A salicylic acid derivative extends the lifespan of Caenorhabditis elegans by activating autophagy and the mitochondrial unfolded protein response

Abstract Plant extracts containing salicylates are probably the most ancient remedies to reduce fever and ease aches of all kind. Recently, it has been shown that salicylates activate adenosine monophosphate‐activated kinase (AMPK), which is now considered as a promising target to slow down aging and prevent age‐related diseases in humans. Beneficial effects of AMPK activation on lifespan have been discovered in the model organism Caenorhabditis elegans (C. elegans). Indeed, salicylic acid and acetylsalicylic acid extend lifespan in worms by activating AMPK and the forkhead transcription factor DAF‐16/FOXO. Here, we investigated whether another salicylic acid derivative 5‐octanoyl salicylic acid (C8‐SA), developed as a controlled skin exfoliating ingredient, had similar properties using C. elegans as a model. We show that C8‐SA increases lifespan of C. elegans and that a variety of pathways and genes are required for C8‐SA‐mediated lifespan extension. C8‐SA activates AMPK and inhibits TOR both in nematodes and in primary human keratinocytes. We also show that C8‐SA can induce both autophagy and the mitochondrial unfolded protein response (UPRmit) in nematodes. This induction of both processes is fully required for lifespan extension in the worm. In addition, we found that the activation of autophagy by C8‐SA fails to occur in worms with compromised UPRmit, suggesting a mechanistic link between these two processes. Mutants that are defective in the mitochondrial unfolded protein response exhibit constitutive high autophagy levels. Taken together, these data therefore suggest that C8‐SA positively impacts longevity in worms through induction of autophagy and the UPRmit.

This may in part explain why they held promise as treatments to improve insulin resistance and type 2 diabetes. AMPK has also been suggested to control the aging process in general (Salminen & Kaarniranta, 2012), and targeting AMPK has been discussed as a potential strategy to slow down aging in humans (Longo et al., 2015). Interestingly, ASA has recently been revealed as a lifespan-extending treatment in both mice and nematodes (Ayyadevara et al., 2013;Strong et al., 2008;Wan, Zheng, Wu, & Luo, 2013). Salicylic acid also extends lifespan of C. elegans, albeit with a less pronounced effect than ASA (Ayyadevara et al., 2013). Work on the molecular mechanism in C. elegans has shown that activation of AAK-2/AMPK and DAF-16/FOXO was required for the lifespan-extending activity of ASA (Wan et al., 2013). These results led us to investigate in the present work another salicylic acid derivate, 5-octanoyl salicylic acid (referred to as C8-SA), which was developed for its controlled skin exfoliating activity, as described in Saint-Leger, Lévêque, & Verschoore, 2007. C8-SA is a salicylic acid derivative containing an octanoyl group in meta-position to the acid group (Supporting information Figure S1). Unlike for ASA or salicylic acid, no anti-inflammatory activity has been detected for C8-SA. However, we were able to show that C8-SA displays a similar activity to ASA with regard to lifespan in the roundworm Caenorhabditis elegans.

| C8-SA extends Caenorhabditis elegans' lifespan and health span by acting in somatic cells
We first tested the impact of C8-SA on lifespan using C. elegans as a model. This compound shows similar effects in C. elegans to other salicylic acid derivatives published earlier such as ASA and salicylic acid itself (Ayyadevara et al., 2013;Wan et al., 2013). Worms exposed to C8-SA lived on average 19% longer than untreated controls ( Figure 1a and Table 1) and remained healthier for longer periods of time. When we measured the number of body bends per second, we detected that treated animals moved more intensely than untreated animals especially at later stage in their lifespan (Figure 1b; Day 11 and Day 18 of adulthood). This tendency is statistically significant, and the positive effect of C8-SA was dose-dependent since 400 μM had a more pronounced effect than 100 μM (Figure 1b).
Treating animals with 400 μM of C8-SA did not lead to further enhancement of their lifespan (data not shown) In addition, treated animals showed reduced levels of carbonylated proteins, a hallmark of aging in C. elegans (Figure 1b). We observed that C8-SA extended lifespan through its action on somatic cells. Indeed, we compared the effect of C8-SA on the lifespan of wild-type worms and of mutants carrying the glp-1(e2141ts) allele that are defective in germline stem cell proliferation and grow to be fully germline less at 25°C (Arantes-Oliveira, Apfeld, . We found that C8-SA extends the lifespan of the glp-1(e2141ts) mutants to an extent similar to what we observed in wild type (Supporting information Figure S2 and Table 1). The finding that C8-SA increases the lifespan of germline-less worms is important because ablation of the germline has been shown to increase lifespan (Hsin & Kenyon, 1999).

