All animal experiments were carried out according to the regulations of the local animal research council (Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit, Oldenburg, Germany) and legislation of the State of Lower Saxony, Germany. High-copy number B6/SJL-Tg(SOD1 G93A)1Gur/J (Gurney et al., 1994) were obtained from Jackson Labs (Stock Number 002726; Bar Harbor, USA). The colony was maintained by crossing B6/SJL males harboring the transgene with wild-type B6/SJL females. Housing of animals was performed under a 12 h light/12 h dark cycle with free access to food and water. For genotyping, tail biopsies of 21-day-old mice were subjected to a standardized PCR protocol (Jackson Labs). Probe sequences were: hSOD1-forward, CATCAGCCCTAATCCATCTGA; hSOD1-reverse, CGCGACTAACAATCAAAGTGA; both SOD1^G93A female and male mice were used in the experiments. Fasudil (“F-4660,” LC Labs, Woburn, Massachusetts, USA) was orally administered at a concentration of 30 mg/kg body weight per day (subsequently termed as Fas30) or at a concentration of 100 mg/kg body weight per day (Fas100) in drinking water. Control groups received drinking water without supplementation (Veh).

Nerve conduction velocities (NCV) and compound muscle action potentials (CMAP) were determined as previously described (Zielasek et al., 1996). Time points of analysis were 70 and 100 days of age. Briefly, mice were anesthetized using an intraperitoneal injection of ketamine (100 mg/kg body weight) and xylazine (8 mg/kg body weight). CMAP and distal motor latencies (DML) were recorded from the intrinsic foot muscles using steel electrodes. NCV were calculated automatically based on the two stimulation sites.

Animals were killed by CO₂ insufflation at disease end-stage in accordance with the local guidelines. Thereafter, they were transcardially perfused with PBS solution followed by 4% paraformaldehyde. Spinal cords were removed as described before (Günther et al., 2012) and were post-fixed for 2 h in 4% paraformaldehyde. Thereafter, the tissue was dehydrated in 30% sucrose overnight and kept at −20°C until further processing. Coronal sections of the lumbar spinal cord (L3–L6) with a thickness of 20 μm were prepared using a cryostat (Leica, Wetzlar, Germany) and collected on gelatine-coated glass slides.

Spinal cord sections were immunolabeled with the primary antibodies (anti-ChAT, 1:100, Millipore; anti-Iba1, 1:500, WAKO, Osaka, Japan) and subsequently with the respective secondary antibodies (Cy3 anti-goat or Cy2/Dylight anti-rabbit; both Dianova).

In order to quantify MN numbers in the ventral horn of coronal cryosections from the lumbar spinal cord, cells in the ventral horn were counted as MN if they had a clearly identifiable nucleolus, were at least 200 μm² in size and were ChAT-positive. At least ten sections per mouse spinal cord including both ventral horns being at least 100 μm apart over a length of at least 1000 μm from the lumbar spinal cord were counted. In the treatment groups a total number of nine mice were analyzed from vehicle (n = 3), Fas30 (n = 3) and Fas100 (n = 3). Microgliosis was evaluated by the quantification of Iba1 positive cells in the ventral horn of lumbar spinal cord sections. In contrast to the MN analysis MG were not confined to ventral horn region. Therefore, the ventral horn area was measured and Iba1 positive cells were manually counted and plotted as cell numbers per 0.2 mm². A total number of nine mice each gender was analyzed from vehicle (n = 3), Fas30 (n = 3) and Fas100 (n = 3). Means of both ventral horns were calculated and the mean of all sections of each animal was included in the final statistical analysis for both ChAT and Iba1 cell numbers.

For the analysis of electrophysiological quantifications between two groups the unpaired Student's t-test was applied. Multiple group comparisons of immunohistological cell counts were done by one-way ANOVA followed by Tukey's post-hoc test. Behavioral data were subjected to a multifactorial ANOVA analysis and post-hoc Fisher's test. Survival data were analyzed using Kaplan-Meier survival fit analysis with Log-Rank tests for statistical significance. Statistical analyses were performed using Statistica 10.0 (StatSoft, Hamburg, Germany) and Kyplot 2.0 (KyensLab Inc, Tokyo, Japan). Data are represented as mean ± standard error of the mean. Differences were considered significant with ^*P < 0.05; ^**P < 0.01; ^***P < 0.001.

