OUP user menu

Fumaric acid esters exert neuroprotective effects in neuroinflammation via activation of the Nrf2 antioxidant pathway

Ralf A. Linker, De-Hyung Lee, Sarah Ryan, Anne M. van Dam, Rebecca Conrad, Pradeep Bista, Weike Zeng, Xiaoping Hronowsky, Alex Buko, Sowmya Chollate, Gisa Ellrichmann, Wolfgang Brück, Kate Dawson, Susan Goelz, Stefan Wiese, Robert H. Scannevin, Matvey Lukashev, Ralf Gold
DOI: http://dx.doi.org/10.1093/brain/awq386 678-692 First published online: 25 February 2011


Inflammation and oxidative stress are thought to promote tissue damage in multiple sclerosis. Thus, novel therapeutics enhancing cellular resistance to free radicals could prove useful for multiple sclerosis treatment. BG00012 is an oral formulation of dimethylfumarate. In a phase II multiple sclerosis trial, BG00012 demonstrated beneficial effects on relapse rate and magnetic resonance imaging markers indicative of inflammation as well as axonal destruction. First we have studied effects of dimethylfumarate on the disease course, central nervous system, tissue integrity and the molecular mechanism of action in an animal model of chronic multiple sclerosis: myelin oligodendrocyte glycoprotein induced experimental autoimmune encephalomyelitis in C57BL/6 mice. In the chronic phase of experimental autoimmune encephalomyelitis, preventive or therapeutic application of dimethylfumarate ameliorated the disease course and improved preservation of myelin, axons and neurons. In vitro, the application of fumarates increased murine neuronal survival and protected human or rodent astrocytes against oxidative stress. Application of dimethylfumarate led to stabilization of the transcription factor nuclear factor (erythroid-derived 2)-related factor 2, activation of nuclear factor (erythroid-derived 2)-related factor 2-dependent transcriptional activity and accumulation of NADP(H) quinoline oxidoreductase-1 as a prototypical target gene. Furthermore, the immediate metabolite of dimethylfumarate, monomethylfumarate, leads to direct modification of the inhibitor of nuclear factor (erythroid-derived 2)-related factor 2, Kelch-like ECH-associated protein 1, at cysteine residue 151. In turn, increased levels of nuclear factor (erythroid-derived 2)-related factor 2 and reduced protein nitrosylation were detected in the central nervous sytem of dimethylfumarate-treated mice. Nuclear factor (erythroid-derived 2)-related factor 2 was also upregulated in the spinal cord of autopsy specimens from untreated patients with multiple sclerosis. In dimethylfumarate-treated mice suffering from experimental autoimmune encephalomyelitis, increased immunoreactivity for nuclear factor (erythroid-derived 2)-related factor 2 was detected by confocal microscopy in neurons of the motor cortex and the brainstem as well as in oligodendrocytes and astrocytes. In mice deficient for nuclear factor (erythroid-derived 2)-related factor 2 on the same genetic background, the dimethylfumarate mediated beneficial effects on clinical course, axon preservation and astrocyte activation were almost completely abolished thus proving the functional relevance of this transcription factor for the neuroprotective mechanism of action. We conclude that the ability of dimethylfumarate to activate nuclear factor (erythroid-derived 2)-related factor 2 may offer a novel cytoprotective modality that further augments the natural antioxidant responses in multiple sclerosis tissue and is not yet targeted by other multiple sclerosis therapies.

  • neuroprotection
  • astrocyte
  • transcription factor
  • oxidative stress
  • autoimmune demyelination
  • multiple sclerosis
  • oral therapy


In recent years, significant advances in the development of disease modifying drugs for relapsing–remitting multiple sclerosis have been made. Yet, the therapeutic efficacy of all available modalities is still incomplete and all hitherto licensed compounds require parenteral application. Thus, there is a high unmet need for the development of highly effective and orally available treatment options. In view of the neurodegenerative features of multiple sclerosis that correlate with permanent disability in chronic phases of the disease (Wujek et al., 2002), neuroprotective treatment approaches are highly warranted. While several oral immunomodulators or immunsuppressors are currently under investigation in phase III multiple sclerosis studies (Linker et al., 2008a), none of these compounds demonstrated neuroprotective properties or long-term safety.

This study explores whether fumaric acid esters may provide novel effects on pathomechanisms of multiple sclerosis. In earlier studies, an oral formulation of fumaric acid esters (Fumaderm®) was shown to be effective in patients with psoriasis, a Th1 mediated skin disease (Mrowietz et al., 1998). In a pilot study involving relapsing–remitting patients with multiple sclerosis, this formulation reduced the number of gadolinium enhancing lesions on brain MRI scans (Schimrigk et al., 2006). Recently, the efficacy and safety of BG00012, a new oral formulation of dimethylfumarate, was demonstrated in a multicentre, randomized, double-blind, placebo-controlled, dose-ranging, phase IIb study in patients with relapsing–remitting multiple sclerosis (Kappos et al., 2008). BG00012 also reduced black holes that are indicative of tissue destruction due to persistent demyelination and axonal loss. In vitro experiments and data from dermatology studies revealed that dimethylfumarate, as well as its primary metabolite monomethylfumarate, can exert immunomodulatory effects on T cell subsets as well as on antigen presenting cells (Linker et al., 2008b). In particular, fumaric acid esters may influence the expression of molecules that are involved in inflammatory cascades. In dermatologic in vitro studies, these effects comprised a Th2 shift, a reduction of proinflammatory cytokines such as interleukin (IL)-2, TNF-α, as well as a downregulation of adhesion molecules like ICAM-1 or E-selectin (de Jong et al., 1996; Vandermeeren et al., 1997; Ockenfels et al., 1998; Loewe et al., 2002; Litjens et al., 2004). At higher dosages, fumaric acid esters were also shown to induce apoptosis in vitro (Treumer et al., 2003).

In the acute phase of experimental autoimmune encephalomyelitis (EAE), earlier studies have shown that treatment with dimethylfumarate results in a significant reduction of macrophage/microglia infiltration, but no significant effects on T cells (Schilling et al., 2006). Here we investigate the effects of dimethylfumarate in the chronic phase of myelin oligodendrocyte glycoprotein (MOG) induced EAE, a model of multiple sclerosis that also displays neurodegenerative features (Gold et al., 2006; Herrero-Herranz et al., 2008). Our data reveal that at the later stage of the disease, fumaric acid esters exert neuroprotective effects dependent on the function of the nuclear-factor (erythroid-derived 2) related factor-2 (Nrf2) oxidative stress response pathway.

