ASM-3 Acid Sphingomyelinase Functions as a Positive Regulator of the DAF-2/AGE-1 Signaling Pathway and Serves as a Novel Anti-Aging Target

In C. elegans, the highly conserved DAF-2/insulin/insulin-like growth factor 1 receptor signaling (IIS) pathway regulates longevity, metabolism, reproduction and development. In mammals, acid sphingomyelinase (ASM) is an enzyme that hydrolyzes sphingomyelin to produce ceramide. ASM has been implicated in CD95 death receptor signaling under certain stress conditions. However, the involvement of ASM in growth factor receptor signaling under physiological conditions is not known. Here, we report that in vivo ASM functions as a positive regulator of the DAF-2/IIS pathway in C. elegans. We have shown that inactivation of asm-3 extends animal lifespan and promotes dauer arrest, an alternative developmental process. A significant cooperative effect on lifespan is observed between asm-3 deficiency and loss-of-function alleles of the age-1/PI 3-kinase, with the asm-3; age-1 double mutant animals having a mean lifespan 259% greater than that of the wild-type animals. The lifespan extension phenotypes caused by the loss of asm-3 are dependent on the functions of daf-16/FOXO and daf-18/PTEN. We have demonstrated that inactivation of asm-3 causes nuclear translocation of DAF-16::GFP protein, up-regulates endogenous DAF-16 protein levels and activates the downstream targeting genes of DAF-16. Together, our findings reveal a novel role of asm-3 in regulation of lifespan and diapause by modulating IIS pathway. Importantly, we have found that two drugs known to inhibit mammalian ASM activities, desipramine and clomipramine, markedly extend the lifespan of wild-type animals, in a manner similar to that achieved by genetic inactivation of the asm genes. Our studies illustrate a novel strategy of anti-aging by targeting ASM, which may potentially be extended to mammals.

In mammals, acid sphingomyelinase (ASM) is a phosphodiesterase that hydrolyzes sphingomyelin to produce ceramide and phosphorylcholine [16]. When acting in lipid rafts, which are plasma membrane microdomains [17], ASM leads to the production of ceramide-enriched lipid rafts [18,19]. Because ceramides have a physical property of self-association, it is likely that ceramide-enriched lipid rafts can provide a unique local microenvironment for protein-protein interactions. Indeed, ASM-dependent and ceramide-enriched lipid rafts have been shown to facilitate the oligomerization and signaling of CD95 receptors, leading to apoptosis [18,19]. However, most recent studies have shown that CD95, previously known as death receptors, also possess pro-proliferation and pro-survival function in vivo under physiological conditions [20]. It is not known if ASM is involved in the pro-proliferation and pro-survival signaling of the CD95 receptors. In addition to ASM, neutral sphingomyelinase (NSM) can also hydrolyze sphingomyelin to produce ceramide, although the exact site of NSM action is not clear [21,22]. Ceramides can also be synthesized by a de novo biosynthetic pathway, which occurs in the endoplastic reticulum and Golgi apparatus [21,22]. Ceramides produced through biosynthesis are also known to affect the cellular processes such as cell signaling, stress resistance, and apoptosis, although the detailed molecular mechanisms are not clear [21,22]. As ceramides produced by hydrolysis of sphingomyelin and through biosynthesis are localized in different cellular compartments, it is possible that they may have different cellular functions. In C. elegans, while the ceramides produced through de novo biosynthesis are known to be involved in the radiation-induced apoptosis in the germline [23], the roles of ceramides produced through hydrolysis of sphingomyelin are not clear, although C. elegans have been shown to contain ASM homologs but no NSM homologs [23,24].
In the current study, we used a genetic approach to study the role of asm in signal transduction in the C. elegans model system. Our results show that C. elegans ASM homologs, in particular, ASM-3, is a critical and positive regulator of the DAF-2/IIS pathway and controls both animal lifespan and dauer formation. Using pharmacological agents, we further demonstrated that inhibition of CeASM, similar to genetic inactivation of the asm genes, leads to significant animal lifespan extension. Our studies illustrate a novel anti-aging strategy of targeting CeASMs to downregulate the DAF-2/IIS pathway, which could potentially be extended to mammals. RNAi extended animal lifespan with the mean lifespan 19% greater than that of the control in the rrf-3(pk1426) background (P,0.0001). (B) asm-3(ok1744) mutants had 14% longer lifespan than wild-type (N2) animals (P = 0.0141). (C) Knockdown of asm-1, asm-2 or asm-3 by RNAi each extended lifespan with the mean lifespan 12%, 10% or 19% greater than that of the vector control in the rrf-3(pk1426) background, respectively (P = 0.0068 for asm-1 RNAi, P = 0.0258 for asm-2 RNAi and P,0.0001 for asm-3 RNAi). (D) Experiments were carried out in the asm-3(+);rrf-3(pk1426) or asm-3(ok1744);rrf-3(pk1426) background. asm-3(ok1744) mutation extended lifespan with the mean lifespan 15% greater than that of the control (P = 0.0018), and the lifespan of asm-3(ok1744) mutant was further enhanced by RNAi of asm-1 or asm-2 with the mean lifespan 30% or 28% greater than that of the control (P,0.0001 for asm-1 RNAi or asm-2 RNAi). Lifespan extension produced by asm-3 mutation was inhibited by daf-16 RNAi (P,0.0001). (E) daf-16(mgDf47) null mutation completely abolished lifespan extension phenotype of asm-3 mutants. (F) Lifespan extension phenotype by asm-1, asm-2 or asm-3 RNAi shown in (C) was completely abolished by daf-18(nr2037) null mutation in the rrf-3(pk1426);daf-18(nr2037) background. Mean lifespan, P values and other details for these experiments are listed in Table 1 and Table 2. doi:10.1371/journal.pone.0045890.g001

