The cell non-autonomous function of ATG-18 is essential for neuroendocrine regulation of Caenorhabditis elegans lifespan

Dietary restriction (DR) and reduced insulin growth factor (IGF) signaling extend lifespan in Caenorhabditis elegans and other eukaryotic organisms. Autophagy, an evolutionarily conserved lysosomal degradation pathway, has emerged as a central pathway regulated by various longevity signals including DR and IGF signaling in promoting longevity in a variety of eukaryotic organisms. However, the mechanism remains unclear. Here we show that the autophagy protein ATG-18 acts cell non-autonomously in neuronal and intestinal tissues to maintain C. elegans wildtype lifespan and to respond to DR and IGF-mediated longevity signaling. Moreover, ATG-18 activity in chemosensory neurons that are involved in food detection sufficiently mediates the effect of these longevity pathways. Additionally, ATG-18-mediated cell non-autonomous signaling depends on the release of neurotransmitters and neuropeptides. Interestingly, our data suggest that neuronal and intestinal ATG-18 acts in parallel and converges on unidentified neurons that secrete neuropeptides to regulate C. elegans lifespan through the transcription factor DAF-16/FOXO in response to reduced IGF signaling.


Author summary
Restriction of dietary food without malnutrition robustly extends lifespan in more than twenty species. It was also reported that fruit flies (Drosophila melanogaster) deficient in olfactory function live longer and that the longevity induced by food restriction is partially due to decreased olfaction. These observations suggest food assimilation through the gastrointestinal tract and food smell detected by olfactory neurons influence lifespan. The insulin growth factor signaling pathway is regulated by nutrient levels and has been shown to mediate the lifespan extension conferred by food restriction and defective gustatory neurons in the nematode Caenorhabditis elegans. However, the mechanism remains unclear. Autophagy is a lysosomal degradation pathway and is sensitive to nutrient availability. We found autophagy activity in the intestine and food sensory neurons acts in parallel to mediate food restriction and insulin signaling regulated lifespan extension in Caenorhabditis elegans. Moreover, intestinal and neuronal autophagy converges on unidentified neurons to control the secretion of neuropeptides that regulate lifespan. These data suggest autophagy

Introduction
Limitation of food without causing malnutrition (DR: dietary restriction) extends lifespan in a wide range of species [1]. In C. elegans, multiple signaling transduction pathways are involved in response to DR, including the insulin-like growth factor pathway (IGF) [1]. The C. elegans daf-2 gene encodes an insulin receptor-like tyrosine kinase [2]. DAF-16, a FOXO (forkhead box O) transcription factor, is a major target of DAF-2 signaling and its activity is required for the longevity of daf-2 mutants [3,4]. When DAF-2 signaling is inhibited, DAF-16 enters the nucleus and turns on the transcription of many longevity genes to increase lifespan [5]. Under certain DR conditions DAF-2 signaling mediates the response to DR [6]. However, the IGF and DR pathways can be separated as DAF-2 is not required for lifespan extension of C. elegans under all DR conditions [7,8]. DAF-2 signaling has also been linked to lifespan extension resulted from defective sensory neurons [9]. C. elegans use amphid sensory neurons to detect environmental cues. The worm has one pair of thermosensory neurons and eleven pairs of bilaterally symmetric chemosensory neurons. The chemosensory neurons can detect a variety of olfactory and gustatory signals through their sensory cilia that penetrate the cuticle and are exposed to the external environment [10,11]. These neurons allow worms to process and transmit environmental cues to neuronal or nonneuronal endocrine cells that can modulate hormonal signaling to adjust physiology to maintain homeostasis. Interestingly, several studies have shown that chemosensory neurons regulate lifespan in C. elegans. Mutations that disrupt sensory cilia or laser ablation of their support cells, resulting in defective sensory perception, were able to increase lifespan [12]. Laser ablation of either ASI or ASG gustatory neurons or AWA or AWC olfactory neurons alone extends lifespan [13]. Moreover, the olfactory and gustatory neurons act in parallel to control lifespan [13]. Lifespan extension due to ablating gustatory neurons but not olfactory neurons is dependent on DAF-16 and ablation of gustatory neurons cannot further increase the lifespan of daf-2 mutants, indicating gustatory neurons function through DAF-2 [13]. Similar to C. elegans, in the fruit fly Drosophila melanogaster loss of olfactory function increases lifespan. Furthermore, DR was less effective when olfactory mutant flies were exposed to food odors, suggesting DR longevity is partially due to decreased olfaction [14]. Despite these observations, the molecular mechanism by which environmental food cues and food nutrients coordinate to influence lifespan remains elusive.