Materials and methods

Murine motoneuron survival assay and human astrocyte viability assay

Embryonic Day 14 mouse motoneurons were isolated as described before (Wiese et al., 2010) with the modification that the p75NTR panning antibody was substituted by 10 µg/ml Lectin (Sigma, Deisenhofen, Germany). The resulting purified cells were >92% islet-positive cells. These cells were plated in Greiner 4-well culture dishes at a density of 3000 cells/well. Ciliary neurotrophic factor (Biomol, Hamburg, Germany) was added at 10 ng/ml, 1 h after plating. Initial counting of plated cells was carried out when all cells were attached to the bottom of the well. Human spinal cord astrocytes were purchased from ScienCell Research Laboratories (Carlsbad, CA, USA) and maintained according to supplier specifications. For viability assays, astrocytes were plated into Corning poly-l-lysine coated clear bottom 96-well plates. Oxidative challenge of motoneurons with H2O2 was performed as described earlier (Su et al., 1999). Briefly, cells were cultured in the presence of monomethylfumarate, at various concentrations, for 24 h. After incubation with 50 µM H2O2 for 1 h, cells were cultured for another 24 h before immunostaining for caspase-3 as marker for apoptosis (anti-caspase-3 antibody C8487, Sigma). Total cells on the plate and caspase-3 immunoreactive cells were counted in a blinded manner and data expressed as percent reduction of caspase-3 positive cells in comparison with addition of H2O2 alone. Human spinal cord astrocytes were cultured in the presence of monomethylfumarate at various concentrations for 24 h, and then challenged with 500 μM H2O2 for 2 h. Cells were then washed and incubated with normal growth media for an additional 24 h. Astrocyte viability was assessed using the Live/Dead Viability/Cytotoxicity kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. Calcein acetoxymethylester cleavage by live cells was measured using a FlexStation II plate reader (Molecular Devices, Silicon Valley, CA, USA). Relative fluorescence from wells receiving no H2O2 was normalized to 100%, and all other data were expressed as a percent of control ± SD. Cells were imaged on an Axio Observer Z1 inverted fluorescence microsope (Zeiss, Thornwood, NJ, USA).


C57/BL6 mice were obtained from Harlan Laboratories (Harlan Winkelmann, Borchen, Germany), Nrf2 knockout mice were from Riken Laboratories (Japan) and backcrossed to the C57BL/6 background for more than 10 generations (Itoh et al., 1997). Mice were kept under pathogen-free conditions. All experiments were approved by the North-Rhine-Westphalia authorities for animal experimentation.

Induction of MOG-EAE and treatment protocol

Ten-week-old female mice received a subcutaneous injection of 200 µg MOG 35–55 peptide (Charité, Berlin, Germany) emulsified in complete Freund`s adjuvant containing 200 µg Mycobacteria tuberculosis (Difco/BD Biosciences, Heidelberg, Germany). On Days 0 and 2 post immunization, 200 ng pertussis toxin (List/Quadratec, UK) were applied by intraperitoneal injection. Mice were weighted and scored for clinical signs daily according to a 10-scale score as described by Linker et al. (2002). dimethylfumarate was applied in a preventive setting starting from Day 0 post injection at a dosage of 15 mg/kg body weight twice a day via oral gavage, as described earlier (Schilling et al., 2006). Control animals received methocel (methylcellulose) orally twice a day as sham treatment. For treatment in a therapeutic setting starting on Day 18 post injection, pairs of mice with matched scores (n = 7 per group) were treated with dimethylfumarate (15 mg/kg) or methocel as control and followed until Day 36 post injection.


After perfusion with 4% paraformaldehyde (Sigma) on Days 41 and 74 post injection, the spinal cord and spleen were removed and embedded in paraffin. Routine stainings comprised Luxol Fast Blue for myelin and Bielschowksy silver impregnation for axons. Immunohistochemistry was performed on 3 µm thick paraffin sections as described to label T cells (α-CD3 1:200; Serotec; Wiesbaden, Germany), macrophages (α-Mac-3 1:200; BD), astrocytes (α-GFAP 1:1000; DAKO, Hamburg, Germany), myelin/oligodendrocytes (CNPase 1:200; Millipore, Schwalbach, Germany), oligodendrocytes (α-NogoA 1:100; Santa Cruz Biotechnologies, Heidelberg, Germany) and Nrf2 (1:75; Santa Cruz, Biotechnologies). Neurons were identified by cresyl violet staining or immunohistochemistry for NeuN (1:200; Millipore). Visualization was performed by diaminobenzidine for light microscopy or with anti-mouse Alexa 488 (1:1000) and Alexa 647 (1:1000) conjugated secondary antibodies and confocal laser scanning microscopy (Zeiss). Omitting the primary antibody or pre-absorption with a specific peptide (Nrf2) served as negative control.

Spinal cord tissue from six multiple sclerosis autopsy cases and two control cases without any neurological disease were stained with the Nrf2 antibody (Santa Cruz, Biotechnologies; dilution 1:25) after pre-treatment of the sections with 1% Triton and microwave (850 W for 38 min). Visualization was performed with diaminobenzidine.

Blinded quantification of axons and immune cells was performed on serial sections and included six independent spinal cord cross sections per mouse. Immune cells were counted by means of overlaying a stereological grid onto sections from and counting inflammatory infiltrates per mm2 white matter. Quantification of demyelinated areas on Luxol Fast Blue or CNPase stained sections was performed semi-automatically with the help of CellD Software (Olympus, Hamburg, Germany). Areas with complete demyelination and homogeneous immune cell infiltration were identified as lesion. For quantification of axons in lesions, silver impregnated profiles were counted on six visual fields involving cervical, thoracic and lumbar spinal cord using a grid 100 μm diameter (modified from Mews et al., 1998). Data are presented as relative axonal densities.

Proliferation assay and cytokine measurement

For spleen cell proliferation assays, single cell suspensions of spleen from dimethylfumarate or methocel treated mice were prepared 9 days after immunization of mice with 200 µg MOG 35–55. Cells (2 × 105) were seeded in 96-well microtitre plates (Nunc, Wiesbaden, Germany) in 100 µl medium with addition of antigen or mitogen. Concentrations were 20 µg/ml for MOG and 1.25 µg/ml for ConA. Triplicate cultures were maintained at 37°C in a humidified atmosphere with 5% CO2 for 56 h and harvested following a 16 h pulse with 0.2 µCi/well 3H-dT (tritiated thymidine, Amersham-Buchler, Braunschweig, Germany). Cells were collected on fibreglass filter paper with a 96-well harvester (Pharmacia, Freiburg, Germany), and radioactivity was measured with a 96-well Betaplate liquid scintillation counter (Pharmacia). Supernantants were harvested after 3 days of culture. IFN-γ concentrations were determined by sandwich enzyme linked immunosorbent assay (BD, Heidelberg, Germany), as described earlier (Linker et al., 2010).

