Apoptosis of Oligodendrocytes via Fas and TNF-R1 Is a Key Event in the Induction of Experimental Autoimmune Encephalomyelitis¹

Multiple sclerosis (MS)⁶ is a human autoimmune disease characterized by accumulation of lymphocytes and macrophages in the CNS (1) leading to demyelination, destruction of axons (2) and paralysis. Experimental autoimmune encephalomyelitis (EAE) is a T cell-mediated disease that is used as a model for the study of MS. In EAE, cells secreting Th1 cytokines predominate and mediate inflammatory damage while cells secreting Th2 cytokines have been associated with remission and recovery from disease (for review, see Ref.3). Immunization with myelin Ags or adoptive transfer of encephalitogenic CD4⁺ T cells (4) can induce EAE. T cells infiltrating into the CNS mediate an inflammatory process, resulting in demyelination and death of resident ODCs (5).

Considerable evidence has identified an important role for the apoptosis mediating ligand/receptor pairs Fas ligand (FasL)/Fas and TNF α (TNF-α)/TNF-R1 in EAE. These molecules are members of the TNF and TNF-R superfamilies, respectively (6). Triggering of Fas and TNF-R1 activates overlapping signal transduction pathways including Fas-associated death domain protein and caspase-8, leading to the induction of target cell apoptosis (7).

In EAE experiments, mice mutant for TNF-α (8) or FasL (gld mice) showed a significant delay in disease onset and a remarkable reduction in demyelination (9, 10, 11). Adoptive transfer of FasL-deficient encephalitogenic T cells to wild-type (wt) mice resulted in partial reduction of EAE severity compared with the transfer of wt encephalitogenic T cells, proving that FasL expression by T lymphocytes is important for EAE pathogenesis (9, 12).

Similarly, mice deficient for the receptors Fas or TNF-R1 were shown to be partially resistant to EAE induction by immunization with myelin Ags (12, 13, 14). Adoptive transfer of myelin ODC glycoprotein (MOG)-specific wt T cells into mice harboring a natural mutation in the fas gene (lpr mice) (11) or a mutation in the tnf-r1 gene (15) showed that the presence of Fas and TNF-R1 on non-T cells is required for full development of EAE. Hence, encephalitogenic T cells activating TNF-R1 and Fas signal transduction are crucial for EAE pathogenesis.

To conclusively investigate the role of Fas in ODC destruction during the course of EAE, we generated a mouse strain, MOGi-cre/fas^fl/fl, that lacks Fas expression exclusively in ODCs by the use of homologous (16) and Cre-mediated recombination (17), expressing the Cre recombinase under the control of the ODC-specific MOG promoter. The ODC-specific deletion of a loxP-flanked fas allele resulted in significant protection of the mutant mice from the induction of EAE. To investigate whether the remaining disease is mediated by TNF-R1, we analyzed mice lacking additionally the tnf-r1 gene. The mice deficient for TNF-R1 and Fas expression in ODCs were almost completely resistant to EAE induction. ODC demyelination and apoptosis was virtually absent, although a degree of lymphocyte infiltration into the CNS persisted. Thus, we show that apoptosis via Fas expressed on ODC is an important mechanism for EAE induction.

The open reading frame of Cre was introduced into the first exon of the mog gene followed by the polyadenylation signal of the human growth hormone. The neomycin resistance (neo) and thymidine kinase genes were used as positive and negative selection markers, respectively. C57BL/6-derived embryonic stem (ES) cells (18) were transfected with the targeting vector according to standard protocols (19). ES cells carrying the insertion were injected into CB20 blastocysts to generate the MOGi-cre mouse strain. Intercrossing of this mouse with a mouse strain carrying a loxP-flanked exon 9 of the fas gene (20) resulted in offspring harboring ODC-specific Fas deletion. To delete both Fas and TNF-R1 receptors, ODC-specific Fas-deficient mice were further bred to TNF-R1-deficient mice that were kindly provided by Dr. K. Pfeffer (21).