In order to detect an early impairment of motor nerve function as indicator for disease onset we examined wildtype and SOD1^G93A mice with electrophysiological methods at day of life 70 (d70) and at day of life 100 (d100). At d70, the NCV of the sciatic nerve was already significantly delayed in SOD1^G93A animals if compared to wildtype controls (SOD1^WT: 30.63 ± 0.95 m/s; SOD1^G93A: 24.07 ± 1.10 m/s) (Figure 1A). This NCV difference was also persistent at d100 (SOD1^WT: 26.31 ± 1.33 m/s; SOD1^G93A: 20.72 ± 0.98 m/s) (Figure 1E). The CMAP at d70 was reduced in SOD1^G93A mice in comparison to SOD1^WT [proximal CMAP: SOD1^WT 9.10 ± 1.44 mV; SOD1^G93A 5.17 ± 0.57 mV (Figure 1B); distal CMAP: SOD1^WT 12.10 ± 1.98 mV; SOD1^G93A 8.78 ± 1.23 mV (Figure 1C)]. At d100, this difference increased even more [proximal CMAP: SOD1^WT 9.35 ± 0.93 mV; SOD1^G93A 4.58 ± 1.45 mV (Figure 1F); distal CMAP: SOD1^WT 11.30 ± 1.61 mV; SOD1^G93A 5.91 ± 1.73 mV (Figure 1G)]. The DML was not significantly different at both time points [d70: SOD1^WT 1.28 ± 0.09 ms; SOD1^G93A 1.17 ± 0.07 ms (Figure 1D); d100: SOD1^WT 1.22 ± 0.05 ms; SOD1^G93A 1.33 ± 0.06 ms (Figure 1H)].

Oral Fasudil treatment was initiated at d80 in symptomatic SOD1^G93A mice in two different dosages Fas30 and Fas100, both in female and in male mice. Veh-treated SOD1^G93A female and male cohorts served as control. Among all cohorts differentiated by gender, there were no significant differences in weight dynamics. In females, the weight maximum for the Veh group was reached at day 104 (19.54 ± 0.42 g), for the Fas30 group at day 101 (19.98 ± 0.42 g) and for the Fas100 group at day 92 (19.86 ± 0.46 g). This corresponds to a mean maximum weight gain from day 80 on in the Veh group of 1.94%, in the Fas30 group of 3.24% and in the Fas100 group of 3.90%. In males, the weight maximum for the Veh group was reached at day 92 (25.26 ± 0.48 g), for the Fas30 group at day 101 (24.89 ± 0.67 g) and for the Fas100 group at day 95 (24.94 ± 0.43 g) corresponding to a mean maximum weight gain from day 80 of 0.41% (Veh), 2.01% (Fas30) and 1.60% (Fas100). A progressive weight loss in the female veh group was observed starting on day 104 and in both female treatment groups starting on day 107. In contrast, progressive weight loss in males started on day 92 in veh group, on day 101 in the Fas30 group and on day 95 in the Fas100 group. The overall dynamics of weight loss did not significantly differ in female or male groups (Figure 2).

We repeatedly performed neurological scoring over the entire treatment period. Here, application of Fas30 or Fas100 did not significantly modify the progression of clinical disease in both female and male mice (Figures 3A,C). Similarly, the average duration of disease was not significantly different between female (Veh 45,57 ± 3,12 d; Fas30 41,57 ± 1,89 d; Fas100 41,60 ± 1,48 d) and male (Veh 44,92 ± 3,31 d; Fas30 41,64 ± 3,54 d; Fas100 43,50 ± 2,64 d) treatment groups (Figures 3B,D).

In the analysis of overall survival, the mean times of survival were not significantly different in any treatment groups, neither in female (Veh: 139.1 ± 2.7 days; Fas30 135.9 ± 1.8 days; Fas100 136.7 ± 1.7 days) nor in male mice (Veh: 131.7 ± 2.8 days; Fas30 131.9 ± 1.8 days; Fas100 131.9 ± 2.1 days) (Figure 4).