Western blotting and small interfering RNA experiments

Cellular levels of Nrf2 were monitored in total cell lysates by western blotting using rabbit polyclonal Nrf2 antibodies (courtesy of Dr Cecil Picket, or C20, Santa Cruz Biotechnologies). For small interfering RNA mediated depletion of Nrf2, cells were transfected with Nrf2-specific siGENOME SMARTPool small interfering RNA pool at 100–200 nM using Lipofectamine 2000 (Invitrogen) 24 h prior to dimethylfumarate simulation. Nrf2 depletion efficiency was verified by quantitative polymerase chain reaction and western blotting. Control and dimethylfumarate-treated astrocyte cultures were briefly rinsed with cold phosphate buffered saline and lysed in the PureLink RNA lysis buffer (Invitrogen). Total RNA was isolated using the PureLink RNA mini silica column kits (Invitrogen) according to the manufacturer's instructions. Quantitative polymerase chain reaction was performed in coupled reverse transcriptase–polymerase chain reactions (100 ng total RNA input per sample) using commercial TaqMan probe sets (Applied Biosystems). FAM-labelled probesets Hs00252524_m1 and Hs00168547_m1 were used to detect AKR1B10 and NQO1, respectively. GAPDH (VIC-labelled GAPDH probeset Hs03929097_g1) was used as internal housekeeping normalization target. Relative target quantitation and fold-change calculation were performed using the standard ΔΔCt method.

Analysis of NADP(H) quinoline oxidoreductase-1 activity

Cerebellum and liver were prepared from rats immunized with MOG or incomplete Freund`s adjuvant after intraperitoneal treatment with 200 mg/kg dimethylfumarate or methocel as control. Tissue was homogenized in ice cold 25 mM Tris–HCl and 1 mM EDTA (pH 7.4). After 5 s of sonification, the homogenates were centrifuged. The enzymatic activity of NQO-1 was determined spectrophotometrically by measuring the NADPH dependent, menadiol-mediated reduction of 3-(4,5-dimethylthiazol-2yl)-2,5-diohenyltetrazolium bromide as described (Prochaska et al., 1988) with some slight modifications (van Muiswinkel et al., 2000).

Luciferase assay

A stable reporter cell line was prepared by transfection of colon carcinoma d-lactate dehydrogenase 1 cells with a luciferase reporter plasmid harbouring eight copies of the glutathione-S-transferase 2 antioxidative response elements cloned upstream of the luciferase complementary DNA into the pGL4.26 vector. The reporter cells were stimulated with dimethylfumarate for 24 h and luciferase activity was determined in the cell lysates using the Promega luciferase detection kit (Promega, Madison, WI, USA).

Mass spectrometry analysis of monomethylfumarate interaction with Keap1

C-terminal V5-tagged rat Kelch-like ECH-associated protein 1 (Keap1) (Nioi et al., 2005) was transfected into 293 cells (ATCC) using Lipofectamine 2000 (Invitrogen). Forty-eight hours after transfection, cells were treated with 30 µM dimethylfumarate for 6 h. Samples were lysed in radioimmune precipitation buffer [137 mM NaCl, 20 mM Tris, 1 mM MgCl2, 1 mM CaCl2, 1% NP-40, 0.5% desoxycholate, 0.1% sodium dodecyl sulphate, pH 7.4, complete protease inhibitor cocktail (Roche, Basel, Switzerland), 1 mM sodium orthovanadate, 10 mM sodium fluoride], and Keap1-V5 was immunoprecipitated from clarified supernatants using immobilized anti-V5 agarose beads (Sigma). Five washes were performed using spin columns (Thermo Scientific, Rockford, IL, USA), and protein was eluted from the column using hot Laemmli- sodium dodecyl sulphate buffer under reducing conditions. The samples were run on 4–12% Bis–Tris gradient gel (Invitrogen), with MES [2-(N-morpholino)ethanesulfonic acid] buffer, stained (0.2% Coomassie blue, 0.5% methanol, 20% ethanol), destained (30% methanol), and then finally stored in 0.5% acetic acid for band retrieval and analysis by mass spectroscopy. The 65 kD rat Keap1 gel bands were excised into ∼1 × 1 mm slices. The gel slices were destained with 50 mM ammonium bicarbonate/methanol (1:1 v/v) for 30 min and then dehydrated with acetonitrile. The dried gel slices were rehydrated and reduced with 20 mM dithiothreitol in 25 mM ammonium bicarbonate/acetonitrile (9:1 v/v) at 50°C for 45 min. The solution was discarded and 50 mM iodoacetamide in 25 mM ammonium bicarbonate/acetonitrile (9:1 v/v) was added and incubated at dark for 30 min for cysteine alkylation. The alkylation solution was discarded and the gel slices were washed with water/acetonitrile (1:1), followed by dehydration with acetonitrile. The resulting gel pieces were then digested with 0.3 µg of trypsin in 25 mM ammonium bicarbonate/acetonitrile (9:1 v/v) 0.01% ProteasMax (Prozyme, Hayward, CA, USA) at 37°C for 3.5 h. N-glycanase (2.5 mU) was then added and the solution was incubated at 37°C overnight. The peptide digestion product was extracted with water/acetonitrile (9:1 v/v) 5% formic acid.

Liquid chromatography–tandem mass spectrometry

The peptide pools were analysed using a Thermo Fisher Scientific LTQ-FT ICR Ultra Hybrid high resolution mass spectrometer, which was equipped with a nano-spray source and in-line with a Dionex nanoHPLC system The peptide pools were separated by a C18 column (0.075 × 150 mm) using a linear gradient from 5 to 55% solvent B over 120 min and then 55–75% over 5 min with solvent A, water/0.1% formic acid/0.005% heptafluorobutyric acid, and solvent B, acetonitrile/0.1% formic acid/0.005% heptafluorobutyric acid. The mass spectrometer was operated in a data-dependent MS/MS mode. These data were processed using SpectrumMill on rat Keap1 protein sequences. The sites of Keap1 alkylation were identified based on the database search and manual inspection of MS/MS spectra.

Statistical analyses

For statistical evaluation of the clinical course, data were pooled from different experiments. Analysis was performed using Mann–Whitney U test for clinical courses and histological quantifications. Significant differences in cell viability were assessed using ANOVA analysis followed by Dunnett’s Multiple Comparison test. All data are given as mean values ± SEM. P-values were considered significant at *P < 0.05 and highly significant at **P < 0.01 or *** P < 0.001.