| C8-SA extends Caenorhabditis elegans' lifespan through the insulin signaling pathway
We next asked whether any commonly known "aging pathways" were required for C8-SA-mediated lifespan extension. We first investigated the role of the insulin signaling pathway, which is known to regulate lifespan in multiple species (Kenyon, 2010). Similar to previous results using other salicylic acid derivatives (Ayyadevara et al., 2013;Wan et al., 2013), worms defective for the insulin signaling pathway, daf-16(mu86) mutants, were unresponsive to C8-SA. Indeed, daf-16(mu86) mutants harbored similar mean and maximal lifespan when exposed to C8-SA or the vehicle ( Figure 2a and Table 1). We also found that C8-SA treatment increased the endogenous mRNA levels of sod-3, a DAF-16/FOXO-specific target (Figure 2b), even though SOD-3 protein levels, when measured with the SOD-3::GFP transgene, seemed to remain unchanged (Figure 2c,d). We also failed to detect the translocation of DAF-16::GFP from the cytosol to nuclei of intestinal cells (Figure 2e). Although this may seem at odd with DAF-16's requirement for C8-SA-mediated lifespan extension and its action on sod-3 mRNA levels, it was already shown that DAF-16 transcriptional activity could be boosted without altering DAF-16::GFP localization (Xiao et al., 2013). Our data suggest that the presence of C8-SA may activate DAF-16-dependent transcription without significantly altering the intracellular localization of the transgene.

| C8-SA extends Caenorhabditis elegans'
lifespan through the mitochondrial signaling pathway Next, we asked whether C8-SA extended lifespan through the mitochondrial pathway. It has been shown previously that altering the expression of some subunits of the mitochondrial complexes that mediate the oxidative phosphorylation within the inner membrane of the mitochondria could significantly extend the lifespan of C. elegans Lee et al., 2003). Interestingly, these interventions were then found to trigger the mitochondrial unfolded protein response (UPR mit ) that contributes to the observed lifespan extension (Durieux, Wolff, & Dillin, 2011). Similar observations were also described in mice (Houtkooper et al., 2013;Jovaisaite, Mouchiroud, & Auwerx, 2014).
We tested the impact of C8-SA on worms mutated in isp-1 shown before to be long-lived (Feng, Bussière, & Hekimi, 2001). The isp-1 gene encodes the Rieske iron-sulfur protein of the mitochondrial respiratory chain complex III, and its mutation leads to reduced mitochondrial respiration (Feng et al., 2001). We found that, similar to daf-16(m86)-null mutants, C8-SA failed to extend the lifespan of isp-1 mutants ( Figure 3 and Table 1), suggesting that C8-SA promotes lifespan at least in part through the iron-sulfur protein isp-1.

