RXR agonist, 9-cis-13,14-dihydroretinoic acid (9CDHRA), reduces damage and protects from demyelination in transsynaptic degeneration model

Neurodegenerative and demyelinating disease, such as multiple sclerosis (MS) are at the forefront of medical research and the discovery of new drugs and therapeutics. One phenomenon of degeneration seen in these diseases is transsynaptic degeneration (TSD), where damage from one axon spreads to the other axons that are connected to it synaptically. It has previously been found that demyelination occurs prior to neuronal loss in an experimental form of induced TSD. Retinoid-x receptor (RXR) agonists have been shown to promote remyelination. Therefore, this study aimed to reveal the effects of a novel endogenous RXR-γ agonist, 9-cis-13,14-dihydroretinoic acid (9CDHRA), on preventing or restoring the effects of TSD. 9CDHRA was administered to mice following optic nerve crush (ONC) procedures, and electrophysiology (visual evoked potential, VEP) and histological (immunofluorescent) assessments were performed. It was found that 9CDHRA treatment effectively delayed glial activation and reduced the presence of apoptosis at the site of injury and further anterogradely in the visual system, including the lateral geniculate nucleus (LGN) and primary visual cortex (V1). Most notably, 9CDHRA was able to maintain myelin levels following ONC, and effectively protected from demyelination. This was corroborated by VEP recordings with improved P1 latency. The promising findings regarding the injury attenuating and myelin protecting properties of 9CDHRA necessitates further investigations into the potential therapeutic uses of this compound.


Introduction
Many neurodegenerative diseases have several forms and aggravating components of injury which lead to the degeneration and symptoms seen, such as those in Alzheimer's disease, multiple sclerosis, or Parkinson's disease (Sharma et al., 2022).A form of neurodegeneration found in these diseases is transsynaptic degeneration (TSD), where injury found in one neuron or axon can be spread to the other neurons synaptically connected to it (Sharma et al., 2022).For example, damage found in the posterior pathway spreading retrogradely (retrograde TSD, rTSD) to the retina has been quantified in MS by Gabilondo and colleagues (2014).Alongside evidence of retrograde TSD, anterograde or forward moving TSD was also evidenced in.Similar findings regarding TSD have also been implicated in other degenerative diseases including Parkinson's disease and Alzheimer's disease (Sharma et al., 2022).This further highlights the importance of understanding the process and mechanisms behind TSD.
There are theories and evidence pointing to a relationship between demyelination, and the presence of TSD, which has previously been investigated to aid in the foundations of this study (Parrilla et al., 2024).This has been seen in animal models of multiple sclerosis and Parkinson's (Ferreira et al., 2021;Wagenknecht et al., 2016), as well as in human case studies (Beatty et al., 1982).For example, in an animal model of optic nerve injury it was found that demyelination and glial activation in the posterior visual system occurs prior to axonal loss (You et al., 2019).This indicates that demyelination and their associated processes are a potential driving force of TSD.These results were paralleled in observations regarding changes in myelin integrity and the presence of TSD in glaucoma, multiple sclerosis, and optic neuritis patients.Following this, if myelin integrity can be maintained in the face of neurodegeneration, such as that of TSD, it may also protect neuronal integrity.Lastly, it has been noticed that the detection of neurodegeneration in TSD is a long, subtle process that occurs secondarily following injury or other exacerbating circumstances.This is supported by the fact that other features are seen as degenerated prior to axonal loss (e.g.demyelination occurs prior to neurodegeneration), and that secondary evidence of TSD in patients is detected over years of progression (Murphy et al., 2022;You et al., 2019).
Therapeutics or preventative therapies targeting remyelination are of interest in the same line of neurodegenerative diseases, specifically as a therapeutic for demyelination seen in multiple sclerosis.For example, retinoid X receptors (RXR) and their agonists have been shown to promote remyelination in various animal models of neurodegenerative diseases (Huang et al., 2011;Santos-Gil et al., 2021).This has been attributed by RXR's function as a transcription factor and gene transcription regulation, specifically a positive regulation on OLG precursor differentiation which encourages remyelination (Dawson and Xia, 2012;Huang et al., 2011).Bexarotene, a synthetic RXR agonist, is an approved cancer therapeutic, but has also been shown in animal models to promote remyelination (Santos-Gil et al., 2021).In a clinical trial, bexarotene was administered to relapsing-remitting multiple sclerosis patients and revealed improvements in electrophysiological assessments (Brown et al., 2021).This includes improved latency in visual evoked potential (VEP) responses, which has been previously shown to accurately reflect myelin integrity in the visual system (Leocani et al., 2018;You et al., 2011).However, most patients on bexarotene treatment suffered various adverse events and the drug was deemed unsuitable for multiple sclerosis patients (Brown et al., 2021).One reason for this may be the unspecific nature of bexarotene as a non-selective RXR agonist, and the unspecific agonism of all RXR isoforms (alpha, beta, and gamma) could have potentiated widespread negative symptoms (Dawson and Xia, 2012).A more selective agonist towards the RXR gamma (RXR-γ) pathway may be more beneficial as it is most widely found in the brain compared to other RXR forms, and has been shown that specific RXR-γ agonists promote OLG differentiation and further remyelination (Huang et al., 2011;Yang et al., 2017).
This study sought to further extrapolate the relationship between TSD and myelin, such that preventing demyelination may also prevent neurodegeneration.This study also aimed to determine the efficacy of a novel RXR-γ endogenous ligand, 9-cis-13,14-dihydroretinoic acid (9CDHRA) in promoting remyelination.Utilising an animal model of TSD, 9CDHRA was administered to mice who underwent an optic nerve crush (ONC) to determine if they would be protected from demyelination and subsequent neuronal loss through TSD.Electrophysiology (VEP) data and postmortem histological analyses were compared to animals who did not receive the drug or who did not undergo ONC to determine the extent of damage from TSD and the level of protection from demyelination occurred with 9CDHRA treatment.

