The Motor Neuron Disease Mouse Model hSOD1G93A Presents a Non-canonical Prole of Senescence Biomarkers in the Spinal Cord

Recent evidence demonstrates a pathological role for senescent cells in Alzheimer’s and Parkinson’s diseases. The present study aimed to show senescence mechanisms including senescence-associated secretory phenotype (SASP) in the familial amyotrophic lateral sclerosis (ALS) transgenic mouse model hSOD1-G93A. We evaluated, as senescence biomarkers, the expression of p16 and p21 with reverse-transcriptase quantitative PCR (RT-qPCR), immunouorescence (IF), and immunohistochemistry (IHC), as well as the senescence-associated β galactosidase (SA-β-gal) activity in the lumbar spinal cords (LSC) of this model. As SASP markers, we quantied the mRNA levels of Il1a, Il6, Ifna, and Ifnb. Furthermore, we explored if an alteration of alternative splicing is associated with senescence phenomena in this model. Thus, we quantied the Adipor2 cryptic exon inclusion levels, a specic splicing variant repressed by TAR-DNA binding of 43 kDa (TDP-43), using RT-qPCR. Our results show an atypical senescence-prole in LSC from transgenic mice, increasing p16 and p21 mRNA and protein levels in glial cells with a mostly cytoplasmic pattern, without the canonical increase in SA-beta-gal activity in these cells. Consistent with enhanced SASP, there is an increase in Il1a and Il6 expression. Also, TDP-43 splicing activity is compromised in this ALS model, in a direct relationship with the increase in p16 expression. However, senolytic drug Navitoclax -with reported benets in Alzheimer and Parkinson disease mouse models does not alter the present model’s disease progression. Navitoclax neither eliminates cells expressing senescence and nor represses the expression of SASP related genes. Globally, our ndings support the existence of a non-canonical prole of senescence biomarkers in the LSC of the ALS model hSOD1-G93A.


Introduction
Aging is a major risk factor for developing amyotrophic lateral sclerosis (ALS) (Niccoli et al. 2017). ALS is a neurodegenerative disease characterized by the loss of motor neurons with an unfavorable outcome (< 5% survival at 5 years after diagnosis). Cellular senescence was rst described by Hay ick in the 1960s as a limitation on division of normal cells in vitro (Hay ick & Moorhead 1961). The cellular mechanisms behind this phenomenon were later described, with an important role for cell cycle inhibitors, highlighting p16-INK4a as the major contributor (Serrano et al. 1997). Another hallmark of senescent cells is the increase in β-galactosidase, commonly known as SA-β-gal which is associated with an increase in lysosomal biogenesis (Kurz et al. 2000). Cellular senescence has been described as a barrier against oncogenesis, with a tradeoff where these cells can develop a pro-in ammatory status known as SASP. This process re ects an attempt to induce tissue repair in which senescent cells, usually accumulating DNA damage, can stimulate its clearance by the immune system. Regarding neurodegenerative diseases, several groups have independently demonstrated the presence of senescent glial cells and SASP in the central nervous system (CNS).
Another process related to aging is the change in alternative splicing (AS), a conserved mechanism that increases the complexity of the proteome. TDP-43 regulates a large number of AS events in a complex way (Tollervey et al. 2011). Several evidence support the role of TDP-43 pathology in age-related neurodegenerative processes and physiological aging (McAleese et al. 2017). Most AS events regulated by TDP-43 involve the repression of a set of non-conserved (cryptic) exons which are abnormally incorporated into mRNA in ALS (Ling et al. 2015). In this line, we previously quanti ed the rate of inclusion of cryptic exons in nervous tissue from ALS donors and cellular models and found a positive correlation with age at death (Torres et al. 2018).

