Trimethylamine modulates dauer formation, neurodegeneration, and lifespan through tyra‐3/daf‐11 signaling in Caenorhabditis elegans

Abstract In the nematode Caenorhabditis elegans, signals derived from bacteria in the diet, the animal's major nutrient source, can modulate both behavior and healthspan. Here we describe a dual role for trimethylamine (TMA), a human gut flora metabolite, which acts as a nutrient signal and a neurotoxin. TMA and its associated metabolites are produced by the human gut microbiome and have been suggested to serve as risk biomarkers for diabetes and cardiovascular diseases. We demonstrate that the tyramine receptor TYRA‐3, a conserved G protein‐coupled receptor (GPCR), is required to sense TMA and mediate its responses. TMA activates guanylyl cyclase DAF‐11 signaling through TYRA‐3 in amphid neurons (ASK) and ciliated neurons (BAG) to mediate food‐sensing behavior. Bacterial mutants deficient in TMA production enhance dauer formation, extend lifespan, and are less preferred as a food source. Increased levels of TMA lead to neural damage in models of Parkinson's disease and shorten lifespan. Our results reveal conserved signaling pathways modulated by TMA in C. elegans that are likely to be relevant for its effects in mammalian systems.

(ILS) pathways. These ultimately converge to modulate the activity of key molecular factors such as forkhead transcription factor (DAF-16/ FOXO) (Avery & You, 2012). DAF-16 activity is pivotal in determining the organism's response to either proceed with reproductive growth or to enter the dauer stage (Hu, 2007;Ogg et al., 1997). Although the molecular consequences of changes in food availability have been well studied, there is little understanding of specific bacterial signals that direct these choices in the host. C. elegans is known to respond to signals emanating from bacteria to detect food sources, and bacterial fatty acids can act as a signal that modulates recovery from the dauer stage (Kaul et al., 2014). A genetic screen for bacterial genes has identified several other signals including cAMP that in part direct these choices (Khanna et al., 2016;Urrutia et al., 2020).
In this study, we describe a role for trimethylamine (TMA) as a bacterial signal that both promotes food-seeking behavior and activates nutrient signaling pathways. A gut microbiome-derived metabolite, TMA, is produced from dietary phosphatidylcholine and is converted to trimethylamine N-oxide (TMAO) in the body (Gipson et al., 2008;Z. Wang et al., 2011). Multiple studies have identified TMA or TMAO as a risk factor for type II diabetes and cardiovascular diseases in humans (Gipson et al., 2008;Z. Wang et al., 2011). Recent study reported increased gut-derived TMAO in the cerebrospinal fluid of Alzheimer's disease (AD) patients with mild cognitive impairment emphasizing the role of this metabolite in connecting gut-brain axis (Vogt et al., 2018). However, the mechanisms by which TMA and TMAO influence host molecular pathways and their physiological consequences remain poorly understood.
Here, we demonstrate that TMA acts as a nutrient signal in C. elegans sensed by the G protein-coupled receptor (GPCR), TYRA-3, and the guanylyl cyclase, DAF-11, to modulate food perception, dauer formation, and lifespan. We additionally show that TMA causes dopaminergic neuronal loss in a TYRA-3-dependent manner. Given the conservation of molecular pathways modulating nutrient perception and signaling (Hu, 2007), C. elegans constitutes a powerful system for understanding how bacterial signals interact with their host, which is likely to be relevant for better understanding host-microbiome interactions and their influence on healthspan in humans.