Animal wellbeing and treatment
Mice (male n = 9) were acquired (Ozgene ARC, Perth, Australia) at 6 weeks old.This number of animals was chosen to ensure an even number of mice per treatment group (n = 6) while maintaining an equal representation of male and female animals with sufficient in-group replicates (n = 3 per group).These mice were randomly allocated to one of three groups such that there were three female mice and three male mice per group • Group D (9CDHRA): administered 9CDHRA only • Group C (ONC): ONC operation only • Group B (ONC+9CDHRA): administered 9CDHRA and ONC operation All mice were monitored for general wellbeing throughout the study, including weight and overall appearance (coat appearance, gait, breathing, etc.).Food and water were provided ad libitum and animals were given standard enrichment in their housing cages.Half of the mice from each group (n = 9) were culled four weeks after acquisition, and the remaining mice (n = 9) were culled at five weeks.A timeline for study procedures has been supplied in supplementary data (supplementary Fig. 1).All procedures were conducted in compliance with the Australian code for the care and use of animals for scientific purposes as defined by the National Health and Medical Research Council (NHRMC), and in accordance with the Macquarie University Animal Ethics Committee (AEC 2020/007-11).

Optic nerve crush procedure
The mice that received ONC procedures (ONC+9CDHRA and ONC only) underwent the operation two weeks after acquisition (week three).This was done by first incising into the conjunctiva of one eye to gain access to the surrounding orbital muscles.The orbital muscles were blunt dissected to create a point of access to the optic nerve.Once the optic nerve was visible, it was crushed by hand at approximately 2 mm retrobulbar for five seconds using blunt forceps.This procedure has similarly been defined in previous studies such as that by Du and colleagues (2020).All mice were given ciloxan eyedrops 0.3% (Alcon Laboratories, NSW, Australia) and maxidex 0.1% (Novartis Pharmaceuticals, NSW, Australia) for post-operative care.The mice were regularly monitored into the following day and any symptoms or adverse events were recorded as needed.
One male mouse from the ONC only group began to bleed during the procedure which was terminated before it could be completed to ensure the mouse's safety.The mouse was given all post-operation care per protocol, and there were no further complications following this.All VEP recordings and histological analyses from the afflicted eye were omitted from the following results.The fellow eye has been included as it was not impacted and remained suitable as a control.9-cis-13,14-dihydroretinoic acid administration 9CDHRA (Acanthus, cat.no.ACA-160929-0001) was prepared into a 10 μM dose in 3% ethanol (ETOH) and 3% dimethyl sulfoxide (DMSO, Sigma-Aldrich ref. no D2650) in sunflower oil.50 μL was administered to the ONC+9CDHRA and 9CDHRA only mice immediately prior to VEP recordings conducted on weeks two and four.3 mL/kg of this solution was administered via intraperitoneal injection immediately following anaesthesia procedures as defined in the following section.Mice in the ONC only group were administered 50 μL of saline (NaCl) in the same vehicle and at the same timepoints.

Visual evoked potential
All animals were dark adapted starting the evening before VEP procedures (for at least 12 h) in a red wavelength lit room and remained in these conditions throughout all recordings.Mice were anaesthetised prior to recordings with 50 mg/kg ketamine (Provet, NSW, Australia) and 0.5 mg/kg medetomidine (Troy Laboratories, NSW, Australia), and were administered either drug or vehicle according to group allocation via intraperitoneal injection.1% Tropicamide (Alcon Laboratories, NSW, Australia) was administered to both eyes for pupil dilation.Once the mouse was sufficiently unconscious and did not respond to tactile stimulus, they were placed onto a heated stage (Phoenix Research Labs) within a cage also housing the ERG machine (Phoenix Ganzfeld ERG, Phoenix Research Laboratories).One reference electrode was inserted subcutaneously between the eyes pointing inferiorly towards the nose.A second recording electrode was inserted subcutaneously on either side of the skull such that the electrode was positioned superior to the contralateral visual cortex of the eye being stimulated, as optimised in previous studies (Liu et al., 2022).A final grounding electrode was inserted at the base of the tail.The mouse's position was adjusted on the stage such that the objective lens of the ERG machine could be centred over the eye to be recorded.Once properly positioned, flash stimuli were produced and responses recorded through LabScribe2 software (Phoenix Research Labs ERG module).VEP measurements utilised 60 green LED flash stimulus (intensity 3.1 log (cd sec / m 2 )) shown for 1 millisecond (ms) across a total of three runs, with 90 sweeps per run.This includes a 2 ms delay between stimuli.After recording from both eyes, the mice were removed from the cage and subcutaneously administered 0.75 mg/ kg atipamezole (Troy Laboratories, NSW, Australia) to reverse anaesthesia.All mice remained in the darkened environment until all recordings had been completed and were monitored as needed.
The information measured from VEP's were analysed to determine the P1 latency and N1-P1 amplitude.P1 latency was determined by the amount of time (ms) from baseline to the first significant P1 incline, occurring around 100-120 ms.N1-P1 amplitude refers to the average amplitude recorded from the first N1 negative peak to P1.Within figures, P1 latency from individual groups are graphically as a horizontal scatter plot with bars, dictating the mean and standard deviation (SD).The average P1 latency per group at baseline to two weeks post ONC are displayed as connected scatter plots and denotes the mean and SD.Average P1 latencies from baseline to two weeks post ONC are shown in connected scatter plots.N1-P1 amplitude utilises a vertical scatter plot with error bars, also showing mean and SD.Average N1-P1 amplitude from baseline to two weeks post ONC are shown in connected scatter plots with SD.Figures displaying summarised VEP responses from each group and eye are shown in supplementary materials (supplementary Fig. 2).

