Efficacy of Fumaric Acid Esters in the R6/2 and YAC128 Models of Huntington's Disease

Huntington's disease (HD) is an autosomal dominantly inherited progressive neurodegenerative disease. The exact sequel of events finally resulting in neurodegeneration is only partially understood and there is no established protective treatment so far. Some lines of evidence speak for the contribution of oxidative stress to neuronal tissue damage. The fumaric acid ester dimethylfumarate (DMF) is a new disease modifying therapy currently in phase III studies for relapsing-remitting multiple sclerosis. DMF potentially exerts neuroprotective effects via induction of the transcription factor “nuclear factor E2-related factor 2” (Nrf2) and detoxification pathways. Thus, we investigated here the therapeutic efficacy of DMF in R6/2 and YAC128 HD transgenic mice which mimic many aspects of HD and are characterized by an enhanced generation of free radicals in neurons. Treatment with DMF significantly prevented weight loss in R6/2 mice between postnatal days 80–90. At the same time, DMF treatment led to an attenuated motor impairment as measured by the clasping score. Average survival in the DMF group was 100.5 days vs. 94.0 days in the placebo group. In the histological analysis on day 80, DMF treatment resulted in a significant preservation of morphologically intact neurons in the striatum as well as in the motor cortex. DMF treatment resulted in an increased Nrf2 immunoreactivity in neuronal subpopulations, but not in astrocytes. These beneficial effects were corroborated in YAC128 mice which, after one year of DMF treatment, also displayed reduced dyskinesia as well as a preservation of neurons. In conclusion, DMF may exert beneficial effects in mouse models of HD. Given its excellent side effect profile, further studies with DMF as new therapeutic approach in HD and other neurodegenerative diseases are warranted.


Introduction
Huntington's Disease (HD) is an autosomal dominantly inherited neurodegenerative disorder caused by a trinucleotide CAG repeat expansion $36 in the HD gene located on chromosome 4. The course of HD is characterized by progressive motor dysfunction, cognitive impairment, affective disorders and personality changes. The most striking neuropathological feature of HD is the progressive atrophy of the striatum and cortex accompanied by neuronal cell loss in these regions [1]. The major histopathological hallmark is the accumulation of intracellular huntingtin (Htt) aggregates, which continuously increase as the disease progresses. The underlying mechanisms of neurodegeneration and Htt aggregation are far less well understood. Here, lack of trophic factors, mitochondrial dysfunction and oxidative stress may play a role [2]. So far, therapeutic options for HD are limited to symptomatic treatment approaches and there is no cure for this devastating disease.
Dimethylfumarate (DMF) is an orally bioavailable fumaric acid ester (FAE) which is metabolized to methyl hydrogen fumarate [3]. Since 50 years, fumaric acid esters are now in clinical use for the treatment of psoriasis [4]. In the recent years, DMF was successfully introduced into the therapy of relapsing-remitting multiple sclerosis (MS). In a first phase IIb study the modified FAE BG00012 was successfully tested and phase III studies are currently ongoing to further explore its therapeutic potential in this indication. While several dermatologic in vitro studies point at immunomodulatory effects [5], another primary mode of action of DMF may be the induction of the nuclear factor E2-related factor 2 (Nrf2) pathway which activates detoxifying phase II enzymes [6]. This pathway may thus play a critical role for cellular protection in an environment of oxidative stress and may present an interesting therapeutic target in HD.
Here we explore the potential of DMF in the R6/2 and YAC128 transgenic murine models of HD. The R6/2 and YAC128 mouse strains are well characterized animal models mimicking many histopathological aspects of HD. R6/2 mice trangenetically express the exon 1 of the human HD gene with 141-157 CAG repeats [7]. In R6/2 mice, motor symptoms like dyskinesia, ataxia, clasping behaviour, epileptic seizures and spontaneous shivering movements start at the age of about 6 weeks. Continous weight loss leads to death between 11-14 weeks of age. From the age of 9-10 weeks, there is a significant neuronal dysfunction and mice display neuronal atrophy in the striatum [8].
In YAC128 mice, a yeast artificial chromosome (YAC) transgene is employed to express a full-length mutant Htt with about 120 CAG repeats [9,10]. In YAC128 mice, hypoactivity is first seen at the age of 8 months. Additionally, progressive gait abnormalities, ataxia, hindlimb clasping and a progressive decline in the forced motor function as measured by the rotarod test occur over time [10,11].
In the models of R6/2 and YAC128 mice, we show that DMF exerts beneficial effects and preserves motor functions as well as intact neurons which may involve activation of the Nrf2 signalling pathway.

