Akt/Foxo pathway activation switches apoptosis to senescence in short telomere zebrafish

Progressive telomere shortening during lifespan is associated with increased genome instability, block to cell proliferation and aging. Apoptosis and senescence are the two main cellular outcomes upon irreversible cell damage. In this study, we show a transition between apoptosis to senescence in cells of two independent tissues in telomerase zebrafish mutants. In young mutants, proliferative tissues exhibit defects in cell proliferation and p53-dependent apoptosis, but no senescence. Progressively, these tissues display signs of tissue dysfunction, loss of cellularity and increased senescence. These alterations are accompanied by an activation of pro-proliferative stimulus mediated by AKT. Consequently, FoxO1 and FoxO4 transcriptional factors are inactivated, reducing SOD2 levels, causing an increase in ROS. These alterations elicit the activation of the zebrafish p16/15 and senescence. Thus, upon telomere shortening in aging, early apoptosis induces compensatory proliferation. However, progressive decline in cell proliferation results in tissue damage and proliferative signals, promoting a switch to senescence.

* These authors contributed equally in this work  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  INTRODUCTION  RESULTS   119 tert-/zebrafish proliferative tissues undergo a time-dependent switch from apoptosis to 120 cell senescence 121 Apoptosis is a process in which programmed cell death allows for clearance of damaged 122 cells (Hawkins and Devitt, 2013). In contrast, replicative senescence is a state of terminal 123 proliferation arrest, associated to gradual telomeres attrition occurring during cell division 124 (Olovnikov, 1973;Shay and Wright, 2000). To explore the molecular mechanisms underlying 125 the cell-fate decision between apoptosis and senescence, we used telomere attrition as a 126 trigger of these two possible outcomes. 127 First-generation tert-/-zebrafish have shorter telomeres than their wild-type (WT) 128 siblings, develop several degenerative conditions affecting mainly highly proliferative tissues, 129 such as the testis and gut, and die prematurely (Anchelin et al., 2013;Carneiro et al., 2016a; 130 Henriques et al., 2013). At 3 months of age, tert-/-fish are macroscopically similar to their WT 131 siblings (Carneiro et al., 2016a), with testis and gut being histologically indistinguishable from 132 WT ( Figure 1A). However, at this early age, average telomere length is short and trigger the 133 onset of DDR and increased apoptosis in tert mutants (Carneiro et al., 2016a). We analyzed the 134 presence of apoptotic cells in 3 months-old tert-/-gut and testis using the TUNEL assay. We 135 confirmed that, even in the absence of macroscopic defects, tert-/-gut and testis exhibits a 136 higher number of apoptotic TUNEL-positive cells compared to their WT siblings ( Figure 1C). In 137 order to confirm the activation of DDR, we analyzed the phosphorylation levels of the DNA 138 damage marker gH2A.X on whole cell lysates from gut and testis of 3 month-old tert-/-zebrafish 139 ( Figure 1E, quantification in Supp. Figure 1). We detected a significant increase of the ATM-140 dependent phosphorylated form of gH2A.X in ser139 in both 3 month-old tert-/-gut and testis 141 ( Figure 1E, quantification in Supp. Figure 1). As expected, we observed a concomitant increase 142 p53 protein levels in both tissues of tert-/-zebrafish ( Figure 1E, quantification in Supp. Figure 1). 