Introduction
Recent studies suggest a major role of oxidative injury in the pathogenesis of demyelination and neurodegeneration in multiple sclerosis (MS). This view is based on the presence of oxidized lipids, proteins and DNA in MS lesions [
4,
43], in particular in oligodendrocytes within active lesions [
21] and in neurons within areas of severe active cortical demyelination and neurodegeneration [
15]. Oxidative injury is associated with the presence of activated microglia-expressing enzymes involved in oxygen radical production such as NADPH oxidases [
14,
15] and myeloperoxidase [
20]. Additionally, oxidative tissue damage may be aggravated via the Haber–Weiss Fenton reactions catalyzed by redox-active iron, which has been shown to be present in and around a subset of classical active and slowly expanding lesions, respectively [
22]. Furthermore, mitochondrial injury, which is prominent within MS lesions and the normal-appearing cortex [
8,
29], may be induced by oxidative injury but can also amplify oxidative stress [
32]. The data suggest that in early MS, oxidative injury may be driven by inflammation and radical production in activated microglia and amplified in the progressive stage of MS by age- and disease burden-related additional mechanisms [
27,
30].
A prominent defense mechanism counteracting oxidative stress is the induction of the transcription factor nuclear factor (erythroid-derived 2)-like 2 (Nrf2; encoded by the
NFE2L2 gene), which triggers the expression of a variety of anti-oxidative defense molecules when translocated into the nucleus [
9,
10]. Experimental autoimmune encephalomyelitis was more severe in Nrf2-deficient mice in comparison with wild-type littermates [
24], and induction of Nrf2 expression by treatment with fumarate was protective in this animal model [
28]. For these reasons and since the induction of endogenous anti-oxidant defense mechanisms might be insufficient in MS lesions, therapeutic trials testing dimethyl fumarate for its neuroprotective effects have been initiated and in part already conducted in MS patients [
16,
18,
37,
39]. So far, results show a reduction of relapses and magnetic resonance imaging (MRI) activity in patients with relapsing–remitting MS (RRMS), which was associated with a significant reduction of sustained disability progression in one of the two phase III trials.
Nrf2 has been described to be consistently up-regulated in multiple sclerosis plaques, but it was only detected in astrocytes and macrophages [
41] and in some spinal cord neurons [
28]. It has not been detected in oligodendrocytes so far, which are the prime target for oxidative damage within demyelinating lesions. In our present study, we report a high nuclear Nrf2 reactivity in oligodendrocytes in actively demyelinating lesions in patients with acute, relapsing as well as progressive disease. Most extensive Nrf2 expression was present in degenerating cells. Our observations raise doubts that further stimulation of Nrf2 by fumarate-induced cell stress protects oligodendrocytes in active lesions.
Materials and methods
The study was performed on formalin-fixed paraffin-embedded (FFPE) autopsy brain tissue derived from 28 controls without neurological disease and 23 MS cases (Table
1). The MS cohort included 6 acute MS (AMS) cases defined by a clinical course leading to death within 12 months after disease onset. One patient died during disease exacerbation in the RRMS stage. Nine patients presented with secondary progressive MS (SPMS) and seven with primary progressive MS (PPMS). The control cases included patients without neurological disease and neuropathologically detectable lesions in the central nervous system. In addition we included 2 brain biopsies of progressive multifocal leukoencephalopathy (PML) patients, who were treated daily with 4 (patient 1) and 5 (patient 2) tablets of Fumaderm (120 mg dimethyl fumarate and 95 mg monomethyl fumarate) until the last day before diagnostic brain biopsy. For control reasons, we included 2 brain biopsies from PML patients without fumarate therapy (Table
1). To determine the effect of systemic inflammation on Nrf2 expression in the brain, the control cohort also included patients, who died under septic conditions. Clinical and pathological information on the patients is summarized in Table