The evaluation of motor performance on a rotarod device reflects both muscular strength and motor coordination. While female mice showed similar results in all treatment groups, male mice were significantly better in both Fas30 and Fas100 treatment groups. The most significant differences to the control group were observed at intermediate d110 (endurance on rotarod: Veh 81.64 ± 22.42 s; Fas30 133.91 ± 17.28 s; Fas100 153.92 ± 15.14 s) (Figure 5).

In order to evaluate underlying neuropathological alterations, we quantified MN survival and MG cell infiltration in the anterior horn of the lumbar spinal cord. There were no significant differences in the number of ChAT-immunopositive MN in both female (Veh 6.1 ± 0.53 MN/anterior horn; Fas30 5.39 ± 0.36 MN/anterior horn; Fas100 7.35 ± 0.29 MN/anterior horn) and male mice (ChAT numbers Veh 7.15 ± 0.61 MN/anterior horn; Fas30 8.43 ± 0.37 MN/anterior horn; Fas100 6.92 ± 0.29 MN/anterior horn) (Figure 6).

Iba1-immunopositive MG cells were found in the spinal cord anterior horn at similar numbers in all treatment groups in female (Veh 64.64 ± 2.65 MG/section; Fas30 56.52 ± 10.42 MG/section; Fas100 52.61 ± 7.53 MG/section) and male mice (Veh 54.38 ± 4.54 MG/section; Fas30 59.22 ± 4.41 MG/section; Fas100 56.91 ± 5.96 MG/section) (Figure 7).

Several parameters can be used to indicate disease onset in mouse models of ALS. One parameter is the clinical phenotype of the mice. It is well described that around d80 first clinical symptoms start to appear with trembling of hind limbs (Turner and Talbot, 2008). Consequently, detection of trembling of hind limbs is included in a commonly used neurological scoring system for this mouse model (Miana-Mena et al., 2005) representing the first disease stage (Weydt et al., 2003). In previous studies, we had observed a decline from the healthy phenotype (score 4) to the first disease stage (score 3) between d80 and d100 in both male and female animals. At d100 nearly all animals had reached disease onset (score 3) (Günther et al., 2014). Furthermore, motor testing with a rotarod device resulted in a decline of performance between d80 and d100. Applying a hanging wire test we observed a decrease of muscle strength already after d70 (Günther et al., 2014; Tönges et al., 2014). Concerning these functional parameters it is important to note that male mice are impaired earlier than females with regard to clinical scoring and motor behavior. Taking into account the genetic background of the animals, differences between females and males in the SOD1^G93A have been previously observed (Oliván et al., 2015; Pfohl et al., 2015). While both, the C57BL/6 and the B6SJL background, lead to delayed disease onset in females, a prolonged survival was only observed in females with the B6SJL background in comparison to their male counterparts (Pfohl et al., 2015). Similar to previous reports, female animals in our study had a prolonged survival time and a delayed decline in motor performance as compared to their male counterparts (Miana-Mena et al., 2005; Alves et al., 2011; Mancuso et al., 2012; Pfohl et al., 2015). Another quantitative method to assess initiation or progression of disease is the examination based on electrophysiological measurements, which was used in this study and represents also an important diagnostic tool for human ALS (Brooks et al., 2000). Considerable differences between SOD1^WT and SOD1^G93A have recently been reported in a previous study on time point d100 and additional comprehensive evaluations of electrophysiological measurements have been done (Alves et al., 2011; Mancuso et al., 2012; Tönges et al., 2014). Mancuso et al. determined onset of disease between 70 and 90 days of age in the SOD1 mouse model which was associated with CMAP alterations of the tibialis anterior muscle suggesting that electrophysiological measures can be a marker for disease onset (Mancuso et al., 2012). In our own electrophysiological analysis we also used neurographic measurements and could detect deteriorations in CMAPs and a significant deceleration of NCVs compared to SOD1^WT already on d70. In line with progressive disease, CMAPs deteriorated and became significantly reduced at d100. These results confirm a disease onset in electrophysiological terms at around d70 and the presence of progressive disease until d100 in SOD1^G93A animals. In order to define a distinct starting point for treatment in our study we then chose d80 which additionally incorporates interindividual variances in SOD1^G93A mice and thus represent a clear time point of manifest disease.

