Inhibition of the neuromuscular acetylcholine receptor with atracurium activates FOXO/DAF‐16‐induced longevity

Abstract Transcriptome‐based drug screening is emerging as a powerful tool to identify geroprotective compounds to intervene in age‐related disease. We hypothesized that, by mimicking the transcriptional signature of the highly conserved longevity intervention of FOXO3 (daf‐16 in worms) overexpression, we could identify and repurpose compounds with similar downstream effects to increase longevity. Our in silico screen, utilizing the LINCS transcriptome database of genetic and compound interventions, identified several FDA‐approved compounds that activate FOXO downstream targets in mammalian cells. These included the neuromuscular blocker atracurium, which also robustly extends both lifespan and healthspan in Caenorhabditis elegans. This longevity is dependent on both daf‐16 signaling and inhibition of the neuromuscular acetylcholine receptor subunit unc‐38. We found unc‐38 RNAi to improve healthspan, lifespan, and stimulate DAF‐16 nuclear localization, similar to atracurium treatment. Finally, using RNA‐seq transcriptomics, we identify atracurium activation of DAF‐16 downstream effectors. Together, these data demonstrate the capacity to mimic genetic lifespan interventions with drugs, and in doing so, reveal that the neuromuscular acetylcholine receptor regulates the highly conserved FOXO/DAF‐16 longevity pathway.


| INTRODUC TI ON
Aging is a major risk factor for disease, and in recent years, cellular pathways and interventions influencing the aging process have steadily been uncovered. These are classified as the hallmarks of aging (López-Otín et al., 2013), which range from intracellular alterations, such as deregulated nutrient signaling pathways, to altered intercellular communication (López-Otín et al., 2013). Many genetic interventions have been identified to slow the progression of aging processes, and thereby progression of disease ( Figure 1a).
Identification of pharmacological means to address this risk would be of great benefit to society (Figure 1a) (Partridge et al., 2020).
Another hallmark of aging, altered intercellular communication, is most often discussed in relation to inflammation. Yet, more recently, the role of other forms of intercellular communication in the aging process has begun to emerge. One that is particularly relevant to this study is acetylcholine (ACh) signaling at the neuromuscular junction (NMJ). In Caenorhabditis elegans, ACh supplementation has been shown to improve stress tolerance (Furuhashi & Sakamoto, 2016). However, increased cholinergic transmission was shown to accelerate neurodegenerative phenotypes in mice (Sugita et al., 2016). There is currently limited insight into the specific role of NMJ ACh signaling in aging and longevity, yet a connection may be present.
The identification of geroprotectors, compounds that can slow the aging process to delay the onset of age-related diseases, is a relatively new field. Despite the development of some highthroughput methods, screening for these compounds is time consuming, due to the need to evaluate full lifespans of model organisms (Moskalev et al., 2016). Though commonly used aging models such as the nematode Caenorhabditis elegans have relatively short lifespans compared to mammals, testing compounds on a large scale is still difficult.
An alternative to large scale in vivo screening is transcriptomebased in silico drug screening. Here, the majority of the screening process is performed computationally, thereby considerably narrowing down the candidate list of compounds requiring validation.
This method, when applied to the transcriptional signature of the well-known longevity intervention, dietary restriction (DR), identified the DR mimetic allantoin (Calvert et al., 2016). Similarly, members of our group applied machine learning algorithms trained on human aging transcriptome datasets to drug-response transcriptomes. This method allowed the identification of HSP90 inhibitors, including monorden and tanespimycin, as geroprotectors (Janssens et al., 2019).
Here, using a transcriptomics-based in silico drug screen focused on the conserved genetic longevity intervention of overexpression of FOXO3, we identified a number of FDA-approved geroprotectors. From there, we focused on the neuromuscular blocker, atracurium, which activated FOXO targets in mammalian cell culture and extended both lifespan and healthspan in worms.
This extension was dependent on FOXO (daf-16 in worms) and antagonism of the acetylcholine receptor (AChR). Collectively, these results demonstrate a conserved longevity cascade in which inhibition of the AChR by atracurium leads to activation of FOXO/ DAF-16 and therefore extended longevity.

