1. Genetics and Genomics
  2. Neuroscience
Download icon

Sleep-promoting effects of threonine link amino acid metabolism in Drosophila neuron to GABAergic control of sleep drive

  1. Yoonhee Ki
  2. Chunghun Lim  Is a corresponding author
  1. Ulsan National Institute of Science and Technology, Republic of Korea
Research Communication
  • Cited 0
  • Views 1,318
  • Annotations
Cite this article as: eLife 2019;8:e40593 doi: 10.7554/eLife.40593

Abstract

Emerging evidence indicates the role of amino acid metabolism in sleep regulation. Here we demonstrate sleep-promoting effects of dietary threonine (SPET) in Drosophila. Dietary threonine markedly increased daily sleep amount and decreased the latency to sleep onset in a dose-dependent manner. High levels of synaptic GABA or pharmacological activation of metabotropic GABA receptors (GABAB-R) suppressed SPET. By contrast, synaptic blockade of GABAergic neurons or transgenic depletion of GABAB-R in the ellipsoid body R2 neurons enhanced sleep drive non-additively with SPET. Dietary threonine reduced GABA levels, weakened metabotropic GABA responses in R2 neurons, and ameliorated memory deficits in plasticity mutants. Moreover, genetic elevation of neuronal threonine levels was sufficient for facilitating sleep onset. Taken together, these data define threonine as a physiologically relevant, sleep-promoting molecule that may intimately link neuronal metabolism of amino acids to GABAergic control of sleep drive via the neuronal substrate of sleep homeostasis.

Editorial note: This article has been through an editorial process in which the authors decide how to respond to the issues raised during peer review. The Reviewing Editor's assessment is that all the issues have been addressed (see decision letter).

https://doi.org/10.7554/eLife.40593.001

Introduction

The circadian clock and sleep homeostasis are two key regulators that shape daily sleep behaviors in animals (Borbély, 1982). In stark contrast to the homeostatic nature of sleep, the internal machinery of sleep is vulnerable to external (e.g., environmental change) or internal conditions (e.g., genetic mutation) that lead to adaptive changes in sleep behaviors. Sleep behavior is conserved among mammals, insects, and even lower eukaryotes (Allada and Siegel, 2008; Joiner, 2016). Since the identification of the voltage-gated potassium channel Shaker as a sleep-regulatory gene in Drosophila (Cirelli et al., 2005), fruit flies have been one of the most advantageous genetic models to dissect molecular and neural components that are important for sleep homeostasis and plasticity.

To date, a number of sleep-regulatory genes and neurotransmitters have been identified in animal models as well as in humans (Allada et al., 2017; Artiushin and Sehgal, 2017; Tomita et al., 2017). For instance, the inhibitory neurotransmitter gamma-aminobutyric acid (GABA) is known to have a sleep-promoting role that is conserved in invertebrates and vertebrates. Hypomorphic mutations in mitochondrial GABA-transaminase (GABA-T) elevate GABA levels and lengthen baseline sleep in flies (Chen et al., 2015). The long sleep phenotype in GABA-T mutants accompanies higher sleep consolidation and shorter latency to sleep onset, consistent with the observations that pharmacological enhancement of GABAergic transmission facilitates sleep in flies and mammals, including humans (Holmes and Sugden, 1975; Lancel et al., 1998; Schneider et al., 1977). In addition, resistance to dieldrin (Rdl), a Drosophila homolog of the ionotropic GABA receptor, suppresses wake-promoting circadian pacemaker neurons in adult flies to exert sleep-promoting effects (Agosto et al., 2008; Chung et al., 2009; Liu et al., 2014a; Parisky et al., 2008). Similarly, 4,5,6,7-tetrahydroisoxazolo[5,4 c]pyridin-3-ol (THIP), an agonist of the ionotropic GABA receptor, promotes sleep in insects and mammals (Dissel et al., 2015; Faulhaber et al., 1997; Lancel, 1997).

Many sleep medications modulate GABAergic transmission. A prominent side effect of anti-epileptic drugs relevant to GABA is causing drowsiness (Jain and Glauser, 2014). Conversely, glycine supplements improve sleep quality in a way distinct from traditional hypnotic drugs, minimizing deleterious cognitive problems or addiction (Bannai and Kawai, 2012; Yamadera et al., 2007). In fact, glycine or D-serine acts as a co-agonist of N-methyl-D-aspartate receptors (NMDARs) and promotes sleep through the sub-type of ionotropic glutamate receptors (Dai et al., 2019; Kawai et al., 2015; Tomita et al., 2015). Emerging evidence further supports the roles of amino acid transporters and metabolic enzymes in sleep regulation (Aboudhiaf et al., 2018; Sonn et al., 2018; Stahl et al., 2018). In particular, we have demonstrated that starvation induces the expression of metabolic enzymes for serine biosynthesis in Drosophila brains, and elevates free serine levels to suppress sleep via cholinergic signaling (Sonn et al., 2018). These observations prompted us to hypothesize that other amino acids may also display neuro-modulatory effects on sleep behaviors.

Results

Dietary threonine promotes sleep and facilitates sleep onset

To determine if amino acid supplements modulate sleep in Drosophila, we employed an infrared beam-based Drosophila activity monitor (DAM) that detects locomotor activity in individual flies (Pfeiffenberger et al., 2010). Sleep behaviors in wild-type flies fed 5% sucrose containing 17.5 mM of each amino acid were quantitatively assessed in 12 hr light:12 hr dark (LD) cycles at 25°C. The strongest impact on sleep quantity and quality was observed with cysteine supplementation (Figure 1A and Figure 1—figure supplement 1). However, dietary cysteine compromised locomotion and caused high lethality during our sleep assay (see Figure 1—figure supplement 6). We thus excluded it from further analyses. Intriguingly, threonine supplementation potently elevated total sleep amount by increasing the number of sleep bouts (Figure 1A and Figure 1—figure supplement 1). In addition, dietary threonine evidently shortened the latency to sleep onset after lights-off. The sleep-promoting effects of dietary threonine (SPET) were dose-dependent and observed in both male and female flies (Figure 1—figure supplement 2). Transgenic silencing of sensory neurons that express either gustatory receptors (Gr66a, Gr33a, and Gr5a) or olfactory co-receptor (Lone et al., 2016) negligibly affected SPET as compared to relevant heterozygous controls (Figure 1—figure supplement 3). These results suggest that sensory perception of dietary threonine is less likely responsible for SPET. We further found that flies fed nutrient-rich food containing additional protein sources (e.g., cornmeal, yeast) also exhibited SPET, although higher concentrations of threonine were required (Figure 1—figure supplement 4). We reason that flies may ingest smaller volume of daily food on nutrient-rich diet than on sucrose-only diet as a compensation for their difference in calories per volume (Carvalho et al., 2005). Nonetheless, these data indicate that SPET is not limited to carbohydrate-only diets.

Figure 1 with 8 supplements see all
Dietary threonine promotes sleep and facilitates sleep onset.

(A) Wild-type male flies were individually loaded on to 5% sucrose food containing 17.5 mM of each amino acid (day 0) and entrained in LD cycles at 25°C. Total sleep amount (top) and latency to sleep onset after lights-off (bottom) were calculated in individual flies on day 4 and averaged for each amino acid. Essential and non-essential amino acids are grouped separately as shown at the top. The width of a violin plot indicates the density of samples. The violins are restricted by the observed ranges. Error bars indicate mean ±95% confidence interval (CI) (n = 29–213). *p<0.05, **p<0.01, ***p<0.001 to control (i.e., no amino acid supplement) as determined by one-way ANOVA, Dunnett’s multiple comparisons test. (B) Control- and threonine-fed flies were awakened by a range of mechanical stimuli 4 hr after lights-off on day 4. Aroused flies were defined if they displayed no activity for >5 min prior to the stimulus but showed any locomotor response within 10 min. The percentage of aroused flies per condition was averaged from three independent experiments (top). Error bars indicate mean ± SEM (n = 3). Sleep latency after arousal was calculated in individual flies and averaged for each condition (bottom). Error bars indicate mean ±95% CI (n = 12–27). Two-way ANOVA detected significant effects of dietary threonine on sleep latency after arousal (F[1,119]=20.43, p<0.0001) but not on % aroused flies (F[1,16]=0.227, p=0.6402). n.s., not significant; **p<0.01, ***p<0.001 as determined by Bonferroni’s multiple comparisons test.

https://doi.org/10.7554/eLife.40593.002

It has previously been shown that flies exhibit a positional preference relative to their food source, depending on sleep-wake cycles or genetic backgrounds (Donelson et al., 2012). These observations raised the possibility that threonine supplementation might have affected the positional preference in wild-type flies, thereby leading to the overestimation of their sleep amount by the DAM-based analyses (Figure 1—figure supplement 5A). To exclude this possibility, we placed individual flies into circular arenas in which food is provided unilaterally from the whole floor (Figure 1—figure supplement 5B and C). Locomotor activities of individual flies were then video-recorded in LD cycles. The video-based assessment of sleep behaviors in control- versus threonine-fed flies further confirmed SPET (Figure 1—figure supplement 5D). Lower waking activity (i.e., beam crosses per minute during wakefulness) was observed in threonine-fed flies by the DAM analysis (Figure 1—figure supplement 1). Dietary threonine actually shortened total traveling distance, but it did not significantly affect moving speed in the video analysis (Figure 1—figure supplement 5D and Figure 1—figure supplement 6). Therefore, it is unlikely that threonine supplementation causes general locomotor impairment responsible for low waking activity or long sleep phenotypes. It is also noteworthy that low waking activity does not necessarily associate with long sleep phenotypes as observed with tryptophan supplementation (Figure 1A and Figure 1—figure supplement 1).

To examine if SPET affects arousal threshold (i.e., sleep depth), we quantified arousal responses to sensory stimuli during sleep. Control- and threonine-fed flies displayed no significant differences in the percentage of flies aroused by a given range of mechanical stimuli in the middle of night (Figure 1B). However, threonine-fed flies displayed shorter latency to the first post-stimulus bout of sleep. Consistent results were obtained when nighttime sleep was interrupted by a pulse of light (Figure 1—figure supplement 7). Taken together, these data suggest that a higher sleep drive, but not a change in sleep depth, may contribute to SPET.

Circadian clock-dependent control of sleep onset is dispensable for SPET

Rdl and wide awake (wake) are two evolutionarily conserved genes that contribute to circadian clock-dependent control of sleep onset in Drosophila (Agosto et al., 2008; Liu et al., 2014a). A circadian transcription factor, CLOCK (CLK), drives daily rhythmic transcription of wake, particularly in a subset of clock neurons that express the circadian neuropeptide PIGMENT-DISPERSING FACTOR (PDF) (Liu et al., 2014a). Subsequently, WAKE acts as a clock output molecule that interacts with RDL, silences the wake-promoting PDF neurons, and facilitates sleep onset. Therefore, we asked whether circadian clocks and their regulation of sleep drive would be necessary for SPET.

We first confirmed that female mutants trans-heterozygous for hypomorphic Rdl alleles had shorter sleep latency in control-fed condition than their heterozygous controls (Agosto et al., 2008) (Figure 2—figure supplement 1). Dietary threonine, however, shortened sleep latency additively with the loss of Rdl function (p=0.084, by two-way ANOVA). In addition, trans-heterozygous Rdl mutation did not compromise SPET on daily sleep amount compared to either heterozygous controls. We next examined if SPET was suppressed in arrhythmic clock mutants. Loss of Clk function caused long sleep latency in fed condition (Liu et al., 2014a), and SPET had additive effects on the latency phenotype in Clk mutants (Figure 2A, p=0.14 by two-way ANOVA). On the other hand, short sleep latency in per mutants (Liu et al., 2014a) likely caused a floor effect, leading to no significant SPET on their sleep latency (Figure 2A). Nonetheless, wild-type and both clock mutants showed comparable SPET on daily sleep amount (Figure 2A, p=0.8367 for Clk mutants; p=0.2573 for per mutants by two-way ANOVA). Finally, it has been shown that overexpression of dominant-negative CLK proteins (CLKDN) in PDF neurons is sufficient to abolish free-running circadian locomotor rhythms (Tanoue et al., 2004) and lengthen sleep latency (Liu et al., 2014a). We observed consistent effects of CLKDN overexpression in PDF neurons on sleep drive in control-fed condition, but it did not suppress SPET (Figure 2B). These lines of our genetic evidence suggest that SPET does not require clock-dependent control of sleep onset by circadian clock genes or PDF neurons.

Figure 2 with 2 supplements see all
Circadian rhythms and clock-dependent control of sleep onset are dispensable for SPET.

