Transmembrane tumour necrosis factor is neuroprotective and regulates experimental autoimmune encephalomyelitis via neuronal nuclear factor- B

doi:10.1093/brain/awr203

Brain 2011: 134; 2722–2735

| 2722

BRAIN A JOURNAL OF NEUROLOGY

Transmembrane tumour necrosis factor is neuroprotective and regulates experimental autoimmune encephalomyelitis via neuronal nuclear factor-B Era Taoufik,1 Vivian Tseveleki,1 Seung Y. Chu,2 Theodore Tselios,3 Michael Karin,4 Hans Lassmann,5 David E. Szymkowski2 and Lesley Probert1 Laboratory of Molecular Genetics, Hellenic Pasteur Institute, 11521 Athens, Greece Xencor, 111 West Lemon Avenue, Monrovia, CA 91016, USA Department of Chemistry, University of Patras, 26504 Rio, Greece Laboratory of Gene Regulation and Signal Transduction, Department of Pharmacology, University of California at San Diego, La Jolla, CA 92093, USA 5 Division of Neuroimmunology, Brain Research Institute, A-1090 Vienna, Austria Correspondence to: Dr Lesley Probert, Laboratory of Molecular Genetics, Hellenic Pasteur Institute, 11521 Athens, Greece E-mail: [email protected]

Tumour necrosis factor mediates chronic inflammatory pathologies including those affecting the central nervous system, but non-selective tumour necrosis factor inhibitors exacerbate multiple sclerosis. In addition, TNF receptor SF1A, which encodes one of the tumour necrosis factor receptors, has recently been identified as a multiple sclerosis susceptibility locus in genome-wide association studies in large patient cohorts. These clinical data have emphasized the need for a better understanding of the beneficial effects of tumour necrosis factor during central nervous system inflammation. In this study, we present evidence that the soluble and transmembrane forms of tumour necrosis factor exert opposing deleterious and beneficial effects, respectively, in a multiple sclerosis model. We compared the effects, in experimental autoimmune encephalomyelitis, of selectively inhibiting soluble tumour necrosis factor, and of both soluble and transmembrane tumour necrosis factor. Blocking the action of soluble tumour necrosis factor, but not of soluble tumour necrosis factor and transmembrane tumour necrosis factor, protected mice against the clinical symptoms of experimental autoimmune encephalomyelitis. Therapeutic benefit was independent of changes in antigen-specific immune responses and focal inflammatory spinal cord lesions, but was associated with reduced overall central nervous system immunoreactivity, increased expression of neuroprotective molecules, and was dependent upon the activity of neuronal nuclear factor-B, a downstream mediator of neuroprotective tumour necrosis factor/tumour necrosis factor receptor signalling, because mice lacking IB kinase b in glutamatergic neurons were not protected by soluble tumour necrosis factor blockade. Furthermore, blocking the action of soluble tumour necrosis factor, but not of soluble tumour necrosis factor and transmembrane tumour necrosis factor, protected neurons in astrocyte–neuron co-cultures against glucose deprivation, an in vitro neurodegeneration model relevant for multiple sclerosis, and this was dependent upon contact between the two cell types. Our results show that soluble tumour necrosis factor promotes central nervous system inflammation, while transmembrane tumour necrosis factor is neuroprotective, and suggest that selective inhibition of soluble tumour necrosis factor may provide a new way forward for the treatment of multiple sclerosis and possibly other inflammatory central nervous system disorders.

Received May 11, 2011. Revised July 13, 2011. Accepted July 22, 2011 ß The Author (2011). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

1 2 3 4

EAE therapy by soluble TNF blockade

Brain 2011: 134; 2722–2735

| 2723

Keywords: multiple sclerosis; EAE; TNF; NF-B; therapy Abbreviations: EAE = experimental autoimmune encephalomyelitis; FLIP = FLICE-inhibitory protein; IKKb = IB kinase b; IKKbF/F = loxP-flanked Ikbkb; MOG35–55 = myelin oligodendrocyte glycoprotein peptide 35–55; NF-B = nuclear factor-B; RT-PCR = reverse transcription-polymerase chain reaction; TNF = tumour necrosis factor

Introduction

Materials and methods Animals Mice containing a conditional IB kinase b (IKKb) allele in which exon 3 of the Ikbkb gene, encoding the IKKb activation loop, is flanked by loxP sites (IKKbF/F) have previously been described (Park et al., 2002; Li et al., 2003). IKKb is the main activating kinase for NF-B in the canonical pathway (Ghosh and Karin, 2002). Mice with a selective deletion of IKKb in CNS neurons (nIKKbKO) were generated by crossing IKKbF/F mice with mice that express a neuronal calmodulin-kinase IIa promoter-driven Cre recombinase (CamkIICre; Minichiello et al., 1999). Mice were kept under specific pathogen-free conditions in the experimental animal unit of Hellenic Pasteur Institute. All animal procedures were approved by national authorities and conformed to European Community guidelines.

Experimental autoimmune encephalomyelitis induction and evaluation EAE in female C57BL/6, IKKbF/F and nIKKbKO mice was induced by subcutaneous tail base injection of 30 mg of the 35–33 peptide sequence of rat myelin oligodendrocyte glycoprotein (MOG35–55) dissolved in 100 ml saline emulsified in an equal volume of complete Freund’s adjuvant supplemented with 400 mg of H37Ra Mycobacterium tuberculosis (Difco). Mice also received an intraperitoneal injection of 200 ng of pertussis toxin (Sigma-Aldrich) on Days 0 and 2 post-immunization. Groups of mice were treated with twice weekly subcutaneous injections of XPro1595 (10 mg/kg; Xencor; Steed et al., 2003), etanercept (10 mg/kg; Amgen; Murray and Dahl, 1997) or saline vehicle in either prophylactic (starting on the day of immunization) or therapeutic (starting at the onset of clinical signs) protocols. Mice were assessed daily for clinical signs of disease according to the following scale: 0, normal; 1, limp tail; 2, hind limb weakness; 3, hind limb paralysis; 4, forelimb weakness; and 5, moribund or dead (0.5 gradations represent intermediate scores). Moribund animals were sacrificed and given a clinical score of 5 for the remaining days

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

Tumour necrosis factor (TNF) is a highly regulated proinflammatory cytokine, the aberrant production of which promotes a wide range of chronic inflammatory pathologies, including those affecting the CNS (Ruddle et al., 1990; Selmaj et al., 1991; Baker et al., 1994; Korner et al., 1997; Akassoglou et al., 1998; Centonze et al., 2009). However, while TNF inhibitors are effective treatments for human diseases such as rheumatoid arthritis (Taylor and Feldmann, 2009), they provide no benefit to patients with multiple sclerosis; indeed, they exacerbate multiple sclerosis (van Oosten et al., 1996; The Lenercept Multiple Sclerosis Study Group and The University of British Columbia MS/MRI Analysis Group, 1999) and can even induce demyelinating disease and neuropathies (Mohan et al., 2001; Stubgen, 2008). Taken together, clinical data show that reduced TNF activity is associated with the onset of CNS inflammatory lesions. Experimental studies have provided several lines of evidence for beneficial functions of TNF in the CNS. Under physiological conditions, TNF is important for regulating synaptic strength at excitatory and inhibitory synapses through mechanisms of synaptic scaling (Beattie et al., 2002; Stellwagen et al., 2005) and gliotransmission (Santello et al., 2011) as well as learning and memory processes in the hippocampus (Albensi and Mattson, 2000). Under conditions of pathology, TNF supports CNS cells and tissue by inducing neuroprotection mechanisms through its two membrane-bound receptors, TNF receptor I (Gary et al., 1998; Taoufik et al., 2007; Lambertsen et al., 2009) and TNF receptor II (Marchetti et al., 2004) in models for stroke and excitotoxicity, and by promoting remyelination via TNF receptor II signalling in a model of CNS demyelination (Arnett et al., 2001). However, the cellular and molecular mechanisms of beneficial TNF function are still poorly defined. TNF is produced in two biologically active forms, transmembrane TNF and a cleaved soluble form (soluble TNF; Wajant et al., 2003). Transmembrane TNF signals efficiently through both TNF receptors I and II, while soluble TNF selectively signals through TNF receptor I, leaving transmembrane TNF as the main TNF receptor II ligand (Grell et al., 1995). Both receptors are widely expressed, including in cells of the CNS under normal and pathological conditions (Aranguez et al., 1995; Botchkina et al., 1997; Wajant et al., 2003). Genetic and pharmacological studies addressing the individual functions of soluble and transmembrane TNF have shown that transmembrane TNF mainly supports beneficial processes such as lymphoid organ structure and host defence responses to infections (Ruuls et al., 2001; Alexopoulou et al., 2006; Zalevsky et al., 2007), while soluble TNF drives optimal inflammatory responses in models of endotoxic shock and autoimmune disease (Ruuls et al., 2001; Steed et al., 2003; Alexopoulou et al., 2006; Zalevsky et al., 2007). However, little is known about the function of transmembrane TNF and soluble TNF in the CNS. Here, we investigated the effects of

soluble and transmembrane TNF in a multiple sclerosis model (experimental autoimmune encephalomyelitis, EAE) producing chronic CNS inflammation and demyelination (Gold et al., 2006). We used a novel dominant-negative TNF analogue, XPro1595, to selectively block the effects of soluble TNF (Steed et al., 2003; Zalevsky et al., 2007), and compared its effects with those of blocking the actions of both soluble and transmembrane TNF using etanercept, a TNF receptor II-IgG1 Fc fusion protein (Murray and Dahl, 1997). We provide evidence that transmembrane TNF directly protects neurons against in vitro ischaemic injury and that selective blockade of soluble TNF, but not soluble TNF and transmembrane TNF, is therapeutic in EAE via a neuronal NF-B-dependent mechanism.