| In silico drug screen identifies compounds that mimic transcriptional signature of FOXO3 overexpression
In order to identify compounds that produce similar effects as FOXO3 overexpression, we utilized the library of integrated network-based cellular signatures (LINCS), a database and software suite containing transcriptional signatures of both drug-treated human cell lines and genetic perturbations (Keenan et al., 2018;Subramanian et al., 2017). LINCS includes the transcriptional signatures of 2837 compounds, many of which are FDA-approved, originating from a core set of eight cell lines (PC3, VCAP, A375, HA1E, HCC515, HT29, MCF7, and HEPG2).
With the FOXO3 overexpression transcriptional signature available in this database, we used the online software available within LINCS to search for compounds producing the most similar transcriptional effects to FOXO3 overexpression ( Figure 1c). For each compound, a standardized score was generated which ranged from −100 (highly opposing transcriptional signature to query candidate) to 100 (highly similar transcriptional signature to query candidate), and a summary score consolidating cell line data was generated for each compound. Summary scores greater than 90 were used as cutoff criteria to designate compounds with similar transcriptional signatures, as recommended by LINCS, and the ranked results were downloaded for further evaluation (Table S1).
This approach generated a list of 129 candidate drugs mimicking FOXO3 overexpression transcriptionally ( Figure 1d; Table S1).
To categorize commonalities within this list, which contained both FDA-approved drugs and compounds developed for research use, we used the drug descriptions associated with each compound to perform enrichment analysis. This allowed us to see which drug classes were most prevalent in our list, relative to the entire dataset.
We found that both mTOR inhibitors and PI3Kinase inhibitors were strongly enriched as drug classes that matched the FOXO3 overexpression mRNA signature (Figure 1e). Because mTOR and PI3Kinase inhibitors are both known to increase lifespan, and act within the FOXO longevity pathway (Babar et al., 1999;Hay, 2011), we concluded that our approach could indeed identify compounds that are known to extend lifespan through the FOXO signaling pathway.
As it is more straightforward to repurpose drugs already used in a clinical setting than to develop new ones, we continued investigation with our top five hits from the screen that already possessed FDA approval ( Figure 1f). These compounds included two positive controls known to extend model organism lifespan: the serotonin antagonist cyproheptadine (ranked 1st on the FDA-approved subset; ranked 4th overall) (Petrascheck et al., 2007) and the mTOR inhibitor rapamycin (2nd FDA-approved; 14th overall) (Robida-Stubbs et al., 2012) F I G U R E 1 In silico drug screen identifies compounds that mimic transcriptional signature of FOXO3 overexpression. (a) Schematic depicting the increased incidence of disease throughout the aging process. Many genetic interventions (orange) have been described that intervene in the aging process to promote health and increase lifespan. Drug interventions (blue) are less prevalent, but here we describe efforts to identify compounds capable of producing the same effects. (b) FOXO's conserved role in longevity regulation across the phylogenetic tree. FOXO and its orthologs directly regulate, or are associated with, aging in organisms as distantly related as yeast and humans. (c) The library of integrated network-based cellular signatures (LINCS), a database that holds a core set of transcriptional signatures from at least five different commonly used cell lines. Transcriptional signatures for over 2500 drugs, over 3500 gene knock-downs and over 2000 gene over expressions are available. We used the FOXO3 overexpression transcriptional signature present in LINCS to query for mimetic compounds. (d) The top 15 hits mimicking FOXO3 overexpression from the LINCS core compounds and their summary score integrating results from all cell lines tested. All compound hits can be found in Table S1. (e) Enrichment of drug classes in the candidate list, shows mTOR and PI3K inhibitors as highly enriched drug classes for their potential to mimic FOXO3 overexpression among the candidate drugs. Names of drug classes are only displayed if passing significance of enrichment (adjusted p < 0.05, Y axis). Number of drug hits from the drug class is depicted on the X axis, while sizes of the dots in the scatterplot denote what percentage of the drug class were hits (i.e., 5 hits out of a class of 10 would denote a 50% coverage). (f) Schematic of in vitro validation method to test candidate drugs. HL1 mouse cardiac cells were cultured with compounds for 24 h of exposure. FOXO3 and target gene expression was assessed by qPCR (mRNA level normalized to reference gene Gapdh). The top 5 ranked compounds from the candidate list (figure d) that also had FDA approval status were used for testing. (g-k) The expression levels of Foxo3 and its target genes-Gadd45a, Ccng2, Hspa1b, Cat, Bcl6, Scp2 as assessed by qPCR-in either control cells (gray) or cells treated with drug candidates (blue)-cyproheptadine (g), rapamycin (h), atracurium (i), sildenafil (j), and zidovudine (k). All drugs were assessed at a concentration of 50 µM, except for rapamycin that was assessed at 10 µM to avoid toxicity. Asterisk indicates significantly increased expression for each gene (p < 0.05, t test) with the drug relative to control
The single FOXO ortholog is thought to be crucial for self-renewal and immortal capacity (Boehm et al., 2012).
Foxo3 is required for life-extending effects of dietary restriction in M. musculus (Shimokawa et al., 2015).  Table S3 for healthspan statistics. (f) Survival curves showing significant lifespan extension with atracurium at increasing doses (5 μM, 50 μM, 500 μM). While some dose-effect seems present, and all doses significantly extend lifespan relative to water vehicle treatment, the differences between concentrations are not statistically significant. All statistical comparisons of survival curves are determined by log-rank tests. For each condition, n = 100 worms. See Table S2 for lifespan statistics. ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, ns-not significant Foxo3 target genes ( Figure 1f). Target genes included Gadd45a, a DNA repair gene directly stimulated by FOXO3 activity (Tran et al., 2002), Ccng2, and Bcl6, both cell cycle related genes also directly regulated by FOXO3 (Fernandez de Mattos et al., 2004;Fu & Peng, 2011), the catalase Cat and the sterol carrier Scp2, which are part of the cellular antioxidant defense system regulated by Foxo transcription factor activity in general (Dansen et al., 2004;Klotz et al., 2015), and Hspa1a, which encodes a heat shock protein regulated by FOXO transcription factors, conserved across species (Webb et al., 2016). Relative to untreated cells, we found that most compounds could significantly activate the expression of Foxo3 itself as well as the majority of the targets we tested (Figure 1g-k). These data suggest that our compound screen was successful in identifying compounds that mimic overexpression of FOXO3 by activating Foxo3 target genes.