(A) Arrhythmic clock mutants were loaded on to 5% sucrose food containing the indicated amount of threonine (day 0) and entrained in LD cycles at 25°C. Sleep behaviors in individual female flies were analyzed similarly to the data presented in Figure 1A. Two-way ANOVA detected no significant interaction of SPET with per01 (F[1,159]=1.293, p=0.2573 for sleep amount) or ClkJrk (F[1,160]=0.0426, p=0.8367 for sleep amount; F[1,160]=2.199, p=0.14 for sleep latency). Error bars indicate mean ±95% CI (n = 35–46). (B) PDF neuron-specific overexpression of dominant-negative CLK proteins (CLKDN) lengthened sleep latency in female flies fed control food (5% sucrose) but it did not suppress SPET. Error bars indicate mean ±95% CI (n = 26–42). (C and D) Wild-type flies were loaded on to 5% sucrose food containing the indicated amount of threonine (day 0) and then entrained in LD or constant light (LL) cycles at 25°C. For sleep analyses in constant dark (DD), LD-entrained flies were transferred to DD at the end of day 4 and their sleep was monitored during the first DD cycle (day 5). Averaged sleep profiles (% sleep per 30 min bin) on day 4 (LD or LL) or day 5 (DD) were shown at the top. Data represent mean ± SEM (n = 25–46). Error bars in the violin plots indicate mean ±95% CI (n = 25–46). n.s., not significant; *p<0.05, **p<0.01, ***p<0.001 as determined by two-way ANOVA, Tukey’s multiple comparisons test.

https://doi.org/10.7554/eLife.40593.011

To further test the implication of circadian clocks in SPET, we compared SPET in different light-dark conditions. Constant dark (DD) following LD entrainment eliminates masking behaviors in direct response to the light transitions while allowing free-running circadian rhythms by endogenous clocks (Allada and Chung, 2010). We found that DD did not suppress SPET but rather exaggerated it particularly in male flies (Figure 2C and D, p<0.0001 to SPET on sleep amount or sleep latency in LD by two-way ANOVA). SPET was thus evident even in the absence of light. By contrast, constant light (LL) abolishes circadian rhythms in wild-type flies (Emery et al., 2000). Consequently, control-fed flies completely lost their daily rhythms in sleep-wake cycles (Figure 2C and D) and dampened their sleep latency in LL (Figure 2—figure supplement 2). This caused a floor effect whereby SPET was barely detectable, particularly on sleep latency at the transition of subjective day and night in LL as compared to LD. Nonetheless, we observed significant effects of dietary threonine on sleep latency (i.e., shorter sleep latency in threonine-fed flies) when SPET on sleep latency was compared among different time-points in LL (p=0.0003 to control-fed male in LL; p<0.0001 to control-fed female in LL by two-way ANOVA). Dietary threonine also increased daily sleep amount in LL (Figure 2C and D). In fact, male flies displayed comparable SPET on daily sleep amount in LD and LL (p=0.1835 by two-way ANOVA). Collectively, these data support that higher sleep drive by SPET likely operates in a manner independent of circadian clocks and their control of sleep onset.

Genetic or pharmacological elevation of synaptic GABA suppresses SPET

To elucidate genetic and neural mechanisms underlying SPET, we examined effects of dietary threonine on sleep behaviors in loss-of-function mutants of other sleep-regulatory genes. Interestingly, SPET was potently suppressed in GABA-T mutants trans-heterozygous for a null allele (GABA-TPL) over chromosomal deficiency (Figure 3—figure supplement 1). Their sensitivity to SPET was partially but significantly rescued by transgenic overexpression of wild-type GABA-T (Chen et al., 2015). However, the trans-heterozygosity of these strong GABA-T alleles promoted sleep in control-fed condition (Chen et al., 2015), raising the possibility that a ceiling effect may mask SPET. We thus tested SPET in weaker allelic combinations of GABA-T mutations. GABA-T mutants trans-heterozygous for null over hypomorphic alleles (GABA-TF or GABA-TLL) (Chen et al., 2015) did not significantly affect baseline sleep in control-fed condition as compared to their heterozygous controls (Figure 3A). Nonetheless, these mutants still exhibited the resistance to SPET (p<0.0001 to SPET on sleep amount or sleep latency in heterozygous controls by two-way ANOVA).

Figure 3 with 1 supplement see all
Genetic or pharmacological elevation of GABA suppresses SPET.

(A) GABA-T trans-heterozygous mutants were resistant to SPET. Sleep behaviors in individual male flies were analyzed similarly to the data presented in Figure 1A. Two-way ANOVA detected significant suppression of SPET in GABA-T trans-heterozygous mutants on sleep amount (F[2,403]=39.21, p<0.0001 for GABA-TPL/GABA-TF; F[2,430]=32.28, p<0.0001 for GABA-TPL/GABA-TLL; F[2,454]=13.99, p<0.0001 for GABA-TF/GABA-TLL) and sleep latency (F[2,403]=13.35, p<0.0001 for GABA-TPL/GABA-TF; F[2,430]=15.97, p<0.0001 for GABA-TPL/GABA-TLL; F[2,454]=9.324, p=0.0001 for GABA-TF/GABA-TLL), compared to their heterozygous controls. Error bars indicate mean ±95% CI (n = 32–114).(B) Co-administration of GABA-T inhibitor (EOS) or GABA transporter inhibitor (NipA) with threonine blocked SPET in wild-type flies. Where indicated, EOS or NipA was added to the sucrose food containing the increasing amounts of threonine. Sleep behaviors were analyzed as described above. Two-way ANOVA detected significant interaction of SPET with EOS (F[2,155]=14.07, p<0.0001 for sleep amount; F[2,155]=11.2, p<0.0001 for sleep latency) or NipA (F[2,162]=13.09, p<0.0001 for sleep amount; F[2,162]=26.58, p<0.0001 for sleep latency). Error bars indicate mean ±95% CI (n = 22–37) .n.s., not significant; *p<0.05, **p<0.01, ***p<0.001 as determined by Tukey’s multiple comparisons test.

https://doi.org/10.7554/eLife.40593.014

To independently confirm the implication of GABA-T function in SPET, we pharmacologically silenced the enzymatic activity of GABA-T in wild-type flies by oral administration of ethanolamine O-sulfate (EOS), a GABA-T inhibitor. EOS supplement did not significantly increase daily sleep amount at a given dose in our sleep assay, but it modestly shortened sleep latency in wild-type flies fed control food (Figure 3B). However, SPET was suppressed in EOS-fed flies (p<0.0001 to SPET on sleep amount or sleep latency in control flies by two-way ANOVA) similarly as in GABA-T mutants. Considering that GABA-T is a mitochondrial enzyme which metabolizes GABA into succinic semialdehyde (Chen et al., 2015), we hypothesized that high GABA levels at GABAergic synapses might interfere with sleep drive by dietary threonine. This idea was further supported by our observation that nipecotic acid (NipA), which blocks GABA reuptake from synaptic clefts (Leal and Neckameyer, 2002), comparably suppressed SPET (Figure 3B, p<0.0001 to SPET on sleep amount or sleep latency in control flies by two-way ANOVA). Collectively, these genetic and pharmacological data suggest that SPET may involve a sleep drive relevant to GABA. In addition, evidence from our adult-specific manipulations of GABA levels excludes possible developmental effects of GABA-T mutation or GABA on SPET.

Dietary threonine decreases GABA and glutamate levels

Genetic deficit in the metabolic conversion of GABA to glutamate leads to high levels of GABA in GABA-T mutants while they have low levels of glutamate and alpha-ketoglutarate, a glutamate derivative that enters tricarboxylic cycle (Maguire et al., 2015). These changes in GABA-derived metabolites are accompanied with impairment in energy homeostasis as supported by the high ratio of NAD+/NADH levels and low ATP levels in GABA-T mutants. Consequently, GABA-T mutants cannot survive on carbohydrate-based food (i.e., 5% sucrose +1.5% agar) but their metabolic stress phenotypes are rescued by the supplement of glutamate and other amino acids that can be metabolized to glutamate. We thus asked if dietary threonine would induce relevant metabolic changes that may be responsible for SPET.

Dietary threonine did not significantly affect ATP levels or the ratio of NAD+/NADH levels (Figure 4—figure supplement 1A). However, pyruvate levels were selectively elevated in threonine-fed flies (p<0.0001 to succinate by one-way ANOVA, Dunnett’s multiple comparisons test), possibly due to the metabolism of dietary threonine into pyruvate via L-2-amino-acetoacetate (Figure 4—figure supplement 2). Nonetheless, dietary pyruvate itself did not promote sleep (Figure 4—figure supplement 1B). Quantification of free amino acids further revealed that dietary threonine reduced the relative levels of proline, histidine, alanine, glutamate, and GABA among other amino acids (Figure 4A, p<0.05 or p<0.001 to arginine by one-way ANOVA, Dunnett’s multiple comparisons test). Since it has been shown that glutamate acts as either a wake- or sleep-promoting neurotransmitter in Drosophila (Guo et al., 2016; Robinson et al., 2016; Tomita et al., 2015; Zimmerman et al., 2017), we asked if co-administration of threonine and glutamate could suppress SPET. Glutamate supplement, however, negligibly affected SPET (Figure 4—figure supplement 3, p=0.91 for sleep amount; p=0.516 for sleep latency by two-way ANOVA), suggesting that dietary threonine may not limit glutamate levels to promote sleep. It is noteworthy that glutamate supplement can rescue metabolic stress, but not sleep phenotypes, in GABA-T mutants, indicating independent regulatory pathways of GABA-relevant metabolism and sleep (Maguire et al., 2015).

Figure 4 with 6 supplements see all
Down-regulation of metabotropic GABA transmission likely mediates SPET.

(A) Dietary threonine decreased the relative levels of select amino acids including GABA and glutamate. Wild-type male flies were loaded on to standard cornmeal-yeast-agar food containing either 0 mM (control) or 50 mM threonine, and then entrained in LD cycles at 25°C for 4 days before harvest. Relative levels of free amino acids in head extracts from threonine-fed flies were measured using ion exchange chromatography and then normalized to those in head extracts from control-fed flies. Error bars indicate mean ± SEM (n = 3). n.s., not significant; *p<0.05, ***p<0.001 to the relative levels of arginine as determined by one-way ANOVA, Dunnett’s multiple comparisons test. (B) Conditional blockade of GABAergic transmission promoted sleep in control-fed condition and masked SPET. Transgenic flies expressing a temperature-sensitive allele of shibire (shibirets) were loaded on to 5% sucrose food containing the indicated amount of threonine (day 0) and entrained in LD cycles at 29°C (restrictive) or 21°C (permissive). Sleep behaviors in individual male flies were analyzed on day 3 (29°C) or day 6 (21°C) since low temperature delayed SPET even in heterozygous controls. Two-way ANOVA detected significant masking of SPET by shibirets overexpression in GAD1-expressing cells at 29°C (F[2,197]=14.06, p<0.0001 for sleep amount; F[2,196]=6.953, p=0.0012 for sleep latency), but not at 21°C (F[2,184]=2.055, p=0.131 for sleep amount; F[2,184]=0.1835, p=0.8325 for sleep latency) as compared to their heterozygous controls. Error bars indicate mean ±95% CI (n = 14–63 for 29°C; n = 25–35 for 21°C). (C) Pan-neuronal deletion of metabotropic GABA receptors (GABAB-R1 and GABAB-R3) by transgenic RNA interference (RNAi) increased daily sleep amount in control-fed condition and masked SPET. Locomotor activities in individual male flies were monitored similarly to the data presented in Figure 1A. Sleep behaviors were analyzed on day 3 to better compare the sensitivity to SPET among different genotypes. Two-way ANOVA detected significant masking of SPET by the pan-neuronal RNAi on sleep amount (F[2,236]=8.913, p=0.0002 for GABAB-R1RNAi#2; F[2,317]=16.78; F[2,193]=4.594, p=0.0112 for GABAB-R3RNAi#1) and sleep latency (F[2,193]=3.267, p=0.0403 for GABAB-R3RNAi#1) as compared to their heterozygous controls. Error bars indicate mean ±95% CI (n = 24–50). (D) Oral administration of SKF-97541 (an agonist of metabotropic GABA receptor), but not of THIP (an agonist of ionotropic GABA receptor), suppressed SPET. Sleep behaviors in individual male flies were analyzed as described above. Where indicated, THIP (5 μg/ml) or SKF-97541 (1 μg/ml) was added to the behavior food. Two-way ANOVA detected significant effects of SKF-97541 on SPET (F[1,143]=17.39, p<0.0001 for sleep amount; F[1,143]=6.898, p=0.0096 for sleep latency).Error bars indicate mean ±95% CI (n = 27–53). n.s., not significant; *p<0.05, **p<0.01, ***p<0.001 as determined by Tukey’s multiple comparisons.