Macroautophagy (hereafter autophagy) has emerged as a central pathway regulating aging in various model organisms including C. elegans and is essential for both DR and IGF signaling induced longevity [15][16][17]. Autophagy is highly sensitive to food availability and strongly induced by starvation. Upon induction, double-membrane vesicles referred to as autophagosomes form, engulf cellular components and then fuse with lysosomes allowing the sequestered contents to be degraded. The degraded products are recycled to synthesize new macromolecules and maintain cellular energy homeostasis [18,19].
The autophagosome is derived from the elongation and closure of a membrane precursor called the phagophore [19]. ATG-18 is a member of the WIPI (WD repeat protein interacting with phosphoinositides) protein family that binds to phosphatidylinositol 3-phosphate to promote formation and elongation of the phagophore [20]. WIPI1 and WIPI2 are mammalian Atg18 homologs and both are protein components of autophagosomes [21,22]. Autophagy is suppressed in the loss-of-function C. elegans null mutant atg-18(gk378), as indicated by the accumulation of p62/sequestosome 1 (SQSTM1) (a cargo receptor for autophagic degradation of ubiquitinated protein aggregates) [23], the inefficient clearance of cell corpses [24,25], exacerbation of polyglutamine protein toxicity [26] and decreased lifespan [27]. Interestingly, mammalian WIPI proteins are also involved in the pathogenesis of age-related human diseases such as cancer and neurodegenerative disorders [28], implying that the function of ATG-18 is evolutionarily conserved from C. elegans to humans, which highlights the importance of understanding the mechanisms by which ATG-18/WIPI proteins regulate the aging process.
Here we show that autophagy protein ATG-18 acts cell non-autonomously in the neurons and intestine to maintain C. elegans wildtype lifespan and respond to DR and DAF-2 longevity signals. Furthermore, ATG-18 activity in chemosensory neurons is sufficient to mediate the effect of IGF signaling and DR on lifespan. In addition, we found that neuronal and intestinal ATG-18 converges on unidentified neurons to control C. elegans lifespan through neuropeptide signals.

Tissue-specific requirement of ATG-18 in maintenance of wildtype lifespan
We examined the tissue requirement of atg-18 for regulating C. elegans wildtype lifespan. ATG-18 is expressed in almost every tissue including the body wall muscles, hypodermis, neurons and intestine [29]. As the atg-18 mutant is short-lived [27], we used a complementation approach to express atg-18 in different tissues to examine where autophagy is required for maintenance of a wildtype lifespan. We found the full-length genomic DNA of atg-18 whose expression is driven by its native promoter (Patg-18::atg-18, Fig 1A, S1 Table), a pan-neuronal promoter (Punc-119::atg-18, Fig 1B) and an intestine-specific promoter (Pges-1::atg-18, Fig  1C) fully restores the lifespan of atg-18 mutants to that of wildtype N2 worms. Expression of atg-18 in body wall muscles (Pmyo-3::atg-18, Fig 1D) modestly rescues the short lifespan of atg-18 mutants. However, hypodermal expression of atg-18 (Pdpy-7::atg-18, Fig 1E) has no effect on wildtype lifespan. These results indicate that autophagy activity in neurons and intestinal cells is essential for maintenance of wildtype lifespan. We also expressed atg-18 in the tissues where it is natively expressed in N2 animals and found that this overexpression of atg-18 decreases the wildtype lifespan (S1 Fig, S2 Table).
Tissue-specific requirement of ATG-18 in response to dietary restriction Autophagy genes are required for DR-mediated lifespan extension in C. elegans [30,31]. We examined the tissue requirement of atg-18 in response to solid DR (sDR: bacteria are serially diluted on plates) [32]. Similar to previous studies [32], the lifespan of N2 worms treated by sDR (bacteria concentration: 5X10 9 cells/mL) is significantly increased compared to control animals fed ad libitum (AL: bacteria concentration: 5X10 11 cells/mL) (Fig 2A, S3 Table) (P<0.0001, log-rank test). However, the atg-18(gk378) mutation completely abrogates the response to sDR (Fig 2B). We found the lifespan of atg-18 worms with atg-18 rescued in either neurons or intestinal cells is significantly increased after sDR treatment (Fig 2C and 2D). However, expression of atg-18 in the body wall muscles or hypodermis fails to restore the DR longevity (Fig 2E and 2F). Thus, atg-18 expression in neurons and intestinal cells plays a major role in the DR response.