| A conserved GPCR, tyra-3, mediates TMA sensing in C. elegans
C. elegans possesses an intricate chemosensory system capable of distinguishing various chemical stimuli. C. elegans' response to food sources, toxins, or bacterial-derived metabolites can be measured via chemotaxis assays (Bargmann, 2006;Bargmann et al., 1993;Margie et al., 2013). To evaluate whether TMA acts as a chemoattractant or a nutritional signal, we used a chemotaxis assay to measure C. elegans behavior in the presence of TMA. Wild-type C. elegans (N2) were found in this assay to be attracted to TMA in the absence of bacteria in a dose-dependent manner (Figure 1a; Table S1a).
C. elegans detect various chemical stimuli through specialized biogenic amine receptors (Troemel et al., 1995). These are typically GPCRs, and their signaling is conserved between C. elegans and vertebrates (Sengupta, 2007). To identify putative receptors involved in sensing TMA, we tested candidate mutants deficient in biogenic amine receptors using the chemotaxis assay. Observed results demonstrated that the tyramine receptor mutant tyra-3(ok325) was unable to sense TMA. Similarly, RNAi-mediated knockdown of tyra-2 reduced chemotaxis behavior toward TMA (Figure 1b). C. elegans daf-11(m47) mutants containing a mutation in a transmembrane guanylyl cyclase also displayed defective chemosensory behavior ( Figure 1b). This suggests that multiple receptors in C. elegans may sense TMA. In contrast, our candidate screen showed that loss of function in odr-4(n2144) (odorant response abnormal protein 4) and the serotonin/octopamine receptor family genes ser-5(ok3087) and ser-7(tm1325) did not alter chemotaxis behavior toward TMA ( Figure 1b). In order to rule out that changes in levels of tyra-3 accounted for lack of change in chemotaxis behavior in these mutants, we quantified tyra-3 expression via qRT-PCR. Our results show that tyra-3 expression is unchanged in ser-5(ok3087), ser-7(tm1325), and odr-4(n2144) mutants ( Figure S1).
C. elegans tyra-3 is predicted to encode a GPCR for the invertebrate neurotransmitter tyramine (Wragg et al., 2007) and shares high conservation with the transmembrane domain of the human trace amine receptor (TAAR5) (WormBase). CB4856, a C. elegans strain isolated in Hawaii (HW), encodes a tyra-3 variant shown to be partially responsible for increased food leaving behavior (Bendesky et al., 2011). Since C. elegans tendency to leave or remain on a lawn of bacteria is linked to their capability to sense external stimuli, we performed chemotaxis assay on the HW worm strain. We found that HW worm was not attracted to TMA, further linking tyra-3 to the sensing of TMA (Figure 1c; Table S1). TYRA-3 is expressed in the gonads, vulva, and several head neurons and in interneuron SDQ. To identify the functional loci involved in TYRA-3-mediated response to TMA, we used transgenic HW worms expressing N2-derived tyra-3b promoter (N2 tyra-3b) in order to rescue variations in the non-coding region of the HW strain (Bendesky et al., 2011). This sequence was driven by cell type-specific promoters for BAG, CEP, ASK, or ADL neurons in biogenic amine receptor mutant worms. Transgenic expression of N2 tyra-3b in ASK, BAG, and CEP neurons, but not ADL, partially rescued TMA attraction in HW worms ( Figure 1c; Table S1).
This confirms the necessity of a functional TYRA-3 in ASK, BAG, and CEP neurons for positive chemotaxis behavior in response to TMA.
C. elegans display behavioral responses to bacterial food based on both the quality of food available and the animal's ability to detect the food. If food is easily available, the worm will spend more time on the bacterial lawn, a phenotype called dwelling. If the food is not of a preferred variety, worms avoid the bacterial colony and search for more preferable food, a phenotype called roaming (Shtonda & Avery, 2006). The food intake in C. elegans can be measured by counting the pharyngeal pumping rate. We used lawn-leaving behavior and pharyngeal pumping to test the ability of TMA to alter both food-seeking and food intake behaviors.
Results demonstrate that addition of 2 mM TMA to bacterial food significantly reduced lawn-leaving behavior ( Figure 1d) without altering food intake ( Figure 1e).
Increased food availability or favorable changes in the environment initiate feeding and molting of L2d larvae to L4. Given the role of TMA as a food signal, we examined whether TMA could act as a nutrient signal to influence dauer formation. To avoid confounding effects of dietary amines (or lack thereof), we employed minimal media conditions (no glucose) and high temperature (Lee et al., 2009) to examine the effects of exogenous TMA (2 mM) on dauer formation.
The addition of TMA to K12 bacteria lawns significantly reduced dauer formation in N2 (by ~23%) compared to control, whereas treatment of tyra-3(ok325) mutant worms with TMA showed no significant decrease in dauer formation ( Figure 2a; Table S2a). This supports the hypothesis that TMA requires tyra-3 to influence dauer signaling.
Taken together, our results show that TMA is sensed by TYRA-3 and modulates dauer formation through DAF-11 signaling while also interacting with DAF-2 pathway.
Guanylyl cyclase, TGFβ, and insulin-like signaling (ILS) pathways are known to independently modulate DAF-16, a C. elegans forkhead (FOXO) homolog that influences both dauer formation and lifespan (Hahm et al., 2009;Hu, 2007). Upon mild heat shock, DAF-16 translocate from the cytoplasm to the nucleus to modulate gene expression including those involved in dauer formation (Lee et al., 2009). To interrogate the potential role of TMA in the DAF-16 pathway, we utilized a DAF-16 reporter strain and explored its response to mild temperature increase in the presence of TMA. We observed that exposure to TMA significantly increased cytoplasmic retention of DAF-16 in worms shifted from 15ºC to 21.5ºC in a dose-dependent manner ( Figure S2).
Together, these results indicate that TYRA-3 acts upstream of the dauer signaling pathway(s) to modulate both dauer formation and chemotaxis behavior in response to biogenic amines like TMA.

