Background

Disrupting mammalian sleep during development correlates with negative effects on physical, cognitive, and social health, suggesting that sleep is important for appropriate development [1,2,3]. Nonlethal sleep deprivation also causes lasting neurological and behavioral deficits in Drosophila melanogaster [4]. However, a mechanistic grasp of why inadequate sleep during development is particularly deleterious is lacking.

Caenorhabditis elegans exhibits developmentally timed sleep during lethargus, a 2- to 3-h-long period at the termination of each larval stage [5,6,7]. Similar to mammalian sleep, lethargus is characterized by locomotion and feeding quiescence, sensory gating, a typical posture, rebound sleep, and deeply conserved regulation [6, 8,9,10,11,12,13,14,15].

Severe sleep deprivation activates DAF-16/FoxO, the C. elegans Forkhead box O (FoxO) transcription factor. FoxOs function broadly in regulating metabolism, lifespan, and responses to environmental stressors [16,17,18,19,20,21,22,23,24]. Nuclear translocation of DAF-16/FoxO is inhibited by the insulin/insulin-like growth factor signaling (IIS) pathway [17, 18]. In response to prolonged and continuous sleep deprivation, DAF-16 translocates to the nucleus to mitigate or delay lethality [10]. DAF-16 is also required for rebound sleep following much weaker disruptions [11].

A particular proteotoxic outcome of sleep deprivation is endoplasmic reticulum (ER) stress. In all species examined, including rodents and flies, expression of the ER chaperone immunoglobulin binding protein (BiP/Grp78) from the heat shock 70 protein family is upregulated upon sleep deprivation [25,26,27,28,29,30]. The C. elegans homolog of mammalian BiP is HSP-4 [31,32,33]. BiP/HSP-4 upregulation requires the action of the ribonuclease inositol-requiring protein-1 (IRE-1), a key receptor for sensing misfolded ER proteins [32,33,34,35,36,37]. IRE-1 signaling activates the XBP-1 transcription factor, which changes the expression of BiP and other genes in the deeply conserved unfolded protein response (UPRER) pathway [25, 28, 30, 38,39,40,41].

Prolonged wakefulness increases daily energy expenditure [42,43,44], for instance, in the brain [45, 46]. Consequently, energy production by the mitochondrial oxidative phosphorylation system is upregulated [25, 30, 47,48,49,50]. Nevertheless, the roles of the mitochondrial UPR (UPRmt) following sleep deprivation are largely unknown, although one study found that sleep deprivation induces mitochondrial chaperones (to a lesser degree than BiP) in rat cerebral cortexes [30].

Upon mitochondrial stress, expression of ubl-5, encoding a ubiquitin-like protein, is upregulated, and UBL-5 plays a key role in activating dedicated chaperones and proteases of the UPRmt [51,52,53,54]. In C. elegans, chemically induced mitochondrial stress upregulates the mitochondrion-specific chaperones HSP-6 and HSP-60 (from the Hsp70 and Hsp10/16 superfamilies) [51, 55].

Tractable model organisms have been prominently used to study responses to environmental stressors, such as oxidation or heat [56, 57]. In contrast, C. elegans sleep deprivation was minimally explored, and its lasting impacts, other than lethality, were never characterized. Here we established an automated approach to inflicting severe yet nonlethal deprivation of developmentally timed sleep in C. elegans. We found that worm sleep deprivation inflicts both mitochondrial and ER stress, as indicated by the triggering of the UPRmt and the UPRER. When the UPRs were genetically impeded, lasting defects in feeding, fecundity, and egg-laying physiology were detected. Moreover, different UPRs protected different tissues from the impacts of deprivation.

To assay feeding, we measured the pumping motion of the pharynx, a neuromuscular organ that takes in bacterial food, expels excess liquid, and passes food to the intestine [58, 59]. Maintaining these functions requires speed and regularity and is energetically costly. The pharyngeal nervous system consists of 20 neurons, is isolated from the rest of the animal by the basal lamina, and can operate independently [60, 61]. The rate of pumping depends on feeding history, quality of food, and endogenous serotonin levels [62,63,64,65]. In addition, pumping can be stimulated with exogenous serotonin [66, 67]. We found that sleep deprivation-induced mitochondrial stress impacts pharyngeal neurons and slows pumping.

In contrast, sleep deprivation-induced ER stress resulted in germ cell apoptosis and abnormal activity in the egg-laying circuit. The key determinant of C. elegans brood size is the number of available sperm [68]. Germ cell apoptosis can be triggered to protect sperm against DNA damage or environmental stressors that are not directly genotoxic. In both cases, highly conserved core apoptotic genes are strictly required for the initiation of programmed cell death [69,70,71,72,73]. One of these is CED-3, a cysteine-aspartate protease essential for execution of apoptosis [74,75,76,77,78,79]. Conveniently, this process can be visualized: the transmembrane receptor CED-1 mediates engulfment of early apoptotic corpses by surrounding sheath cells [80, 81]. Thus, the translational ced-1::gfp reporter is used to indicate the occurrence of germ cell apoptosis [82].

In addition, the egg-laying circuit exhibited a post-deprivation defect similar to the outcome of genetically ablating an egg-laying command neuron [83]. Opposite to the case of the feeding circuit, the UPRER (but not the UPRmt) mitigated the impact of deprivation in the egg-laying circuit. Collectively, these findings implicate two UPRs and insulin/insulin-like growth factor signaling in mitigating the impacts of disrupting worm sleep. They show that developmentally timed sleep is a vulnerable period: external stimuli that are benign outside of lethargus are proteotoxic when administered during lethargus. Adequate sleep promotes normal functions in tissues differing in developmental dynamics and physiological activity, and distinct UPRs mitigate different impacts of nonlethal deprivation.

Results

Substantial deprivation of lethargus quiescence can be automatically inflicted

Forced locomotion inflicted during C. elegans sleep by manually delivering harsh touch was previously shown to be lethal [6, 10]. Considerably more gentle mechanical vibrations can transiently force motion. We previously identified rebound sleep when vibrations were applied infrequently [11] and found that worms desensitized to vibrations delivered too frequently. In response to a 3-min on/off cycle of vibrations, we measured a 50% reduction in mean quiescence (Fig. 1a, b). This indicated that, in contrast to previously published conditions, the 3-min on/off cycle robustly overwhelmed the capacity of the worms to compensate for excess motion. We further found that this disruption affected wild-type animals and daf-16 mutants similarly.