Beneficial effects of dimethylfumarate in chronic autoimmune demyelination are associated with preservation of myelin and axons

In a phase IIb study in relapsing–remitting multiple sclerosis, dimethylfumarate displayed beneficial effects on numbers of hypointense T1 lesions, a parameter that is putatively associated with axonal destruction (Kappos et al., 2008). Thus, we were interested in investigating the protective potential of dimethylfumarate during chronic autoimmune demyelination. In a preventive setting, we found that dimethylfumarate treatment (15 mg/kg body weight) of MOG immunized C57BL/6 mice ameliorated the clinical course of EAE until Day 74 post injection. In a pool of four experiments, beneficial effects became especially evident during the chronic phase of the disease (e.g. between Days 40 and 74 post injection). During this period, mice treated with vehicle suffered from severe paraparesis while dimethylfumarate-treated mice showed only gait ataxia (Fig. 1A). In another EAE experiment, dimethylfumarate treatment started after the first maximum of disease on Day 18 post injection. Similar to the preventive approach described above, dimethylfumarate exerted a significant beneficial effect that became apparent 10 days after start of treatment (Fig. 1B, Supplementary Fig. 1).

Figure 1

Clinical course of chronic EAE under dimethylfumarate treatment. (A) Preventive dimethylfumarate treatment (grey curve) significantly ameliorates the course of disease as compared with application of methocel as control (black curve, ***P < 0.001). Data represent the mean of four independent experiments (n = 36/42 mice per group) ± SEM. (B) Therapeutic application of dimethylfumarate (grey curve) after the first maximum of disease (Day 18 post injection, marked by arrows) significantly ameliorates the course of disease as compared with application of methocel as control (black curve, *P < 0.05; n = 7 per group). p.i. = post injection; DMF = dimethylfumarate.

In order to determine the cellular changes governing the beneficial clinical effect, especially in the chronic phase of EAE, we performed a histopathological analysis of the spinal cord on Day 74 post injection, after preventive therapy with dimethylfumarate. Interestingly, there were no differences in the number of infiltrating T cells and macrophages/microglia (Fig. 2A–D; Table 1) which is in contrast to observations during the early phase of MOG-EAE (Schilling et al., 2006). To further elaborate on the effects of dimethylfumarate on T cells, we performed a T cell proliferation assay and analysed the interferon gamma production in a MOG peptide-recall assay ex vivo (Supplementary Fig. 2). In these experiments, no differences in T cell proliferation and interferon gamma production were observed after in vivo application of dimethylfumarate as compared with control mice receiving the carrier only. In contrast to these immunological data and upon blinded analysis of demyelinated areas, we noted a significant preservation of myelin after dimethylfumarate treatment with a reduction of demyelination by ∼60% in comparison with vehicle-treated controls (Fig. 2E and F; Table 1). Silver impregnation revealed a significant preservation of axons in inflamed lesions with twice as much silver impregnated profiles present after dimethylfumarate treatment (Fig. 2G and H; Table 1). Finally, application of dimethylfumarate resulted in a reduced activation of astrocytes as measured by glial fibrillary acidic protein staining both in the white matter (data not shown) as well as grey matter of spinal cord cross sections (Fig. 2I and J; Table 1).

Figure 2

Dimethylfumarate treatment results in myelin and axonal preservation as well as reduction of astrocyte activation. Representative images of thoracic spinal cord cross sections on Day 74 post injection are shown. Serial sections from the same animal and anatomical region are depicted for the different stainings. (A and B) No difference in Mac-3 positive macrophages and microglia could be detected between dimethylfumarate treated mice and controls. (C and D) Similarly, there was no difference in numbers of infiltrating CD3 positive T cells. (E and F) In contrast, Luxol Fast Blue staining revealed a preservation of myelin after dimethylfumarate treatment, arrows in (E) mark demyelinated areas. (G and H) Silver impregnation studies disclosed a preservation of axonal densities after dimethylfumarate treatment. (I and J) In the grey matter, dimethylfumarate therapy resulted in a reduced activation of astrocytes (arrows mark activated astrocytes). Bar = 200 µm for A, B, and F and 50 µm for C, D and G–J. DMF = dimethylfumarate.

View this table:
Table 1

Dimethylfumarate treatment in chronic MOG-EAE does not influence inflammatory infiltration, but leads to preservation of myelin and axons as well as reduced astrocyte activation

Macrophages/microglia (Mac-3+ cells/mm2 ± SEM)307 ± 30.0329 ± 38.5n.s.
T cells (CD3+ cells/mm2 ± SEM)114 ± 17.8156 ± 25.4n.s.
Demyelination (% of white matter ± SEM)12.7 ± 1.64.5 ± 0.9P = 0.0001
Axonal density ± SEM2.1 ± 0.44.2 ± 0.7P = 0.02
Activated astrocytes (GFAP+ cells/mm2 ± SEM)176.2 ± 13.6117 ± 13.5P = 0.005
  • Blinded quantification of CD3 positive T cells, Mac-3 positive macrophages/microglia, demyelination after Luxol Fast Blue staining, axonal numbers after silver impregnation and astrocyte activation after glial fibrillary acidic protein (GFAP) staining were performed on Day 74 p.i. Data are presented as cells/mm2 for CD3 and Mac-3 positive cells as well as glial fibrillary acidic protein positive activated astrocytes, percent demyelination for Luxol Fast Blue staining and relative axonal densities for silver impregnated axons. n.s = not significant.

In summary, dimethylfumarate treatment led to persistent beneficial effects during chronic autoimmune demyelination, which were associated with apparent neuroprotective effects involving myelin, axons and astrocytes.

Dimethylfumarate treatment results in a reduced loss of central nervous system neurons and glial cells in experimental autoimmune encephalomyelitis

Since dimethylfumarate appeared to preserve myelin and axons in the spinal cord during an autoimmune attack, we were interested if dimethylfumarate also exerts effects on neurons in the CNS. To this end, we performed an independent MOG-EAE experiment and analysed the number of neurons in brainstem and cerebellar nuclei on Day 41 post injection. At this time point, there was clear inflammation in the brainstem and cerebellum in both dimethylfumarate-treated and control animals. In particular, direct contact between mononuclear immune cells and neurons could be delineated (Fig. 3C). Cresyl violet staining and blinded quantification of neuronal cells in the spinal cord and cerebellar nuclei revealed significantly higher numbers of neurons in dimethylfumarate-treated mice (Fig. 3A, B, D and E). This enhanced number of neurons in dimethylfumarate-treated mice was selective for spinal cord, brainstem and cerebellum, and was not present in the midbrain or cerebellar cortex (data not shown). Finally, we investigated a possible mechanism underlying the observed dimethylfumarate-mediated neuronal preservation by performing immunohistochemistry for nitrotyrosine residues as an indicator for oxygen radical induced damage in the CNS. After dimethylfumarate treatment, there was a significant reduction of nitrotyrosine immunoreactivity in EAE lesions of the spinal cord (Fig. 3F and G).

