Melatonin promotes sleep by activating the BK channel in C. elegans

Significance

Melatonin (Mel) promotes sleep through G protein-coupled Mel receptors. However, the downstream molecular target(s) of the Mel receptors to produce the sleep effect remains enigmatic. The study shows that a potassium channel, the BK channel, plays a role in sleep and that Mel promotes sleep by activating this channel through a specific Mel receptor.

Abstract

Melatonin (Mel) promotes sleep through G protein-coupled receptors. However, the downstream molecular target(s) is unknown. We identified the Caenorhabditis elegans BK channel SLO-1 as a molecular target of the Mel receptor PCDR-1-. Knockout of pcdr-1, slo-1, or homt-1 (a gene required for Mel synthesis) causes substantially increased neurotransmitter release and shortened sleep duration, and these effects are nonadditive in double knockouts. Exogenous Mel inhibits neurotransmitter release and promotes sleep in wild-type (WT) but not pcdr-1 and slo-1 mutants. In a heterologous expression system, Mel activates the human BK channel (hSlo1) in a membrane-delimited manner in the presence of the Mel receptor MT₁ but not MT₂. A peptide acting to release free Gβγ also activates hSlo1 in a MT₁-dependent and membrane-delimited manner, whereas a Gβλ inhibitor abolishes the stimulating effect of Mel. Our results suggest that Mel promotes sleep by activating the BK channel through a specific Mel receptor and Gβλ.

Melatonin (Mel) is generally known as a hormone of darkness or a sleep hormone because it plays a pivotal role in sleep. In the human brain, Mel is secreted by the pineal gland with low levels during the day and high levels at night. This rhythmic secretion of Mel is under the control of neurons in the suprachiasmatic nucleus (SCN), which is the site of a circadian master clock, and receives synaptic inputs from photosensitive neurons in the retina (1–3). Many people take Mel supplements as a sleep aid.

Mel is believed to produce its sleep effect through Mel receptors. Two Mel receptors exist in mammals: MT₁ and MT₂ (4). Experiments with knockout mice indicate that MT₁ and MT₂ deficiencies compromise rapid eye movement (REM) sleep and non-REM sleep, respectively (5). MT₁ signaling is also important to circadian rhythmic expression of several clock genes (6, 7). However, it is unclear whether MT₁ and MT₂ play similar roles in humans, mainly due to a lack of sufficiently selective receptor agonists and antagonists (4, 8). Although it is well established that MT₁ and MT₂ function through G_i-type G proteins (9), the downstream molecular targets leading to its sleep-promoting effect remain mysterious. Protein interactome mining indicates that MT₁ but not MT₂ is an integral component of a presynaptic protein complex (10), but the physiological functions of presynaptic MT₁ are undetermined.

The BK channel is a large-conductance potassium channel gated by membrane voltage and cytosolic Ca²⁺. BK channels are ubiquitously expressed in the nervous system where they colocalize with voltage-gated Ca²⁺ channels at presynaptic sites of neurons and play a major role in downregulating neurotransmitter release (11–13). Results of previous studies indicate that the BK channel may play important roles in circadian behaviors. In mice, expression level of the BK channel in the SCN oscillates according to daily light and dark cycles (14), and this oscillation regulates the SCN neuron firing rate and circadian behavioral rhythms (15). In rats, the BK channel serves to inhibit spontaneous Ca²⁺ oscillations in pinealocytes (16). In flies, inhibition of the BK channel promotes wakefulness (17). However, it is unknown whether the BK channel plays a role in sleep.

C. elegans goes through four larval stages (L1–L4) before becoming an adult. A behavioral quiescence period, known as lethargus, exists between consecutive larval stages and between L4 and adult. Lethargus is considered a sleep state of worms because it bears major behavioral and molecular similarities to sleep states of other species (18–23). During the lethargus, behavioral quiescence is frequently interrupted by brief movements (19). Although C. elegans also produces Mel (24) and has one putative Mel receptor (F59D12.1/PCDR-1) (25), neither Mel nor the receptor has been implicated in the lethargus.