Fig. 1
figure 1

A periodic mechanical stimulus can partially and nonlethally reduce lethargus quiescence and induce translocation of DAF-16. a The fraction of quiescence measured during L4 lethargus of undisrupted (gray) and deprived (orange) wild-type animals and daf-16 mutants. Locomotion was forced using a square wave of mechanical vibrations with a 6-min period and 50% duty cycle. N = 30 animals, shaded areas depict mean ± standard error of the mean (SEM). b The total time in which quiescence was observed integrated over the data presented in a. Error bars depict mean ± SEM. c DAF-16::GFP fluorescence in the intestine before and after 1 h of sleep deprivation. Top: examples of green fluorescent protein (GFP) fluorescence in unperturbed and partially deprived animals. Arrowheads point to bright particles indicating concentration of DAF-16::GFP. Bottom: Histograms of the number of bright particles per animal identified under each set of conditions. Sample sizes are denoted in parentheses. Single and double asterisks denote significant differences with p < 0.05 and p < 0.01, respectively

Full size image

Manually forced locomotion during worm sleep drives translocation of DAF-16 to nuclei of intestinal and body wall muscle cells [10]. We asked whether our 3-min on/off disruptions would induce similar translocation of DAF-16. To address this, we exposed animals expressing fluorescently labeled DAF-16 to 1 h of the vibration stimuli during the first half of the fourth lethargus stage (L4 lethargus). In agreement with the manual (lethal) deprivation protocol, we observed nuclear localization in intestinal cells (Fig. 1c). Translocation was not observed following a mock perturbation protocol, where vibrations were not applied during an equivalent 1-h period. In our hands, a clear translocation response was not observed in body wall muscles.

In contrast to the consequences of continuous manual deprivation [10], we did not observe any molting defects or lethality following our deprivation conditions. Possibly, this was a consequence of not depriving the animals of quiescence for a continuous period that exceeded 3 min. Collectively, these data demonstrate the ability to automatically and severely disrupt quiescence to a stressful yet nonlethal degree.

Considering these results, the experiments described throughout the manuscript employ three types of deprivation conditions: 1 h of disruptions for acute responses in individual animals expressing fluorescent markers (“mock” animals were loaded to identical observation chambers), 4 h of disruptions for gene expression assays in small groups of tightly synchronized animals, and 12 h of disruptions for lasting effects of deprivation assayed in large groups of animals. To control for nonspecific effects of the prolonged stimulation period, we compared sleep-deprived animals to those exposed to vibrations outside of lethargus. These groups were labeled “control” (see Methods section for details). In the latter two protocols (4 h and 12 h), sleep was disrupted for no more than the 3-h duration of lethargus. To identify effects of vibrations that are nonspecific to lethargus, we assayed a subset of strains without exposing them to any vibrations. These groups were labeled “unperturbed,” and they consistently exhibited similar phenotypes to those of the “control” animals. Thus, nonspecific effects of vibrations were found to be minor and genetically separable from impacts of sleep deprivation.

Two unfolded protein responses are triggered by nonlethal deprivation of C. elegans sleep

DAF-16/FoxO is associated with a broad spectrum of stress responses, and ER stress in particular was detected in previously examined sleep-deprived animals [25, 38,39,40]. To address whether the UPRER was triggered by worm sleep deprivation, we used a transcriptional reporter for HSP-4/BiP expression, hsp-4p∷gfp (zcIs4) [33, 51]. Animals expressing this indicator were subjected to a 1-h deprivation protocol. During L4 lethargus, hsp-4 was notably expressed in the epithelial seam of undisrupted animals (Fig. 2a). This array of hypodermal stem cells, termed seam cells, regulate hypodermal/cuticle formation and transform to their adult fate at the time of the fourth molt [7, 84,85,86]. We observed that hsp-4 expression in seam cells coincided with the generation of the alae — the adult cuticular ridges.

Fig. 2
figure 2

Worm sleep deprivation triggers the UPRER. a Example of pre- (top) and post- (bottom) deprivation fluorescence of the hsp-4p::gfp reporter. Prior to deprivation the reporter was prominently observed in the seam cells. b Fluorescence of the hsp-4p::gfp reporter before and after deprivation, mock deprivation, and stimulation of mid-L4 larvae. c Quantification of hsp-4 expression using real-time PCR. Error bars depict 99% confidence intervals (1 biological replicate, 20 animals per sample). d Fluorescence of the hsp-4p::gfp reporter in mutants where the function of the UPRER genes ire-1 and xbp-1 was lost. On these mutant backgrounds sleep deprivation did not upregulate the expression of hsp-4. In all box plots, horizontal lines, boxes, and bars depict medians, 1st and 3rd quartiles, and 5th and 95th percentiles, respectively. Crosses denote outliers. Single and double asterisks denote significant differences with p < 0.05 and p < 0.01, respectively

Full size image

Neither mock deprivation nor 1 h of vibration stimuli prior to lethargus affected hsp-4 expression. In contrast, sleep-deprived animals significantly upregulated the expression of hsp-4 (Fig. 2b). Consistently, hsp-4 expression remained elevated after a 4-h period of disruptions, as indicated by quantitative PCR (Fig 2c). Upregulation of hsp-4 requires essential components of the UPRER including IRE-1 and XBP-1 [33, 34, 87]. Consistent with induction of ER stress by environmental stressors or genetic perturbations, sleep deprivation failed to upregulate hsp-4 expression on ire-1 or xbp-1 mutant backgrounds (Fig. 2d).

The roles of the UPRmt following sleep deprivation are less understood, although evidence of mitochondrial stress was detected in sleep-deprived rodents [30]. Therefore, we similarly used fluorescent indicators to assay the activation of the UPRmt [51,52,53]. The ubl-5 translational reporter of the UPRmt expresses broadly at low levels and brightly in the posterior bulb of the pharynx, the posterior of the intestine, and the anterior edge of the intestine near the pharyngeal-intestinal valve. Following 1 h of sleep deprivation, we observed a small but significant upregulation of intestinal Publ-5::ubl-5::gfp expression. No increase in reporter fluorescence was observed following the mock protocol or when vibrations were applied at the mid-L4 larval stage (Fig. 3a).

