C. elegans electrotaxis behavior is modulated by heat shock response and unfolded protein response signaling pathways

The nematode C. elegans is a leading model to investigate the mechanisms of stress-induced behavioral changes coupled with biochemical mechanisms. Our group has previously characterized C. elegans behavior using a microfluidic-based electrotaxis device, and showed that worms display directional motion in the presence of a mild electric field. In this study, we describe the effects of various forms of genetic and environmental stress on the electrotactic movement of animals. Using exposure to chemicals, such as paraquat and tunicamycin, as well as mitochondrial and endoplasmic reticulum (ER) unfolded protein response (UPR) mutants, we demonstrate that chronic stress causes abnormal movement. Additionally, we report that pqe-1 (human RNA exonuclease 1 homolog) is necessary for the maintenance of multiple stress response signaling and electrotaxis behavior of animals. Further, exposure of C. elegans to several environmental stress-inducing conditions revealed that while chronic heat and dietary restriction caused electrotaxis speed deficits due to prolonged stress, daily exercise had a beneficial effect on the animals, likely due to improved muscle health and transient activation of UPR. Overall, these data demonstrate that the electrotaxis behavior of worms is susceptible to cytosolic, mitochondrial, and ER stress, and that multiple stress response pathways contribute to its preservation in the face of stressful stimuli.


Results
Paraquat-induced oxidative stress reduces the electrotaxis speed of C. elegans. Oxidative stress is known to affect the behavioral responses of C. elegans 17,18,25 . To investigate whether this form of stress can also alter electrotaxis in animals, we used paraquat (PQ), a herbicide that disrupts mitochondrial function by inducing the generation of superoxides 26,27 . High concentrations of PQ have a detrimental effect on the lifespan of worms and cause damage to dopaminergic (DA) and other neurons by increasing ROS levels 17,18,[26][27][28] . We examined the effect of PQ-induced stress on electrotaxis by exposing animals to different doses of the chemical. While exposures to low or moderate concentrations ranging from 50 to 125 µM had no effect on the movement of day-1 adults, 250 µM of PQ caused a significant defect, resulting in a slower speed in the microfluidic channel (Fig. 1A). In addition to electrotaxis, PQ-exposed animals showed increased expression of mitochondrial chaperones. Specifically, the mitochondrial stress response reporters hsp-6::GFP (HSP70 family) and hsp-60::GFP (HSP60 family) were significantly upregulated (Fig. 1B,C).
To determine whether PQ affects other processes, we examined the expression of two other chaperones for ER and cytosolic stress, namely HSP-4 (BiP/GRP78 family member, ER stress) and HSP-16.2 (HSP16 family, cytosolic stress) following PQ treatment. Although the GFP reporters for both these chaperons showed a statistically significant increase in fluorescence, the fold change of the ER GFP reporter was much higher than that of mitochondrial reporters (Fig. 1D,E). Similar to changes in fluorescence reporters, the transcript analysis also showed higher levels of hsp-6 and hsp-16.2 following PQ exposure (Fig. 1F). We did not see a significant increase in hsp-4 transcript. These results are consistent with the previously reported finding that PQ induces both mitochondrial and ER stress in different animal models [29][30][31] . Overall, PQ causes multiple stresses in C. elegans that contribute to a decrease in electrotaxis speed.

Mutations in genes regulating mitochondrial function affect electrotaxis.
To further investigate whether the electrotactic response is perturbed by mitochondrial dysfunction, we conducted a survey of mutants affecting mitochondrial processes and UPR MT . The set of genes consisted of an ortholog of ubiquinolcytochrome C reductase (isp-1, Rieske iron sulfur protein subunit of mitochondrial complex III), a subunit of mitochondrial complex I (gas-1), a subunit of mitochondrial complex II (mev-1, cytochrome b), PTEN-induced kinase 1 (pink-1), mitochondrial uncoupling protein (UCP) ortholog (ucp-4), two mitochondrial superoxide dismutase orthologs (sod-2 and sod-3), coenzyme Q7 ortholog (clk-1), and atfs-1. Some of these genes are reported to affect motility of C. elegans on solid media and thrashing in liquid environment [32][33][34][35] .