| Zidovudine and atracurium both extend lifespan in Caenorhabditis elegans, but only atracurium depends on daf-16
To test the effects of the identified compounds on lifespan, we turned to C. elegans, a simple and well-described model organism for aging.
As noted above, cyproheptadine and rapamycin are known lifespan extenders (Petrascheck et al., 2007;Robida-Stubbs et al., 2012), so we proceeded to test lifespan effects of atracurium, sildenafil, and zidovudine. Of these three, zidovudine and atracurium robustly and reproducibly extended lifespan at a concentration of 50 μM (Figure 2a,b; Table S2), whereas sildenafil led to no consistent difference in lifespan with the concentration and conditions we tested (Table S2).
Because our compound screen was based on overexpression of is dependent on daf-16, while zidovudine is not. We expect that the highly conserved nature of FOXO/daf-16 may be of significant translational relevance and therefore continued investigation of atracurium to further understand the upstream signaling at play.
Healthy aging is not only determined by lifespan, but also healthspan. One way to measure the health of worms is through mobility, or average crawling speed (Pierce-Shimomura et al., 2008). We titrated the concentration of atracurium to additionally test healthspan for dose-responses. At day 10 of adulthood, atracurium significantly increased age-related mobility in a dose-dependent manner ( Figure 2e; Table S3), suggesting that atracurium extends not only lifespan, but healthspan as well. We then titrated the concentration of atracurium to determine if there was a similar dose-response in lifespan extension. Each concentration tested extended lifespan when compared to the water vehicle ( Figure 2f). However, while higher concentrations of atracurium further extended median lifespan (Table S2), the differences between the various doses on survival were not statistically significant ( Figure 2f). Therefore, while healthspan increases in a dosedependent manner, we are unable to conclude that there is a significant dose-response effect in lifespan with atracurium.