| Bacteria-specific DMSO reductase mutant modulates nutrient sensing and dauer formation in C. elegans
Bacteria utilize dimethyl sulfoxide (DMSO) and TMA as oxidizing agents. The bacterial enzymes DMSO reductase (dms) and trimethylamine N-oxide (TMAO) reductase are involved in catalyzing the reduction of various N-oxides to amines, such as TMAO to TMA (Baba et al., 2006;Cammack & Weiner, 1990;Sambasivarao & Weiner, 1991). To examine whether bacteria with altered amine production are sufficient to influence nutrient-sensing pathways in C. elegans, we screened single-gene mutants ( Figure S3a) of DMSO reductase (dms)/TMAO reductase and genes involved in choline and betaine metabolism (Table S2b). Using a bacterial choice assay (Glater et al., 2014), we found that N2 worms were more attracted to K12 (control) bacteria than to DMSO reductase mutant bacteria (dmsA, dmsB, F I G U R E 2 TMA modulates dauer formation in C. elegans. a. Agesynchronized worms were tested for dauer formation on minimal media plates without glucose on K12 bacterium lawns at high temperature (as specified in materials and methods) in the presence and absence of TMA (2 mM). b. Under similar assay conditions, dauer formation was quantitated in the presence of cGMP (2 mM) in eggs from daf-11(m47), and tax-4;daf-11 double mutant worms. c. Dauer assays were performed with synchronized daf-2(e1370) mutant worms grown on bacteria expressing worm RNAi for tyra-3, tyra-2, and vector control. Eggs were exposed to K12 bacteria in the presence and absence of TMA, and dauers were scored. **p < 0.001; average ±std. dev (n = 3) dmsC) with or without TMA ( Figure 3a; Table S2b). Consistent with their defect in TMA perception, tyra-3(ok325) and HW animals showed no difference in attraction to these two different bacterial strains ( Figure 3b). However, HW transgenic worms expressing N2 tyra-3b allele in either ASK, BAG, or CEP neurons preferred K12 over dmsC mutant bacteria (Figure 3b). Neither daf-11 nor tyra-3 mutants displayed a significant preference for K12 bacteria over dmsC mutant bacteria as compared to N2 worms ( Figure S3b, c). We observed a significant decline in lawn-leaving events in N2 worms when dmsC bacterial lawns were supplemented with TMA ( Figure S3d).
Next, we assessed the potential role of DMSO reductases in dauer formation in N2, daf-11(m47), daf-7(e1372), and daf-2(e1370) mutant worms. Since the goal of this assay is to understand the role of bacterial enzymes and TMA, we performed dauer assay using minimal media conditions (no glucose) at 21.5ºC. We observed a significant increase in dauer formation in N2 worms fed with bacteria lacking DMSO/TMAO reductases. Interestingly, supplementing dmsC mutant lawns with TMA (2 mM) significantly decreased dauer formation compared to dmsC mutant controls (Figure 3c). We also observed a significant increase in dauer formation in daf-7(1372) and daf-2(e1370) mutant worms which was rescued by addition of TMA (2 mM). No significant change was observed in daf-11(m47) mutant worms under these conditions, further supporting a role for DAF-11 signaling in TMA-dependent dauer formation ( Figure 3c). The HW parent strain does not sense food and did not show dauer formation when fed with dmsC mutant bacteria. However, HW worms expressing the N2 tyra-3b allele in ASK, BAG, or CEP neurons showed an increase in dauer formation when worms were fed with dmsC mutant bacteria ( Table S2c). Addition of 2 mM TMA significantly rescued this dauer phenotype ( Table S2c), suggesting that TMA act as nutrient signal in rescuing C. elegans from dauer development and that tyr-3 is required to sense TMA.
The addition of TMA (2 mM) or a cGMP analog (2 mM) to dmsC mutant bacterial lawns did not decrease dauer formation in the tax-4(p678);daf-11(m47) double mutant under semi-permissive conditions ( Figure S3e). Similarly, we observed that TMA (2 mM) did not rescue dauer formation in daf-2(e1370) worms with a tyra-3 knock down compared to daf-2(e1370) worms or daf-2(e1370) worms with a tyra-2 knockdown when fed on dmsC mutant bacteria (Figure 3d), confirming the role of daf-11 signaling and emphasizing the requirement of tyra-3 for sensing TMA. In addition, we observed increased DAF-16 nuclear localization in worms fed with dmsC mutant bacteria which F I G U R E 3 Bacteria-specific DMSO reductase mutant modulates nutrient sensing and dauer formation in C. elegans. a. Bacterial choice index for N2 worms exposed to K12 and DMSO reductase (dms) mutants in the presence or absence of TMA (2 mM). b. Bacterial choice index for HW, tyra-3(ok325) mutant, and HW transgenic worms expressing the N2 tyra-3b allele in ASK, BAG, or CEP neurons in the presence of wild-type (K12) bacteria as compared to dmsC mutant bacteria. Worms n ~ 300, n ≥ 6 assays. **p < 0.01; average ±std. dev (n = 3). c. Dauer assays were performed on minimal media plates without glucose with K12 bacterium lawns at 21.5°C in the presence and absence of TMA (2 mM). Synchronized eggs were used from N2 (wild-type), daf-2(e1370), daf-7(e1372), and daf-11(m47) mutant worms. d. Dauer assays were performed with eggs from daf-2(e1370) mutants worms kept on bacteria expressing worm RNAi for tyra-3, tyra-2, and vector control. Eggs were exposed to K12 and dmsC mutant bacteria in the presence and absence of TMA and dauers scored. e. DAF-16::GFP nuclear localization in worms fed K12 vsdmsC mutant bacteria in the presence or absence of TMA (2 mM). Worms n ~ 300, n ≥ 6 assays, **p < 0.01; average ±std. dev (n = 3) was rescued by addition of exogenous TMA (2 mM) (Figure 3e). No dauers were observed in the dauer defective, daf-16(mu86), worms fed with dmsC mutant bacteria, indicating that DAF-16 is required to manifest the effects of dmsC mutant bacteria on dauer formation (Data Not shown). These results suggest that TMA triggers a neuroendocrine signal that modulates DAF-16 activity to promote reproductive growth and inhibit dauer formation. Our results support the idea that biogenic amines including TMA play an important role in C.
elegans nutrient-sensing pathways by signaling the presence of bacteria and that TMA signaling is mediated in part through TYRA-3 to modulate both dauer formation and chemotaxis behavior.