We chose a subset of mitochondrial genes for further investigation. Specifically, we explored the possibility that mutant animals may not exhibit an electrotaxis phenotype on their own, but may do so when subjected to further stress. Consistent with this, mev-1(kn1) and pink-1(ok3538), which show sensitivity to PQ 18,38 , displayed electrotaxis defects at exposure concentrations that do not alter the speed of control animals. Specifically, mev-1(kn1) animals exhibited speed deficits when cultured under chronic exposures to either 50 μM or 75 μM PQ (Fig. 2B). Similarly, pink-1(ok3538) animals were also affected, displaying significantly reduced electrotaxis speed at 125 μM PQ (Fig. 2C). Interestingly, ucp-4(ok195) adults displayed increased sensitivity to PQ under our chronic exposure paradigm, exhibiting slow electrotaxis at 75 μM (Fig. 2D). Thus, PQ-induced stress and mitochondrial defects perturb C. elegans electrotaxis speed.
To further investigate the effect of perturbations in UPR MT on electrotaxis, we examined the phenotypes of two different atfs-1 alleles, a genetic null atfs-1(gk3094) and a gain-of-function atfs-1(et15). While the gk3094 mutation prevents the UPR MT response, an opposite phenotype is caused by et15 [40][41][42] . Additionally, the HSR is activated in gk3094 animals 41 . Both atfs-1 mutants displayed significantly slower speeds than wild-type controls, although the null allele phenotype was most severe (Fig. 2E). We also knocked down atfs-1 in isp-1 and gas-1 mutant strains and found that while the electrotaxis speed of isp-1(qm150) was reduced, there was no change in gas-1(fc21) animals (Fig. 2F). Collectively, these data demonstrate that electrotaxis speed is sensitive to disruptions in mitochondrial function, UPR MT , and HSR. Impairment of UPR ER causes electrotaxis speed deficits. Our finding that PQ induces the expression of ER chaperone hsp-4::GFP (Fig. 1C) led us to investigate the impact of ER stress on electrotaxis using tunicamycin, a chemical that promotes protein misfolding by inhibiting protein glycosylation 43,44 . As shown in Fig. 3A, chronic exposure to tunicamycin at concentrations of 5 μg/mL or greater significantly reduced electrotaxis speed. These results are supported by the analysis of hsp-4::GFP reporter expression and hsp-4 transcript (Fig. 3B,C), suggesting the detrimental effect of tunicamycin exposure 40 . As expected, hsp-6 level was unchanged (Fig. 3C). We did not test electrotactic responses of animals at concentrations higher than 10 μg/mL due to significant toxicity, as judged by the highly pronounced larval lethality and slow growth of escapers at this concentration (data not shown).    (E) pek-1: n = 18, pek-1 + 2 μg/mL Tuni: n = 20, ire-1: n = 17, ire-1 + 2 μg/mL Tuni: n = 21, atf-6: n = 33, atf-6 + 2 μg/ mL Tuni: n = 44. One-way ANOVA was performed followed by Dunnett's post hoc test for A and B, which showed that all conditions except one (i.e., control vs. 2 µg in panel B, p = 0.7410) were significant. Student's unpaired t-test was used for the remaining drug conditions. Data in C was analyzed using one-way ANOVA with Tukey's post hoc test.  22 . Therefore, we examined whether tunicamycin-induced ER stress will affect these neurons in our assay. The analysis of dat-1::YFP reporter revealed visible defects in neuronal morphology in treated animals illustrated by dendritic/ axonal abnormalities and loss of cell bodies (Fig. S1A). A similar result was obtained following PQ treatment (Fig. S1B). These data suggest that toxin-induced neuronal damage may contribute to electrotaxis deficits in animals. www.nature.com/scientificreports/ We further characterized ER disruption by testing the phenotypes of UPR ER pathway mutants, specifically ire-1, pek-1, atf-6, and xbp-1. The speed was significantly lower in ire-1(v33), xbp-1(zc12), and pek-1(ok275) but not atf-6(ok551) animals compared to the control, (Fig. 3D). Interestingly, the defect was most pronounced in xbp-1 mutants, which may be due to the gene being regulated by multiple UPR ER arms. We also examined the sensitivity of UPR ER mutants by exposing them to tunicamycin. Treatment with 2 μg/mL caused a significant decrease in the mobility of all three strains (Fig. 3E). A higher concentration (5 μg/mL) was found to be toxic to worms, resulting in lethality (data not shown). Taken together, the results show that electrotaxis is susceptible to ER stress-inducing conditions, and all three UPR ER transducers function to facilitate this behavior.