We have been using a forward genetics approach to identify proteins important to in vivo functions of SLO-1, the C. elegans BK channel (26–28). Here, we report that PCDR-1, one of the proteins identified from our genetic screen, is an indispensable molecule for SLO-1 physiological function in neurons, and that it allows Mel to promote sleep by activating SLO-1. Furthermore, we found that the human BK channel hSlo1 is activated by Mel through MT₁ but not MT₂, and this effect of MT₁ occurs through a membrane-delimited mechanism requiring Gβγ subunits and may be produced by Mel at concentrations (EC₅₀ < 1 nM) within the reported Mel concentration range in the human cerebrospinal fluid (29). Our results suggest that BK channels may play an evolutionarily conserved role in mediating the sleep effect of Mel.

Results

Neuronal SLO-1 Function Depends on PCDR-1.

In an unbiased genetic screen with C. elegans for mutants that ameliorate a sluggish phenotype caused by a hyperactive or gain-of-function (gf) SLO-1 (26), we isolated a mutant zw82, which was mapped to a gene (F59D12.1) encoding a G protein-coupled receptor (GPCR). F59D12.1 was initially predicted as a Mel receptor based on phylogenetic analyses (25) but was recently named PCDR-1 (pathogen clearance-defective receptor 1) because mutations of the gene are associated with a pathogen clearance defect phenotype (30). PCDR-1 shows significant sequence homology to human MT₁ and MT₂ receptors (SI Appendix, Fig. S1A). In zw82, cysteine 148, which is located in a putative transmembrane domain of PCDR-1, is mutated to tyrosine (SI Appendix, Fig. S1 A and B).

Neurotransmitter release is increased in the slo-1 loss-of-function (lf) mutant but decreased in the slo-1(gf) mutant (26, 31). To determine whether and how PCDR-1 regulates the SLO-1 function, we assessed the role of PCDR-1 in neurotransmitter release by analyzing evoked postsynaptic currents (ePSCs) and miniature postsynaptic currents (minis) at the neuromuscular junction. To avoid potential ambiguities associated with the missense mutation of zw82, we used two putative pcdr-1 null alleles, gk1122 and gk1000 (SI Appendix, Fig. S1B) from the Caenorhabditis Genetics Center. Compared with WT, both mutants showed larger ePSC amplitudes and higher mini frequencies without a change in the mean amplitude of minis (Fig. 1 A and B). These phenotypes are similar to those of slo-1(lf) and were not aggravated in the slo-1(lf);pcdr-1(lf) double mutant (Fig. 1 A and B). In the slo-1(gf) mutant, both the amplitude of the ePSCs and the frequency of the minis were greatly decreased compared with WT, and these phenotypes were eliminated in the slo-1(gf);pcdr-1(gk1122) double mutant (SI Appendix, Fig. S2). The mutant effects of pcdr-1(lf) on synaptic transmission did not result from a change in either expression or subcellular localization of SLO-1 (SI Appendix, Fig. S3). Taken together, these results suggest that neuronal SLO-1 function depends on PCDR-1.

Fig. 1.

PCDR-1 and SLO-1 act together to regulate neurotransmitter release. (A and B) Loss-of-function mutations of pcdr-1 and slo-1 similarly augmented the amplitude of ePSCs (A) and the frequency of minis (B) at the neuromuscular junction, and these mutant effects were nonadditive in the slo-1;pcdr-1 double mutant. (C) PCDR-1 expression and subcellular localization patterns. Top, pcdr-1 is expressed in many head neurons, ventral nerve cord (VNC) motor neurons, and vulval muscles (VMs) based on the expression of the GFP reporter under the control of the pcdr-1 promoter. Shown are corresponding dark field and differential interference contrast images of the entirety (Left), the head (Middle), and the midportion (Right) of a worm (Scale bar, 20 μm.) Bottom, dark field images of the VNC (indicated by an arrow) and dorsal nerve cord (DNC) (indicated by an arrow) of transgenic worms expressing a PCDR-1::GFP fusion protein under the control of the pan-neuronal rab-3 promoter (Prab-3) (scale bar, 10 μm.) Arrow heads indicate motor neuron cell bodies. (D and E) The synaptic phenotypes of the pcdr-1 mutant at the neuromuscular junction were completely rescued by presynaptic expression of WT PCDR-1 under the control of Prab-3. The numbers inside the bar graphs indicate sample sizes. The ** and *** symbols indicate statistically significant differences compared with WT at P < 0.01 and P < 0.001 levels, respectively (one-way ANOVA with Tukey’s post hoc test).