Histology Tissue collection and preparation
All animals were anaesthetised with isoflurane via inhalation until sufficiently unconscious and then culled by cervical dislocation.Sacrificed mice were intracardially perfused with saline and the brains and eyes with optic nerves attached were harvested.All tissue was drop fixed in paraformaldehyde (PFA, 4% in 0.1 M sodium phosphate solution), with collected brains remaining in PFA for 24 h and eyes for 2 h.They were then moved to sucrose in PBS until fully saturated in preparation for freezing.
The first 1/3 from the rostral end of the brain and the cerebellum were cut from the brain with a scalpel, and the remaining tissue arranged in a specimen mould such that the rostral (cut) side was facing downwards and the dorsal side up.The specimen was covered in optimal cutting temperature compound (OCT compound,Sakura,ref. 4583) and placed on dry ice until fully frozen.
The optic nerve was removed from the most proximal end to the eye with Noyes scissors.The optic nerve was then put in a specimen mould filled with OCT, with the proximal (cut) end facing downwards and oriented vertically.This was also placed on dry ice until fully frozen.
All samples were stored at − 80C until ready to be used.Both optic nerves and brains were cut coronally into cross-sections at 15 μm thickness using a cryostat (Leica CM 1860 or Leica CM 1950).The prepared slides were kept at − 80C in a slide box until needed for immunofluorescent analyses.

Immunofluorescent methods
Slides were thawed from − 80C to room temperature and then rinsed with phosphate buffer solution (PBS).Blocking solution, which included 5% donkey serum and 0.3% Triton X-100 (Lomb Scientific Pty Ltd), was applied for three hours at room temperature under foil.The following primary antibody diluents were applied in a solution containing bovine serum albumin (BSA) and 0.3% Triton X-100 in PBS: • Rabbit anti-myelin basic protein (MBP) (ab40390, Abcam) at 1:500 • Rabbit anti-cleaved caspase 3 (CC3) (CS9661, Cell Signalling Technology) at 1:500 • Mouse anti-glial fibrillary acidic protein (GFAP) (CS3670, Cell Signalling Technology) at 1:500 These primary antibodies were chosen for different purposes.MBP is one of the most abundantly expressed extrinsic membrane protein and is commonly used to detect and measure the presence of myelin (Deber and Reynolds, 1991).The presence of MBP was measured to determine the presence and extent of demyelination.GFAP is an intermediate protein found in astrocytes within the central nervous system (CNS) and is closely involved in the maintenance of normal glial activities and of activated (astrogliosis) states (Yang and Wang, 2015).Reactive astrogliosis is the state of hypertrophy and proliferation that glial cells undertake following injury, such that the expression and size of GFAP increases (Yang and Wang, 2015).The immunofluorescent intensity and measurement of GFAP fluorescent area was used to determine the presence of astrogliosis, such that an increase would indicate reactive glia in response to damage or injury.Lastly, CC3 is one of many caspases responsible for the cleavage that occurs during the apoptosis process, and the fragments of these caspases are visible following this cleavage (Crowley and Waterhouse, 2016;Strasser et al., 2011).Apoptosis is utilised here to detect and measure cell death and overall, nonspecific degeneration.
Multiple quality control measurements were performed prior to and throughout the immunofluorescent analyses for this study.Antibodies that have been previously validated (used in previously published research or used in previous studies within the same lab).Antibody targets were tested to ensure that the target specificity was to standard and there was minimal background staining in control tissues.There were also multiple replicates done of the defined immunofluorescent methods as "practice" runs.

Microscopy and image acquisition
All images were acquired using a fluorescent microscope (Zeiss Axio Imager 2) and accompanying software Zeiss ZEN Blue 3.5 (RRID: SCR_013672).Representative images of each optic nerve were acquired at 20x magnification in a single or 2x2 tiled image as necessary.Images of the lateral geniculate nucleus and visual cortex area were acquired at 10x and 20x from each side of the brain.These images were split by their individual fluorescence channel and were converted to 8Bit TIF image types.In ImageJ/FIJI, the fluorescence of each channel was determined by first thresholding the image: The integrated density (IntDen) of each image was calculated by: mean grey value × area For the purposes of this manuscript, all example immunofluorescence images have been chosen based on overall quality (e.g.not blurry, able to be resized without warping, accurately reflects findings, etc.).No images have been altered with or adjusted except for the MBP images which have been digitally recoloured to violet for better identification.
All images have been automatically adjusted for best visual quality in Zeiss ZEN Blue 3.5 by choosing "auto" under histogram options.

GFAP/Glial morphology
Slides stained for GFAP were converted into a binary image in ImageJ/FIJI.The threshold level of each image was automatically derived using the previously defined methods.The pixels that fall above the average background threshold are made white in the binary image.This is then skeletonised such that pixels along the edge of each object are removed until a single pixel "skeleton" of the object remains.The area of this skeleton is then measured for the GFAP positive area within the LGN and optic nerves.Examples of the skeletonised images are shown in supplementary materials Fig. 4.These methods were adapted from Bosco and colleagues (2016).

Statistics
All statistical analyses were performed in GraphPad version 10.1.2.As the majority of data sets acquired in this study were not normally distributed, all statistical analyses have been performed using nonparametric tests (Scheff, 2016).Details regarding the normality of data sets that support this decision have been provided in supplementary data (supplementary Table 1).
Tests performed between two treatment groups with paired samples used Wilcoxon matched-pairs signed rank test and are reported with the sum of signed ranks (W) and p-value, and confidence intervals (CI) will be included for significant results.Similar tests between treatment groups with unpaired samples utilise Mann-Whitney tests, and are reported as Mann-Whitney U-value (U) and p-value, with CI reported as needed.The specific test used in graphical representations (connected scatter plot denoting change over time) has been detailed in the figure caption, and the distribution of rank sums is represented by the error bars within the graphs.
Paired tests to determine variance across multiple treatment groups were conducted using Friedman tests, followed by Dunn's multiple comparison test, which gives a Friedman statistic (M), Z value, and adjusted p-value and are reported as such (Friedman, 1937).Unpaired tests similarly conducted across multiple groups utilised Kruskal-Wallis test and were also followed by Dunn's multiple comparison test.This also renders a Z value and adjusted p-value, with a Kruskal-Wallis statistic (H).It should be noted that this test calculates the mean rank difference, while the previously mentioned Friedman with Dunn's test measures rank sum difference.These will be reported following significant results.Figures for both tests use box and whisker plots, with the specific test used denoted in the figure caption, except for VEP data previously defined.
Within figures, statistical analyses that rendered a "significant" pvalue are indicated with an asterisk, where the number of asterisks represents a range of p-values.A p-value less than or equal to 0.05 has one asterisk (*), p-values less than or equal to 0.01 has two asterisks (**), p-values less than or equal to 0.001 has three asterisks (***), and any pvalue less than 0.0001 has four asterisks (****).The specific value of each significant p-value is also specified in the caption of each figure.
A linear progression was used to create a graph that would represent a change over time, such as that used to represent the increase in CC3 positive cells within the LGN (Fig. 6).The equation for the fitted line is y = mx + b, where m = the slope of the line and b = the y-intercept of the line.The slopes were also assessed to determine if they were significantly different from zero, and the p-value is reported following the equation for the line.Lastly, the slopes of multiple lines were analysed using ANCOVA to establish if they are significantly different from each other, and the corresponding F statistic (F) and p-values are reported as necessary.