| C8-SA extends Caenorhabditis elegans' lifespan through the dietary restriction pathway
We also tested the impact of C8-SA on the dietary restriction response. The restriction of food intake is one of the most powerful and robust ways to extend lifespan in a wide variety of organisms.
Although this observation was first made by McCay in 1935(McCay, Crowell, & Maynard, 1935, molecular mechanisms of nutritional-mediated lifespan extension only started to be unraveled in 2007 (Bishop & Guarente, 2007;Panowski, Wolff, Aguilaniu, Durieux, & Dillin, 2007). We analyzed the lifespan of worms that were fed ad libitum or subjected to two different regimens of dietary restriction. In the first instance, we diluted the amount of bacteria (provided as a nutritional source for the nematodes) to which the nematodes were exposed. This intervention was already shown to extend lifespan (Greer et al., 2007). We also subjected worms to a complete bacterial deprivation as of their first day of adulthood. This experiment must be performed in the presence of fluorodeoxyuridine (FUDR) to inhibit matricidal hatching and results in an even more enhanced lifespan extension (Lee et al., 2006;Sutphin & Kaeberlein, 2008). In the pres-  Table 1). In summary, C8-SA seems to act on all longevity pathways known so far with the exception of the germline pathway.

| C8-SA activates various processes usually linked to dietary restriction
We next investigated the role of genes specifically known to be involved in the dietary restriction response. We first tested the impact of the forkhead transcription factor PHA-4/FOXA known to be required for the dietary restriction response (Panowski et al., 2007). We found that pha-4 mRNA levels were induced upon dietary restriction in controls, but not in C8-SA-treated worms (Supporting information Figure S3a). In line with this, we also found that pha-4 targets were induced upon dietary restriction, but not in the presence of C8-SA (Supporting information Figure S3b,c). Thus, C8-SA treatment impairs the activation of the PHA-4 pathway. This may explain, at least in part, why the lifespan extension observed upon DR is hampered when C8-SA is administered to the worms.
We next confirmed that the AMPK pathway was also required for C8-SA-mediated lifespan extension, as it was shown for ASA (Wan et al., 2013). When C8-SA was administered to mutants that  Table 1). The mean lifespan observed in these experiments was similar to that of wild-type worms and therefore suggested that C8-SA activates AMPK. To confirm that C8-SA activates AMPK, we tested the impact of C8-SA on the phosphorylation of its target ACC in primary normal human epidermal keratinocytes (pNHEK). We found that treatment with C8-SA increased the level of pACC after 2 hr (Supporting information Figure S4c), suggesting that the C8-SA effect on AMPK activity can be observed across species.
Since C8-SA seems to act through most aging pathways and all of these pathways except the mitochondrial pathway implicate autophagy, we next investigated the action of C8-SA on TOR signaling (that controls autophagy) and on autophagy itself. When we exposed the worms to C8-SA, we could detect a significant reduction in the eukaryotic translation initiation factor 4H gene drr-2 that correlates with reduced TOR signaling ( Figure 5a). This indicates that C8-SA likely inhibits TOR signaling (Ching, Paal, Mehta, Zhong, & Hsu, 2010). Again, we tested the impact of C8-SA in pNHEK on the phosphorylation state of the (ribosomal S6 protein), a key mTOR target. Accordingly, we also found that C8-SA diminished the phosphorylation of the ribosomal S6 protein by mTOR ( Figure 5b). Next, we found that C8-SA induced autophagy in pNHEK by measuring LC3 I/II levels as well as in nematodes as measured by lgg-1/LC3::GFP (Figure 5c-e). To test whether Addition of 100 µM of C8-SA extends the lifespan of wild-type nematode Caenorhabditis elegans (N2). Two independent lifespan curves of each treatment are presented in the figure. (b) C8-SA also improves the worms' capacity to move at Day 11 and Day 18 of adulthood. In addition, this effect is more pronounced with 400 µM (p < 0.001) than with 100 µM (p < 0.001). (c) Concomitantly, the addition of the same amount of C8-SA reduces significantly (p < 0.05) the overall level of carbonylated proteins elegans, we then performed a lifespan experiment on worms treated with RNAi against the autophagy mediator beclin (bec-1). We found that, under these conditions, C8-SA no longer extended lifespan (Figure 5f). To further confirm these results, we next treated mutants that were either lacking or overexpressing the transcription factor hlh-30.
These worms are known to be impaired for autophagy or to display constitutively increased autophagy levels, respectively (Lapierre et al., 2013). In the two cases, C8-SA lost its capacity to increase lifespan ( Figure 5g,h). These results suggest that the positive impact of C8-SA on lifespan and on health span may be transferable in human cells.
Indeed, autophagy is induced both in worms and in human cells and is required for lifespan extension in worms. It will be interesting to gain further insights into the molecular mechanisms through which autophagy is activated by C8-SA in future studies.