2724

| Brain 2011: 134; 2722–2735

of the experiment. Mice were allowed access to food and water ad libitum throughout the experiment.

Histopathological analysis Mice were transcardially perfused with ice-cold 4% paraformaldehyde in phosphate-buffered saline under deep anaesthesia. CNS tissues were post-fixed in the same fixative overnight at 4 C, embedded in paraffin and processed for standard histopathological and immunohistochemical analyses. Inflammation was visualized by haematoxylin and eosin staining, demyelination by Kluver-Barrera Luxol fast blue and axonal damage by Bielschowsky silver impregnation according to standard procedures. Immunostaining for CD3 (Thermo Scientific; RM9107) and amyloid precursor protein (Millipore; MAB 348) was performed on paraffin sections using an avidin–biotin detection technique. Antigen retrieval was performed by placing sections in a food steamer for 1 h with citrate buffer (pH 6.0).

Biomarker analysis

T cell priming and proliferation assays T cells were primed in vivo by subcutaneous immunization of C57BL/6 mice with 30 mg of MOG35–55 peptide dissolved in 100 ml saline and emulsified in an equal volume of complete Freund’s adjuvant (SigmaAldrich) supplemented with 400 mg H37Ra M. tuberculosis (Difco). Draining lymph nodes and spleens were removed 10 days later. Isolated mononuclear cells were stimulated in triplicate at 4  106 cells/ml in round-bottom, 96-well plates (Costar) for 72 h in RPMI 1640 (Biochrom) containing 10% heat-inactivated foetal calf serum, 50 mM 2-mercaptoethanol and increasing concentrations of MOG35– 3 55. Cells were pulsed with 0.5 mCi [ H]thymidine (Amersham 5 Radiochemicals)/5  10 cells for the last 16 h of culture, and [3H]thymidine incorporation was measured by liquid scintillation counting (Wallac). Results are expressed as a stimulation index to MOG35–55 peptide (ratio between radioactivity counts of cells cultured in the presence of peptide and cells cultured with medium alone).

Extraction and characterization of central nervous system-infiltrating cells Splenocytes and spinal cord-infiltrating mononuclear cells were isolated from mice at the peak of EAE and Day 35 post-immunization. Spinal cord mononuclear cells were isolated by Percoll gradient centrifugation as previously described (Krishnamoorthy et al., 2006). For detection of cell surface markers, cells were washed and fixed in 2% paraformaldehyde solution in phosphate-buffered saline for 15 min at room temperature and stained with fluorochrome-labelled antibodies (antiNK1.1, clone PK136; anti-CD11b/Mac1, clone M1/70; BD Biosciences). For detection of the CD4 surface marker and intracellular

cytokines by fluorescence-activated cell sorting analysis, after isolation, cells were restimulated with 10 ng/ml phorbol 12-myristate 13-acetate and 1 mg/ml ionomycin in the presence of 5 mg/ml brefeldin A (Sigma-Aldrich) for 3 h before fixation, as above. Cells were permeabilized with 0.5% w/v saponin and stained with fluorochrome-labelled antibodies (anti-CD4, clone L3T4; anti-IFN- , clone XMG1.2; antiIL-17, clone TC11-18H10; BD Biosciences). Data acquisition and analysis were done with a FACSCalibur cytometer and CellQuest software (BD Biosciences).

Neuron/astrocyte cultures and experimental treatments Astrocyte cultures were prepared from post-natal Days 1–3 cortical tissues of C57BL/6 mice by mild mechanical trituration. Cells were grown on poly-D-lysine (0.1 mg/ml) and laminin (20 mg/ml)-coated 24-well plates in medium containing Dulbecco’s Modified Eagle Medium containing 4.5 g/L glucose and supplemented with 10% foetal calf serum, 10% horse serum, 2 mM Glutamax, 1 mM sodium pyruvate, 100 mM non-essential amino acids, 50 U/ml penicillin and 50 mg/ml streptomycin. A confluent layer of astrocytes was used as a feeder layer for primary cortical neurons. Cortical neurons were prepared from embryonic Day 15 mice as previously described (Nicole et al., 2001), plated on the astrocyte feeder layer, and used for experiments starting on neuronal Days 7 and 8 in vitro. In some experiments, the astrocytes were seeded on Transwell membrane inserts (BD Falcon) and co-cultured with neurons without direct contact. Glucose deprivation was performed as previously described (Taoufik et al., 2007). Human recombinant TNF (R&D; 50 ng/ml), which selectively signals through murine TNF receptor I (Lewis et al., 1991) and is sufficient to activate murine astrocytes (Aranguez et al., 1995), was added to co-cultures 24 h prior to the onset of and during glucose deprivation. XPro1595 (Xencor; 200 ng/ml; Steed et al., 2003) or etanercept (Amgen; 100 ng/ml; Murray and Dahl, 1997) was added to the co-cultures 4 h after the onset of glucose deprivation. BMS 345 541 (Merck; 25 mM) was added to the co-cultures 2 h before glucose deprivation. Mixed cultures were stained with 0.4% trypan blue dye solution (Sigma) and cells were considered viable if they excluded the dye. Cells with neuron morphology were counted in 10 different fields per well, in at least three separate cultures per condition, by phase contrast microscopy (  40 objective).

RNA isolation and semi-quantitative reverse transcription–polymerase chain reaction Total RNA was extracted with TRIzolÕ (Invitrogen) according to the manufacturer’s instructions. For semi-quantitative reverse transcription– polymerase chain reaction (RT-PCR), DNase-treated (Promega) RNA was reverse transcribed with M-MLV Reverse Transcriptase (Promega) and random hexamers (Roche). Primers were used for the detection of FLICE-inhibitory protein (FLIP) (forward: 5’-GAA GAG TGT CTT GAT GAA GA-3’ and reverse: 5’-GAA AAG CTG GAT ATG ATA GC-3’), vascular endothelial growth factor (forward: 5’-GCG GGC TGC CTC GCA GTC-3’ and reverse: 5’-TCA CCG CCT TGG CTT GTC AC-3’) and colony stimulating factor-1 receptor (forward: 5’GAC CTG CTC CAC TTC TCC AG-3’ and reverse: 5’-GGG TTC AGA CCA AGC GAG AAG-3’). Mouse b-actin was amplified as a loading control. Densitometric analysis was performed using Image

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

Mouse brain, spinal cord and serum from representative mice were dissected and frozen in liquid nitrogen. Tissue homogenates and serum were analysed by multi-analyte profiling technology (Rules Based Medicine). Tissues were assayed using the RodentMAPÕ v. 2.0 multiplex immunoassays. Bead-based immunoassays were developed and optimized at Rules Based Medicine and read out on a Luminex 200 system (Luminex). The panel contained 59 biomarkers, including key cytokines, chemokines and proteases. Markers not shown for spinal cord versus brain and serum could not be assayed in spinal cord due to limited material.

E. Taoufik et al.

EAE therapy by soluble TNF blockade Quant 5.2 (Molecular Dynamics Storm Scanner 600) and relative band intensities were determined.

Western blot analysis Total protein extracts from spinal cords of selected mice were prepared as previously described (Taoufik et al., 2007). Thirty micrograms of total protein extracts were resolved on NuPAGE Novex Bis–Tris Gels (Invitrogen) and transferred onto nitrocellulose membranes (Schleicher and Schuell). Blots were probed with antibodies against the nonphosphorylated form of neurofilament-H (SMI-32) (1:400, Covance), glial fibrillary acidic protein (1:1500, Chemicon), myelin basic protein (1:200, Chemicon), phospho-IB (1:500, Cell Signalling), phospho-p65 (1:250, Cell Signalling) and caspase 3 (1:2000, Santa Cruz). The secondary antibodies used were horseradish peroxidaseconjugated anti-mouse and anti-rabbit IgG (1:2000 up to 1:5000, Jackson Immunoresearch Laboratories). Antibody binding was detected using the ECL Plus detection system (Amersham Pharmacia). To normalize for protein content, membranes were stripped and reprobed with anti-b-tubulin antibody (1:1000, Pharmingen).