| Atracurium extends healthspan in Caenorhabditis elegans by antagonizing the neuromuscular acetylcholine receptor
Atracurium is a neuromuscular blocker and is used in patient care as an anesthetic muscle relaxant. We therefore considered the possibility that, if the pharyngeal muscles were relaxed by atracurium treatment, the worm would be unable to pump bacteria efficiently into its body, consequently reducing food intake and inducing a state of dietary restriction (DR). We therefore tested pharyngeal pumping on day 1 of adulthood. Contrary to our expectation, worms treated with atracurium showed a significantly increased rate of pumping when compared to their control counterparts ( Figure S1A). Also contrary to a DR phenotype, the body area of the worms at day 10 of adulthood was increased by atracurium relative to untreated worms in a dose-dependent manner ( Figure S1B). These data together suggest that DR is not the cause of atracurium's extension of healthspan.
At the NMJ in physiological conditions, ACh crosses the synapse to bind to the AChR, causing the channel to depolarize and the muscle to contract (Rand, 2007). Atracurium competitively antagonizes the AChR (Lee, 2001;Wishart et al., 2018), preventing this contraction and allowing the muscle to relax instead ( Figure 3a). We questioned if atracurium extends healthspan through this pathway, or instead through an off-target effect. Atracurium binds an alpha subunit of the AChR, encoded by CHRNA2 in humans (Wishart et al., 2018). The closest homolog to this subunit in C. elegans is encoded by the gene unc-38 ( Figure S1C) (Fleming et al., 1997;Harris et al., 2020). There are multiple strains of C. elegans available with mutant unc-38, so we chose to utilize both unc-38(e264) and unc-38(x20). As the AChR system is not entirely homologous between human and C. elegans (Rand, 2007), we also included mutants of a range of genes involved in NMJ signaling. There are 27 genes identified as AChR genes in C.
elegans, which have been divided into five classes of protein subunits based on sequence similarity (Rand, 2007). We wished to test all five classes, and therefore included mutants for unc-29(e193), acr-8(ok1240), acr-16(ok789), and deg-3(u662) (Figure 3b, Figure S1C). To determine if the effects we saw with atracurium were dependent on any of these subunits, we tested healthspan at day 10 of adulthood.
We saw again increased healthspan in wild type N2 worms treated with atracurium compared to those untreated ( Figure 3e). However, in both unc-38(e264) and unc-38(x20) mutant worms, atracurium was no longer able to prolong healthspan ( Figure 3b). This effect was specific to unc-38, as atracurium was still able to increase healthspan in the other four NMJ mutants tested ( Figure 3b). Therefore, these data show that unc-38 is specifically required for atracurium-induced improvements in healthspan.
We next tested if absence of unc-38 also prevented atracuriuminduced lifespan extension. Indeed, unlike treatment in N2 worms ( Figures 2a, 3c), atracurium was unable to extend lifespan in either of these mutants (Figure 3d,e). In fact, in the unc-38(e264) mutant, atracurium appeared to have detrimental effects on lifespan ( Figure 3e). Altogether, these data suggest that atracurium specifically requires the unc-38 subunit of the AChR to enact improvements on healthspan and lifespan in C. elegans, and these effects are not due to an off-target response.

| Atracurium and unc-38 RNAi similarly stimulate DAF-16 nuclear localization and extend healthspan and lifespan
Having determined that the lifespan extension induced by atra-  As it appeared that unc-38 RNAi caused a similar stimulation of DAF-16 nuclear localization as atracurium treatment, we sought to F I G U R E 3 Atracurium extends healthspan and lifespan by antagonizing the neuromuscular acetylcholine receptor. (a) Representation of the canonical mechanism of atracurium at the neuromuscular junction (NMJ) relevant to this study. In humans, atracurium is used to relax muscles, and does so by antagonizing the acetylcholine receptor (AChR). Under normal conditions, acetylcholine (ACh) binds the receptor, causing depolarization of the channel, and contraction of the muscle. Atracurium binds the AChR in place of ACh, preventing this depolarization and contraction. (b) Mobility of atracurium-treated (50 μM) N2 worms and mutants of players in NMJ ACh signaling at day 10 of adulthood. Atracurium improves healthspan in N2 worms, but this effect is abolished in worms with mutant unc-38. unc-38 encodes an alpha subunit of the AChR, and is the worm ortholog to human CHRNA2, which binds atracurium in humans. The other ACh pathway mutants are still responsive to atracurium, suggesting that the increased healthspan is dependent on this specific subunit of the AChR. For each condition, n = ~50-90 measurements of ~50 worms. Statistical significance for each mutant is determined by an individual Mann-Whitney U test comparing treated and untreated. See Table S3 Table S2 for lifespan statistics. ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, ns -not significant RNAi of unc-38 also extended lifespan, though we observed shortened lifespan upon combined unc-38 RNAi and atracurium treatment ( Figure 4d). Altogether, these data suggest that inhibition of UNC-38 with atracurium, or knockdown of unc-38 using RNAi, both cause DAF-16 nuclear localization through a similar pathway to improve healthspan and lifespan in worms.