Influence of tissue-specific expression of atg-18 on autophagy activity
We examined autophagy activity in the daf-2 mutant background and in sDR-treated animals. The GFP::LGG-1 reporter was used to visualize autophagosomes that appear as fluorescence puncta in hypodermal seam cells [33,35]. We found that the number of autophagosomes in   Similar to observations about autophagy activity in daf-2;atg-18 mutants, we found that sDR induces autophagy in wildtype N2 animals and this autophagy induction is suppressed in atg-18 mutants as judged by the number of autophagosomes in hypodermal seam cells (Fig 4B, S6  Table, S4I Fig, S4J Fig, S4K Fig, S4L Fig). Although expression of atg-18 in neurons, intestinal cells and muscles cannot restore autophagy induction in the seam cells of atg-18 mutants upon sDR, autophagy is induced in the seam cells of atg-18 mutants carrying hypodermally expressed atg-18 ( Fig 4B, S6 Table, S4M Fig, S4N Fig, S4O Fig, S4P Fig, S4Q Fig, S4R Fig, S4S Fig, S4T  Fig). These data suggest that tissue-specific expression of atg-18 in the atg-18 mutant background can only restore autophagy in the tissues where the atg-18 gene is expressed. Thus, under this condition ATG-18 in specific tissues should generate a cell non-autonomous signal to influence C. elegans lifespan instead of relying on autophagy induction in other tissues.  Table). We conclude that atg-18 activity in ADF, ADL, ASG and AWA neurons is essential to mediate the effect of IGF signaling on lifespan. We expressed atg-18 in ASG (Podr-2::atg-18), AWA (Podr-10::atg-18), ADL (Psrh-220::atg-18) and ADF (Ptph-1::atg-18) neurons individually and found it significantly increases daf-2;atg-18 lifespan (Figs 5B-4E). Notably, ATG-18 in ASG gustatory neurons fully restores daf-2(e1370) longevity in daf-2;atg-18 worms.
To confirm the expression of atg-18 in individual chemosensory neurons, we inserted the gfp coding sequence upstream of atg-18 genomic DNA in the same plasmids used for the individual neuronal rescue lifespan experiments. The expression pattern of these gfp reporter genes is consistent with previous publications. Two promoters, odr-10 and srh-220, drive the  [36]. The Ptph-1:: gfp::atg-18 reporter gene is expressed in serotonergic neurons including ADF, NSM and HSN neurons as reported previously (S6D Fig, S6E Fig) [37]. For the odr-2 promoter, we used the same promoter reported by Chou et al. [38]. This odr-2(2b) promoter is active in AIZ, AIB, AVG, RIF, PVP, and RIV interneurons, SIAV motor neurons, and IL2 and ASG sensory neurons [38]. The expression the Podr-2::gfp::atg-18 reporter gene was observed in six neurons posterior to the nerve ring. Based on their relative positions in the head, three neurons (ASG, AIB and AIZ) are labelled in S6A Fig. A similar expression pattern has been reported previously [38]. As expression of atg-18 in the chemosensory neurons under control of the gpa-3 promoter fully rescues the short lifespan of daf-2;atg-18 mutants (Fig 5A), we conclude that it is the expression of atg-18 in ASG neurons not the interneurons driven by the odr-2 promoter that mediates the influence of daf-2 signaling on the lifespan. Taken together, these data indicate that expression of atg-18 in four chemosensory neurons is essential to mediate the effect of daf-2 insulin-like signaling on C. elegans lifespan.
Next, we expressed atg-18 in the chemosensory neurons of atg-18 mutants under control of the gpa-3 promoter. We found that the Pgpa-3::atg-18 transgene only partially rescues the short lifespan of atg-18 mutants (S7 Fig, S8 Table), which is in contrast to the observation that the Pgpa-3::atg-18 transgene fully restores the lifespan of daf-2;atg-18 mutants to the daf-2 mutant level ( Fig 5A). Thus, expression of atg-18 in chemosensory neurons plays a more significant role in regulation of the daf-2 mutant lifespan than that of atg-18 mutants. Interestingly, atg-18 mutants carrying an atg-18 transgene in chemosensory neurons (Pgpa-3::atg-18) show increased lifespans after sDR (S8 Fig, S9 Table). Hence, ATG-18 in chemosensory neurons also influences the aging process regulated by DR in C. elegans.