Materials and Methods
Animal models and treatment procedure Male R6/2 and YAC128 mice were obtained from the Jackson Laboratories (Bar Harbor, Maine, USA) and bred at the local animal care facilities under standardized conditions. Mice were given food and water at libidum and were weighed daily to obtain weight curves. In the R6/2 strain, stability of the transgene over at least 5 generations of backcrossing was proven by PCR (data not shown). All transgenic mice were heterozygous. All animal experiments were approved by the local authorities for animal experimentation (approval ID: 8.87-50.10.32.08.032, 1 8 Protection of Animals Act). For treatment, mice received either DMF 30 mg/kg body weight in 0.08% methocel as carrier solution or the same amount of methocel alone as placebo treatment twice daily via oral gavage. For survival analysis, cohorts of mice were followed over the course of disease with moribund mice sacrificed according to animal protection laws.

Behavioral analyses
Over the course of the disease, mice were subjected to behavioral analyses every second week starting from week 3. An accelerating rotarod was used to analyse motor coordination and balance (Ugo Basile, Biological Research Apparatus, Varese, Italy). Mice were trained twice on the rotarod with 10 rpm for a maximum of 240 s. During test conditions, the rotarod was accelerated from 4 to 40 rpm with two trials on the same day.
To analyse limb dyskinesia, the clasping behaviour was assessed in a blinded manner as described previously [12,13]. Briefly, mice were vertically suspended for 45 s approximately 40 cm above ground. Scoring was based on the duration of clasping as follows: clasping .10 s = score 3, 5-10 s = score 2, 1-5 s = score 1, no clasping = score 0.
Gait analysis was performed by footprint analysis as described previously [14]. Analysis included stride-length, hind-base width, front-base width and overlap indicating the distance from left or right front footprint to hind footprint.
Cohorts with a maximal number of 47 R6/2 mice per group (only male mice) and a maximal number of 23 YAC128 mice per group (only male mice) were tested.
For electron microscopical analysis, removed brains were adjusted in a plexiglass frame following the orientation of Paxinos and Franklin [18], embedded in 2% agarose in PBS and cut into 1.5 mm frontal brain slices with a vibratome cutter. All slices were photodocumented for later orientation, postfixed with 2% osmium tetroxide in PBS for 3 hours and embedded in araldite (Serva, Heidelberg, Germany) as recently described [19]. Blocks with the striatum and motor cortex were prepared for semithin section series, cut with a Leica Ultracut UCT microtome and stained with 1% toluidine (pH 9.3). Ultrathin sections (80 mm) were contrasted with uranyl acetate and lead citrate. Light microscopical pictures were taken by a CP71 camera (Olympus, Hamburg, Germany). Ultrathin sections were viewed in a Philips EM 420 electron microscope documented by the digital system DITABIS (Digital Biomedical Imaging System, Pforzheim, Germany).

Detection of oxidative chemistry markers in tissue samples
The presence of reactive oxygen species and other oxidants such as ONOO 2 was visualized on frozen mouse brain sections from R6/2 mice and wild-type controls using dihydroethidium as described previously [20]. After transcardial perfusion with phosphate buffered saline (PBS, pH 7.4), we applied 5 mM hydroethidine (Polysciences, CAS Number 104821-25-2) for 30 minutes. Sections were analyzed by confocal laser scanning microscopy.

Quantification and statistical analyses
All neuropathological and behavioral analyses were performed completely blinded with respect to treatment allocation. In the striatum or motor cortex of both hemispheres, quantification of neuronal densities was performed on standardized sections comprising an area of 1 mm 2 . On 2 sections per animal, 4 standardized areas were analyzed under 40 fold magnification. Intact neurons were defined by intact shape, regular appearance and intact processes without formation of vacuoles. Data are presented as mean 6 SEM. For histological evaluations, statistical analysis was performed by Mann-Whitney U-test. To assess the survival of mice, a Kaplan-Meier analysis with Log rank test was used (all analyses done by Graph Pad Prism 5, San Diego, CA, USA). A probability level of *p,0.05, **p,0.01, ***p,0.001 was considered to be statistically significant for all tests.