asm Genes Regulate Animal Lifespan as Novel Longevity Genes
The C. elegans genome encodes three ASM homologs, asm-1, asm-2 and asm-3. Among them, ASM-3 is most closely related to human ASM, and the two proteins share 42% identity in the predicted C-terminal catalytic domain ( Figure S1). The asm-3 gene was initially discovered in our recent genome-wide RNAi screen for new genes regulating aging [25]. In the current study, we found that RNAi-mediated inactivation of asm-3 produces a lifespan extension phenotype ( Figure 1A). The mean lifespan of animals treated with asm-3 RNAi was 19% longer than that of the vector control (Table 1, Set #1). To verify lifespan extension phenotype in asm-3 mutant, we also investigated a chromosomal mutation of asm-3, asm-3(ok1744), a putative null allele with a 1558 bp deletion and a 7 bp insertion in the predicted catalytic domain ( Figure S2). Consistent with RNAi experiments, the asm-3(ok1744) mutants also extended animal lifespan by 14% as compared to the wildtype ( Figure 1B; Table 2, Set #1). In C. elegans, two other paralogs, asm-1 and asm-2, encode polypeptides highly homologous to ASM-3 ( Figure S3). We thus examined their roles in lifespan regulation. When asm-1 or asm-2 was inactivated by RNAi knockdown, a modest lifespan extension phenotype was observed ( Figure 1C), with the mean lifespan 12% or 10% greater than that of the control, respectively (Table 1, Set #2). We speculated that there may be functional redundancy between asm-3, asm-1 and asm-2. Indeed, double inactivation of asm-3 and asm-1 or of asm-3 and asm-2 further extended lifespan, with the mean lifespan 30% or 28% greater than that of the control, respectively ( Figure 1D; Table 1, Set #3). These results suggest that asm-3, asm-1 and asm-2 each contributes to the regulation of animal lifespan, and inactivating two asm genes produces an additive effect on lifespan extension. Together, our results demonstrate the importance of the asm gene family in the regulation of lifespan and highlight asm-3 as the most prominent member in this process.
Genetic Interactions of asm-3 with Negative Regulators of the daf-2/IIS Pathway To investigate whether asm-3 functions in the daf-2/IIS pathway, we examined genetic interactions of asm-3 with daf-16 and daf-18, which negatively regulate the daf-2/IIS pathway. Inactivation of daf-2 dramatically extends animal lifespan, and this effect can be completely suppressed by loss-of-function daf-16 mutations [1]. We found that daf-16 RNAi abolished the lifespan extension phenotype caused by asm-3(ok1744) mutation ( Figure 1D; Table 1, Set #3). To verify this result, we constructed double mutant strain carrying both asm-3(ok1744) and daf-16(mgDf47) null alleles. We found that the lifespan extension phenotype of the asm-3(ok1744) mutants was completely suppressed by the null mutation of daf-16 ( Figure 1E; Table 2, Set #2).
On the other hand, inactivation of daf-18, encoding C. elegans homolog of human PTEN tumor suppressor, is known to suppress lifespan extension and constitutive dauer arrest caused by deficiency in daf-2 or age-1 [14,15]. To test the epistatic relationship between asm-3 and daf-18, we first attempted to construct a double mutant strain carrying both asm-3(ok1744) and daf-18(nr2037) alleles but our effort failed, as these two alleles are located in close proximity on the same chromosome IV region. Instead, we have carried out RNAi experiments in the rrf-3(pk1426);daf-18(nr2037) mutant strain for the epistatic analysis of asm-3 and daf-18. The presence of the rrf-3(pk1426) allele is to enhance the RNAi efficiency. Consequently, we observed that the lifespan extension phenotypes produced by inactivation of asm-1, asm-2 or asm-3 gene ( Figure 1C; Table 1, Set #2) were all suppressed by a null mutation of daf-18, daf-18(nr2037) ( Figure 1F; Table 1, Set #4). Thus, these results show that asm-3 behaves similarly to daf-2 in that the lifespan extension phenotypes caused by asm-3 or daf-2 loss-of-function mutations require daf-16 and daf-18 gene activity.
Genetic Interactions of asm-3 with Major Components of the daf-2/IIS Pathway To investigate how asm-3 regulates lifespan in the daf-2/IIS pathway, we examined genetic interactions between asm-3 and major components of the daf-2/IIS pathway. While partial loss-offunction daf-2(e1370) mutants had dramatically extended lifespan at 20uC [1], we found that the lifespan of daf-2 mutants was not further extended by the loss of asm-3 ( Figure 2A; Table 2, Set #3). In the asm-3(ok1744);daf-2(e1370) double mutant, it is possible that daf-2 signaling was reduced to below threshold levels by the daf-2(e1370) mutation, and thus the asm-3(ok1744) mutation could not further dampen the daf-2 signaling outputs. Thus, this result suggests that asm-3 and daf-2 function in the same pathway. We also examined genetic interaction between asm-3 and age-1. The age-1 gene encodes a homolog of mammalian PI-3 kinase catalytic subunit [26]. Partial loss-of-function age-1(mg305) mutants have a dramatically extended lifespan phenotype [27]. Remarkably, our study showed that asm-3(ok1744);age-1(mg305) double mutants had mean lifespan 67% greater than that of age-1(mg305) single mutants, or 259% greater than that of wild-type animals ( Figure 2B; Table 2, Set #4). Consistently, we also observed a cooperative effect on animal lifespan extension when asm-3 was inactivated in another partial loss-of-function age-1 mutant, age-1(hx546) [26] (data not shown). Subsequently, we analyzed the genetic interaction between asm-3 and aap-1. It has been previously reported that aap-1, encoding a C. elegans homolog of mammalian p85 regulatory subunit of PI-3 kinase, functions in the same pathway as age-1 [28]. While partial loss-of-function aap-1(m889) mutants exhibited longer lifespan phenotype than wild-type animals, silencing of asm-3 in the aap-1(m889) mutants further extended the mean lifespan by 21% ( Figure 2C; Table 1, Set #5). These genetic results indicate that loss of asm-3 cooperates with an age-1 or aap-1 loss-of-function mutation to extend animal lifespan, suggesting that asm-3 potentiates age-1/aap-1 signaling. We also tested asm-3 interaction with pdk-1, which encodes a kinase that acts downstream of AGE-1 but upstream AKT-1/2 in the DAF-2/ IIS pathway [29]. Loss of asm-3 did not further extend the lifespan in the partial loss-of-function pdk-1(sa709) [29] mutant background ( Figure 2D; Table 2, Set #5). This result suggests that asm-3 and pdk-1 likely function in the same pathway. In addition, we tested the genetic interaction of asm-3(ok1744) with a null mutation of akt-1, akt-1(mg306) [30]. We observed that the asm-3(ok1744) mutation shortened the lifespan of the longer lived akt-1(mg306) mutant animals, but the akt-1(mg306) mutation did not seem to affect the lifespan extension phenotype of the asm-3(ok1744) mutant ( Figure 2E; Table 2, Set #6). The fact that the asm-3(ok1744) mutation can suppress the lifespan extension phenotype of the akt-1(mg306) mutant suggests that there may be additional genetic interactions between the gene families of akt and asm. Taken together, these results show that asm-3 plays an important role in the daf-2/IIS pathway to regulate animal lifespan.