Figure 3

Dimethylfumarate treatment leads to neuronal preservation in vivo. (A and B) Representative cresyl violet staining of cerebellar nuclei neurons after treatment with methocel (A) or dimethylfumarate (B) on Day 41 of MOG-EAE are shown (bar 100 µM). Arrows indicate intact neurons. Note the higher number of neurons after dimethylfumarate treatment. (C) High power photograph of neuron in close contact with mononuclear cell (arrow), bar 20 µM. (D) Blinded quantification of cerebellar nuclei neurons revealed a preservation of neuronal numbers after dimethylfumarate treatment. **P < 0.01. (E) Blinded quantification of spinal cord alpha motoneurons revealed a preservation of neuronal numbers after dimethylfumarate treatment. *P < 0.05. (F and G) Staining for nitrotyrosine residues indicating free radical induced cell stress. Representative spinal cord cross sections are shown, bar 100 µM. Note the increased immunoreactivity for nitrotyrosine in lesions of methocel-treated control mice (marked by arrows). DMF = dimethylfumarate.

Fumarates rescue neurons and glia from oxidative stress-induced cell death

We then investigated whether the observed neuroprotective effects of fumaric acid esters were only indirect or rather due to direct effects on neurons or glial cells. To analyse the direct neuroprotective capacity, the direct metabolite of dimethylfumarate, monomethylfumarate, was tested in a murine motoneuron survival assay and in human astrocyte culture after induction of oxidative stress by addition of H2O2. While the application of H2O2 significantly enhanced neuronal death, treatment of murine motoneuron cultures with monomethylfumarate at a dose of 1 µg/ml led to a significant reduction of caspase-3 positive neurons as indicator of apoptosis (Fig. 4A) while 10 µg/ml failed to show beneficial effects. To shed further light on the role of Nrf2 in neuronal anti-oxidant pathways, we performed a motoneuron survival assay using Nrf2 deficient motoneurons. In this assay, the Nrf2 deficient cells were hypersensitive to H2O2 leading to immediate cell death in all groups including those with monomethylfumarate treatment (data not shown). Moreover, monomethylfumarate was tested in human primary astrocyte culture. monomethylfumarate itself did not show any cytotoxic effects (data not shown), but caused a dose-dependent protection with almost complete survival from H2O2-induced cell death at highest concentrations (Fig. 4B and C).

Figure 4

Fumaric acid ester treatment leads to direct neuroprotection from oxidative stress in vitro. (A) A motoneuron survival assay after challenge with 50 µM H2O2 is shown. Motoneurons were cultured in the presence of monomethylfumarate (dark grey and black bar), addition of ciliary neurotrophic factor (CNTF) (10 µg/ml) served as a positive control (light grey bar). Addition of monomethylfumarate at 1 µg/ml was able to significantly support motoneuron survival 24 h after wash-out of H2O2. Data are pooled from three independent experiments; data are shown as percent reduction of caspase-3 immunoreactive cells ± SEM. *P < 0.05. (B and C) monomethylfumarate protects astrocytes against oxidative injury and promotes cell viability. Primary human spinal cord astrocytes were treated for 24 h with a titration of monomethylfumarate, then challenged with 500 µM H2O2 for 2 h. Cells were labelled with a live/dead labelling kit after a 24 h wash-out of H2O2. Live cells were labelled in green and nuclei of dying cells in red, representative cell culture images are shown (B). Quantification of total green fluorescence indicating live cells reveals a dose-dependent protective effect of monomethylfumarate in astroglial cultures (C). Data represented as percent of no H2O2 control, ±SD, **P < 0.01. MMF = monomethylfumarate.

Fumarates activate the Nrf2 oxidative stress response pathway in vitro and in vivo

dimethylfumarate is a known inducer of phase II detoxifying enzymes (Begleiter et al., 2003; Wierinckx et al., 2005) and both dimethylfumarate and its primary metabolite monomethylfumarate are thiol-reactive electrophiles (Schmidt et al., 2007). This combination of properties suggests that fumarates could activate the Nrf2 transcriptional pathway known to mediate induction of phase II genes by thiol-reactive electrophiles and play a major role in cell and tissue defence against oxidative stress. To determine whether dimethylfumarate and monomethylfumarate could activate Nrf2, we first analysed their effects on Nrf2, its inhibitor Keap1, and Nrf2-dependent gene expression in cultured cells. In human and rodent astrocyte cultures, both dimethylfumarate and monomethylfumarate led to a concentration-dependent increase in cellular levels of Nrf2 (Fig. 5A, Nrf2 upregulation is shown separately for human, mouse and rat astrocytes) accompanied by accumulation of prototypical Nrf2 target genes such as NADP(H) quinoline oxidoreductase-1 (NQO-1) or aldo–keto reductase family 1 member B10 (AKR1B10), as shown in human astrocytes on the messenger RNA level (Fig. 5B) or in a cell line on the protein level (Supplementary Fig. 3). To verify the functional dependence of increased NQO1 expression on Nrf2, we performed small interfering RNA mediated depletion of Nrf2 prior to stimulation of cells with dimethylfumarate. Depletion of Nrf2 from dimethylfumarate stimulated cells prevented dimethylfumarate-induced accumulation of NQO-1 (Fig. 5C). Functional activation of Nrf2 was further verified in reporter assays showing concentration-dependent activation of Nrf2-dependent antioxidative response elements-mediated transcriptional activity in dimethylfumarate-treated reporter cells. Upon stimulation of the cells with dimethylfumarate at concentrations ranging from 1.5–15 µM, we observed a strong induction of antioxidative response elements-driven luciferase expression in a dose-dependent manner (Fig. 5D). Pharmacokinetic analyses reveal that application of dimethylfumarate at a dose of 15 mg/kg body weight, mimicking the approach in human trials, results in a maximal concentration of 4.4 µM monomethylfumarate in CNS tissue (data not shown). This is exactly in the range as used in the in vitro experiments and supports the in vivo relevance of our findings. In order to further elucidate whether these in vitro effects of dimethylfumarate may also play a role in vivo, studies were performed in EAE, an animal model of multiple sclerosis. In comparison with application of vehicle, treatment with dimethylfumarate led to a significant induction of NQO-1 activity in liver and cerebellum of both MOG and adjuvant immunized control rats (Fig. 5E and F). Finally, we investigated direct effects of fumaric acid esters on the inhibitor of Nrf2, Keap1 in vitro. Mass spectroscopy revealed a covalent modification at the cysteine residue 151 of the Keap1 protein after monomethylfumarate treatment (Fig. 5G). The matched Keap1 peptides (peptide score >7) from a SpectrumMill database search covered 95% of the rat Keap1 protein sequence. After stimulation with monomethylfumarate, a 130 Da mass increase of the Keap1 tryptic peptide CVLHVMNGAVMYQIDSVVR was observed. The mass increase of 130 Da is consistent with a Michael addition of a free cysteine sulphydryl group across the monomethylfumarate double bond. The MS/MS spectrum of the peptide confirms this modification at cysteine 151 (Fig. 5G). The assignment of 130 Da to the monomethylfumarate modification is consistent (within 0.14 mmu) with a chemical formula of C5H6O4 (130.02661 exp. and 130.02647 cal.). The ion intensity ratio of the modified and unmodified peptides are ∼69 and 31%, respectively. The monomethylfumarate induced 130 Da modification was also observed––to a much lower extent––in other cysteine-containing Keap1 peptides (Table 2).