ATG-18-mediated cell non-autonomous signal depends on the release of neurotransmitters and neuropeptides
We investigated whether neurotransmitters and/or neuropeptides act downstream of ATG-18 to control C. elegans lifespan cell non-autonomously. Mutations in unc-64, the gene that encodes the worm ortholog of vertebrate syntaxin 1A, block the release of neurotransmitters [39]. We found the atg-18 mutation decreases the lifespan of unc-64(e246) mutants (Fig 6, S10 Table). Expression of atg-18 in neurons or intestinal cells only partially rescues the short lifespan of unc-64;atg-18 ( Fig 6A and 6B), which suggests neurotransmitters are required for neuronal and intestinal ATG-18 to maintain wildtype lifespan. To confirm the involvement of neurotransmitters in ATG-18-mediated neuronal signaling, we constructed unc-13(e51);atg-18 (gk378) mutants. The C. elegans UNC-13 protein and its mammalian orthologs are important for normal neurotransmitter release. Mutations in unc-13 block release of most neurotransmitters [40]. We found that the eggs of unc-13;atg-18 mutants cannot hatch. Neuronal but not intestinal expression of atg-18 can rescue this lethality, indicating neuronal ATG-18 influences neurotransmitter release.
We next attempted to test the influence of UNC-31 on the rescue of daf-2;atg-18 lifespan by neuronal or intestinal ATG-18. We found that daf-2;unc-31;atg-18 mutant eggs cannot hatch and expression of atg-18 in either neurons or intestinal cells cannot rescue this lethality. In the daf-2unc-64;atg-18 mutant background, we also observed that the fertility is significantly reduced and only 18.99±2.98% (mean±s.e.m., n = 463) of eggs develop to adults. By contrast, 91.54±1.06% (mean±s.e.m., n = 1275) of daf-2unc-64 eggs develop to adults in four days at 20˚C. Interestingly, neuronal or intestinal expression of atg-18 only partially rescues this developmental defect (for neuronal rescue, 38.36±3.99%, mean±s.e.m., n = 672; for intestinal rescue, 27.70±3.21%, mean±s.e.m., n = 240), which is consistent with the observation that UNC-64 is required for the rescue of daf-2;atg-18 short lifespan by neuronal or intestinal ATG-18. Next, we tested if UNC-64 and UNC-31 are required for the cell non-autonomous regulation of C. elegans lifespan under sDR. We found that the lifespan of unc-64 and unc-31 mutants is increased after sDR treatment (S9 Fig, S11 Table) and atg-18 mutations block this DR response (Fig 7A and 7B, S11 Table). Moreover, neuronal or intestinal expressed ATG-18 confers DR response in unc-64;atg-18 mutants (Fig 7C and 7D), suggesting that neurotransmitter release might be dispensable for neuronal and intestinal ATG-18-mediated DR response. We also found that the unc-31 mutation fails to block neuronal ATG-18-mediated DR response since the lifespan of unc-31;atg-18 mutants carrying neuronal expressed ATG-18 is increased after DR treatment ( Fig 7E). Interestingly, the unc-31 mutation completely suppresses the intestinal ATG-18-mediated DR response (Fig 7F), which supports the cell non-autonomous function of intestinal ATG-18 in regulation of DR response.
Finally, to test whether ATG-18 influences the release of neurotransmitters we performed an aldicarb assay. Aldicarb inhibits acetylcholinesterase, preventing the breakdown of acetylcholine in the synaptic cleft, which causes paralysis in C. elegans. Gene mutations that block synaptic transmission render worms resistant to aldicarb-induced paralysis [43]. Similar to a previous report [39], while 100% of N2 worms become paralyzed after 3-hour treatment by 1mM aldicarb, none of unc-64 mutants are paralyzed (S10 Fig). Interestingly atg-18 mutants are as sensitive to aldicarb treatment as N2 worms (S10 Fig), indicating the atg-18 mutation neither blocks acetylcholine release nor influences release of all neurotransmitters in general. This is consistent with the fact that sensory neurons, where ATG-18 acts, are generally not cholinergic [44]. Moreover, unc-64;atg-18 mutants are resistant to aldicarb treatment indicating unc-64 is downstream of ATG-18 in neurotransmitter signaling.
