Abstract
To identify molecular mechanisms that function in G-protein signaling, we have performed molecular genetic studies of a simple behavior of the nematode Caenorhabditis elegans, egg laying, which is driven by a pair of serotonergic neurons, the hermaphrodite-specific neurons (HSNs). The activity of the HSNs is regulated by the Go-coupled receptor EGL-6, which mediates inhibition of the HSNs by neuropeptides. We report here that this inhibition requires one of three inwardly rectifying K+ channels encoded by the C. elegans genome: IRK-1. Using ChannelRhodopsin-2-mediated stimulation of HSNs, we observed roles for egl-6 and irk-1 in regulating the excitability of HSNs. Although irk-1 is required for inhibition of HSNs by EGL-6 signaling, we found that other Go signaling pathways that inhibit HSNs involve irk-1 little or not at all. These findings suggest that the neuropeptide receptor EGL-6 regulates the potassium channel IRK-1 via a dedicated pool of Go not involved in other Go-mediated signaling. We conclude that G-protein-coupled receptors that signal through the same G-protein in the same cell might activate distinct effectors and that specific coupling of a G-protein-coupled receptor to its effectors can be determined by factors other than its associated G-proteins.
Introduction
G-protein-coupled receptors (GPCRs) constitute a large family of cell surface receptors for diverse small molecules and peptides. In neurons, GPCRs act as receptors for most neurotransmitters and neuromodulators. Accordingly, GPCRs have critical functions in brain circuits, and they are targets of therapeutics used for the treatment of numerous psychiatric disorders, including schizophrenia, depression, anxiety disorders, and addiction (Iversen et al., 2009). GPCRs exert their effects on cell physiology by activating associated heterotrimeric G-proteins, which in turn regulate effectors such as protein kinases, ion channels, and enzymes that generate second-messenger signals. Although G-protein signaling has been linked to many such effectors, how GPCR signaling alters cell and circuit function in vivo to change behavior is still poorly understood.
The egg-laying system of the nematode Caenorhabditis elegans is a powerful model for the study of G-protein signaling in the control of circuit function and behavior. Egg laying is driven by two motor neurons, the serotonergic hermaphrodite-specific neurons (HSNs), which innervate vulval muscles (White et al., 1986). Both the HSNs and vulval muscles are regulated by G-protein signaling. Vulval muscles express two Gq-coupled serotonin receptors that promote muscle contraction: SER-1 and SER-7 (Carnell et al., 2005; Hobson et al., 2006). Egg-laying behavior is inhibited by Go signaling in the HSNs (Mendel et al., 1995; Ségalat et al., 1995; Tanis et al., 2008). HSNs express several receptors that inhibit egg laying in a Go-dependent manner: the muscarinic acetylcholine receptor GAR-2 (Lee et al., 2000; Bany et al., 2003), the orphan receptor EGL-47 (Moresco and Koelle, 2004), and the neuropeptide receptor EGL-6 (Ringstad and Horvitz, 2008). These Go-coupled receptors might coordinate egg laying with the activity of other neuronal circuits. For example, sources of RFamide neuropeptide ligands for EGL-6 include the BAG sensory neurons (Kim and Li, 2004; Ringstad and Horvitz, 2008), which are activated by hypoxia (Zimmer et al., 2009) and carbon dioxide (Hallem et al., 2011). Go signaling in the HSNs likely integrates inhibitory signals from multiple circuits outside of the reproductive neuromusculature, including sensory systems, to modulate egg-laying behavior.
Although Go signaling has long been known to mediate inhibition of the HSNs, the mechanism by which this inhibition occurs is unclear. At cholinergic neuromuscular synapses, Go signaling reduces levels of diacylglycerol, which potentiates synaptic vesicle fusion via the fusion regulator UNC-13 (Nurrish et al., 1999) and also promotes dense-core vesicle fusion by activating protein kinase C (Sieburth et al., 2007). In other contexts, Go can regulate neuronal function through other mechanisms. Vertebrate G-protein βγ subunits once liberated from Goα directly inhibits voltage-sensitive calcium channels (Herlitze et al., 1996) and activates potassium channels (Reuveny et al., 1994), reducing both neuronal excitability and presynaptic function. To better understand how Go signaling modulates neuronal function in vivo, we sought to identify genes required for inhibition of HSNs by the Go-coupled neuropeptide receptor EGL-6. We report here a role for nematode Kcnj family potassium channels in the modulation of HSNs by inhibitory Go signaling.