Figure 5

Fumaric acid esters activate Nrf2 dependent anti-oxidative pathways in vitro. (A) Upon stimulation of cells with 5–50 µM dimethylfumarate and monomethylfumarate, a western blot analysis revealed a dose-dependent upregulation of Nrf2 in human, mouse and rat astrocyte culture. Different species in the graph are separated by dashed lines. (B) Reverse transcriptase–polymerase chain reaction analysis of the Nrf2 dependent target genes NADP(H) quinoline oxidoreductase-1 (NQO-1) or aldo-keto reductase family 1 member B10 (AKR1B10) reveals a significant up-regulation of both genes after treatment of astrocytes with 30 µM dimethylfumarate. (C) Small interfering RNA experiment. Small interfering RNA mediated inhibition of Nrf2 expression in dimethylfumarate stimulated cell culture resulted in a complete inhibition of NQO-1 induction thus proving the direct regulation of Nrf2 by dimethylfumarate. (D) Luciferase assay experiments. Upon addition of dimethylfumarate at dosages from 1.5–15 µM, there was a strong, dose-dependent induction of antioxidative response elements activation in a reporter cell line. (E and F) dimethylfumarate mediated regulation of NQO-1 activity in vivo in immunized animals. In comparison with application of vehicle, treatment with dimethylfumarate led to a significant induction of NQO-1 enzyme activity in liver and cerebellum of both MOG/CFA, and also incomplete Freund’s adjuant only immunized rats. *P < 0.05. (G) Mass spectrometry showing modification of the Keap1 cysteine residue cys151 by monomethylfumarate (modifications at 130 Da are marked by arrows).

View this table:
Table 2

Keap1 peptides modified by 130 Da supported by MS/MS tandem mass spectrometry analysis

Observed m/zKEAP1 Observed MMF Cys-modified peptidesPercent MMF modification
  • MMF = monomethylfumarate.

In summary, fumaric acid ester treatment leads to the induction of the Nrf2 mediated anti-oxidative pathways via direct interaction of monomethylfumarate with Keap1.

Dimethylfumarate leads to induction of Nrf2 in glial cells and neurons

After having established that dimethylfumarate is a potent inducer of Nrf2 mediated, anti-oxidative pathways in vitro and that dimethylfumarate conferred neuroprotective effects in MOG-EAE in vivo, we were interested in the analysis of Nrf2 expression during autoimmune demyelination. On Day 74 of MOG-EAE, we analysed the expression of Nrf2 in the spinal cord by immunohistochemistry. In these experiments, the specificity of Nrf2 immunohistochemistry was proven by pre-adsorption with an Nrf2 peptide (Supplementary Fig. 4). After dimethylfumarate treatment, there was a clear increase of Nrf2 immunoreactivity in the brain, especially in the grey matter. In the spinal cord, dimethylfumarate treatment led to an increased Nrf2 immunoreactivity of grey matter interneurons, as defined by morphological criteria (Supplementary Fig. 4). In contrast, alpha-motoneurons in the anterior horn of naïve mice were slightly Nrf2 positive (data not shown). Nrf2 expression in motoneurons was augmented by the induction of EAE and even further increased by dimethylfumarate treatment (Supplementary Fig. 4). To further delineate Nrf2 expressing cell-types in and around inflammatory lesions, we performed laser scanning confocal imaging after staining for Nrf2 and different cell markers. Double labelling of Nrf2 with markers for macrophages/microglia and T cells did not disclose an increased Nrf2 expression in those cell types after dimethylfumarate treatment (data not shown). In contrast, double staining of Nrf2 with the neuronal marker NeuN disclosed that dimethylfumarate treatment led to a clear increase in Nrf2 immunoreactivity in different neuronal subpopulations of the brain, especially in the cerebellar or brainstem nuclei and in the motor cortex (Fig. 6A–D). Moreover, double labelling of Nrf2 with NogoA as marker for oligodendrocytes and glial fibrillary acidic protein as marker for astrocytes also identified an increase in Nrf2 immunoreactivity in oligodendroglia as well as astroglia after treatment with dimethylfumarate (Fig. 6E–H). This effect was more prominent and ubiquitous in oligodendrocytes while only selected astrocytes in dimethylfumarate-treated mice were Nrf2 positive. In summary, dimethylfumarate treatment in chronic MOG-EAE leads to an increase in free Nrf2 in glial cells and neuronal subpopulations that are relevant for motor function and project to spinal cord pathways. To verify these findings for its relevance in multiple sclerosis, spinal cord sections from six multiple sclerosis autopsy cases and controls were stained for Nrf2. All patients with multiple sclerosis (male to female ratio 1:1; mean age 46 ± 4.5, mean disease duration 9.0 ± 1.3 years, mean expanded disability status scale 7.8 ± 0.2) were in the chronic-progressive disease stage of the disease and not treated with fumarates. In control spinal cord, alpha-motoneurons expressed Nrf2 (Fig. 6I). In multiple sclerosis spinal cord, Nrf2 immunoreactivity of alpha-motoneurons was increased. In addition, grey matter interneurons were also Nrf2-positive in multiple sclerosis spinal cord, as was described above in EAE (Fig. 6J).