9CDHRA safe in mice
After administering 9CDHRA to mice, there was no negative affect on their physiology.This includes observations on gait, coat appearance, and breathing.There were no observations outside of the normal or healthy expectation for any mouse throughout the study time.Some noticeable differences were seen between sex and between groups.The mice that underwent ONC but did not receive 9CDHRA treatment had significantly greater weights than mice that did receive treatment at one week post ONC (+0.5185 g, W = 21.00,p = 0.0312, CI 96% − 1.000, 0.000) and at two weeks post (+2.259,W = 45.00,p = 0.0039, CI 96% 1.333, 3.000) (Supplementary Fig. 3A).The average weight of female mice in control group D (9CDHRA only) was significantly greater than that of female mice in group B (ONC+9CDHRA) (+4.296 g, Z = 2.593, M = 10.57,p = 0.0190) (Supplementary Figure 3B).Details of these results are supplied in supplementary materials (Supplementary Table 2).
No mice from any group or any sex experienced sudden weight loss or gain greater than 5% of their previously measured bodyweight.

P1 latency restored after two weeks
The P1 latency, or the amount of time until the first major positive deflection following the N1 deflection, was measured at the three defined timepoints (baseline, one week post ONC, and two weeks post ONC).There were no significant differences between either eye, between the three groups at baseline (Fig. 1A).After one week, there was an increase in P1 latency for both the treated and untreated mice crushed eyes compared to the control group, however it was not significant (9CDHRA vs ONC+9CDHRA Z = 1.443,M = 2.333, p = 0.2978, and 9CDHRA vs ONC Z = 0.2887, p > 0.9999) (Fig. 1B).Further, the untreated mice had a non-significantly slower P1 latency in both the crushed and control eye compared to treated mice (crushed eye + 10.98 ms ± 24.17, W = -2.000,p = 0.5625, and control eye + 10.57 ms ± 14.99, W = -7.000,p = 0.8438) (Fig. 1B).
After two weeks, the untreated mice continued to have a prolonged P1 latency in both the crushed and control eye compared to the control mice (crushed eye +23.27 ms ± 15.21, Z = 0.000, M = 2.000, and control eye +34.87 ms ± 20.52, Z = 0.4082, M = 0.6667, p > 0.9999 for both values).However, the mice given 9CDHRA (ONC+9CDHRA) had restored P1 latency back to their baseline levels (crushed eye 2.567 ms difference, control eye 4.533 ms difference) (Fig. 1A).
Untreated mice had a slight but not significantly prolonged P1 latency in the crushed eye compared to control eye at one week post (+8.980 ms ± 23.98, W = 5.000, p = 0.6875), which was reversed in week two such that the control eye had a more prolonged latency (+13.30ms ± 40.22) (Fig. 1B).The ONC+9CDHRA mice had more stable P1 latency times between the crushed and control eyes at one week post (+8.567 ms ± 15.23) and two weeks post ONC (+1.467 ms ± 9.758) (Fig. 1B).

N1-P1 amplitude reduced after optic nerve crush
The amplitude between points N1 and P1 for mice that underwent ONC procedures, with or without 9CDHRA treatment, were compared to the control mice.One week following ONC both groups had decreased N1-P1 amplitudes for the crushed eye (ONC+9CDHRA − 13.38 μV and ONC − 14.34 μV).This was not significantly different from the control group for either (Z = 1.732,M = 4.000, and p = 0.1665 for both values) (Fig. 1C).
All ONC afflicted mice had further reduced N1-P1 amplitudes at two weeks post ONC in the crushed eye (ONC+9CDHRA − 1.52 μV and ONC − 2.11 μV).This reduction over time was significant for the treated mice (Z = 2.309, M = 9.333, p = 0.0418 from baseline to one week post, and Z = 2.887, p = 0.0078 at two weeks post) (Fig. 1D).The reduction for the group without the drug was not significant at one week post (Z = 1.588,M = 7.913, p = 0.2247) but was significant at two weeks post ONC (Z = 2.742, p = 0.0122) (Fig. 1C).This was also the case for the uncrushed eye in untreated mice at two weeks post (− 15.40200 μV, Z = 2.309, M = 6.333, p = 0.0418).There were no significant differences between ONC afflicted mice for any eye (crushed or control) or at any timepoint (one week or two weeks post ONC) (Fig. 1C, D).