| The UPR mit is required for C8-SA-mediated lifespan extension
As mentioned before, it has been shown that mutations in mitochondrial respiratory chain can trigger the mitochondrial unfolded protein  (Durieux et al., 2011). Because C8-SA also acts through the mitochondrial pathway (Figure 3), similar to ASA (Wan et al., 2013), we next set out to test the action of C8-SA on the UPR mit .
We could detect a small increase in the UPR mit readout, HSP-6::GFP, even though not statistically significant (Figure 6a). To better test the role of the UPR mit on C8-SA-mediated lifespan extension, the impact of C8-SA on ubl-5 mutants that are defective in the UPR mit (Durieux et al., 2011) was evaluated. We found that C8-SA was unable to extend the lifespan of ubl-5 mutant worms (Figure 6b and

| Autophagy and UPR mit are linked
Finally, we wondered whether the respective activation of Autophagy and the UPR mit was somewhat linked. Although some stresses such as dietary restriction are known to trigger the UPR mit and autophagy in parallel in worms (Bennett et al., 2017) and the UPR mit We found that the ubl-5 mutation harbored significantly higher autophagy levels than control worms (Figure 7). In addition, C8-SA did not further increase autophagy levels in these UPR mit -defective mutants (Figure 7). C8-SA therefore activates both UPR mit and autophagy in wild-type animals, and both of these processes are required for C8-SA-mediated lifespan extension.
Taken together, our data suggest that autophagy is either inhibited by the UPR mit or that, in the absence of a functional UPR mit , autophagy is constitutively activated in the worm. In addition, C8-SA fails to further activate autophagy in the absence of UPR mit . Thus, C8-SA requires the presence of both autophagy and UPR mit in order to affect lifespan in a positive way; however, these requirements do not seem to be independent of one another. It is particularly intriguing to observe that, in C8-SA-treated ubl-5 mutants, autophagy levels are not different from that of C8-SA-treated wild-type animals (Figure 7). Our data therefore suggest that activating autophagy is insufficient for C8-SA to increase lifespan. Instead, C8-SA requires a functional UPR mit to be effective even if this implies lower autophagy basal levels.