Fig. 3
figure 3

Worm sleep deprivation triggers the UPRmt. a Intestinal fluorescence of the ubl-5p::ubl-5::gfp UPRmt reporter before and after deprivation, mock deprivation, and stimulation of mid-L4 larvae. b Intestinal fluorescence of the mitochondrial chaperone hsp-6p::gfp reporter after deprivation and mock deprivation. Horizontal lines, boxes, and bars depict medians, 1st and 3rd quartiles, and 5th and 95th percentiles, respectively. Single and double asterisks denote significant differences with p < 0.05 and p < 0.01, respectively. c Relative expression of ubl-5 and hsp-6 in deprived as compared to undisrupted wild-type animals (2 biological replicates, 20 animals per sample). The mechanical stimulus was applied for 4 h (which included L4 lethargus), and RNA was prepared immediately after this period. Error bars depict mean ± SEM

Full size image

We similarly assayed two additional indicators of the UPRmt: the hsp-6 and hsp-60 transcriptional reporters. Expression of hsp-6 was broad and most clearly visible in the intestine. Strong expression of hsp-6 or an accumulation of the reporter led to bright staining of a posterior segment of the intestine. Sleep deprivation did not affect the posterior bright patch, but expression in the rest of the intestine was upregulated after the 1-h deprivation protocol. Mock deprivation did not affect hsp-6 expression (Fig. 3b). Real-time PCR was used to assay the relative expression of these genes after a 4-h period of administering the disruptive stimuli that included L4 lethargus. Consistently, we observed elevated expression of ubl-5 and hsp-6 in deprived animals (Fig. 3c). In our hands, we could not detect upregulation of hsp-60 post-deprivation (Additional file 1: Figure S1). Combined, these results show that nonlethal deprivation of worm sleep is proteotoxic and induces both mitochondrial and ER stress. Comparable stimulation outside of lethargus did not trigger these UPRs, indicating that the period of sleep is particularly vulnerable.

The UPRmt plays a role in mitigating effects of nonlethal sleep deprivation on pumping

Pharyngeal pumping for the purpose of feeding is an energetically demanding behavior that can readily be quantified [65]. To identify lasting effects of sleep deprivation on pumping, each animal was assayed for 60 min at a food concentration corresponding to an optical density (OD600) = 2.5 of the bacterial suspension, where pumping activity was intermediate (Additional file 2: Figure S2). Since control and unperturbed animals (as defined above) exhibited nearly identical feeding behaviors, deprivation conditions were typically compared to stimulation outside of lethargus (control).

Pharyngeal pumping can be adequately described as bursts of rapid pumping interspersed with pauses [65]. Similar to a previous study, we measured the mean instantaneous pumping rate and the duty ratio of rapid pumping [88]. These summary statistics did not reveal significant differences between deprived and control wild-type animals. However, sleep deprivation reduced pumping in daf-16 mutants, and the native promoter rescue of DAF-16 function restored the wild-type phenotype (Fig. 4a and Additional file 3: Figure S3). These results suggest that worm sleep deprivation can negatively impact feeding in a DAF-16-dependent manner.

Fig. 4
figure 4

Mitigating mitochondrial (but not ER) stress is required for maintaining normal post-deprivation pumping rates. a Box plots of average pumping rates for control (i.e., stimulated outside of lethargus) and deprived animals obtained with the optical tracking method. Loss of DAF-16 or mec-10 expressing neurons confers a broad defect in stress responses or the loss of gentle mechanosensation, respectively. b Average pumping rates of animals deficient in the UPRmt (left) and the UPRER (right). Horizontal lines, boxes, and bars depict medians, 1st and 3rd quartiles, and 5th and 95th percentiles, respectively. Sample sizes are noted in parentheses; asterisks and double asterisks denote significant differences (p < 0.05 and p < 0.01, respectively)

Full size image

To control for potential biomechanical impacts of the vibrations, we assayed worms expressing deg-3(u662), a degeneration-causing constitutively active nicotinic acetylcholine receptor (nAChR) channel subunit, in touch neurons expressing the mec-10 gene [89,90,91,92,93]. These transgenics did not respond to vibrations (Additional file 4: Figure S4). In the absence of mec-10 expressing neurons, vibration stimuli during lethargus did not affect subsequent pumping on both wild-type and daf-16 mutant backgrounds (Fig. 4a and Additional file 3: Figure S3). These data indicate that mechanosensation by mec-10 expressing neurons and the ensuing loss of quiescence are required for pumping fatigue following sleep deprivation.

To test whether the UPRmt plays a role in mitigating consequences of nonlethal sleep deprivation, we examined feeding in ubl-5 mutants. Deprived ubl-5 mutants exhibited a decrease in pumping rate as compared to the control group. This pumping defect was rescued by expressing the Publ-5::ubl-5::gfp translational reporter (Fig. 4b and Additional file 3: Figure S3). In contrast, UPRER defective worms did not exhibit pumping fatigue: animals carrying two ire-1 putative null alleles and hsp-4 mutants did not significantly change their mean pumping rate post-deprivation (Fig. 4b). A mild change in the duty ratio of deprived ire-1 mutants may indicate a weak contribution of the UPRER maintaining post-deprivation pumping (Additional file 3: Figure S3). These results demonstrate that ER proteotoxic stress, unlike mitochondrial stress, does not play a major role in mitigating pumping fatigue following sleep deprivation.

Complementarily, electropharyngeograms (EPGs) enabled us to precisely time contractions and relaxations of the pharyngeal corpus and terminal bulb. We therefore used EPGs to measure durations of individual pumps and to confirm our optical measurements of pumping rates (Additional file 5: Figure S5, Additional file 6: Figure S6, and Additional file 7: Figure S7). We found that repeated mechanical stimuli extended the duration of individual pumps. However, the extension of single pumps was not specific to sleep deprivation; deprived animals and controls exposed to the mechanical stimuli before and after lethargus exhibited similar pump durations. The extended duration of individual pumps in ubl-5 mutants was comparable to that of wild-type animals and daf-16 mutants (Additional file 6: Figure S6). Thus, we could differentiate between deprivation-related and nonspecific impacts of mechanical stimuli, and we did not identify a role for the UPRmt in mitigating the nonspecific effect. Combined, our results indicate that the UPRmt (but not the UPRER) mitigates lasting effects of sleep deprivation on pharyngeal pumping.

Nonlethal sleep deprivation impacts pharyngeal pumping by affecting regulatory neurons

The pharynx is isolated from the rest of the animal and can exhibit pumping tens of minutes after it has been dissected out [60, 61]. Pumping defects induced by sleep deprivation can thus originate from pharyngeal regulatory neurons or pharyngeal muscles. We note that our mechanical stimuli do not noticeably affect the buccal plug, a cap of extracellular material that prevents food from entering the pharynx during lethargus [7], and they do not induce pumping. Consequently, the stimuli do not activate the pharyngeal muscles during lethargus, and “wear and tear” damage caused by anachronistic muscle activation is unlikely.