| TYRA-3 mediates TMA-dependent dopaminergic neuronal damage
TYRA-3 is expressed in several neurons including the CEP dopaminergic (DA) neurons (Bendesky et al., 2011). Our results in C. elegans placed in liquid demonstrates specific locomotor behavior dubbed "body bends" or "thrashing." Thrashing behavior or body bends are regulated by motor neurons and therefore can be used to assess motor neuron function (Matthies et al., 2006). We counted the number of body bends per 30 seconds in the presence and absence of TMA. Here, we observed a significant decline in body bends in WT and αsyn OE worms treated with 2 mM TMA compared to the non-treated (Figure 4d). Conversely, tyra-3 mutant worms treated with 2 mM TMA did not decrease body bend rate, implying that TYRA-3 is relevant for the TMA-induced loss of neurons ( Figure 4d). Our results demonstrate that TMA causes an increase in alpha-synuclein aggregation and a loss in DA neurons through its F I G U R E 4 TYRA-3 mediates TMA-dependent dopaminergic neuronal damage. a. Histogram shows the percentage of intact CEP neurons/ all worms scored on day 1, 3, and 5 in αsyn OE and αsyn OE;tyra-3 transgenic worms after exposure to TMA (2 mM). b. Representative images of αsyn OE expressing αsyn and GFP specifically in the four anterior DA neurons in Control (water) and TMA (2 mM). c. Histogram shows the number of aggregates in DA dendrites after exposure to TMA for 5 days in αsyn OE and αsyn OE;tyra-3. d. Number of body bends are presented as a histogram for wild-type worms (N2), tyra-3 (ok325), αsyn OE, and αsyn OE;tyra-3 transgenic in absence and presence of TMA (2 mM). Worms n ~ 120, n ≥ 6 assays. **p < 0.01; average ±std. dev (n = 3) receptor TYRA-3. Further, knockdown of tyra-3 was sufficient to negate the effects of TMA-induced alpha-synuclein aggregation and loss of neuronal function (Figure 4d).