| RNA sequencing demonstrates induction of protective stress responses and metabolic remodeling by FOXO/DAF-16 downstream effectors
Having concluded that atracurium causes nuclear localization of DAF-16, we wished to examine the downstream transcriptional effects. Therefore, we performed RNA sequencing on worms treated for 24 h (from L4 stage to day 1 of adulthood) with atracurium compared to those untreated. The two treatment groups F I G U R E 4 Atracurium and unc-38 RNAi similarly stimulate DAF-16 nuclear localization and extend healthspan and lifespan. (a) Representative images of day 1 adult worms expressing DAF-16::GFP upon feeding with either empty vector or unc-38 RNAi HT115 bacteria, and treatment with water vehicle or atracurium (50 μM). The left panel of each condition shows fluorescent images of DAF-16::GFP nuclear localization, while the right panel displays overlay of DAF-16::GFP fluorescent images with bright field images of worms. Scale bar measures 25 μM and can be applied to all images. (b) Quantification of A. n = ~20-30 different animals, pooled from two independent experiments. Treatments are compared to heat stress at 35°C for 3 h as a positive control. A chi-square test of independence was performed to assess the significance of differences in the levels of DAF-16 nuclear enrichment between each of the treatments. Both atracurium treatment and unc-38 RNAi significantly increase the proportion of worms with nuclear-localized DAF-16 when compared to the untreated empty vector condition (p < 0.0001). There is no significant difference between worms fed unc-38 RNAi with and without atracurium. (c) Mobility of N2 worms upon feeding with either empty vector or unc-38 RNAi, and treatment with water vehicle or atracurium (50 μM) at day 12 of adulthood. Both atracurium (50 μM) and unc-38 RNAi improve healthspan in empty vector-treated worms. For each condition, n = ~85-110 measurements of ~50 worms. Statistical significance is determined by one-way ANOVA. Healthspan statistics can be found in Table S3. (d) Survival curves showing that both atracurium (50 μM) and unc-38 RNAi extend lifespan in N2 worms, but atracurium shortens lifespan when combined with unc-38 RNAi. For each condition, n = 120. Lifespan statistics are determined by log-rank tests. See Table S2 for lifespan statistics. ****p < 0.0001, ***p < 0.001, *p < 0.05, ns -not significant were readily distinguishable and separable using partial least squares discriminant analysis (PLS-DA; Figure 5a). Differential analysis showed a number of significant hits after correcting for multiple hypothesis testing (Figure 5b; Table S4). We noted that many of the significantly upregulated genes are FOXO downstream targets, for instance the alcohol dehydrogenase sodh-1, the heat shock protein hsp-70, and several members of the cytochrome P450 family (Figure 5b).
In order to form a more comprehensive view of the changing processes, we investigated the differentially expressed genes further, using the Database for Annotation, Visualization and Integrated Discovery (DAVID) bioinformatics resource (Huang et al., 2009).
We performed functional annotation clustering, using a less strict significance cutoff (unadjusted p < 0.01). Enriched annotation clusters included functional categories, GO-terms, KEGG pathway features and protein domains (Table S5). Upregulated clusters included protective stress responses, for instance those relating to oxidative stress and proteostasis, as well as a number of metabolic processes, such as carbon metabolism and the TCA cycle (Figure 5c).

| Atracurium activates DAF-16 target genes, dependent on unc-38
Activation of oxidative stress responses and proteostasis, as well as modulation of metabolic processes, are also key downstream effects  Table S4. (c) Functional annotation clustering of the significantly (using unadjusted p-value < 0.01) up-and downregulated genes performed using the DAVID Bioinformatics Database with an enrichment score >1.3. (df) Known DAF-16-regulated oxidative stress response (d), proteostasis (e), and metabolic genes (f) that are upregulated with atracurium treatment, as indicated by RNA-seq. For each fold change shown, unadjusted p-value is <0.01. (g) Hypothesized mechanism for atracuriummediated longevity. We have shown that atracurium requires the AChR subunit UNC-38 to extend healthspan and lifespan, thereby competitively antagonizing the receptor and this inhibition causes DAF-16-mediated lifespan extension. Longevity is likely caused by DAF-16 activation of genes known to induce longevity such as those involved in protective oxidative stress and proteostasis responses, as well as metabolic remodeling  Oxidative stress genes (p<0.01) of DAF-16 activation, and the majority of genes within the clusters we identified are also known to be upregulated by DAF-16 (Harris et al., 2020;Honda & Honda, 1999;Webb et al., 2016). To quantify this, we compared the differentially expressed genes in our RNAsequencing dataset to a previously described dataset of DAF-16 target genes (Webb et al., 2016). In doing so, we saw significant overlap of differentially expressed genes upon atracurium with those described as regulated by DAF-16 ( Figure S2A).