To determine the expression pattern of pcdr-1, we expressed a green fluorescent protein (GFP) reporter in worms under the control of the pcdr-1 promoter. In transgenic worms, the GFP signal was detected in many neurons, including ventral cord motor neurons and head neurons but not in body-wall muscles (Fig. 1C), suggesting that PCDR-1 likely functions presynaptically to regulate synaptic transmission. Consistently, the neuromuscular synaptic phenotypes of pcdr-1(lf) could be rescued by expressing WT PCDR-1 in neurons alone (Fig. 1 D and E).

PCDR-1 Is a Mel Receptor.

Although C. elegans produces Mel (24, 32) and PCDR-1 has been predicted as a Mel receptor (25), no Mel receptor has been experimentally confirmed in worms. We examined the possibility of PCDR-1 being a Mel receptor by testing the effects of Mel and Mel receptor antagonists on ePSCs and minis. In WT worms, Mel (100 nM) reduced both the amplitude of ePSCs and the frequency of minis without an effect on the mean amplitude of minis (Fig. 2 A and B). These effects of Mel on synaptic transmission are similar to those of SLO-1(gf) (SI Appendix, Fig. S2). In pcdr-1(lf) and slo-1(lf) mutants, however, Mel showed no effect on either ePSCs or minis (Fig. 2 A and B). These results suggest that PCDR-1 is a Mel receptor that inhibits neurotransmitter release through SLO-1.

Fig. 2.

PCDR-1 is activated by Mel but blocked by cis-4-phenyl-2-propionamidotetralin (4-P-PDOT). (A and B) Mel (100 nM) reduced the amplitude of ePSCs and the frequency of minis without altering the mean amplitude of minis in WT but had no effect in pcdr-1 and slo-1 mutants. (C and D) 4-P-PDOP (4-P, 100 nM) increased the amplitude of ePSCs and the frequency of minis without altering the mean amplitude of minis in WT but had no effect in the pcdr-1 mutant. The numbers inside the bar graphs indicate sample sizes. The asterisks indicate statistically significant differences (*P < 0.05; **P < 0.01; ***P < 0.001) whereas “ns” stands for “no significant difference” compared with the vehicle control group (unpaired t test).

To determine whether acute blockade of PCDR-1 alters synaptic transmission, we first tested the effect of the nonselective Mel receptor antagonist luzindole (3) because pretreatment with luzindole counteracts an inhibitory effect of exogenous Mel on worm locomotion (24). However, application of luzindole (1 μM) to the bath solution inhibited both the amplitude of ePSCs and the frequency of minis (SI Appendix, Fig. S4), which cannot be explained by a blocking effect on PCDR-1. We, then, tested the effect of the MT₂-selective antagonist 4-P-PDOT (3). In WT worms, 4-P-PDOT increased both the amplitude of ePSCs and the frequency of minis without altering the mean amplitude of minis (Fig. 2 C and D). These effects of 4-P-PDOT on synaptic transmission are similar to those of pcdr-1(lf) and slo-1(lf) mutants (Fig. 1 A and B). In contrast, 4-P-PDOT showed no effect on synaptic transmission in the pcdr-1(lf) mutant (Fig. 2 C and D), suggesting that it augments synaptic transmission by blocking PCDR-1. We, subsequently, tested the effect of the MT₂ receptor agonist IIK7 (3) (300 nM) in WT but did not detect any effect on either ePSCs or minis (SI Appendix, Fig. S4). Finally, we tested the Mel precursor serotonin (100 nM) and found that it did not alter either ePSCs or minis (SI Appendix, Fig. S5). Collectively, our results suggest that PCDR-1 is a Mel receptor with pharmacological properties distinct from mammalian MT₁ and MT₂.

Endogenous Mel Acts on PCDR-1.

Mel is synthesized from serotonin through sequential actions of serotonin N-acetyltransferase (SNAT) and hydroxyindole-O-methyltransferase (HIOMT) (33) (Fig. 3A). The worm HIOMT is encoded by homt-1 (24). To determine whether the function of PCDR-1 depends on endogenous Mel, we created a homt-1(lf) strain using the CRISPR/Cas9 approach and determined the effect of HOMT-1 deficiency on synaptic transmission. Compared with WT, homt-1(lf) worms showed a 55% increase in the amplitude of ePSCs and a 102% increase in the frequency of minis, and these synaptic phenotypes were nonadditive with those of pcdr-1(lf) (Fig. 3 B and C), suggesting that HOMT-1 and PCDR-1 act in the same pathway in regulating neurotransmitter release and that the function of PCDR-1 depends on endogenous Mel.