3-and daf-11-dependent manner
As nutrient and sensory cues play an important role in modulating longevity (Bargmann, 2006;Bendesky et al., 2011;Libina et al., 2003), we examined the effects of TMA on worm lifespan. We observed a significant increase in the mean lifespan (15.61%) of N2 worms fed dmsC mutant bacteria compared to K12 bacterial-fed worms ( Figure 5a). In addition, we found that mean lifespan was significantly reduced (16.82%) in worms fed K12 bacteria supplemented with TMA. (Figure 5a). Consistent with our observations in the chemotaxis and dauer formation assays, there was no significant increase in the lifespan of tyra-3(ok325) mutant worms fed with dmsC mutant bacteria compared to K12-fed worms (Figure 5b).
However, on continuous exposure to TMA (2 mM), tyra-3(ok325) mutants had a decreased mean lifespan by ~21% on K12 and 19% on dmsc mutant, respectively, implying that there may be additional receptors that mediate the reduction of lifespan by TMA (Figure 5b).
No significant increase or decrease in lifespan was observed in daf-11(m47) worms fed with dmsC bacterial mutant or TMA, respectively (4%), implying an important role for DAF-11 (Figure 5c). We observed a small increase in lifespan of daf-16(mu86) worms fed with dmsC bacterial mutant (8.11%). We also observed that TMA treatment reduced the lifespan in daf-16(mu86) worms (9.67%) suggesting partial dependence of TMA on DAF-16 activity in mediating lifespan changes (Figure 5d).