Materials and Methods
Strains.
C. elegans hermaphrodites were grown on nematode growth medium (NGM) agar at 20°C on Escherichia coli OP50 as described previously (Brenner, 1974). Wild-type C. elegans were the Bristol N2 strain. Other strains used in this study were as follows: MT15934 irk-1(n4895), FQ295 irk-2(n4896), MT17360 irk-3(n5049), MT1222 egl-6(n592), MT16668 egl-6(n592) irk-1(n4895), MT16693 egl-6(n592) irk-2(4896), FQ308 egl-6(n592sd) irk-3(n5049), FQ36 Promceh-24::dsRed; Promirk-1::cameleon, FQ420 wzIs124[Promegl-6a::gfp::irk-1], FQ351 irk-1(n4895); wzEx46 [Promegl-6a::irk-1 Promttx-3::dsRed], FQ352 irk-1(n4895); wzEx47 [Promegl-6a::irk-1 Promttx-3::dsRed], FQ353 irk-1(n4895); wzEx48 [Promegl-6a::irk-1 Promttx-3::dsRed], FQ370 egl-6(n592); wzEx52 [Promegl-6a::irk-1RNAi sense Promegl-6a::irk-1RNAi antisense Promttx-3::dsRed], FQ372 egl-6(n592); wzEx54 [Promegl-6a::irk-1RNAi sense Promegl-6a::irk-1RNAi antisense Promttx-3::dsRed], FQ373 egl-6(n592sd); wzEx55[Promegl-6a::irk-1RNAi sense Promegl-6a::irk-1RNAi antisense Promttx-3::dsRed], FQ422 egl-6(n592); wzEx[Promceh-24::irk-1RNAi sense Promceh-24::irk-1RNAi antisense Promttx-3::dsRed], FQ342 wzEx41[Promegl-6a::irk-1 Promttx-3::dsRed], FQ343 egl-6(n4536); wzEx41 [Promegl-6a::irk-1 Promttx-3::dsRed], FQ84 wzIs30 [Promegl-6a::ChR2::yfp lin-15(+)]; lin-15AB(n765), FQ287 egl-1(n487); wzIs30 [Promegl-6a::ChR2::yfp lin-15(+)]; lin-15AB(n765), FQ200 irk-1(n4895); wzIs30 [Promegl-6a::ChR2::yfp lin-15(+)]; lin-15AB(n765), FQ201 irk-3(n5049) irk-1(n4895) irk-2(4896); wzIs30[Promegl-6a::ChR2::yfp lin-15(+)]; lin-15AB(n765), FQ293 egl-10(n692); wzIs30[Promegl-6a::ChR2::yfp lin-15(+)]; lin-15AB(n765), FQ291 egl-6(n592); wzIs30[Promegl-6a::ChR2::yfp lin-15(+)]; lin-15AB(n765), FQ310 egl-6(n592) irk-1(n4895); wzIs30 [Promegl-6a::ChR2::yfp lin-15(+)]; lin-15AB(n765), MT16698 goa-1(n1134); egl-6(n592), FQ309 goa-1(n1134); egl-6(n592) irk-1(n4895), FQ281 egl-10(n692);) irk-1(n4895), FQ280 egl-47(n1081); irk-1(n4895).
irk-1, irk-2, and irk-3 deletion alleles and egl-6(n592), egl-10(n692), egl-47(n1081), and goa-1(n1134) mutations were followed and verified using PCR; primers that specifically amplified sequences containing point mutations were designed as previously described (Drenkard et al., 2000).