Myelin was prepared from spinal cord or brain as described previously (22). Equal amounts of myelin were loaded on a 15% SDS-polyacrylamide gel and transferred to a Protran BA 83 nitrocellulose membrane (Schleicher & Schuell BioScience). Blots were incubated with rabbit antiserum against the Ig-like extracellular domain of mouse MOG (1/1000) or anti-cyclic nucleotide phosphodiesterase (CNPase) mAb (1/3000) (Chemicon) for 2 h, followed by corresponding secondary peroxidase-conjugated anti-IgG (mouse and rabbit) (Amersham Biosciences) and developed according to the ECL protocol (Amersham).

Ab conjugates specific to the following Ags were purchased from BD Pharmingen: B220-cychrome; CD4-FITC; CD8-PE; CD90.2-PE; and TCRαβ-PE. Cells from lymphoid organs were stained with the Ab conjugates for flow cytometric analysis on a FACSCalibur (BD Biosciences). The data were analyzed using CellQuest software (BD Biosciences).

EAE was induced in 8- to 12-wk-old mice by injecting 50 μg of MOG p_35–55 peptide (Neosystem; Ref.23) supplemented with 8 mg/ml heat-inactivated Mycobacterium tuberculosis (H37Ra strain; Difco Laboratories) in CFA (Difco Laboratories) on day 0 in the tail base. 200 ng of Bordetella pertussis toxin (List Biologicals) were administered i.p. on the day of immunization and 2 days later. Mice were monitored daily for clinical signs of EAE, graded on a scale from 0 to 5 as follows: 0 = clinically normal; 0.5 = limp tail or isolated weakness of gait; 1 = loss of tail tonicity; 2 = hind leg paralysis; 3 = hind and front leg paralysis; 4 = total hind and front leg paralysis and partial body paralysis; 5 = moribund or dead animals. Moribund animals were sacrificed.

To assess the specificity of the Cre recombinase activity in the MOGi-cre strain, these mice were crossed to the indicator mice ROSA26-LacZ (24) or to the Z/EG mouse strain (25).

For cryosectioning, perfused whole brains of 44-day-old MOGi-cre/Z/EG mice were sectioned at 30 μm. Sections were washed three times in 0.1 M phosphate buffer, pH 7.4, for 15 min before two washings in TBS (50 mM Tris (pH 7.4), 150 mM NaCl) for 15 min. After blocking for 1 h at room temperature in blocking solution (10% normal horse serum, 0.3% carrageenin, 0.5% Triton X-100 in TBS), sections were incubated with primary Abs in 1% normal horse serum, 0.3% carrageenin, 0.5% Triton X-100 in TBS for 48 h at 4°C. Sections were washed three times for 10 min each in TBS and subsequently incubated with secondary Abs in TBS overnight at 4°C. After a washing with TBS for 10, 15, and 30 min, sections were placed on slides and allowed to dry before mounting with Vectashield mounting medium with 4′,6′-diamidino-2-phenylindole.

Abs used include: primary Abs against CNPase (monoclonal, 1/100; Chemicon), S100 (polyclonal, 1:200; Sigma-Aldrich), and neuron-specific nuclear protein (NeuN; monoclonal, 1/100; Chemicon); fluorescent secondary Abs: Alexa Fluor 594 goat anti-mouse IgG (1/1000; Molecular Probes); and Texas Red AffiniPure goat anti-rabbit IgG (1/200; Jackson ImmunoResearch).

Sections were analyzed using a Leica NTS confocal microscope (FITC channel to detect native enhanced green fluorescent protein (EGFP) signal and tetramethylrhodamine isothiocyanate channel to detect immunofluorescence).

To analyze the primary T cell response to the MOG peptide, inguinal and popliteal lymph nodes (LN) were harvested from mice 10–12 days after immunization with MOG peptide-CFA (as described above, but without the pertussis toxin injection). A single-cell suspension was prepared, and 2.5 × 10⁴ lymphocytes were incubated with 2 × 10⁵ irradiated wt splenocytes in the presence of MOG p_35–55 for 72 h. Cell proliferation was assessed in triplicate cultures by adding 0.5 μCi/well [³H]thymidine (DuPont). After 18 h, cells were harvested with a Harvester 96 (Tomtec), and [³H]thymidine incorporation was determined using a Matrix 9600 direct beta counter (Packard BioScience). Cytokines were measured by ELISA as described previously (26).