| DISCUSSION
In this work, we investigated the impact of 5-octanoyl salicylic acid on the health and the lifespan of the nematode Caenorhabditis elegans. It is important to note that there have been earlier reports showing that both acetylsalicylic and salicylic acids could extend C. elegans' lifespan (Ayyadevara et al., 2013;Wan et al., 2013). In these reports, several genes were shown to be important for the lifespan effect of these compounds, including the AMPK, the forkhead transcription factor DAF-16/FOXO and other genes involved in the oxidative stress response. Here, using yet another salicylic acid derivative (C8-SA), we confirm that these molecules do extend lifespan in a remarkable way. Indeed, we show that C8-SA extends lifespan solely acting on somatic tissues since its effect is still observed in worms lacking a germline. We also found that most aging pathways were required for C8-SA-mediated lifespan extension. Mutants defective for daf-16 were incapable to respond to C8-SA, although C8-SA failed to induce DAF-16 translocation from the cytosol to the nuclei. We also found that isp-1 and aak-2/AMPK mutants were unresponsive to C8-SA. Taken together, these observations are puzzling as they indicate that C8-SA acts through many of the known longevity genes.
We also detected that C8-SA exposure led to the activation of AMPK and to the inhibition of TOR signaling in worms and in primary human keratinocytes. This led us to investigate the role of C8-SA on the induction of autophagy and to confirm that C8-SA significantly activates autophagy both in worms and in primary human keratinocytes. Also, when treated with bec-1 RNAi that disrupts autophagy, C8-SA becomes less effective, suggesting that its effect on lifespan requires this process. Although it is possible that more specific processes such as mitophagy are activated by C8-SA, our experiments do not allow to discriminate between autophagy and mitophagy. It has been described before that autophagy plays an important role in the keratinization process during epidermal differentiation (for review see Li, Chen, & Gu, 2016). It is therefore possible that C8-SA might have antiaging benefits on the skin beyond its controlled exfoliating activity.
The requirement of the isp-1 for C8-SA-mediated lifespan extension also led us to investigate the role of the UPR mit . Although, the UPR mit readout HSP-6::GFP only exhibited a small induction upon C8-SA treatment, we found that C8-SA was unable to enhance the lifespan of ubl-5 mutants that are UPR mit -defective. Thus, both autophagy and UPR mit are required processes for C8-SA-mediated lifespan extension.
We finally asked whether these processes were linked or independent. To this end, we constructed transgenic worms defective in the mitochondrial unfolded protein response (ubl-5 mutants) carrying an autophagy reporter (lgg-1::GFP). Thanks to these, we found that the ablation of the UPR mit led to a constitutive increase in autophagy. One explanation for this observation is that autophagy may take over when UPR mit is abrogated to compensate the loss of its machinery. Interestingly, when these nematodes were treated with (a) (b) F I G U R E 4 Treatment with C8-SA strongly impairs the dietary restriction response. In this experiment, we used either bacterial deprivation (abbreviated BD; green curves) which is considered like a severe regimen of dietary restriction or bacterial dilution (abbreviated DR; red curves), considered as a mild dietary restriction. (a, b) It is striking that, in the presence of C8-SA (b), the dietary restriction response is either partially or completely suppressed, depending on the severity of the regimen. This suggests that C8-SA-treated worms may be similar to wildtype worms in dietary restriction C8-SA, autophagy levels remained unaffected. These data suggest that the induction of autophagy by C8-SA requires the presence of an integer UPR mit and therefore advocates in favor of an intrinsic link between the autophagy and the UPR mit processes. To our knowledge, the data presented here provide the first evidence for the existence of such a link.  P-values were calculated using a log-rank test.

| Caenorhabditis elegans movement assay
Wild-type N2 C. elegans worms were grown on Nematode Growth Medium (NGM Agar) prepared in 24-well plates (1.5 ml per well).
Treatment compounds were prepared as solutions in DMSO, of which 10 µl was added to each well to achieve the final desired treatment concentration in the media. Once the treatments had dried, each well was seeded with OP50 E. coli culture, and plates were incubated at 20°C for 24 hr before approximately 15 C. elegans eggs were added to each well. Plates were maintained in these con- Worms were allowed 30 s to adapt, before 30-s videos were captured at 30 frames per second with StreamPix 6 software (Norpix), using an Infinity 2 CCD camera (Luminera) on top of a Motic SMZ-171 microscope. Images were then analyzed using ImageJ according to the C. elegans motility analysis protocol (Pedersen, 2017). The data were filtered to remove any worms for which less than 15 s of movement data was available before further statistical analysis was completed.

(a) (b)
F I G U R E 6 (a) C8-SA activates the UPR mit (cco-1 RNAi is used as a positive control) (b) Furthermore, C8-SA fails to extend the lifespan of ubl-5 mutant worms that are defective for UPR mit . These results suggest that C8-SA alters lifespan also by activating the UPR mit F I G U R E 7 C8-SA activates autophagy in wild-type worms, but fails to increase autophagy in worms that are defective for the UPR mit (ubl-5 mutants). Interestingly, the sole genetic ablation of ubl-5 sufficed to induce autophagy monitored by increased LGG-1::GFP levels. This reveals an unforeseen interaction between autophagy and the UPR mit Three replicate experiments were completed, and the data were analyzed in R (version 3.4.1). A fixed-effects model was fitted to the data to estimate the effect of treatments on the number of BBPS using the lme4 package, with replicate experiment included as a random effect, and Tukey posttests completed to allow specific comparisons between treatments using the emmeans package.