Transgenic animals were created according to Mello et al. (1991). The Promttx-3::dsRed coinjection marker was used at 40 ng μl−1. A lin-15 rescuing plasmid pL15EK was used at 30 ng μl−1. Promirk-1::cameleon was injected at 100 ng μl−1. Promegl-6a::irk-1 and Promegl-6a::gfp::irk-1 were injected at 30 ng μl−1. irk-1 RNAi constructs were injected at 50 ng μl−1 each. Promegl-6a::ChR2::yfp was injected at 100 ng μl−1. In some cases, extrachromosomal transgenes were integrated using gamma radiation (5000 rad).
Isolation of deletion alleles.
Libraries of mutagenized animals were constructed and screened by PCR for deletion alleles of irk-1, irk-2, and irk-3, essentially as described previously (Jansen et al., 1997). Deletion mutants were isolated from frozen stocks and backcrossed to wild-type worms at least four times.
Molecular biology.
PCR-amplified promoter sequences and cDNAs were initially cloned into the pGEM-T Easy vector (Promega) and sequenced. Promoter sequences, cDNA sequences, and coding sequences used for expression of dsRNA targeting irk-1 were subcloned into the modular C. elegans expression vector pPD49.26 (Mello and Fire, 1995) using standard methods. cRNA for expression in Xenopus oocytes was generated from cDNAs clones into the vector pGEMHE (Liman et al., 1992). irk-1 cDNA was amplified by PCR from cDNA created using the SuperScript III First-Strand Synthesis kit (Invitrogen). Primers used to amplify irk-1 cDNA were as follows: 5′-ATGACGCTTTCAGTCCCGGATTGC-3′ and 5′-CTACGTCTCGCTGACGATTTCAATG-3′. Primers used to amplify a 6.5 kb irk-1 promoter were as follows: 5′-CAACATAAACTTTGGTTTCGAATGTGG-3′ and 5′-CATTGATCTGAAAAATGTATG-3′. Primers used to amplify a 3.5 kb egl-6a promoter were as follows: 5′-GCAAAAATCTAGAGGTCTG-3′ and 5′-TGTGTTCTATGATGTTCTCCATC-3′.
Microscopy.
Young adults were anesthetized with 30 mm sodium azide and mounted on 2% agarose pads made in M9 medium. Z stacks were obtained with a Zeiss LSM510 confocal microscope, and maximum-projection images were created using ImageJ (W. S. Rasband, ImageJ, National Institutes of Health, Bethesda, MD; http://imagej.nih.gov/ij/; 1997–2011). HSNs were identified on the basis of cell body position and neurite morphology. Other irk-1-expressing cells were identified as neurons on the basis of their nuclear morphology observed using differential interference contrast (DIC) optics.
Xenopus oocyte electrophysiology.
Capped irk-1 cRNAs were prepared using the mMessage Machine kit (Ambion). X. laevis oocytes were injected with 50 ng μl−1 irk-1 sense cRNA. Injected oocytes were incubated at room temperature in ND96 medium (96 mm NaCl, 2.5 mm KCl, 1 mm MgCl2, and 5 mm HEPES, pH 7.6) for 2–5 d before recording. Whole-cell current recordings were made using a two-electrode voltage-clamp (Warner Instruments). Oocytes were continuously superfused with ND96 during measurements of the resting membrane potential. To assay activity of IRK channels, we equilibrated oocytes in high-K+ medium (96 mm KCl and 2.5 mm NaCl instead of 2.5 mm KCl and 96 mm NaCl) to permit measurement of inward currents. Data were acquired with a Digidata 1440 digitizer (Molecular Devices) and analyzed off-line with Clampfit (Molecular Devices). All experiments using Xenopus laevis were performed according to guidelines of the Committee on Animal Care at New York University Medical Center.
Behavioral assays.
All behavioral assays were performed on C. elegans hermaphrodites. For assays of egg-laying behavior, five young nongravid hermaphrodites were transferred to a fresh plate seeded with OP50 bacteria and allowed to lay eggs for 1 h. These experiments were performed with at least 20 animals. Newly laid eggs were transferred to a 2% agarose pad in M9 buffer and observed at 100× (late stage) or 1000× (early stage) magnification using a Zeiss Axioskop equipped with Nomarski/DIC optics and scored as being in one of seven categories: 1–8 cell embryos, 9–25 cell embryos, embryos with >26 cells that had not undergone overt morphogenesis, comma stage embryos, twofold embryos, threefold embryos, and embryos that hatched during the observation period and appeared as L1 larvae.