Brain 2011: 134; 2722–2735

| 2725

control in both therapeutic (Fig. 1A and Table 1) and prophylactic (Fig. 1B and Table 1) protocols, demonstrating disease-promoting effects of soluble TNF. In contrast, inhibition of both soluble TNF and transmembrane TNF with etanercept did not provide benefit when compared with controls in either treatment protocols and even exacerbated disease. Comparison between the two treatment groups showed that etanercept-treated mice displayed significantly increased clinical defects compared with XPro1595-treated mice (Fig. 1A and B; Table 1), indicating that transmembrane TNF exerts significant beneficial effects in EAE. Mice-treated prophylactically with either XPro1595 or etanercept showed a significant delay in disease onset (Fig. 1B and Table 1) confirming a disease-advancing role for soluble TNF. All subsequent analyses were performed using mice treated with the therapeutic protocol.

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

Statistical analysis Statistical analyses were performed with Sigma Stat 2.0 for Windows. All data are given as mean  SEM. To compare clinical scores among treated groups of mice at each time point and neuropathological changes (inflammation, demyelination, axon dystrophy and axon loss), the Mann–Whitney rank sum test was performed. For biomarker analysis, Student’s t-test was performed. For semi-quantitative protein and RT–PCR analyses, a measurement of the band intensity was performed with ImageQuant 5.2 (Molecular Dynamics Storm Scanner 600) and expressed as pixel intensity per unit area. Values were normalized using the b-actin or b-tubulin values, respectively, and were compared using one-way ANOVA followed by Bonferroni t-test for pairwise comparisons. For cell viability, ANOVA on Ranks followed by Bonferroni t-test for pair-wise comparisons was used. Fluorescenceactivated cell sorting analysis and proliferative responses of splenocytes and lymph node cells were analysed by Mann–Whitney rank sum test. P 5 0.05 were considered statistically significant.

Results Inhibition of soluble tumour necrosis factor, but not soluble and transmembrane tumour necrosis factor, protects mice against experimental autoimmune encephalomyelitis EAE was induced in female C57BL/6 mice by immunization with MOG35–55 peptide (Gold et al., 2006), and the effects of twice weekly subcutaneous injection of XPro1595 (10 mg/kg), or etanercept (10 mg/kg) or of saline vehicle were tested in both prophylactic (starting on the day of immunization) and therapeutic (starting at the onset of clinical signs) protocols. Both inhibitors have been characterized previously to show equivalent potency and effectiveness in a murine model of rheumatoid arthritis Listeria monocytogenes infection (Zalevsky et al., 2007). XPro1595 provided significant clinical benefits compared with

Figure 1 EAE is ameliorated by the inhibition of soluble TNF, but not of soluble TNF and transmembrane TNF. Active EAE was induced in C57BL/6 mice by immunization with MOG35–55 peptide. (A) Clinical EAE scores (means  SEM) over time in mice-treated therapeutically with XPro1595 (filled triangle, n = 9), etanercept (open circle, n = 8) or vehicle (filled square, n = 8) from the onset of clinical signs (arrow). Data are from one representative experiment of four. (B) Clinical EAE scores (means  SEM) over time in mice-treated prophylactically with XPro1595 (filled triangle), etanercept (open circle) or vehicle (filled square) (n = 10 per group) from the day of immunization. Arrows indicate first day of administration. *P 5 0.05 for comparisons between vehicle- versus XPro1595- or etanercepttreated mice. nP 5 0.05 for comparisons between XPro1595and etanercept-treated mice.

2726

| Brain 2011: 134; 2722–2735

E. Taoufik et al.

Table 1 Incidence, clinical severity and mortality rate of MOG35–55 induced EAE in vehicle-, XPro1595- and etanercept-treated mice Mice

Therapeutic Experiment 1 Vehicle XPro1595 Etanercept Experiment 2 Vehicle XPro1595 Etanercept Prophylactic Vehicle XPro1595 Etanercept

Incidence (%)

Day of onset, mean  SEM

Maximal score, mean  SEM

Clinical score at first peak, mean  SEM

Cumulative score (up to day post-immunization)

Mortality rate (%)

8/8 (100) 6/9 (67) 8/8 (100)

14.75  0.79 15.8  1.07 15.63  0.99

4  0.33a 2.14  0.95b 4.57  0.28

2.57  0.39 1.31  0.51b 2.93  0.41

614 (53) 343 (53) 836 (53)

3/8 (38) 2/6 (33.3) 5/8 (62.5)

10.4  0.73 11.2  0.51 10.7  0.21

4  0.4 2.75  0.16 4  0.32

3.1  0.38 2.25  0.13b 3.1  0.27

832 (39) 565 (39) 737 (39)

3/10 (30) 0/10 (0) 3/10 (30)

11.3  0.49c 17.2  1.21 15.33  0.44

3.4  0.38 4  0.47 4.25  0.15

2.85  0.38 2.7  0.33 3.56  0.15

754 (41) 671 (41) 698 (41)

1/10 (10) 3/10 (30) 2/10 (20)

10/10 (100) 10/10 (100) 10/10 (100) 10/10 (100%) 10/10 (100%) 10/10 (100%)

Spinal cord sections taken at disease peak in the vehicle group (Fig. 2) and Day 35 post-immunization with MOG peptide, when block of soluble TNF with XPro1595 gave maximal protection (Supplementary Fig. 1), showed equivalent levels of inflammatory cell infiltration and similar lesion size in the different treatment groups at both time points, as measured at peak by immunostaining for CD3 (Fig. 2A, D and G) and Day 35 post-immunization by haematoxylin and eosin staining (Supplementary Fig. 1A, D and G). The area of local structural damage in lesions, as measured by Luxol fast blue staining for demyelination (Fig. 2B, E and H; Supplementary Fig. 1B, E and H) and the number of amyloid precursor protein-immunostained dystrophic axons in the anterior column (Fig. 2C, F and I), was also similar in XPro1595treated compared with vehicle-treated samples at both time points. Surprisingly, myelin and axon damage were reduced in etanercept-treated compared with vehicle-treated samples at peak (Fig. 2B, C, H and I), but not Day 35 post-immunization (Supplementary Fig. 1B, C, H and I), even though these mice presented the most severe clinical scores. These results revealed a dissociation between the effects of TNF inhibitors on clinical disease and on early inflammatory spinal cord lesions, and indicate that the therapeutic benefit of soluble TNF inhibition is at least partly independent of levels of initial CNS infiltration by inflammatory cells.

Tumour necrosis factor blockade does not compromise MOG35–55-specific effector T cell responses We next examined whether TNF blockade affected immune responses by analysing primary T cell responses to MOG35–55 in mice that had soluble TNF, or soluble TNF and transmembrane TNF, blocked. No significant differences in the recall proliferation

responses of splenocytes or of draining lymph node cells to MOG35–55 (Supplementary Fig. 2A) or maturation of CD4 + IFN- + and CD4 + IL-17 + effector T cells in spleen (Supplementary Fig. 2B) were detected among the different treatment groups. We also compared secondary T cell responses to MOG35–55 in splenocytes and spinal cord-infiltrating mononuclear cells isolated from mice at peak and chronic phases of EAE. Again, no differences in splenocyte or CNS-infiltrating CD4 + IFN- + T cells, CD11b + cells or NK1.1 + were detected among the treatment groups, except for an increase in CD4 + IFN- + splenocytes in XPro1595-treated versus vehicle-treated mice at peak (Supplementary Fig. 2B). Thus, overall immune responses are not affected by either of the TNF inhibitors.

Inhibition of soluble tumour necrosis factor, not soluble and transmembrane tumour necrosis factor, reduces the production of pro-inflammatory mediators and chemokines in the central nervous system To explore mechanisms that might mediate the differences in clinical score between treatment groups, we performed a broad biomarker analysis on naı¨ve mice and EAE mice treated with vehicle or TNF blockers, at the time of disease peak in the vehicle group. A standard biomarker panel from Rules Based Medicine was used to quantitatively assess expression of 59 key protein markers in whole brain and spinal cord extracts and serum. Many proinflammatory and chemotactic molecules showed upregulation in the spinal cord (Fig. 3A) and brain (Fig. 3B) but not serum (Fig. 3C) of vehicle-treated EAE compared with naı¨ve mice with similar expression patterns, although the amplitude of effects

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

a P 5 0.05 for comparison between mice treated with vehicle or XPro1595. b P 5 0.05 for comparison between mice treated with XPro1595 or etanercept. c P 5 0.05 for comparison between mice treated with vehicle or XPro1595 and vehicle or etanercept.