Serotonin or 5-hydroxytryptamine (5-HT) robustly activates rapid pumping through the action of the neuronally expressed SER-7/5-HT7 receptor. Several labs have shown that 5-HT-induced rapid pumping is abolished in ser-7 null mutants [64, 66, 67, 88]. We similarly activated pharyngeal neurons with 10 mM 5-HT instead of food and assayed pumping fatigue. We found that 5-HT-induced pumping was rapid in undisrupted and deprived daf-16(mgDf50) and ubl-5 mutants (Fig. 5a).

Fig. 5
figure 5

A neuronal deficiency underlies the post-deprivation slowdown of feeding. a Box plots of pumping rates for 5-HT-triggered pumping in undisrupted and deprived daf-16(mgDf50), ubl-5, and ser-1 mutants. b Left: distributions of pumping rates for deprived and control daf-16 mutants where neuronal function of DAF-16 was restored. Arrow points to the indistinguishable left tails of the distributions, indicating similar statistics of sporadic pumping. Right: mean pumping rates under these conditions. Horizontal lines, boxes, and bars depict medians, 1st and 3rd quartiles, and 5th and 95th percentiles, respectively. Sample sizes are noted in parentheses; asterisks and double asterisks denote significant differences (p < 0.05 and p < 0.01, respectively)

Full size image

An additional serotonin receptor, SER-1/5-HT2, is expressed in pharyngeal muscles [94] but is not required for rapid pumping, whether induced by food or by 5-HT [64, 95]. Consistently, mutants carrying the putative null allele ser-1(ok345) and treated with 5-HT exhibited high pumping rates and no pumping fatigue (Fig. 5a). The absence and presence of rapid 5-HT-induced pumping in ser-7 and ser-1 mutants, respectively, indicates that 5-HT acts through activating pharyngeal neurons. The ability of deprived daf-16 and ubl-5 mutants to pump rapidly suggests that their deficits, exhibited in the presence of food, are the result of regulation rather than a biomechanical defect. In addition, neuronal rescue of DAF-16 function abolished the post-deprivation phenotype (Fig. 5b). Combined, these data indicate that rapid pumping is mechanically possible even in sleep-deprived mutants upon activation of pharyngeal neurons. Thus, sleep deprivation likely leads to lasting deficits in the neural circuit regulating pumping.

Nonlethal deprivation results in UPRER-dependent reduction in brood size

Sleep deprivation was recently implicated in affecting fertility in rodents [96, 97]. To address whether nonlethal sleep deprivation impacts C. elegans fecundity, we compared brood sizes of deprived and control animals using the 12-h protocol. We found that brood size in the control group was indistinguishable from that of undisrupted animals. However, sleep deprivation reduced wild-type brood size by 10% (Fig. 6a). The negative impact of nonlethal deprivation was exacerbated in daf-16(mu86) mutants, where brood size was reduced by 29% (Fig. 6b). In daf-16(mgDf50) null mutants, where brood size was overall lower, we observed an 18% reduction (Fig. 6c). When the function of DAF-16 was restored by driving expression with its native promoter, brood size was not reduced following deprivation (Fig. 6d).

Fig. 6
figure 6

Worm sleep deprivation reduces brood size. a Brood sizes of wild-type animals during the first 3 days after L4 lethargus (t = 0 is 10–12 h after the fourth molt). Prior to the assay, animals were exposed to the 12-h deprivation protocol. The control group was stimulated ouside of lethargus, and the undisrupted group was not stimulated (see Methods). The dotted line depicts the brood size of deprived animals at the latest time point (see also Fig. 7a). Inset: undisrupted and control animals were indistinguishable. b, c The same as a for daf-16(mu86) and daf-16(mgDf50) mutants. d The same as a for daf-16(mgDf50) mutants where expression of daf-16 cDNA was driven by the daf-16 native promoter. e, f The same as a for UPRER deficient ire-1 mutants and UPRmt deficient ubl-5 mutants. The wild-type phenotype was exhibited by ubl-5 mutants, but fecundity was not reduced in ire-1 mutants. g Quiescence fraction during L4 lethargus of animals deficient in touch sensation (mec-10p::deg-3(gf)), whose responses to vibrations were mostly or entirely abolished (see also Additional file 4: Figure S4). h, i The same as a for transgenic animals deficient for touch sensation on wild-type or daf-16 mutant backgrounds. Shaded areas depict mean ± SEM, numbers of animals assayed are noted in parentheses, and double asterisks denote a significant difference in brood size at t = 60 h (p < 0.01)

Full size image

To address the possibility of a floor effect in the daf-16(mgDf50) animals, we increased the brood size of daf-16(mgDf50) mutants through male mating [68] and observed a 24% reduction in brood size when both hermaphrodites and males were deprived. Disrupting sleep of either hermaphrodites or males resulted in intermediate phenotypes (Additional file 8: Figure S8A).

Since germ cell apoptosis can be enhanced by pharmacologically or genetically induced ER stress [98], we hypothesized that the UPRER may mediate an adverse impact of sleep deprivation on fecundity. To test this, we compared ire-1 mutants deficient in the UPRER to wild-type animals, daf-16 mutants, and UPRmt deficient ubl-5 mutants. In contrast to the other genotypes, where deprivation reduced fecundity, brood size was not reduced in the absence of IRE-1. Rather, deprived, control, and unperturbed ire-1 mutants were indistinguishable (Fig. 6e). Similarly, fecundity was impervious to sleep deprivation when ASI neurons, required for UPRER-mediated germ cell apoptosis, were genetically ablated (Additional file 8: Figure S8B) [98, 99]. The wild-type phenotype exhibited by ubl-5 mutants served as a negative control and indicated that the UPRmt did not mediate the impact of deprivation on brood size (Fig. 6f).

To control for potential biomechanical effects of the vibrations, we assayed deg-3(u662) transgenics. Failure to respond to vibrations abolished the reduction in brood size in the presence of the stimuli (Fig. 6h, i). Forced locomotion also did not appreciably increase the number of retained eggs (Additional file 9: Figure S9). Overall, these results suggest that nonlethal sleep deprivation in C. elegans negatively impacts brood size and that the UPRER and IIS play distinct roles in this process. Together with the relevance of the UPRmt to pumping fatigue, these data demonstrate distinct UPRs for mitigating impacts of sleep deprivation in different tissues.