Optic nerve assessments shows delayed response to ONC damage following 9CDHRA treatment
The optic nerves of mice were analysed to determine if 9CDHRA had any local affects at the site of injury following ONC (example images shown in Fig. 2D).GFAP fluorescence intensity in the crushed optic nerve was determined to have increased for both ONC afflicted groups (with or without 9CDHRA treatment) at one week post compared to control group, but this was not significant for either (ONC+9CDHRA +824907 AU, Z = 2.092, H = 4.954, p = 0.0728 and ONC +983846 AU, Z = 2.101, p = 0.0713, respectively) (Fig. 2A).
The surface area of the optic nerve cross section occupied by glial processes were also measured (example skeletonised images shown in supplementary Fig. 4).Compared to the control group, the mice without 9CDHRA treatment had significantly more activated glia area within the optic nerve at one week (crushed +4394 μm 2 , Z = 3.358, H = 11.32,p = 0.0016, and control +2891 μm 2 , Z = 2.812, H = 7.950, p = 0.0098) and two weeks post (crushed +2953 μm 2 , Z = 2.305, H = 5.331, p = 0.0424) (Fig. 2B).The mice with 9CDHRA treatment also experienced increased active glia within the crushed eye area at one week and two weeks post which was not significantly greater compared to the control mice (one week +3101 μm 2 , Z = 2.169, and two weeks +2738 μm 2 , Z = 1.956 respectively) (Fig. 2B).
Although the area of positive GFAP staining was less in the crushed optic nerves of mice that received 9CDHRA than the mice that did not, this difference was not significant one or two weeks after the procedure (one week − 1060 μm, U = 21, and two weeks − 122.7 μm, U = 41) (Fig. 2Bi).There was also greater disparity between the crushed and control eyes in mice that did not receive 9CDHRA at one week post operation (crushed eye +1503 μm), which stabilised by week two (crushed eye +22 μm).Mice with 9CDHRA had a similar experience (one week crushed eye +1022 μm, and two weeks crushed eye +1 μm) (Fig. 2Bi).There were no significant differences between ONC afflicted mice that did or did not receive 9CDHRA at one week or two weeks following ONC.
The presence of CC3, a marker of cells that are dying of or have died from apoptosis, was also measured (example images shown in Fig. 2D).One week following ONC, the crushed optic nerves without 9CDHRA treatment had significantly greater CC3 fluorescence intensity compared Fig. 1.P1 latency restored in 9CDHRA treated mice at two weeks post ONC, but did not influence N1-P1 amplitude.Note: (A-B) While the P1 latencies for all mice were stable at baseline (A), the crushed optic nerves from 9CDHRA treated and untreated mice decreased one week following ONC procedures.However, the following week while the untreated mice in group C continued to have lengthened P1 latencies, the treated mice in group B had latencies restored to their baseline levels.(B) At both timepoints following ONC, the untreated mice (ONC) had longer P1 latencies than the treated mice (ONC+9CDHRA).(C-D) One week following ONC, the mice that underwent ONC procedures (ONC+9CDHRA and ONC only) had reduced N1-P1 amplitudes compared to those in the control mice (C).This reduction was more dramatic the following week for the treated and untreated mice.This decrease over time was significant for the ONC+9CDHRA mice at one week and two weeks post ONC (Friedman test and Dunn's multiple comparison test, *p = 0.0418 and **p = 0.0078), but was only significant in the untreated mice at two weeks post ONC (crush *p = 0.0122 and control *p = 0.0418) (D).

Evidence of protracted glial activation in the lateral geniculate nucleus
Following the trajectory of the visual system from the optic nerve, the LGN was also assessed to determine the presence of aTSD.GFAP fluorescence intensity was measured at one week and two weeks post ONC with the same procedures previously described for the optic nerves (example images shown in Fig. 7).After one week, untreated mice had greater fluorescence intensity within the LGN on the opposite side of the brain from the crushed eye (contralateral LGN) than the control group (+163581 AU, Z = 2.417), while the treated mice was slightly decreased (− 19583 AU, Z = 0.7306) (Fig. 5A).
For the ONC+9CDHRA mice, there was a slightly greater GFAP positive area within the contralateral LGN compared to the ipsilateral LGN at week one (+692 µm) and week two (+693 µm) (Fig. 5B).The ONC only mice maintained a more symmetrical area at week one (+37 µm) and week two (+229 µm).These differences between LGNs were not significantly different (week one ONC+9CDHRA Z = 0.5812 and ONC Z = 0.2997, and two weeks ONC+9CDHRA Z = 0.3935 and ONC Z = 0.2997).

9CDHRA treatment reduced apoptosis in lateral geniculate nucleus
The presence of CC3 was quantified by measuring the fluorescence intensity and manually counting the number of CC3 positive cells within the LGN (example images shown in Fig. 7).One week following ONC, the group without 9CDHRA treatment had slightly greater fluorescence intensity in the contralateral LGN than the group with treatment (ONC vs ONC+9CDHRA +136910 AU) and the control group (9CDHRA vs ONC − 495243 AU, Z = 1.375) (Fig. 6A, Ai).The ipsilateral LGN also exhibited increased fluorescence intensity for untreated mice compared to the treated mice (+2228396 AU, U = 101, p = 0.0016, CI 95% − 4750205, − 1079967).The mice given 9CDHRA was more similar to that of the control group (9CDHRA vs ONC+9CDHRA ±179166.5, Z = 0.8915) (Fig. 6A).
To further validate this, a linear progression was fit to determine the change of CC3 positive cells over time (Fig. 6C).The slope of the lines for both the contralateral and ipsilateral LGN in the untreated mice were negative (m = − 154.7, p = 0.0530, and m = − 82.39, p = 0.0004, respectively).The slope for the group that received treatment (contralateral m = 87.13,and ipsilateral m = 124.9)and for the control mice (ipsilateral m = 126.4,and contralateral m = 111.9)were positive and all significantly non-zero (ONC+9CDHRA contralateral p = 0.00048, ONC+9CDHRA ipsilateral p = 0.0342, 9CDHRA ipsilateral p = 0.0028, and 9CDHRA ipsilateral p = 0.0047).However, there was no significant difference between these slopes for either the contralateral or the ipsilateral LGN in 9CDHRA+ONC treated mice when compared to control mice (p = 0.9768 and p = 0.6246 respectively) (Fig. 6C).The slopes for the ipsilateral lines of ONC+9CDHRA mice (m = 124.9)compared to the ONC only mice (m = -82.39)were significantly different from each other (F = 14.25, p = 0.0007).This was also the case for the contralateral lines (ONC+9CDHRA m = 87.13vs ONC m = -154.7,F = 23.40,p < 0.0001).