EAE therapy by soluble TNF blockade

Brain 2011: 134; 2722–2735

| 2727

was greater in the spinal cord. Notably, a large number of these molecules are products of activated macrophages/ microglia (MPO, MCP-1/CCL2, RANTES, IL-1, KC/GRO/CXCL1; Miyagishi et al., 1997; Filipovic et al., 2003; Simi et al., 2007; Gray et al., 2008) and astrocytes (MCP-1, MCP-3/CCL7, IP-10, RANTES, KC/GRO; Ransohoff et al., 1993; Miyagishi et al., 1997; Omari et al., 2006; Thompson and Van Eldik, 2009) under conditions of inflammation; several are key mediators of EAE (MPO, MCP-1, IL-6; Eugster et al., 1998; Mahad and Ransohoff, 2003; Chen et al., 2008) and are upregulated in multiple sclerosis tissues (MPO, MCP-1, KC/GRO; Filipovic et al., 2003; Mahad and

Ransohoff, 2003; Omari et al., 2006; Gray et al., 2008). Interestingly, inhibition of soluble TNF and transmembrane TNF by etanercept further enhanced expression of several of these mediators in the spinal cord (Fig. 3A) and brain (Fig. 3B) compared with vehicle-treated EAE mice. In contrast, selective soluble TNF inhibition by XPro1595 generally reduced expression relative to vehicle-treated mice in both spinal cord (Fig. 3A) and brain (Fig. 3B). These results show that transmembrane TNF acts to downregulate the production of pro-inflammatory mediators within CNS tissues during EAE, while soluble TNF increases their expression.

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

Figure 2 TNF inhibitors do not alter the extent of inflammatory cell infiltration in the spinal cord at EAE peak. Neuropathological analysis of spinal cord sections from representative vehicle- (top), XPro1595- (middle) and etanercept-(bottom) treated mice at disease peak (vehicle, n = 2; XPro1595, n = 10; etanercept, n = 8) stained with anti-CD3 antibody for T cells (A, D and G), Luxol fast blue for myelin (B, E and H) and anti-amyloid precursor protein antibody for axonal dystrophy (C, F and I; arrows show damaged axons). The clinical scores of the representative mice at the time of sacrifice were: vehicle = 3.5; XPro1595 = 3.0; and etanercept = 3.5. The amyloid precursor protein-immunostained areas shown on the right are indicated as insets on serial sections of the spinal cord. Scale bars: A, B, D, E, G and H = 1 mm; C, F, I = 100 mm. (J) Quantitative representation of neuropathological changes in treated spinal cords at disease peak. Inflammation was scored in haematoxylin and eosin-stained sections as number of perivascular inflammatory infiltrates per average spinal cord section (n = 10 sections per animal). Demyelination was scored in Luxol fast blue-stained sections by measuring the area of demyelination with a morphometric grid and representing it as a percentage of the total spinal cord area (n = 10 sections per animal). Axon dystrophy was measured as the numbers of amyloid precursor protein + -immunoreactive axons in the lesions per high power field ( 650 magnification; n = 6 microscope fields in three representative lesions per animal). Statistical comparisons are made between levels of pathology in vehicle- versus XPro1595- or etanercept-treated tissues.

2728

| Brain 2011: 134; 2722–2735

E. Taoufik et al.

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

Figure 3 Non-selective TNF inhibition exacerbates, and soluble TNF-selective inhibition suppresses, inflammation markers in the CNS at EAE peak. Multiplex immunoassay analysis of 59 key protein markers was done on spinal cord (A), brain (B) and serum (C) of naive and three treatment groups of EAE mice at peak of disease. Mean is shown for XPro1595 (green triangle) and etanercept (red circle) (n = 3 mice per group); for naı¨ve (dotted line) and vehicle (open square) (n = 2 mice per group). Markers are rank-ordered by foldinduction for etanercept group versus naive samples. EAE induces expression of almost all tested markers, while etanercept amplifies and XPro1595 suppresses expression versus vehicle-treated samples. *P 5 0.05 for comparison between etanercept and naı¨ve samples.

EAE therapy by soluble TNF blockade

Inhibition of soluble tumour necrosis factor, not soluble and transmembrane tumour necrosis factor, maintains the expression of neuroprotective proteins in the spinal cord Immunoblot and densitometry analysis of total spinal cord lysates from naı¨ve and treated EAE mice at disease peak for CNS cell marker proteins showed that levels of the non-phosphorylated form of neurofilament-H (SMI-32), a marker of neurodegeneration, and glial fibrillary acidic protein, a marker of astrocytosis, were significantly upregulated in both vehicle-treated and soluble TNF plus transmembrane TNF blocker (etanercept)-treated compared with naı¨ve tissue (Fig. 4A). In contrast, soluble TNF blocker (XPro1595)-treated tissues showed no changes in SMI-32 levels

Brain 2011: 134; 2722–2735

| 2729

compared with naı¨ve tissue and a smaller upregulation of glial fibrillary acidic protein compared with vehicle- and etanercepttreated tissue (Fig. 4A). Levels of myelin basic protein were markedly reduced in tissues from all treatment groups compared with naı¨ve, although this change was again less severe in XPro1595treated compared with vehicle- and etanercept-treated tissues (Fig. 4A). We next analysed the phosphorylated forms of the inhibitor of NF-B, which becomes degraded thereby releasing active NF-B (Ghosh and Karin 2002), and of the p65 subunit of NF-B, which is specifically induced by TNF (Ghosh and Karin, 2002) and associated with TNF-mediated neuroprotection (Camandola and Mattson, 2007). The levels of both phosphorylated inhibitor of NF-B and phosphorylated p65 NF-B subunit were significantly reduced in vehicle- and etanercept treated compared with naı¨ve spinal cord, implying lower levels of NF-B activity. However,

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

Figure 4 Blockade of soluble TNF, but not soluble TNF and transmembrane TNF, is sufficient to maintain NF-B activity and expression of other neuroprotective mediators in the spinal cord during EAE. (A) Immunoblot analysis of SMI-32, glial fibrillary acidic protein (GFAP) and myelin basic protein (MBP) expression in extracts of spinal cord from naı¨ve and treated EAE mice at disease peak. Representative blots are shown and quantification of protein expression using b-tubulin for normalization was done using values from n = 3 mice per group. Statistical comparisons are made between protein levels in treated versus naı¨ve samples. (B) Similar analysis of phosphorylated-inhibitor of NF-B (p-IB) and phosphorylated p65 NF-B subunit (p-p65; arrowhead) protein levels in spinal cord extracts from naı¨ve and Day 22 EAE mice from different treatment groups (XPro1595, n = 4; etanercept, n = 4; vehicle, n = 3; naı¨ve = 5). Statistical comparisons are made between protein levels in treated versus naı¨ve samples. (C) FLIP, vascular endothelial growth factor (VEGF) and CSF-1 receptor messenger RNA transcript levels in spinal cord extracts from EAE mice at peak and Day 39 post-immunization by semi-quantitative RT-polymerase chain reaction using b-actin for normalization. Data show mean mRNA expression  SEM *P 5 0.05, **P 5 0.01.

2730

| Brain 2011: 134; 2722–2735

Transmembrane tumour necrosis factor mediates neuroprotection against glucose deprivation in astrocyte–neuron co-cultures To examine whether the protective effects of transmembrane TNF in EAE include the enhanced survival of neurons, we studied the effects of TNF blockade in astrocyte-neuron co-cultures subjected to glucose deprivation-induced death, a model for ischaemic neuronal damage that has relevance to multiple sclerosis (Lassmann, 2003). Astrocytes were included as a source of murine transmembrane TNF and soluble TNF. As previously shown in nearly pure cortical neuron cultures (55% astrocytes; Taoufik et al., 2007), glucose deprivation induced the death of neurons that were cultured in direct contact with astrocytes, and this was inhibited by soluble TNF blockade (XPro1595), but not soluble TNF plus transmembrane TNF blockade (etanercept) starting 4 h after the onset of glucose deprivation (Fig. 5A and B). Astrocyte viability was not affected by glucose deprivation (data not shown). To activate astrocytes and to induce the production of transmembrane TNF, astrocyte-neuron co-cultures were pretreated with human TNF for

24 h prior to glucose deprivation, conditions that induce strong neuroprotection in nearly pure neuronal cultures (Taoufik et al., 2007). These conditions significantly reduced glucose deprivationinduced neuronal death, and this protection was inhibited by etanercept but not XPro1595, indicating that it is mediated by transmembrane TNF. A neuroprotective role for induced transmembrane TNF was further confirmed by the finding that XPro1595-treated cultures showed less neuronal death after human TNF pretreatment (Fig. 5B). When neurons were cultured with astrocytes separated from them by Transwells, however, soluble TNF blockade (XPro1595) starting 4 h after the onset of glucose deprivation was ineffective in protecting neurons against glucose deprivation-induced death (Fig. 5C). This shows that transmembrane TNF neuroprotection is dependent upon astrocyte–neuron contact and that astrocytes are the source of neuroprotective transmembrane TNF under these conditions. Immunoblot and densitometry analysis for phosphorylated inhibitor of NF-B and the phosphorylated p65 NF-B subunit of NF-B showed that, as in EAE spinal cord, their levels were maintained in XPro1595-treated, but not etanercept-treated, glucose deprivation-challenged astrocyte–neuron cultures when measured 24 h after glucose deprivation (Supplementary Fig. 3). Levels of active caspase 3 (p20 subunit) were also significantly reduced in XPro1595- compared with etanercept-treated cells at 12 h after glucose deprivation (Supplementary Fig. 3).