Fig. 3.

PCDR-1 synaptic function depends on endogenous Mel. (A) Mel synthesis pathway. (B and C) Loss-of-function mutation of homt-1 augmented the amplitude of ePSCs and the frequency of minis without altering the mean amplitude of minis at the neuromuscular junction. These effects were nonadditive with those of the pcdr-1 mutation and could be rescued completely by expressing WT HOMT-1 in either the PVT neuron using the zig-2 promoter or the intestine using the inx-16 promoter. The numbers inside the bar graphs indicate sample sizes. The ** and *** indicate statistically significant differences at P < 0.01 and P < 0.001 levels, respectively, compared with WT (one-way ANOVA with Tukey’s post hoc test). (D) homt-1 is expressed in the intestine, PVT neuron, DVB neuron, and the pharynx based on the expression of the GFP reporter under the control of the homt-1 promoter. (E) Diagram showing spatial relations of ventral cord motor neurons, the intestine, PVT neuron, and DVB neuron.

To determine the source of Mel for activating PCDR-1, we assessed the expression pattern of homt-1 by expressing the GFP reporter under the control of the homt-1 promoter. In transgenic worms, GFP expression was observed in the pharynx, intestine, and two neurons in the tail (DVB and PVT) (Fig. 3 D and E). While DVB is a γ-aminobutyric acid (GABA)ergic motor neuron that innervates enteric muscles (34), PVT is an interneuron that projects along the ventral nerve cord to the nerve ring (35) (https://www.wormatlas.org/). Because both the intestine and the neurite of PVT are close to ventral cord motor neurons (Fig. 3E), we determined whether intestine- or PVT-targeted expression of WT HOMT-1 can rescue homt-1(lf) with respect to its synaptic phenotypes. We found that expression of WT HOMT-1 in either the intestine or the PVT neuron obliterated the synaptic phenotypes of homt-1(lf) (Fig. 3 B and C), suggesting that Mel release from both the intestine and the PVT regulates neurotransmitter release from ventral cord motor neurons. These results are consistent with a hormonal effect of Mel because ventral cord motor neurons do not receive direct synaptic inputs from either the PVT or the intestine.

Mel Promotes Sleep through PCDR-1 and SLO-1.

We then determined whether the Mel/PCDR-1/SLO-1 pathway plays a role in sleep. We analyzed the effects of mutations of slo-1, pcdr-1, and homt-1 on the sleep behavior between L4 and adult using a custom-made sleep chamber (SI Appendix, Fig. S6). In WT worms, the total sleep duration, the motionless sleep duration, and the frequency of active events during sleep were 125 ± 4.83, 105 ± 4.65 min, and 30 ± 3.70/h, respectively (Fig. 4A). Compared with WT, mutants of slo-1(lf), pcdr-1(lf), and homt-1(lf) showed 20–43% decreases in both the total and the motionless sleep durations but an approximately twofold increase in the frequency of active events (Fig. 4A). These mutant phenotypes were not aggravated in pcdr-1(lf);slo-1(lf) and pcdr-1(lf);homt-1(lf) double mutants (Fig. 4A). The sleep phenotypes of pcdr-1(lf) could be rescued completely by expressing WT PCDR-1 specifically in neurons and those of homt-1(lf) by expressing WT HOMT-1 specifically in either PVT neuron or intestine (Fig. 4A). These results suggest that HOMT-1-dependent release of endogenous Mel can enhance sleep through PCDR-1 and SLO-1. We also tested whether SLO-2, a paralog of SLO-1 (36) with a regulatory role in neurotransmitter release (37), is important to sleep. We observed similar sleep behaviors between the WT and a putative slo-2 null mutant (38) (SI Appendix, Fig. S7), suggesting that SLO-2 does not play a role in sleep.

Fig. 4.

Mel promotes sleep through PCDR-1 and SLO-1. (A) Compared with WT, loss-of-function mutants of slo-1, pcdr-1, and homt-1 showed shorter durations of both total sleep and motionless sleep, and higher frequencies of active events during sleep. This phenotype of pcdr-1 could be rescued by expressing WT PCDR-1 in neurons under the control of the rab-3 promoter and that of homt-1 by expressing WT HOMT-1 in either the PVT neuron using the zig-2 promoter or the intestine using the inx-16 promoter. (B) Exogenous Mel (10 µM) prolonged the durations of both total sleep and motionless sleep and reduced the frequency of active events during sleep in WT and the homt-1 mutant but had no effect in slo-1 and pcdr-1 mutants. The numbers inside the bar graphs indicate sample sizes. The ** indicate statistically significant differences (P < 0.01) compared with the vehicle control group based on either one-way ANOVA with Tukey’s post hoc test (A) or unpaired t test (B).