Autophagy deficiency influences the expression levels of insulin-like peptide (ILP) genes DAF-16 is required for lifespan extension by inhibition of certain chemosensory neurons [13]. We found inhibition of DAF-16 blocks the rescue effect of either neuronal or intestinal ATG-18 on daf-2;atg-18 lifespan (Fig 8A, S12 Table), which suggests both neuronal and intestinal ATG-18 act through DAF-16 to regulate C. elegans lifespan in response to reduced IGF signaling. Since DAF-16 is the major target for IGF signaling, we postulated that autophagy deficiency might influence the expression levels of ILP genes. We examined the expression levels of 40 ILP genes in daf-2 and daf-2;atg-18 mutants. We found that the expression levels of six ILP genes (ins-1, ins-3, ins-29, ins-35, ins-37 and daf-28) are significantly changed in daf-2;atg-18 mutants compared to those of daf-2 mutants (Fig 8B, S13 Table). The expression levels of ins-1, ins-3 and ins-35 are increased in daf-2;atg-18 mutants, while the expression levels of ins-29, ins-37 and daf-28 are decreased (Fig 8B, S13 Table). Then we tested if neuronal or intestinal expression of atg-18 can revert the expression of these six ILPs genes in daf-2;atg-18 mutants. We found that neuronal ATG-18 has no significant effect on the expression levels of ins-1 (p = 0.1038 and 0.2315 for two trials), ins-3 (p = 0.0548 and 0.74513 for two trials) and ins-37 (p = 0.3611 and 0.6729 for two trials). Moreover, contradictory to our prediction, neuronal ATG-18 further increases the expression levels of ins-35 and further decreases the expression levels of ins-29, and daf-28 in daf-2;atg-18 mutants (Fig 8C). These data indicate that neuronal expression of atg-18 fails to revert the change in expression of six ILP genes in daf-2;atg-18 mutants.

Discussion
Autophagy is required for IGF signaling and DR to extend lifespan in C. elegans [30,31,33]. However, the mechanisms are not well understood. Here we show that autophagy is an essential component in the neuroendocrine regulation of C. elegans lifespan. The DAF-2 insulinlike receptor is highly expressed in neurons, as detected by an anti-DAF-2 antibody, and the daf-2 promoter is active in both neurons and intestinal cells [45,46]. Apfeld and Kenyon found that neuronal DAF-2 acts cell non-autonomously to regulate longevity [47], while Wolkow et al. reported that DAF-2 expressed in either neurons or intestinal cells prevents the long lifespan of daf-2 mutants [48]. Interestingly, DAF-16, the major target of DAF-2, primarily functions in the intestine to mediate DAF-2 regulated longevity while neuronal DAF-16 is dispensable [49]. This suggests DAF-2 can work in both neurons and intestinal cells while DAF-16 acts mainly in the intestine to control lifespan. Autophagy is induced in daf-2 mutants and acts downstream of DAF-2 in a DAF-16 independent manner [31,33]. Therefore, it is reasonable to speculate that DAF-2 regulates autophagy in neurons. Our data indicate ATG-18 in chemosensory neurons and intestine acts in parallel to mediate the influence of DAF-2 on lifespan through DAF-16. Since autophagy is highly sensitive to nutrients, we propose neuronal and intestinal autophagy responds to environmental food cues and assimilated food nutrients respectively to regulate C. elegans lifespan in response to reduced IGF signaling (Fig 9). In this model, environmental food cues inhibit autophagy though DAF-2 in chemosensory neurons. Autophagy negatively influences the availability of neurotransmitters that require UNC-64 for release. These neurotransmitters inhibit the function of unidentified (herein XX) neurons which secrete neuropeptides that negatively influence C. elegans lifespan by inhibition of DAF-16. Indeed, DAF-16 is required for the lifespan extension resulting from mutations in odor receptors or ablation of food sensory neurons [12,13]. In parallel to the neuronal regulation of autophagy by environmental food