Transmembrane tumour necrosis factor-mediated protection in experimental autoimmune encephalomyelitis is dependent upon neuronal NF-B activity We next tested whether transmembrane TNF-mediated neuroprotection through NF-B signalling in neurons may be one mechanism of XPro1595’s clinical efficacy in vivo in the EAE model. In a previous study, we showed that neuronal IKKb, which phosphorylates and mediates the degradation of IB, thereby activating NF-B (Ghosh and Karin 2002), is sufficient to mediate neuroprotection and suppression of CNS inflammation in EAE (Emmanouil et al., 2009). Here, we tested whether XPro1595 was capable of exerting its therapeutic effects in conditional IKKb mice (Park et al., 2002; Li et al., 2003) that selectively lack IKKb in CamkII-expressing neurons in the brain and spinal cord (nIKKbKO). Similar to results in wild-type C57BL/6 mice, control IKKbF/F mice, in which exon 3 of the Ikbkb gene is flanked by loxP sites but not deleted by Cre recombinase, were protected from the clinical symptoms of EAE during the chronic phase by XPro1595 administration (Fig. 6A). In contrast, XPro1595 administration provided no protection in nIKKbKO mice, and disease followed a severe non-remitting course in both vehicle- and XPro1595treated groups until the last time point studied (Fig. 6A). Beneficial effects of XPro1595 in IKKbF/F, but not nIKKbKO, mice were also seen on animal survival (Fig. 6B). Consistent with these results, a selective IKKb inhibitor, BMS 345 541, abolished the neuroprotective effects of XPro1595 against glucose

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

blocking the activity of soluble TNF with XPro1595 treatment maintained the phosphorylation levels of these proteins close to those in naı¨ve tissue (Fig. 4B). This shows that soluble TNF contributes to the loss of spinal cord NF-B activity and that transmembrane TNF has an opposing effect during EAE. We next compared expression of the NF-B-induced neuroprotective proteins FLIP (Kreuz et al., 2001) and vascular endothelial growth factor (Schmidt et al., 2007), and of the receptor for colony stimulating factor-1 (Berezovskaya et al., 1996), CSF-1R, by semi-quantitative RT–PCR of total RNA isolated at peak and Day 39 post-immunization (Fig. 4C). FLIP expression was significantly reduced in vehicle-treated spinal cord at both EAE time points compared with naı¨ve tissue (Fig. 4C). In contrast, its expression was preserved in XPro1595- and etanercept-treated cord, showing that soluble TNF contributes to EAE-induced loss of FLIP expression. Pair-wise analysis between treatment groups at each time point showed that the soluble TNF blocker XPro1595 was more effective than etanercept at maintaining FLIP levels in the spinal cord during EAE, indicating that transmembrane TNF is sufficient for this. Vascular endothelial growth factor expression was not altered in vehicle-treated spinal cord during EAE compared with naı¨ve tissue (Fig. 4C). However, pair-wise analysis between groups at each time point showed marked upregulation of expression in XPro1595-treated tissues at both time points, and in etanercept-treated tissues at peak, showing that soluble TNF inhibits vascular endothelial growth factor induction during EAE and that XPro1595 most effectively blocks this inhibition (Fig. 4C). Colony stimulating factor-1 receptor expression was induced in the spinal cord during EAE independently of treatment, although XPro1595 further enhanced its expression at Day 39 post-immunization (Fig. 4C). These results show that inhibition of soluble TNF and maintenance of transmembrane TNF with XPro1595 supports the expression of neuroprotective molecules in the spinal cord during EAE.

E. Taoufik et al.

EAE therapy by soluble TNF blockade

Brain 2011: 134; 2722–2735

| 2731

deprivation in neuron–astrocyte co-cultures (Fig. 6C), possibly through combined inhibition of astrocyte activation (and therefore beneficial transmembrane TNF production) and neuroprotection. Overall, these results show that the therapeutic effects of the soluble TNF blocker XPro1595 in EAE are directly dependent upon neuronal NF-B activity, and suggest that transmembrane TNF is a neuroprotective mediator that signals via neuronal NF-B to induce CNS tolerance and possibly also repair during the chronic phase of EAE.

Discussion TNF is constitutively expressed in the CNS, unlike in most other cells and tissues where TNF production is tightly regulated and inducible (Breder et al., 1993). During inflammation, local CNS production of TNF is further enhanced by infiltrating immune cells and activated glial cells. However, the functions of TNF and its receptors in the CNS during health and disease are still poorly

understood. To probe the different effects of soluble TNF and transmembrane TNF in the CNS during EAE, we used XPro1595, a dominant negative TNF inhibitor that selectively inhibits soluble TNF while preserving the function of transmembrane TNF (Steed et al. 2003; Zalevsky et al., 2007; Olleros et al., 2009) and compared its effects to those of a non-selective TNF inhibitor, etanercept. We found that XPro1595 significantly ameliorated the clinical symptoms of EAE in both prophylactic and therapeutic protocols, demonstrating that soluble TNF has disease-promoting effects. In contrast, the inhibition of both soluble TNF and transmembrane TNF by etanercept did not provide protection and even exacerbated disease, thus revealing a disease-protective function of transmembrane TNF. The therapeutic effects of XPro1595 were associated with reduced production of inflammatory mediators and sustained or induced levels of neuroprotection mediators in the brain and spinal cord tissues, and were dependent upon neuronal NF-B activity since mice lacking the NF-B regulator IKKb in glutamatergic neurons were not protected by XPro1595. The beneficial effects of XPro1595 were independent of changes in

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

Figure 5 Transmembrane TNF mediates neuroprotection in neuron–astrocyte co-cultures that is dependent on contact between the two cell types. (A) glucose deprivation-induced neuron death in astrocyte–neuron co-cultures that were incubated with XPro1595 (200 ng/ml), etanercept (100 ng/ml) or saline starting 4 h after the onset of glucose deprivation (GD). Neuron death is shown as the percentage of trypan blue (TB)-positive cells (dead) to total cell number in each experimental condition. (B) Glucose deprivation-induced neuron death in astrocyte–neuron co-cultures that have been pretreated for 24 h with human TNF (50 ng/ml) prior to glucose deprivation and incubated with XPro1595, etanercept or saline starting 4 h after the onset of glucose deprivation. (C) Glucose deprivation-induced neuron death in astrocyte–neuron co-cultures where astrocytes were separated from neurons by Transwells, treated with XPro1595, etanercept or saline starting 4 h after the onset of glucose deprivation. All in vitro experiments were performed at least three times and data from representative experiments are shown.