Next, the role of pqe-1 was investigated based on its potential link to ER stress. Earlier, pqe-1 mutants were reported to cause an increase in transgene expression and enhance polyglutamine (poly-Q) neurotoxicity 23,24 . We speculated that both these phenotypes may be caused by proteotoxic stress due to a failure to regulate global protein synthesis. In support of this, pqe-1(ok1983) animals showed an increase in the expression of hsp-4 chaperon and xbp-1 (Fig. 4A, Fig. S2A). Interestingly, hsp-6 and hsp16.2 levels were also upregulated (Fig. 4A). As expected, the electrotaxis speed of mutant animals was significantly reduced (Fig. 4B), and the phenotype was exacerbated by treatment with PQ and tunicamycin (Fig. 4C,D). The sensitivity towards tunicamycin was particularly pronounced; although wild-type animals showed no abnormal electrotaxis phenotype at 2 μg/mL tunicamycin, pqe-1 mutants exhibited a significantly reduced speed (compare Fig. 4D with 3A). Concentrations of tunicamycin above 2 μg/mL induced growth arrest and lethality in pqe-1(ok1983) worms (data not shown). The mutant animals also exhibited an increased degeneration of dopaminergic neurons and shorter lifespan (Fig. S2B,C).
We investigated whether reducing protein translation would lower proteotoxicity leading to a rescue of the electrotaxis defect of pqe-1 mutant animals. This was tested by knocking down the eIF4G homolog, ifg-1, that encodes a component of the eIF4 complex 45 . The results showed that ifg-1 RNAi was unable to suppress the phenotype of pqe-1 mutant (Figs. 4E, S3). Overall, these observations provide the first evidence that pqe-1 plays an essential role in maintaining multiple stress response pathways and electrotactic response of C. elegans.
Chronic heat impairs the electrotactic response. Studies have shown that heat treatments can have both positive and negative effects on animals due to activated HSR 7,46 and UPR ER7,47 . We investigated the electrotactic movement of young adults that were subjected to acute and chronic heat conditions. Worms exposed to two types of heat pulses, namely 5 h at 25 °C and 1 h at 33 °C, showed no change in their speed (Fig. 5A). Three additional treatments with a longer exposure time (i.e., 8 h each at 15 °C, 25 °C, and 28 °C), did not affect the speed either (Fig. 5B); however, all three heat shock chaperons (hsp-4, hsp-6, and hsp-16.2) were significantly upregulated (Figs. 5C, S4), suggesting the activation of UPR pathways. By contrast, chronic heat treatment at 28 °C for 3 days significantly reduced the speed (Fig. 5D). This demonstrates that unlike shorter duration (up to 8 h) of heat pulses, prolonged heat stress affects the electrotaxis speed of C. elegans.
One of the ways by which heat can induce stress is by interfering with the function of the heat shock transcription factor (HSF) 7 . In C. elegans, the HSF ortholog hsf-1 is transcriptionally activated by a variety of stress conditions, including heat exposure 48 . In support of hsf-1′s involvement, we found that hsf-1(sy441) mutants have a significant electrotaxis speed deficit (Fig. 5E), which was further enhanced by a 1 h heat pulse at 33 °C but not by 5 h at 25 °C (Fig. 5F). There was no change in hsp-6 and hsp-16.2 levels in hsf-1 mutant animals, although hsp-4 was significantly down-regulated (Fig. 5G). Given that hsf-1 regulates the expression of a large number of genes 49 , we speculate that other heat shock chaperon(s) might mediate hsf-1 function in electrotaxis. We also found that the hsf-1(sy441) phenotype was exacerbated by PQ-induced stress and L1 starvation (Figs. S5, S6), which indicates that hsf-1 is a key regulator of diverse stress-induced responses.