Results
The symptomatic disease stage in R6/2 mice is associated with an increased formation of oxygen radicals In a first step, we examined the formation of free radicals in R6/ 2 mice at the age of 3-4 months, i.e. at a time point where mice are also clinically symptomatic and exhibit severe neurodegeneration [21]. Therefore, we stained brain sections with dihydroethidine as a broad-spectrum indicator of oxidative stress. As compared to wild-type mice, R6/2 mice showed a significantly larger amount of reactive oxygen species in neurons, as detected by dihydroethidium staining (Fig. 1).
In summary, neurodegeneration in R6/2 mice is associated with higher levels of oxidative stress in neurons which is well in line with previous studies [22,23].
The antioxidant DMF treatment leads to a prolonged survival and preserved body weight in R6/2 mice Since we found a considerable amount of oxidative stress in the R6/2 mouse model of HD, we were interested in the clinical effects of DMF, a compound which has been recently shown to exert neuroprotective effects via the induction of antioxidant pathways [6]. After initiation of treatment with DMF at 30 mg/kg body weight four weeks after birth, R6/2 mice displayed a maximum weight between days 50 to 70 of age and then lost body weight concurrent with the progression of motor deficits. Upon comparison of DMF treated R6/2 mice with controls treated with carrier solution alone, there was a trend to preserved body weight after DMF treatment with a significant preservation of body weight on days 85 and 90 of life ( Fig. 2A). Moreover, DMF treated R6/2 mice displayed a 7% increase in life span. In a Kaplan-Meier analysis, the median survival after DMF treatment was 100.5 days in contrast to 94.0 days after oral gavage with vehicle only (Fig. 2B).
In summary, DMF treatment leads to a preserved body weight and prolonged survival in R6/2 mice.

DMF treatment preserves motor functions in R6/2 mice
In a next step, we correlated the beneficial clinical effects of DMF in the R6/2 mouse model with behavioral tests of motor functions. While motor functions generally decreased over time, analysis of forced motor behaviour in the rotarod test from weeks 10-12 revealed that DMF treatment led to a trend towards increased time on the rotarod (Fig 3A). Furthermore, R6/2 mice were tested weekly for clasping behaviour as a marker of dyskinesia. Up to the age of 9 weeks, there was no difference in clasping scores between DMF treated mice or methocel treated controls (Fig. 3C). At 10-12 weeks of age, matched with the beginning of motor impairment, DMF treatment resulted in a   trend towards reduced clasping scores. In mice older than 12 weeks, DMF treatment led to significantly reduced clasping scores as compared to methocel treated controls (Fig. 3B,C). Additionally, gait abnormalities were assessed by analyzing footprint patterns of DMF or sham treated R6/2 mice walking along a narrow tunnel. Analysis of gait width, stride length or overlap did not reveal any significant differences between both groups, but also failed to convincingly show differences over the course of disease between sham-treated R6/2 mice and wild-type controls (data not shown). In summary, DMF treatment resulted in a preservation of motor functions in R6/2 mice.

DMF treatment preserves the morphology of neurons in the striatum and the motor cortex
Since DMF treatment exerts beneficial effects on survival and motor behaviour in R6/2 mice, we were interested in the histopathological changes under DMF treatment. Numbers of morphologically intact neurons were analyzed after cresyl violet staining or immunohistochemistry for the NeuN antigen on day 80. Morphological criteria were applied to a blinded stereological analysis of neurons in the striatum and motor cortex. In the analysis of both NeuN positive or cresyl violet stained cells, DMF treatment resulted in 1.5-2 fold higher numbers of intact neurons both in the striatum as well as in the motor cortex (Tab. 1 and Fig. 4A-H).
Since it is well known that there is no complete neuronal loss, but still severe neurodegeneration in the striatum and the motor cortex of R6/2 mice, the light microscopical data were corroborated by semithin-and ultrasection analysis (Fig. 5). The integrity of striatal and cortical neurons was relatively preserved in DMF treated R6/2 mice (Fig. 5A,C,E,G,I,L) when compared with the corresponding areas in the methocel treated controls (Fig. 5B,D,F,H,K,M). In semithin sections of the striatum (Fig. 5A,C) and of the motor cortex (Fig. 5G,I) of DMF treated mice, most neurons appeared intact in the striatum and motor cortex (Fig. 5A,C,G,I). In methocel treated mice, dark cells were dispersed among intact neurons in the striatum and the motor cortex (Fig. 5B,D,H,K). Electron microscopical analysis confirmed the presence of degenerating neurons with darkened and shrunken appearance in the striatum (Fig. 5F) and in the motor cortex (Fig. 5M) when compared with intact light appearing neurons in the respective areas of DMF treated mice (Fig. 5E,L). Nuclear  aggregates were present within intact neurons (Fig. 5L) and also in neurons undergoing dark cell degeneration (Fig. 5M). In summary, DMF treatment led to preservation of intact neurons in the striatum and the motor cortex.