asm-3 Regulates Dauer Arrest by Modulating the daf-2/IIS Pathway
In C. elegans, dauer arrest is controlled by a variety of signal transduction pathway [3,10]. Dauer, a hibernation-like state, is normally induced by starvation, high population density or high temperature [3,10]. Mutants with reduced signaling in the daf-2/ IIS pathway form dauer even under favorable growth conditions [3,10]. We investigated whether asm-3 participates in the daf-2/IIS pathway to regulate dauer formation. As previously reported [28,38,39], temperature-sensitive daf-2(e1370) mutants are prone to induction of dauer arrest at 25uC and less so at 22.5uC. Consistent with previous reports, our results showed that the daf-2(e1370) mutants formed dauer constitutively at non-permissive temperature 25uC, but formed dauer less efficiently at semipermissive 22.5uC ( Figure 3A; Table S1). However, the dauer arrest at 22.5uC in daf-2(e1370) mutants was markedly enhanced from 46% to 86% by the presence of the asm-3(ok1744) allele ( Figure 3A; Table S1). In addition, age-1(mg305) mutants possessed a constitutive dauer phenotype at 25uC [27] but rarely displayed dauer arrest phenotype at 22.5uC ( Figure 3B; Table S1). However, loss of asm-3 dramatically enhanced dauer arrest of age-1(mg305) mutants at 22.5uC from 3% to 99% ( Figure 3B; Table S1). On the other hand, loss of asm-3 by itself did not induce dauer arrest at 25uC (Figure 3A, 3B; Table S1). Therefore, these data indicate that the asm-3 gene activity potentiates daf-2 and age-1 signaling to regulate dauer arrest. In comparison, because dauer formation phenotype of the pdk-1(sa709) mutant was so penetrant at 27uC [29] ( Figure 3C; Table S1), we could not detect any effects of asm-3(ok1744) mutation on pdk-1(sa709) at this temperature ( Figure 3C; RNAi treatments, initiated at L1 stage, were continued throughout the assay time. Lifespan assays were performed at 20uC. Each experiment was individually grouped and statistical analyses of the data sets were carried out using the vector (L4440) control as the reference in each group (P value). In Set #3, the second set of statistical analyses was carried out for all the survivor data derived from the asm-3(ok1744);rrf-3(pk1426) strain, using the corresponding vector (L4440) control in this strain (marked as *) as the reference for this subgroup (*P values). Each set of the lifespan experiments was repeated at least two independent times and similar results were obtained. Data from representative sets of experiments are shown. Greater than 50 worms were counted for each RNAi-inducing condition in each experiment. doi:10.1371/journal.pone.0045890.t001 Table S1). Loss of asm-3 did not trigger dauer formation in pdk-1(sa709) mutants at 25uC ( Figure 3C; Table S1). However, for akt-1(mg306) null mutants, which were shown to form dauer efficiently at 27uC but not at 25uC [30,40], loss of asm-3 partially suppressed dauer formation of akt-1(mg306) mutants at 27uC ( Figure 3D; Table S1). The genetic interactions of asm-3 with pdk-1 or akt-1 in dauer formation appear to be consistent with those in lifespan regulation ( Figure 3C, 2D; Figure 3D, 2E). Taken together, our genetic analyses suggest that asm-3 acts in the daf-2/IIS pathway to regulate dauer formation.
In addition to the daf-2/IIS pathway, the daf-7/TGF-b-like signaling pathway is also involved in the regulation of dauer formation. The daf-7 gene encodes a ligand related to mammalian transforming growth factor beta (TGF-b) [41]. The daf-7(e1372) mutants are known to constitutively form dauer at 25uC [41]. We found that, at either 25uC or 22.5uC, loss of asm-3 did not affect the constitutive dauer arrest phenotype of daf-7(e1372) mutants ( Figure 3E; Table S1). These results thus indicate that dauer regulation via the daf-7/TGF-b signaling pathway does not require asm-3 gene activity. These results suggest that the involvement of asm-3 in the daf-2/IIS pathway is specific.Loss of asm Genes Results in Nuclear Localization of DAF- 16 In live animals, the downstream signaling output of the DAF-2/IIS pathway can be assayed by the intracellular localization of DAF-16/FOXO. DAF-16 is normally sequestered in the cytoplasm after phosphorylation by AKT-1/AKT-2, while mutations causing reduced signaling in the daf-2 pathway all lead to translocation of DAF-16 to the nucleus [3,9]. To further investigate whether asm-3 directly regulates the molecular signaling cascade of DAF-2 to DAF-16, we used a strain carrying a daf-16::gfp transgene to monitor the intracellular distribution of DAF-16::GFP fusion protein [9]. We constructed mutant strains of rrf-3(pk1426);daf-16::gfp and asm-3(ok1744);rrf-3(pk1426);daf-16::gfp with the presence of the rrf-3(pk1426) allele to enhance the RNAi efficiency. We found that inactivation of individual asm genes each induced nuclear localization of DAF- All the assays were carried out at 20uC and on regular NGM plates. For Set #4, assays were carried out on plates containing FuDR (50 mg/ml) to avoid the internal hatching of the asm-3(ok1744);age-1(mg305) mutant animals. Each experiment was individually grouped and statistical analyses were carried out for the experiment data set and the wild-type control data set in each group (P values). In addition, statistical analyses were carried out for the survival data of the assayed single mutant (marked as *) and the corresponding double mutant containing the asm-3(ok1744) allele (*P values). In Set #6, P value between asm-3(ok1744) and asm-3(ok1744);akt-1(mg306) is 0.064. Each set of the lifespan experiments was repeated at least two independent times and similar results were obtained. Data from representative sets of experiments are shown. Greater than 50 worms were counted for each strain in each experiment. doi:10.1371/journal.pone.0045890.t002 16::GFP in the rrf-3(pk1426);daf-16::gfp background ( Figure 4A; Figure S5). As expected, age-1, daf-2, or akt-1 RNAi, which were used as positive controls, all resulted in the nuclear localization of DAF-16::GFP ( Figure 4A; Figure S5). In contrast, control (L4440) RNAi and daf-18 RNAi led to a diffused distribution of DAF-16::GFP in cytoplasm, while daf-16 RNAi greatly decreased fluorescence intensity of DAF-16::GFP ( Figure 4A; Figure S5). We also tested the effects of multiple knockdowns of asm genes. We found that combined inactivation of 2 or 3 asm genes caused a further increase in the nuclear localization of DAF-16::GFP in the asm-3(ok1744);rrf-3(pk1426);daf-16::gfp mutant background ( Figure 4B; Figure S6). In addition, the nuclear translocation of DAF-16::GFP in the asm-3(ok1744) mutant was abolished by daf-16 RNAi or daf-18 RNAi ( Figure S6). These results are consistent with the observations that daf-16 and daf-18 gene activities are required for the asm-mediated lifespan regulation ( Figure 1D-1F). Taken together, these results indicate that asm-3 regulates DAF-16 through controlling its intracellular localization.