| Dietary restriction
DR was performed through bacterial deprivation or bacterial dilution from the first day of adulthood.

| Bleaching method
Worms were suspended in 5 ml M9 buffer and washed once. After adding 200 µl 5 N NaOH and 500 µl bleaching solution, the worms were vortexed for 6 min and centrifuged for 5 min at maximum speed (400g) to explode adult worms. M9 buffer was then added up to 10 ml, centrifuged at maximum speed for 2 min at +4°C, and the pellet was washed four times with 10 ml M9 buffer. Eggs were then transferred to a 250-ml flask with 50 ml M9 buffer and incubated at 20°C overnight.

| qRT-PCR
For each gene, analyses were performed on triplicate biological samples and for each sample, two technical replicates.

| RNA extraction and purification
Total RNA was isolated from synchronized populations of Day 1 adult worms (about 3,000 individuals per condition) using the following method. Worms were harvested and washed three times with M9 buffer and twice with DEPC water. TRIzol reagent (MRC) was added to the worm pellet (TRIzol/worm pellet ratio was 2/1), and the mixture was vigorously shaken for at least 1 min. The mixture was frozen at −80°C overnight or for a longer period before the next RNA extraction steps.
Frozen worms were then placed on ice, vortexed for 5 min, and settled at room temperature.

| Analysis of autophagy levels using an LGG-1 reporter strain
Autophagy was monitored using an lgg-1::GFP translational reporter.
GFP-positive puncta in seam cells was counted in L4 transgenic worms using a Leica DMI 6,000 B microscope (Leica, Nanterre, France) at 1,000× magnification. All worms were kept at 20°C and were treated with ethanol (as a control), and C8-SA and then their progeny were assayed at the L3 stage. | 9 of 11 aprotinin 2 µg/ml (Roche), leupeptin 1 µg/ml (Roche), and Pefebloc 1 µg/ml (Roche). Lysates were cleared by centrifugation, and protein concentration was assessed using the Pierce BCA protein assay (Thermo Scientific, Waltham, MA, USA). 30 µg of total protein extracts was loaded on 4%-12% Criterion XT Bis-Tris precast gels (Bio-Rad, Mississauga, Canada). Proteins were immobilized on PVDF membranes, which were blocked in 5% nonfat dry milk in
Total GFP signal for each worm was quantified by ImageJ software.
Data shown are the average number of pixels in the transgenic C. elegans (n = 10-15) at each indicated treatment. Data are presented as the mean ± SEM.

| Protein carbonylation
Protein extracts were prepared from 1-mL pellet of synchronized Day 2 adult nematodes grown on ht115 plates. All protein samples were resolved by electrophoresis through 10% gradient SDS-polyacrylamide gel. Proteins were detected by immunoblot using the OxyBlot Protein Oxidation Detection Kit following the provided protocol (Aguilaniu, Gustafsson, Rigoulet, & Nyström, 2003). Detections were accomplished using the ECL Plus Western Blotting Detection System following the provided instructions. Band intensities were quantified with ImageJ. OxyBlot analyses were repeated twice with lysates from separate nematode preparations.

ACKNOWLEDGMENT
We thank members of the Aguilaniu laboratory for comments of the manuscript. We are grateful to the Caenorhabditis Genetics Center for kindly providing strains. We thank the PLATIM (Plateau Technique Imagerie/Microscopie IFR 128). We also wish to acknowledge Sibelius and the ChronoscreenTM platform. This work was supported by grants from the L'Oréal Research and Innovation Center to HA and MS. HA is a consultant for L'Oréal Research and Innovation.