Germ cell apoptosis causes the brood size reduction following nonlethal deprivation

The requirement of the UPRER for the reduction in brood size suggested that sleep deprivation may trigger germ cell apoptosis [98]. To test this, we assayed mutants lacking core apoptotic machinery caspase, quantified a fluorescent indicator of germ cell apoptosis, and counted fluorescently labeled germ cells. Animals carrying the putative null allele ced-3(n1286) were crossed to daf-16(mu86) mutants, which exhibited the most pronounced effect on fecundity. In both ced-3 single and daf-16; ced-3 double mutants the post-deprivation reduction in brood size was eliminated; i.e., the egg-laying dynamics of deprived, control, and unperturbed animals were indistinguishable (Fig. 7a). Moreover, brood sizes of daf-16; ced-3 double mutants were identical to those of undisrupted daf-16(mu86) single mutants. This suggested that the ced-3 mutation did not affect fecundity independently of sleep deprivation on a daf-16 background.

Fig. 7
figure 7

Worm sleep deprivation causes germ cell apoptosis. a Brood sizes of daf-16(mu86); ced-3(n1286) double mutants during the first 3 days after L4 lethargus (t = 0 is 10–12 h after the fourth molt). Deprivation failed to induce a reduction in brood size. The dotted line depicts the brood size of deprived daf-16(mu86) mutants at t = 60 h from Fig. 6a. b Box plots of ced-1::gfp fluorescence after deprivation (left) or mock deprivation (right), normalized by the mean pre-treatment fluorescence. Horizontal lines, boxes, and bars depict medians, 1st and 3rd quartiles, and 5th and 95th percentiles, respectively. Sample sizes are denoted in parentheses; double asterisks depict a significant difference (p < 0.01). c Left: the number of sperm detected in a single gonad per animal. Horizontal lines, boxes, and bars depict means, 95% confidence intervals, and standard deviations, respectively. Inset: a confocal image of adult hermaphrodite sperm nuclei, specifically labeled by GFP-histone fusion driven by the Pcomp-1 promoter. Right: the ratio between the sperm count of deprived and control animals. Error bars depict mean ± SEM. Sample sizes are denoted in parentheses; single and double asterisks depict significant differences (between the mean and 1 or between genotypes) with p < 0.05 or p < 0.01, respectively

Full size image

The ced-1::gfp translational reporter is an established indicator of germ cell apoptosis [82]. Animals expressing this reporter were subjected to the 1-h deprivation protocol. Deprivation resulted in a significant increase in ced-1::gfp fluorescence, suggesting that germ cell corpses were actively being disposed of. In contrast, the mock deprivation group, i.e., worms not exposed to vibration stimuli, did not exhibit a change in reporter fluorescence during the equivalent period (Fig. 7b).

Next, we counted fluorescently labeled sperm in self-fertilized deprived and control hermaphrodites as previously described [100, 101]. We observed reductions in sperm count following forced locomotion during L4 lethargus as compared to the control groups in wild-type and the two daf-16 strains (Fig. 7c). The reduced sperm counts were sufficient to explain the corresponding reductions in brood sizes. The reductions in daf-16(mgDf50) and daf-16(mu86) mutants were similar, perhaps due to the variability of the measurement, the 3-day assay having fallen short of revealing the full extent of the brood size deficiency, or further loss of sperm during adulthood. Direct indication of germ cell corpse engulfment mediated by CED-1, the requirement of the caspase CED-3, and the reduced sperm count provide three consistent lines of evidence. Together, they show that sleep deprivation induces germ cell apoptosis, consistent with the role of the UPRER in brood size reduction.

The UPRER (but not the UPRmt) mitigates the effects of nonlethal sleep deprivation on activity in the egg-laying circuit

The absence of hermaphrodite-specific neurons (HSNs) suppresses egg laying and doubles the frequency of calcium transients in vulval muscles (vms) [83, 102]. As germ cell apoptosis does not preclude an independent impact of deprivation on the egg-laying circuit, we asked whether sleep deprivation can cause abnormally high calcium activity in the vms. To address this, we assayed calcium dynamics indicative of vm twitching using a ratiometric reporter: co-expression of the calcium indicator GCaMP5 and the red fluorescent protein mCherry in the vms [83, 103] (Fig. 8a, b).

Fig. 8
figure 8

Worm sleep deprivation results in excess twitching of vulval muscles. a A schematic representation of the imaging setup. Top: fluorescently labeled anterior and posterior vulval muscle groups. Worms crawl on their left or right side such that their dorsoventral axis is parallel to the imaging plane. b Example traces of the ratio of GCaMP5 to mCherry fluorescence from anterior and posterior vms. c The total number of calcium transients in vms of undisrupted and sleep-deprived worms. An average increase of 30% in the number of vm twitches was detected in ire-1 (UPRER deficient) mutants. No significant changes were detected in wild-type animals and ubl-5 (UPRmt deficient) mutants. Horizontal lines, boxes, and bars depict medians, 1st and 3rd quartiles, and 5th and 95th percentiles, respectively. Sample sizes are noted in parentheses, and the asterisk denotes a significant difference (p < 0.02)

Full size image

Opposite to the pharyngeal circuit, the egg-laying circuit was affected by ER but not mitochondrial stress. Neither wild-type animals nor UPRmt deficient ubl-5 mutants exhibited abnormal post-deprivation activity. When UPRER deficient ire-1 mutants were deprived, their mean number of vm twitches increased by 30%, mirroring the trend reported for HSN deficient animals (Fig. 8c).

Physiological activity in the egg-laying circuit is coupled with body posture and locomotion during brief periods around egg-laying events [83, 103, 104]. Therefore, we could not rule out the possibility that differences in locomotion over long timescales may also correlate with activity in the egg-laying circuit. Specifically, we asked whether a potential effect of deprivation on locomotion may indirectly cause the vm twitching phenotype observed in ire-1 mutants. To address this, we assayed the effect of sleep deprivation on mean velocity and measured the correlations between mean velocity and physiological activity (twitches) in the vms.

In all genotypes assayed, the mean velocity did not vary significantly between undisrupted and deprived animals (Additional file 10: Figure S10). While correlations between the mean velocity and vm twitching were found in wild-type, UPRER deficient, and UPRmt deficient worms, they were weakest in ire-1 mutants (Additional file 11: Figure S11). Moreover, the differences between mean or median velocities of deprived and nondeprived worms were insufficient to account for the 30% increase in vm twitching. Thus, excess vm twitching was not an indirect consequence of elevated locomotion activity in our assays.