Experimental mice were perfused with PBS-heparin followed by 4% paraformaldehyde (in PBS, pH 7.4). Paraffin-embedded sections from brains and spinal cords were stained with HE, with Luxol fast blue, or with Bielschowsky’s silver stain. Immunohistochemistry was performed as previously described (27). Primary Abs against the following targets were used: T lymphocytes (anti-CD3 from Serotec and anti-CD90 from BD Pharmingen); macrophages (anti-Mac-1 and anti-Mac-3; BD Pharmingen) and ODC (anti-CNPase and SMI-91; Sternberger Monoclonals, Lutherville, MD). The extent of inflammation was quantified by counting the perivascular inflammatory infiltrates in 10 randomly selected sections of thoracic, cervical, and lumbar spinal cord. The size of demyelinated lesions were determined by overlaying the spinal cord sections with a morphometric grid and counting the area of demyelination in relation to the total area of the spinal cord.

Apoptosis was determined by morphological criteria such as nuclear condensation and nuclear fragmentation. For evaluation of ODC apoptosis, CNPase-positive ODCs were counted in six inflamed and demyelinated areas of 66 mm², and the average number was determined. The mean values were multiplied by 15 to determine the density of apoptotic ODCs per square millimeter.

In the course of EAE, the death of ODCs was observed and reported to be responsible for demyelination and axonal death (28). The underlying mechanisms of this degeneration remain insufficiently understood and controversial (12, 28, 29, 30). Morphologically, ODCs in MS lesions were shown to exhibit cytoplasmic swelling, organelle disruption, and membrane damage (31, 32). Similarly, apoptotic cells with oligodendroglial features have been described in EAE lesions after immunization with myelin basic protein (MBP) (28, 33). To determine whether apoptosis of ODCs can also be detected in our EAE model, brain sections of C57BL/6 mice immunized with MOG p_35–55 in CFA were stained for the ODC marker CNPase and analyzed for morphological hallmarks of apoptosis (Fig. 1). We could detect all consecutive stages of apoptosis, namely condensation (Fig. 1,b), deformation (Fig. 1,c) and fragmentation of the nuclei in CNPase-positive cells (Fig. 1 d). Thus, different phases of apoptotic death were present among ODCs in the lesions of C57BL/6 mice immunized with MOG. To investigate whether Fas is directly involved in the induction of apoptosis in these cells, we decided to delete the fas gene specifically in ODCs using Cre-mediated recombination and investigated EAE pathology in these mice.

We therefore generated a mouse strain in which the Cre recombinase is expressed specifically in ODCs. Because expression of MOG was shown to be tightly restricted to ODCs in contrast to other myelin proteins, like phospholipoprotein and MBP (34, 35), we inserted the Cre open reading frame into the mog gene of C57BL/6-derived ES cells via homologous recombination (Fig. 2 b, clone 2F5).

Insertion of the Cre recombinase gene under the MOG promoter results in inactivation of the mog gene. As a result, the generated MOGi-cre mice are heterozygously deficient for the mog gene. Because EAE is induced by immunization with MOG peptide and mediated by MOG-specific T cells, different endogenous expression levels of MOG could theoretically alter the development of the autoimmune disease. We therefore tested whether heterogeneity of the mog gene in the MOGi-cre mice influences the expression levels of the MOG protein by Western blot analysis using CNS homogenate. We found that the amount of MOG from the brain expressed in heterozygous MOGi-cre mice was comparable with that in wt controls (Fig. 2,c). As expected, homozygous MOGi-cre mice lacked MOG expression completely (Fig. 2 c).