cues, food nutrients absorbed by the intestine also inhibit autophagy. In the absence of this inhibition, autophagy promotes longevity by modulating the function of chemosensory neurons. According to this model, the change of physiological levels of insulin peptides due to autophagy deficiency enhances DAF-2 activity and suppresses the  28. Asterisks indicate statistically significant differences relative to controls (* P < 0.05, ** P < 0.01, t-test). Details of the statistical analysis of the qPCR data is found in S13 Table. https://doi.org/10.1371/journal.pgen.1006764.g008 ATG-18 regulates C. elegans lifespan cell non-autonomously daf-2 longevity phenotype. This model is supported by the following experimental data in the present study. Firstly, both neuronal and intestinal expressed atg-18 only partially rescue the short lifespan of daf-2unc-64;atg-18 mutants, which indicates that UNC-64-mediated neurotransmitter release is required for ATG-18 to fully restore the lifespan of daf-2unc-64;atg-18 mutant to the daf-2unc-64 mutant level. Secondly, the fertility of daf-2unc-64;atg-18 mutants is significantly reduced and neuronal or intestinal expression of atg-18 only partially rescues this developmental defect. This is consistent with the observation that UNC-64 is required for the rescue of daf-2;atg-18 short lifespan by neuronal or intestinal ATG-18. Thirdly, although unc-13(e51) is viable, the unc-13(e51);atg-18(gk378) double mutant is lethal. Additionally, it has been reported that severe unc-13 mutants are lethal [50]. If unc-13 and atg-18 are in a parallel pathway, we expect that either neuronal or intestinal ATG-18 should rescue the lethality as expression of atg-18 in these two tissues fully rescues the short lifespan of atg-18 mutants. However, we found that only neuronal ATG-18 can rescue the lethality, which indicates that UNC-13 and ATG-18 act in a linear pathway in neurons and UNC-13 is downstream of ATG-18. Fourthly, because the daf-2;unc-31;atg-18 mutant is lethal we could not test the requirement of UNC-31 for neuronal or intestinal ATG-18 to rescue the short lifespan of daf-2;atg-18. However, UNC-31 is necessary for neuronal or intestinal ATG-18 to rescue the short lifespan of atg-18. Moreover, neuronal or intestinal ATG-18 fails to rescue the lethality of daf-2;unc-31; atg-18 mutants, indicating that secreted neuropeptides are the final target of neuronal and intestinal ATG-18. Finally, inhibition of daf-16 by RNAi completely blocks the ability of neuronal or intestinal ATG-18 to rescue the lifespan phenotype of daf-2;atg-18 mutants, indicating DAF-16 is the target regulated by the secreted neuropeptides.
Our model is supported by a recent study showing that ILPs secreted by chemosensory neurons facilitated by UNC-31 enable L1 C. elegans larva to survive from starvation via DAF-16 [51]. Indeed, autophagy is essential for survival of animals under starvation conditions [18]. Therefore, it is reasonable to speculate that neuronal insulin-like peptides act downstream of neuronal and intestinal ATG-18 to regulate C. elegans lifespan. C. elegans encodes 40 ILPs most of which are not characterized [52]. Some of ILPs are agonists for DAF-2 while others are antagonists [52]. A recent study indicates that the nervous system secretes a specific group of ILPs and that the agonistic-antagonistic ILP balance determines the activity of intestinal DAF-2 during the larval development in response to environmental food availability [53]. Thus, autophagy may influence the physiological levels of both agonist and antagonist ILPs from XX neurons instead of only agonist ILPs as shown in the model. Interestingly, although most of the ILPs are expressed in neurons, some are expressed in the hypodermis [54], which may explain our data showing that expression of ATG-18 in the hypodermis fully rescues daf-2;atg-18 lifespan. However hypodermal ATG-18 has no influence on wildtype lifespan.