Interestingly, the correlations between activity in the egg-laying circuit and locomotion were stronger in deprived wild-type animals and ubl-5 mutants as compared to their respective undisrupted groups. This trend was reversed in ire-1 mutants (Additional file 11: Figure S11). This observation may indicate that elevated coordination between distinct behaviors during stress may require secreted proteins, such as neuropeptides, whose function depends on processing in the ER [105]. These results support the conclusion that sleep deprivation causes ER stress which, if not mitigated, impacts the egg-laying circuit. Taken together, our data demonstrate that distinct types of proteotoxic stress caused by nonlethal deprivation of worm sleep adversely affected different circuits.

Discussion

The cognitive, physiological, and behavioral changes resulting from deprivation of human sleep can be subtle and elude superficial phenotyping. Detecting them requires functional imaging and/or proper design of the task being assayed, a clear definition of the sleep deprivation conditions, and careful measurements. Interpreting such results involves consideration of details such as differences in vulnerability to deprivation between individuals and whether the task was monotonous or complex [106,107,108,109]. Similarly, wild-type rodents and insects do not typically exhibit gross defects or substantial damage to brain cells following nonlethal deprivation protocols [4, 110,111,112]. This may be due to protective responses, activated by sleep deprivation, with the capacity to effectively prevent or repair the resulting damage.

To address the lasting impact of sleep deprivation in C. elegans, we established an experimental method enabling severe reduction in quiescence with no lethality or molting defects. Our periodic stimulus allowed for some quiescent behavior to take place throughout lethargus. Possibly, not forcing locomotion for extended continuous periods was key to avoiding a lethal outcome. In contrast, previous work assessed the impact of total sleep deprivation, i.e., consecutive forced movement for 30 min, which resulted in lethal molting defects. The impact on molting was interpreted to indicate a defect in metabolic regulation during lethargus, as the loss of DAF-16 sensitized the animals to this effect [10]. Lasting defects in surviving animals were not previously assayed.

The complete lack of lethality despite a loss of 50% of quiescence during lethargus suggests that quiescence, in and of itself, is an imperfect measure of the quality and restorative benefits of developmentally timed sleep in C. elegans. If homeostatic compensation can affect the quality of sleep [13], periodically allowing for rebound could confer greater restoration per unit time as compared to uninterrupted sleep. Testing this hypothesis will be key for understanding lethargus and may promote our understanding of additional quiescent states.

Identifying protective responses to worm sleep deprivation may clarify ancient functions of sleep. Cellular damage and repair can vary in molecular detail and occur at various rates. A comparison between tissues with different functions, developmental states, and metabolic demands may resolve requirements for a rest period due to accumulating damage. We found that deprivation triggers the UPRmt, which protects feeding behavior in sleep-deprived worms. In contrast to the UPRER, the UPRmt was not strongly associated with sleep deprivation previously. This may be partly due to a focus on sleep-related changes in gene expression in the brain [25, 27, 28, 38,39,40, 113,114,115,116]. One study reported that the mitochondrial chaperones Hsp60 and glucose-regulated protein 75 (Grp75, from the Hsp70 superfamily) were upregulated in the cerebral cortex of rats after sleep deprivation, although not as much as BiP [30].

Broad transcriptional responses to sleep deprivation were reported in mouse livers, lungs, and hearts [117, 118]. These studies demonstrate that the molecular consequences of disrupting sleep are not limited to the brain and that sleep contributes to normal function in a manner that may vary between different tissues or organs. For instance, the pharynx of C. elegans is a highly active organ, required to repeatedly generate powerful contractions. Neurons that regulate pumping may face a distinct cellular metabolic challenge and therefore may particularly benefit from the UPRmt post-deprivation.

Nonlethal sleep deprivation upregulates the expression of BiP in rodents and flies, thus indicating the activation of the UPRER [25, 38,39,40]. We have shown a similar upregulation of HSP-4/BiP in response to disrupting developmentally timed worm sleep. Furthermore, loss of function of the misfolded ER protein receptor IRE-1 affected both fecundity and the egg-laying circuit post-deprivation. These findings in C. elegans show that the activation of the UPRER is a deeply conserved response to sleep deprivation [15]. It remains to be determined whether the complementing set of phenotypes exhibited by the pharyngeal and the egg-laying circuits, whose function depended on the UPRmt and the UPRER, respectively, is associated with differences in their developmental states, energy expenditure, or additional factors.

Finally, we characterized outcomes of nonlethal sleep deprivation with respect to feeding, fecundity, and egg-laying physiology. Consistent with previous findings, our deprivation protocol triggered translocation of DAF-16/FoxO into intestinal cell nuclei [10] and implicated it in mitigating lasting adverse effects of deprivation. Three lines of evidence indicated that fecundity was reduced due to germ cell apoptosis: a dependence of the effect on CED-3 and IRE-1, a low sperm count in sleep-deprived worms, and the engulfment of early germ cell corpses by surrounding sheath cells as visualized by the ced-1 reporter [74, 75, 81, 98]. Interestingly, sperm quality was recently shown to be negatively impacted by disruptions to sleep in rodents [96, 97].

Conclusions

Hypotheses explaining the core functions of sleep include the notion that its utility may differ across species [119]. The synaptic homeostasis hypothesis proposes that continuous learning during wakefulness is bound to saturate synaptic connections. Therefore, renormalization of net synaptic strength during sleep is required to restore homeostasis [120, 121]. Other suggestions focus on “wear and tear” in physiologically active neurons during wakefulness due to accumulation of protein fragments, unfolded proteins, or other molecular stressors [112, 122,123,124,125]. On a larger scale, metabolite clearance from the brain can increase during mammalian sleep [126]. In addition, disrupting sleep is linked to abnormal glucose metabolism and appetite regulation. These findings suggest that sleep is key to normal metabolic and hormonal processes outside the brain [127, 128].

This work describes multiple pathways by which sleep deprivation can upset cellular proteostasis, likely by creating unfavorable biochemical conditions. Specific characteristics of physiological activity and development could affect the balance between accumulation and relief of allostatic load. To the best of our knowledge, how this balance might scale with different types of metabolic loads has not been systematically studied, let alone connected to sleep. Protection of post-deprivation pharyngeal function by the UPRmt is consistent with the notions that sleep reduces cellular metabolic stress and that highly active organs may invoke distinct responses in this context. Implicating the UPRs in mitigating consequences of worm sleep loss indicates that these responses are deeply conserved.