The MOGi-cre mouse strain was intercrossed with the fas^fl/fl strain in which exon 9 of the fas gene is flanked by loxP sites (20). Because specificity and efficiency of Cre-mediated recombination vary between loxP-flanked target genes, we analyzed the Cre-mediated recombination from DNA of various organs from the MOGi-cre/fas^fl/+ mice by PCR. A 240-bp product corresponding to the deletion of exon 9 was observed solely in DNA taken from brain, thus confirming CNS-specific recombination in this mouse strain (Fig. 2 d).

To test the specificity of Cre-mediated recombination, we crossed the MOGi-cre mouse strain to the Z/EG reporter strain (25), in which upon Cre-recombination expression of EGFP is achieved. Brain sections from young MOGi-cre/Z/EG mice were analyzed for EGFP fluorescence to establish where Cre-mediated recombination occurred. EGFP signal was detected in cells and fiber bundles of the white matter resembling the expected distribution of ODCs (Fig. 3). To rule out the possibility of recombination occurring in neurons, brain sections were stained with NeuN (that reacts with most neuronal cell types throughout the nervous system), and confocal analysis revealed no colocalization of the two signals (Fig. 3, a–c). In contrast, the ODC-specific marker CNPase overlapped with the EGFP signal in fiber bundles and ODC cell somas as shown in Fig. 3, d–i. In addition, glial fibrillary acidic protein, a marker to identify astrocytes, did not colocalize with EGFP-positive cells (data not shown).

Moreover, recombination is occurring in mature ODCs and not in precursors as revealed by S100 immunostainings of MOGi-cre/Z/EG brain sections (Fig. 3 j). The S100 protein is expressed in oligodendroglial progenitors committed to differentiate into OL lineage. In the adult, S100 expression is down-regulated in mature ODCs that have established contacts with their axonal targets (36). Furthermore, the MOGi-cre-expressing mouse strain was crossed to the reporter strain ROSA26 (24), in which β-galactosidase is expressed after Cre-mediated excision of a transcriptional stop cassette. Brain sections from MOGi-cre/ROSA26 mice were analyzed for the expression of β-galactosidase using 5-bromo-4-chloro-3-indolyl β-D-galactoside staining, to determine where in the CNS Cre-mediated recombination occurred. Blue-stained cells were found primarily in the white matter of the brain, resembling the expected distribution of ODCs (data not shown). To rule out Cre activity within lymphoid organs, spleen, thymus, and lymph node cells were stained with fluorodeoxyglucose, a substance that is rendered fluorescent by β-galactosidase, and analyzed by FACS. In this assay, we did not detect Cre activity in lymphoid organs (data not shown).

Thus, in contrast to the previously published MBP-Cre mice (37), the MOGi-cre mouse does not exhibit Cre-mediated recombination activity in any tested cell type other than mature ODCs.

The immune system of fas gene mutant (lpr) mice and FasL mutant (gld) mice suffers from abnormal development and homeostasis of T and B cells. This leads to lymphoproliferative disease and accumulation of T cells expressing the B220 (CD45R) marker (38, 39). To confirm that the immune system in the MOGi-cre/fas^fl/fl mice develops normally, we stained for the expression of B220 on the surface of LN T cells. In contrast to lpr mice, exhibiting the previously described population of B220 expressing T cells (Fig. 4,a) (38), the fraction of TCRαβ⁺B220⁺ cells in LNs of the MOGi-cre/fas^fl/fl was unaltered compared with wt mice (Fig. 4,a). Also in the thymus we found a normal distribution of CD4/CD8 thymocytes in the different mutant mice (Fig. 4 a). Together, these results demonstrate that the MOGi-cre/fas^fl/fl mice do not exhibit defects in T cell development and T cell homeostasis.