Electrotactic movement decreases with reduced diet but increases after exercise. Another
form of stress that we tested is starvation. Initially, the effects of two different starvation exposure protocols on electrotaxis were investigated. In one case, L1 larvae were subjected to starvation for 168 h and subsequently allowed to resume development in the presence of food. In the second case, day-1 adults were exposed to acute starvation conditions. Both nutrient deprivation treatments showed that the effect on electrotaxis was comparable to that of the controls (Fig. 6A,B).
We also investigated a chronic form of nutrient deprivation that involves reduced daily caloric intake. This treatment, termed 'dietary restriction (DR), ' has been shown to be beneficial in animals such as lifespan extension, increased metabolic fitness, and reduced severity of age-related physiological decline [50][51][52] . To this end eat-2 mutants were examined. eat-2(ad1116) animals have a longer lifespan that is attributed to chronic DR due to their slower pharyngeal pumping 33 . We found that the mutants had a significantly slower speed, suggesting that DR affects electrotaxis behavior (Fig. 6C). Since eat-2 mutants show higher fluorescence based on hsp-4::GFP reporter analysis 53 but no change in hsp-6::GFP 54 , we conclude that the sensitivity of electrotactic response to dietary restriction is likely due to the detrimental effect of persistent ER stress.
In addition to starvation and DR, we explored the effect of dietary changes on electrotaxis. Two of the bacterial strains that were initially tested were Streptomyces venezuelae and Bacillus thuringiensis. The metabolites secreted by these bacteria cause intracellular stress in C. elegans 55,56 . We found that neither bacteria had an impact on electrotaxis (Fig. S7). Next, worms were fed with two plant pathogens, Agrobacterium tumefaciens and Pectobacterium carotovorum 57-59 and a nonpathogenic fungus, Cryptococcus aquaticus. Plant pathogens have been reported to have a detrimental effect on the lifespan of worms 60 . None of these cultures had an obvious effect on electrotactic movement (Fig. S8). We also tested three commonly used E. coli strains, HB101, HT115, and DH5α, but, once again, observed no change in movement (Fig. S8). These results led us to conclude that electrotaxis behavior is unaffected by the above selected set of microorganisms, although the possibility of reduced nutrient   www.nature.com/scientificreports/ intake in some or all cases cannot be ruled out. In the future, a larger set of microbial diets should be tested to investigate their effect on animals more thoroughly. Finally, we investigated the effect of physical activity on the electrotactic movement of animals. The beneficial effects of exercise in humans and model organisms are well documented. In addition to improving muscle fitness, exercise also promotes neuronal health [61][62][63] . Since swimming activity in C. elegans induces known features of exercise reported in mammalian systems 62,64,65 , worms were allowed to swim in M9 for 30 min daily, starting at the L3 larval stage, for a period of 7 days (see Methods for details). The results showed that animals had a significantly faster electrotaxis speed compared to controls of the same age, although they were slower compared to day-1 controls (Fig. 7A). Consistent with this, all three heat shock chaperons, hsp-4, hsp-6, and hsp-16.2, were upregulated (Fig. 7B). Additionally, we found that exercise caused a significant increase in the proportion of animals with tubular muscle mitochondria 62,66 (Fig. 7C,D). There was no change in dopaminergic neurons (Fig. S9).

Discussion
The experiments described herein are the first to demonstrate that the electrotactic response of C. elegans is affected by mutations and environmental conditions that increase cytosolic, mitochondrial, and ER stress. We found that the electrotaxis speed was altered by treatment with PQ and tunicamycin, two chemicals that are known to negatively impact the locomotory behavior of worms on solid media 17,67 . PQ exposure causes generation of ROS, which activates the expression of multiple chaperones such as hsp-4 (ER stress), hsp-6 and hsp-60 (both mitochondrial stress), and hsp-16.2 (cytosolic stress). The effects of PQ on mitochondrial and ER stress are well documented; hence, the effects of PQ on electrotaxis are likely due to increased stress associated with these two organelles 13,30,68 .