To measure locomotory behavior of C. elegans strains, 30 young adult hermaphrodites were transferred to a fresh NGM agar plate coated with a thin lawn of E. coli OP50 and allowed to recover for 1 h. Animals were visualized with a custom-built LED ring light. Video was captured at 7.5 frames per second using a Unibrain Fire-i785 camera and the MATLAB Image Acquisition Toolbox (MathWorks). Individual worm trajectories were computed using the Mosaic plugin for ImageJ (Sbalzarini and Koumoutsakos, 2005). Centroid speeds of worms that were tracked for at least 140 consecutive frames were calculated using MATLAB (MathWorks).
For structured illumination of immobilized animals expressing ChannelRhodopsin-2 (ChR2), animals were placed on pads composed of 2% agarose in NGM buffer and immobilized using 0.1-μm-diameter polystyrene beads as described previously (Fang-Yen et al., 2012). To create structured illumination, we used a 473 nm laser coupled through a Leica inverted microscope via a digital micromirror device (Texas Instruments) in an optical setup similar to one previously described (Leifer et al., 2011). Worms were illuminated at an irradiance of 3 mW/mm2 in one of two spatiotemporal patterns: (1) illumination of a 60 μm-square region centered around the vulva for 10 s followed by illumination of the entire field of view for 10 s, or (2) illumination of the entire field of view except for the 60 μm-square region around the vulva for 10 s, followed by 10 s illumination of the entire field of view. A worm was scored as responsive if it laid one or more eggs during the first period of illumination.
For photostimulation of unrestrained animals, animals expressing ChR2 were placed on OP50 plates either with or without all-trans-retinal (ATR), prepared as previously described (Guo et al., 2009). Animals were transferred from these plates to conventional OP50 plates the next day and immediately assayed for egg laying in response to blue light. Assays were conducted using a Leica M165FC fluorescence stereo microscope equipped with a GFP filter set. Fluorescence illumination was provided by a Leica EL6000 light source with an internal shutter controlled by a programmable timer (Koyo KT-V). Irradiance was altered using filters and by varying the diameter of the light beam using the zoom optics of the microscope. Irradiance was calculated using power measurements and beam diameter at the surface of the cultivation plate.
Statistical methods.
Values of p from pairwise comparisons of egg stage data from different strains were calculated using the Wilcoxon Mann–Whitney rank sum test as implemented in the Coin package of the R statistical analysis program (R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org). These p values are presented in Table 1. Other statistical analyses were performed using GraphPad Prism software.
Results
One of three inward rectifier potassium channels encoded by the C. elegans genome is required for inhibition mediated by the neuropeptide receptor EGL-6
n592 is a gain-of-function (gf) mutation that increases function of the neuropeptide receptor EGL-6, which inhibits the HSN neurons of the egg-laying system. Importantly, the n592 mutant receptor is still regulated by endogenous RFamide peptide ligands and requires Go-proteins (Ringstad and Horvitz, 2008). We sought to identify genes that function in Go signaling by identifying genetic suppressors of the egg-laying defects of egl-6(gf) mutants. In the mammalian nervous system, inhibitory GPCRs regulate potassium channels of the Kcnj family, which are inwardly rectified tetrameric potassium channels comprising subunits with two transmembrane domains flanking a pore-forming loop (Luján et al., 2009). The C. elegans genome is predicted to encode three such channel subunits (Bargmann, 1998): IRK-1, IRK-2, and IRK-3 (Fig. 1A). To determine whether any of these channels is required for inhibitory EGL-6 signaling, we isolated deletion alleles of each irk gene (Fig. 1B) and tested whether loss of IRK channels suppressed the egg-laying defect caused by egl-6(gf).