1. This study was approved by the ethics committee of the Medical University of Vienna (EK No. 048/01/2014).
Table 1
Clinical demographics
MS 1 | AMS | 78 | M | 2 | – | No | MSR |
MS 2 | AMS | 34 | F | 4 | – | S | MSR |
MS 3 | AMS | 45 | M | 0.2 | – | S | MSR |
MS 4 | AMS | 69 | F | 2 | – | S | MSR |
MS 5 | AMS | 35 | M | 1.5 | – | S | MSR |
MS 6 | AMS | 45 | M | 1,5 | – | S | MSR |
MS 7 | RRMS | 40 | F | 120 | – | S | UK |
MS 8 | SPMS | 41 | M | 137 | 99 | Mit, IFNβ, S | CV |
MS 9 | SPMS | 34 | M | 120 | n.a. | No | UK |
MS 10 | SPMS | 42 | F | 241 | 87 | No | UK |
MS 11 | SPMS | 53 | F | 241 | 104 | S | P |
MS 12 | SPMS | 62 | F | 144 | n.a. | S | P |
MS 13 | SPMS | 48 | F | 410 | 182 | No | UK |
MS 14 | SPMS | 46 | F | 444 | 228 | No | CV |
MS 15 | SPMS | 56 | M | 372 | 132 | No | CV |
MS 16 | SPMS | 59 | F | 492 | 168 | S | CV |
MS 17 | PPMS | 55 | F | 60 | 60 | No | PE |
MS 18 | PPMS | 36 | M | 61 | 61 | No | PE |
MS 19 | PPMS | 71 | F | 264 | 264 | No | PE |
MS 20 | PPMS | 53 | M | 168 | 168 | No | CV |
MS 21 | PPMS | 77 | F | 168 | 168 | No | CV |
MS 22 | PPMS | 67 | M | 87 | 87 | No | CV |
MS 23 | PPMS | 54 | F | 72 | 72 | No | CV |
Control 1 | Control | 39 | F | – | – | – | |
Control 2 | Control | 36 | F | – | – | – | |
Control 3 | Control | 46 | M | – | – | – | |
Control 4 | Control | 45 | F | – | – | – | |
Control 5 | Control | 47 | F | – | – | – | |
Control 6 | Control | 29 | F | – | – | – | |
Control 7 | Control | 57 | M | – | – | – | |
Control 8 | Control | 42 | F | – | – | – | |
Control 9 | Control | 46 | M | – | – | – | |
Control 10 | Control | 72 | M | – | – | – | |
Control 11 | Control | 65 | M | – | – | – | |
Control 12 | Control | 67 | M | – | – | – | |
Control 13 | Control | 73 | M | – | – | – | |
Control 14 | Control | 71 | F | – | – | – | |
Control 15 | Control | 80 | F | – | – | – | |
Control 16 | Control | 84 | F | – | – | – | |
PML + fumarate 1 | PML | 77 | M | – | – | Fumarates for psoriasis | Brain biopsy |
PML + fumarate 2 | PML | 68 | M | – | – | Fumarates for psoriasis | Brain biopsy |
PML control 1 | PML | 60 | M | – | – | – | Brain biopsy |
PML control 2 | PML | 85 | M | – | – | – | Brain biopsy |
Septic control 1 | Septic control | 97 | F | – | – | – | |
Septic control 2 | Septic control | 37 | M | – | – | – | |
Septic control 3 | Septic control | 71 | F | – | – | – | |
Septic control 4 | Septic control | 45 | M | – | – | – | |
Septic control 5 | Septic control | 74 | M | – | – | – | |
Septic control 6 | Septic control | 51 | F | – | – | – | |
Septic control 7 | Septic control | 70 | M | – | – | – | |
Septic control 8 | Septic control | 71 | M | – | – | – | |
Septic control 9 | Septic control | 95 | F | – | – | – | |
Septic control 10 | Septic control | 42 | F | – | – | – | |
Septic control 11 | Septic control | 89 | F | – | – | – | |
Septic control 12 | Septic control | 79 | F | – | – | – | |
Neuropathology and immunohistochemistry
All disease and control autopsy cases underwent detailed neuropathological examination of multiple tissue blocks covering various brain regions. Lesion activity was evaluated as previously described [
17,
26]. Routine immunohistochemistry was performed on paraffin sections according to established techniques [
2,
25]. For a detailed list of primary antibodies, dilutions and respective pre-treatment of tissue sections see Table
2. For Nrf2 stainings, we used two different primary antibodies, which resulted in very similar staining patterns. However, in combination with FFPE tissue, the antibody from Abcam showed a bit higher degree of sensitivity and less background staining and was, thus, used for quantitative analysis of Nrf2 expression. Cells with nuclear DNA fragmentation were identified via TUNEL staining [
19].