143 In light of the differences found between the INK4a/ARF locus in zebrafish and 144 mammals, we decided to test the conservation of the protein and the validity of the mammalian 145 anti-p16 antibody (sc-1661, Santa Cruz Biotechnology) used for the senescence analysis. To 146 this purpose, we designed antisense morpholino oligonucleotides (p15/16 MOs) and injected 147 increasing amounts in 1 cell-stage embryos (Supp. Figure 2). A control morpholino sequence 148 was included as negative control (CTR MO). At 3dpf upon morpholino injection, larvae were 149 collected and tested for the expression of the p15/16 protein in zebrafish. Western Blot analysis 150 revealed that injection of increasing concentration of p15/16 MOs causes a reduction in the 151 amount of the protein recognized by the anti-p16 antibody, indicating that the sequence of the 152 protein associated to senescence is conserved from mammals to zebrafish (Supp. Figure 2). 153 Strikingly, even though DDR is active in 3-month-old tert-/-, the analyzed tissues did not 154 exhibit signs of cellular senescence. We were unable to detect senescence-associated beta-155 galactosidase (SA-beta-Gal) activity in both gut and testis in tert-/-fish ( Figure 1C). Accordingly, 156 we observed no differences in expression of the senescence marker p15/16 by qRT-PCR, 157 western blot ( Figure 1E) or by immunofluorescence staining ( Figure 1C). Therefore, at 3 month 158 tert-/-in which tissue integrity is retained, telomere dependent DDR signalling predominantly 159 induces apoptosis but no detectable cell senescence. 160 To investigate the consequences of telomere erosion and chronic DDR activation in 161 aging, we analysed testis and gut of older tert-/-animals (9 month of age). Contrary to what we 162 observed in 3 month old fish, older tert-/-zebrafish exhibit tissue morphological defects ( Figure  163 1B), including testis atrophy and width lengthening of the gut lamina propria (as described 164 previously -(Carneiro et al., 2016a)-). 165 Because telomere shortening is known to induce both apoptosis and cell senescence, 166 we wondered if the decline in tissue homeostasis represented a change in cell fate. Surprisingly, 167 at 9 month of age, we could not observe clear differences in p53 levels between WT and tert-/-168 ( Figure 1F). In fact, tert-/-gut and testis exhibited a decline in apoptosis in 9 month-compared 169 to 3 month-old fish, denoting a decrease in TUNEL positive cells ( Figure 1D). In contrast, at this 170 stage, these tissues exhibited a clear accumulation of senescent cells in tert-/-compared to WT, 171 as revealed by SA-beta-Gal staining ( Figure 1D). Increased senescence was confirmed by an 172 increase in p15/p16 by immunofluorescence ( Figure 1D), mRNA and protein levels ( Figure 1E). 173 In addition, we observed that reduction of apoptotic cells and increase of senescent cells was 174 concomitant with higher levels of expression of Bcl-XL mRNA suggesting an activation of anti-175 apoptotic pathways in old tert-/-fish (Supp. Figure 3). Taken together, these results show in vivo 176 a switch from apoptosis to senescence during aging of tert mutant fish, and that this switch 177 associates with age-dependent tissue degeneration. 178

ROS accumulation and mitochondrial dysfunction become apparent upon short 180
telomere-induced senescence 181 In mammalian systems, similarly to what we observe in zebrafish, DNA damage initially 182 halts cell-cycle progression through a p53/p21-mediated cell-cycle arrest (Rodriguez and Meuth,183 2006) ( Figure 1E). But if lesions persist, expression of p16Ink4a predominates as a 184 consequence of mitochondrial dysfunction and ROS production (Freund et al., 2011;Passos et 185 al., 2010). Late generation telomerase knockout mice were observed to induce mitochondrial 186 dysfunction through p53-dependent suppression of the master regulator of mitochondrial 187 biogenesis, PGC1α (Sahin et al., 2011). G4 mTERT deficient mice exhibit significant alterations 188 in mitochondrial morphology, accumulation of ROS and reduced ATP generation (Sahin et al., 189 2011). 