https://doi.org/10.7554/eLife.40593.016

Down-regulation of GABA transmission via metabotropic GABA receptors induces sleep and masks SPET

To determine if dietary threonine affects GABA transmission, we examined intracellular Ca2+ levels in glutamate decarboxylase 1 (GAD1)-expressing GABAergic neurons as a quantitative proxy for their neural activity. Since threonine supplementation exhibited cumulative effects on baseline sleep in LD cycles, we reasoned that it might be necessary to monitor the long-term changes in neural activity associated with threonine diet. Accordingly, we employed a transgenic reporter of the calcium-sensitive transcriptional activator LexA (CaLexA) that translocates into nucleus in a calcium-dependent manner and induces the expression of green fluorescent protein (GFP) (Masuyama et al., 2012). Confocal microscopy of adult fly brains revealed the strongest GFP expression by the GABAergic CaLexA in neurons projecting into antennal lobe (AL), medial antenno-cerebral tract (mACT), and lateral horn (LH) among other GAD1-expressing neurons (Figure 4—figure supplement 4A). These observations suggest a heterogeneity in baseline Ca2+ levels among GABAergic neuron subsets. Interestingly, threonine, but not arginine, induced the CaLexA signal in a subset of GABAergic neurons adjacent to the antennal lobe (LN, lateral neurons) (Figure 4—figure supplement 4B and C). By contrast, no detectable changes were observed in the CaLexA signals from other sleep-regulatory loci such as mushroom body or dopaminergic neurons upon threonine diet (Figure 4—figure supplement 4D and E). Although the sensitivity of CaLexA may limit the detectable size and duration of Ca2+ changes in our experimental condition, these results support the relative specificity of Ca2+ response in LN to the threonine diet. Given that dietary threonine decreased GABA levels, GABAergic LN may selectively display a compensatory increase in their neural activity. Alternatively, it is possible that auto-inhibitory GABA receptors (Pinard et al., 2010) are expressed more strongly in these LN than other GABAergic neurons. Low GABA levels in threonine-fed flies may then relieve this negative feedback and stimulate their neural activity upon threonine diet. In either case, these results prompted us to ask if GABAergic transmission would be necessary for SPET.

To further validate the implication of GABAergic transmission in SPET, we expressed a shibirets transgene (Kitamoto, 2001) in GAD1-expressing GABAergic neurons. The shibirets is a temperature-sensitive mutant allele in a Drosophila homolog of dynamin that interferes with synaptic vesicle recycling and thus, blocks synaptic transmission at restrictive (29°C) but not permissive (21°C) temperature. The conditional blockade of synaptic transmission in GABAergic neurons induced sleep in control-fed condition (Figure 4B), and it significantly masked SPET (p<0.0001 for sleep amount; p=0.0012 for sleep latency by two-way ANOVA). These long sleep phenotypes were partially but consistently observed by the pan-neuronal depletion of metabotropic GABA receptor R1 or R3 (GABAB-R1 or GABAB-R3) (Figure 4C). However, their effects were in contrast with those observed by hypomorphic GABA-T mutations that suppressed SPET but did not promote baseline sleep in control-fed condition. We further found that co-administration of an agonist of metabotropic GABA receptors (SKF-97541), but not of ionotropic GABA receptors (THIP), with threonine suppressed SPET particularly on sleep latency (Figure 4D, p=0.1285 for THIP; p=0.0096 for SKF-97541 by two-way ANOVA). Adult-specific manipulations of GABAergic transmission by the temperature-sensitive allele or by the oral administration of receptor-specific agonists excluded possible developmental effects of GABA on SPET. Collectively, these data suggest a possible model that SPET involves the down-regulation of metabotropic GABA transmission to induce sleep whereas genetic or pharmacological elevation of the GABA transmission interferes with this process to suppress SPET. Nonetheless, the multimeric nature of GABA receptors and their expression in either wake- or sleep-promoting neurons likely complicate the net effects of general activation or silencing of GABA transmission on sleep. We thus asked if more specific suppression of the metabotropic GABA transmission in a dedicated neural locus would induce sleep and mask SPET, thereby supporting our hypothesis above.

Metabotropic GABA transmission in ellipsoid body R2 neurons contributes to SPET

A previous study mapped a subset of ellipsoid body (EB) neurons in the adult fly brain (hereafter referred to as R2 EB neurons) as a neural locus important for sleep homeostasis (Liu et al., 2016). Neural activity in R2 EB neurons positively correlates to sleep need, and the transgenic excitation of R2 EB neurons is sufficient to induce rebound sleep. Considering that SPET involves a higher sleep drive, we hypothesized that dietary threonine might affect the activity of R2 EB neurons via metabotropic GABA transmission. Since the intracellular signaling downstream of metabotropic GABA receptors triggers cAMP synthesis (Onali et al., 2003), we employed Epac1-camps, a transgenic fluorescence resonance energy transfer (FRET) sensor for cyclic adenosine monophosphate (cAMP) (Shafer et al., 2008) (Figure 5A). Our live-brain imaging of the Epac1-camps in R2 EB neurons detected a dose-dependent increase in cAMP levels by a bath application of GABA (Figure 5—figure supplement 1A). Pre-incubation with tetrodotoxin did not affect the GABA-induced elevation of cAMP levels, indicating cell-autonomous effects of GABA on these R2 EB neurons (Figure 5—figure supplement 1B). We further found that dietary threonine modestly, but significantly, dampened the GABA response in R2 EB neurons (Figure 5B), validating that dietary threonine modulates the neural activity of this homeostatic sleep driver.

Figure 5 with 2 supplements see all
Metabotropic GABA transmission in ellipsoid body R2 neurons contributes to SPET.

(A) A representative live-brain image of Epac1-camps (a transgenic FRET sensor for cAMP) expressed in R2 EB neurons by 58H05-Gal4 driver (left). An inverse correlation between intracellular cAMP levels and FRET intensity was depicted on the right. CFP, cyan fluorescent protein; YFP, yellow fluorescent protein. (B) Transgenic flies (58H05 > Epac1 camps) were fed on control or threonine-containing food for 4 days in LD cycles at 25°C. Whole brains were dissected out and transferred to an imaging chamber. A time series of the fluorescence images was recorded using a multi-photon microscopy. Where indicated, 100 mM GABA was batch-applied to the imaging medium. FRET analysis was performed using ZEN software. Averaged histograms of the relative changes in FRET intensity (top) and their averaged median values (bottom) were shown. Data represent mean ± SEM (n = 10–14). *p<0.05 as determined by Student’s t-test. (C) The RNAi-mediated deletion of metabotropic GABA receptors (GABAB-R2 and GABAB-R3) in R2 EB neurons induced sleep in control-fed condition and masked SPET. Sleep behaviors in individual male flies were monitored similarly to the data presented in Figure 4C. Two-way ANOVA detected significant masking of SPET by the overexpression of RNAi transgenes in R2 EB neurons on sleep amount (F[2,161]=4.818, p=0.0093 for GABAB-R2RNAi#2; F[2,133]=7.669, p=0.0007 for GABAB-R3RNAi#1) and sleep latency (F[2,161]=5.088, p=0.0072 for GABAB-R2RNAi#2; F[2,133]=14.65, p<0.0001 for GABAB-R3RNAi#1) as compared to their heterozygous controls. Error bars indicate mean ±95% CI (n = 17–34). n.s., not significant; *p<0.05, **p<0.01, ***p<0.001 as determined by Tukey’s multiple comparisons.

https://doi.org/10.7554/eLife.40593.023

We next asked if metabotropic GABA transmission in R2 EB neurons would contribute to SPET. The RNAi-mediated depletion of GABAB-R2 or GABAB-R3 in R2 EB neurons modestly promoted sleep in control-fed conditions (Figure 5C). Moreover, it significantly masked SPET on sleep amount (p=0.0093 for GABAB-R2; p=0.0007 for GABAB-R3 by two-way ANOVA) and on sleep latency (p=0.0072 for GABAB-R2; p<0.0001 for GABAB-R3 by two-way ANOVA), as compared to heterozygous controls. The GABAB-R3 RNAi phenotypes were consistent with those observed by the pan-neuronal depletion of GABAB-R3 (Figure 4C). On the other hand, no detectable phenotypes were observed by the pan-neuronal overexpression of the GABAB-R2 RNAi transgenes, likely due to insufficient depletion of GABAB-R2 in R2 EB neurons by the pan-neuronal driver (Figure 4—figure supplement 5). Nonetheless, these results indicate that genetic suppression of the metabotropic GABA transmission in R2 EB neurons phenotypically mimics SPET at the levels of neural activity (i.e., weaker GABA responses) and sleep behaviors (i.e., higher sleep drive). The sleep phenotypes by the pan-neuronal, but not R2 EB-specific, depletion of GABAB-R1 (Figure 4C and Figure 5—figure supplement 2) further suggest that this sub-type of metabotropic GABA receptors may be expressed in non-R2 EB neurons to mediate sleep-regulatory transmission relevant to SPET.

SPET rescues short-term memory in fly mutants with memory deficit

Inhibitory effects of dietary threonine on metabotropic GABA transmission in R2 EB neurons support that SPET enhances sleep drive via a physiologically relevant neural locus. Nonetheless, the operational definition of a sleep episode in our behavioral assays (i.e., no movement for longer than 5 min) could mislead threonine-induced behavioral quiescence into SPET. Therefore, we took two independent approaches to validate that SPET is physiologically relevant to sleep. Sleep deprivation impairs learning in Drosophila (Seugnet et al., 2008). By contrast, genetic or pharmacological induction of sleep ameliorates memory deficits in plasticity mutants (Dissel et al., 2015). These observations have convincingly demonstrated the physiological benefits of sleep in memory formation, and we thus hypothesized that dietary threonine should rescue memory mutants if it would induce physiologically relevant sleep. To test this hypothesis, we employed a short-term memory (STM) test that was based on aversive phototaxic suppression (Seugnet et al., 2009) (Figure 6A), and examined possible effects of dietary threonine on STM.

Figure 6 with 1 supplement see all
Dietary threonine rescues short-term memory in dumb mutants with memory deficit in a sleep-dependent manner.

(A) An experimental design of the short-term memory (STM) test after three cycles of training on aversive phototaxis suppression. Wild-type (Canton S) or dumb2 mutant flies were individually loaded on to 5% sucrose food containing either 0 mM (control) or 25 mM threonine (day 0), and then entrained for 3 days in LD cycles at 25°C. Where indicated, 0.5 mg/ml of caffeine was added to the behavior food. Locomotor activity in individual male flies was monitored using the DAM system to analyze sleep behaviors on day 3 prior to the STM test on day 4. (B) The performance index during the test session was calculated in individual flies and averaged for each condition. Two-way ANOVA detected no significant effects of threonine or caffeine on STM in control flies (F[1,41]=0.9644, p=0.3318 for threonine; [1,41]=0.1433, p=0.7070 for caffeine). By contrast, two-way ANOVA detected significant interaction between threonine and caffeine on STM in dumb mutants (F[1,43]=4.329, p=0.0435). Data represent average ± SEM (n = 10–13). (C) Sleep behaviors in individual male flies were analyzed similarly to the data presented in Figure 1A. Two-way ANOVA detected significant effects of threonine or caffeine on daily sleep amount in control flies (F[1,62]=18.41, p<0.0001 for threonine; F[1,62]=22.26, p<0.0001 for caffeine), but not their significant interaction (F[1,62]=0.2836, p=0.5963). Additive effects of threonine and caffeine on daily sleep amount were also observed in dumb mutants (F[1,56]=1.091, p=0.3007 by two-way ANOVA). Error bars indicate mean ±95% CI (n = 11–19). n.s., not significant; *p<0.05 as determined by Tukey’s multiple comparisons test.

https://doi.org/10.7554/eLife.40593.026

Hypomorphic mutants of D1-like dopamine receptor 1 (dumb2) displayed impairment in STM (Figure 6B), consistent with previous observation (Seugnet et al., 2008). Dietary threonine substantially improved STM in dumb mutants (Figure 6B), and comparably rescued memory deficit in plasticity mutants of rutabaga, a Drosophila homolog of adenylate cyclase (Dissel et al., 2015) (Figure 6—figure supplement 1). To confirm that memory rescue actually requires threonine-induced sleep, we pharmacologically deprived sleep in dumb mutants by oral administration of caffeine (Andretic et al., 2008; Nall et al., 2016; Wu et al., 2009), and tested its effects on the threonine-dependent rescue of STM in dumb mutants. SPET and caffeine-induced arousal displayed additive effects on daily sleep amount in control flies (Figure 6C, p=0.5963 by two-way ANOVA) while negligibly affecting their performance index in the memory test (Figure 6B). Consistent with the implication of dopaminergic activation in caffeine-induced arousal (Andretic et al., 2008; Nall et al., 2016), baseline sleep in dumb mutants were relatively insensitive to caffeine. Co-administration of caffeine and threonine, however, suppressed dumb mutant sleep more evidently than caffeine alone (Figure 6B), and blocked the improvement of their memory deficit by dietary threonine (Figure 6C, p=0.0435 by two-way ANOVA).