Table 2
Primary antibodies and antigen retrieval
CA II | Sheep (pAB) | Carbonic anhydrase II | PC076; BindingSite | 1:1500; E |
CD68 | Mouse (mAB) | Cluster of differentiation 68 | M0814; Dako | 1:100; E |
E06 | Mouse (mAB) | Oxidized phospholipids | | 10 µg/ml; none |
GFAP | Mouse (mAB) | Glial fibrillary acid protein | 0410080; ThermoSc | 1:200; E |
HLA-DP, DQ, DR | Mouse (mAB) | Human leukocyte antigen | M0775; Dako | 1:100; E |
HO-1 | Rabbit (pAB) | Heme oxygenase-1 | ab13243; Abcam | 1:2000; CSA; C |
KEAP1 | Mouse (mAb) | Kelch-like ECH-associated protein 1 | NBP2-03319, Novus Biologicals | 1:250; E |
MAG | Mouse (mAB) | Myelin-associated glycoprotein | ab89780; Abcam | 1:1000; E |
MAP 2 | Mouse (mAB) | Microtubule-associated protein 2 | M4403; Sigma | 1:100; E |
Nrf2 | Rabbit (mAB) | Nuclear factor (erythroid-derived 2)-like 2 | ab62352; Abcam | 1:100; E |
Nrf2 | Rabbit (pAB) | Nuclear factor (erythroid-derived 2)-like 2 | C-20; Santa Cruz | 1:500; C |
p22phox | Rabbit (pAB) | NADPH oxidase | Sc-20781; Santa Cruz | 1:100; C |
PLP | Mouse (mAB) | Proteolipid protein | MCA8394; AbD Serotec | 1:1000; E |
Tppp/p25 | Rat (pAB) | Tubulin polymerization-promoting protein | | 1:100; E |
For fluorescent double and triple labeling of Nrf2 with cell type-specific markers and heme oxygenase 1, tissue sections were routinely dewaxed and rehydrated. After the respective antigen retrieval, primary antibodies for Nrf2 and cell type-specific markers were applied simultaneously at 4 °C overnight. After washing with TBS, secondary antibody mixes consisting of biotinylated-anti-rabbit (Jackson ImmunoResearch, 1:2000) and anti-mouse, -rat, -sheep or -goat antibodies conjugated to Cy3 or Cy5 (Jackson ImmunoResearch; 1:100) were applied simultaneously for 1 h at room temperature. Subsequently, the staining procedure was finished by incubation with streptavidin-Cy2 (Jackson ImmunoResearch; 1:100) for 1 h at room temperature. Fluorescent preparations were examined using a Leica SP2 confocal laser scan microscope (Mannheim, Germany).
Quantitative analysis of Nrf2 expression and statistical evaluation
For quantitative analysis, all respective tissue slides were stained for Nrf2 at the same time. This allowed not only to separate cells with intense versus weak expression within different areas of a single section, but also between sections of different MS or control cases. The following regions of interest were analyzed in the white matter: In and around active white matter plaques defined by the presence of macrophages with early myelin degradation products [
5], we differentiated between the normal-appearing white matter (NAWM; at least 1 cm distant from the plaque edge), the periplaque white matter (PPWM; 0.2–0.5 mm distant from the plaque edge), areas of initial demyelination (prephagocytic area characterized by profound microglia activation, scattered macrophages with early myelin degradation products, myelin sheaths with initial stages of dissolution and oligodendrocyte apoptosis) [
1,
26] and the demyelinated plaque center densely packed with macrophages containing early and late myelin degradation products [
5]. In slowly expanding (smoldering) plaques, NAWM and PPWM were defined similarly as in active plaques. The zone of active demyelination was defined by the presence of numerous activated microglia and scattered macrophages with early myelin degradation products. Additionally, the inactive plaque center was analyzed as well. Inactive lesions were defined by their sharp plaque edge, the absence of a rim containing activated microglia and the lack of macrophages with myelin degradation products. In addition, we quantified neurons with nuclear Nrf2 expression in the cerebral cortex of MS patients with progressive disease and presence of cortical lesions, control cases and in biopsies from patients with progressive multifocal leukoencephalopathy with or without treatment with fumarates. Nrf2-positive nuclei were counted at the microscope using the 40× objective with a counting grid in one of the ocular lenses. In the white matter, 10 visual fields of 0.25 mm
2 were counted, thus the values shown in the figures represent cells per 2.5 mm
2. Separate counts are provided for different regions of interest in white matter lesions. Analysis of Nrf2-positive neurons was performed in the same way as described before for white matter lesions. However, since we did not see differences in the presence of Nrf2-positive neurons within and outside cortical lesions in MS, the values were pooled per case.