Subsequently, we tested whether exogenous Mel may enhance worm sleep. Compared with the vehicle control groups, Mel (100 nM) enhanced sleep as indicated by increased durations of total sleep and motionless sleep in the WT and homt-1(lf) mutant (Fig. 4B). In contrast, Mel had no effect on the sleep behaviors when applied to slo-1(lf) and pcdr-1(lf) mutants (Fig. 4B). These results indicate that both PCDR-1 and SLO-1 are required for the sleep-enhancing effects of exogenous Mel.

Finally, we tested whether light and darkness conditions may have different effects on Mel’s action because the BK channel level in mammals varies with the day/night cycles (14). We subjected worms to either a darkness or an illuminated condition (from standard fluorescence ceiling lights) for 16–22 h prior to recording ePSCs and minis at the neuromuscular junction. The light intensities were 1.02 and 0.69 µW/mm² at 473-nm and 635-nm wavelengths, respectively, at the worm culture plate location. Both the amplitude of ePSCs and the frequency of minis were enhanced by the light exposure. Treatment with Mel (100 nM) inhibited ePSC amplitude and mini frequency in worms exposed to both the dark and light conditions. The inhibitory effect on ePSCs was greater in the darkness-exposed worms (ePSC amplitude reduction 66% versus 22%), although the difference did not reach statistical significance (P = 0.058) (SI Appendix, Fig. S8). These results suggest that SLO-1 expression might be down-regulated by light.

Mel Activates hSlo1 through MT₁.

We considered whether mammalian BK channels are also activated by Mel. We started by testing the effect of Mel on macroscopic currents in inside–out patches from Xenopus oocytes coexpressing human hSlo1 and mouse MT₁. In these experiments, Mel (100 nM) was included in the pipette solution, and macroscopic currents were induced by voltage steps from −80 mV to +180 mV at 20-mV intervals. The currents were converted to conductance (G) to quantify V₅₀ from a normalized G-voltage (V) relationship. Mel caused a significant decrease in the voltage for half-maximal channel activation (V₅₀) compared with the vehicle control in patches coexpressing hSlo1 and MT₁ but not in patches expressing hSlo1 alone (SI Appendix, Fig. S9 A and B). These results suggest that Mel can activate hSlo1 in a MT₁-dependent manner.

We, then, performed single-channel recordings on inside–out patches containing only one channel to determine whether Mel may activate hSlo1 through MT₁ and MT₂. Compared with the vehicle control, inclusion of Mel (100 nM) in the pipette solution caused a large leftward shift in the open probability (P_o)-V relationship in patches coexpressing hSlo1 and mouse MT₁ (V₅₀ 22.25 ± 3.28 versus −2.88 ± 3.57 mV) (Fig. 5A) but not in patches expressing either hSlo1 alone or hSlo1 and human MT₂ (SI Appendix, Fig. S10 A and B). Mel did not alter hSlo1 single-channel conductance in patches coexpressing hSlo1 and MT₁ as indicated by similar single-channel current (I)-V relationships between the control and the Mel groups (Fig. 5A). These results indicate that Mel can activate hSlo1 through MT₁ but not MT₂ and that the increased hSlo1 activity results from an increased P_o. To determine how Mel increases P_o in the presence of MT₁, we analyzed dwell times of open and closed events at +10-mV holding voltage. We found that both the open and the closed dwell times could be fit by two exponential terms (τ1 and τ2) and that Mel augmented hSlo1 P_o mainly by prolonging the τ2 of open events and reducing the τ2 of closed events (Fig. 5B).

Fig. 5.