Genetic elevation of endogenous threonine levels facilitates sleep onset

We next asked if a physiologically relevant increase in threonine levels could act as an endogenous promoter of sleep. We hypothesized that genetic mutations in threonine-metabolizing enzymes might elevate the steady-state levels of endogenous threonine. CG5955 is a fly homolog of threonine 3-dehydrogenase that converts threonine and NAD+ into L-2-amino-acetoacetate, NADH, and H+ (Figure 7A). We identified a transposable P-element insertion in the proximal promoter region of the CG5955 locus that reduced the relative levels of CG5955 mRNA (Figure 7B and C). Biochemical analyses of fly extracts confirmed that CG5955 mutants trans-heterozygous for the hypomorphic allele over chromosomal deficiency displayed a higher ratio of threonine to total protein levels than heterozygous controls (Figure 7C). Behavioral analyses revealed that either homozygous or trans-heterozygous mutation in CG5955 increased daily sleep amounts compared to heterozygous controls (Figure 7D). Moreover, the latency to sleep onset after lights-off was strongly shortened in CG5955 mutants, indicating a high sleep drive. We further found that pan-neuronal depletion of CG5955 expression was sufficient to mimic CG5955 mutants in terms of their sleep latency phenotype (Figure 7E). Since genetic manipulations of a metabolic enzyme can lead to compensating changes in relevant metabolic pathways or development, we do not exclude the possibility that these indirect effects may have contributed to the higher sleep drive observed in CG5955 mutants. Nonetheless, our genetic, biochemical, and behavioral evidence supports that threonine metabolism in the brain modulates sleep drive in flies.

Genetic suppression of threonine 3-dehydrogenase elevates endogenous threonine levels and facilitates sleep onset.

(A) A threonine metabolism catalyzed by threonine 3-dehydrogenase (CG5955). (B) A hypomorphic mutant allele of the P element insertion ([GS20382]) in the CG5955 locus. An amplicon used in quantitative PCR was depicted by a gray line. (C) Trans-heterozygous mutations in CG5955 reduced CG5955 mRNA levels (left, normalized to polyA-binding protein mRNA levels) but elevated endogenous threonine levels (right, normalized to protein levels). Data represent mean ± SEM (n = 3). *p<0.05, **p<0.01, ***p<0.001 as determined by one-way ANOVA, Tukey’s multiple comparisons test. (D) Loss-of-function mutations in CG5955 promoted sleep. CG5955 mutants were loaded on to 5% sucrose (day 0) and entrained in LD cycles at 25°C. Sleep behaviors in individual female flies were analyzed on day 3 and averaged for each genotype. Error bars indicate mean ±95% CI (n = 26–76). *p<0.05, ***p<0.001 as determined by one-way ANOVA, Tukey’s multiple comparisons test. (E) Pan-neuronal depletion of CG5955 expression shortened sleep latency. Sleep behaviors in individual male flies were analyzed as described above due to the X-chromosomal insertion of the pan-neuronal ELAV-Gal4 driver. DICER-2 was co-expressed with each of two independent RNAi transgenes (CG5955RNAi #1 and CG5955RNAi #2) to enhance the RNAi effects. Error bars indicate mean ±95% CI (n = 20–64). n.s., not significant; *p<0.05, **p<0.01, ***p<0.001 as determined by one-way ANOVA, Dunnett’s multiple comparisons test.

https://doi.org/10.7554/eLife.40593.028

Discussion

The molecular and neural machinery of sleep regulation intimately interacts with external (e.g., light, temperature) and internal sleep cues (e.g., sleep pressure, metabolic state) to adjust the sleep architecture in animals. Using a Drosophila genetic model, we have investigated whether dietary amino acids could affect sleep behaviors and thereby discovered SPET. Previous studies have demonstrated that the wake-promoting circadian pacemaker neurons are crucial for timing sleep onset after lights-off in LD cycles (Agosto et al., 2008; Chung et al., 2009; Liu et al., 2014a; Parisky et al., 2008). In addition, WAKE-dependent silencing of clock neurons and its collaborative function with RDL have been suggested as a key mechanism in the circadian control of sleep onset (Liu et al., 2014a). However, our evidence indicates that SPET facilitates sleep onset in a manner independent of circadian clocks. We further elucidate that SPET operates likely via the down-regulation of metabotropic GABA transmission in R2 EB neurons, a neural locus for generating homeostatic sleep drive (Liu et al., 2016).

Both food availability and nutritional quality substantially affect sleep behaviors in Drosophila. Sucrose contents in food and their gustatory perception dominate over dietary protein to affect daily sleep (Catterson et al., 2010; Linford et al., 2012; Linford et al., 2015). Starvation promotes arousal in a manner dependent on the circadian clock genes Clock and cycle (Keene et al., 2010) as well as neuropeptide F (NPF), which is a fly ortholog of mammalian neuropeptide Y (Chung et al., 2017). On the other hand, protein is one of the nutrients that contribute to the postprandial sleep drive in Drosophila (Murphy et al., 2016) and this observation is possibly relevant to SPET. While Leucokinin (Lk) and Lk receptor (Lkr) play important roles in dietary protein-induced postprandial sleep (Murphy et al., 2016) and in starvation-induced arousal (Murakami et al., 2016), we observed comparable SPET between hypomorphic mutants of Lk or Lkr and their heterozygous controls (Figure 1—figure supplement 8). Therefore, SPET and its neural basis reveal a sleep-regulatory mechanism distinct from those involved in sleep plasticity relevant to food intake.

What will be the molecular basis of SPET? Given the general implication of GABA in sleep promotion, a simple model will be that a molecular sensor expressed in a subset of GABAergic neurons (i.e., LN) directly responds to an increase in threonine levels, activates GABA transmission, and thereby induces sleep. Several lines of our evidence, however, favored the other model that dietary threonine actually down-regulates metabotropic GABA transmission in R2 EB neurons, de-represses the neural locus for generating homeostatic sleep drive, and thereby enhances sleep drive. The latter model does not necessarily conflict with sleep-promoting effects of genetic or pharmacological conditions that generally elevate GABA levels or enhance GABAergic transmission since those effects will be the net outcome of activated GABA transmission via various sub-types of GABA receptors expressed in either wake- or sleep-promoting neurons and their circuitry.

The structural homology among threonine, GABA, and their metabolic derivatives (e.g., alpha-ketobutyrate and gamma-hydroxybutyrate) led us to the hypothesis that these relevant chemicals may act as competitive substrates in enzymatic reactions for their overlapping metabolism (Figure 4—figure supplement 6). Consequently, dietary threonine may limit the total flux of GABA-glutamate-glutamine cycle possibly through substrate competition, decreases the size of available GABA pool, and thereby down-scales GABA transmission for SPET. This accounts for why genetic or pharmacological elevation of GABA levels rather suppresses SPET. Threonine, GABA, and their derivatives may also act as competitive ligands for metabotropic GABA receptors, explaining weak GABA responses in R2 EB neurons of threonine-fed flies. Biochemical and neural evidence supportive of this hypothesis is quite abundant. It has been previously shown that alpha-ketobutyrate, GABA, and the ketone body beta-hydroxybutyrate act as competitive substrates in common enzymatic reactions (Beyerinck and Brass, 1987; Lund et al., 2011; Suzuki et al., 2009). Moreover, functional interactions of beta-hydroxybutyrate or gamma-hydroxybutyrate with GABAergic signaling have been well documented (Absalom et al., 2012; Carter et al., 2009; Carter et al., 2005; Lund et al., 2011; Nasrallah et al., 2010; Snead and Gibson, 2005; Suzuki et al., 2009). Finally, threonine and GABA derivatives have anti-convulsive effects (Growdon et al., 1991; Hauser et al., 1992; Lee et al., 1990), which further support their common structural and functional relevance to GABAergic signaling.

The removal of the amino group is the initial step for amino acid metabolism, and various transaminases mediate its transfer between amino acids and alpha-keto acids. On the other hand, a group of amino acids (i.e., glutamate, glycine, serine, and threonine) has their own deaminases that can selectively remove the amino group (Bender, 2014). The presence of these specific deaminases is indicative of active mechanisms that individually fine-tune the baseline levels of these amino acids in metabolism, and possibly in the context of other physiological processes as well. This idea is further supported by the conserved roles of glutamate, glycine, and serine as neurotransmitters or neuromodulators important for brain function, including sleep regulation (Kawai et al., 2015; Tomita et al., 2017; Zimmerman et al., 2017). In fact, serine, glycine, and threonine constitute a common metabolic pathway (Figure 4—figure supplement 2), and threonine may contribute indirectly to glycine- or serine-dependent activation of sleep-promoting NMDAR (Kawai et al., 2015; Tomita et al., 2015). Nonetheless, we found that sleep-modulatory effects of dietary glycine were distinct from SPET and thus, we speculate that threonine may act as an independent neuromodulator, similar to other amino acids with their dedicated deaminases.

