A bioisostere of Dimebon/Latrepirdine delays the onset and slows the progression of pathology in FUS transgenic mice

Abstract Aims To assess effects of DF402, a bioisostere of Dimebon/Latrepirdine, on the disease progression in the transgenic model of amyotrophic lateral sclerosis (ALS) caused by expression of pathogenic truncated form of human FUS protein. Methods Mice received DF402 from the age of 42 days and the onset of clinical signs, the disease duration and animal lifespan were monitored for experimental and control animals, and multiple parameters of their gait were assessed throughout the pre‐symptomatic stage using CatWalk system followed by a bioinformatic analysis. RNA‐seq was used to compare the spinal cord transcriptomes of wild‐type, untreated, and DF402‐treated FUS transgenic mice. Results DF402 delays the onset and slows the progression of pathology. We developed a CatWalk analysis protocol that allows detection of gait changes in FUS transgenic mice and the effect of DF402 on their gait already at early pre‐symptomatic stage. At this stage, a limited number of genes significantly change expression in transgenic mice and for 60% of these genes, DF402 treatment causes the reversion of the expression pattern. Conclusion DF402 slows down the disease progression in the mouse model of ALS, which is consistent with previously reported neuroprotective properties of Dimebon and its other bioisosteres. These results suggest that these structures can be considered as lead compounds for further optimization to obtain novel medicines that might be used as components of complex ALS therapy.


| INTRODUC TI ON
Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease primarily characterized by dysfunction and death of lower and upper motor neurons. There is no cure for this fatal disease, and the efficacy of Riluzole and Edaravone, the only drugs approved for treatment of ALS patients, is very low. Suggested mechanisms of action for these two drugs are different; Riluzole is believed to be neuroprotective because of its ability to block glutamate excitotoxicity, whereas Edaravone acts as a scavenger of oxygen radicals (reviewed in Refs1,2). A growing number of re-purposed and new drugs as well as alternative therapeutic approaches entering clinical trials (for recent review, see Refs3,4) raise hope that efficient treatment of different forms of familial and sporadic ALS could be achieved by designing tailored combinations of traditional pharmacological and modern biomedical approaches. Therefore, a search for potential components of efficient ALS therapy should include various types of potentially neuroprotective drugs, particularly those with pleiotropic effects on physiological and pathological processes in the nervous system.
In the last decade, gamma-carbolines have attracted attention as potential neuroprotectors in the nervous system affected by neurodegenerative processes. [5][6][7][8][9][10][11][12][13][14] Their neuroprotective activity can be explained by a combination of multiple mechanisms of action described for these compounds. 11,[15][16][17][18][19][20][21][22][23][24][25][26][27][28] Irreversible pathological aggregation of certain proteins contributes to pathogenesis of all types of ALS and therefore, it is an obvious target for therapeutic intervention. The first gamma-carboline found to affect pathological protein aggregation was Dimebon (also known as Latrepirdine), originally approved and for many years used as an antihistamine drug. An ability of Dimebon to prevent accumulation of cytoplasmic proteinaceous inclusions has been first demonstrated in cultured neurons expressing a highly aggregation-prone variant of RNA-binding protein TDP- 43. 29 Further studies in various cellular models of proteinopathies confirmed anti-aggregation properties of Dimebon and demonstrated its ability to activate autophagic mechanism of pathological aggregate elimination. 10,11,21,22,27,30 Consistently, chronic treatment with Dimebon, when started at a pre-symptomatic stage of the disease, delayed the onset of clinical signs, slowed down disease progression, and increased animal lifespan in several mouse models of neurodegeneration, including those that recapitulate key features of ALS. 7,9,28,31,32 However, the efficacy of Dimebon in ameliorating pathology in these models was low and in attempt to produce more efficient compounds, several bioisosteres of Dimebon have been synthesized. Neuroprotective properties have been demonstrated for some of fluorinated bioisosteres in mouse models. 9,12 Here, we studied effects of one of these bioisosteres, DF402, on the disease triggered in transgenic mice by neuron-specific expression of a highly aggregation-prone truncated form of human FUS. 33 2 | MATERIAL S AND ME THODS