Insulin-like peptides (ILPs) have been shown to regulate DAF-16
activity, but are also themselves regulated by DAF-16 as part of both positive and negative feedback loops (Liu et al., 2004). ILPs also regulate the dauer entry and exit signals by functioning as either agonist or antagonist (Cornils et al., 2011). Certain ILPs are upregulated in daf-2 mutants raised on Escherichia coli OP50 (Mueller et al., 2014).
Using qPCR, we tested the expression of ILP genes in N2 worms fed with either K12 or dmsC mutant bacteria and found significant upregulation of ins-35 in dmsC fed worms. The increased expression of ins-35 was rescued by addition of TMA (2 mM) to dmsC bacteria ( Figure S4a). We next investigated the role of ILS pathway in dauer formation. Unlike wild-type N2 worms, ins-35(ok3297) mutant worms did not show significant alterations in dauer formation when exposed to TMA in the presence of dmsC bacteria compared to worms fed with K12 bacteria ( Figure S4b). These results suggest that TMA modulates the expression of insulin-like peptides, including INS-35, to influence dauer formation in C. elegans. Overall, our results suggest that TMA potentially triggers a neuroendocrine signal that modulates DAF-16 activity to influence dauer formation and lifespan. These results support the hypothesis that bacteria with reduced TMA levels are sensed in part through TYRA-3 and DAF-11 to modulate lifespan in C. elegans.
TMA is a colorless tertiary amine and has a strong "fishy" odor (Messenger et al., 2013). In humans, TMA is produced by gut bacterial metabolism and is associated with trimethylaminuria (odorous sweat), bad breath vaginal odor due to bacterial vaginosis, and pathology of AD (Messenger et al., 2013). Bacterial TMAO and DMSO reductases are both shown to reduce TMAO to TMA, but they are genetically distinct from each other. E. coli DMSO reductase is a trimeric enzyme composed of an integral membrane anchor (dmsC) and a catalytic dimer (dmsAB). These enzymes catalyze the reduction of a number of S-and N-oxide compounds, including DMSO and TMAO. DMSO reductase activity is conserved among a wide variety of micro-organisms including prokaryotes and eukaryotes, aerobes and anaerobes (Zinder & Brock, 1978). Our results demonstrate that single-gene mutant bacteria strains from the Keio library (Baba et al., 2006) can be used in conjunction with C. elegans to successfully identify physiologically relevant bacterial signals, including different type of amines. We demonstrated here that TMA from bacteria acts as a nutrient signal for C. elegans that influences development by activation of a conserved guanylyl cyclase pathway in neurons and regulates the ILS pathway through the peptide INS-35. TMA-derived metabolites are associated with increased risk for type II diabetes, cardiovascular disease, and atherosclerosis in humans (Gipson et al., 2008;Z. Wang et al., 2011). Inhibiting the production of TMA by targeting the microbial enzymes is considered as potential therapeutic approach for TMA-associated metabolic diseases (Z. Wang et al., 2015). Given the importance of nutrient signaling pathways in modulating these diseases, we speculate that TMA may also modulate these conserved signaling pathways to alter host physiology in higher organisms.
Microbes play a critical role in various host functions, including nutrient responses and immunity, and have also been linked to various human diseases (J. Wang et al., 2013). Many species have evolved to have a commensal relationship with microbes by physically harboring certain microbial species, while some species like C. elegans possess elaborate systems to seek these microbial products from their environment as a source of nutrients. Our data suggest that bacterial signals can directly modulate nutrient signaling pathways critical for development, aging, and neurodegeneration in worms (Figure 6), pathways which are also strongly linked with diseases such as atherosclerosis and diabetes in humans. The evolution of the gut from a relatively simple tube in ancient cyclostomatida to the complex gastrointestinal tract of mammalian species has led to a concomitant increase in diversity and complexity of microbial populations. The use of C. elegans with a reduced microbial complexity can help understand how specific bacterial metabolites can regulate host physiology and behavior.

| E. coli strains
E. coli mutant strains, including the parent K12 BW25113 strain, were obtained from the Keio mutant collection (Baba et al., 2006).
The strain library was shipped and stored as a frozen stock in 96well plates. Stock plates were pin-replicated to produce a working copy plate containing Luria Broth (LB) and 25 μg/ml kanamycin.
For assays used in this study, E. coli strains were stab-cultured and grown overnight with shaking at 37°C in minimal media as described below. Overnight cultures were seeded on minimal media plates and dried at 37°C for 12 hours prior to transfer of C. elegans.

| C. elegans and E. coli media
Luria broth (LB) and nematode growth media (NGM) were used to maintain stock cultures of E. coli and C. elegans strains, respectively.
Minimal media was used for assays in this study as described previously (Kim et al., 2014). Liquid media for bacterial growth contained