190 We investigated if mitochondrial dysfunction could play a part in the apoptosis to 191 senescence switch observed in tert-/-zebrafish. First, we started by examining if p53 activation 192 triggers the repression of PGC1α in zebrafish. Curiously, despite significant accumulation of 193 p15/16 and p53 ( Figure 1F), we did not observe differences in terms of RNA or protein levels of 194 PGC1α in older tert-/-gut extracts (Supp. Figure 4). However, we did detect a robust increase in 195 oxidative damage with age. By 3 months of age, the levels of ROS in tert-/-gut and testis do not 196 differ significantly from their WT siblings ( Figure 2A). Later, we observed a gradual and 197 significant accumulation of ROS in both tissues from 6 months onward in tert-/-compared to WT 198 controls ( Figure 2A). Production of ROS, especially superoxide, is a necessary by-product of 199 mitochondrial respiration (Murphy, 2009). Mitochondrial dysfunction is characterized by 200 concurrent high superoxide production leading to a breakdown of membrane potential that 201 compromises energy production and cellular metabolism (Balaban et al., 2005). In agreement 202 with previous findings, we observed that testis mitochondrial ultrastructure became significantly 203 fragmented in older tert-/-zebrafish (arrows, Figure 2B). Similarly, gut mitochondrial morphology 204 became increasingly rounded and swollen with the appearance of perturbed crystal structure 205 (arrows, Figure 2B). Consistently, we observed a significant reduction of levels of ATP in both 206 tissues of tert mutants ( Figure 2C). Together, these results indicate that mitochondrial function 207 declines dramatically during aging of tert-/-proliferative tissues, supporting the idea that a 208 change in mitochondrial homeostasis may dictate the tissue's cell-fate decision. To gain mechanistic insights into the nature of the oxidative damage observed in the 218 tert-/-zebrafish, we decided to analyse the expression levels of this important antioxidant 219 defence enzyme. Western blot analysis of 9 months-old testis and gut samples, showed a 220 significant decrease in terms of protein levels of SOD2 in tert-/-mutants compared to WT 221 ( Figure 3A). In contrast, SOD2 levels were not affected in tert-/-at 3 months of age (Supp Fig.  222 5). This result suggests that the mechanism that copes with superoxide production is 223 compromised in older tert-/-mutants and, therefore, possibly responsible for the accumulation of 224 oxidative damage in the affected tissues.  Phosphorylation of FoxO by AKT triggers the rapid relocalization of FoxO from the nucleus to 231 the cytoplasm, with the consequent downregulation of FoxO target genes. We, therefore, 232 hypothesised that activation of AKT/FoxO signalling was responsible for the increased oxidative 233 stress in older tert-/-zebrafish. As expected, increased phosphorylation levels of FoxO1 and 234 FoxO4 were correlated with lower expression levels of SOD2 ( Figure 3A, quantification in Supp. 235 Figure 6). Phosphorylation of FoxO proteins suggests that inactivation of these transcription 236 factors may be the cause for the down-regulation of SOD2 in older tert-/-gut and testis. 237 AKT is a highly conserved central regulator of growth-promoting signals in multiple cell 238 types. The kinase activity and substrate selectivity of AKT are principally controlled by 239 phosphorylation sites. Phosphorylation of serine 473 (pAKT-Ser473), is a consequence of 240 activation of mammalian target of rapamycin complex 2 (mTORC2) (Sarbassov et al., 2005). 241 pAKT-Ser473 is required for phosphorylation and inactivation of the FoxOs (Guertin et al., 242 2006). Accordingly, while we did not observe differences in total AKT protein levels, we detected 243 a significant increase in the phosphorylated levels of pAKT-Ser473 denoting full activation of 244 AKT in older but not younger tert-/-zebrafish ( Figure 3A, quantification in Supp. Figure 6). 