Independent of ROS, our data show that mutations that cause mitochondrial stress also lead to decreased electrotaxis speed. The slow speed of atfs-1(gf) mutants indicates that the UPR MT contributes to maintaining the wild-type behavior. Analysis of ucp-4, mev-1, and pink-1 mutants revealed that exposure to a relatively   36,72 . Our finding that ucp-4 mutants have defects in electrotaxis suggests that the movement response of animals may be sensitive to ROS levels. Future studies directly measuring oxidative damage following PQ exposure in different genetic backgrounds will be helpful in illuminating the relationship between ROS and electrotaxis. isp-1 and gas-1 mutants exhibit reduced electrotaxis speed. While mutations in both these genes are also associated with electron transport hindrance and increased ROS generation, the isp-1 mutant actually exhibits oxidative stress resistance due to low oxygen consumption and increased expression of mitochondrial superoxide dismutase SOD-3 36 . On the other hand, gas-1 mutants are hypersensitive to oxidative stresses such as PQ 36,37 . The lifespan phenotypes of isp-1 and gas-1 mutants are also opposite, wherein isp-1 animals have a long lifespan and gas-1 are short-lived, possibly due to differences in ROS production and ROS-associated enzymatic activity 36 . These data suggest that electrotaxis speed is independent of pathways involved in lifespan maintenance. The isp-1(qm150) and gas-1(fc21) also both activate the UPR MT73,74 , which led us to investigate the role of atfs-1. We found that atfs-1 null mutant and gain-of-function mutant had a slower speed in our assay. Moreover, RNAi knock-down of atfs-1 reduced the speed of isp-1 mutant but not gas-1 mutant animals. Together with known roles of atfs-1 in regulating the expression of mitochondrial and cytosolic heat shock chaperons, these results support the conclusion that multiple stress response signaling affect the electrotaxis behavior of C. elegans.
The effect of ROS induced by PQ and mitochondrial mutants could also impact UPR ER . Thus, we investigated the responses of animals treated with tunicamycin and observed defects in electrotaxis speed that were accompanied by an increase in the expression of the ER chaperone GRP78/BiP homolog hsp-4. Similar electrotaxis www.nature.com/scientificreports/ phenotypes were also observed in UPR ER mutants, indicating that the ER genes play a crucial role in mediating normal movement behavior. The analysis of mutants affecting the three major arms of UPR ER revealed that while ire-1, pek-1, and xbp-1 are needed to maintain a normal response, atf-6 is not essential. Differences between UPR ER pathway components in mediating some of the biological processes have been reported earlier. Shen et al. (2001) found that while knockdown of pek-1 and ire-1 affected larval development, there was no such phenotype in the case of atf-6 75 . Additionally, ire-1 and xbp-1 mutants have a short lifespan and slower development, but pek-1 and atf-6 mutants appear normal 56,76 . Finally, animals lacking ire-1, xbp-1, and atf-6 function show high sensitivity to pore forming bacterial toxin Cry5B but pek-1 does not appear to be involved 46 . In spite of these differences, we found that increasing the ER stress load by tunicamycin treatment exacerbated the electrotaxis defects of all three UPR ER pathway mutants. In addition to the known UPR ER genes, we also tested the less well-characterized gene pqe-1. Mutations in pqe-1 are thought to cause proteotoxic ER stress, possibly due to the upregulation of global protein synthesis 77 . Consistent with this idea, our results showed that pqe-1 mutants have an abnormal electrotactic response. Studies in yeast have shown that the REXO1 (RNA exonuclease homolog 1) domain of PQE-1 functions in the processing of the 3′ ends of rRNA and tRNA 78,79 , which may suggest a role of pqe-1 in the regulation of protein translation. Since pqe-1 localizes to the cell nucleus 23,24 , it is possible that it affects genes whose products modulate the function of translation machinery. To investigate this further, we knocked-down the eIF4G homolog, ifg-1, a component of the initiation factor 4F (eIF4F) complex 45 , but saw no suppression of the pqe-1 mutant phenotype. This result combined with high sensitivity of mutant animals to tunicamycin and changes in the expression of multiple chaperons and xbp-1 lead us to suggest that pqe-1 plays an essential role in maintaining both HSR and organelle-specific stress response (UPR) pathways in C. elegans.
One of the ways by which stress can affect electrotaxis is by causing damage to neurons. Consistent with this, animals exposed to PQ and tunicamycin exhibited degenerated dopaminergic neurons (Fig. S1). A similar phenotype was also observed in pqe-1 mutants (Fig. S2). These findings, along with our previous work demonstrating the role of dopaminergic neuron signaling in mediating electrotaxis behavior 22 , led us to conclude that neuronal defects may in part contribute to the electrotaxis phenotype in animals as a result of stress-inducing conditions.