The development of EAE depends on pathogenic Th1 type T cells (40), secreting IL-2 and IFN-γ. The levels of these cytokines were shown to correlate with disease severity (41). To exclude defects in T cell polarization toward the Th1 type and cytokine secretion, we tested activation and cytokine secretion of T cells from MOGi-cre/fas^fl/fl and fas^fl/fl control mice after immunization with MOG. Ten days after immunization, LN cells were reactivated in vitro with MOG p_35–55, and the proliferative response as well as cytokine secretion was analyzed. No significant differences in the proliferative response of LN cells taken from MOGi-cre/fas^fl/fl or fas^fl/fl mice were detected (Fig. 4,b). Furthermore, there was no difference in the amount of IL-2 and IFN-γ secreted from wt- and mutant-derived LN cells (Fig. 4 b).

Thus, absence of Fas in ODCs did not affect MOG-specific T cell response in the periphery. It also did not influence the ability of T cells to differentiate into IFN-γ secreting Th1 cells after immunization and restimulation with MOG p_35–55.

To measure EAE susceptibility, we immunized MOGi-cre/fas^fl/fl and fas^fl/fl control mice with the encephalitogenic MOG p_35–55 (23). The mice lacking Fas expression on ODCs show a significant delay of disease onset and attenuation of disease severity in the fas-deficient mice in comparison with control animals (Fig. 5,a and Table I). In contrast, the mice lacking Fas expression on ODCs do not show a decrease in disease incidence compared with control mice (Table I). Because the heterozygous MOGi-cre mice express equal amounts of the MOG protein compared with wt mice in the CNS, it was unlikely that heterozygosity of the MOG locus could influence disease severity. To further rule out possible effects of Cre expression in ODCs on the course of the disease, the EAE experiment was repeated with heterozygous MOGi-cre/+ mice as a third control group (Fig. 5,b). These mice were as susceptible to EAE as the different control mice used in this study. (Fig. 5,b and Table I). This result confirms that the milder disease observed in MOGi-cre/fas^fl/fl mice is in fact due to the lack of Fas expression in ODCs and not to mog heterozygosity or the Cre expression.

Histological analysis was performed to assess whether the reduction of clinical disease severity correlates with reduced demyelination and/or inflammation. Twenty-four days after EAE induction, brains and spinal cords from MOGi-cre/fas^fl/fl and fas^fl/fl control mice were analyzed histologically. Luxol fast blue staining showed a reduction of demyelination by almost 70% in the MOGi-cre/fas^fl/fl mice compared with controls (Fig. 5,c). In addition, we found that mice lacking Fas specifically in ODCs showed reduced histological signs of CNS inflammation such as perivascular and parenchymal tissue infiltration by T lymphocytes and macrophages visualized by H&E staining (Fig. 5 c). Compared with control mice, the infiltrated area of analyzed CNS sections from MOGi-cre/fas^fl/fl mice was reduced by one-third. A similar reduction was detected for both demyelination and inflammation (data not shown). These results indicate that in addition to the direct damage of the myelin, ODC apoptosis in EAE further increases the inflammatory infiltration and thus propagates CNS pathology in EAE.

Apart from Fas-FasL, TNF-R1 was reported to play a role in the induction of ODC apoptosis in the course of EAE (42, 43, 44). It was shown that passive EAE could be induced in wt mice by adoptive transfer of TNF-R1^−/− T cells (15). In contrast, TNF-R1^−/− mice that were adoptively transferred with wt T cells were resistant to EAE induction (15), indicating that TNF-R1 expression by resident cells in the CNS and TNF-α secreted by the invading T lymphocytes and macrophages are critical for the development of disease. To investigate whether the residual ODC apoptosis we detected in MOGi-cre/fas^fl/fl mice during EAE is mediated by activation of TNF-R1, we crossed the MOGi-cre/fas^fl/fl mice to TNF-R1 deficient mice.