Mel augments hSlo1 single-channel open probability (P_o) in isolated membrane patches coexpressing human Slo1 (hSlo1) and mouse MT₁. (A) Mel (100 nM) increased hSlo1 P_o without altering single-channel current amplitudes in inside–out patches containing only one channel (symmetrical K⁺ solutions). Shown are sample current traces, the open probability (P_o) voltage (V) relationships, the voltage for half-maximal channel activation (V₅₀), and the current (I)-V relationships. Single-channel conductance (quantified from the I-V relationships) was 237.8 ± 5.3 pS in control and 242.1 ± 2.3 in Mel. (B) Mel prolonged the τ2 of open events and reduced the τ2 of closed events as determined by fitting open and closed times of hSlo1 at the +10-mV step in A. (C) Mel increased hSlo1 Ca²⁺ sensitivity. Increasing concentrations of Ca²⁺ were applied to the cytosolic side of inside–out patches (containing either one or two hSlo1 channels) held constantly at +20 mV. In A–C, the asterisks indicate statistically significant differences compared with the vehicle control group (*P < 0.05; **P < 0.01; ***P < 0.001, unpaired t test). (D) Mel increased hSlo1 activity in a concentration-dependent manner in outside–out patches containing multiple channels (asymmetrical K⁺ solutions). Increasing concentrations of Mel (two applications at each concentration) were applied to the extracellular side of the patch held at −80 mV (equal to the K⁺ equilibrium potential) through perfusion. hSlo1 currents were induced by stepping to +20 mV (50 ms). Shown are a sample current trace (with one Mel-induced current response displayed at an expanded time scale) and the Mel concentration—response curve fitted to the Hill’s equation E = E_max/(1 + (EC₅₀/A))^nH, where E is the observed response and nH is the Hill coefficient reflecting the slope of the curve.

We next tested the effect of Mel on hSlo1 P_o at different Ca²⁺ concentrations. Increasing concentrations of Ca²⁺ were applied to the cytosolic surface of inside–out patches coexpressing hSlo1 and MT₁ at a constant holding voltage. Compared with the vehicle control, Mel (100 nM) caused significant increases in hSlo1 P_o at all of the tested Ca²⁺ concentrations that were ≥3 µM (Fig. 5C), which indicates that MT₁ activation increases hSlo1 apparent Ca²⁺ sensitivity.

The concentration of Mel used in the experiments described above (100 nM) was chosen based on what has been commonly used in in vitro experiments. Because physiological concentrations of Mel might be much lower than 100 nM, we tested the effects of different concentrations of Mel on macroscopic currents in outside–out patches coexpressing hSlo1 and MT₁. Asymmetrical K⁺ solutions were used in these experiments with the patch held constant at −80 mV (equal to the K⁺ equilibrium potential). Perfusion of Mel to the extracellular side of the patch membrane at increasing concentrations (10 pM to 10 µM in logarithmic increments) caused concentration-dependent increases in outward currents. Fitting the concentration-response data to Hill’s equation gave rise to a curve with an EC₅₀ of 0.77 ± 0.09 nM (Fig. 5D). These results indicate that Mel can produce its stimulatory effect on hSlo1 at concentrations much lower than 100 nM.

MT₁ Activates hSlo1 through Gβγ.

To determine whether and how a G protein is involved in the activation of hSlo1 by MT₁, we tested the effect of mSIRK, a peptide that dissociates the trimeric G proteins into Gα and Gβγ subunits without inducing guanosine diphosphate/guanosine triphosphate (GDP/GTP) exchange in the Gα subunit (39). In inside–out patches coexpressing hSlo1 and MT₁, application of mSIRK to the bath solution augmented hSlo P_o without an effect on single-channel conductance (Fig. 6A). The effect of mSIRK on P_o was nonadditive with that of Mel (100 nM in the pipette solution) and mainly due to an increased τ2 of open events and a decreased τ2 of closed events (Fig. 6 A and B). This effect is similar to that of Mel, which also increased P_o by increasing τ2 of open events and decreasing τ2 of closed events. These results suggest that mSIRK and Mel activate hSlo1 through a common mechanism, presumably through the release of free Gβγ subunits.

Fig. 6.

Mel activates hSlo1 through Gβγ subunits. (A) mSIRK increased hSlo1 P_o, and this effect was nonadditive with that of Mel. Mel (100 nM) was included in the pipette solution whereas mSIRK (10 µM) was added to the bath solution. (B) mSIRK prolonged the τ2 of open events and reduced the τ2 of closed events as determined from analyzing the data at +10 mV in A. The asterisks indicate statistically significant differences (*P < 0.05; **P < 0.01; ***P < 0.001) compared with the vehicle control group whereas the pound sign (#) indicates a statistically significant difference between the indicated groups (one-way ANOVA with Tukey’s post hoc test). (C) Gallein blocked the stimulatory effect of Mel on hSlo1. Mel (100 nM) was included in the pipette solution whereas gallein (10 µM) was applied to the bath solution. The control had dimethyl sulfoxide (vehicle) in both the pipette and the bath solutions. *** indicate a statistically significant difference (P < 0.001, one-way ANOVA with Tukey’s post hoc test). All recordings were from inside–out patches coexpressing hSlo1 and MT₁ with only one channel.