We performed qPCR to measure the expression levels of 40 ILPs and found that the expression levels of ins-1, ins-3 and ins-35 are increased in daf-2;atg-18 mutants, while the expression levels of ins-29, ins-37 and daf-28 are decreased compared to those of daf-2 mutants. DAF-28, INS-3 and INS-35 are classified as agonists for DAF-2 and INS-1 is an antagonist [55,56]. The interaction of INS-29 and INS-35 with DAF-2 has not been reported. If neuronal ATG-18 acts through these ILPs, we expect that expression of neuronal ATG-18 in daf-2;atg-18 mutants will revert the expression of these genes to the daf-2 levels. However, this is not our observation. There are two possible interpretations for this result. It is possible that insulin peptides are not involved in ATG-18 mediated cell non-autonomous signaling. Alternatively, atg-18 mutations influence the physiological levels of certain insulin peptides that regulate intestinal DAF-16 activity. However, this influence fails to revert the transcription levels of the six aforementioned ILPs genes to daf-2 levels. We favor this possibility because both neuronal and intestinal autophagy act though DAF-16 and intestinal DAF-16 acts downstream of DAF-2 signaling to regulate C. elegans lifespan [49]. A future study can be performed to systematically examine the production and secretion of all 40 insulin peptides in daf-2;atg-18 mutants, which might help to determine whether insulin peptides are involved in neuronal ATG-18 mediated cell non-autonomous signaling.
Interestingly, we found that intestinal ATG-18 reverts the expression of ins-1 in daf-2;atg-18 mutants to the daf-2 level, suggesting that INS-1 is a candidate insulin peptide that may be involved in cell non-autonomous regulation of lifespan by intestinal ATG-18. INS-1 is most related to mammalian insulin [55]. It functions redundantly with other insulin-like peptides to regulate reproductive growth and lifespan [55]. GFP expression under the control of the ins-1 promoter is detected in some amphid sensory neurons including ASI and ASJ which regulate dauer formation [55]. Dauer larva is a dispersal stage of C. elegans that forms under starvation conditions and INS-1 promotes dauer entry [55,57]. A functional INS-1::GFP reporter gene is strongly expressed in sensory neurons but not in intestinal cells [53]. Interestingly INS-1 has been reported to act on intestinal DAF-2 to promote dauer formation through DAF-16 [53]. Our data show that intestinal ATG-18 generates cell non-autonomous signal molecules that communicate with neurons to regulate lifespan. Thus, INS-1 could be a candidate neuropeptide secreted by neurons in response to this communication. In support of our data, a crosstalk between neuronal and intestinal autophagy has been reported to influence Drosophila lifespan, which is linked to altered ILP levels [58].
Previous studies indicate that DR can act in parallel to IGF signaling to increase C. elegans lifespan and that DAF-16 is not required for lifespan extension under certain DR protocols [5]. We found that ATG-18 in chemosensory neurons and intestinal cells enables atg-18 mutants to respond to DR. Moreover, we found that neurotransmitters may not be required for neuronal ATG-18 to mediate the DR response as the unc-64(e246) mutation does not block the lifespan extension of atg-18 mutants conferred by neuronal expression of ATG-18 under DR treatment. However, since the unc-31(e298) mutation completely block the DR response of atg-18 mutants carrying intestinal ATG-18, neuropeptides act downstream of intestinal ATG-18 to mediate the DR response. In addition, we found that ATG-18 in muscles plays a minor role in maintaining C. elegans wildtype lifespan and response to reduced IGF signaling but ATG-18 in the muscles or hypodermis has no contribution to the DR response. These results indicate that ATG-18 in individual tissues has different roles in regulating C. elegans lifespan under specific conditions.
Recently, Gelino et al. reported that intestinal knockdown of atg-18 shortens the lifespan of the DR mutant eat-2, demonstrating that under this DR condition neuronal atg-18 cannot sufficiently mediate the DR response [59]. By contrast, our data show that neuronal ATG-18 confers the DR response in atg-18 mutants. This discrepancy might be caused by the difference of DR protocols used in these two studies. Indeed, while DAF-16 is required for the sDR response [32], it is dispensable for the longevity phenotype of eat-2 mutants [60]. Nevertheless, intestinal ATG-18 clearly is essential for lifespan extension in response to DR under these two different DR protocols.
ATG-18 and its mammalian homolog, WIPI1 and WIPI2, are essential for the autophagy process [20][21][22]. Our data show that atg-18 mutations suppress autophagy induction in daf-2 mutants and wildtype animals treated by DR, indicating that the autophagy function of ATG-18 controls C. elegans lifespan. Since autophagy is evolutionarily conserved from C. elegans to mammals, the cell non-autonomous function of autophagy in the chemosensory neurons and intestine to control aging via a neuroendocrine mechanism may be conserved.