Loss of asm-3 Promotes the Expression of daf-16 Target Genes
As described above, inactivation of asm-3 triggers nuclear translocation of DAF-16/FOXO and up-regulates endogenous DAF-16 protein levels. However, as previously reported [45][46][47], nuclear translocation of DAF-16/FOXO is not sufficient for activation of the DAF-16/FOXO transcription activity. We therefore examined whether loss of asm-3 indeed results in an increase of activity of daf-16/FOXO transcription factor, by analyzing the expression levels of several known DAF-16/FOXO target genes [48,49]. The sod-3 gene, encoding superoxide dismutase, is directly regulated by DAF-16; and in partial loss-of-function daf-2 mutants, a GFP transgene under control of the sod-3 promoter (sod-3p::gfp), is highly expressed [11]. We therefore examined transcriptional expression of endogenous sod-3 by using qRT-PCR analysis. We observed that sod-3 mRNA levels were increased by about 3-fold in asm-3(ok1744) mutants over the wildtype ( Figure 5A). The expression of another DAF-16-regulated gene, mtl-1 [48], encoding metallothioneins, was also modestly increased in asm-3 mutants ( Figure 5B). These results indicate that asm-3 regulates expression of daf-16 target genes through controlling DAF-16/FOXO transcriptional activity. In addition, we used a strain carrying the sod-3p::gfp transgene to monitor the daf-16/FOXO transcriptional activity. Inactivation of a single asm gene has only a modest effect on the sod-3p:gfp reporter gene expression (data not shown). However, inactivation of 2 or 3 asm genes, achieved through RNAi of asm-1, asm-2 or asm-1 and asm-2 together in the asm-3(ok1744);rrf-3(pk1426);sod-3p::gfp mutant Figure 3. Effects of asm-3 on dauer formation regulation in various mutants defective in the daf-2 signaling. (A) Loss of asm-3 enhanced dauer formation of daf-2(e1370) mutants at the semi-permissive temperature 22.5uC. (B) asm-3 mutation greatly enhanced dauer arrest phenotype of age-1(mg305) mutants at 22.5uC. (C) asm-3 mutation did not affect dauer arrest induced by the pdk-1(sa709) mutation at 27uC. The mutant animals carrying sa709 allele formed dauer at 27uC but not at 25uC. (D) asm-3 mutation partially suppressed the dauer arrest phenotype of akt-1(mg306) mutants at 27uC. No dauers at 25uC were observed for the akt-1(mg306) mutant animals with or without the presence of the asm-3(ok1744) allele. (E) asm-3 mutation had no effect on dauer arrest phenotype of daf-7(e1372) mutants at either 22.5uC or 25uC. The asm-3(ok1744) allele by itself did not induce dauer formation at either 22.5uC or 25uC. Error bars indicate standard deviation from triplicates. Details including total worm numbers used in the assay are listed in Table S1. doi:10.1371/journal.pone.0045890.g003 background, each led to a marked up-regulation of the sod-3p::gfp expression as compared to control vector (L4440) RNAi ( Figure 5C). As negative controls, daf-16 RNAi and daf-18 RNAi each caused a dramatic reduction of the sod-3p::gfp expression when assayed in parallel ( Figure 5C). These data indicate that asm genes regulate the transcriptional output of daf-16.

Chemical Inhibitors of ASM Delay Aging in C. elegans
Our genetic analyses suggest that ASM may be a potential target for anti-aging at the organismal level. It has been reported that mammalian ASM activity can be inhibited by the chemicals desipramine and clomipramine [50][51][52], which are clinically approved drugs for anti-depression. Desipramine and clomipramine have been used for studying the role of human ASM in CD95 death receptor signaling [53,54]. Indeed, we found that these drugs inhibited mammalian ASM activity in cultured cells (data not shown). We then tested whether desipramine and clomipramine had any effects on C. elegans lifespan. Remarkably, we found that treatment with desipramine (30 mM) or clomipramine (5 mM) each extended the lifespan of wild-type animals, resulting in a mean lifespan 24% or 14% greater than that of control animals, respectively (Figure 6A, 6B; Table 3, Set #1, #2). Furthermore, the effects of these drugs on lifespan extension were dependent on daf-16 gene activity, as no lifespan extension was observed in daf-16(mgDf47) null mutants ( Figure 6C, 6D; Table 3, Set #3, #4). These results are consistent with the data shown in Figure 1. To verify that these drugs indeed target CeASMs, we performed CeASMs activity assays using lysates prepared from drug-treated or vehicle-treated animals. We found that treatment with desipramine or clomipramine each decreased CeASM activity by 78% or 77% as compared to vehicle control, respectively ( Figure 6E). This result indicates that these chemicals effectively inhibit CeASMs. Our studies thus demonstrate that inhibition of CeASM activity by chemical compounds extends animal lifespan to an extent similar to that achieved by inactivating the asm genes.