Late apoptosis fluorescence intensity increase in visual cortex of mice administered 9CDHRA
The fluorescence intensity of CC3 within V1 was measured on the contralateral side of both groups of mice that underwent ONC G.E. Parrilla et al. procedures and compared to the control group of mice.Mice that did not receive 9CDHRA had greater fluorescence intensity within the contralateral LGN than treated mice (+1779659 AU, U = 11) and control mice (+1424545 AU, Z = 0.02942, H = 0.1524) (Fig. 8B, Bi).This was similarly the case two weeks post ONC between treated and untreated mice (+1512280 AU, U = 31) and between untreated and control mice (+5401011 AU, Z = 1.833,H = 3.880) (Fig. 8B, Bi).

9CDHRA treatment does, and does not, align with prior research into RXR agonists
The aim of this study was to determine any protective or restorative effects of 9CDHRA, an RXR-γ agonist, on the transsynaptic degeneration effects seen after optic nerve crush.Within the optic nerve, 9CDHRA treatment delayed glial activation and apoptosis at the site of injury, and the effects of both were less severe.Similar effects were seen anterogradely along the path of the visual system.This includes the delayed glial activation and decreased apoptosis seen in the optic nerve, as well as maintenance of myelin.These results were reflected in the latency and amplitude measurements across VEP recordings.
Following ONC, it was found that 9CDHRA attenuated the insult to the surrounding area.This has similarly been seen in other animal models of traumatic injury.For example, in a mouse model of traumatic brain injury, treatment with a non-specific RXR agonist, bexarotene, promoted repair and reduced the overall insult (He et al., 2020).This included increased MBP expression, increased axonal sprouting, and reduction of pro-inflammatory microglia and macrophages.This was attributed by the authors to interactions with peroxisome proliferatoractivated receptors, which can promote an anti-inflammatory environment (He et al., 2020).Similar use of bexarotene has been useful in other models of ischemic or inflammatory neuronal injury.In a study from Ishihara and colleagues (2019), rat hippocampal slices were injured through oxygen-glucose deprivation or by treatment with lipopolysaccharide.They found that bexarotene treatment protected neurons by attenuating oxidative stress and neurotoxicity following either treatment (Ishihara et al., 2019).The "damage control" abilities from a similar RXR agonist in different models of traumatic injury could also protect from the further spread of damage through TSD.
Proceeding this, the LGN of brains affected by TSD but given 9CDHRA treatment had significantly decreased markers of apoptosis and delayed glial activation.This was also the case further along the visual system within V1.These effects, especially the attenuation of reactive astrocytes and activated microglia, have similarly been seen in other animal models of traumatic brain injury and AD.In the previously mentioned study from He and colleagues (2020), following controlled cortical impact bexarotene treatment was able to not only down regulate pro-inflammatory microglia and macrophages, but it can also promote anti-inflammatory ones.This effect of RXR agonism may be evidenced by the findings in this study that GFAP fluorescence intensity was delayed, while the area occupied by GFAP positive processes remained.Similar results regarding decreased inflammation and astrogliosis, or reduction of reactive microglia have also been seen in animal models of AD treated with bexarotene (Mariani et al., 2017;Munoz-Cabrera et al., 2019).
In lieu of direct measurements of neuronal density, CC3 presence was utilised to measure degeneration, as the presence of neurodegeneration in TSD is a subtle, secondary effect.Within the LGN and V1 of mice that underwent ONC procedures and received 9CDHRA treatment, there was a delay in CC3 fluorescence intensity increase to two weeks post ONC.This contrasts with the mice that underwent ONC procedures without treatment, as they exhibited an already increased fluorescence intensity.A similar pattern was seen for the number of CC3 positive cells in ONC+9CDHRA mice, which significantly increased from one week to two weeks post.Lastly, N1-P1 amplitude derived from the VEP measurements are also indicative of neuronal integrity (You et al., 2011).This also enables an indirect measurement of damage to neurons within the visual system, as these measurements have been shown to be correlated.
The culmination of results indicating damage and degeneration following ONC procedures in the treated and untreated mice aligns with some prior research regarding RXR agonists and their neuroprotective or neuroregenerative abilities.There is a wealth of studies that show RXR agonists can promote axonal regeneration in animal disease models.For example, bexarotene was able to improve neuronal survival in a mouse model of AD, as well as increasing presynaptic and postsynaptic protein expression (Mariani et al., 2017).Similar treatment in a different AD mouse model also increased neuronal nuclei (NeuN) reactivity along with improved synaptic transmission and plasticity as neurons recovered (Munoz-Cabrera et al., 2019).However, there are many reports that RXR agonism has no effect on neuronal survival or regeneration, such that the signs of neurodegeneration (brain atrophy, decreased mature neurons) are unaltered (Balducci et al., 2015;Tachibana et al., 2016).At a cellular level, stem cell subjected to 9cRA experienced decreased gene expression of neuronal biomarkers while promoting astrocyte markers.This indicates that these cells were being directed towards glial differentiation rather than neural differentiation (Kubickova et al., 2023).Also, RXR agonism may improve expression of certain neuronal related proteins, such as postsynaptic proteins and apolipoprotein E (apoE), without altering mature neurons (Tachibana et al., 2016).It is possible that there were no neural regenerative effects found, as CC3 levels increased and N1-P1 amplitudes decreased, because there were other, untested factors that went unnoticed, such as the documented increase of postsynaptic and presynaptic proteins.Another possibility refers to the length of time mice were allowed post ONC.Some studies that find bexarotene or other RXR agonists to restore or protect neurons do not detect this until later stages.For example, bexarotene has been shown to promote microglial and astrocyte activation at earlier timepoints than if they also detect neural (NeuN) activation (Saibro-Girardi et al., 2023;Tachibana et al., 2016).It is possible that more or longer time points would have revealed neural regeneration since increased glial activation was detected at two weeks post ONC.The potential benefits of 9CDHRA could be improved with evidence from further research regarding its specific interactions with other processes, such as interactions with glial activation, and more long-term effects on neuronal survival following damage.
Another notable finding was the ability of 9CDHRA to not only prevent demyelination through TSD, but also maintain myelin integrity throughout the course of this study in mice that underwent ONC procedures.This has been expressed in vitro, as 9cRA is able to stimulate the transcription of MBP promoters and protein concentrations in cell cultures by promoting OPC differentiation (Huang et al., 2011;Pombo et al., 1999).Animal studies utilising disease models have also successfully used bexarotene to encourage MBP expression, which has also been attributed to RXR agonists ability to increase not only OPC proliferation, but it's differentiation into OLG's (He et al., 2020;Huang et al., 2011;Santos-Gil et al., 2021).This ability has also been specifically credited to RXR-γ, further validating the ability of 9CDHRA to protect myelin.For example, the effects of 9cRA on OPC differentiation into OLGs was reversed when an RXR-γ antagonist (HX531) was introduced (Yang et al., 2017).RXR-γ is also expressed by OLG cells during remyelination, such as those found in post-mortem MS lesions, and in animal models of focal demyelination and subsequent remyelination (Huang et al., 2011).This was further validated when, in an animal knockdown model that does not express RXR-γ, was treated with 9cRA and increased remyelination (Huang et al., 2011).
Lastly, the weight observations between groups revealed that mice that received 9CDHRA treatment had lower weight then those who did not.This was true for the group as a whole and when controlled for sex.This may be related to patient studies that utilise RXR agonists, such as bexarotene, and found significant adverse effects.In a study where MS patients were administered bexarotene, most patients experienced negative side effects including hypothyroidism, increased triglycerides, and neutropenia (Brown et al., 2021).Although the researchers suggest that a specific RXR-γ agonist would reduce some of these effects, they also note that thyroid dysfunctions were still a likely possibility (Brown et al., 2021;Sharma et al., 2006).Further research could investigate thyroid specific antibodies, such as thyroid peroxidase antibodies, to determine if there are discrepancies in treated or untreated mice that may be causing this weight disparity.