To construct the Podr-2::gfp::atg-18 reporter gene, the gfp coding sequence was amplified from the pPD95.75 plasmid. The gfp fragment flanked by the KpnI site at both ends was cloned into pGEMT and sequenced to confirm that there are no mutations in the gfp sequence. The gfp fragment was then inserted downstream of the odr-2 promoter and upstream of the atg-18 genomic DNA. The orientation of the gfp sequence was confirmed by sequencing and the gfp sequence was fused in frame with atg-18 gene. A similar strategy was used to construct the following three plasmids: Podr-10::gfp::atg-18, Ptph-1::gfp::atg-18, Psrh-220::gfp::atg-18.

Lifespan experiment
Animals were grown on NG agar plates seeded with E. coli strain OP50. Well-fed L4 hermaphrodites were picked up for lifespan experiments on day zero. OP50 bacteria were grown overnight at 37˚C with shaking. For the lifespan experiments using solid dietary restriction (sDR) assay [32], the overnight culture was washed and re-suspended in S Basal media without cholesterol. Serial dilutions were performed to achieve bacterial concentrations of 5.0 X 10 11 cells/ml (AL: ad libitum) and 5.0 X 10 9 cells/ml (DR: dietary restriction). 250 μl of these diluted bacterial cultures was spotted on each 60 mm sDR agar plate on the day of transfer. sDR plates were made by modifying the standard NGM plate formula to exclude peptone, increase agar concentration to 2.0% and include ampicillin (50μg/mL). L4 larval stage worms were picked up from animals grown on regular OP50 food plates. On the next day, designated day 1, adults were placed on fresh sDR plates with 30 worms on each plate. Three plates were prepared and there were 90 worms total for each group at the beginning of experiments.
For the daf-16 RNAi lifespan experiment, RNAi feeding was performed as described by Kamath et al [65]. Briefly, L4 hermaphrodites were put on RNAi or control plates for thirty-six hours. The worms were then transferred to fresh RNAi or control plates and allowed to lay eggs. The L4 hermaphrodites of the RNAi-treated progeny were collected for experiments.
All lifespan experiments were performed at 20˚C; animals were transferred to fresh plates every day during the reproductive period, and every other day thereafter. Animals were scored as dead when they failed to respond to touch. Animals that died from internal hatching or vulval rupture were censored from statistical analysis. GraphPad Prism 5 (GraphPad Software, La Jolla, CA) was used to generate survival curves, to obtain the median survival times in each group and to perform log-rank survival analyses (Kaplan-Meier method).

qRT-PCR
Worms were allowed to grow on OP50 food for 3 days. These animals were harvested from the plate and washed with M9 buffer to remove all bacteria on the third day. RNA extraction was then done using the Trizol kit (Zymo). First strand cDNA was synthesized from the RNA samples by using the SuperScript First-Strand Synthesis System (Invitrogen). The expression levels of the ILPs were measured in triplicate using the SYBR green dye (Quanta) on an ABI Prism 7000 real-time PCR machine (Applied Biosystems). The relative-fold changes were calculated using the 2 −ΔΔCT method. The entire experiment was repeated at least once with RNA extracted from new samples to confirm the original results. The primer sequence used for qRT-PCR was derived from a previous publication [66].

Autophagy activity analysis
Autophagy activity was analyzed by counting autophagosomes that appear as fluorescence puncta in hypodermal seam cells [33,35]. Animals were grown at 20˚C on OP50 or sDR food plates (preparation of sDR and control food plates were described in the above Lifespan Experiment section). L3-L4 larva were collected, mounted on agar pad slides and observed under a Zeiss upright fluorescence microscope (Axio Imager A2). Pictures were captured using the Zeiss AxioCam digital camera at 1,000X magnification. Approximately 20 animals were examined for each strain in each experiment.

Aldicarb assay
The experiment was performed as described previously [43]. In brief, L4 worms were picked and allowed to develop to adults on OP50 food plates. On the following day, day 1, adults were transferred to freshly prepared aldicarb plates to begin the assay. The aldicarb plates were prepared on the day of the experiment by spreading aldicarb (Sigma) on NGM plates to achieve a concentration of 1mM. A spot of OP50 was added to the center of the plates to keep the worms concentrated. Each strain was tested in triplicate with 20 worms per plate. Paralysis of animals was scored thirty minutes after the worms were transferred to the aldicarb plates.
Worms were considered paralyzed once they showed no movement in response to being prodded three times by a platinum wire. GraphPad Prism 5 was used to generate the graph and to perform statistical analysis.