While several lines of our data support that threonine is likely to be an endogenous sleep driver in fed conditions, we have recently demonstrated that starvation induces serine biosynthesis in the brain and neuronal serine subsequently suppresses sleep via cholinergic signaling (Sonn et al., 2018). These two pieces of our relevant works establish a compelling model that the metabolic pathway of serine-glycine-threonine functions as a key sleep-regulatory module in response to metabolic sleep cues (e.g., food ingredients and dietary stress). We further hypothesize that the adaptive control of sleep behaviors by select amino acids and their conserved metabolic pathway suggests an ancestral nature of their sleep regulation. Future studies should address if the serine-glycine-threonine metabolic pathway constitutes the sleep homeostat that can sense and respond to different types of sleep needs. In addition, it will be interesting to determine if this metabolic regulation of sleep is conserved among other animals, including humans.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional
information
Genetic reagent (D. melanogaster)w1118Bloomington Drosophila Stock CenterRRID:BDSC_5905
Genetic reagent (D. melanogaster)Canton SKorea Drosophila Resource CenterStock #K211
Genetic reagent (D. melanogaster)CG5955GS20382Kyoto Drosophila Genomics and Genetics ResourcesRRID:DGGR_201409
Genetic reagent (D. melanogaster)Df(3L)BSC797Bloomington Drosophila Stock CenterRRID:BDSC_27369CG5955 deficiency
Genetic reagent (D. melanogaster)Df(3L)BSC839Bloomington Drosophila Stock CenterRRID:BDSC_27917CG5955 deficiency
Genetic reagent (D. melanogaster)rut2080Bloomington Drosophila Stock CenterRRID:BDSC_9405
Genetic reagent (D. melanogaster)DA1dumb2Harvard Medical SchoolRRID:
FlyBase_FBst1017920
Dop1R1f02676
Genetic reagent (D. melanogaster)ELAV-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_458
Genetic reagent (D. melanogaster)GAD1-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_51630
Genetic reagent (D. melanogaster)58H05-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_39198
Genetic reagent (D. melanogaster)Gr5a-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_57591
Genetic reagent (D. melanogaster)Gr33a-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_31425
Genetic reagent (D. melanogaster)Gr66a-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_28801
Genetic reagent (D. melanogaster)Orco-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_26818
Genetic reagent (D. melanogaster)Lkc275Bloomington Drosophila Stock CenterRRID:BDSC_16324
Genetic reagent (D. melanogaster)Df(3L)Exel6123Bloomington Drosophila Stock CenterRRID:BDSC_7602Lk deficiency
Genetic
reagent (D. melanogaster)
Lkrc003Bloomington Drosophila Stock CenterRRID:BDSC_16250
Genetic reagent (D. melanogaster)Df(3L)BSC557Bloomington Drosophila Stock CenterRRID:BDSC_25119Lkr deficiency
Genetic reagent (D. melanogaster)per01PMID: 9630223RRID:BDSC_80917
Genetic reagent (D. melanogaster)ClkJrkPMID: 9630223RRID:BDSC_24515
Genetic reagent (D. melanogaster)PDF-Gal4PMID: 10619432
Genetic reagent (D. melanogaster)UAS-ClkDNTanoue et al., 2004RRID:BDSC_36318
Genetic
reagent (D. melanogaster)
RdlMDRRKyoto Drosophila Genomics and Genetics ResourcesRRID:DGGR_106444
Genetic reagent (D. melanogaster)Rdl1Kyoto Drosophila Genomics and Genetics ResourcesRRID:DGGR_106453
Genetic reagent (D. melanogaster)GABA-TPLBloomington Drosophila Stock CenterRRID:BDSC_19461GABATPL00338, null mutants
Genetic reagent (D. melanogaster)GABA-TFHarvard Medical SchoolRRID:FlyBase_FBst101711GABATf01602, hypomorphic
Genetic reagent (D. melanogaster)GABA-TLLKyoto Drosophila Genomics and Genetics ResourcesRRID:DGGR_141269GABATLL04492, hypomorphic
Genetic reagent (D. melanogaster)UAS-GABA-TChen et al., 2015RRID:FlyBase_FBst0491743
Genetic reagent (D. melanogaster)Df(3L)BSC731Bloomington Drosophila Stock CenterRRID:BDSC_26829GABA-T deficiency
Genetic reagent (D. melanogaster)UAS-shibiretsKitamoto, 2001
Genetic reagent (D. melanogaster)30Y-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_30818
Genetic reagent (D. melanogaster)TH-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_8848
Genetic reagent (D. melanogaster)UAS-mLexA-VP16-NFATMasuyama et al., 2012RRID:BDSC_66542
Genetic reagent
(D. melanogaster)
UAS-Epac1-campsBloomington Drosophila Stock CenterRRID:BDSC_25407
Genetic reagent (D. melanogaster)UAS-CG5955RNAi#1Vienna Drosophila Resource CenterRRID:FlyBase_FBst0452036V15838
Genetic reagent (D. melanogaster)UAS-CG5955RNAi#2Bloomington Drosophila Stock CenterRRID:BDSC_64566
Genetic reagent (D. melanogaster)UAS-KirPMID: 11222642
Genetic reagent (D. melanogaster)UAS-GABAB-R1RNAi#1Vienna Drosophila Resource CenterRRID:FlyBase_FBst0473313V101440
Genetic reagent (D. melanogaster)UAS-GABAB-R1RNAi#2Vienna Drosophila Resource CenterRRID:FlyBase_FBst0490977V330042
Genetic reagent
(D. melanogaster)
UAS-GABAB-R1RNAi#3Bloomington Drosophila Stock CenterRRID:BDSC_51817T51817
Genetic reagent (D. melanogaster)UAS-GABAB-R2RNAi#1Vienna Drosophila Resource CenterRRID:FlyBase_FBst0452890V1784
Genetic reagent (D. melanogaster)UAS-GABAB-R2RNAi#2Vienna Drosophila Resource CenterRRID:FlyBase_FBst0452896V1785
Genetic reagent (D. melanogaster)UAS-GABAB-R3RNAi#1Vienna Drosophila Resource CenterRRID:FlyBase_FBst0468888V50176
Genetic reagent (D. melanogaster)UAS-GABAB-R3RNAi#2Vienna Drosophila Resource CenterRRID:FlyBase_FBst0477558V108036
Chemical compound, drugEOSTokyo Chemical IndustryCat. No. S0445
Chemical compound, drugNipASigmaCat. No. 211672
Chemical compound, drugTHIPTocrisCat. No. 0807Also known as
gaboxadol, 2000x stock
Chemical compound, drugSKF-97541TocrisCat. No. 037910000x stock
Chemical
compound, drug
GABAAcrosCat. No. AC10328025010x stock
Chemical compound, drugPyruvateSigmaCat. No. P2256
Chemical compound, drugTetrodotoxin (TTX)Alomone LabsCat. No. T-5501000x stock
Chemical compound, drugcaffeineAlfa AesarCat. No. A104311000x stock
AntibodyMouse anti-GFP, monoclonalUC Davis/NIH NeuroMab FacilityRRID:AB_106719551:1000 dilution
AntibodyRabbit anti-GABA, polyclonalSigmaRRID:AB_4776521:2000 dilution
AntibodyRabbit anti-TH, polyclonalMilliporeRRID:AB_3902041:1000 dilution
AntibodyDonkey anti-Mouse AF488Jackson ImmunoresearchRRID:AB_23408461:600 dilution
AntibodyDonkey anti-Rabbit AF594Jackson ImmunoresearchRRID:AB_23406211:600 dilution

Sleep analyses

Request a detailed protocol

All behavioral tests were performed using individual male flies, unless otherwise indicated. Each fly was housed in a 65 × 5 mm glass tube containing 5% sucrose and 2% agar (behavior food). For amino acid supplements, the indicated amount of each amino acid was dissolved in the behavior food. For oral administration of GABA-T or GABA transporter inhibitors, 10 mM of EOS (Tokyo Chemical Industry) or 10 mg/ml of NipA (Sigma) was directly dissolved in the behavior food containing the indicated amount of threonine. For oral administration of GABA receptor agonists, 10 mg/ml of THIP (Tocris) or SKF-97541 (Tocris) stock solution was diluted into the behavior food at the indicated final concentration. Flies were fed on amino acid- and/or drug-containing behavior food in LD cycles at 25°C for 4.5 days. Locomotor activity was recorded using the DAM system (Trikinetics) and quantified by the number of infrared beam crosses per minute. Sleep bouts were defined as no activity for >5 min. Sleep parameters were analyzed using an Excel macro (Pfeiffenberger et al., 2010).

Measurements of arousal threshold and sleep latency after arousal

Request a detailed protocol

The arousal threshold to mechanical stimuli was measured as described previously (Wu et al., 2008) with minor modifications. Locomotor activities were recorded similarly as in the sleep analyses, while behavioral test tubes containing individual male flies were scraped with a thin wood stick at zeitgeber (ZT) 16 (lights-on at ZT0; lights-off at ZT12) during the fourth LD cycle. Mechanical stimuli used in our tests include: 1) scraping sound and vibration without direct scraping (a weak stimulus), 2) gentle scraping (a moderate stimulus), and 3) hard scraping repeated 3–4 times (a strong stimulus). Flies were defined as aroused if they displayed inactivity for >5 min prior to the stimulus but showed any stimulus-induced locomotor response within 10 min. The percentage of aroused flies was calculated per each group in individual experiments and averaged from three independent experiments. Latency to sleep onset after the arousal was calculated in individual flies and averaged per each group. To measure the arousal threshold to a light stimulus, LD-entrained flies were exposed to a 1 min light pulse at ZT16 instead of the mechanical stimuli. The percentage of light-aroused flies and sleep latency after the light-induced arousal were measured similarly as above.

Video analyses of sleep and locomotor activity

Request a detailed protocol

Wild-type male flies were placed individually into the video-tracking arena (diameter x height = 16 mm x 2 mm) in a 24-well plate filled with the behavior food (5% sucrose +2% agar±25 mM threonine) (day 0). Flies were entrained in 12 hr light:12 hr dim red light (red LED) cycles at 25°C before 24 hr time-lapse images were obtained at 0.3–1 Hz using HandyAVI software (AZcendant) on day 4. Their positional changes in X- and Y-axes were calculated from two consecutive frames of the time-lapse images per each arena. Any positional difference larger than eight pixels was considered as a movement. A window of the time frames with no positional change for >5 min was defined as a sleep bout. Additional parameters for sleep or locomotor activity were analyzed using Excel. For the higher-resolution analysis of locomotor behaviors, male flies were pre-fed on control or amino acid-containing behavior food for four LD cycles at 25°C. After brief anesthetization, flies were individually placed into 6-well plates (diameter x height = 35 mm x 2 mm). After 25 min of habituation, time-lapse images were obtained at 10 Hz using HandyAVI software (AZcendant). Approximately 3000 frames (corresponding to a 5 min video recording) were analyzed using ImageJ software to quantify locomotor activity in individual flies as described above.

Aversive phototaxic suppression (APS)

Request a detailed protocol

An APS-based short-term memory test was performed as described previously (Dissel et al., 2015; Seugnet et al., 2009). Briefly, adult male flies were individually housed and fed either control or threonine-containing behavior food for four LD cycles. A single fly was placed in the dark chamber of a T-maze without anesthesia. A filter paper (3M) was soaked with 180 μL of 1 μM quinine hydrochloride solution (Sigma) and was placed in the light chamber to give aversive condition in concordance with a light stimulus. After 1 min of habituation in the T-maze, a middle bridge between two chambers was opened and the light source was gradually turned on. Any fly which did not move to the light chamber at the first trial was excluded from further analysis. If a fly entered the light chamber within 20 secs, it was considered as a pass. The whole procedure was repeated 16 times in four sessions (four trials/session) at 1 min intervals. A performance index was calculated per each fly by the percentage of ‘non-pass’ in the last session.

Whole-brain imaging

Request a detailed protocol

Transgenic flies were fed on control or amino acid-containing behavior food for four LD cycles at 25°C prior to imaging experiments. Whole brains were dissected in phosphate-buffered saline (PBS) and fixed in PBS containing 3.7% formaldehyde. Fixed brains were washed three times in PBS containing 0.3% Triton X-100 (PBS-T), blocked in PBS-T containing 0.5% normal goat serum, and then incubated with mouse anti-GFP (NeuroMab) and rabbit anti-GABA (Sigma) antibodies for 2 days at 4°C. After washing three times in PBS-T, brains were further incubated with anti-mouse Alexa Fluor 488 and anti-rabbit Alexa Fluor 594 antibodies (Jackson ImmunoResearch) for 1 day at 4°C, washed three times with PBS-T, and then mounted in VECTASHIELD mounting medium (Vector Laboratories). Confocal images of whole-mount brains were acquired using a Multi-Photon Confocal Microscope (LSM780NLO, Carl Zeiss) and analyzed using ImageJ software.

In vivo Epac1-camps imaging

Request a detailed protocol

Transgenic flies fed either control or threonine-containing behavior food were anesthetized in ice. A whole brain was briefly dissected in hemolymph-like HL3 solution (5 mM HEPES pH 7.2, 70 mM NaCl, 5 mM KCl, 1.5 mM CaCl2, 20 mM MgCl2, 10 mM NaHCO3, 5 mM Trehalose, 115 mM Sucrose) and then placed on a 25 mm round coverslip. A magnetic imaging chamber (Chamlide CMB, Live Cell Instrument) was assembled on the coverslip and filled with 900 μl of HL3 solution. Where indicated, 100 μl of 10x GABA stock solution in HL3 was added to the imaging samples. Live-brain images were acquired at ~1 Hz using a multi-photon confocal microscope (LSM780NLO, Carl Zeiss) with a Plan-Apochromat 40x/1.3 oil lens. The power of a 458 nm-laser projection was 3% at a pixel resolution of 256 × 256. Each frame constituted two slices by ~5 µm of step sizes. Gallium arsenide phosphide (GaAsP) detectors were set by two ranges (473–491 nm and 509–535 nm) for ECFP and EYFP channels, respectively. Pinhole was fully opened (599 µm) to avoid any subtle z-drift during the image acquirement. The fluorescence intensities of CFP and YFP were quantified using ZEN software (Carl Zeiss) and any changes in FRET signals were calculated in Excel.

Quantitative PCR

Request a detailed protocol

Total RNA was purified from 10 flies per each genotype (five males and five females) using Trizol Reagent, according to the manufacturer’s instructions (Thermo Fisher Scientific). cDNA was prepared from DNase I-treated RNA samples using the M-MLV Reverse Transcriptase reagent (Promega) and random hexamers. Diluted cDNA samples were quantitatively analyzed by SYBR Green-based Prime Q-Mastermix (GeNet Bio) and gene-specific primers using the LightCycler 480 real-time PCR system (Roche). To validate the efficiency of transgenic RNA interference, total RNAs from head or body extracts were analyzed similarly.

Quantification of threonine levels

Request a detailed protocol

Quantitative measurement of threonine was performed as described previously (Liu et al., 2014b) with minor modifications. Briefly, 30 female flies were homogenized in 200 μL of PBS containing 0.05% Triton X-100. Whole-body extracts were clarified twice by centrifugation, and total proteins in the extracts were quantified using the Pierce BCA Protein Assay Kit according to manufacturer’s instructions (Thermo Fisher Scientific). After boiling, soluble extracts were further clarified by centrifugation and subjected to an enzymatic reaction. Each reaction mixture included 40 μL of 5 × HEPES reaction buffer (500 mM HEPES pH 8.0, 1 mM NADH, 0.25 mM pyridoxal 5-phosphate, and 5 mM dithiothreitol), 160 μL of soluble body extracts, and 1 U of alcohol dehydrogenase (Sigma). In parallel, control reactions with a serial dilution of threonine stock solution (16 mM) were used to generate a standard curve for quantification. The enzymatic reactions were set up in a 96-well microplate (Corning) and incubated for 30 min at 4°C followed by 10 min incubation at 25°C. Absorbance at 340 nm was measured for each reaction mixture using an Infinite M200 microplate reader (Tecan) before 1 µL of bacterially purified L-threonine aldolase (LTA) was added to each reaction mixture. The reaction mixture was further incubated at 37°C for 5 min and post-LTA absorbance was measured to calculate decreases in NADH levels.