Statistical analysis was performed with nonparametric tests. Statistical difference between two groups was assessed by Mann–Whitney U test. For the comparison of multiple groups with controls, the Kruskal–Wallis test was used, followed by pairwise Mann–Whitney U tests and Bonferroni–Holm correction. A P value smaller or equal to 0.05 was considered as statistically significant. Data were presented as scatter plots showing all actual data points as well as the median of each group.
Whole-genome microarrays
For gene expression analysis of Nrf2-responsive genes in active white matter and cortical lesions, we used whole-genome microarray datasets already created in previous studies [
14,
15,
22]. For microarray analysis of active white matter lesions, four cases with acute MS (MS2, MS3, MS4, MS6) and four normal controls were chosen [
14,
22]. From the MS material, three different regions of interest were microdissected: (a) NAWM (at least 0.5 mm apart from the lesion edge); (b) zone of initial demyelination showing partial demyelination and tissue infiltration with macrophages containing early myelin degradation products; and (c) demyelinated plaque centers, which were still densely packed with macrophages containing late myelin degradation products (late active lesions; [
5]). For microarray analysis of cortical lesions, three patients with secondary progressive MS (MS8, MS10, MS13), who showed active demyelination in the cerebral cortex, and three controls without neurological disease and brain lesions were chosen [
15]. At least the outer four cortical layers of lesions with subpial demyelination or all cortical layers were microdissected from MS or control tissue, respectively. The cortical lesion sample contained material from one MS patient (MS10, previously described as MS1 in [
15]), who presented with epileptic seizures prior to death and showed highly inflammatory active subpial lesions in many cortical regions. Data from this patient were additionally analyzed separately since they derived from lesions with an extreme degree of active inflammation, microglia activation and oxidative injury. RNA isolation, amplification and microarray technology have been described in detail in previous studies [
14,
15]. All microarray datasets have been deposited at NCBI’s Gene Expression Omnibus (GEO accession numbers GSE32915 and GSE32645).
Discussion
The concept of neuroprotective treatment of multiple sclerosis patients with fumarates is based on their anti-inflammatory actions [
38], mediated for instance by inhibition of microglia activation [
35] and on their induction of the transcription factor Nrf2 and of endogenous downstream anti-oxidant defense mechanisms [
28,
37]. This treatment paradigm is supported by the observation that oxidative injury is a major factor involved in demyelination and neurodegeneration in the CNS of MS patients.
The potential anti-oxidative protective effect of fumarates may depend on the pre-existent expression levels of Nrf2 in cells or tissues. Fumarates induce Nrf2 and are neuroprotective in conditions of no or low basic expression of Nrf2, such as during EAE [
28]. However, in conditions of pre-existing profound oxidative stress, high basic expression of Nrf2 can be expected, which may not be further increased by fumarate treatment. Additional fumarate-induced cells stress in oligodendrocytes or neurons already exposed to severe oxidative injury might even increase demyelination or neurodegeneration. Further induction of Nrf2 in cells, which already express this protein at high levels, may propagate cell death by interacting with another promotor element through the induction of KLF-9 [
46]. Thus, anti-oxidant effects of fumarates may differ profoundly depending on the degree of oxidative injury and pre-existing expression levels of Nrf2 in the tissue.