VEP results go hand in hand with histological results
The electrophysiological assessments help validate the histological results, and vice versa.VEP's have been previously shown to accurately reflect demyelination and axonal damage.Mice with induced demyelination revealed strong correlations between recorded P1 latency delays and extent of demyelination (You et al., 2011).Contrariwise, shortening P1 latency indicates remyelination, as the presence of myelin allows for faster transmission across axons and require less energy to do so (Nave, 2010).There are also strong correlations between the N1-P1 peak amplitude and axonal loss, as it accurately reflects the number of functional optic nerve fibres (You et al., 2011).
Mice treated with 9CDHRA following ONC saw their P1 latency restored to their baseline levels (prior to ONC), such that their response time was more similar to control mice than untreated mice.Although multifocal VEP (mfVEP) has been shown to increase with age, the increase detected in untreated mice was greater than that seen in control mice (Brown et al., 2018).This corroborates 9CDHRA's ability to protect myelin.For example, animals with induced demyelination through a cuprizone diet experience increased VEP latency, which is improved when the cuprizone diet is ended and mice experience remyelination (Marenna et al., 2022).Remyelinating treatments, such as the antihistamine clemastine, have a similar effect to restoring latency delays as axons are remyelinated within the optic nerve.This has been expressed not only in an animal model of MS treated with clemastine (Cordano et al., 2022), but also in clinical trials for MS patients administered clemastine (Green et al., 2017).
The N1-P1 amplitude significantly reduced for all mice that underwent ONC (regardless of 9CDHRA treatment) from baseline to one week post, and further decreased the following week.Although the untreated mice had a slight, nonsignificant greater decrease than the treated mice, the rate of amplitude decay for the treated mice is more similar to that of the untreated mice and cannot be explained by normal reduction that occurs with age.Similar N1-P1 amplitude reduction has been seen in animal studies, such that the number of surviving axons is positively correlated with the N1-P1 amplitude response (Castoldi et al., 2022;You et al., 2011).This is also the case in animal models that experience remyelination, such as following cuprizone diet cessation, where the decreasing VEP amplitude indicates developing axonal degeneration (Marenna et al., 2022).However, this may be another time related issue, as previous studies utilising ONC in mice found that VEP amplitude can improve up to 80 days post injury onset (Du et al., 2020).This has similarly been seen in patients with optic neuritis, such that the changes to amplitude and latency seen after an optic neuritis episode can recover over the following three-month period (Alshowaeir et al., 2015).Therefore, a more longitudinal study design may have benefited this and allowed for regeneration to take its course.
The effects of 9CDHRA on protecting and/or attenuating the damage seen through transsynaptic degeneration following traumatic injury to the optic nerve further validates the therapeutic properties of RXR agonists and calls for further research to improve this area.Although the results of this study are promising, they also reveal some limitations that can be remedied in future investigations.For example, uncertainty regarding the validity of some results could be remedied by more comprehensive and longitudinal assessments.This was touched on previously, as it was suggested that more markers for pre-and postsynaptic proteins should be investigated for signs of axonal support through RXR agonist treatment.Furthermore, the lack of previous research utilising 9CDHRA leaves some questions regarding therapeutic doses and half-life of the endogenous ligand within animal models.Although the dose level utilised here was validated by two previous studies (Ruhl et al., 2015;Wietrzych-Schindler et al., 2011), a different regiment with more doses or doses of higher concentration may be required for the changes in insult (such as glial activity and apoptosis) to translate into neuronal survival.Not only would this improve the growing work of remyelinating and neuro-regenerative targets, but it would also promote the use of RXR agonists and further clarify its therapeutic qualities.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 2 .
Fig. 2. Glial activation and apoptosis attenuated within the optic nerves of mice treated with 9CDHRA.Note: (A) The fluorescence intensity within the optic nerves of all mice were compared at one week and two weeks following ONC.While there was an increase across ONC affected mice at one week, this was only significantly greater than the control mice for mice treated mice the following week (Kruskal-Wallis test and Dunn's multiple comparison test, **p = 0.0046 and *p = 0.0350).This represents a conflicting change between the treated and untreated mice, such that treated mice significantly increased (Mann Whitney test, *p = 0.0200 and **p = 0.0013) while untreated mice decreased (Ai).At this point, the fluorescence intensity for treated mice in group B was significantly greater than untreated mice in group C (**p = 0.0087).(B) The surface area occupied by active glial processes was also determined to be increased in ONC affected mice at one week and two weeks post.Mice with and without 9CDHRA treatment had a greater area at one week post, and this was significantly greater than the control mice for mice without 9CDHRA treatment (Kruskal-Wallis test and Dunn's multiple comparison test crushed **p = 0.0016 and control **p = 0.0098), and was also significantly greater at two weeks post (*p = 0.0424).The mice treated with 9CDHRA were not significantly different from the untreated or control mice (Bi).(C, Ci) The CC3 fluorescence intensity for mice that received 9CDHRA was not dissimilar to those of the control mice one week following ONC, while the untreated mice had significantly greater fluorescence intensity (Kruskal-Wallis test with Dunn's multiple comparison test, *p = 0.0355) At two weeks post ONC, the treated mice had significantly greater fluorescence intensity in the crushed optic nerves and control, uncrushed nerves (crush *p = 0.0429 and control *p = 0.0447).The increase for mice in group B was significant on both sides (Mann-Whitney test, **p = 0.0076 and *p = 0.0140).Meanwhile, the crushed optic nerves in group C mice had a significant decrease in fluorescence intensity (***p = 0.0007).From this, the mice in group B had significantly greater CC3 fluorescence intensity than group C (**p = 0.0024), contrary to the significantly lower fluorescence intensity seen the previous week (*p = 0.0101).(D) Representative images from the crushed eyes of 9CDHRA treated and untreated mice at one week and two weeks post ONC, in comparison with a control optic nerve showing GFAP (green) and CC3 (orange).Scale bars indicate 50 µm.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4 .
Fig. 4. MBP fluorescent images from the contralateral LGN.Note: MBP immunofluorescence (violet) images from the contralateral (TSD affected) LGN's of mice in group B (ONC+9CDHRA) and C (ONC only), compared to mice in group D (9CDHRA only).Scale bars within images indicate 100 μm.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5 .
Fig. 5. GFAP within the lateral geniculate nucleus (LGN) of 9CDHRA treated mice reveal prolonged glial activation.Note: (A) While untreated mice in group C exhibited increased GFAP fluorescence intensity when compared to control mice (Kruskal-Wallis test and Dunn's multiple comparison test, *p = 0.0313), the treated mice in group B were slightly decreased.Both groups of mice that underwent ONC procedures experienced increases in fluorescence intensity at two weeks post operation, and were greater than that of the control mice (ns).(Ai) This was a significant increase for both the ipsilateral LGN and contralateral LGN for mice that received 9CDHRA (Mann-Whitney test, contralateral **p = 0.0044 and ipsilateral **p = 0.0023) and the mice untreated mice (***p = 0.0007 and **p = 0.0022).(B) The surface area exhibiting GFAP fluorescence was slightly greater within the LGN's of treated mice compared to untreated mice at one week post ONC.However, both groups experienced increases such that they were similar at two weeks post ONC.Example images are shown in Fig. 7.