We measured the egg-laying behavior of the wild-type, egl-6(gf), and egl-6(gf) irk double mutants by scoring the developmental stage of newly laid eggs, which reflects the time embryos spent in utero before being released. As expected (Ringstad and Horvitz, 2008), embryos laid by egl-6(gf) mutants were abnormally late stage, having developed to the threefold stage and in some cases having already hatched by the time of observation (Fig. 1C). Deletion of irk-2 did not suppress the egl-6(gf) phenotype, and deletion of irk-3 only slightly modified the egl-6(gf) phenotype (Fig. 1C). irk-1 deletion, however, strongly suppressed the egl-6(gf) phenotype: egl-6(gf) irk-1(Δ) mutants laid most embryos at a normal developmental stage (Fig. 1C), and egl-6(gf) irk-1(Δ) mutants were not bloated with unlaid eggs as young adults (Fig. 1D).
In contrast to its strong action as a suppressor of the egl-6(gf) phenotype, we found that the irk-1 deletion in a wild-type background conferred only a very mild defect in egg-laying behavior (Fig. 1C), suggesting that the constitutive role IRK-1 plays in regulating egg laying under normal culture conditions is minor. These results are consistent with the observation that vertebrate Kcnj channels expressed by neurons demonstrate low constitutive activity in the absence of activation of neurotransmitter signaling pathways (Takigawa and Alzheimer, 2002). Neither irk-2(Δ) nor irk-3(Δ) animals had overt defects in egg-laying behavior (Fig. 1C). These data suggest that, of the three Kcnj-like subunits encoded by the C. elegans genome, IRK-1 plays a role in modulating the function of the egg-laying system of the adult hermaphrodite, while IRK-2 and IRK-3 subunits either have minor roles in the egg-laying system or function in other tissues that weakly modulate the egg-laying system.
IRK-1 channels are required in HSN neurons for inhibition by EGL-6 signaling
Where does irk-1 function to regulate egg laying by C. elegans? We determined the cellular expression pattern of irk-1 using a reporter transgene that drives yfp expression with the irk-1 promoter. We observed reporter transgene expression in neurons in the head and tail (Fig. 2A,B); we also observed expression of irk-1 in the egg-laying system (Fig. 2C). We observed strong expression of Promirk-1::yfp in the HSN motor neurons of the egg-laying system (Fig. 2C), and faint expression of this reporter in vulval muscles, which are postsynaptic to the HSNs.
To determine the subcellular localization of IRK-1 in the HSNs, we fused GFP to the IRK-1 N terminus and used the egl-6a promoter to drive expression of the fusion protein in HSNs. The HSNs are monopolar neurons and form synapses with target muscles and other neurons in the egg-laying system at stereotyped positions along the neurite (Fig. 2D). We observed GFP::IRK-1 on both the HSN soma and neurite (Fig. 2D). GFP::IRK-1 was not appreciably enriched at HSN presynaptic specializations. An EGL-6::GFP translational reporter is similarly expressed on the HSN soma and neurite without being markedly enriched in specific membrane domains (e.g., synapses) (Ringstad and Horvitz, 2008). These data suggest that IRK-1 channels are uniformly expressed on the HSN plasma membrane and not restricted to a specific region of the cell. It is possible, however, that the reporter transgenes used for these experiments are not localized to subcellular compartments either because the GFP tag interferes with such localization or because they are expressed at levels that saturate mechanisms required for their localization.
To determine whether irk-1 is required for EGL-6 signaling in the HSNs, where EGL-6 is expressed, or whether loss of irk-1 function in vulval muscles bypasses the inhibition of HSNs caused by EGL-6(gf), we created transgenes that express double-stranded irk-1 RNAs in a cell type-specific manner. Expression of irk-1 dsRNA in the HSN neurons potently suppressed the egg-laying defects of egl-6(gf) mutants, recapitulating the phenotype of irk-1 mutants (Fig. 2E). By contrast, expression of the same irk-1 dsRNA using a promoter that drives expression in vulval muscles did not suppress the egl-6(gf) phenotype. Together, these data indicate that loss of irk-1 function in the HSN neurons alleviates the inhibition caused by increased EGL-6 signaling.