245 Therefore, AKT activation correlates with increased levels of FoxO inhibitory phosphorylation 246 and concomitant decrease in SOD2 protein levels in tert-/-mutants when compared to WT. 247 Collectively, our results suggest that, upon tissue damage in older tert-/-zebrafish, activation of 248 a pro-proliferative signaling pathway leads to AKT-dependent inactivation of FoxO1 and Activation of a pro-proliferative pathway in an organism that exhibits defects in cell 256 proliferation was somehow surprising. However, one major difference between 3 and 9 months-257 old gut and testis was the increasing tissue damage (Figure 1 A Figure 4D and 4H). Taken together, our results demonstrate that 278 p53 is required for AKT activation and the onset of senescence in older tert-/-fish. Moreover, 279 they suggest that the age-dependent switch from apoptosis to senescence is intimately linked to 280 the loss of tissue homeostasis. In 3 month-old tert-/-zebrafish, telomeres are sufficiently short to 281 trigger DDR and p53-dependent apoptosis. However, no tissue damage is observed in younger 282 animals and this becomes apparent with age-dependent decline in cell proliferation. An older 283 tissue with short telomeres and limited proliferative capacity responds by promoting mitogenic 284 signalling thereby activating the AKT/FoxO pathway and consequent mitochondrial dysfunction. 285 286 287

Inhibition of AKT activity prevents senescence in G1 and G2 tert-/-mutants 288
Our data indicates that activation of AKT in older tert-/-zebrafish correlates with the 289 appearance of the senescence phenotype. To understand the direct role of the AKT/FoxO 290 pathway in modulating the p15/p16-mediated cell-cycle arrest, we decided to test if AKT 291 activation was causal to cell senescence. Our hypothesis would dictate that AKT 292 phosphorylation inhibition would prevent p15/p16 expression and preserve tissue homeostasis. 293 AKT phosphorylation is mediated by the mTORC2 complex, which the main component 294 is the mTOR (mammalian Target Of Rapamycin) protein (Laplante and Sabatini, 2009). To 295 analyze the role of AKT activation in inducing senescence upon telomere shortening, we 296 created a double mutant bearing a mutation in the tert gene combined with a mutation in the 297 mTOR zebrafish homologue (zTOR). Previous work showed that zTOR is essential for 298 development and zTOR -/-zebrafish are larval lethal (Ding et al., 2011). However, zTOR+/-299 mutants are haploinsufficient, with the lack of one functional copy being sufficient to reduce AKT 300 phosphorylation (Ding et al., 2011). Thus, we decided to test our hypothesis in tert-/-ztor+/-301 mutant zebrafish. As expected, tert-/-ztor+/-present reduction in AKT phosphorylation 302 compared to tert-/-single mutants in 11 month-old fish ( Figure 5). Consistent with our 303 hypothesis, this reduction is associated with a reduction in the expression of p15/16 ( Figure 5), 304 suggesting that preventing the activation of AKT can be sufficient to reduce aging-associated 305 senescence ( Figure 5 A-B). Given the incomplete nature of zTOR inhibition, haploinsufficiency 306 for ztor in a tert-/-mutant background was insufficient to restore tissue morphology in testis and 307 tert-/-mutant gut defects (Supp. Figure 7). Our data corroborates previous reports that 308 disruption of zTOR partially inhibits AKT activation and, consequently, reduces p15/16 309 expression and with an amelioration tissue morphology of older tert-/-mutant. 310 Given the previous incomplete AKT inhibition in older tert-/-zebrafish, we decided to 311 attempt a chemical inhibition in tert-/-fish with very short telomeres. Second generation 312 telomerase-deficient zebrafish (G2 tert-/-), obtained from incross of homozygous tert mutants, Consistent with the phenotypical recapitulation of older G1 tert-/-mutants, we observed that G2 316 tert-/-exhibited a marked increase of senescence revealed by SA-beta-Gal staining and 317 expression of senescence associated markers, p15/p16 ( Figure 5A-B). Similar to older G1 tert-/-318 zebrafish, analysis of G2 tert-/-larvae showed that increase of senescence by p15/16 319 expression is concomitant with increased pAKT phosphorylation, decreased SOD2 and, 320 