Fig. 6 .
Fig. 6.The fluorescence intensity and amount of CC3 positive cells in 9CDHRA treated mice experienced a delayed activation.Note: (A) The CC3 fluorescence intensity within the LGN of untreated mice in group C was greater than those of the control mice at one week post ONC (Kruskal-Wallis test and Dunn's repeated measures test, **p = 0.0055) and two weeks post ONC (control **p = 0.0069 and TSD **p = 0.0060).This is despite a decrease from one week to two weeks post (Mann-Whitney test, ***p = 0.0003).(Ai) The mice in group B also significantly decreased (****p < 0.0001 for both sides), but managed to have a significantly higher fluorescence intensity than group C mice at this time point (****p < 0.0001 for both sides).This is in contrast to the previous week, where the untreated mice had greater fluorescence intensity than the treated mice (**p = 0.0016).(B) One week following ONC procedures, the number of CC3 + DAPI double positive cells was significantly greater in untreated mice when compared to the treated mice (Mann-Whitney test ipsilateral ****p < 0.0001, contralateral **p = 0.0012) and the control mice (Kruskal-Wallis test and Dunn's repeated measures test, ipsilateral **p = 0.0015 and contralateral **p = 0.0096).(C) However, all mice from all groups had a similar number of CC3 positive cells at the two weeks post timepoint.This is due to a significant increase for ONC+9CDHRA (Mann-Whitney test, *p = 0.0106 and *p = 0.0407) and group D (**p = 0.0079 and *p = 0.0111), and a significant decrease for ONC mice (***p = 0.0007 and *p = 0.0360).This is further represented through a linear regression showing that while the slopes for untreated mice are negative, the slopes for the treated and control mice were both positive.Additionally, all slopes were significantly non-zero (as shown within the figure with the line equations) (C).

Fig. 7 .
Fig. 7. Images of GFAP (red) and CC3 (orange) immunofluorescence within the contralateral LGN's of mice.Note: For conciseness, only images of the contralateral (TSD affected) LGN for the treated and untreated mice are shown.LGN images from the control mice are also shown at the bottom.The images are organised from top to bottom as follows: ONC+9CDHRA 1 week post; ONC+9CDHRA 2 weeks post; ONC 1 week post; ONC 2 weeks post; 9CDHRA (control).Scale bars in the combined images denote 100 μm.
G.E.Parrilla et al.

Fig. 8 .
Fig. 8. GFAP and CC3 fluorescence intensity in the visual cortex (V1).Note: (A, Ai) The GFAP fluorescence intensity within the ipsilateral V1 of untreated mice was greater than that of the control mice (Kruskal-Wallis test with Dunn's multiple comparison test *p = 0.0192) and the treated mice (Mann-Whitney test p = 0.0480, not shown in figure).This further increased at two weeks post ONC, not only for the untreated mice (***p = 0.0007 and **p = 0.0047) but also the ONC+9CDHRA mice (ipsilateral **p = 0.0044 and contralateral **p = 0.0023).(B, Bi) The CC3 fluorescence intensity revealed a similar pattern, such that at one week post ONC the untreated mice had a greater fluorescence intensity, but the 9CDHRA treated mice increased to a similar level of the untreated mice the following week.