2732

| Brain 2011: 134; 2722–2735

in vivo in EAE is dependent upon neuronal NF-B signalling. (A) Clinical EAE scores (means  SEM) over time for nIKKbKO mice treated with XPro1595 (filled square; n = 9) or vehicle (filled triangle; n = 10) and IKKbF/F mice treated with XPro1595 (open square; n = 9) or vehicle (open triangle; n = 10). *P 5 0.05 for comparisons between vehicle- and XPro1595-treated mice of the same genotype. (B) Survival (Kaplan–Meier) curve showing viability of mice (groups described in A) during disease progression. (C) Neuron viability in astrocyte–neuron co-cultures that were pretreated for 2 h prior to glucose deprivation with the selective IKKb inhibitor, BMS 34 5541 (25 mM) and then treated with XPro1595 (200 ng/ml) or etanercept (100 ng/ml) starting 4 h after glucose deprivation. Data are from one representative experiment out of two performed.

antigen-specific immune responses and of early CNS infiltration and structural damage by inflammatory cells, indicating that transmembrane TNF acts by directly protecting CNS tissues rather than modulating CNS-directed immune responses. This was supported by the finding that XPro1595, but not etanercept, fully preserved neuroprotective TNF signalling against glucose deprivation in

astrocyte–neuron co-cultures and that this protection was dependent upon cell contact between the two cell types. Our results provide the first evidence that soluble TNF and transmembrane TNF are both functional in the CNS during EAE and have different roles; soluble TNF promotes innate CNS inflammation and clinical symptoms, while transmembrane TNF mediates neuroprotection and therapeutic benefits. This should help to resolve apparently conflicting reports describing deleterious and protective effects for TNF in CNS pathology (Shohami et al., 1999; Steinman and Zamvil, 2005). Importantly, it suggests that selective inhibition of soluble TNF could provide a promising new strategy in the treatment of multiple sclerosis and other inflammatory CNS disorders. Outside the CNS, previous studies of gene-targeted mice, in which the wild-type TNF gene is replaced by one expressing a functional, normally regulated, uncleavable transmembrane TNF protein (TNFKI) (Ruuls et al., 2001; Alexopoulou et al., 2006) or using dominant-negative TNF inhibitors that selectively block soluble TNF (Steed et al., 2003), have been critical for demonstrating that transmembrane TNF and soluble TNF mediate different functions. In transmembrane TNFKI mice, transmembrane TNF was sufficient to support several of the physiological and beneficial activities previously assigned to TNF, such as target cell killing through either TNF receptor I or TNF receptor II and normal splenocyte chemokine expression (Ruuls et al., 2001; Alexopoulou et al., 2006) and was partially able to support secondary lymphoid structure and host defense responses to Listeria monocytogenes and M. tuberculosis (Ruuls et al., 2001; Fremond et al., 2005; Saunders et al., 2005; Torres et al. 2005; Alexopoulou et al., 2006). In contrast, transmembrane TNF, in the absence of soluble TNF, did not induce the development of full-blown disease in models of endotoxic shock, arthritis or EAE (Ruuls et al., 2001; Alexopoulou et al., 2006) implying that soluble TNF plays a dominant role in the development of inflammation pathology. In a model of pulmonary inflammation, soluble TNF also played a detrimental role in mediating the transition from inflammation to fibrosis (Oikonomou et al., 2006). The use of dominant-negative TNF inhibitors confirmed and extended these results, showing that soluble TNF was necessary for the full clinical expression of autoimmune arthritis (Steed et al., 2003; Zalevsky et al., 2007) and endotoxin-induced liver injury (Olleros et al., 2010) but was not required for host immune responses to mycobacterial infections where the presence of a functional transmembrane TNF was critical (Olleros et al., 2009), thereby establishing these or similar reagents as promising therapeutic reagents for treatment of inflammatory diseases. Current understanding of TNF action in the CNS is characterized by the apparent paradox of opposing deleterious and beneficial effects of TNF in pathology that cannot be resolved easily on the basis of available knowledge of TNF and TNF receptor function. Disease-promoting, pro-inflammatory effects of TNF and both TNF receptor I and TNF receptor II receptors that are relevant to multiple sclerosis pathology were shown in early studies using TNF transgenic mice (Akassoglou et al., 1998, 2003) and EAE studies using TNF inhibitors or mice genetically deficient in TNF or its receptors (Ruddle et al., 1990; Selmaj et al., 1991; Baker et al., 1994; Frei et al., 1997; Korner et al., 1997). However, these

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

Figure 6 Transmembrane TNF-mediated neuroprotection

E. Taoufik et al.

EAE therapy by soluble TNF blockade

| 2733

TNF receptor II in mediating direct neuroprotection by transmembrane TNF as its level of expression on neurons may be below the level of detection by immunocytochemical techniques. Also, additional protective effects of soluble TNF blockade, such as the preservation of myelin integrity and compaction and the promotion of remyelination and axon preservation described by Brambilla and colleagues (2011), are also likely to be mediated by transmembrane TNF and might involve direct effects via TNF receptor II localized on oligodendrocytes and their precursors. The evidence here that soluble TNF and transmembrane TNF mediate different pro-inflammatory and neuroprotective roles respectively in EAE should have direct relevance for multiple sclerosis and suggests that selective soluble TNF inhibitors like XPro1595 might be useful in the treatment of multiple sclerosis and other inflammatory CNS disorders. This study revealed a direct neuroprotective role for transmembrane TNF in EAE and an intriguing dissociation between the beneficial effects of XPro1595 on neurological function and immune cell infiltration in the spinal cord. We have recently shown that NF-B in CamkII-expressing (glutamatergic) neurons is necessary for clinical remission, immunosuppression and neuroprotection in EAE (Emmanouil et al., 2009) and here we found that NF-B activity in these neurons was also necessary for the therapeutic effects of XPro1595, implying that brain neuronal systems significantly control the clinical and molecular features of CNS inflammatory diseases like EAE. Our data show that the neuroprotective effects of TNF that have been previously described in diverse types of neurodegeneration (Gary et al., 1998; Eugster et al., 1999; Marchetti et al., 2004; Taoufik et al., 2007; Lambertsen et al., 2009) may now extend to EAE and point to the possibility that transmembrane TNF might play a more general role in mediating CNS neuroprotection.

Acknowledgements We thank David Attwell for critical reading of the manuscript and Richard Ransohoff for helpful discussions. We also thank Dimitris Antoniou for artwork.

Funding This study was supported by the 6th Framework Program of the European Union through the integrated project entitled NeuroproMiSe (Grant LSHM-CT-2005-018637) and by the European Co-operation in Science and Technology (COST) Action entitled, Inflammation in Brain Disease (NEURINFNET) (Grant BM0603) through a short-term scientific mission grant (to V.T.).

Supplementary material Supplementary material is available at Brain online.

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

promising data from animal models, even coupled with the clinical success of TNF inhibitors in rheumatoid arthritis, were unable to predict the failure of non-selective TNF inhibitors in multiple sclerosis (van Oosten et al., 1997; The Lenercept Multiple Sclerosis Study Group and The University of British Columbia MS/MRI Analysis Group, 1999). More recently, a specific role for soluble TNF in inflammatory cell migration in CNS tissues during EAE and for the full development of clinical symptoms was demonstrated (Ruuls et al., 2001; Alexopoulou et al., 2006; Nomura et al., 2011). This was supported in the present study by the observation that disease onset was significantly delayed in mice treated prophylactically with either XPro1595 or etanercept. Also, a role for soluble TNF in neuro-inflammation and dopaminergic neuron loss in Parkinson’s disease models was shown using a similar dominant-negative inhibitor of soluble TNF to that used in this study (McCoy et al., 2006). In contrast, a few studies have detected disease-protective effects for TNF and its receptors during conditions of CNS inflammation. In one study, TNF deficiency converted EAE-resistant mouse strains to high susceptibility (Liu et al., 1998) and resulted in abnormally-prolonged clinical disease and T cell responses (Kassiotis and Kollias, 2001), while mice deficient in TNF receptor I showed enhanced CNS inflammation (Frei et al., 1997) and reduced clearance of T cells from lesions (Eugster et al., 1999). TNF and TNF receptor II promoted the proliferation of oligodendrocyte precursor cells and nerve remyelination in the cuprizone demyelination model (Arnett et al., 2001), indicating for the first time a specific beneficial role for transmembrane TNF, the primary ligand for TNF receptor II, in myelin maintenance. The neuroprotective effects of transmembrane TNF described in this study implicate both TNF receptor I and TNF receptor II in mediating the direct protection of neurons. Transmembrane TNF is known to signal efficiently through both TNF receptor I and TNF receptor II, while soluble TNF selectively signals through TNF receptor I, meaning that transmembrane TNF is the primary ligand for TNF receptor II (Grell et al., 1995). Studies in TNF receptor gene-targeted mice showed that TNF receptor I (Taoufik et al., 2007) and TNF receptor II (Marchetti et al., 2004) are capable of mediating TNF neuroprotection against ischaemic and excitotoxic injury in isolated mouse neurons and that in both cases this is associated with NF-B activation, so it remains to be determined whether the neuroprotective effects of transmembrane TNF in vivo are mediated through one or both TNF receptors. In this context it is interesting that in a parallel study to ours by Brambilla and colleagues (2011) (this issue), where similar therapeutic effects of soluble TNF blockade in EAE are described, localization studies of TNF receptor in spinal cord tissues from EAE mice and patients with multiple sclerosis revealed high levels of TNF receptor I on neurons, astrocytes and oligodendrocytes, while TNF receptor II was localized to oligodendrocytes and oligodendrocyte precursor cells (only mouse) and astrocytes, but not neurons. Collectively our data, combined with recent evidence that the gene for TNF receptor I is associated with multiple sclerosis (De Jager et al., 2009), provide strong support for the hypothesis that transmembrane TNF mediates direct neuroprotective signals via TNF receptor I in CNS inflammatory diseases like multiple sclerosis. However, we cannot exclude an additional involvement of