Results And Discussion
To clarify whether senescence-associated phenomena and TDP-43 dysfunction could be implicated in ALS, we measured the abovementioned variables in the familial ALS transgenic mouse model hSOD1-G93A at different disease stages. The senescence markers p16 and p21, typical biomarkers of senescent cells (Coppé et al. 2010), were analyzed in LSC. Using two different technics (IHC and IF), we characterized the cellular expression pattern of p16. The results show that the expression of p16 mRNA was progressively increased during disease evolution (Fig. 1A), whereas p21 mRNA levels were only higher at the end-stage (Fig. 1B). p16 and p21 exhibited a predominantly cytoplasmic pattern (Fig. 1C), in contrast to recent results from an ALS rat model where it was mainly nuclear (Trias et al. 2019). As shown by IF, p16 positive cells were microglia (Iba1 + cells) ( Fig. 2A) and astroglia (GFAP + cells) (Fig. 2B).
These results indicate dynamic changes in cellular senescence-associated markers and SASP related to disease evolution. p16 expression is highly expressed before the symptomatology in our transgenic mice, similarly to p16 + microglia in LSC from transgenic rats (Trias et al. 2019). Both facts suggest a role for p16 in disease initiation and progression. Interestingly, senescence-associated cell cycle arrest in an early symptomatic stage (120d) is driven exclusively by p16, whereas p21 only increases later in this model. This may be seen as a result of the late-onset activation of p53 and the DNA damage response pathway, similar to what occurs in the senescence process in microglia (Stojiljkovic et al. 2019). In contrast to p21 (related to reversible cell cycle arrest or quiescence), the senescence process depends heavily on prolonged p16 expression. Strictly speaking, our work and most published articles on 'senescence' do not demonstrate an always irreversible cell cycle arrest. There may be divergent processes sharing common biomarkers (Sharpless & Sherr 2015). This is the case with macrophage polarization, in which p16 expression and SA-β-gal activity are physiological, reversible, and not associated with cellular senescence (Hall et al. 2017). In this line, cytoplasmic p16 can regulate cell migration in a manner similar to cyclin D1 (Chen et al. 2013). This evidence re ects a convergent pathway of cell cycle-and senescence-associated We also analyzed another senescence canonical biomarker: SA-β-gal activity. The main cellular populations expressing SA-β-gal in ventral LSC are the motor neuron cells (Nissl + cells in the ventral horn, with a motor-neuron compatible cellular size). Neurons of other LSC locations and the vast majority of Nissl-do not show SA-β-gal activity ( Fig. 1D and S1). Interestingly, SA-β-gal activity was reduced during disease progression in motor neurons and in a small fraction of Nissl-cells (compatible with glia). Our ndings agree with previously shown data demonstrating that SA-β-gal activity in neurons is not associated with senescence, although it is increased in aging mouse brain (Piechota et al. 2016). Our results suggest that motor neurons contain more lysosomes in cell body than other cells, and that their biogenesis is compromised in this ALS mouse model. In this line, lysosomal mass de cit has already described in this model, highlighting a role of hSOD1 aggregates disturbing lysosomal biogenesis (Xie et al. 2015) and potentially explaining our results from the SA-β-gal activity assay.
Another marker commonly employed in senescence description is the increase in cytokines linked to SASP. In this case, we quanti ed the expression of typical SASP markers Il1a and Il6. We analyzed as well the expression of Ifna and Ifnb (corresponding to type-I IFN response) as they are postulated as latesenescence markers and could be helpful in determining senescence progression in the LSC of this model. The expression of Ifna was not detected in any of the analyzed samples (data not shown). We observed a different pattern of expression between Il1a (Fig. 3A) and Il6 (Fig. 3B). Il1a is increased in the pre-symptomatic stage and is known to be the upstream regulator of IL-6 in SASP (Orjalo et al. 2009). IL-6 is increased in cerebrospinal uid in ALS, Alzheimer's, and Parkinson's disease (Chen et al. 2018). In contrast, Ifnb expression ( Fig. 3C) is not altered, which could indicate that senescence in this model does not evolve a late phase. Overall, this might re ect a complex interaction between senescence, SASP, and changes in reactive glial cells and neurodegeneration.