c t l-3 c t l-1 c y p -2 5 A 2 c y p -3 3 C 5 g s t -4 g s t -5 g s t -3 1 g s t -3 5 c y p -2 9
Significantly upregulated oxidative stress response genes in-

| Lifespan and healthspan increases with atracurium depend on daf-16 and skn-1, but not germ cell proliferation
GO-term analysis of our RNA-sequencing dataset provided not only evidence of DAF-16 involvement, but possible other mechanisms or interactions. Firstly, our most enriched upregulated cluster was related to oxidative stress. While DAF-16 is highly connected to this process (Honda & Honda, 1999), SKN-1 may be considered the canonical transcription factor related to oxidative stress (Blackwell et al., 2015). Therefore, we performed RNAi of daf-16 or skn-1 to determine how this influenced lifespan and healthspan in combination with atracurium treatment. We observed that knockdown of either transcription factor prevented the increase in age-related mobility seen with atracurium treatment ( Figure S2C). In addition, like the lifespan abrogation we saw with daf-16 mutation (Figure 2d), there was no lifespan extension with atracurium upon daf-16 RNAi ( Figure S2D). Similarly, no lifespan extension was present upon skn-1 RNAi ( Figure S2E).
Together, these data suggest that the healthspan and lifespan effects induced by atracurium treatment are not only dependent on daf-16, but also skn-1.
Lastly, the largest GO-term cluster downregulated in our RNA sequencing was spermatogenesis. Therefore, we sought to determine if germ cell proliferation played a role in the lifespan extension we observe with atracurium. First, we tested if the progeny-disrupting compound 5-fluorouracil (5FU) -used in all other lifespan experiments to prevent egg hatching-was necessary for this extension by performing lifespans in a novel microfluidic device which does not require treating worms with 5FU. In the absence of 5FU, atracurium still significantly extended lifespan ( Figure S2F). Additionally, we tested if starting treatment of atracurium at various time points would influence its effects. When treatment began at day 1 and day 3 of adulthood, like our normal treatment beginning at L4, atracurium significantly extended lifespan ( Figure S2G). However, that lifespan extension was no longer significant when treatment began at day 5 of adulthood ( Figure S2G). Eliminating the germ-line precursor cells through day 1 of adulthood can activate DAF-16 and extend lifespan in worms (Arantes-Oliveira et al., 2002). Because atracurium extends lifespan when treatment begins at day 3, after the majority of germ cell proliferation has occurred, we conclude that disruption of germ cell proliferation is not directly responsible for the lifespan extension we observe.