Brain 2011: 134; 2722–2735

2734

| Brain 2011: 134; 2722–2735

References

Eugster HP, Frei K, Bachmann R, Bluethmann H, Lassmann H, Fontana A. Severity of symptoms and demyelination in MOG-induced EAE depends on TNFR1. Eur J Immunol 1999; 29: 626–32. Eugster HP, Frei K, Kopf M, Lassmann H, Fontana A. IL-6-deficient mice resist myelin oligodendrocyte glycoprotein-induced autoimmune encephalomyelitis. Eur J Immunol 1998; 28: 2178–87. Filipovic R, Jakovcevski I, Zecevic N. GRO-alpha and CXCR2 in the human fetal brain and multiple sclerosis lesions. Dev Neurosci 2003; 25: 279–90. Frei K, Eugster HP, Bopst M, Constantinescu CS, Lavi E, Fontana A. Tumor necrosis factor alpha and lymphotoxin alpha are not required for induction of acute experimental autoimmune encephalomyelitis. J Exp Med 1997; 185: 2177–82. Fremond C, Allie N, Dambuza I, Grivennikov SI, Yeremeev V, Quesniaux VF, et al. Membrane TNF confers protection to acute mycobacterial infection. Respir Res 2005; 6: 136. Gary DS, Bruce-Keller AJ, Kindy MS, Mattson MP. Ischemic and excitotoxic brain injury is enhanced in mice lacking the p55 tumor necrosis factor receptor. J Cereb Blood Flow Metab 1998; 18: 1283–7. Ghosh S, Karin M. Missing pieces in the NF-kappaB puzzle. Cell 2002; 109 (Suppl): S81–96. Gold R, Linington C, Lassmann H. Understanding pathogenesis and therapy of multiple sclerosis via animal models: 70 years of merits and culprits in experimental autoimmune encephalomyelitis research. Brain 2006; 129: 1953–71. Gray E, Thomas TL, Betmouni S, Scolding N, Love S. Elevated activity and microglial expression of myeloperoxidase in demyelinated cerebral cortex in multiple sclerosis. Brain Pathol 2008; 18: 86–95. Grell M, Douni E, Wajant H, Lo¨hden M, Clauss M, Maxeiner B, et al. The transmembrane form of tumor necrosis factor is the prime activating ligand of the 80 kDa tumor necrosis factor receptor. Cell 1995; 83: 793–802. Kassiotis G, Kollias G. Uncoupling the proinflammatory from the immunosuppressive properties of tumor necrosis factor (TNF) at the p55 TNF receptor level: implications for pathogenesis and therapy of autoimmune demyelination. J Exp Med 2001; 193: 427–34. Korner H, Riminton DS, Strickland DH, Lemckert FA, Pollard JD, Sedgwick JD. Critical points of tumor necrosis factor action in central nervous system autoimmune inflammation defined by gene targeting. J Exp Med 1997; 186: 1585–90. Kreuz S, Siegmund D, Scheurich P, Wajant H. NF-kappaB inducers upregulate cFLIP, a cycloheximide-sensitive inhibitor of death receptor signaling. Mol Cell Biol 2001; 21: 3964–73. Krishnamoorthy G, Lassmann H, Wekerle H, Holz A. Spontaneous opticospinal encephalomyelitis in a double-transgenic mouse model of autoimmune T cell/B cell cooperation. J Clin Invest 2006; 116: 2385–92. Lambertsen KL, Clausen BH, Babcock AA, Gregersen R, Fenger C, Nielsen HH, et al. Microglia protect neurons against ischemia by synthesis of tumor necrosis factor. J Neurosci 2009; 29: 1319–30. Lassmann H. Hypoxia-like tissue injury as a component of multiple sclerosis lesions. J Neurol Sci 2003; 206: 187–91. Lewis M, Tartaglia LA, Lee A, Bennett GL, Rice GC, Wong GH, et al. Cloning and expression of cDNAs for two distinct murine tumor. Proc Natl Acad Sci USA 1991; 88: 2830–4. Li ZW, Omori SA, Labuda T, Karin M, Rickert RC. IKK beta is required for peripheral B cell survival and proliferation. J Immunol 2003; 170: 4630–7. Liu J, Marino MW, Wong G, Grail D, Dunn A, Bettadapura J, et al. TNF is a potent anti-inflammatory cytokine in autoimmune-mediated demyelination. Nat Med 1998; 4: 78–83. Mahad DJ, Ransohoff RM. The role of MCP-1 (CCL2) and CCR2 in multiple sclerosis and experimental autoimmune encephalomyelitis (EAE). Semin Immunol 2003; 15: 23–32. Marchetti L, Klein M, Schlett K, Pfizenmaier K, Eisel UL. Tumor necrosis factor (TNF)-mediated neuroprotection against glutamate-induced excitotoxicity is enhanced by N-methyl-D-aspartate receptor activation. Essential role of a TNF receptor 2-mediated phosphatidylinositol

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

Akassoglou K, Bauer J, Kassiotis G, Pasparakis M, Lassmann H, Kollias G, et al. Oligodendrocyte apoptosis and primary demyelination induced by local TNF/p55TNF receptor signaling in the central nervous system of transgenic mice: models for multiple sclerosis with primary oligodendrogliopathy. Am J Pathol 1998; 153: 801–13. Akassoglou K, Douni E, Bauer J, Lassmann H, Kollias G, Probert L. Exclusive tumor necrosis factor (TNF) signaling by the p75TNF receptor triggers inflammatory ischemia in the CNS of transgenic mice. Proc Natl Acad Sci USA 2003; 100: 709–14. Albensi BC, Mattson MP. Evidence for the involvement of TNF and NF-kappaB in hippocampal synaptic plasticity. Synapse 2000; 35: 151–9. Alexopoulou L, Kranidioti K, Xanthoulea S, Denis M, Kotanidou A, Douni E, et al. Transmembrane TNF protects mutant mice against intracellular bacterial infections, chronic inflammation and autoimmunity. Eur J Immunol 2006; 36: 2768–80. Ara´nguez I, Torres C, Rubio N. The receptor for tumor necrosis factor on murine astrocytes: characterization, intracellular degradation, and regulation by cytokines and Theiler’s murine encephalomyelitis virus. Glia 1995; 13: 185–94. Arnett HA, Mason J, Marino M, Suzuki K, Matsushima GK, Ting JP. TNF alpha promotes proliferation of oligodendrocyte progenitors and remyelination. Nat Neurosci 2001; 4: 1116–22. Baker D, Butler D, Scallon BJ, O’Neill JK, Turk JL, Feldmann M. Control of established experimental allergic encephalomyelitis by inhibition of tumor necrosis factor (TNF) activity within the central nervous system using monoclonal antibodies and TNF receptor-immunoglobulin fusion proteins. Eur J Immunol 1994; 24: 2040–8. Beattie EC, Stellwagen D, Morishita W, Bresnahan JC, Ha BK, Von Zastrow M, et al. Control of synaptic strength by glial TNFalpha. Science 2002; 295: 2282–5. Berezovskaya O, Maysinger D, Fedoroff S. Colony stimulating factor-1 potentiates neuronal survival in cerebral cortex ischemic lesion. Acta Neuropathol 1996; 92: 479–86. Botchkina GI, Meistrell ME, Botchkina IL, Tracey KJ. Expression of TNF and TNF receptors (p55 and p75) in the rat brain after focal cerebral ischemia. Mol Med 1997; 3: 765–81. Brambilla R, Ashbaugh JJ, Magliozzi R, Dellarole A, Karmally S, Szymkowski DE, et al. Inhibition of soluble tumour necrosis factor is therapeutic in experimental autoimmune encephalomyelitis and promotes axon preservation and remyelination. Brain 2011; 134: 2736–54. Breder CD, Tsujimoto M, Terano Y, Scott DW, Saper CB. Distribution and characterization of tumor necrosis factor-alpha-like immunoreactivity in the murine central nervous system. J Comp Neurol 1993; 337: 543–67. Camandola S, Mattson MP. NF-kappa B as a therapeutic target in neurodegenerative diseases. Expert Opin Ther Targets 2007; 11: 123–32. Centonze D, Muzio L, Rossi S, Cavasinni F, De Chiara V, Bergami A, et al. Inflammation triggers synaptic alteration and degeneration in experimental autoimmune encephalomyelitis. J Neurosci 2009; 29: 3442–52. Chen JW, Breckwoldt MO, Aikawa E, Chiang G, Weissleder R. Myeloperoxidase-targeted imaging of active inflammatory lesions in murine experimental autoimmune encephalomyelitis. Brain 2008; 131: 1123–33. De Jager PL, Jia X, Wang J, de Bakker PI, Ottoboni L, Aggarwal NT, et al. Meta-analysis of genome scans and replication identify CD6, IRF8 and TNFRSF1A as new multiple sclerosis susceptibility loci. Nat Genet 2009; 41: 776–82. Emmanouil M, Taoufik E, Tseveleki V, Vamvakas SS, Tselios T, Karin M, et al. Neuronal I kappa B kinase beta protects mice from autoimmune encephalomyelitis by mediating neuroprotective and immunosuppressive effects in the central nervous system. J Immunol 2009; 183: 7877–89.