consequently increase of ROS species ( Figure 5C-D). Our data thus indicates that G2 tert-/-321 larvae recapitulates aging-associated AKT activation and senescence observed in old 322 telomerase-deficient fish. 323 We decided to use the G2 tert-/-model to assess a direct link between AKT activation 324 and increase in cell senescence, by testing whether direct AKT kinase inhibition would be 325 sufficient to prevent p16 expression. For this purpose, we daily treated tert-/-and WT larvae 326 with an AKT inhibitor (AKT 1/2 kinase inhibitor, Santa Cruz) for 2 days ( Figure 5E). At 6 dpf, 327 larvae were collected and analyzed for AKT activation and expression of senescence markers. In the present study, we describe that young (3 month-old) telomerase deficient 348 zebrafish already exhibit active DDRs and p53 activation. At this stage, apoptosis is the 349 predominant cell fate. Even though DNA damage is present in proliferative tissues, such as gut 350 and testis, no signs of cell senescence could be detected. However, we observed a switch 351 between apoptosis and senescence in older tert-/-fish. In these animals, senescence becomes 352 the most prevalent cellular response, exhibiting SA-beta-Gal and p15/p16 positive cells and 353 elevated p15/p16 and p21 levels. This observation underscores the fact that the same tissue 354 can undergo different cellular fates, apoptosis or senescence, depending on the animal's age. 355 The p53 transcription factor is described as a "master regulator" of several cellular 356 processes, including cell cycle arrest, apoptosis, senescence and autophagy (Farnebo et al., 357 2010). p53 was first shown to trigger apoptosis in response to cellular stress (Vogelstein et al., 358 2000). However, it is now acknowledged that p53 modulates genes involved in senescence 359 depending on the stress inflicted or cell type (Murray-Zmijewski et al., 2008). In late generation 360 telomerase knockout mice, p53 was shown to be responsible for down-regulating of PGC1α and 361 PGC1β and mitochondrial dysfunction upon telomere shortening (Sahin et al., 2011). In our 362 study, early p53 activation in tert-/-zebrafish does not visibly alter mitochondria function. We propose that, upon telomere shortening and p53 activation, loss of tissue integrity 397 triggers the AKT-dependent pro-proliferative pathway (Figure 3). The combination of these 398 antagonistic forces in the cell would result in cellular senescence. We tested this hypothesis on 399 both pathways. By genetically removing tp53, we were able to rescue tissue degeneration and 400 avoid activation of AKT, accumulation of ROS and induction of senescence. On a second level, 401 we inhibited TOR/Akt genetically by dampening the ztor pathway and, chemically, by directly 402 inhibiting Akt in G2 tert-/-larvae. In both cases, we were able to reduce the effects of telomere 403 shortening. Collectively, our results show that the crosstalk between two pathways telomere 404 shortening/DDR and AKT/FoxO signalling regulate a apoptosis-to-senescence switch and 405 contributes to tissue homeostasis in vivo.  for 30 minutes at room temperature (RT), washed three times in dH20 for 5 minutes each and 495 blocked for 1 hour at RT in 1% BSA, 0,5% Tween 20 in PBST (Triton 0.5%). Subsequently the 496 slides were incubated over-night with anti-p16 (F-12) (1:50, Santa Cruz Biotechnology, sc-497 1661), followed by 3x10 minute PBS washes. Incubation with the secondary antibody Alexa 498

MATERIALS AND METHODS