Methods

Strains

Wild-type, transgenic, and mutant C. elegans strains were cultivated with OP50 bacteria according to standard protocols at 20 °C. The following strains were used: N2 Bristol (wild type), CF1038 daf-16(mu86), GR1307 daf-16(mgDf50), NQ116 muIs211 [pNL213(ges-1p::GFP::daf-16) + rol-6(su1006)] (gift from D. Raizen), NQ441 daf-16(mgDf50); qnIs45[Pdaf-16:GFP::daf-16; Pmyo-2:mCherry] (gift from D. Raizen at the University of Pennsylvania), SJ4151 zcIs19 [ubl-5p::ubl-5::gfp], SJ4100 zcIs13 [hsp-6p::gfp], SJ4058 zcIs9 [hsp-60p::GFP + lin-15(+)], VC2564 ubl-5(ok3389), SJ4200 zcIs41 [ubl-5p::3xmyc-His tag::ubl-5 + myo-3p::gfp]; SJ4151 zcIs19 [ubl-5p::ubl-5::gfp], NQ128 muEx169[unc-119p::GFP::daf-16 + rol-6(su1006)] (gift from D. Raizen at the University of Pennsylvania), DA184 ser-1(ok345), Is[Pmec-10::deg-3(u662)] (gift from M. Treinin at the Hebrew University of Jerusalem), MT3002 ced-3(n1286), MD701 bcIs39 [lim-7p::ced-1::GFP + lin-15(+)], SJ4005 zcIs4 [hsp-4p::gfp], RE666 ire-1(v33), SJ17 xbp-1(zc12), SJ30 ire-1(zc14) II; zcIs4 V, SJ17 xbp-1(zc12) III; zcIs4 V, RB925 ire-1(ok799), VC1099 hsp-4(gk514)II, UX564 jnSi118[Pcomp-1::GFP::H2B::3’comp-1; Cb-unc-119(+)]; him-5(ok1896) (gift from G. Stanfield at the University of Utah), LX1938 egl-1(n986dm)V; vsIs164 X; lite-1(ce314) X; lin-15(n765ts) X (gift from K. Collins at the University of Miami), LX1918 vsIs164 X; lite-1(ce314) X; lin-15(n765ts) X (gift from K. Collins at the University of Miami), IV205 ueEx122 [str-3::TeTx::GFP; elt-2::sl2GFP] (gift from S. Chalasani at the Salk Institute for Biological Studies).

Severe nonlethal deprivation protocol

Motion and quiescence were identified using the image difference method as described in [129]. To disrupt quiescence during lethargus, mid-L4 animals were transferred to 60-mm nematode growth media (NGM) plates containing 10 mL of medium and seeded with 50 μL of OP50 culture that was grown overnight at 37 °C. Vibrations (1 kHz) were delivered using the mechanical clamps described in [11]. In contrast to the brief and well-spaced disruptions described previously, the stimuli described here were composed of 3 min of vibrations interspersed with 3 min of “quiet” intervals, i.e., a period of 6 min and a duty cycle of 50%.

Synchronization was performed in two steps. Gravid adults were allowed to lay eggs on a fresh plate for 2 h. Of the resulting larvae, mid and early L4s were manually picked for the deprived and control groups, respectively. Manually picking the L4 larvae reduced the variability of the timing of lethargus onset to 2 h.

Assays described throughout the manuscript employ one of three deprivation protocols: (1) fluorescent markers were assayed in individual worms during the first half of lethargus following a 1-h period of 3-min on/off disruptions. Control animals were either not stimulated at all (“mock”) or stimulated for an hour prior to lethargus. (2) Gene expression was assayed in small groups of tightly synchronized animals. Deprived groups were exposed to 3-min on/off disruptions for a period of 4 h that included lethargus, and control groups were exposed to 4 h of vibrations outside of lethargus. (3) Lasting defects were assayed in large groups of animals following a 12-h period of 3-min on/off stimulation. The stimulation period was initiated prior to lethargus and terminated at the young adult stage. Control animals were exposed to vibrations 5 h before lethargus, not stimulated for 5 h that included lethargus, and stimulated again for 5 h after lethargus. In cases when animals that were never stimulated were assayed, they were labeled “unperturbed” to differentiate them from the standard “control” groups.

Optical measurements of pharyngeal pumping

Post-stimulus (see above) young adults were picked into liquid NGM and loaded into a WormSpa microfluidic device [65, 130]. An Escherichia coli OP50 overnight culture, concentration-adjusted in NGM to OD600 = 2.5 (an intermediate food concentration), was flown through the device at a constant rate throughout the assay. After 1 h of acclimation in the device, the animals were imaged for an additional 1 h at a magnification of 10× and 62.5 frames per second using a Basler acA1920-25um complementary metal-oxide-semiconductor (CMOS) camera mounted on Celestron 44104 microscopes [131]. Pumping events were identified using a custom Python script which aligned and subtracted consecutive images and calculated the entropy of the difference image. A pumping event resulted in a characteristic spike in this entropy [65, 132].

Electropharyngeograms (EPGs)

Worms were loaded to a NemaMetrix ScreenChip System microfluidic device (NemaMetrix, Inc., Eugene, OR, USA) on a standard dissection microscope and allowed to acclimate for at least 30 min before recording. The device was filled with either bacterial food (OD600 = 2.5) or a solution of 10 mM 5-HT in NGM buffer. Immediately before the onset of the measurement, the output tube was disconnected to reduce electromagnetic noise. The pumping frequency was measured as described in the ScreenChip User Guide, and each recording lasted 5–10 min (measurements that terminated prematurely were discarded).

Imaging of green fluorescent protein (GFP) reporters

The 1-h deprivation protocol consisted of 3-min-long vibration pulses interspersed with 3-min-long pauses, starting during the first 30 min of L4 lethargus and lasting a total of 1 h. Animals expressing a fluorescent reporter were exposed to these mechanical stimuli in artificial dirt microfluidic devices placed inside a standard petri dish. They were imaged in the device immediately before and after the 1-h period of disruption. Imaging was performed at a magnification of 20× (0.5 numerical aperture, NA) using a Nikon Eclipse Ti microscope (Nikon Inc., Melville, NY, USA) and an Andor iXon X3 EMCCD camera (Andor, Belfast, UK). Fluorescence intensity was determined using custom Python scripts.