For these reasons, it is important to know in detail the extent and patterns of Nrf2 expression in MS patients at different disease stages and in different lesion types. So far, only two studies have addressed this question and concluded that its low expression may be insufficient to protect the tissue against oxidative injury. In our present study, we found very prominent Nrf2 expression predominantly in oligodendrocytes at sites of initial demyelination in active MS lesions. Within these cells, Nrf2 appeared to be functionally active, as indicated by the expression of HO-1 in the same cells and the selective up-regulation of other Nrf2-responsive genes in microdissected lesion areas with initial stages of demyelination and tissue injury. As in our study, presence of HO-1 in oligodendrocytes has been previously described in active MS lesions [
40]. Excessive Nrf2 expression was seen in cells with signs of cell degeneration, such as the loss of cytoplasmic marker proteins and nuclear DNA fragmentation. Furthermore, the liberation of AIF from mitochondrial stores into the cytoplasm and nucleus, as seen in cells with excessive Nrf2 expression, has been already described as a major pathway for oxidative cell death in MS lesions [
15,
44]. High Nrf2 expression was present in active lesions of acute and relapsing MS as well as in slowly expanding lesions in progressive MS. Nrf2-positive astrocytes and macrophages were seen mainly in later stages of active lesions characterized by complete demyelination and profound oligodendrocyte loss. Such a pattern of Nrf2 immunoreactivity has been reported previously by van Horssen et al. [
41]. In contrast to this report, we also included in our study patients with acute MS and highly active lesions and we also used very stringent criteria for identification of lesional activity in patients with progressive MS based on the presence of early myelin degradation products in macrophages. Nrf2 expression in MS is fundamentally different from that seen in EAE, where it is sparse or absent even in highly active lesions, but is induced in oligodendrocytes, astrocytes, macrophages and neurons after fumarate treatment [
28].
So far, we convincingly found Nrf2 in mature oligodendrocytes in active MS lesions, but we cannot draw conclusions on its expression in oligodendrocyte progenitor cells (OPCs). Although a variety of markers for OPCs are currently available, none of them work with convincing reliability in archival autopsy tissue with variable post mortem autolysis time and formaldehyde fixation time.
Expression of Nrf2 in neurons has been described in control patients and in various neurodegenerative diseases, but it was concluded that its expression may be insufficient to prevent neuronal degeneration [
36]. In our study, the nuclear Nrf2 reactivity in neurons in MS lesions was low or absent, even in gray matter lesions with extensive acute oxidative injury. Only exceptionally, neurons with morphological features of acute (terminal) hypoxic injury revealed prominent Nrf2 up-regulation and nuclear translocation. Interestingly, within the same sections and lesion areas, oligodendrocytes with abundant Nrf2 expression were located in close vicinity to Nrf2-negative neurons. Thus, the lack of Nrf2 expression in neurons is not due to different lesion environments between gray and white matter, but seems to reflect an intrinsic difference in the reaction to inflammation, demyelinating and oxidative stress between neurons and oligodendrocytes. The low induction of Nrf2 in cortical neurons was also reflected by the lack of induction of most Nrf2-responsive anti-oxidant molecules and may also be related to the down-regulated expression of another transcription factor (peroxisome proliferator-activated receptor gamma, coactivator 1 alpha, PGC-1α, encoded by the
PPARGC1A gene), which regulates mitochondrial anti-oxidant mechanisms [
45]. In addition, it has been recently shown that Nrf2 expression is repressed in neurons in the adult nervous system by epigenetic inactivation of its promotor [
3]. Profound induction of Nrf2 has been seen in neurons in vitro [
37] and in vivo in EAE animals treated with fumarates [
28] and protected neurons against oxidative injury [
37]. However, as in MS patients, in PML patients treated with fumarate we did not observe an increase of nuclear Nrf2 reactivity in cortical neurons, despite increased nuclear immunoreactivity within the white matter lesions of the same patients [
31]. This may be of major importance for the interpretation of treatment effects in patients with progressive MS, since gray matter damage appears to play a key role in disability progression [
7,
13].
A limitation of our study is that pathology does not provide direct information on the temporal pattern of Nrf2 expression. Thus, induction of Nrf2 expression by fumarates in oligodendrocytes before new lesions are formed may be protective, while its further stimulation in active lesions might have no or even opposite effects. In addition, up-regulation of anti-oxidant defense molecules in astrocyte may indirectly exert neuroprotective effects in adjacent cells such as neurons [
33]. This view is further supported by previous studies, which showed that anti-oxidant molecules such as NAD(P)H:quinone oxidoreductase 1, thioredoxin-2, or thioredoxin reductase are mainly expressed in astrocytes, while these anti-oxidant defense mechanisms are less effective in neurons [
6,
12,
33,
42]. Overall, our study provides further evidence for an important role of oxidative stress as a mechanism of demyelination and neurodegeneration in MS and for extensive up-regulation of anti-oxidant defense molecules at least in active white matter lesions.