asm-3 Functions in the daf-2/IIS Pathway to Regulate Animal Lifespan and Dauer Formation
In this study, we have shown that asm-3 regulates both animal lifespan and dauer formation in C. elegans. Further, several lines of novel evidence suggest that asm-3 functions in the daf-2/IIS pathway. First, the ability of asm-3 deficiency to extend animal lifespan is suppressed by loss-of-function mutations in either daf-16 or daf-18, two well-known negative regulators of the daf-2 pathway ( Figure 1C-1F). Second, the asm-3(ok1744) loss-of-function allele does not further extend the lifespan of daf-2(e1370) or pdk-1(sa709) loss-of-function mutants (Figure 2A, 2D), placing asm-3 in the daf-2 and pdk-1 pathway. It is possible that the daf-2(e1370) or pdk-1(sa709) mutation has already sufficiently reduced the gene activity of daf-2 or pdk-1 to below the threshold levels, and thus the presence of the asm-3(ok1744) allele in these mutants cannot further dampen daf-2 signaling output. Third, asm-3 deficiency strongly enhances the lifespan extension phenotype of the age-1(mg305) or aap-1(m889) mutants ( Figure 2B, 2C). As age-1(mg305) and aap-1(m889) mutations are partial loss-of-function alleles, it is possible that the daf-2 signaling strength can be further reduced in these strains by the asm-3(ok1744) mutation, resulting in an increase in lifespan. Fourth, inactivation of asm-3, either alone or in combination with inactivation of asm-1 or asm-2 paralog, induces the nuclear translocation of DAF-16::GFP fusion protein, in a result similar to that achieved by inactivation of daf-2, age-1 or akt-1 ( Figure 4A, 4B; Figure S5, S6). Overall, these observations suggest that asm-3 normally functions as a positive regulator of daf-2 signaling.
Of particular interest are the observations that the effects of asm-3 inactivation in age-1 or aap-1 mutant backgrounds are very strong. The asm-3(ok1744) allele can further increase the mean lifespan of the age-1(mg305) mutants by 67%, and asm-3(ok1744);age-1(mg305) double mutants have a mean lifespan 259% greater than that of wild-type animals ( Figure 2B). Inactivation of asm-3 by RNAi also potently extends the lifespan of aap-1(m889) mutants ( Figure 2C). Consistently, asm-3(ok1744) enhanced the dauer formation efficiency of age-1(mg305) mutants from 3% to 99% at an intermediate temperature of 22.5uC ( Figure 3B). Genetically, these studies suggest that asm-3 acts in parallel to age-1 or aap-1, in the daf-2/IIS pathway (see model in Figure 7A). Since both age-1(mg305) and aap-1(m889) are partial loss-of-function alleles, it is also possible that asm-3 normally functions together with age-1 or aap-1, through potentiating age-1 and aap-1 gene activities (see model in Figure 7B).  Table 3. (E) Desipramine and clomipramine each inhibited CeASM activity. Synchronized L1 animals were treated with desipramine, clomipramine or vehicle control (see Material and Methods for details). Lysates were prepared from L4 animals and assays were performed in duplicates using [ 14 C]-sphingomyelin. CeASM activities from all samples were normalized against that in the vehicle control (100%). Desipramine or clomipramine treatment each decreased CeASM activity by 78% or 77%, respectively (T-test *P,0.001 for desipramine and **P,0.001 clomipramine vs. vehicle control, respectively). doi:10.1371/journal.pone.0045890.g006 Based on our genetic data on both lifespan and dauer regulation, we propose a model in which asm-3, either functioning in parallel of age-1/aap-1 ( Figure 7A) or acting together with age-1/aap-1 ( Figure 7B), activates pdk-1 and then akt-1, culminating in the suppression of daf-16 to regulate animal lifespan and dauer arrest.

asm-3 Regulates the Localization, Transcriptional Activity, and Protein Expression Level of DAF-16/FOXO
We provide several lines of evidence suggesting that asm-3 modulates daf-2 signaling through affecting the DAF-16 forkhead transcription factor. We have shown that DAF-16::GFP fusion protein is translocated into the nucleus when asm-3 is inactivated by the asm-3(ok1744) mutation or by RNAi ( Figure 4A, 4B; Figure  S5). Stronger effects have been observed when asm-3 is co-inactivated with asm-1 or asm-2 ( Figure 4B; Figure S6). Nuclear translocation of DAF-16::GFP fusion protein has also been observed when daf-2, age-1 or akt-1 is inactivated by RNAi ( Figure 4A; Figure S5). Consistently, qRT-PCR analysis has shown that several known DAF-16 downstream target genes, including sod-3, are also up-regulated when asm-3 or daf-2 is inactivated by mutations ( Figure 5A). Together, our data indicate that asm-3 gene activity regulates the sub-cellular localization and transcriptional activity of DAF-16/FOXO in a manner similar to daf-2.
In addition, we have found that DAF-16 protein levels are upregulated not only in the asm-3(ok1744) mutants and in animals treated with asm-3(RNAi), but also in the daf-2(e1370) mutants or daf-2(RNAi) animals ( Figure 4C, 4D). It is possible that an additional role of DAF-2 in regulation of DAF-16 is mediated through the regulation of DAF-16 protein levels, and that this  process also involves ASM-3. Up-regulation of DAF-16 protein levels has been previously observed in rle-1 mutants and eak-7 mutants, both known to have a longer animal lifespan [12,40]. Future studies are needed to determine how DAF-2 and ASM-3 regulate DAF-16 protein levels.

