The ethical guidelines established by the municipal government (Regierungspräsidium Karlsruhe, Germany) were strictly adhered to throughout the animal experimentation process. This study was approved by the animal care and use committee, Regierungspräsidium Karlsruhe, Referat 35, Germany, AZ 35.9185.81/G-190/19. Heterozygous SOD1G93A transgenic mice expressing the human SOD1G93A mutant on a C57BL/6 J background (www.jax.org/strain/004435) were utilized. The maintenance of heterozygosity involved breeding heterozygous transgenic males with C57BL/6 wild-type females, a service provided by the Interdisciplinary Neurobehavioral Core (INBC) at Heidelberg University. All experiments employed both transgenic and age-matched (20 weeks) wild-type male mice, which were group-housed (up to three mice per cage) at the INBC, where pattern electroretinogram (pERG) experiments took place. Standard cages (15 cm × 21 cm × 13.5 cm) under a 12:12 h light:dark cycle, with unrestricted access to food, water, and nesting material, were provided for all mice. Treatment assignments were randomized within litter blocks. ALZET pumps (ALZET 2004) were filled with either vehicle (40% propylene glycol, P4347, Sigma-Aldrich) or FP802 (166 mg/ml in 40% propylene glycol) and placed in sterile saline at 37 °C for 48 h to equilibrate. Mice were anesthetized with isoflurane inhalation for the entire surgery, with the depth of anesthesia confirmed by the loss of pedal reflex. The insertion site (back of the mouse) was shaved and cleaned with an antiseptic solution, before making a small incision (about 1 cm) with a disinfected scissor. A subcutaneous pocket was then created using blunt, sterile forceps to a size large enough to accommodate the pump. The osmotic pump was then inserted with the releasing side to the opposite direction of incision before closing the incision. The mouse was then placed in a warm recovery cage to maintain body temperature until it regained consciousness. Mice were subsequently monitored for two days to ensure the incision did not reopen. The experimenter remained blinded to the experimental group assignment from implantation until completion of all analyses. A total of 46 heterozygous SOD1G93A and 16 wild-type littermates were used in this study. For the pERG measurement, 10 wild-type and 26 SOD1G93A mice at 20 weeks were used with one SOD1G93A excluded as it died during the pERG measurement. The left and right eyes were collected on the same day after pERG measurement for immunostaining with Brn3A and qPCR analysis, respectively. Another six wild-type and 11 SOD1G93A mice at 20 weeks were used for the determination of the thickness of the retinal nerve fiber layer (RNFL). In addition, 8 SOD1G93A mice at 15 weeks were sacrificed without any treatment for the quantification of RGCs.

FP802 measurements using mass spectrometry were established by Charles River Laboratories (CRL) (Groningen, Netherland). In mice that were not given FP802, FP802 was neither detected in serum samples nor in brain tissue. The lower limit of quantification (LLOQ) of the methodology used by CRL was 1 nM in serum and 3 pmol/g for brain tissue. To determine FP802 penetrance of the retina, osmotic pump insertion was carried out in wild-type mice according to the same procedure as described for the SOD1G93A mice at the same dose of 40 mg/kg/day. The concentration of FP802 in the retina was measured after 10 days of continuous supply of FP802 using the osmotic pump. Mice were perfused with 20 ml cold PBS to avoid contamination by FP802 present in the blood. The eyes were removed and washed 3 times in cold PBS before the retina was dissected and washed again 3 times in cold PBS. Retinas were frozen on dry ice before being submitted for analysis of FP802 concentration.

Retinas were prepared from animals euthanized by an intraperitoneal dose of 400 mg/kg BW pentobarbital (Narcoren; Merial, Hallbergmoos, Germany). The eyes were removed and fixed in 10% formalin for 15 min. The cornea, iris, lens, and vitreous body were then removed under a stereomicroscope. The retinas were dissected by being separated from the pigment epithelium and sclera. For retinal whole-mount immunolabelling, retinas were post-fixed in 10% formalin for 15 min before being placed in PBS. Retinal whole-mounts were then immunolabelled using the free-floating method. Retinas were blocked at room temperature in 1% Triton X-100, 10% FBS and 0.02% sodium azide in PBS for 6 h, followed by overnight incubation with mouse anti-Brn3a (1:200; sc-8429; Santa Cruz Biotechnology, Dallas, TX, USA) at room temperature. Following washing 3 times in PBS, goat anti-mouse IgG (H + L) Alexa Fluor 594 secondary antibody (1:500; A11005, Life Technologies) was applied and incubated overnight at 4 °C. Retinas were then washed again, cut, and mounted onto slides. For each whole-mount, images were equally obtained from eight fields (375 μm × 375 μm) to minimize the influence of location-associated variability in RGC density on cell counts. Retinal whole-mount images were obtained using Las X software via an HC PL APO 20 × objective on a Leica TCS SP8LIA in a DM6 CFS upright confocal microscope. Brn3a-positive cells were detected and quantified automatically using a self-developed macro in CellProfiler with subsequent manual review to eliminate false cell recognition due to non-specific staining. Data analysis was conducted in a single-blind manner, ensuring unawareness of the group assignments.