In addition to examining the effects of genetic mutations and compounds, we investigated whether environmental perturbations alter the electrotactic response. Among other effects , heat is reported to induce mitochondrial stress 7 . High temperatures have also been shown to modulate ER and cytosolic stress. This may involve multiple mechanisms, such as activation or repression of PERK depending on the severity of the heat regimen 47 and changes in global protein synthesis and apoptosis 47 . We exposed animals to acute heat and found that their electrotaxis was unaffected. Chronic heat, however, severely reduced the speed, suggesting that behavior is affected by prolonged heat exposure, which is consistent with cellular stress playing a role in modulating electrotaxis. The resistance to heat-induced stress appears to depend on the transcription factor hsf-1, since hsf-1 mutants exhibit electrotaxis defects and are sensitive to heat treatments.
As diet can also play a role in the UPR maintenance 53,80 , we subjected worms to different forms of starvation. When C. elegans are starved during the L1 larval stage, they enter L1 arrest, which halts the development and reproductive growth, enhances stress resistance, modifies feeding behavior, and alters metabolic flux 81,82 . Starvation during the L4 larval and adult stages has beneficial effects such as extended life span, germline cell reduction, delayed reproduction, and thermotolerance 83 . At the cellular level, starvation induces both mitochondrial and ER stress 53,80,84 . Mitochondria become fragmented in starved animals and are cleared through mitophagy 10,80,85 . Zhang et al. 86 reported that starvation caused ER stress and led to activation of PERK/eIF2α in an astrocyte cell line. Other groups have shown changes in the regulation of different ER proteins, such as chaperones 84 . Our work demonstrates that short-term DR does not affect the electrotactic response. However, eat-2 mutants that are chronically starved, and therefore mimic a long-term DR condition, show a significantly slower electrotaxis speed. It is conceivable that a reduction in speed may aid in the long life span of eat-2 mutants as the animals are slow feeders 53 . This line of reasoning agrees with previous findings that young adults lacking eat-2 function move slower on solid media 33 . Furthermore, eat-2 mutants show increased expression of hsp-4 53 but not hsp-6 54 , suggesting higher ER stress levels.
The last stress condition we investigated was a daily exercise regimen. The benefits of exercise on muscular and neuronal health are well demonstrated. Healthier muscles through exercise allow animals to increase their athletic capacity 63 . At the molecular level, exercise activates the UPR ER response as a protective mechanism 87 . Beneficial effects of exercise have also been demonstrated in C. elegans, such as a longer lifespan and improved neuronal and muscular health 62,64,65,88 . Laranjeiro et al. (2017) reported that the ER chaperone hsp-4 is downregulated in animals subjected to exercise, which suggests a conserved role of UPR ER in exercise-induced benefits 64 . Our lab showed earlier that defects in muscles reduce the electrotaxis speed of C. elegans 22 . Consistent with the above studies, we found that daily exercise treatment significantly improved the electrotaxis speed of C. elegans and resulted in increased expression of heat shock chaperons as well as better preservation of mitochondrial morphology.
In summary, the work described in this paper shows that electrotactic movement of worms is reduced by mutations in genes that affect HSR, UPR MT , and UPR ER as well as following chronic exposure to chemicals, heat, and reduced diet. In contrast, short bursts of daily exercise are beneficial since they resulted in a higher speed of animals and better preservation of mitochondrial morphology. Collectively, these findings lead us to conclude that conditions that elevate cellular stress for prolonged periods cause detrimental effects, whereas transient stress imparts a beneficial effect on health. The results established that a microfluidic-based assay is a reliable output of locomotory circuits and demonstrate that electrotaxis can be used to study stress response pathways in C. elegans.  89 . Except where indicated, animals were grown and maintained at 20 °C on nematode growth medium (NGM) agar plates containing E. coli OP50 culture using previously described methods 15 . All experiments used agesynchronous populations obtained by bleach treatment 15 . Experiments involving wild-type and untreated controls were done using young adults unless stated otherwise. Growth times for all animals bearing mutations or undergoing treatments that affect developmental rate were appropriately adjusted.