In contrast, mSIRK had no effect on the P_o in patches expressing hSlo1 alone (SI Appendix, Fig. S11). This result, in combination with the result described above, suggests that MT₁ might localize G proteins to the vicinity of hSlo1 to allow G protein-mediated channel activation. To obtain further evidence in support of the role of Gβγ subunits, we determined whether inclusion of gallein, a Gβγ inhibitor (40, 41), in the bath solution can prevent the stimulatory effect of Mel on channel activities in inside–out patches coexpressing MT₁ and hSlo1. Indeed, gallein obliterated the stimulatory effect of Mel on hSlo1 (Fig. 6C). Collectively, our results indicate that Mel activates hSlo1 through MT₁-dependent release of free Gβγ subunits.

Discussion

The present study shows that the BK channel is a molecular target through which Mel promotes sleep and that Mel activates the channel through PCDR-1 in worms and MT₁ in mammals. Our results support a model in which a Mel receptor localizes a specific type of G protein to the BK channel, and binding of Mel to the receptor causes the release of free Gβγ subunits to activate the channel (Fig. 7). Our results also suggest that the BK channel is a key molecular target for Mel to produce its sleep-promoting effect in worms and that it might play a similar role in mammals. These conclusions are supported by the similar and nonadditive effects of slo-1 and pcdr-1 mutations on neurotransmitter release and sleep behavior in C. elegans and by MT₁-dependent activation of hSlo1 in a membrane-delimited manner in the Xenopus oocyte heterologous expression system. The putative roles of MT₁ in neurotransmitter release and sleep are also in agreement with the earlier reports that MT₁ is a molecular component of presynaptic protein complexes (10) and that both sleep and circadian rhythmic expression of clock proteins depend on MT₁ (5–7) in mice.

Fig. 7.

Model of BK channel activation by Mel. Binding of Mel to MT₁ causes the exchange of GDP by GTP in the α subunit of a Gi-type G protein, the dissociation of the trimeric G protein into G_iα-GTP and Gβγ subunits, and the binding of free Gβγ subunits to the BK channel to cause channel activation.

Sleep in C. elegans is a global state with the majority of neurons showing reduced activities during sleep (42). However, some neurons are active during sleep, including sleep-promoting neurons (42). Among the sleep-promoting neurons, RIS plays a pivotal role in sleep onset by releasing FLP-11, a neuropeptide, and GABA (43, 44), and its activity is regulated by interneurons, such as the bilateral pairs of PVC and RIM interneurons (45). While PVC activates RIS, RIM may either activate or inhibit RIS depending on RIM’s activity level (45). How might SLO-1 contribute to sleep? One possibility is that Mel may activate SLO-1 in many neurons to produce a global inhibitory effect. Another possibility is that SLO-1 activation in inhibitory neurons presynaptic to RIS may promote sleep through a disinhibitory effect. A third possibility is that SLO-1 functions in wake-promoting neurons, such as those in the RMG circuit (46), to reduce their antagonistic effects on sleep. How exactly SLO-1 functions in sleep can potentially be addressed by analyzing the effects of cell-targeted knockdown of slo-1 or pcdr-1 on RIS activity and sleep behavior.

In flies, octopamine promotes wakefulness by inhibiting the BK channel (17). Octopamine is also a signaling molecule in worms (47). The contrasting effects of octopamine in flies and Mel in worms suggest that the BK channel might help to switch between sleep and wake states depending on the levels of octopamine and Mel. The putative bidirectional regulation of sleep behavior through the BK channel might also exist in mammals, although the signaling molecule is perhaps norepinephrine instead of octopamine because norepinephrine, which is a vertebrate equivalent of octopamine (48), has a wake-promoting effect in vertebrates (49, 50).