DMF treatment in R6/2 mice targets the Nrf2 pathway
After having established that DMF exerts beneficial effects on neuronal integrity in R6/2 mice, we next analyzed the mechanisms of DMF action. First, we investigated the effects of DMF on the formation of Htt aggregates. To this end, we performed immunohistochemistry for ubiquitin which reliably detects Htt inclusions in R6/2 mice [15]. DMF treatment did not affect the formation of Htt aggregates as measured by Htt or ubiquitin immunoreactivity (Tab. 2). Similarly, staining for Mac-3 as a marker of activated phagocytes did not disclose a major bystander immune reaction which could be modulated by DMF treatment (data not shown).
DMF is a known inducer of phase 2 detoxifying enzymes and both DMF and its primary metabolite monomethlyfumarate (MMF) are thiol-reactive electrophiles [24,25,26]. This combination of properties suggests that the DMF may activate the Nrf2 transcriptional pathway known to mediate induction of phase 2 genes by SH-reactive electrophiles and play a major role in cell and tissue defense against oxidative stress. Immunohistochemistry for Nrf2 revealed a higher number of Nrf2 positive cells in the striatum of DMF treated mice (Fig. 6A,B and Tab. 3). We next  analyzed the effects of DMF on Nrf2 expression in different cell types of the striatum and motor cortex by double label immunfluorescence and laser scanning confocal microscopy. Double staining of Nrf2 with the neuronal marker NeuN revealed an increased Nrf2 immunoreactivity after DMF treatment in different neuronal subpopulations of the brain, especially in the striatum and in the motor cortex (Fig. 6 C,D). After double staining of Nrf2 with the astroglial marker GFAP, double labelled cell were neither identified in DMF nor in methocel treated mice (Fig. 6 E, F). In summary, DMF treatment in R6/2 mice led to an increase in Nrf2 staining in neuronal subpopulations which are relevant for motor functions.

DMF exerts positive effects in the YAC128 model of HD
Finally, the beneficial effects of DMF in the R6/2 mice were corroborated in the YAC128 mouse as another well established mouse model of HD. Analysing the forced motor behaviour on a rotarod confirmed a trend towards preserved motor functions after DMF treatment of YAC128 mice older than 22 weeks (Fig. 7A). While there were no differences in body weight when analysing mice up to a life span of one year (data not shown), DMF treatment in YAC128 mice led to a reduction of limb dyskinesia as measured by the clasping score (Fig. 7B). These beneficial effects on the motor behavior were also seen in a histopathological analysis on day 365. Blinded quantification of intact neurons in the  striatum and motor cortex after cresyl violet staining revealed a preservation of neuronal number after DMF treatment (Fig. 7C,D). In summary, the beneficial effects of DMF in R6/2 mice were also observed in YAC128 mice as another transgenic mouse model of HD.