Further, we determined whether the additional deletion of TNF-R1 has an impact on T cell development and immunity. We observed normal B and T cell distribution among the lymphocytes in the LNs of the MOGi-cre/fas^fl/fl TNF-R1^−/− mice. In addition, the double-deficient mice did not exhibit the population of T cells expressing the B220 marker found in lpr mice (Fig. 6,a). To assess the role of TNF-α and FasL-mediated apoptosis of ODCs during EAE, we induced EAE in the MOGi-cre/fas^fl/fl TNF-R1^−/− mice by immunization with MOG p_35–35. In mice lacking Fas on ODCs, TNF-R1, or both receptors, the disease onset was delayed (Fig. 6,b and Table II). The mice with a single mutation in either Fas or TNF-R1 showed clinical signs of EAE, although much more mildly than wt mice (Fig. 6,b and results summarized in Table II). However, mice deficient for the expression of both receptors are almost completely resistant to EAE induction (Fig. 6 b). Thus, we conclude that the remaining disease observed in lpr mice or mice with deletion of Fas in ODCs in EAE is a result of TNF-R1 expression in resident CNS cells and that Fas and TNF-R1 are the pivotal mediators in ODC apoptosis in EAE.

To investigate whether the absence of the proapoptotic receptors Fas and TNF-R1 on the surface of ODCs will indeed lead to the reduction of ODC apoptosis during EAE, we immunized the different strains with MOG p_33–35 and analyzed nuclear condensation and fragmentation to identify apoptosis. Mice deficient for Fas in ODCs have a mild reduction of ODC apoptosis after EAE induction, whereas mice deficient for TNF-R1 show only 25% of the ODC apoptosis observed in wt mice (Fig. 6 c). Mice lacking both TNF-R1 expression and Fas expression in ODCs had only occasional apoptotic ODCs in the spinal cord after immunization, indicating that Fas and TNF-R1 are the major receptors mediating ODC apoptosis during EAE.

Although the clinical course of the disease was similar for mice with Fas deficiency on ODC and TNF-R1^−/− mice, respectively, the latter showed a greater reduction in demyelination in comparison to wt mice (Figs. 6,c and 7,h compared with Fig. 7, b and e). The double-deficient mice exhibit even less demyelination, as summarized in Fig. 6,c (see also Fig. 7 k). We found that demyelination mirrors to a large extent the findings of ODC apoptosis in the CNS of the different mice.

Because we showed that in MOG p_35–55 immunized MOGi-cre/fas^fl/fl mice the reduction in demyelination is accompanied by reduced inflammatory infiltrates (Fig. 5), we questioned whether infiltrating lymphocytes in the CNS are further reduced or absent in the double-deficient mice. Histological analysis of CNS tissue from the experimental mice was performed. Mice deficient for TNF-R1 or for Fas in ODCs exhibited reduced levels of inflammation in comparison with wt mice (Fig. 6,c, and exemplified in Fig. 7). In contrast to the difference in the level of demyelination in the two single-mutant mouse strains, they show comparable levels of inflammation, in both cases reduced compared with results in wt mice. Although the double-deficient mice have a merely moderate reduction of inflammation in the CNS compared with the single mutant mice, they exhibit only a very mild clinical course of EAE.

Further, we analyzed the extent of neuronal damage in the CNS of the different mice. Spinal cords of the mice taken 24 days after EAE induction were analyzed for axonal injury by Bielschowsky silver staining. We demonstrate that MOGi-cre/fas^fl/fl TNF-R1^−/− mice show minimal axonal injury (compare Fig. 7,l with Fig. 7 c).

Together, these results demonstrate that in the MOGi-cre/fas^fl/fl TNF-R1^−/− model lymphocytes can invade and populate the CNS, but the lack of apoptosis-mediating receptors on the surface of ODCs is conducive to cell survival and diminished EAE severity.