Figure 6

Dimethylfumarate leads to induction of Nrf2 in glial cells and neurons in EAE. Representative sections of motorcortex (A and B), brainstem (C and D), spinal cord grey matter (E and F) and spinal cord white matter (G and H) from methocel (left) or dimethylfumarate (right) treated mice are shown. (A–D) Arrows mark Nrf2 double labelled cells with dimethylfumarate therapy and Nrf2 negative cells with methocel application. Dimethylfumarate led to a clearly increased Nrf2 immunoreactivity (red) in nuclei of NeuN positive neurons of the motorcortex and brainstem (green). (E and F) dimethylfumarate resulted in increased Nrf2 immunoreactivity in glial fibrillary acidic protein positive astrocytes in the spinal cord grey matter (astrocytes marked in green). (G and H) dimethylfumarate also led to increased Nrf2 immunoreactivity in nuclei of NogoA positive oligodendrocytes (oligodendrocytes marked in green). (I and J) Nrf2 expression in control and multiple sclerosis spinal cord. Representative sections from control (I) and multiple sclerosis (J) spinal cord. There is basal expression of Nrf2 in control alpha motoneurons (I), marked by arrows. In multiple sclerosis spinal cord, Nrf2 expression is increased in alpha motoneurons and additionally seen in spinal cord interneurons (J) see arrows. Bar 20 µM for (A–F) and 100 µM for (G) and (H). MS = multiple sclerosis.

Neuroprotective efficacy of dimethylfumarate during autoimmune demyelination is dependent on Nrf2

The observation that dimethylfumarate led to induction of the Nrf2 pathway in vitro and in vivo prompted us to determine if functional Nrf2 was required for therapeutic efficacy of dimethylfumarate in chronic autoimmune demyelination. To this end, we employed Nrf2 knockout mice on a C57BL/6 background. At a young age, these mice are grossly normal, show no overt immunological abnormalities, but are reported to be hypersensitive to toll-like receptor stimulation (Thimmulappa et al., 2006). In our study, these mutant mice displayed a significantly more severe course of MOG-EAE after immunization with 200 µg MOG/200 µg CFA and suffered from severe tetraparesis while wild-type mice only showed a mild paraparesis (Supplementary Fig. 5), which is in line with earlier results (Johnson et al., 2010). To delineate effects of dimethylfumarate in these MOG-sensitive mice, we developed an attenuated protocol with immunization of 200 µg MOG/ 50 µg CFA on Days 0 and 30 post injection. With this protocol, there was only a mild difference in the severity of MOG-EAE between Nrf2 knockout mice and wild-type controls until the early chronic phase of MOG-EAE (Day 41 post injection). Both groups suffered from different degrees of paraparesis. Yet, upon oral dimethylfumarate treatment at 15 mg/kg body weight twice a day, solely wild-type mice responded to therapy and only displayed gait ataxia while no therapeutic effect was seen in Nrf2 knockout mice (Fig. 7A). We then performed a histopathological analysis on Day 41 post injection. Compared with sham treated wild-type mice, axonal densities in demyelinated lesions were significantly reduced in Nrf2 knockout mice while demyelination and astrocyte activation was attenuated in dimethylfumarate-treated wild-type mice (Fig. 7B–D). dimethylfumarate treatment failed to preserve axon densities or reduce astrocyte activation in Nrf2-deficient mice but resulted in axon protection and reduction of astrocyte activation in wild-type mice (Fig. 7B–D). Analyses of inflammatory infiltration revealed no differences in infiltrating T cells (Table 3). Nrf2 knockout mice were characterized by an enhanced infiltration of macrophages and microglia (Table 3), which was not modulated by dimethylfumarate treatment. In summary, these data strongly suggest a role for Nrf2 in dimethylfumarate-mediated axon protection independent from inflammatory infiltration.

Figure 7

Neuroprotective activity of dimethylfumarate during autoimmune demyelination is dependent on Nrf2. (A) Clinical course of chronic EAE under dimethylfumarate treatment. EAE was induced with an attenuated protocol to allow for studying the chronic disease phase in Nrf2−/− mice. The clinical course after a boost immunization is shown. Dimethylfumarate treatment (grey curve) significantly ameliorates the course of disease only in wild-type mice (blue curves), but not in Nrf2−/− mice (red curves, P < 0.01). Data are pooled from three independent experiments (n = 12/11/14/15 mice), error bars represent SEM. (B–D) In a blinded analysis of axonal densities (B), demyelination (C) and astrocyte activation (D) on spinal cord cross sections, treatment with dimethylfumarate results in preservation of axons and myelin as well as reduced astrocyte activation only in wild-type (black bars), but not Nrf2−/− mice (white bars). *P < 0.05; **P < 0.01; ***P < 0.001. WT = wild-type.

View this table:
Table 3

Blinded quantification of inflammatory infiltration after dimethylfumarate treatment of Nrf2 knockout mice

Wild-type methocelWild-type DMFNrf2−/− methocelNrf2−/− DMFP-value
CD3399 ± 36334 ± 51440 ± 35459 ± 37n.s.
Mac-31036 ± 1071097 ± 1141587 ± 1271905 ± 92P < 0.001 for methocel groups P < 0.01 for DMF groups
  • Analysis of CD3 positive T cells and Mac-3 positive macrophages/microglia was performed on Day 41 p.i. Data are presented as cells/mm2 ± SEM. While Nrf2 knockout mice display an enhanced macrophage infiltration, dimethylfumarate neither influences immune cell infiltration in wild-type or knockout mice in chronic EAE. DMF = dimethylfumarate.


Here we show that fumarates exert neuroprotective effects dependent on Nrf2 mediated anti-oxidative pathways. In vitro, monomethylfumarate, the primary metabolite of dimethylfumarate, protected cultured neurons and astrocytes from H2O2 induced cell death and we provide evidence by mass spectrometry that monomethylfumarate leads to direct modification of Keap1 at cysteine residue 151. In MOG-EAE, dimethylfumarate exerts protective effects on oligodendrocytes, myelin, axons and neurons in vivo and reduces oxidative stress as measured by protein nitrosylation. Activation of the Nrf2 pathway by dimethylfumarate or monomethylfumarate and the evidence that Nrf2 function is required for the therapeutic effect of dimethylfumarate suggest that the CNS-protective effects of dimethylfumarate involve activation of Nrf2-mediated oxidative stress response mechanisms previously implicated as important for protection of the CNS in a variety of pathological conditions. The relevance of our findings is underlined by studies on multiple sclerosis autopsy tissue showing that neuronal Nrf2 is also activated during the natural course of multiple sclerosis, similar to our findings in untreated EAE. This observation is in line with recently published data by van Horssen et al. (2010).