Assays and treatments. The following methods were used to analyze worms in this study. Additional details are provided in supplementary methods.
Chemical treatments. Paraquat dichloride was obtained from Sigma-Aldrich (St. Louis, MO, USA) and tunicamycin was obtained from Bioshop Canada Inc (Burlington, ON, CA). Paraquat treatments were first prepared as 20 × solutions in M9 buffer; subsequently, 1 × treatment plates were produced by spreading 500 μL of the 20 × solutions across the surface of plates containing 10 mL of NGM agar. Tunicamycin treatments were first prepared as 5 mg/mL stock solutions in 100% DMSO, then diluted with water to make 20 × solutions; subsequently, 1 × treatment plates were produced by spreading 500 μL of the 20 × solutions across the surface of plates containing 10 mL of NGM agar. For chemical treatments, synchronized worms were grown on chemical-containing plates from L1 until adulthood. Young day-1 adults, based on direct observation, were used for experiments.
Heat stress. Animals were first grown at 20 °C until young adulthood. Plates were then sealed with parafilm and exposed to specific temperatures in a water bath for desired durations. Control plates were also sealed but kept at 20 °C. After the treatment, parafilm was removed and animals were allowed to recover at 20 °C for 30 min to an hour before electrotaxis assays. For 28 °C treatment, synchronized larvae were allowed to grow at 20 °C till day-1 adulthood and then shifted to 28 °C for 3 days.
Feeding bacterial and fungal cultures. C. elegans plates were seeded with overnight grown cultures of microorganisms. For E. coli, plant pathogens and fungus, L1 animals were transferred on plates and allowed to grow till adulthood. Day-1 adults were used for electrotaxis assays. In the case of Streptomyces venezuelae and Bacillius thuringiensis, L1 larvae were first fed with E. coli OP50 till adulthood. One day-old adults were transferred on maltose yeast extract medium (MYM) plates containing culture of S. venezuelae for 24 h. Bacillus thuringiensis culture was grown on NGM agar and fed to one day-old adults for 24 h. Electrotaxis was performed the following day. Control worms were grown in the presence of OP50 culture on NGM-Agar and MYM-Agar plates.
Starvation. Animals were first grown on OP50 until day-1 adulthood. They were then transferred to an unseeded NGM agar plate (no food) for 8 h to mimic acute starvation. Following the starvation period, electrotaxis assay was performed.
Exercise. Worms were subjected to exercise starting the L3 larval stage until day-6 adulthood (for a total of 7 days). Two different treatments were performed, each of which consisted of 30 min of daily swimming, followed by recovery on LB-agar plates. The first exercise regimen involved a belly dancer shaker in which worms, suspended in M9, were subjected to 100 rpm rotation. The second regimen consisted of keeping worms in M9 on the countertop without shaking, thereby allowing them to perform natural swimming.
Microchannel fabrication and electrotaxis assay. Microfluidic channels were fabricated as previously described 21,90 . The channel design was printed on a transparency sheet using high-resolution photoplotting to create a photomask, which was then used in conjunction with SU-8 100 negative photoresist (MicroChem Corp., MA, USA) to lithographically pattern the design onto a silicon wafer. Microchannels were then casted by pouring polydimethylsiloxane (PDMS) pre-polymer (Sylgard 184 Kit, Dow Corning Corp., MI, USA; 10:1 ratio of base and cross-linker) onto the resultant master mold and allowing 24 h for curing. The channel was then excised from the PDMS replica and fluid access ports were punched into each end. Next, the channel, a blank PDMS strip and a glass slide were oxidized via exposure to oxygen plasma for 40 s at 40 W power and stuck together to seal the microchannel. Lastly, plastic tubing and insulated copper wire were affixed to the punched reservoirs and secured with PDMS pre-polymer.
The electrotaxis assay protocol has also been described 21,90 . A syringe was attached to one of the inlet/outlet tubes of the PDMS microchannel to facilitate worm loading at the other tube. A power supply was connected via insulated copper wiring to the electrodes of the microchannel device to provide worms with electrical stimulus. A microscope, camera and monitor allowed visualization and recording of the electrotaxis experiment. www.nature.com/scientificreports/ In preparation for the assay, worms were washed off of their culture plates, cleaned, and suspended in M9 buffer. Animals were then aspirated into the channel using the syringe pump. Both tubes were then laid flat at the same elevation to eliminate pressure-induced flow. Next, a 3 V/cm DC electric field was applied, and the worm's resultant behaviour recorded by camera.