Previous reports demonstrated a role for the death receptors Fas and TNF-R1 and their ligands in the induction and development of EAE. The disruption of Fas-FasL interaction in gld, as well as lpr mice, partially protected them from EAE pathogenesis (10, 45). To further investigate the role of Fas-FasL interactions in EAE pathogenesis, we generated mice deficient for Fas expression specifically in ODC and tested their susceptibility to ODC apoptosis, demyelination, brain inflammation, and clinical disease after induction of EAE. In addition, we combined the ODC-specific Fas mutation with a mutation in the TNF-R1 gene. We assumed that these two signaling pathways together are responsible for ODC apoptosis and further speculated that this process is critical for disease development. Because the use of gld and lpr mice for the analysis of EAE is hampered by defects in the immune system, leading to lymphoproliferative disease (38), we analyzed actively induced EAE in mice deficient for Fas specifically in ODCs and mice deficient for TNF-R1. As expected, we found no evidence for alterations of the immune system neither in MOGi-cre/fas^fl/fl mice nor in TNF-R1^−/− mice. Similarly, MOGi-cre/fas^fl/fl TNF-R1^−/− double-deficient mice did not display alterations in the immune system. T cell proliferation or IL-2 and IFN-γ secretion by T cells after activation with MOG p_35–55 were also normal. These data confirm that the MOGi-cre/fas^fl/fl and the TNF-R1/Fas double-deficientmice are suitable mouse models to analyze EAE development and pathogenesis.

Upon induction of EAE with MOG p35–55, we found decreased disease severity as well as delayed onset of EAE in ODC-specific Fas-deficient mice. Thus, absence of Fas on ODCs can partially protect mice from EAE. This result compliments published data that showed that absence of Fas (lpr mice) on non-bone marrow-derived cells protects mice from EAE (9). Thus, we have now shown that Fas-FasL interaction between cells of the immune system (presumably MOG-specific T cells) and ODCs plays an important role in the induction of ODC death. Consequently, we strengthened the published data indicating that ODCs are critical target cells of the autoimmune response in the primary step of EAE, because preventing their apoptosis reduces the clinical course of the disease. Although a contribution of other cell types expressing Fas (i.e., neurons and astrocytes) cannot be ruled out completely, our data are supported by the finding that the transfer of wt encephalitogenic T cells into FasL-deficient mice resulted in EAE, proving a main role of T cell-ODC interaction, involving Fas-FasL (12).

Although we have shown that death of ODCs via Fas-FasL interaction is necessary for full induction of EAE, the mice lacking Fas on ODC suffered from a milder form of the disease. In accordance, Sabelko-Downes et al. (9) have previously published that mice lacking Fas expression are still susceptible to EAE, albeit to a lower degree than wt mice. This milder disease therefore had to be mediated by alternative factors. Similar to Fas, TNF-R1 was shown to promote apoptosis of ODCs during EAE (14). Thus, we reasoned that TNF-TNF-R1 interaction is responsible for mediating the remaining disease observed in mice lacking Fas expression on ODCs. We demonstrate that mice with double deficiency of Fas in ODCs and TNF-R1 are almost completely resistant to EAE induction.

The mice lacking these two major apoptosis-mediating receptors, Fas and TNF-R1, still exhibited moderate levels of inflammation in the CNS. In contrast, we found very little damage in the CNS in those mice. Together, these data are consistent with the fact that damage to ODCs is crucial for clinical signs of EAE and that inflammation of the CNS will not indefinitely result in EAE pathogenesis, because mice deficient for Fas on ODCs are protected from apoptosis by the invading lymphocytes although the latter are able to enter the CNS.

Fas expression was detected on ODCs of MS patients, where highest expression was seen at the edges of the lesion (46). It was shown that during the inflammatory reaction in the CNS of MS patients, Fas expression is up-regulated in CNS cells in a process involving IFN-γ (47, 48, 49). FasL expression in the human brain was reported for microglia, indicating a role of these cells in disease development. In addition, in the course of MS, apoptotic death of ODCs in acute plaques was shown, although the vast majority of apoptotic cells were found to be T cells (50, 51).

We thank Anke Leinhaas, Angela Egert, and Claudia Uthoff-Hachenberg for excellent technical assistance. We thank Dr. Ulrich Kalinke, Dr. Jack Zapulla, Dr. Stefan Brocke, Christine Tertilt, and Iana Parvanova, as well as Andrew Croxford and Carsten Merkwirth for critically reading the manuscript. Further, we thank Alexander Pozhitkov for help with statistical analysis.

The authors have no financial conflict of interest.