| DISCUSS ION
Increasing evidence demonstrates that slowing the progression of aging processes with geroprotective compounds can treat agerelated disease, thereby necessitating the discovery and deeper understanding of geroprotectors (Moskalev et al., 2016;Partridge et al., 2020). We have utilized an in silico drug screening strategy to identify compounds that mimic the FOXO3 overexpression transcriptome. In doing so, we identified a number of FDA-approved compounds, both known and unknown previously to defer aging.
The top five compounds of the screen activated at least some Foxo3 targets in mammalian cell culture. When we tested for longevity effects in vivo, both the NRTI zidovudine and the neuromuscular blocker atracurium robustly extended lifespan in C. elegans, but only atracurium was dependent on FOXO/daf-16. Atracurium extended not only lifespan, but also healthspan in worms. We found that these healthspan effects were dependent on the AChR subunit encoded by unc-38, the closest worm homolog to human CHRNA2, which encodes the subunit known to bind atracurium upon treatment. Knockdown of unc-38 seemed to genetically mimic the pharmacological intervention of atracurium treatment, stimulating DAF-16 nuclear localization, as well as lifespan and healthspan benefits. RNA sequencing revealed differential expression of genes known both to be regulated by DAF-16 signaling and to modulate aging.
We chose to base our in silico screen on overexpression of FOXO3, as it is one of the most widely recognized and conserved genetic interventions to promote longevity (Martins et al., 2016).
We therefore hypothesized that compounds mimicking this transcriptome, particularly by activating FOXO/DAF-16 itself, would also induce conserved longevity effects. Specifically, FOXO3 expression has been associated with long-lived populations of humans in several independent cohorts (Broer et al., 2015;Morris et al., 2015;Soerensen et al., 2015) suggesting that activation of this pathway is one of the most promising routes to benefiting The healthspan and lifespan extensions induced by atracurium were dependent on the AChR subunit unc-38, demonstrating that atracurium works to promote longevity through its canonical mechanism. Though the specific role of AChR antagonists was previously unstudied, agonists of the AChR were shown to promote recovery from the developmentally arrested dauer phase (Tissenbaum et al., 2000). This effect is dependent on insulin-like neuroendocrine signaling (Tissenbaum et al., 2000), a major regulator of DAF-16 (Gottlieb & Ruvkun, 1994). In that study, the authors suggest a mechanism by which ACh activation of the AChR activates DAF-2, which would then inhibit activation of DAF-16. This seems in line with our data, which demonstrates inactivation of the AChR activates DAF-16.
There is now more recent evidence that altered NMJ signaling can influence aging. In worms, mutation of the potassium channel gene slo-1, which dampens synaptic release from motor neurons, also led to increased healthspan and lifespan in worms, dependent on DAF-16 (Li et al., 2019). Interestingly, RNAi of slo-1 improved healthspan and lifespan when started at day 5 or day 7 or adulthood, but this effect was not present when RNAi was started at day 1 or day 3. When combined with our findings that atracurium treatment extends lifespan when started at L4, day 1 or day 3, these results suggest that for worms there is a point (seemingly around day 5) when it becomes beneficial for longevity to stimulate NMJ signaling, as opposed to inhibit it.
In a mouse model comparable to the slo-1 knockout, mice with overexpression of the vesicular acetylcholine transporter (VAChT) allowed for analysis of how increased release of ACh affected development and aging (Sugita et al., 2016). In this model, increased ACh accelerated aging of the NMJ, measured by fragmentation, denervation, sprouting of motor axon nerve endings, and innervation by multiple motor axons (Sugita et al., 2016). Degeneration of the NMJ preceded motor deficits and muscular atrophy in the overexpression mice, and, upon crossing mice of this genotype with a mouse model for ALS, increased ACh transport led to accelerated pathology and early death (Sugita et al., 2016). These data provide insight into the converse mechanism-overactivation of the AChR as a mechanism for premature aging-and align with our findings that inhibition of the AChR promotes longevity. We observed dependence of the lifespan and healthspan benefits caused by atracurium on skn-1, as well as daf-16. Additional longevity interventions have been identified that depend on these two pathways simultaneously (Robida-Stubbs et al., 2012;Wan et al., 2020), and DAF-16 itself has been shown to activate SKN-1 (Tullet et al., 2017). While the focus of our study was identifying FOXOmimetics, and attempting to activate this pathway specifically to promote longevity, the possible interaction between these two transcription factors merits further investigation.
Altogether, our results demonstrate the power of in silico drug screening, particularly the potential to mimic genetic longevity interventions with compounds. In doing so, we identified the neuromuscular blocker atracurium as a geroprotector, a compound that activates FOXO3 targets in vitro in mammalian cell culture, and promotes longevity in vivo in worms. Using atracurium, we show that neuromuscular acetylcholine signaling acts as a regulator of the conserved FOXO longevity pathway. This finding facilitates deeper understanding of this aging mechanism and subsequent identification of potential targets for the treatment of age-related diseases.

| LINCS database compound screen
The online library of integrated network-based cellular signatures (LINCS) (Keenan et al., 2018;Subramanian et al., 2017) Table S1. A cutoff was applied to the ranked list whereby compounds with a score greater than 90 were considered to match the FOXO3 oe transcriptional signature. The top 5 FDA-approved drugs within this list were used for further in vitro and in vivo evaluation.

| Extraction of mRNA from cells and quantitative real-time PCR (qPCR)
Isolation of total mRNA from cells was performed with TRI-

| RNAi experiments
Escherichia coli HT115 (DE3) with the Empty Vector (EV) L4440 was obtained from the Caenorhabditis Genetics Center. Bacterial feed- ing RNAi experiments were performed as described (Kamath et al., 2001). RNAi E. coli feeding clones used were unc-38(F21F3.5), daf-16(R13H8.1), and skn-1(T19E7.2) derived from the Ahringer RNAi library (Lee et al., 2003). All these clones were confirmed by sequencing and knockdown efficiency was confirmed by qPCR. In all RNAi experiments described in this study, worms were subjected to RNAi bacteria from the time of hatching.

| Pharmacological treatment of Caenorhabditis elegans
All chemicals were obtained from Sigma-Aldrich. Atracurium besylate was dissolved in water at a concentration of 16.67 mM.
Zidovudine and sildenafil citrate salt were dissolved in DMSO at a concentration of 50 mM. These compounds were added to plates just before pouring at the concentrations described. Unless otherwise described, worms were treated with compounds from L4 stage onwards. Plates were changed at least once a week to ensure consistent exposure to the compound.