E. Taoufik et al.

EAE therapy by soluble TNF blockade

| 2735

Ruuls SR, Hoek RM, Ngo VN, McNeil T, Lucian LA, Janatpour MJ, et al. Membrane-bound TNF supports secondary lymphoid organ structure but is subservient to secreted TNF in driving autoimmune inflammation. Immunity 2001; 15: 533–43. Santello M, Bezzi P, Volterra A. TNFalpha controls glutamatergic gliotransmission in the hippocampal dentate gyrus. Neuron 2011; 69: 988–1001. Saunders BM, Tran S, Ruuls S, Sedgwick JD, Briscoe H, Britton WJ. Transmembrane TNF is sufficient to initiate cell migration and granuloma formation and provide acute, but not long-term, control of Mycobacterium tuberculosis infection. J Immunol 2005; 174: 4852–9. Schmidt D, Textor B, Pein OT, Licht AH, Andrecht S, Sator-Schmitt M, et al. Critical role for NF-kappaB-induced JunB in VEGF regulation and tumor angiogenesis. EMBO J 2007; 26: 710–19. Selmaj K, Raine CS, Cross AH. Anti-tumor necrosis factor therapy abrogates autoimmune demyelination. Ann Neurol 1991; 30: 694–700. Shohami E, Ginis I, Hallenbeck JM. Dual role of tumor necrosis factor alpha in brain injury. Cytokine Growth Factor Rev 1999; 10: 119–30. Simi A, Tsakiri N, Wang P, Rothwell NJ. Interleukin-1 and inflammatory neurodegeneration. Biochem Soc Trans 2007; 35: 1122–6. Steed PM, Tansey MG, Zalevsky J, Zhukovsky EA, Desjarlais JR, Szymkowski DE, et al. Inactivation of TNF signaling by rationally designed dominant-negative TNF variants. Science 2003; 301: 1895–8. Steinman L, Zamvil SS. Virtues and pitfalls of EAE for the development of therapies for multiple sclerosis. Trends Immunol 2005; 26: 565–71. Stellwagen D, Beattie EC, Seo JY, Malenka RC. Differential regulation of AMPA receptor and GABA receptor trafficking by tumor necrosis factor-alpha. J Neurosci 2005; 25: 3219–28. Stubgen JP. Tumor necrosis factor-alpha antagonists and neuropathy. Muscle Nerve 2008; 37: 281–92. Taoufik E, Valable S, Mu¨ller GJ, Roberts ML, Divoux D, Tinel A, et al. FLIP(L) protects neurons against in vivo ischemia and in vitro glucose deprivation-induced cell death. J Neurosci 2007; 27: 6633–46. Taylor PC, Feldmann M. Anti-TNF biologic agents: still the therapy of choice for rheumatoid arthritis. Nat Rev Rheumatol 2009; 5: 578–82. The Lenercept Multiple Sclerosis Study Group and The University of British Columbia MS/MRI Analysis Group. TNF neutralization in MS: results of a randomized, placebo-controlled multicenter study. Neurology 1999; 53: 457–65. Thompson WL, Van Eldik LJ. Inflammatory cytokines stimulate the chemokines CCL2/MCP-1 and CCL7/MCP-3 through NFkB and MAPK dependent pathways in rat astrocytes. Brain Res 2009; 1287: 47–57. Torres D, Janot L, Quesniaux VF, Grivennikov SI, Maillet I, Sedgwick JD, et al. Membrane tumor necrosis factor confers partial protection to Listeria infection. Am J Pathol 2005; 167: 1677–87. van Oosten BW, Barkhof F, Truyen L, Boringa JB, Bertelsmann FW, von Blomberg BM, et al. Increased MRI activity and immune activation in two multiple sclerosis patients treated with the monoclonal anti-tumor necrosis factor antibody cA2. Neurology 1996; 47: 1531–4. Wajant H, Pfizenmaier K, Scheurich P. Tumor necrosis factor signaling. Cell Death Differ 2003; 10: 45–65. Zalevsky J, Secher T, Ezhevsky SA, Janot L, Steed PM, O’Brien C, et al. Dominant-negative inhibitors of soluble TNF attenuate experimental arthritis without suppressing innate immunity to infection. J Immunol 2007; 179: 1872–83.

Downloaded from brain.oxfordjournals.org at Hellenic PASTEUR Institute on September 12, 2011

3-kinase-dependent NF-kappa B pathway. J Biol Chem 2004; 279: 32869–81. McCoy MK, Martinez TN, Ruhn KA, Szymkowski DE, Smith CG, Botterman BR, et al. Blocking soluble tumor necrosis factor signaling with dominant-negative tumor necrosis factor inhibitor attenuates loss of dopaminergic neurons in models of Parkinson’s disease. J Neurosci 2006; 26: 9365–75. Minichiello L, Korte M, Wolfer D, Kuhn R, Unsicker K, Cestari V, et al. Essential role for TrkB receptors in hippocampus-mediated learning. Neuron 1999; 24: 401–14. Miyagishi R, Kikuchi S, Takayama C, Inoue Y, Tashiro K. Identification of cell types producing RANTES, MIP-1 alpha and MIP-1 beta in rat experimental autoimmune encephalomyelitis by in situ hybridization. J Neuroimmunol 1997; 77: 17–26. Mohan N, Edwards ET, Cupps TR, Oliverio PJ, Sandberg G, Crayton H, et al. Demyelination occurring during anti-tumor necrosis factor alpha therapy for inflammatory arthritides. Arthritis Rheum 2001; 44: 2862–9. Murray KM, Dahl SL. Recombinant human tumor necrosis factor receptor (p75) Fc fusion protein (TNFR:Fc) in rheumatoid arthritis. Ann Pharmacother 1997; 31: 1335–8. Nicole O, Ali C, Docagne F, Plawinski L, MacKenzie ET, Vivien D, et al. Neuroprotection mediated by glial cell line-derived neurotrophic factor: involvement of a reduction of NMDA-induced calcium influx by the mitogen-activated protein kinase pathway. J Neurosci 2001; 21: 3024–33. Nomura T, Abe Y, Kamada H, Shibata H, Kayamuro H, Inoue M, et al. Therapeutic effect of PEGylated TNFR1-selective antagonistic mutant TNF in experimental autoimmune encephalomyelitis mice. J Control Release 2011; 149: 8–14. Oikonomou N, Harokopos V, Zalevsky J, Valavanis C, Kotanidou A, Szymkowski DE, et al. Soluble TNF mediates the transition from pulmonary inflammation to fibrosis. PLoSOne 2006; 1: e108. Olleros ML, Vesin D, Fotio AL, Santiago-Raber ML, Tauzin S, Szymkowski DE, et al. Soluble TNF, but not membrane TNF, is critical in LPS-induced hepatitis. J Hepatol 2010; 53: 1059–68. Olleros ML, Vesin D, Lambou AF, Janssens JP, Ryffel B, Rose S, et al. Dominant-negative tumor necrosis factor protects from Mycobacterium bovis Bacillus Calmette Guerin (BCG) and endotoxininduced liver injury without compromising host immunity to BCG and Mycobacterium tuberculosis. J Infect Dis 2009; 199: 1053–63. Omari KM, John G, Lango R, Raine CS. Role for CXCR2 and CXCL1 on glia in multiple sclerosis. Glia 2006; 53: 24–31. Park JM, Greten FR, Li ZW, Karin M. Macrophage apoptosis by anthrax lethal factor through p38 MAP kinase inhibition. Science 2002; 297: 2048–51. Ransohoff RM, Hamilton TA, Tani M, Stoler MH, Shick HE, Major JA, et al. Astrocyte expression of mRNA encoding cytokines IP-10 and JE/ MCP-1 in experimental autoimmune encephalomyelitis. FASEB J 1993; 7: 592–600. Ruddle NH, Bergman CM, McGrath KM, Lingenheld EG, Grunnet ML, Padula SJ, et al. An antibody to lymphotoxin and tumor necrosis factor prevents transfer of experimental allergic encephalomyelitis. J Exp Med 1990; 172: 1193–200.

Brain 2011: 134; 2722–2735