Fluor 568 goat anti-mouse (Invitrogen, UK, 1:500 dilution) overnight at 4°C was followed by 499 three 10 minute PBS washes. The day after the slides were washed 2×5 minutes in PBS and 500 then incubated with TUNEL labelling mix (protocol indicated by the supplier). Washing and 501 mounting were performed by DAPI staining (Sigma, MO, USA) and mounting with DAKO 502 Fluorescence Mounting Medium (Sigma, MO, USA). 503 Images were acquired on a commercial Nikon High Content Screening microscope, based on 504 Nikon Ti equipped with a Andor Zyla 4.2 sCMOS camera, using the a 20x 1.45 NA objective, 505 DAPI + GFP fluorescence filter sets and controlled with the Nikon Elements software. 506 For quantitative and comparative imaging, equivalent image acquisition parameters were used. 507 The percentage of positive nuclei was determined by counting a total of 500-1000 cells per 508 slide, 63x amplification (N = 3-4 zebrafish per time point/genotype). 509 Senescence-associated β-galactosidase assay 510 β-galactosidase assay was performed as previously described (Kishi et al., 2008). Briefly, 511 sacrificed zebrafish adults were fixed for 72h in 4% paraformaldehyde in PBS at 4°C and then 512 washed three times for 1 h in PBS-pH 7.4 and for a further 1 h in PBS-pH 6.0 at 4°C. β-513 galactosidase staining was performed for 24 h at 37°C in 5 mM potassium ferrocyanide, 5 mM 514 potassium ferricyanide, 2mM MgCl2 and 1 mg/ml X-gal, in PBS adjusted to pH 6.0. After 515 staining, fish were washed three times for 5 minutes in PBS pH 7 and processed for de-516 calcification and paraffin embedding as before. Sections were stained with nuclear fast red for 517 nuclear detection and images were acquired in a bright field scan (Leyca, APERIO). 518 519 Statistical and image analysis 520 Image edition was performed using Adobe Photoshop CS5.1 Statistical analysis was performed 521 in GraphPad Prism5, using two-way ANOVA test with Bonferroni post-correction for all 522 experiments comparing WT and tert-/-over time. For real-time quantitative PCR, statistical 523 analysis was performed in GraphPad Prism5, two-way ANOVA with Bonferroni post-correction. 524 A critical value for significance of p<0.05 was used throughout the study. For Western Blot the 525 bands intensities were calculated using FIJI. Statistical analysis was performed using GraphPad 526 Prism6, the significance was assigned according to the Mann-Whitney t-test. A critical value for 527 significance of p<0.05 was used throughout the study. 528 529 Immunoblot analysis 530 Age-and sex-matched adult zebrafish fish were sacrificed in 200 mg/L of MS-222 (Sigma, MO, 531 USA) and portions of each tissue (gonads and gut) were retrieved and immediately snap-frozen 532 in dry ice. 4dpf larvae were sacrificed in ice and collected in 1,5mL Eppendorf tube, minimum 10 533 larvae /tube. Gonads tissues and larvae were then homogenized in RIPA buffer (sodium 534 chloride 150mM; Triton-X-100 1%; sodium deoxycholate 0,5%; SDS 0,1%; Tris 50mM, pH=8.0), 535 including complete protease and phosphatase inhibitor cocktails (Roche diagnostics), with the 536 help of a motor pestle. Protein extracts were incubated on ice for 30 minutes and centrifuged at 537 4ºC, 13.000 rpm, for 10 min. The supernatant was collected and added to 100 mL of protein 538 sample buffer containing DTT. 539 Gut samples were homogenized in TRIzol (Invitrogen, UK) by mashing each individual tissue 540 with a pestle in a 1.5 ml Eppendorf tube. After incubation at RT for 10 minutes in TRIzol, 541 chlorophorm extractions were performed. The organic phase was collected and proteins were 542 precipitated according to the manufacture protocol. The protein pellet was resuspended in 100ul 543 of Lysis Buffer (150mM NaCl, 4%SDS, 50mM TrisHCl pH 8.0, 10mM EDTA, complete protease 544 and phosphatase inhibitor cocktails-Roche diagnostics). 545 For each sample, a fraction of Proteins was separated on 12% SDS-PAGE gels and transferred 546 to Immobilon PVDF membranes (Millipore). The membranes were blocked in 5% milk or 5% 547 BSA (depending on the primary antibody), then incubated with the indicated primary antibody 548 prior to incubation with the appropriate HRP-conjugated secondary antibody. Antibody Assay, Promega). The luminescence was measured on a Victor 3 plate reader (Perkin Elmer). 565 The relative ATP levels were calculated by dividing the luminescence by the total protein 566 concentration, which was determined by the Bradford method. For Bradford assays, samples 567 were diluted (1/50) with extraction buffer. 