Regarding TDP-43 splicing function, in mice it controls the inclusion in Adipor2 mRNA ( Figure S2). In line with loss of TDP-43 function in this model, cryptic exon inclusion in Adipor2 mRNA was higher in lumbar spinal cord in end-stage mice (Fig. 3D) and positively correlated with p16 expression (Fig. 3E). The present data are the rst to show speci c alteration regarding splicing function in this ALS model.
Notably, this process is associated with an increase in the senescence marker p16, and the two processes are likely to be linked in the same pathway. Of note, increased p16 seem restricted to central nervous system, as sciatic nerve does not show these changes (Fig. 3F), in contrast with Adipor2 cryptic exon inclusion, which was also increased in sciatic nerve (Fig. 3G). Noteworthy, increased Adipor2 cryptic exon was associated with loss of Adipor2 mRNA levels, suggesting increased non-sense mediated decay in both locations (Fig. 3H).
We wanted to explore the potential bene ts of senolytic treatment due to the higher expression of senescence related genes in this mouse model. We performed Navitoclax treatment following the protocol described for Alzheimer's disease mouse model (Bussian et al., 2018). The treatment was initiated at 90 days old and nished at end point (Fig. 4A). We estimated the disease progression by weight loss. Navitoclax treatment did not prevent weight loss, neither prolonged survival ( Fig. 4B and 4C).
Finally, we quanti ed senescence and SASP genes in lumbar spinal cord. None of the analyzed genes showed statistically signi cant differences (Fig. 4D). These results suggest differences in molecular effectors between Alzheimer's and ALS.
Navitoclax is an inhibitor of antiapoptotic protein Bcl2 (Zhu et al., 2016). Senescent cells are highly dependent of different antiapoptotic members. Senolysis is achieved when this antiapoptotic protein is inhibited, promoting cell death (Zhu et al., 2015). Navtioclax treatment is not enough to slow the disease progression and does not extend the survival. In contrast with data in Alzheimer's and Parkinson's disease models, this treatment does not prevent the increase of senescence and SASP markers. It suggests that senescence phenotype is not driven by Bcl2 expression of stressed or aged cells in this model (Zhu et al., 2015). Further studies are warranted to determine whether senescence-linked phenomena are mechanistically involved in this fatal disease, clearing the pathway for therapeutic development.
In the case of ALS, we speculate that Navitoclax treatment is not e cient because Bcl2 is not overactivated in our G93A mouse model (Vukosavic, Dubois-Dauphin, Romero, & Przedborski, 1999).
However, Bcl-XL, a Bcl-2 family member, is overactive in astrocytes and provides pro-survival input and may mediate the activation of toxic astroglia (Lee, Kannagi, Ferrante, Kowall, & Ryu, 2009). It suggests that an speci c inhibition Bcl-XL could have greater effects on disease progression.

Conclusions
The LSC from the hSOD1-G93A mouse, a model of familial ALS, exhibits a non-canonical pro le of senescence biomarkers. This pro le is characterized by an early increase in p16 and a late increase in p21, with both displaying a mainly cytoplasmic pattern in glial cells without an increase in SA-β-gal activity. In the case of SASP, it also has a dynamic pro le with increasing levels of Il1a from the presymptomatic stage onward and an acute peak of expression in end-stage transgenic mice. Regarding AS, this tissue shows a dysfunctional splicing activity of TDP-43 in end-stage ALS mice. This is the rst time that senescence markers, SASP, and TDP-43-associated splicing dysfunction have been described in this ALS mouse model.