Protective effects of dimethylfumarate on neurons in cell culture have been described in earlier studies (Duffy et al., 1998; Su et al.,1999). Other studies describing deprivation of glutathione or toxicity of dimethylfumarate and monomethylfumarate in glial cultures used much higher, unphysiological concentrations of both compounds (Thiessen et al., 2010). Moreover, our data provide evidence for dimethylfumarate-mediated neuroprotective effects in chronic MOG-EAE in vivo and additionally provide a mechanism that involves regulation of the anti-oxidative response element via Nrf2. The selective regulation of Nrf2 in neuronal populations of the spinal cord, brainstem and motor cortex may reflect different expression levels of Nrf2 and its inhibitor Keap-1. Several earlier in vitro and in vivo studies revealed that activation of Nrf2 pathways may exert neuroprotective effects (Lee et al., 2003; Shih et al., 2005a; Satoh et al., 2006, 2009a). Ex vivo studies and analyses of neurodegenerative models for motoneuron disorders and Parkinson’s disease or cerebral ischaemia disclosed that Nrf2-mediated neuroprotection may not only be directly mediated, but critically involves effects via astrocytes (Kraft et al., 2004; Shih et al., 2005b; Chen et al., 2009). This notion is further underscored by the observation that the expression of NQO1, an Nrf2 regulated enzyme involved in cellular detoxification, is predominantly present in glial cells. Since NQO-1 is one of several genes involved in the anti-oxidative tissue response, it’s upregulation is representative of a part of the anti-oxidative cascade that plays a role in the effects of dimethylfumarate in EAE.

Moreover, the importance of Nrf2 in glial cells is evidenced in Nrf2 knockout mice that display astrogliosis and myelinopathy in the cerebellum (Hubbs et al., 2007). Nrf2 knockout mice also suffer from a more severe MOG-EAE with increased oxidative damage in the CNS finally leading to enhanced demyelination and a more pronounced axonal loss (Johnson et al., 2010). These data are consistent with the effects of dimethylfumarate treatment in MOG-EAE, where dimethylfumarate application may lead to direct Nrf2 activation, not only in different neuronal populations, but also in oligodendrocytes and astrocytes. This dimethylfumarate-mediated regulation of Nrf2 is associated with reduced astrocyte gliosis as well as preservation of myelin, axons and neurons in wild-type mice. In contrast, this protective effect is not present in Nrf2 knockout mice, thus providing evidence for the functional effect of dimethylfumarate being Nrf2 dependent.

While the exact sequence of molecular events leading to fumaric acid ester-mediated activation of Nrf2 has yet to be clarified, we provide evidence that monomethylfumarate leads to direct modification of Keap1 at cysteine residue 151. It has been shown that activation of Nrf2 results from covalent modification of free cysteine residues in the Nrf2-binding adaptor protein Keap1 that targets Nrf2 for ubiquitin-mediated degradation leading to suppression of Nrf2 function (Rachakonda et al., 2008). Modification of cysteine 151 by electrophiles renders Keap1 incapable of interacting with Nrf2 and thus leads to stabilization of Nrf2, its accumulation in the nucleus and activation of induction of Nrf2-dependent expression of antioxidant and cytoprotective genes.

Our data reveal a dimethylfumarate-mediated regulation of Nrf2 in vitro and in vivo during MOG-EAE. Several earlier studies elaborated on other inducers of Nrf2 which may also exert neuroprotective effects. Such compounds include, among others, ceftriaxone (Lewerenz et al., 2009), ortho/parahydroquinones (Satoh et al., 2009b), sulforaphane (Danilov et al., 2009), 2-cyano-N-methyl-3,12-dioxooleana-1,9(11)-dien-28 amide (Yang et al., 2009), chitosan (Khodagholi et al., 2010) or lindenyl acetate (Li et al., 2009). Further compounds targeting the Nrf2 pathway may be detected via proteomic or genomic screening approaches (Liu et al., 2007). While all these substances may to some extent exert real neuroprotection, fumaric acid esters are additionally characterized by a very beneficial side effect profile not only in animal models, but also in humans with psoriasis or multiple sclerosis (Kappos et al., 2008).

To date, several potentially neuroprotective therapeutic approaches have been described in EAE models. These include targeting of TNF-related apoptosis inducing ligand-mediated cell death (Aktas et al., 2005), NF-κB inhibition via the green tea compound epigallotechicin (Aktas et al., 2004), inhibition of sodium or potassium channels (Bechtold et al., 2004; Bittner et al., 2009), or antagonists of kainate activated glutamate receptors (Pitt et al., 2000). dimethylfumarate may be added to these compounds as another interesting neuroprotective treatment approach whose anti-oxidative effects offer a novel mechanism of action and additionally, has long-term safety data in humans.

In addition to its neuroprotective capacity, dimethylfumarate may also possess immunomodulatory effects, which were at least proposed in several dermatological analyses in vitro. In acute MOG-EAE, we did not observe Th2 modulating effects and the reduction of infiltrating macrophages at this early disease stage (Schilling et al., 2006) may also be secondary to reduced tissue damage via dimethylfumarate-mediated enhanced cell viability. In conjunction with these studies and the present data, we propose a dual mechanism of action for dimethylfumarate in autoimmune demyelination: targeting immune cells on one hand and CNS cells on the other. To further explore the neuroprotective potential of dimethylfumarate, additional studies in models of primary neurodegenerative diseases are warranted.


This work was supported by an unrestricted research grant from Biogen Idec and by intramural funds from Ruhr University Bochum.

Conflict of interest

S.R., P.B., W.Z., X.H., A.B., S.C., K.D., S.G., R.H.S. and M.L. are employees of BiogenIdec; the company involved in marketing of BG-12 (dimethylfumarate). R.L. and R.G. received travel support, speaker's honoraria and research grants from BiogenIdec. R.G. also has board activities for BiogenIdec.

Supplementary material

Supplementary material is available at Brain online.


We thank Silvia Seubert for expert technical assistance, Christiane Reick for help with oral gavage and Prof. Masayuki Yamamoto, Tsukuba, Japan for providing Nrf2 knockout mice. The original enthusiasm of Prof. Altmeyer and Prof. Przuntek, Bochum in introducing FAE to therapy of MS is gratefully acknowledged. We are indebted to Dr. Cecil Pickett, BiogenIdec, Cambridge, MA, USA, for providing an anti-Nrf2 antibody.


  • *These authors contributed equally to this work.

  • Present address: Department of Neurology, University Erlangen, Germany

experimental autoimmune encephalomyelitis
Kelch-like ECH-associated protein 1
myelin oligodendrocyte glycoprotein
NADP(H) quinoline oxidoreductase-1
nuclear-factor (erythroid-derived 2)-related factor-2


View Abstract