Protein purification of L-threonine aldolase

Request a detailed protocol

The coding sequence of LTA was PCR-amplified from genomic DNA of Pseudomonas aeruginosa (a gift from R.J. Mitchell) and cloned into a modified pDuet vector (a gift from C. Lee). Bacterial purification of His-tagged LTA proteins using Ni-NTA Agarose (Qiagen) was performed as described previously (Lee et al., 2017). Purified proteins were dialyzed using a dialysis buffer (50 mM NaH2PO4, pH 8.0, 10 μM pyridoxal 5-phosphate, and 1 mM dithiothreitol), diluted in 50% glycerol, quantified using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific), and stored at −80°C prior to use.

Quantitative analyses of free amino acids and energy metabolites

Request a detailed protocol

Wild-type male flies were loaded on to standard cornmeal-yeast-agar food containing either 0 mM (control) or 50 mM threonine, and then entrained in LD cycles at 25°C for 4 days before harvest. Extracts were prepared from 100 fly heads per condition and the relative levels of free amino acids were measured using ion exchange chromatography as described previously (Sonn et al., 2018). For quantification of energy metabolites, fly heads were homogenized in 400 µl of chloroform/methanol (2/1, v/v) and clarified by centrifugation. The supernatant was dried by vacuum centrifugation, and then reconstituted with 50 μL of 50% acetonitrile prior to liquid chromatography-tandem mass spectrometry analysis using 1290 HPLC (Agilent), Qtrap 5500 (ABSciex), and a reverse phase column (Synergi fusion RP 50 × 2 mm).

Statistics

Request a detailed protocol

Appropriate sample sizes were not determined by statistical computation but based on those reported in previous studies. For all the analyses, ‘n’ refers to the total number of biological replicates which were tested in more than two independent experiments, unless otherwise indicated in figure legends. For immunofluorescence assay, ‘n’ refers to the total number of brain hemispheres which were tested in 2–4 independent experiments. For cAMP imaging, ‘n’ refers to the total number of brains which were tested in 2–3 independent experiments. Individual flies were allocated into each group of biological replicates by their specific diet or genotype. Raw sleep data were collected non-blindly to the conditions but analyzed by an automated macro program. For immunofluorescence assay, GFP-positive neurons were scored in a way of double-blinded to the conditions. Short-term memory tests were performed blindly to the conditions. All the statistical analyses were performed using Prism (GraphPad Software, Inc) as described in figure legends. F distributions with degrees of freedom were indicated by F[DFn, DFd]. All the P values from post hoc tests after one-way or two-way ANOVA were corrected for multiple comparisons. Violin plots present mean ± 95% confidence intervals and were generated using Python with the help of Seaborn library. Bar graphs indicate mean ± SEM and were generated using Excel.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
    Introduction to Nutrition and Metabolism, 267
    1. DA Bender
    (2014)
    Boca Raton: CRC Press.
  11. 11
    Effects of alpha-ketobutyrate and alpha-hydroxybutyrate on the enzymatic determination of acetoacetate and beta-hydroxybutyrate
    1. RA Beyerinck
    2. EP Brass
    (1987)
    Clinical Chemistry 33:1469–1470.
  12. 12
    A two process model of sleep regulation
    1. AA Borbély
    (1982)
    Human Neurobiology 1:195–204.
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
    The effects of GABAtransaminase (GABA-T) inhibition on sleep and behavior of the cat
    1. SW Holmes
    2. D Sugden
    (1975)
    Sleep Res, 4.
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
    The antispastic effect of L-threonine
    1. K-C Lee
    2. V Patterson
    3. G Roberts
    4. E Trimble
    (1990)
    In: G Lubec, G. A Rosenthal, editors. Amino Acids. Dordrecht: Springer. pp. 658–663.
    https://doi.org/10.1007/978-94-011-2262-7_78
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
    Gamma-hydroxybutyric acid
    1. OC Snead
    2. KM Gibson
    (2005)
    The New England Journal of Medicine 352:2721–2732.
    https://doi.org/10.1056/NEJMra044047
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72

Decision letter

  1. K VijayRaghavan
    Senior and Reviewing Editor; National Centre for Biological Sciences, Tata Institute of Fundamental Research, India

In the interests of transparency, eLife includes the editorial decision letter, peer reviews, and accompanying author responses.

[Editorial note: This article has been through an editorial process in which the authors decide how to respond to the issues raised during peer review. The Reviewing Editor's assessment is that all the issues have been addressed.]

Thank you for submitting your article "Threonine enhances sleep drive via a circadian clock-independent GABAergic pathway in for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by K VijayRaghavan as the Reviewing Editor and Senior Editor. The reviewers have opted to remain anonymous.

The Reviewing Editor has highlighted the concerns that require revision and/or responses, and we have included the separate reviews below for your consideration. If you have any questions, please do not hesitate to contact us.

Summary:

This manuscript from the Lim laboratory follows upon their recent work showing that serine modulates sleep in Drosophila. The focus here is on threonine, which appears to be sleep-promoting, likely through an effect on GABA signaling. The idea that metabolic factors are important regulators of sleep is gaining recognition, and this study is an interesting addition in that regard. As written though, this big picture view of the work does not come across until the end of the Discussion; laying it out earlier in the manuscript (e.g. Abstract and Introduction) would significantly boost the appeal of the work. From a scientific perspective, it would be good to have the mechanism nailed. This will require attention to the major points in each of the reviews below. Attending to these will likely involve experiments that will take about 2-3 months in addition to the time needed for re-writing parts of the manuscript.

Major concerns that require revision and/or responses.

These concerns derive from our consultation on the two reviews. They highlight specific points and need to be seen along with the two reviews.

1) The scope of the paper's title gets lost in detail. As pointed out in the summary above, the big overview needs to be linked to the core of every section of the Results.

2) The fundamental matter is that the authors have not nailed the mechanism. Ideally, they should show how GABA signaling is affected, but at the very least address the points made by the reviewers or satisfactorily exclude clock cells

3) The authors suggest that Threonine and GABA are competitive substrates. Yet, the results suggest that an explanation could be at the level of GABA synthesis presynaptically (Figure 4—figure supplement 6). This need to be better discussed.

4) Both reviewers have pointed out the need to connect neuronal CalExa signals with threonine to sleep in a robust way. For example, do GAD+ cells with increased CaLexA signal have anything to do with sleep? Although this may be much more difficult to show. (Are there known GAD+, sleep-relevant drivers the authors could use?)

5) The GAD-Gal4 shibire experiment is not making a persuasive point. Maybe it should be repeated? Please particularly note the points about rutabaga made by reviewer #1. Both reviewers feel that this is an important experimental point to address.

6) With regard to reviewer #2's point 3, our consultation agreed in the general principle that multiple alleles should be used for each gene, but if some of these genes fall into the same pathway and/or corroborate findings made in other ways (pharmacology), then this could be seen as mitigating,

To deal with potential developmental roles, it will be good to see at least one adult- specific manipulation and phenotype of GABA manipulation.

Separate reviews (please respond to each point):

Reviewer #1:

The authors favor the idea that threonine acts through GABA signaling, pointing out that threonine and GABA and their metabolic derivatives are competitive substrates? For what-for the GABA receptor? This would explain why GABA-T mutants are unresponsive to threonine as they have high extracellular GABA, and would be consistent with the modest block produced by threonine in GABA action in an ex vivo preparation. However, how does this explain interactions of threonine with presynaptic GABA-CaLexa signals are increased in GAD neurons by threonine and blocking GAD neurons with shibire is stated to attenuate effects of threonine?

Does the CaLexa signal increase in non-GABA cells with threonine treatment? The manuscript shows specificity for threonine by using another amino acid as control (arginine), but what about specificity for GABAergic cells?

The blocking effect of GAD-GAL4 driven shibire on threonine action is questionable (Figure 4) as the threonine effect in controls in this experiment is even smaller than normal, and even normally it is not great. Is this really relevant (even if significant)?

The comparisons in Figure 4B are hard to follow. If the point is that threonine has no effects when GABAB-R1 is knocked down, then it should be shown that threonine increases sleep in the GAL4 alone control, but not in the knockdown, relative to the same genotypes not treated with threonine (from the data, it looks like threonine does increase sleep, probably significantly, in the knockout). Instead, control and knockdown are compared with each other, both in the presence of threonine. What this says is that knockdown of GABAB-R1 does not have any effect in threonine-treated flies, which is not the point (I think).

If the learning assay with rutabaga is to show that increased sleep (with threonine) rescues learning in rutabaga, then the authors should show that blocking the sleep increase (with deprivation) abrogates rescue. Otherwise, the direct effects of threonine on learning cannot be excluded.

Does threonine or the threonine-increasing mutant display any kind of metabolic phenotype, as does the GABAT mutant (Maguire et al., 2015)? Given that the underlying mechanisms are the same or overlapping, it is worth at least discussing this point.

Additional data files and statistical comments:

The authors use Student's t-test a lot, which is not ideal or stringent enough for most measurements.

Reviewer #2:

The manuscript by Ki and Lim demonstrates sleep-promoting effects of dietary threonine (SPET) in Drosophila. The authors make several interesting claims regarding the mechanism of SPET: 1) decreased sleep latency in response to dietary threonine (Thr) does not require functional clock genes or pacemaker neurons; 2) GABA signaling through GABAB-R1 is involved in SPET; and 3) Thr activates a subset of GABAergic neurons, reduces GABA responsiveness of sleep-relevant R2 EB neurons, and rescues memory defects of rutabaga mutants; and 4) genetic manipulations that increase endogenous Thr levels enhance sleep. These are potentially interesting findings, but additional data are required to strengthen some of the claims.

Major points:

1) To demonstrate the relevance to GABA signaling for SPET, the authors show increased CaLexA signal in response to Thr in neurons lateral to the antennal lobe, which they call LNs. However, they do not show that these neurons have any role in sleep. Nor do they show any increase in CaLexA signal in known sleep centers in response to Thr.

2) The authors claim that a slightly reduced GABA response to Thr in R2 EB neurons implies that threonine-induced behavioral quiescence is physiologically relevant to sleep. However, it is unclear whether the effect is sufficient to contribute significantly to SPET. Does R2 EB specific knockdown of GABAB-R1 affect SPET?

3) The involvement of GABA-T in SPET is supported by a single allelic combination (hypomorph over deficiency) and that of GABAB-R1 is based on a single RNAi construct. The involvement of CG5955 is based on one allelic combination and one RNAi. To show that a gene is involved in a process, it is important to provide converging evidence from multiple sources: e.g., multiple mutant alleles or allelic combinations, multiple RNAi constructs to minimize the potential off-target effects, and rescue with transgene expression.

4) Potential developmental defects could be a confounding factor for the effects of GABA-T, GABAB-R1, and CG5955 on SPET. Adult specific knockdown or developmental rescue could alleviate the concern.

5) The data behind the claim that Rdl and PDF neurons are not involved in SPET are not compelling. In Figure 2—figure supplement 1B, Rdl MMDR/1 mutants seem to have reduced SPET (i.e., the difference between – and + Thr is smaller in mutants than het controls). If the reduction is not statistically significant, it may be due to the unusually large error bar in one of the conditions. Additional data may provide a clearer picture.

A related concern is that without Thr, sleep latency is not reduced in Rdl MMDR/1 mutants relative heterozygous controls (B), as would be expected based on previous reports. Similarly, without Thr, PDF and CRY neuron silencing does not lead to sleep latency reduction (C). These unexpected features of the data raise concerns about the validity of the rest of the data.

6) Rescue of memory defects in rut mutants by Thr is interesting. However, it would strengthen the conclusion if additional memory mutants were tested. Please see reviewer 1 's points on this,

Minor Comments:

1) It is unclear how the effects of Thr vary as a function of time-of-day. It would be informative to show these data in a format where sleep amount per 30min or 1h is plotted as a function of time of day.

2) To further examine the clock- and light-dependence of SPET, it would be helpful to repeat the basic +/- Thr experiments in DD and LL.

3) As the authors note, GABA-T effects on SPET could be due to a floor effect. Although pharmacological data somewhat alleviate this concern, a genetic demonstration (using hypomorphs?) would strengthen the claim.

4) In Figure 1A and related figures, it is unclear whether p values were corrected for multiple comparisons.

5) Figure 1—figure supplement 3. In this and other similar figures, to determine whether SPET is affected by an experimental manipulation (in this case, silencing of gustatory or olfactory neurons), it is important to test whether the difference between + and – Thr is different across the three genotypes.

6) Discussion section: "…SPET facilitates sleep onset in a manner independent of light". Where are the data for light-independence of SPET?

7) Figure 2—figure supplement 1A and B. Some of the error bars are so big that it is difficult to believe the differences are significant at p<0.001 as indicated. In these situations, standard ANOVAs that assume equal variance across conditions are not appropriate.