Experimental procedures Animal Experiments
A colony of the strain B6.Cg-Tg(SOD1*G93A)1Gur/J JAX catalogue stock number 004435; from now on hSOD1G93A or G93A) was purchased at The Jackson Laboratories (Bar Harbor, MN, USA). Mice were maintained in C57BL/6J background. Genotyping was performed following manufacturer's instructions. After genotyping and weaning, animals were placed at 12:12 hours dark / light cycle, at 22 ± 2 °C temperature, 50%±10 relative humidity, in individual cages (at 21 days). For age-related studies, we evaluated. Navitoclax (T2101, Targetmol) was diluted in 60% Phosal 50 PG (Lipoid), 30% PEG400 (Sigma, 91893) 10% EtOH. Navitoclax was administered by oral gavage at a dose of 50 mg kg-1 body during ve consecutive days followed by 16 days of rest (N = 5 per group). Treatment cycles were repeated until clinical endpoint (righting re ex > 20 s). Spinal cords and sciatic nerves were rapidly excised, frozen in liquid N 2 and stored at -80 C. This study was approved by the Animal Research and Ethics Committee at the University of Lleida. RNA extraction, cDNA synthesis and RT-qPCR 1 ml of TRIzol reagent (Thermo Fisher Scienti c, AM9738) was added to 50-100 mg of tissue. The tissue was then mechanically homogenized using T 10 basic ULTRA-TURRAX® (IKA). 200 µl of chloroform was added to each sample and mixed. After 5 minutes of incubation at room temperature, the samples were centrifuged at 12,000 xg (15 min, 4 °C) to separate the phases. The aqueous phase was separated into a new tub and mixed by vortexing with 500 µl of isopropanol. After an incubation of 10 minutes at room temperature, RNA was precipitated through spun at 12,000 xg (10 min, 4 °C). The resulting supernatant was removed, and the pellet was washed with a 75% ethanol. After vortexing, the samples were centrifuged again at 12,000 xg (10 min, 4 °C). The supernatant was discarded, and the RNA pellet was allowed to air dry at room temperature. The RNA was resuspended with 50 µl of RNAse-free water, quanti ed with Nanodrop (Nanodrop technologies, ND-1000 UV/Vis Spectofotometer) and stored at -80 °C unjtil further use. One microgram of RNA was used for retrotranscription employing TaqMan Reverse Transcription Reagent using random hexamers (Thermo Scienti c, N8080234). RT-qPCR experimets were performed using a CFX96 instrument (Bio-Rad, Hercules, California, USA) with SYBR select Master mix (Thermo Fisher Scienti c, 4472908). Each 20 µL of reaction contained 4 µL cDNA, PCR grade water. RT-qPCR run protocol was as follows: 50 °C for 2 minutes and 95 °C for 2 minutes, with the 95 °C for 15 seconds and 60 °C for 1 minute steps repeated for 40 cycles; and a melting curve test from 65 °C to 95 °C at a 0.1 °C/s measuring rate. N = 3-10 mice per group were used for RT-PCR experiments. Primers employed in these experiments are listed in Supplementary Table S1. Actb expression was used as housekeeping to normalize the other genes. Speci c Adipor2 cryptic mRNA (primers annealed with cryptic exon) was normalized with Adipor2 normal transcript (the primers annealed with conserved exons).
Senescence-associated β-galactosidase activity Brie y, paraformaldehyde-xed frozen sections were incubated with X-gal solution (20 mg/ml X-Gal (SIGMA), 5 mM K 3 Fe(CN) 6 , 5 mM K 4 Fe(CN)6 and 2 mM MgCl 2 ) in PBS at pH 6.0 overnight at 37 °C. To allow comparison between specimens, all samples were assayed simultaneously. Then, neurons were stained with Green Fluorescent Nissl Stain (Thermo Fisher Scienti c, N21480) diluted 1:150 in PBS and incubated 20 minutes at room temperature. The slices were then washed 3x with PBS for 10 minutes at room temperature and 1x with PBS for 2 hours at room temperature. Nuclei were stained with DAPI (SIGMA, 32670). Images of the stained sections were taken using inverted microscope (Olympus, IX71S8F-2). Eight randomly selected areas of each mouse (N = 2-3 per group) of the ventral horn of lumbar spinal cord sections were photographed at 20x magni cation for visual analysis. The whole section of SA-β-Gal stained slices was photographed at 4x magni cation.