Electrotaxis was carried out for up to 5 min. Animals were allowed to travel a minimum distance of 5 mm in one direction, towards the cathode, after which the field polarity was reversed to induce a turning response. Locomotory data was extracted from recorded videos using custom MATLAB-based worm tracking software. Electrotaxis speed data was plotted in box plots.
Fluorescence microscopy and quantification. Animals were mounted on 2% agar pad containing glass slides. Before placing the cover slip, they were anesthetized using a 15 μL drop of 30 mM NaN 3 mixed with a drop of M9. GFP fluorescence was visualized using a Zeiss Observer Z1 microscope equipped with an Apotome 2 and X-Cite R 120LED fluorescence illuminator. Fluorescence intensity was quantified using NIH ImageJ (http:// rsbwe b.nih.gov/ij/). Neurodegeneration was manually scored by counting the number of cell bodies of DAergic neurons and by assessing neurite morphologies and trajectories. Worms consist of two pairs of cephalic neurons (CEPs), a pair of anterior deirid neurons (ADE) and one pair of posterior dieirid neurons (PDE).
RNAi. The atfs-1 RNAi plasmid (pGLC171) was constructed by inserting a 759 bp genomic fragment into the L4440 vector using the restriction enzymes KpnI and XbaI. The fragment was obtained by PCR using the forward primer GL1649 (5'-AAG GGT ACC CAC TAC TTG GAG AGC GAC GAC -3') and reverse primer GL1650 (5'-AAG TCT AGA CTA CTT CTT GGA ACT CCC TGC -3'). We used the RNAi gene silencing protocol described by Wu et al. 41 Since authors reported that knock-down in parental isp-1 mutant worms caused an arrest of F1 larvae, L4-stage animals were treated with RNAi and day-2 old adults were analyzed. Synchronized worms were grown on OP50 until L4 stage and then transferred to RNAi plates containing either L4440 empty plasmid or atfs-1 RNAi plasmid. atfs-1 knock-down was confirmed by qPCR, which showed a range of ~ 2 × to 10 × reduction in different batches. For the ifg-1 RNAi, animals were fed with bacteria from Ahringer library. In this RNAi paradigm synchronized animals were grown on OP50 until day-1 adult stage and then transferred to ifg-1 RNAi plates for 48 h and analyzed right after.

Quantitative reverse transcription PCR (RT-qPCR).
For RT-qPCR experiments, synchronized cultures were prepared by bleaching adult hermaphrodites twice in two successive generations to obtain highly synchronous cultures. After the second round of bleaching, eggs were grown until the day-1 adulthood. Total RNA was extracted from these animals using trizol (Catalog Number T9424, Sigma-Aldrich, Canada), according to the manufacturer's instructions. The SensiFAST cDNA Synthesis Kit (Catalog Number BIO-65053, MeridianBioscience, Canada) was used to obtain cDNA according to manufacturer's instructions. RT-qPCR was performed (in triplicate) using the Bio-Rad cycler CFX 96 machine. The primers are listed in the Supplementary table (Table S1). The reaction was set up using SensiFAST SYBR Green Kit (Catalog Number BIO-98005, BIOLINE, USA) according to the manufacturer's instructions. The expression levels of the analyzed genes were normalized to pmp-3. Data analysis. GraphPad Prism 7 was used to plot the graphs. For all assays, data from repeat experiments were pooled and analyzed together. Statistical analyses were done using GraphPad Prism 7 except for RT-qPCR experiments that were analyzed using CFX Maestro 3.1 software (Bio-Rad, Canada; https ://www.bio-rad.com/ en-ca/produ ct/cfx-maest ro-softw are-for-cfx-real-time-pcr-instr ument s). P values less than 0.05 were considered statistically significant.

Data availability
All data generated or analysed during this study are included in this published article (and its Supplementary  Information files). Additional details are available from the corresponding author on reasonable request.