G proteins are trimeric complexes of Gα, Gβ, and Gγ subunits. Before the activation of a GPCR, the three subunits are closely associated with GDP bound to the Gα subunit. Upon activation of the GPCR, a substitution of GDP by GTP in the Gα subunit causes dissociation of the G protein into Gα-GTP and free Gβγ subunits, which can produce their biological effects independently. A variety of ion channels are regulated either directly or indirectly by Gβγ, including G protein-gated inward rectifier K⁺ (GIRK) channels (51), TRP channels (52, 53), piezo channels (54), and voltage-gated Ca²⁺ channels (55). Mechanistically, the regulation of GIRK channels by Gβγ is best understood. Gβγ binds to GIRK channels to activate them through a membrane-delimited mechanism (51, 56–58). In the present study, we show that induction of free Gβγ by mSIRK augments hSlo1 activity whereas inhibition of Gβγ by gallein inhibits hSlo1 activity. The result with gallein is in agreement with that of a very recent study showing that gallein can eliminate a stimulating effect of propionate, a short-chain fatty acid, on the BK channel in arterial smooth muscle cells (59). The opposite effects of gallein and mSIRK on hSlo1 indicate that Gβγ subunits likely mediate the activating effect of Mel.

This study offers insights to some reported effects of exogenous Mel on C. elegans. For example, one study showed that exogenous Mel has a strong inhibitory effect on worm locomotion (24). This effect of Mel resembles those of ethanol intoxication (60) and slo-1 gain-of-function mutation (26) but its molecular mechanism is unknown. The results of this study suggest that, like the intoxicating effect of ethanol (60), SLO-1 activation in neurons is likely a major mechanism for the inhibitory effect of Mel on locomotion. Another study showed that treatment of dystrophin (dys-1) mutant worms with exogenous Mel improves muscular strength, thrashing rate, and mitochondrial integrity (61). Although Mel has also been used to treat muscular dystrophy patients (62), the molecular mechanism for its beneficial muscle effects is unknown. Given that dys-1 deficiencies enhance cholinergic synaptic transmission (63) and that the dys-1-associated protein complex plays a pivotal role in neuronal SLO-1 function by localizing SLO-1 to presynaptic sites (28, 64–66), activation of neuronal SLO-1 might be partially responsible for the beneficial effects of Mel in dys-1-deficient worms.

To summarize, the present study shows that the BK channel is a key molecular target that Mel may act on to produce its sleep-promoting effect and that hSlo1 can also be activated by Mel in a MT₁-dependent manner. These findings suggest that the BK channel might play an evolutionarily conserved role in Mel’s sleep-promoting effect. Investigations into this possibility in mammals will likely lead to major advances in our understanding about sleep biology.

Materials and Methods

C. elegans Culture and Strains.

C. elegans were raised on nematode growth medium plates with a layer of OP50 E. coli at 22 °C inside an environmental chamber. The strains used are described in the SI Appendix.

Generation of the homt-1 Mutant.

homt-1 knockout worms were generated using the CRISPR/Cas9 approach. Details are described in the SI Appendix.

Analysis of Expression Pattern and Subcellular Localization.

The expression patterns of pcdr-1 and homt-1 were assessed by expressing GFP under the control of their promoters. Subcellular localization of PCDR-1 in the neuron was determined by fusing GFP to its carboxyl terminus and expressing the fusion protein under the control of Prab-3. Details are provided in the SI Appendix.

C. elegans Electrophysiology.

All electrophysiological experiments were performed with adult hermaphrodites maintained in a low-temperature incubator at 22 °C. ePSCs and minis at the C. elegans neuromuscular junction were recorded as described previously (31, 67). Further information may be found in the SI Appendix.

Xenopus Oocyte Electrophysiology.

Xenopus oocytes were used to examine the effects of Mel receptors on hSlo1 function. Further information may be found in the SI Appendix.

Sleep Behavior Analysis.

Mid-L4 stage worms were individually placed inside the openings of a polydimethylsiloxane (PDMS) membrane and imaged at 10-s intervals for 10 h. Image acquisition and subsequent analyses were performed using a custom-written program running in Matlab, which may be accessed at GitHub: https://github.com/WanglabNiu/PNAS_sleep-tracking-software. Further information may be found in the SI Appendix.

Statistical Analyses.

Amplitudes of currents, frequencies of minis, and open probability of single-channel events were quantified using the ClampFit software. Dwell time analyses were performed using OriginPro (version 2020, OriginLab, Northampton, MA, USA). Data graphing and statistical analyses were performed with OriginPro. All data are shown as mean ± SE. The sample size (n) equals the number of worms or membrane patches recorded. Further information may be found in the SI Appendix.

Data Availability.

All study data are included in the article and SI Appendix.