A Potential Mechanism of ASM-3 in the Regulation of Trans-membrane Receptor Signaling
The predicted C. elegans ASM-3 peptide shares a strong homology with the human ASM in the presumed catalytic domain of the ASM enzyme ( Figure 1S). In cultured human cells, ASMdependant ceramide production in lipid rafts has been shown to be essential for CD95 death receptor mediated signaling to apoptosis [18,19,53]. However, the endogenous CD95 receptors have recently been found to have a pro-survival and pro-proliferation function in the mouse knockout system [20,21]. It appears that the biological consequence of CD95 receptor signaling is dependent on cell context and stimulus used. It remains to be investigated whether the pro-proliferation and pro-survival signaling by the CD95 receptors requires the participation of ASM.
Our studies have provided strong genetic evidence that ASM-3 is a positive regulator of signaling through the DAF-2 receptors. This is the first report that ASM functions in vivo as a positive regulator for a receptor tyrosine kinase. By analogy with human ASM in the formation of ceramide-enriched lipid rafts, it is possible that similar lipid rafts may be involved in facilitating the DAF-2 receptor signaling in C. elegans. In light of the fact that the strongest phenotypes of asm-3 inactivation are observed in age-1 mutant backgrounds, it can be further speculated that transducing the signals from DAF-2 receptor to the AGE-1 PI 3-kinase homolog may involve ASM-3dependent lipid rafts. In support of this hypothesis, we have observed that, in mammalian systems, human ASM regulates the lipid raft localization of the IGF-1 receptors (P. Ghosh, Y. Kim, X. Xiong and H. Sun, unpublished observations). Taken together, our studies indicate that ASM-3 proteins, along with their products ceramides, play a positive role in facilitating DAF-2 receptor signaling.

asm-3 Serves as a Molecular Target for Anti-aging
Our genetic studies have shown that asm-3 regulates animal lifespan through modulation of the daf-2/IIS pathway. These results suggest the potential of using ASM as a molecular target for anti-aging at organismal level. Indeed, we have shown that two chemical compounds, desipramine and clomipramine, which are known to inhibit mammalian ASM activity [50][51][52], can effectively extend wild-type animal lifespan by 24% or 14%, respectively ( Figure 6A, 6B). Such lifespan extension phenotypes require the presence of daf-16, as no lifespan extension was observed in the daf-16(mgDf47) null mutant background ( Figure 6C, 6D). The effects of these compounds on lifespan are very similar to the lifespan extension resulting from genetic inactivation of the asm gene family (Figure 1). Consistently, our biochemical studies have demonstrated that these compounds indeed greatly decrease the CeASM activities ( Figure 6E). Thus, these data suggest that these two drugs mediate their effects through inhibition of CeASMs. Together, these results further underscore the importance of ASM in the regulation of animal lifespan and its role as a potential target for anti-aging. Moreover, since mammalian ASM is known to localize to the outer leaflet of the plasma membrane [16], ASM presents itself as an even more attractive target for anti-aging drug development.

Lifespan Assay on NGM Agar Plates
Lifespan assays were carried out on NGM plates at 20uC, and plates were seeded with E. coli OP50. In order to synchronize worms, eggs were first isolated from gravid adults and then hatched overnight in the S1 basal media. The synchronized L1 larva were then placed on OP50-seeded NGM agar plates and allowed to grow to adult. During the egg-laying period, adult worms were daily transferred to new OP50-seeded agar plates. Worms were then examined every other day for survival and scored. Where indicated, FuDR was also included in the plates. FuDR, 5-Fluoro-29-deoxyuridine, is used to inhibit DNA synthesis and thus to prevent the production of progeny. Animals hatched from egg preparation were grown on regular OP50-seeded NGM agar plates until L4 stage and then transferred to plates containing 50 mg/ml FuDR. Animal survival was scored every other day and transferred to new plates containing 50 mg/ml FuDR every 2,3 days. Animals were considered dead when failed to respond to gentle touches by a platinum wire. Adult lifespan were counted using the L4 stage as Day 0. For each set of experiments, animals from different strains or conditions were assayed in parallel. Each set of experiments was carried out at least 2 times. In all the lifespan assays performed on regular NGM plates, RNAi-inducing plates or drug-containing plates, at least 50 worms were counted per strain per experiment; while the animals that were bagged, exploded or crawled off the plate were censored from the counts.

Lifespan Assay on RNAi-inducing Plates
RNAi bacteria culture was grown in Luria broth media with 100 mg/ml ampicillin at 37uC for 16,18 hours and then seeded onto NGM plates containing 5 mM IPTG and 50 mg/ml carbanicillin. In case of double knockdown, RNAi bacteria cultures of same density were mixed at 1:1 ratio and were then seeded on RNAi-inducing plates. The plates were kept at room temperature overnight to induce the expression of the RNAi molecules. Lifespan assay was previously described [25]. Briefly, the synchronized L1 larvae animals prepared by egg preparation were plated on the RNAi-inducing plates and allowed to grow to adults. The adult animals were transferred to new plates every day until adult animals stop laying eggs and after that the animals were scored every other day and transferred to new plates every 2,3 days. All the RNAi experiments were carried out in the rrf-3(pk1426) background, a strain known to be hypersensitive to RNAi, with the exception of RNAi experiment in the aap-1(m889) background carrying the rrf-3(+) allele. Lifespan assays were carried out at 20uC. Each set of experiments were conducted at least 2 times.

Lifespan Assay on Drug-containing NGM Plates
To prepare NGM plates containing desipramine or clomipramine, drugs were first dissolved in water as 100 mM stock solutions and sterilized by filtration. Drugs were then properly diluted to the indicated final concentration during the solidification step of the NGM agar plate preparation. The synchronized L1 larvae prepared by egg preparation were placed on NGM agar plates and allowed to develop to L4 larva. Synchronized L4 animals were then transferred to regular NGM (control) plates or plates containing drug desipramine (30 mM) or clomipramine (5 mM). Animals were then transferred to drug-containing plates or regular NGM plates every day until animals stop laying eggs. The populations were scored every other day and transferred to new drug-containing plates or regular NGM plates every 2,3 days. Adult lifespan was counted using the L4 stage as Day 0. Lifespan assays were carried out at 20uC. Drug inhibition experiments have been carried out .3 times and consistent results have been obtained from independent experiments.