Calcium imaging in freely behaving animals

To reduce background fluorescence, calcium imaging was performed in artificial dirt microfluidic devices [133]. Animals co-expressing GCaMP5 and mCherry in their vms were mounted in the presence of bacterial food on an epi-fluorescence Nikon Eclipse Ti inverted microscope. Each worm was imaged at a magnification of 20× (0.5 NA) and a frame rate of 6 frames per second. Images were captured with an Andor iXon X3 EMCCD camera. A Dual View (DV2) two-channel system was used for simultaneous imaging of the red and green channels (Photometrix, Tucson, AZ, USA). Each animal was tracked manually and continuously imaged for a total period of 30 min. Calcium transients were analyzed using custom Matlab scripts (The Mathworks Inc., Natick, MA, USA).

Brood size

Brood size was counted by visual inspection: 10 h after mechanical stimulation ended, animals were transferred to individual 60-mm NGM plates seeded with a 50-μL drop of OP50 (two animals per plate). Plates were scored in the morning and evening of the following 3 days. To avoid the accumulation of hatched offspring, animals were transferred to new plates prior to the evening counts each day. For male mating, two males and two hermaphrodites were transferred to each plate.

Sperm counting

Adult Pcomp-1::GFP::H2B::3’comp-1 animals 24 h post-L4 lethargus were sealed into individual “artificial dirt” chambers filled with NGM and 10 μM levamisole. Confocal images of fluorescently labeled sperm were obtained using a Zeiss LSM 880 microscope with a Plan-Apochromat 40×/1.4 oil differential interference contrast (DIC) objective. Image stacks were analyzed using the FARSIGHT Nucleus Editor (http://www.farsight-toolkit.org/wiki/NucleusEditor). We note that when the fluorescent marker was crossed to a daf-16 mutant background, it was not confined to the nuclei. However, individual sperm cells were still detectable. We observed this in all daf-16(mgDf60) mutants and 25% of the daf-16(mu86) mutants.

Real-time PCR

The lmn-1 gene, encoding the C. elegans nuclear laminin, was chosen as the endogenous control gene. Primers (except for lmn-1, Shaham lab, Rockefeller University, private communication) were designed using Wormbase.org and the National Center for Biotechnology Information (NCBI) Primer-Basic Local Alignment Search Tool (BLAST) software. They were tested for specificity using NCBI BLAST and by agarose gel electrophoresis (using genomic DNA) and purchased from Integrated DNA Technologies (IDT, Coralville, IA, USA). The primers used for lmn-1, hsp-4, ubl-5, and hsp-6 were TCGAGGCGGAAAAGGCTC (Fwd), GCTCCAGCGAGTTCTCTCTC (Rev), GCCGACAAGGAAAAACTCGG (Fwd), GTGGGGTTGGGTTGGGAAA (Rev), ACAAACTGGAACACGATGGGA (Fwd), TCCCTCGTGAATCTCGTAATCC (Rev), AAGAACTCTGGAGGTGACGC (Fwd), and ACGTTGGGGGTTTCTAAAGAT (Rev), respectively. Real-time quantitative PCR amplifications were performed using 25 μL of QuantiTect SYBR Green Real-Time PCR master mix (QIAGEN, Hilden, Germany), 2 μL of diluted reverse transcription product (2 ng/reaction), 1.5 μL each of forward and reverse primer, and 20 μL of DNase/RNase-free water in a total volume of 50 μL. Amplification was carried out in an AB 7900 HT Real-Time PCR cycler (Applied Biosystems, Foster City, CA, USA) with initial polymerase activation at 95 °C for 15 min, followed by cycles of 94 °C for 15 s denaturation, 57 °C for 30 s for primer-specific annealing, and 72 °C for 30 s for extension. A melting curve analysis was carried out (60–95 °C) to verify the specificity of amplicons, i.e., the absence of primer dimers and nonspecific products. Each assay included six technical replicates and a no-template control for every primer pair.

Statistical analysis

Pairwise comparisons of data represented in bar or box plot were done using the Student’s t test. In the case of multiple comparisons, significance was calculated using a one-way analysis of variance (ANOVA) test and the Bonferroni post hoc correction. Distributions represented by histograms were compared using the k-sample Anderson-Darling test and the Bonferroni post hoc correction for multiple comparisons (when applicable).

For each figure panel, the following p values are listed in order of the positions of asterisks and not significant (n.s.) labels (top to bottom and then left to right). Fig. 1b: 1.1 × 10–3, 6.5 × 10–15, 2.2 × 10–15, 9.5 × 10–7. Fig. 1c: 0.59, 0.84, 0.002, 0.001. Fig. 2b: 0.003, 0.41, 0.63. Fig. 2c: 4.9 × 10–5. Fig. 2d: 0.36, 0.98, 0.033, 0.164. Fig. 3a: 0.016, 0.11, 0.34. Fig. 3b: 0.04, 0.22. Fig. 3c: 0.003, 0.001, 0.005, 0.002. Fig. 4a: 0.21, 0.048, 0.010, 0.50, 0.26, 0.46. Fig. 4b: 0.006, 0.40, 0.142, 0.074, 0.42. Fig. 5a: 0.018, 0.095, 0.38. Fig. 5b: 0.50. Fig. 6a: 4.8 × 10–3 . Fig. 6b: 5.4 × 10–14. Fig. 6c: 6.1 × 10–3. Fig. 6d: 0.39. Fig. 6e: 0.33. Fig. 6f: 0.007. Fig. 6h: 0.19. Fig. 6i: 0.48. Fig. 7a: 0.46. 0.18. Fig. 7b: 0.009, 0.44. Fig. 7c: 0.050, 0.043, 0.009, 0.008. Fig. 8c: 0.38, 0.017, 0.79. Additional file 1: Figure S1: 0.43. Additional file 3: Figure S3A: 0.37, 0.44, 0.003, 0.50, 0.34, 0.41. Additional file 3: Figure S3B: 0.048, 0.028, 0.40, 0.070, 0.27, 0.50. Additional file 5: Figure S5: 0.16, 0.050, 0.010. Additional file 6: Figure S6B: 0.035, 0.042, 0.021. Additional file 6: Figure S6C: 0.002. Additional file 6: Figure S6D: 0.061, 0.005, 0.28. Additional file 7: Figure S7 (OD600 = 2.5): 0.15, 0.37, 0.22, 0.34, 0.092, 0.36, 0.30, 0.30. Additional file 7: Figure S7 (10 mM 5-HT): 0.039, 0.043, 0.22, 0.47, 0.27, 0.30. Additional file 8: Figure S8A: 2.5 × 10–4, 0.82. Additional file 8: Figure S8B: 0.77. Additional file 10: Figure S10: 0.10, 0.34, 0.26, 0.29.