| Animals
Transgenic mice expressing C-terminally truncated form of human FUS under control of the neurospecific Thy-1 regulatory sequences (FUS[1-359]-line 6) have been produced and characterized previously. 33 A line S-FUS  produced by backcrossing of the original FUS[1-359] line 6 mice with CD1 mice 34,35 was used in the current study. Cohorts of experimental animals were formed from hemizygous transgenic and wild-type male littermates. Genotyping was carried out using a PCR protocol described previously. 33

| DF402 treatment and monitoring of the disease phenotype
Dimebon bioisostere DF402 (2,8-Dimethyl-5-[2-(6-trifluoromethyl pyridin-3-yl)ethyl]-2,3,4,5-tetrahydro-1H-pyrido [4,3-b]indole dihydrochloride) was synthesized in IPAC RAS as described elsewhere. 37 Similar to DF302, another fluorine-containing and structurally close derivate of Dimebon, that was previously assessed for its ability to ameliorate neurodegenerative processes in various models, 9,12 DF402 is water soluble, stable in aqueous solutions, and non-toxic to cultured cells and experimental animals. Animals received DF402 in the drinking water (70 µg/ml; correspondent to daily dose of 12 mg/ kg/day after adjustment to average daily water intake in these mice) from the age of 42 days. The same administration protocol and a drug dose were used for assessing the efficacy of Dimebon in our previous studies 9,31,38 and in two mouse models of neurodegeneration, a slowing down of pathology progression was achieved. 9,31 To collect data about the disease onset, disease duration, and animal lifespan, mice were assessed daily for the development of first signs of neuronal pathology (paresis of a limb, unstable gait, hunched posture, clasping reflex, decreased motility). Observations continued after the disease onset and animals were humanely killed when their conditions reached the severe level as specified by the Home Office Licence.

| Animal gait analysis
CatWalk XT system Version 10.6 was used according to the manufacturer (Noldus Information Technology, Netherlands) instructions with the Intensity of the Green Walkway Light 16,5 V; Camera Gain 20 dB; Green Intensity Threshold 0,1; and Red Ceiling Light 17,7 V. All animals had one habituation session to the CatWalk apparatus one day before the first test. Each animal was tested every or every second day until the onset of clinical signs. On the testing day, each animal was allowed to explore and walk freely through the apparatus detection "runway" without any rewards. During this test period of maximum 10 minutes, at least 3 videos of an uninterrupted crossing of the recording field of the runway (approximately 35 cm) were automatically recorded.
Runs for analysis were selected based on a minimum of four step cycles in the crossing field, such that each step cycle involved capturing the use of each of the four paws irrespective of the order in which paws were used. After classification of the footprints in these runs using the CatWalk software, qualified data [39][40][41] were exported for further analysis to the RStudio. 42 Given that the disease progression has individual characteristics resulting in a floating date of the manifestation of clinical signs, we have compared changes in the gait of animals in the reverse direction starting from the day when first clinical signs were observed to the day of the beginning of data collection.
In order to combine the data of gait analysis between the animals with different ages, time intervals were adjusted to 25 days for each animal starting from the day of the manifestation of clinical signs, as illustrated in Figure S1. Statistical analysis was done using nonpara- For example, parameter "Right_Hind_MaxIntensity" consists of two components "Right_hind" and "MaxIntensity" representing qualitative and quantitative indicators. The frequency was determined as a ratio of the significantly changed parameters or components included in such parameter (p < 0.05; Benjamini-Hochberg correction) to all CatWalk parameters. For principal component analysis, the values of all parameters were transformed to z-scores ((x-mean(x)/sd(x)). The multivariate analysis of variance 43 was used to find traits with significant changes between experimental groups. Classical (Torgerson) multidimensional scaling (MDS) 44 was performed to estimate clusterization between mice groups. Visualization was performed using custom scripts written in R.