8) In Figure 3A, the CaLexA signal for LNs is not symmetric. Is this common?

9) Figure 4B is formatted differently from all other figures. Why?

10). Some of the N's for behavioral experiments are as low as 8. This seems too low.

11) SEM, which takes into account N's, would more informative than SD.

Additional data files and statistical comments:

As noted in the minor point #7, some of the statistics do not seem appropriate.

https://doi.org/10.7554/eLife.40593.031

Author response

Major concerns that require revision and/or responses.

These concerns derive from our consultation on the two reviews. They highlight specific points and need to be seen along with the two reviews.

1) The scope of the paper's title gets lost in detail. As pointed out in the summary above, the big overview needs to be linked to the core of every section of the Results.

We modified our title, Abstract, and Introduction to better state our view regarding the metabolic control of sleep behaviors in the revised manuscript.

2) The fundamental matter is that the authors have not nailed the mechanism. Ideally, they should show how GABA signaling is affected, but at the very least address the points made by the reviewers or satisfactorily exclude clock cells

In our revised manuscript, we provided new pieces of evidence that exclude the possible requirement of clock-dependent control of sleep in SPET. First, SPET was detectable even in constant light (LL) where wild-type flies lose circadian rhythms and sleep-wake cycles (Figure 2C and D; Figure 2—figure supplement 2). Second, a genetic loss of per or Clk shortened or lengthened sleep latency, respectively, in LD cycles (Liu et al., 2014) while these arrhythmic clock mutants displayed SPET robustly (Figure 2A). Third, PDF neuron-specific overexpression of dominant-negative CLK proteins lengthened sleep latency (Liu et al., 2014) whereas the loss of molecular clocks in PDF neurons did not compromise SPET (Figure 2B).

We further showed that transgenic depletion of metabotropic GABA receptors in R2 EB neurons promoted sleep non-additively with SPET (Figure 5C). Given that dietary threonine reduced metabotropic GABA responses in R2 EB neurons, these results support that SPET may involve the down-regulation of metabotropic GABA transmission in this specific neural locus for generating homeostatic sleep drive. Please see our point-by-point responses to additional points made by each reviewer below.

3) The authors suggest that Threonine and GABA are competitive substrates. Yet, the results suggest that an explanation could be at the level of GABA synthesis presynaptically (Figure 4—figure supplement 6). This need to be better discussed.

We actually found that GABA levels were low in threonine-fed flies compared to control (Figure 4A). While this could be a compensatory decrease, additional lines of our evidence indicated that suppression of the metabotropic GABA transmission indeed induced SPET-like sleep phenotypes in control-fed flies and displayed non-additive effects with SPET in threonine-fed flies (please see our responses to the reviewer comment #1-1 and the reviewer #2, major point #2 below). We thus reasoned that dietary threonine decreased GABA levels likely as a metabolic consequence important for SPET and better discussed it in our revised manuscript.

4) Both reviewers have pointed out the need to connect neuronal CalExa signals with threonine to sleep in a robust way. For example, do GAD+ cells with increased CaLexA signal have anything to do with sleep? Although this may be much more difficult to show. (Are there known GAD+, sleep-relevant drivers the authors could use?)

Given our revised model for the implication of GABA transmission in SPET, we toned down our conclusion from the CaLexA experiment and discussed a possible explanation for the CaLexA phenotypes in the revised manuscript. Please see our specific responses to the reviewer comments #1-1 and #1-2 as well as the reviewer #2, major points #1 and 2 below.

5) The GAD-Gal4 shibire experiment is not making a persuasive point. Maybe it should be repeated?

In our revised manuscript, we provided more convincing data from the GAD1-Gal4 shibire experiments (Figure 4B). Please see our response to the reviewer comment #1-3 below.

Please particularly note the points about rutabaga made by reviewer #1. Both reviewers feel that this is an important experimental point to address.

In our revised manuscript, we provided new data that dietary threonine rescued memory deficits in another memory mutants (Figure 6B). Using caffeine-induced sleep suppression, we further showed that this memory rescue required threonine-induced sleep (Figure 6B and C). Please see our responses to the reviewer comment #1-5 and the reviewer #2, major point #6.

6). With regard to reviewer #2's point 3, our consultation agreed in the general principle that multiple alleles should be used for each gene, but if some of these genes fall into the same pathway and/or corroborate findings made in other ways (pharmacology), then this could be seen as mitigating,

To deal with potential developmental roles, it will be good to see at least one adult- specific manipulation and phenotype of GABA manipulation.

In our revised manuscript, we provided converging evidence from multiple genetic and transgenic sources (e.g., multiple allelic combinations, transgenic rescue, multiple RNAi transgenes) to validate the function of a given gene and the implication of a metabotropic GABA pathway in threonine-relevant sleep regulation. While our pharmacological manipulations of GABA transmission independently confirmed the genetic evidence, oral administration of GABA-relevant inhibitors and agonists to adult flies also excluded the possible developmental effects of GABA manipulations. These adult-specific, pharmacological manipulations of GABA transmission included 1) GABA-T inhibitor (Figure 3B), 2) GABA transporter inhibitor (Figure 3B), and 3) agonists of ionotropic or metabotropic GABA receptors (Figure 4D). We further showed that a conditional blockade of synaptic transmission from adult neurons expressing GAD1 impaired SPET, supporting adult-specific effects of the transgenic GABA manipulation on SPET (Figure 4B). Please see our specific responses to reviewer #2, major points #3 and #4 below.

Separate reviews (please respond to each point):

Reviewer #1:

The authors favor the idea that threonine acts through GABA signaling, pointing out that threonine and GABA and their metabolic derivatives are competitive substrates? For what-for the GABA receptor? This would explain why GABA-T mutants are unresponsive to threonine as they have high extracellular GABA, and would be consistent with the modest block produced by threonine in GABA action in an ex vivo preparation. However, how does this explain interactions of threonine with presynaptic GABA-CaLexa signals are increased in GAD neurons by threonine and blocking GAD neurons with shibire is stated to attenuate effects of threonine?

The structural homology among threonine, GABA, and their derivatives led us to the hypothesis that these relevant chemicals may act as competitive substrates in enzymatic reactions for their overlapping metabolism. As the reviewer suggested above, they may similarly act as competitive ligands for GABA receptors (i.e., metabotropic GABA receptors, in particular), explaining weaker GABA responses in R2 EB neurons of threonine-fed flies than those in control.

In our revised manuscript, we further found that threonine-fed flies displayed low levels of GABA and glutamate compared to those in control-fed flies (Figure 4A). Although this could be a compensatory decrease, a simple hypothesis would be that dietary threonine limits the total flux of GABA/glutamate/glutamine cycle possibly through substrate competition, decreases the size of available GABA pool, and thereby down-scales GABA transmission for SPET. This model is consistent with our observations that genetic suppression of GABA transmission by the conditional blockade of synaptic transmission in GAD1-expressing neurons or by the transgenic depletion of metabotropic GABA receptors in R2 EB neurons drives sleep in control-fed flies (i.e., mimics SPET in control-fed flies) and displays non-additive effects with SPET in threonine-fed flies. By contrast, high levels of extracellular GABA (e.g., GABA-T mutants or EOS/NipA-fed flies) or pharmacological enhancement of the metabotropic GABA transmission (e.g., oral administration of the GABA-B receptor agonist) may interfere with this process, thereby conferring the resistance to SPET.

Given this revised model, we reason that some group of GAD1 neurons (i.e., LN) may express presynaptic GABA receptors to suppress their own GABA transmission via a negative feedback. Threonine diet could then de-repress the auto-inhibitory GABA transmission, elevating the CaLexA signals selectively in those GAD1 neurons. Accordingly, we revised our Results and Discussion to clarify this model and better interpret our results.

Does the CaLexa signal increase in non-GABA cells with threonine treatment? The manuscript shows specificity for threonine by using another amino acid as control (arginine), but what about specificity for GABAergic cells?

As the reviewer suggested, we tested additional Gal4 drivers expressed in other sleep-regulatory loci (i.e., mushroom body and dopaminergic neurons) in the CaLexA experiments and did not see any detectable increase in their CaLexA signals upon threonine diet (Figure 4—figure supplement 4D and E).

The blocking effect of GAD-GAL4 driven shibire on threonine action is questionable (Figure 4) as the threonine effect in controls in this experiment is even smaller than normal, and even normally it is not great. Is this really relevant (even if significant)?

As the editor suggested above, we repeated the shibire experiment in a different incubator with better control of the internal temperature and provided more convincing data in our revised manuscript (Figure 4B). We confirmed that overexpression of the temperature-sensitive shibire in GAD1-expressing neurons induced sleep in control-fed condition and masked SPET only at restrictive temperature. We also noted that low (i.e., permissive) temperature delayed SPET even in heterozygous controls, explaining smaller SPET in our original data.

The comparisons in Figure 4B are hard to follow. If the point is that threonine has no effects when GABAB-R1 is knocked down, then it should be shown that threonine increases sleep in the GAL4 alone control, but not in the knockdown, relative to the same genotypes not treated with threonine (from the data, it looks like threonine does increase sleep, probably significantly, in the knockout). Instead, control and knockdown are compared with each other, both in the presence of threonine. What this says is that knockdown of GABAB-R1 does not have any effect in threonine-treated flies, which is not the point (I think).

We thought that the comparisons in the original Figure 4B would better visualize the effects of neuronal GABAB-R1 depletion on SPET among other genotypes. Given the comments from both reviewers (please also see the reviewer #2, minor comment #9), we realized that the different format was rather confusing. Accordingly, we presented all our behavioral data in the same format in the revised manuscript.

If the learning assay with rutabaga is to show that increased sleep (with threonine) rescues learning in rutabaga, then the authors should show that blocking the sleep increase (with deprivation) abrogates rescue. Otherwise, the direct effects of threonine on learning cannot be excluded.

In our revised manuscript, we showed that dietary threonine also rescued memory deficits in dumb mutants (Figure 6B). Co-administration of caffeine with threonine substantially blocked sleep induction as well as memory rescue in dumb mutants (Figure 6B and C). By contrast, caffeine alone did not significantly affect short-term memory in control and dumb mutants under our experimental condition. These data thus support that dietary threonine rescues memory mutants in a sleep-dependent manner.

Does threonine or the threonine-increasing mutant display any kind of metabolic phenotype, as does the GABAT mutant (Maguire et al., 2015)? Given that the underlying mechanisms are the same or overlapping, it is worth at least discussing this point.

To examine any metabolic phenotypes induced by dietary threonine, we compared the relative levels of free amino acids and energy metabolites between control- and threonine-fed flies. Threonine diet lowered GABA and glutamate levels, but it did not significantly affect ATP levels or the ratio of NAD+ to NADH (Figure 4A and Figure 4—figure supplement 1). While dietary glutamate rescued metabolic phenotypes in GABA-T mutants (Maguire et al., 2015), it affected neither long sleep in GABA-T mutants (Maguire et al., 2015) nor SPET (Figure 4—figure supplement 3), indicating that low glutamate levels do not explain either sleep phenotypes. Accordingly, we compared and discussed these metabolic phenotypes in GABA-T mutants and threonine-fed flies in our revised manuscript.

Additional data files and statistical comments:

The authors use Student's t-test a lot, which is not ideal or stringent enough for most measurements.

Wherever possible, we avoided Student’s t-test to increase the stringency for our statistical analyses in the revised manuscript. We also clarified our statistical analyses in the text, figures, and figure legends of our revised manuscript.

Reviewer #2:

The manuscript by Ki and Lim demonstrates sleep-promoting effects of dietary threonine (SPET) in Drosophila. The authors make several interesting claims regarding the mechanism of SPET: 1) decreased sleep latency in response to dietary threonine (Thr) does not require functional clock genes or pacemaker neurons; 2) GABA signaling through GABAB-R1 is involved in SPET; and 3) Thr activates a subset of GABAergic neurons, reduces GABA responsiveness of sleep-relevant R2 EB neurons, and rescues memory defects of rutabaga mutants; and 4) genetic manipulations that increase endogenous Thr levels enhance sleep. These are potentially interesting findings, but additional data are required to strengthen some of the claims.

Major points:

1) To demonstrate the relevance to GABA signaling for SPET, the authors show increased CaLexA signal in response to Thr in neurons lateral to the antennal lobe, which they call LNs. However, they do not show that these neurons have any role in sleep. Nor do they show any increase in CaLexA signal in known sleep centers in response to Thr.

