteasome. Using Iba-1 as a marker specific for microglia cells, we could identify these cells as the main producers of immunoproteasome in the inflamed brain with strong immunoproteasome expression in the cytosol and the nucleus (Figure 21 + 22). This confirms previous results by Stohwasser et al., who could induce immunoproteasome expression in isolated microglia and a microglia BV-2 cell line upon stimulation with IFN-γ (Stohwasser et al. 2000).

Microglia cells are generally considered to be immunologically quiescient under non-pathological conditions as indicated by their ramified morphology, marginal expression of MHC class I molecules and the complete absence of MHC class II surface expression. Following activation by viral inflammation for example, they transform into highly active effector cells as indicated by their contracted morphology (Figure 20) and their proliferation (Figure 18 +19). Due to an inflammatory increase in

Immunoproteasome assembly in the brain of LCMV-infected mice Discussion

MHC class I and MHC class II surface expression, as well as the up-regulation of co-stimulatory molecules such as B7.1 (CD80), B7.2 (CD86) and ICAM-1, most of the literature considers them to be the most-efficient antigen-presenting cell population in the brain parenchyma (Frei et al. 1987) (Gehrmann et al. 1995). Our finding that following a viral infection these cells adapt their proteasomal composition to the requirements of an optimised MHC class I epitope processing by the expression of immunoproteasome strongly supports this hypothesis and suggests that these cells participate in antigen presentation. The antigen-presentation properties of dendritic cells as well as macrophages are well described (Guermonprez et al. 2002), but how far CNS-associated DCs and macrophages contribute to brain inflammation following a viral infection still needs to be elucidated.

The ventricular ependymal cells including the choroid plexus also showed a positive staining for immunoproteasome, but their potential involvement in antigen-presentation and/or (re-)activation of T-cells needs further investigation.

In accordance with our findings, Ferrington and colleagues could also show an up-regulation of immunoproteasome in the inflamed brain using a model of CTL-induced brain injury (Ferrington et al. 2008). However, they postulate that neurons, astrocytes and oligodendrocytes produce a substantial portion of the immunoproteasome in the brain. We could hardly see any co-localization of immunoproteasomal subunits and neurons (Figure 28) and no co-localization at all with oligodendrocytes (Figure 29), pointing more towards a non-priming function of these cells. Ferrington and collegues assessed the expression of LMP7 and a different expression of LMP7 compared to other subunits cannot completely be excluded, but a more plausible reason for this contradictionary finding maybe based on the different models used to induce inflammation. We induced inflammation by a viral infection, whereas Ferrington et al.

used CTL-mediated injury of β−gal-expressing astrocytes as a model system of inflammation and concerning the applied cell marker molecules we also used a different set.

The antigen presentation capacity of astrocytes, the major glial cell population, remains in question and is heavily discussed. Although increased astrocytic cell numbers are associated with many neurodegenerative diseases and could be shown to present MBP to encephalolitogenic T-cells (Fontana et al. 1984), their presentation capacity remains debatable. Sedgwick and collegues could show that astrocytes can

Immunoproteasome assembly in the brain of LCMV-infected mice Discussion

activate T-cells only if these cells were incubated in the presence of microglia or IL-1 indicating that an additional antigen presenting cell is required (Sedgwick et al. 1991). Analysing the immunoproteasome expression in GFAP⁺ astrocytes, we hardly observe any cytosolic co-localization of MECL-1 or LMP2 in astrocytes (Figure 25).

Based on these findings, we postulate that in the LCMV-infected brain the function of antigen processing and presentation is mainly fulfilled by microglia cells, which represent the most dominant immunoproteasome-expressing cell population in the brain. An additional role of CNS-associated dendritic cells and macrophages was not analysed but cannot be excluded and seems to be quite likely.

Neurons and oligodendrocytes do not seem to play a role for antigen presentation in the context of LCMV infection, and astrocytes are more likely to be involved in shaping the immune response towards inflammation and/ or recovery.