| PCR confirmation of dmsC mutant
Bacterial strains for kanamycin cassette confirmation were streaked onto LB plates containing 25 μg/ml kanamycin. Strains were colony-purified, and genomic PCR was performed as follows: Primer sets were developed for individual knockout strains encompassing sequences from outside the target gene. The forward primer was designed to be within 200 bp upstream of the 5' end and the reverse primer to be within 200 bp downstream of the 3' end of the target gene. Thermal cycler parameters were as follows: initial 5 min at 94°C, then 25 cycles of 30 s at 94°C, 10 s at 56°C, 1 min at 72°C, and final 5 min at 72°C. Products were run on an agarose gel, and amplified products were examined for appropriate length.
Plates were air-dried and stored overnight at 4°C. Prior to the assay, 4 µl of 2 mM TMA and control solvent (water) was freshly spotted equidistant from the center of the plate and was air-dried at 20°C for 10 minutes without lids. Age-synchronized day 1 adult worms were washed and prepared as previously described (Glater et al., 2014).
Worms were placed in the center of the plate, and positions of the worms were observed using stereomicroscope and images taken at 15-minute intervals. The chemo-attraction index was documented as described previously (Lans et al., 2004). washed and prepared as previously described (Glater et al., 2014).

| Bacterial choice assay
The worms were placed in the center of the plate, and positions of the worms were observed using stereomicroscope and images taken at 15-minute interval. The bacterial choice index was documented as described before (Glater et al., 2014).

| Lawn-leaving assay
Age-synchronized day 1 adult animals (n = 20 per plate) were transferred to bacterial lawns cultured under conditions as described above. Assay was performed for 5 minutes, and lawn-leaving events were manually counted by observing under the microscope and averaged to lawn-leaving events per minute as previously described (Bendesky et al., 2011).
Animals were grown at 15°C for 2-3 days and changed to higher temperature for one day, unless otherwise noted. The fluorescence intensity and/or location was captured using reflected light fluorescence microscopy (Olympus IX3), and the images were processed for densitometry analysis using ImageJ.

| Lifespan assay
Late L4 larvae growing at 15°C were transferred to fresh minimal media plates with FUdR (5 µg/ml) added to the bacterial lawn as indicated in "Results." The first day of adulthood is designated as day 1 in survival curves. Animals were scored as alive, dead, or lost every other day. Animals that failed to display touch-provoked movement were scored as dead. Animals that died from causes other than aging, such as sticking to the plate walls, internal hatching, or bursting in the vulval region, were scored as lost. Animals were transferred to fresh plates and fresh E. coli every 2 days with TMA added to have final concentration 10 mM. Plates were dried for 15 minutes at 20°C before the worms were transferred. All lifespan experiments were performed at 21.5°C. Survival curves were plotted, and statistical analyses (log-rank tests) were performed using Prism 6 software (GraphPad Software, Inc., San Diego, CA, USA). Statistical significance between lifespan curves was determined by p-value of <0.05.

| RNAi knockdown of gene expression
RNAi bacterial strains expressing double-stranded RNA that inactivates specified genes were cultured and used as previously described (Timmons et al., 2001). Briefly, eggs isolated from synchronous populations of cultures were placed on fresh RNAi IPTG and antibiotic-supplemented plates and allowed to grow at 15°C; 3 days later, L4 molt nematodes were used for assays as mentioned in "Results."

| Pharyngeal Pumping
Worms were observed using stereomicroscope at 40x magnification. Pumps per minute were counted as described previously (Raizen et al., 2012).

| Statistical Analysis
Student's two-tailed t test was used when comparing two sets of data. One-way ANOVA and multiple comparison test were performed for comparing between groups using Prism 6.0. Survival curves were generated using Prism 6.0, and statistical analysis was performed using log-rank t test.

ACK N OWLED G EM ENTS
This work was funded by grants from the American Federation of Aging Research (P.K.), NIH grants R01AG038688 and R01AG061165 (P.K.) and the Larry L. Hillblom Foundation. We would like to thank members of the Kapahi and Lithgow labs for discussions and suggestions.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict.

AUTH O R CO NTR I B UTI O N S
AK and PK designed the study. AK, MV, and JK did the dauer and lifespan assays. MC helped with transgenic strain generation. DS analyzed data, drafted manuscript, and addressed reviewers' queries.
SC and AR helped with data analysis. CN and RB helped with expression data analysis. HP, JA, and GL gave valuable inputs in the drafting of the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that supports the findings of this study are available in the supplementary material of this article.