Fasudil treatment was thus started at d80 in two different concentrations (Fas30 and Fas100) and was clinically well tolerated. The weight course was comparable to the Veh group. In our treatment paradigm, Fasudil treatment did not prolong survival and we could not observe a statistically relevant benefit in the graded clinical examination. However, if we applied more sensitive examination methods for motor strength and coordination with rotarod testing, male animals showed a benefit from treatment with both dosages.

Others and our own group had found a moderate protection of spinal cord MN in an intermediate analysis on day 100 with identical dosages of Fasudil if treatment was started already at a presymptomatic stage at day 50 (Takata et al., 2013; Tönges et al., 2014). However, the histological analysis in endstage animals did not reveal significant differences of MN survival or in MG infiltration between the groups. In the symptomatic treatment study from day 80, we similarly did not detect clear ameliorations such as an increased motoneuron survival on the neuropathological level. This may be due to an only functional improvement of motoneurons, which, however, was too small or unstable to transfer to a persistant cellular change.

Another reason for the rather small treatment effect observed in our study is most probably the advanced pathology that is already present at the time, when animals become clinically symptomatic (Turner and Talbot, 2008). It is also known, that progression of disease in the SOD1^G93A mouse model is rather rapid and one could hypothesize that a more slowly progressing disease model would yield more favorable results. Many groups have performed therapeutic pilot studies in the SOD1^G93A mouse model in a clinically presymptomatic stage between 30 and 60 days of age, long before clinical symptoms become obvious (Turner and Talbot, 2008), but this strategy is insufficient if no verification with a treatment after symptom onset is performed.

Of interest, significant, but overall only moderate therapeutic effects were found exclusively in male mice in this study. Gender specific effects have also been reported in the Y-27632 presymptomatic treatment study (Günther et al., 2014) and differences in the disease course between male and female mice were previously described (Alves et al., 2011; Oliván et al., 2015; Pfohl et al., 2015). Ovariectomy in the SOD1^G93A mouse model leads to a decrease in survival time and makes female mice progress almost as fast as male mice. In line with this finding, estrogen treatment can counteract the effects of ovariectomy (Groeneveld et al., 2004; Choi et al., 2008). Similarly, ovariectomy in the MPTP model of Parkinson's disease leads to an increase of ROCK activity and treatment with either estrogen or a ROCK inhibitor could inhibit MPTP-induced dopaminergic cell death. Furthermore, ROCK activity was less increased in ovariectomized MPTP mice when treated with estrogen (Rodriguez-Perez et al., 2013). In analogy, ROCK activity could be higher in male mice in advanced disease and therefore the application of Fasudil could have been more successful in male mice in our study.

In multiple other neurodegenerative disease models, ROCK inhibitors and particularly Fasudil showed very promising results. In the MPTP model of Parkinson's disease Fasudil significantly protected both dopaminergic neurons and its axonal nigrostriatal terminals, restoring striatal neurotransmitter levels and attenuating behavioral deficits (Tönges et al., 2012). Furthermore, Fasudil attenuated alpha-synuclein aggregation in vitro and in vivo in a genetic model of Parkinson's disease (Tatenhorst et al., 2016). In a Drosophila model of tauopathy, Fasudil suppressed the rough eye phenotype and mitigated pathogenic tau levels by inducing autophagic pathways (Gentry et al., 2016). Furthermore, significant therapeutic effects were also observed in a mouse model of Huntington's disease, both with Y-27632 and Fasudil treatment (Li et al., 2009, 2013), and in a model of spinal muscle atrophy, a condition with high similarity to ALS. Here the clinical progression was attenuated by Y-27632 and Fasudil (Bowerman et al., 2010, 2012).

In a translational perspective, Fasudil combines several highly promising features. It is an already licensed drug, which has a long-standing clinical history with several tens of thousands of patients treated, demonstrating a favorable safety-profile (Suzuki et al., 2007). Its pharmacokinetics is characterized by excellent solubility in water, sufficient uptake into the CNS and no major interactions with Cytochrome P450 enzymes. The fact that Fasudil is effective in several disease models, which all combine similar pathomechanisms of neurodegeneration, reaffirms its potential for a drug-repositioning approach.

LT and PL have filed a priority patent application on the use of Fasudil for the treatment of ALS. The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.