Statistical Analysis
Statistical analyses of all the survival curves were performed as described [11] using the software GraphPad 5.0. Survival curve of each population was compared with control using Log-rank (Mantel-Cox) test and an experimental data set with a P value ,0.05 was considered to be significantly different from the control data set.

Oxidative Stress Response Assay using Paraquat
The synchronized L1 animals after standard egg preparation were placed on NGM agar plates seeded with E. coli OP50 and allowed to develop to the L4 stage larva at 20uC. A 1-mL 80 mM paraquat solution was then treated on the 6-cm plates at the L4 stage and after soaking for about 20 minutes, the plates were airdried in a chemical hood till plates were dried. The dried plates were shifted to 20uC incubator and survivor of the animals was scored every day. The experiment was repeated at least two independent times. Total worms counted per plate were greater than 100.

Heat Stress Response Assay at 35uC
The synchronized L1 animals after egg preparation were put on NGM agar plates seeded with E. coli OP50 and allowed to grow to Day 1 stage worms (young adult worms) at 20uC. At Day 1, the plates were shifted to 35uC and then scored after 6, 9, 12 or 15 hours afterwards. The plates used for counting in each time point were discarded to avoid the complication of recovery at room temperature on scoring plates. Experiments were conducted for two times, each in triplicates, and the standard deviations were calculated. The number of worms counted per strain per plate was greater than 200.

Dauer Assay
All the strains used for dauer assay were synchronized by egg preparation and the synchronized L1 larvae were placed on NGM agar plates seeded with E. coli OP50. The plates were incubated at 22.5uC, 25uC or 27uC. Numbers of dauer and nondauer in the population were visually scored after 72, 48 and 45 hours later for the assays carried out at 22.5uC, 25uC or 27uC, respectively. All assays were performed in triplicates and experiments were conducted at least 2 times.

RNA Isolation and qRT-PCR
Total RNA was isolated from synchronized adult Day 1 worms using Trizol reagent (Invitrogen). RNA purity was checked by UV absorbance (260/280 ratio). cDNA was synthesized with primer Oligo(dT) 20 by using SuperScript III First-Strand Kit (Invitrogen) according to the manufacturer's protocol. RT-PCR reactions were performed with a 20 ml of Power SYBR PCR Master Mix (Applied Biosystems) using triplicates for each sample. PCR reaction was carried out on Real Time PCR Machine 7500 Fast (Applied Biosystems) under the condition of 95uC for 5 minutes for denaturation, followed by 40 cycles of 95uC for 15 seconds and 60uC for 1 hour. The reaction products were analyzed using software provided by the onboard software from the Real Time PCR machine. The primers used for qRT-PCR are as follows: Primer for act-1 (133 bp in product size) [34,39]; Forward primer: 59-ccaggaattgctgatcgtatgcagaa-39; Reverse primer: 59-tggagagggaagcgaggataga-39: Primer for sod-3 (98 bp in product size) [34,39]; Forward primer: 59-tcgcactgcttcaaagcttgttcaa-39; Reverse primer: 59-ccaaatctgcatagtcgaatgggagat-39: Primer for mtl-1 (111 bp in product size); Forward primer: 59-atggcttgcaagtgtgactg-39; Reverse primer: 59-tttctcactggcctcctcac-39. act-1 was used as an internal control for normalization of input RNA levels. All experiments have been repeated .2 times and consistent results have been obtained from independent experiments.

Preparation of Worm lysate and Western Blot Analysis
Synchronized young adult worms (at adult Day 1) were rinsed three times with M9 buffer to remove bacteria and then washed one time with ddH 2 O to remove salt. 0.4 volume of 5X SDS-PAGE sample buffer (312.5 mM Tris-HCl, pH 6.8, 50% Glycerol and 10% SDS) was added to each tube containing the packed worms. Tubes were subjected to 3 cycles of freezing/heating (CO 2 /ethanol bath and boiling at 95uC for 10 minutes). Samples were then sonicated using water bath sonicator (VWR B1500A-MT) for 3 cycles of 2 minutes per cycle. Lysates were clarified by centrifugation at 12000 rpm for 1 minute and supernatants were transferred to new tubes, and protein concentrations were quantified using BCA protein assay reagent (Thermo Scientific). For experiments involving RNAi-treated samples, adult Day 2 worms from plates seeded with RNAi-inducing bacteria were used. Equal amounts of protein lysates were analyzed by SDS-PAGE analysis, and the proteins were transferred to nitrocellulose membrane. Membranes were blocked with 1X TBST (150 mM NaCl, 20 mM Tris-base, pH 7.4 and 0.05% Tween-20) containing 5% non-fat dry milk for 1 hour at room temperature. Primary antibody were applied and incubated at 4uC overnight. Filters were extensively washed, and then incubated with a properly diluted secondary antibody for 1 hour at room temperature. Following washes, filters were developed with Western lightning Plus-ECL (Perkin Elmer) and exposed to X-ray films. The primary antibodies used were antibodies specific for DAF-16 (