Immuno uorescence
One control and one transgenic lumbar spinal cord were xed in 4% paraformaldehyde made in PBS overnight at 4 °C and cryopreserved in 30% sucrose in PBS 48 hours. The lumbar spinal cord was then cut at a 16 µm section depth and resulting seeded on a gelatin-coated slide. Samples were permeabilized with 0.3% Triton X-100 PBS for 30 min and blocked with 5% BSA in PBS for 1 h at room temperature. The primary antibody 1:100 anti-p16 (abcam, ab54210) and 1:200 anti-GFAP (abcam, ab7260) were diluted in 0.3% Triton X-100 PBS and incubated overnight at 4 °C. The slices were washed with PBS three times for 5 min at room temperature, followed by the secondary antibody (diluted 1:800 in PBS), goat anti-mouse Alexa Fluor 555 (Thermo Fisher Scienti c, A21422) and goat anti-rabbit Alexa Fluor 488 (Thermo Fisher Scienti c, A11008) incubation for 1 h at room temperature in darkness. Sections were nally counterstained with 1 µg/ml 4,6-diamidino-2-phenylindole dihydrochloride (DAPI) in PBS for 10 min at RT and mounted on slides with Fluoromount-G® (Southern Biotech, 0100). Samples were imaged using a laser scanning confocal microscope Olympus FluoView FV10.

Immunohistochemistry
One control and one transgenic paraformaldehyd xed para n embedded tissue slides were dried for 1 h at 65º before pre-treatment procedure of depara nization, rehydration and epitope retrieval in the Pre-Treatment Module (Agilent Technologies-DAKO, PT-LINK) at 95 °C for 20 min in 50x Tris/EDTA buffer, pH 9. For p21 immunohistochemical staining, p21WAF1/Cip1 antibody (Agilent Technologies-DAKO, clone SX118) 1:100 dilution was used. After incubation, the reaction was visualized with the EnVisionTM FLEX Detection Kit (Agilent Technologies-DAKO) using diaminobenzidine chromogen as a substrate. For p16 immunohistochemical staining, p16 INK4a antibody was used using CINtec® Histology Kit (ROCHE, clone E6H4) following manufacturer's instructions. Sections were counterstained with hematoxylin.

Statistical Analysis
All statistical tests and graphs were performed using GraphPad Prism 6 (GraphPad Software). Normalized mRNA expression was analyzed with ordinary Two-way ANOVA test of the variables time and genotype. For multiple comparisons between genotypes, Bonferroni's multiple comparisons test was used. In SA-beta-gal activity experiments, p-value was determined by Chi-square's test. To evaluate relationship between p16 and Adipor2 cryptic mRNA levels a linear regression was tested.   Figure 1 Increase in senescence markers in spinal cord during ALS progression. p16 expression was progressively higher at 120 days and 150 days (a), whereas p21 were only found to be increased at end stage (b). Both p16 and p21 exhibited cytoplasmic staining (c). Nissl+ cells of the ventral horn of the spinal cord (compatible with motor neurons) are the main contributor to SA-beta-gal activity in lumbar spinal cord and are almost depleted in this activity in hSOD1-G93A mice at 150 days, similarly to Nissl-(glia) cells (d). p16 and p21 expression are expressed as mean ±SEM. Ns indicates p > 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 for Student's t test or ANOVA, when appropriate. Red arrows indicate p16 and p21 positive cells. Green scale bar represents 100 µm, yellow scale bar represents 500 µm, and red scale bar represents 2500 µm. n=4 from each genotype and age.  Senolytic treatment using Navitoclax does not slow the disease progression. A chronic treatment was established with ve consecutive doses followed by two weeks of resting (a). Weight loss was not different between Navitoclax and vehicle experimental groups (b). Survival time was also unaltered between groups (c), as senescence and SASP genes (d). n=5 from each genotype.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.