To prepare retinas for sectioning, eyes were enucleated from euthanized mice and prepared as previous described [21]. Briefly, eyes were placed in modified Davidson’s fixative (10% glacial acetic acid, 20% neutral buffered formalin, 30% ethanol, 40% dH2O) at room temperature with gentle shaking for 20 h. Following a further 20 h in 4% paraformaldehyde, eyes were washed in PBS and a window created in the posterior wall using a dissecting microscope (Leica, Wetzlar, Germany). After washing in PBS, eyes were placed in 30% ethanol for 1 h and further dehydrated before embedding in paraffin. 0.5 µm paraffin sections were cut using a microtome (Leica SM2000R) and stained with haematoxylin–eosin (H&E) for morphological visualization of retinal layers. Retinal sections traversing the optic nerve head were used for quantification. Images were taken on an Eclipse 80i upright microscope fitted with a DXM1200C camera (Nikon GmbH, Düsseldorf, Germany) using a 10 × objective. RNFL thickness was assessed using ImageJ (NIH) and defined as extending from the inner limiting membrane of the retina to the outer border of the ganglion cell layer. Measurements were taken at 200 µm intervals from the center of the optic disc in both directions and averaged for each eye.

Following excision of retinas as described above, they were immediately shock-frozen in liquid nitrogen. RNA was isolated using the RNeasy Plus Mini Kit (Qiagen; catalog no.: 74106) with additional on-column DNase I digestion according to the manufacturer's instructions. For complementary DNA synthesis, RNA was reverse transcribed with the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems; catalog no.: 4368813). Quantitative real-time reverse-transcription (qRT–PCR) was performed on a StepOnePlus thermal cycler using TaqMan gene expression assays (Applied Biosystems; catalog no.: 4331182) for the following genes: Bdnf (exon IV; Mm00432069_m1), cFos (Mm00487425_m1), Npas4 (Mm00463644_m1), and Inhibin beta A (Mm00434338_m1). Expression levels of target genes evaluated using TaqMan reagents were normalized to the expression of the housekeeping gene, glucuronidase, beta (GusB) (Mm00446953_m1), and to the expression levels in WT.

pERG recordings were performed using a UTAS Visual Diagnostic System (LKC Technologies, Gaithersburg, MD, USA) as previously described [28]. Mice were dark-adapted for 30 min prior to recording and were subsequently anesthetized with an intraperitoneal injection of ketamine (100 mg/kg; Atarost GmbH and Co., Twistringen, Germany) and xylazine (10 mg/kg; Albrecht, Aulendorf, Germany). Anesthetized mice were placed on a heating pad to maintain body temperature at 37 °C and were centered at 15 cm in front of a 17-inch cathode ray tube. In order to achieve maximum focus of the pattern stimulation, pupils were not dilated. A transparent contact lens electrode with gold contact (LKC Technologies) was placed on the corneal surface, and both electrical contact and prevention of eye desiccation was facilitated through application of saline solution. Reference and ground needle electrodes were placed subcutaneously in the neck and tail, respectively. Vertical square wave grating stimulations (90% contrast, temporal frequency of 1 Hz equivalent to two pattern reversals per second) were presented at different spatial frequencies (0.05, 0.1, 0.2 and 0.4 cyc/deg), and noise levels were measured with the monitor turned off. Due to the small pERG amplitudes, 100 sweeps were averaged to increase the signal-to-noise ratio, band-pass filtered between 0.3 and 300 Hz, and amplified (64 × gain). pERG amplitude (μV) was measured as the amplitude of the second harmonic of the fast Fourier transformation plotted against spatial frequency of stimulation.

GraphPad Prism 7 software (GraphPad Software) was used to calculate, plot, and evaluate mean values and standard deviation (SD). Data was evaluated for normalcy using a Shapiro–Wilk Test, followed by the statistical tests as outlined in the figure legend. Mean values are shown as bars, and the error bars represent the SD. The following values denote statistical significance: ∗p < 0.05, ∗∗p < 0.01.