As mentioned in our response to the reviewer comment #1-1 above, we did not see any detectable increase in the CaLexA signals from sleep-regulatory mushroom body or dopaminergic neurons upon threonine diet (Figure 4—figure supplement 4D and E). Although the sensitivity of CaLexA may intrinsically limit the detectable size and duration of Ca2+ changes in our experiment, these results support the relative specificity of Ca2+ response in LN to the threonine diet. Lack of a specific Gal4 driver that targets these LN only, however, did not allow us to validate that these effects were necessary for SPET. Nonetheless, it led us to find that a conditional blockade of synaptic transmission in GAD1-expressing neurons, including LN, actually induced sleep in control-fed flies, and masked SPET in threonine-fed flies. Given our revised model for the implication of GABA transmission in SPET, we toned down our conclusion from the CaLexA experiment and discussed a possible explanation for the CaLexA phenotypes in the revised manuscript.

2) The authors claim that a slightly reduced GABA response to Thr in R2 EB neurons implies that threonine-induced behavioral quiescence is physiologically relevant to sleep. However, it is unclear whether the effect is sufficient to contribute significantly to SPET. Does R2 EB specific knockdown of GABAB-R1 affect SPET?

In our revised manuscript, we tested several RNAi transgenes to individually deplete metabotropic GABA receptors in R2 EB neurons and examined their effects on SPET. Depletion of GABAB-R2 or GABAB-R3 in R2 EB neurons indeed enhanced sleep drive non-additively with SPET (Figure 5C). This contrasted with our observations that genetic elevation of GABA levels or pharmacological enhancement of the metabotropic GABA transmission suppressed SPET, but it did not comparably affect sleep drive in control-fed flies (Figure 3A and Figure 4D). Considering our additional finding that dietary threonine reduces GABA levels, we revised our model for SPET and proposed that 1) sleep drive by low metabotropic GABA transmission in R2 EB neurons may mediate SPET, and 2) genetic/pharmacological elevation of the GABA transmission may interfere with this process, thereby masking SPET. We revised our manuscript accordingly.

3) The involvement of GABA-T in SPET is supported by a single allelic combination (hypomorph over deficiency) and that of GABAB-R1 is based on a single RNAi construct. The involvement of CG5955 is based on one allelic combination and one RNAi. To show that a gene is involved in a process, it is important to provide converging evidence from multiple sources: e.g., multiple mutant alleles or allelic combinations, multiple RNAi constructs to minimize the potential off-target effects, and rescue with transgene expression.

As the reviewer suggested, we provided converging evidence from multiple sources in the revised manuscript. These included 1) multiple allelic combinations (GABA-T, CG5955), 2) transgenic rescue (GABA-T), 3) multiple RNAi transgenes (metabotropic GABA receptors, CG5955), and 4) adult-specific, pharmacological manipulations of target gene function (GABA-T, GABA transporter, GABA receptors) to exclude any development effects on SPET.

4) Potential developmental defects could be a confounding factor for the effects of GABA-T, GABAB-R1, and CG5955 on SPET. Adult specific knockdown or developmental rescue could alleviate the concern.

We have been testing adult-specific knockdown or developmental rescue using tubGal80ts. Unfortunately, heterozygous controls of the key UAS transgenes (e.g., UAS-GABA-T, UAS-GABA receptor RNAi) displayed Gal4-independent effects on the genetic rescue or SPET at high temperature. Although it was likely due to their leaky expression, this limited our experimental conditions to genetically validate the developmental defects. Nonetheless, we provided a series of compelling evidence that supports adult-specific effects of GABA manipulations on SPET. First, adult-specific administration of GABA-T inhibitor or GABA transporter inhibitor potently suppressed SPET (Figure 3B). Second, an adult-specific blockade of synaptic transmission in GAD1-expressing cells induced sleep drive non-additively with SPET (Figure 4B). Third, adult-specific administration of the agonist of metabotropic GABA receptors, but not of ionotropic GABA receptors, suppressed SPET (Figure 4D). Regarding CG5955, we could not fully exclude the possibility of metabolic compensation or developmental effects using transgenic reagents available to us. So, we toned down our conclusion and stated these possibilities in our revised manuscript.

5) The data behind the claim that Rdl and PDF neurons are not involved in SPET are not compelling. In Figure 2—figure supplement 1B, Rdl MMDR/1 mutants seem to have reduced SPET (i.e., the difference between – and + Thr is smaller in mutants than het controls). If the reduction is not statistically significant, it may be due to the unusually large error bar in one of the conditions. Additional data may provide a clearer picture.

A related concern is that without Thr, sleep latency is not reduced in Rdl MMDR/1 mutants relative heterozygous controls (B), as would be expected based on previous reports. Similarly, without Thr, PDF and CRY neuron silencing does not lead to sleep latency reduction (C). These unexpected features of the data raise concerns about the validity of the rest of the data.

In our original manuscript, we examined sleep behaviors in male trans-heterozygous mutants of Rdl and could not observe their short sleep latency compared to heterozygous controls in control-fed condition. Since the original paper has shown the latency phenotype in female Rdl mutants (Agosto et al., 2008), we examined baseline sleep as well as SPET in female flies. Consistent with the previous finding, female Rdl MMDR/1 mutants displayed shorter sleep latency than their heterozygous controls in control-fed condition (Figure 2—figure supplement 1). We further observed comparable SPET between Rdl mutants and their heterozygous controls (P=0.1381 for sleep amount; P=0.2881 for sleep latency by two-way ANOVA). These data indicate that loss of Rdl function does not impair SPET.

Regarding the transgenic manipulations of circadian pacemaker neurons, we tested male transgenic flies in our original manuscript and could not detect any phenotype in their sleep latency. As the reviewer mentioned above, these data were not consistent with previous observations in female flies that have suggested the wake-promoting role of PDF-expressing neurons (Parisky et al., 2008; Liu et al., 2014). Accordingly, we examined SPET in female transgenic flies that have been validated for clock-dependent control of sleep latency in PDF neurons. Consistent with the published result (Liu et al., 2014), we found that overexpression of dominant-negative CLOCK proteins only in PDF neurons was sufficient to lengthen sleep latency in female flies fed control food (Figure 2B). The loss of molecular clocks in PDF neurons did not compromise SPET on the sleep amount as well as sleep latency. These new pieces of our genetic evidence more convincingly demonstrate that SPET requires neither Rdl- nor PDF clock-dependent control of sleep drive.

6) Rescue of memory defects in rut mutants by Thr is interesting. However, it would strengthen the conclusion if additional memory mutants were tested. Please see reviewer 1 's points on this,

In our revised manuscript, we further showed that dietary threonine rescued memory deficits in dumb mutants likely in a manner dependent on threonine-induced sleep (Figure 6B). Please see our response to the reviewer comment #1-5 above.

Minor Comments:

1) It is unclear how the effects of Thr vary as a function of time-of-day. It would be informative to show these data in a format where sleep amount per 30min or 1h is plotted as a function of time of day.

We included the sleep profiles of threonine-fed flies as well as their sleep latency at different time-points of the day in the revised manuscript (Figure 2C and D, Figure 2—figure supplement 2).

2) To further examine the clock- and light-dependence of SPET, it would be helpful to repeat the basic +/- Thr experiments in DD and LL.

As suggested by the reviewer, we examined SPET in LL or DD condition, and included these new data in the revised manuscript (Figure 2C and D, Figure 2—figure supplement 2).

3) As the authors note, GABA-T effects on SPET could be due to a floor effect. Although pharmacological data somewhat alleviate this concern, a genetic demonstration (using hypomorphs?) would strengthen the claim.

In our revised manuscript, we tested several mutants trans-heterozygous for GABA-T alleles (Figure 3—figure supplement 1A) and found that those harboring weaker allelic combinations did not display a floor effect on sleep latency, but their sleep behaviors were actually resistant to SPET (Figure 3A).

4) In Figure 1A and related figures, it is unclear whether p values were corrected for multiple comparisons.

All the P values from post hoc tests after one-way or two-way ANOVA were corrected for multiple comparisons. To clarify this, we included this statement in Materials and methods (Statistics section) of the revised manuscript.

5) Figure 1—figure supplement 3. In this and other similar figures, to determine whether SPET is affected by an experimental manipulation (in this case, silencing of gustatory or olfactory neurons), it is important to test whether the difference between + and – Thr is different across the three genotypes.

To determine whether or not our experimental manipulations significantly affected SPET, we compared SPET among all necessary heterozygous controls (e.g., Gal4/+, UAS/+) and testing genotypes (e.g., Gal4/UAS) by two-way ANOVA. To clarify this, we modified our text, figures, and figure legends in the revised manuscript to better describe our statistical comparisons among different genotypes and conditions.

6) Discussion section: "…SPET facilitates sleep onset in a manner independent of light". Where are the data for light-independence of SPET?

In our original manuscript, we generally measured the latency to sleep onset right after lights-off in LD cycles (ZT12) but we also observed SPET on the sleep latency after mechanical arousal at ZT16 (i.e., 4 hours after lights-off in LD cycles). These results indicate that SPET does not require the presence of light transition (Figure 1B). In our revised manuscript, we showed that SPET was also detectable in LL or DD (Figure 2C and D, Figure 2—figure supplement 2). Nonetheless, we thought that the wording, “light-independence”, might unnecessarily underestimate possible effects of light on the scaling of SPET. So, we modified our text in the revised manuscript.

7) Figure 2—figure supplement 1A and B. Some of the error bars are so big that it is difficult to believe the differences are significant at p<0.001 as indicated. In these situations, standard ANOVAs that assume equal variance across conditions are not appropriate.

Error bars in our original violin plots indicate SD although large variations in some genotypes (e.g., Clk[Jrk] or Pdf>CLK[DN]) are likely their phenotype. We revised all our violin plots to show 95% confidence intervals (corresponding to 1.98 fold of SEM).

8) In Figure 3A, the CaLexA signal for LNs is not symmetric. Is this common?

We quantified the CaLexA signals from each hemisphere since we often observed asymmetric ones in the whole-brain images. The CaLexA reporter involves a transcriptional amplification step. So, we reason that subtle differences in the levels and/or duration of Ca2+ influx between the same groups of neurons in each hemisphere may lead to asymmetric CaLexA signals at detectable levels.

9) Figure 4B is formatted differently from all other figures. Why?

We thought that the comparisons in the original Figure 4B would better visualize the effects of neuronal GABAB-R1 depletion on SPET among other genotypes. However, the reviewer #1 also mentioned that the original Figure 4B was rather hard to follow (please see the reviewer comment #1-4). Accordingly, we presented all our behavioral data in the same format in the revised manuscript.

10). Some of the N's for behavioral experiments are as low as 8. This seems too low.

We increased N in the relevant experiments.

11) SEM, which takes into account N's, would more informative than SD.

In our revised manuscript, we showed 95% confidence intervals in all the violin plots instead of SD.

Additional data files and statistical comments:

As noted in the minor point #7, some of the statistics do not seem appropriate.

Please see our response to the reviewer comment #1-7 above.

https://doi.org/10.7554/eLife.40593.032

Article and author information

Author details

  1. Yoonhee Ki

    School of Life Sciences, Ulsan National Institute of Science and Technology, Ulsan, Republic of Korea
    Contribution
    Software, Validation, Investigation, Visualization, Methodology, Writing—original draft
    Competing interests
    No competing interests declared
  2. Chunghun Lim

    School of Life Sciences, Ulsan National Institute of Science and Technology, Ulsan, Republic of Korea
    Contribution
    Conceptualization, Supervision, Funding acquisition, Validation, Investigation, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    clim@unist.ac.kr
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8473-9272

Funding

National Research Foundation of Korea (NRF-2017R1E1A2A02066965)

  • Chunghun Lim

National Research Foundation of Korea (NRF-2018R1A5A1024261)

  • Chunghun Lim

Suh Kyungbae Foundation (SUHF-17020101)

  • Chunghun Lim

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank JM Han at Yonsei University College of Pharmacy for conceiving relevant ideas at the initial stage of the research; JY Sonn at KAIST for critical reading of the manuscript and helpful comments; C Lee and RJ Mitchell at UNIST for reagents; EY Suh at Chungnam National University for amino acid analyses; SJ Kim and HJ Yoo at University of Ulsan College of Medicine for energy metabolite analyses; A Sehgal at University of Pennsylvania School of Medicine, Bloomington Drosophila stock center, Korea Drosophila resource center, Kyoto stock center, and Vienna Drosophila resource center for Drosophila strains.

Senior and Reviewing Editor

  1. K VijayRaghavan, National Centre for Biological Sciences, Tata Institute of Fundamental Research, India

Publication history

  1. Received: August 31, 2018
  2. Accepted: June 27, 2019
  3. Version of Record published: July 17, 2019 (version 1)

Copyright

© 2019, Ki and Lim

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,318
    Page views
  • 201
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Genetics and Genomics
    Téo Fournier et al.
    Research Article Updated
    1. Genetics and Genomics
    Joshua S Bloom et al.
    Research Article Updated