Acid Sphingomyelinase Assay
Synchronized wild type L1 larvae animals prepared by egg preparation were placed on regular NGM plates (control) or plates containing drug desipramine (30 mM) or clomipramine (5 mM) and allowed to develop to L3 larva for 30 hours. At L3 stage, worms were further treated with either ddH 2 O, desipramine (30 mM) or clomipramine (20 mM) by adding 500 ml of the corresponding solution to each plate. The plates were dried for 20 minutes in air-ventilated hood and then returned to the 20uC incubator for 18 hours. At L4 stage, worms were harvested using M9 buffer. Worms were washed three times with M9 buffer followed by two times washes with ddH 2 O. To each tube of packed worm (,800 ml), 500 ml of cold lysis buffer (0.2% Triton X-100, 50 mM Tris-HCl pH 7.4, 50 mM sodium acetate, 1 mM EDTA and protease inhibitors) was added and samples were incubated on ice for 30 minutes. Samples were then sonicated using a water bath sonicator (Branson 1510) for 20 seconds then returned to ice for 1 minute and the procedure was repeated 5 times. Lysates were clarified by centrifugation at 8006g for 5 minutes at 4uC. The clarified lysates were measured for protein concentration using Bio-Rad protein assay kit and then normalized. CeASM activity assays were measured according to the procedures described by (24). [ 14 C]-sphingomyelin (Amersham Life Sciences, supplied in organic solvent) was air-dried then reconstituted in the assay buffer (250 mM sodium acetate pH5.0, 1 mM EDTA and 0.1% Triton X-100) and briefly sonicated in water bath sonicator. The assay was initiated by combining 200 ml of lysate with 200 ml assay buffer containing [ 14 C]-sphingomyelin (25 nCi/sample) and then incubated at 37uC for 40 minutes. Reaction mixtures were subsequently extracted with 800 ml chloroform/methanol mixture (2:1 v/v). After centrifugation at 50006g for 5 minutes, an aliquot of the aqueous phase (600 ml) were removed and counted by liquid scintillation counter (Beckman Coulter LS 6500). All samples were assayed in duplicate, and the lysis buffer alones were served as negative controls.

Quantification of DAF-16 Protein Levels
The relative abundance of DAF-16 protein levels were calculated from the western blots using the ImageJ software. The intensity of DAF-16 was normalized against beta-actin, and the numbers were presented as a percentage relative to the N2 or the vector control. The controls were set as 100%. Figure S1 Multiple alignment of C. elegans ASM-3, H. sapiens ASM and D. melanogaster ASM. The C-terminal catalytic domain of CeASM-3 was aligned with that of HsASM and DmASM. Sequences used in the alignment correspond to amino acids 117 to 464 of CeASM-3, 186 to 523 of HsASM and 254 to 596 of DmASM. Identical residues are shaded black and similar residues are shaded gray. CeASM-3 shares 42% identity with HsASM and 39% identity with DmASM. Alignment was performed using Clustal W2 and BioEdit v7.0.5. Consensus symbol ''*'' is used when the residues in all sequences are identical. Symbol '':'' or ''?'' is used for conserved substitutions or semiconserved substitutions, respectively. (TIF) Figure S2 Genomic structures of asm-3 and ok1744 allele. The genomic arrangement of asm-3 is illustrated according to the Worm database (www.wormbase.org). The ok1744 allele contains a deletion of 1558 bp which removes exon VII through exon XIII of asm-3, followed by a 7 bp insertion. As a consequence, such changes shall lead to production of a peptide that is truncated at the amino acid 272 (Pro272). (TIF) Figure S3 Multiple alignment of C. elegans ASM-3, ASM-1 and ASM-2. The predicted C-terminal catalytic domain of CeASM-3 was aligned with that of CeASM-1 and CeASM-2. Sequences used in the alignment correspond to amino acids 117 to 464 of CeASM-3, 138 to 478 of CeASM-1 and 167 to 526 of CeASM-2. Identical or similar residues are shaded black or gray, respectively. CeASM-3 shares 39% identity with CeASM-1 and 41% identity with CeASM-2. Alignment was performed using Clustal W2 and BioEdit v7.0.5. Consensus symbol ''*'' is used when the residues in all sequences are identical. Symbol '':'' or ''?'' is used for conserved substitutions or semi-conserved substitutions, respectively. (TIF) Figure S4 Stress responses. (A) Loss of asm-3 resulted in an increase in resistance towards oxidative stress using paraquat treatment (80 mM). (B) Loss of asm-3 promoted resistance towards heat stress (35C). In (A) and (B), daf-16(mgDf47) mutants were sensitive to oxidative stress or heat stress. Increased resistance of the asm-3(ok1744) mutants was dependent on the daf-16 gene activity. For all stress response assays, experiments were repeated at least two independent times with similar results obtained. Data from representative sets of experiments are shown. For the oxidative stress assays, greater than 100 worms were counted for each strain in each experiment. For the heat stress assays, greater 200 worms per plate per strain were counted and experiments were conducted in triplicates. Error bars indicate standard deviation. (TIF) Figure S5 Examination of DAF-16::GFP subcellular localization. The subcellular localization of DAF-16::GFP fusion protein was examined by fluorescence microscopy. rrf-3(pk1426);daf-16::gfp mutants were treated with the indicated RNAi molecules and examined on adult day 4. For the daf-2 and akt-1 RNAi, animals were examined on adult day 1 and day 4. In the vector alone (L4440) control animals, DAF-16::GFP proteins were localized only in the cytoplasm. In animals treated with RNAi molecules for asm-3, asm-1 or asm-2, DAF-16::GFP proteins were localized mostly in the nucleus. As positive controls, RNAi of daf-2, age-1 or akt-1 each induced the nuclear localization of DAF-16::GFP proteins. RNAi knockdown of daf-16, a negative control, markedly reduced DAF-16::GFP expression. In daf-18 RNAi-treated animals, no nuclear translocation of DAF-16::GFP was observed. (TIF) Figure S6 Synergistic effect of DAF-16::GFP nuclear localization. DAF-16::GFP cellular distributions were examined by fluorescence microscopy. asm-3(ok1744);rrf-3(pk1426);daf-16::gfp mutant animals were examined on adult day 4. In the vector alone treated animals, DAF-16::GFP proteins were partially nuclearlocalized due to the presence of asm-3(ok1744) allele. Further inactivation of asm-1, asm-2, or asm-1 together with asm-2 (asm-1/ asm-2) by RNAi induced more pronounced nuclear localization of DAF-16::GFP. RNAi inactivation of daf-16, a negative control, markedly reduced DAF-16::GFP expression. The daf-18 RNAi prevented the nuclear localization of DAF-16::GFP proteins caused by loss of asm-3.

(TIF)
Table S1 Effects of asm-3 on dauer formation. Dauer formation experiments were carried out for the wild-type animals and various mutants at the indicated temperatures. Dauer formation assays were conducted in triplicates and experiments were repeated at least two times. Data from representative experiments are shown. The average percentages of nondauer and dauer, as well total numbers of animals used for each assay, were listed. (PDF)