568 569 Electron microscopy 570 For electron microscopy analysis, zebrafish tissues were processed according to Schieber et al,571 2010. Briefly, zebrafish were fixed in 2% Paraformaldehyde, 2.5% Glutaraldehyde in 0.1M 572 PHEM buffer for 72h at 4°C. Dissected tissues were then washed 3 times in 0.1M PHEM. 573 Tissues were transferred in 1% Osmium Tetroxide in 0.1M PHEM for 1h fixation on ice. 574 Samples were then dehydrated before being processed for embedding using Epon (Schieber et 575 al., 2010). 70 nm ultrathin sections were cut using Reichert Ultramicrotome. After being 576 counterstained with uranyl acetate and lead, samples were analyzed using a transmission 577 electron microscope (Hitachi H-7650). 578 579 AKT inhibitor larval treatment 580 AKT ½ kinase inhibitor (AKT inh) was purchased from Santa-Cruz (sc-300173). Stock solutions 581 were prepared in DMSO. AKT inh was applied, after a titration, at 2uM concentration between 582 days 3 and 5 post fertilization. Larvae were grown at 28°C and over the incubation periods, 583 replacement of medium with the above mentioned compounds was performed every day, 584 between 3 and 7 PM. Since the compound was dissolved in DMSO, controls were treated with 585 the correspondent dilution of the solvent. The drug was tested in 2 independent trials. Finally, 586 5dpf larvae were sacrificed and collected to perform protein and RNA analysis. and tert-/-siblings (N=3 each). Dashed outlines locate zones of maturing spermocytes (testis) or 604 villi (gut). At 3month, both tissues show an increased number of apoptotic cells in tert-/-605 compared to WT. At that age, no signs of senescence are visible in these tissues. However, 606 senescent cells appear in gut and testis of 9 month-old tert-/-fish depicting a switch between 607 apoptosis and senescence at that age. E-F) Western blot and RT-qPCR analysis for DNA 608 Damage and senescence associated genes in gut and testis of 3 month (E) or 9 month-old (F) 609 WT and tert-/-siblings (N>=6 fish). RT-qPCR graphs are representing mean ±SEM mRNA fold 610 increase after normalisation by RPL13a gene expression levels (* p-value <0.05; ** p-611 value<0.01). At 3 month, gut and testis showed higher levels gH2AX and p53 proteins in tert-/-612 compared to WT but no differences in P15/p16 expression. At 9 month, senescence-associated 613 p15/16 expression increased in tert mutant compared to WT siblings. Compared to WT where difference are seen from 3 to 9 month, gut and testis of tert-/-exhibit a 618 time dependant increase in ROS level and decrease of ATP levels (N>=3 fish per time point per 619 genotype). Representative EM images of these tissues at 9 month revealed fragmented 620 mitochondrial ultrastructure in tert-/-testis and rounded and swollen mitochondria containing 621 perturbed crystal structures (N>=3 fish). Data are represented as mean±SEM. 622 623 Figure 3. Activation of Akt in old tert-/-leads to ROS accumulation by blocking the 624 FoxO1/4-SOD2 Axis and promoting mitochondrial dysfunction. 625 Representative immunoblot of p-Akt, total Akt, pFOXO1, pFOXO4 and SOD2 from testis and gut 626 of 9 month-old tert mutant and WT siblings (N>=9). At 9 month, these proliferative tissues show 627 an increased activation of Akt leading to the inhibition of FOXO-dependant SOD2 expression. 628 629 Figure 4. Genetic inhibition of p53 prevents short telomeres-induced tissue degeneration, 630 Akt activation, ROS accumulation and induction of senescence.

676
RT-qPCR analysis of Bcl-XL in gut and testis of 3 or 9 month-old tert-/-or WT siblings (N=6 677 fish). The graphs are representing mean ±SEM mRNA fold increase after normalisation by 678 RPL13a gene expression levels (* p-value <0.05; ** p-value<0.01) . While no differences are 679 seen at 3 months, Bcl-XL is overexpressed in 9 month-old tert-/-gut and testis compared to WT.