Discussion
We show that DMF treatment exerts beneficial effects on survival time and motor functions in the R6/2 and YAC128 models of HD. The clinical efficacy of DMF in the R6/2 model was corroborated by preservation of intact neurons and less pronounced dark cell degeneration in the striatum and motor cortex as well as an up-regulation of the transcription factor Nrf2 in striatal neurons.
As expected, R6/2 mice display a higher amount of free radicals in the CNS indicating an increase in oxidative stress which may be linked to neurodegeneration. Indeed, oxidative stress is known to be a common underlying mechanism in the pathogenesis of many neurodegenerative diseases causing damage to multiple cellular components such as DNA, proteins and lipids [27,28]. Here, fumaric acid esters which activate antioxidant response pathways may provide neuroprotective effects [6]. Several studies revealed that DMF is a potent inducer of phase 2 detoxifying enzymes [26] and both DMF and its primary metabolite MMF are thiol-reactive electrophiles [25]. These characteristics imply that fumarates may activate the Nrf2 transcriptional pathway which is known to mediate induction of phase 2 genes by SH-reactive electrophiles and to play a major role in cell and tissue defense against oxidative stress [29]. Indeed, several in vitro studies described protective effects of DMF on neurons [30,31]. In cell culture, MMF protected cultured neurons and astrocytes from oxidative stress induced by H 2 O 2 [6]. Other studies describing deprivation of glutathione or toxicity of DMF and MMF in glial cultures used much higher, non-physiological concentrations of both compounds [32,33]. Very recent studies using Nrf2 deficient mice in a chronic model of multiple sclerosis provide direct evidence for DMF mediated neuroprotective effect via Nrf2 mediated regula- Although DMF treatment leads to a trend towards longer times on the accelerating rod, there is no significant difference between both groups. (B) Clasping Score. A cohort of YAC128 mice treated with DMF (n = 23) or methocel (n = 22) is shown. In comparison to control mice (white bars), there is a trend towards reduced clasping behaviour in the DMF treated group (black bars) from the age of 47 weeks. (C) Blinded quantification of neuronal densities in the striatum after creysl violet staining on day 365 (n = 3 per group). There is a significant preservation of neuronal numbers after DMF treatment (black bar) as compared to methocel treatment (white bar, p = 0.03). (D) Blinded quantification of neuronal densities in the motor cortex after creysl violet staining on day 365 (n = 3/5 per group). There is a significant preservation of neuronal numbers after DMF treatment (black bar) as compared to methocel treatment (white bar, p = 0.004). doi:10.1371/journal.pone.0016172.g007 tion of the oxidative stress response [6]. Fumarates directly modify the inhibitor of Nrf2, Keap-1, which leads to transfer of free Nrf2 into the nucleus and activation of anti-oxidative response elements. A result of the activated Nrf2 pathway is cellular detoxification, normalization of the energy metabolism and repair and/or degradation of damaged proteins.
The Nrf2-mediated oxidative stress response was already previously linked to protection of the CNS in a variety of pathological conditions [34,35,36,37]. In particular, the importance of this pathway was recently implicated in the pathogenesis of HD [38]. Furthermore, ex vivo studies and analyses of neurodegenerative models for motor neuron disorders and Parkinson's disease disclosed that Nrf2 mediated neuroprotection may not only be directly mediated, but in some models critically involves effects via astrocytes [39,40,41,42].
In addition to its neuroprotective capacity, DMF may also possess immunomodulatory effects which were at least proposed in several dermatologic analyses in vitro [43,44,45,46,47]. In the R6/2 model, we did not observe a significant bystander immune reactivity in the CNS which could be modulated by DMF. We thus propose that DMF treatment in HD models primarily targets neurons. This concept is in line with the observation that DMF metabolites reach significant concentrations in the CNS [6]. To further explore the neuroprotective potential of DMF, additional studies in other models of HD and neurodegenerative diseases are warranted.
The efficacy of DMF as an inducer of the oxidative stress response in models of HD underlines the importance of free radicals in the pathophysiology of HD. Indeed, mitochondrial dysfunction, oxidative stress and energy depletion have long been implicated in the pathogenesis of HD and stimulated several therapeutic approaches. In particular, the treatment of R6/2 transgenic mice with dichloracetate which increases the activity of the pyruvate dehydrogenase complex revealed positive effects. Further studies employing antioxidants like a-lipoic acid [48] observed beneficial effects and a significantly enhanced survival of R6/2 mice. Similar results were achieved by administration of coenzyme Q 10 [49]. While these effects may be based on the improvement of mitochondrial function or the decrease of oxidative damage, it seems conceivable that different disease stages are governed by different prevailing pathomechanisms. This concept could explain the decreasing efficacy of DMF at the later disease stages in R6/2 mice.
In R6/2 mice, DMF leads to a preservation of intact neurons as revealed by cresyl violet staining and immunohistochemistry for the NeuN antigen. These light microscopy results were corrobo-rated by semi-and ultrathin section analyses revealing a less pronounced dark cell degeneration after DMF treatment. While several reports point at a significant neuronal loss and striatal atrophy in R6/2 mice [7,8,50] or YAC128 HD transgenic mice [9,10], other studies only describe functional changes of neuronal subpopulations in these models [51,52]. This concept was further refined by the characterization of dark cell degeneration in striatal and cortical neurons [8,53,54,55]. In our study, we therefore used several histological approaches to describe a relative preservation of morphologically intact neurons under DMF treatment as compared to sham treated controls. Yet, the study design does not allow conclusions on a putative overall loss of striatal or motor cortex neuron numbers in R6/2 or YAC128 mice.
Many therapeutic approaches were tested in rodent models of HD, including inhibition of histone deacetylation and methylation [56,57], congo red [58,59], disaccharides [60], tiagabine [61], transglutaminase inhibitors [62,63], caspase inhibitors [64] or also tetracyclins [65,66,67]. Yet, many of these approaches were later refuted in clinical trials or could not be reproduced in follow-up studies as recently shown e.g. for minocyline or co-enzyme Q [66,68]. This may in part be explained by the low number of animals used, environmental factors or husbandry [69]. These ambiguous results highlight the importance of a standardized testing which was also implemented in the present study design [69]. Using large cohorts as well as standardized therapy, husbandry and testing batteries, we show that DMF exerts beneficial effects in the two best characterized mouse models of HD. While DMF does not influence the pathogensis of HD, it exerts significant protective effects on the affected tissue. Given its excellent side effect profile, further studies with DMF as a potential neuroprotective approach in HD are warranted. Such trials may also include combination therapy approaches to target different pathomechanisms of HD which are possibly stage-specific.