| Manual lifespan analysis
Gravid adult worms were age-synchronized using alkaline hypochlorite treatment and incubated in M9 buffer overnight. L1 larval stage worms were seeded to NGM plates and grown to L4 stage. At L4 stage, worms were transferred to plates supplemented with compounds as described and 10 µM 5-fluorouracil (Sigma-Aldrich).
10 µM 5-fluorouracil treatment continued for the first 2 weeks of life. All assays were performed at 20°C, and the L4 stage was counted as day 0 of life. Per condition, 100-120 worms were used.
Survival was analyzed every other day and worms were considered dead when they did not respond to repeated prodding. Worms that were missing, displaying internal egg hatching, losing vulva integrity, and burrowing into NGM agar were censored. Statistical analyses of lifespan were calculated by Log-rank (Mantel-Cox) tests on Kaplan-Meier curves in GraphPad Prism. Prep Kit for Illumina" (NEB #E7760S/L). Briefly, mRNA was isolated from total RNA using the oligo-dT magnetic beads. After fragmentation of the mRNA, cDNA synthesis was performed. This was used for ligation with the sequencing adapters and PCR amplification of the resulting product. The quality and yield after sample preparation

| Microfluidic lifespan analysis
was measured with the Fragment Analyzer. The size of the resulting products was consistent with the expected size distribution (a broad peak between 300-500 bp). Clustering and DNA sequencing using the NovaSeq6000 was performed according to manufacturer's protocols. A concentration of 1.1 nM of DNA was used. NovaSeq control software NCS v1.6 was used.

| Read mapping, statistical analyses, and data visualization
Reads were subjected to quality control FastQC (Andrews, 2010) trimmed using Trimmomatic v0.32 (Bolger et al., 2014) and aligned to the C. elegans genome obtained from Ensembl (wbcel235.v91), using HISAT2 v2.1.0 . Counts were obtained using HTSeq (v0.11.0, default parameters) (Anders et al., 2015) using the corresponding GTF taking into account the directions of the reads. Statistical analyses were performed using the edgeR v3.26.8 (Robinson et al., 2010) and limma/voom v 3.40.6 (Ritchie et al., 2015) R packages. All genes with more than 2 counts in at least 3 of the samples were kept. Count data were transformed to log2-counts per million (logCPM), normalized by applying the trimmed mean of M-values method (Robinson et al., 2010) and precision weighted using voom (Law et al., 2014). Differential expression was assessed using an empirical Bayes moderated t test within limma's linear model framework including the precision weights estimated by voom (Law et al., 2014;Ritchie et al., 2015).
Resulting p-values were corrected for multiple testing using the Benjamini-Hochberg false discovery rate. Data processing was performed using R v3.6.1 and Bioconductor v3.9. Partial least squares discriminant analysis (PLS-DA) was performed using mixomics (Rohart et al., 2017) setting a variable of importance (VIP) score of greater than 1 as significant. Resulting p-values (where applicable) were corrected for multiple testing using the Benjamini-Hochberg false discovery rate. Genes were reannotated using biomaRt using the Ensembl genome databases (v91). Data visualization was performed using gplots (Warnes et al., 2015) and ggplot2 (Wickham, 2016) selecting colors from RcolorBrewer (Neuwirth, 2014). Functional annotation clustering of differentially expressed genes was performed using the DAVID bioinformatics resource with all default settings (Huang et al., 2009). Clusters with an enrichment score of greater than 1.3 were considered significant. The RNA-seq data are available on GEO under the ID GSE14 9944.

| Isolation of RNA and qPCR of Caenorhabditis elegans
Worms were synchronized, grown and treated with compounds from L4 stage as described above. Twenty-four hours later, worms were washed from treatment plates, three times in M9 buffer and two times in water before being snap frozen in liquid nitrogen. For each sample, ~1000 worms were collected. For isolation of total mRNA, whole worms were homogenized in TRI-reagent (Sigma-Aldrich) with a 5 mm steel bead using a TissueLyser II (QIAGEN) for 5 min at frequency of 30 times/s. RNA was quantified with a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific) and 1 µg of extracted RNA was reverse transcribed into cDNA according to manufacturer's instructions using the QuantiTect Reverse Transcription Kit (QIAGEN; Venlo, The Netherlands).
Quantitative gene expression analysis was performed using the LightCycler Ⓡ 480 SYBR Green I Master (Roche; Woerden, The Netherlands) and measured using the LightCycler Ⓡ 480 Instrument II (Roche). Gene-specific primers were synthesized according to the sequences in Table S7. The N0 values of target genes were normalized to the geometric mean of reference genes ama-1 and cdc-42.

ACK N OWLED G EM ENTS
The authors thank the Caenorhabditis Genetics Center at the

CO N FLI C T O F I NTE R E S T
MR, SG, MT, and SAV have equity in NemaLife Inc.

S U PP O RTI N G I N FO R M ATI O N
Additional supporting information may be found online in the Supporting Information section.