| RNA sequencing
For RNA-seq analysis, total RNA was extracted from the thoracic and lumbar spinal cords of experimental and control presymptomatic mice (70-day-old, 4 animals per group) using Qiagen RNeasy Plus mini kit. RNA quantification, quality controls, and further steps were performed as described previously. 34  Raw sequence data processing (QC, trimming, alignment, read quantification) was performed with PPLine tool. 45 Differential gene expression analysis was performed with the edgeR package. 46 Gene Ontology and KEGG enrichment analyses were performed using topGO (v.2.36.0) and clusterProfiler packages. 47 Multidimensional scaling (MDS) between all experimental samples was performed with limma package. 48 Sorting of microglial and neuronal genes of the whole spinal cord samples was performed as described previously. 34 RNA-seq data of wild-type animals were deposited previously by the number GSE13 0604.
The real-time RT-qPCR analysis of mRNA expression was carried out as described previously. 34,50

| General statistical analysis
Data sets for the disease onset, disease duration, animal lifespan, and RT-qPCR analysis of RNA expression were assessed for normal distribution and for those that passed D'Agostino and Pearson test, statistical significance of observed difference was evaluated by oneway ANOVA or paired t-test, as appropriate. For not normally distributed data, Kruskal-Wallis ANOVA and/or Mann-Whitney U-test were used. was recorded, and monitoring of animal health was continued until their conditions deteriorated to the level when they were deemed to be premorbid and therefore were euthanized by a Schedule 1 method. We found that chronic DF402 treatment increases animal lifespan by 13% ( Figure 1A,D), which was due to both a later onset ( Figure 1B) and longer duration ( Figure 1C) of the disease (by 11% and 24%, respectively).

| CatWalk analysis of the gait of transgenic mice expressing C-terminally truncated form of human FUS detects impairments of motor function and effects of DF402 at pre-symptomatic stage
To assess whether instrumental analysis of multiple parameters of animal gate would allow detecting the decline of animal motor performance long before obvious manifestation of clinical signs and revealing any improvements of this performance as the result of DF402 treatment, we used CatWalk XT system to systematically monitor gait of DF402-treated and control, untreated S-FUS  transgenic mice as well as their wild-type littermates. For this part of the study, additional cohorts of 24 wild-type and 49 S-FUS  transgenic male littermate mice were produced. The same as above protocol of DF402 administration was used to treat 25 of these S-FUS  mice from the age of 42 days. Four weeks later, that is, at the age of 70 days, four animals from each group were humanely euthanized and tissue samples were collected for transcriptomic analysis (see below). It is important to note that at this age, all transgenic animals were undistinguishable from their wild-type littermates and in this cohort of S-FUS[1-359] mice, a mean age of the disease onset was 124 days for untreated and 137 days for DF402-treated animals.
However, the CatWalk analysis of 177 gait parameters carried out as  Figure S3). Interestingly, the compensation applies to many genes upregulated in the presence of pathogenic FUS protein, resulting in the reversion of expression toward lower levels typical for expression of these genes in the spinal cord of wild-type animals (lower cluster in Figure 4C, Figure S4A-D). In contrast, genes that were downregulated in the spinal cord of transgenic mice showed less compensation (upper cluster in Figure 4C, Figure S4E). Genes from the middle cluster in Figure 4C showed no compensation or even a trend toward overcompensation ( Figure S4F,G).

| Multifactorial analysis of CatWalk testing results detects gait changes in FUS transgenic mice at early pre-symptomatic stage
CatWalk is an effective combination of hardware and software in an experimental system designed for simultaneous analysis of multiple parameters of rodent gait. 53,54 It has been successfully used to assess changes in animal motor function in several transgenic models of neurodegenerative diseases. [55][56][57][58] Here, this system was used for

CO N FLI C T O F I NTE R E S T
The authors have declared no competing interest.

AUTH O R S' CO NTR I B UTI O N
NN, SOB, and VLB conceived and supervised the study and analyzed the data. KC, APR, SF, TAI, EAL, and AYA carried out the experiments. KC, APR, SF, AVD, and MSK analyzed the data. VLB wrote the manuscript with contribution from all authors.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in NCBI GEO database at https://www.ncbi.nlm.nih.gov/geo/, reference numbers GSE16 1680 and GSE13 0604.