We next tested whether increased activity of irk-1 in HSNs mimics the effect of increased EGL-6 signaling (i.e., inhibits egg-laying behavior). Animals carrying a transgene that drives irk-1 expression using the HSN-specific promoter of egl-6a were markedly defective in egg-laying behavior (Fig. 2F). This effect of increased irk-1 gene dosage was not suppressed by deleting the egl-6 gene, suggesting that IRK-1 channels function downstream of or in parallel to the EGL-6 GPCR (Fig. 2F).
The effects of loss of IRK-1 channels and loss of GOA-1/Goα on egg-laying behavior are not additive
Although EGL-6 requires IRK-1 to inhibit HSN neurons, it is possible that this dependency does not involve Go signaling, since some GPCRs can activate signaling pathways in a G-protein-independent manner (Lefkowitz and Shenoy, 2005; Lu et al., 2009). We used a genetic interaction test to determine the functional relationship between IRK-1 channels and Goα. If IRK-1 were required for inhibition of the HSN by the EGL-6 GPCR independently of G-proteins, then irk-1(Δ) would modify the egl-6(gf) phenotype even in the absence of Goα. We used a severe loss-of-function allele of goa-1 to make a triply mutant strain defective in both irk-1 and goa-1 in an egl-6(gf) background, and we compared its egg-laying phenotype with that of egl-6(gf) animals that carried either a goa-1(lf) mutation or an irk-1(Δ) mutation (Fig. 3A). We found that deleting irk-1 did not significantly affect the egg-laying behavior of goa-1(lf); egl-6(gf) mutants (Fig. 3A). The failure of irk-1 deletion to enhance the suppression of egl-6(gf) by goa-1(lf) could not be attributed to a saturating effect of goa-1(lf) on the egg-laying system; goa-1(lf) mutants laid earlier stage eggs than did suppressed egl-6(gf) mutants (Fig. 3A). These data indicate that IRK-1 acts downstream of or in parallel to Goα and suggest that the function of IRK-1 is regulated by G-protein signaling. Indeed, sequences required for regulation by G-proteins of mammalian inward-rectifier potassium channels are conserved between those channels and IRK-1 (Fig. 3B), suggesting that IRK-1 channels might be directly targeted by G-proteins.
The failure of irk-1 to enhance suppression of the Egl-6(gf) phenotype by goa-1 suggests that irk-1 and goa-1 function in a common pathway. We tested whether irk-1 mutants have other behavioral defects in common with goa-1 mutants. In addition to hyperactive egg-laying behavior, goa-1 mutants also display hyperactive locomotory behavior (Mendel et al., 1995; Ségalat et al., 1995). We compared the locomotory behavior of irk-1 mutants with that of goa-1 mutants and found that irk-1 mutants did not display the hyperactivity characteristic of goa-1 mutants (Fig. 3C). The role of irk-1 in Go signaling, therefore, seems to be specific to the egg-laying system of C. elegans.
IRK-1 forms homomeric inwardly rectified K+ channels
We identified C. elegans IRK channels as candidate effectors of EGL-6 signaling by their sequence homology to mammalian Kcnj channels. Nematode IRK channels, however, have not been functionally characterized. To determine the functional properties of IRK-1, we reconstituted IRK-1 channels in Xenopus oocytes. IRK-1 expression conferred upon oocytes a large inwardly rectified current (Fig. 4A,B) that reversed near the reversal potential for potassium, which in our experiment was close to 0 mV. Although IRK-1 currents were rectified, we nonetheless observed significant outward currents in IRK-1-expressing oocytes (Fig. 4A,B), and IRK-1 expression caused the resting membrane potential of oocytes to be close to the reversal potential of potassium (Fig. 4C). In vitro, therefore, IRK-1 subunits are sufficient to form homomeric potassium channels that are strongly, but not completely, inwardly rectified. IRK-1 channels have detectable constitutive activity when overexpressed in Xenopus oocytes, as do other members of the Kcnj channel family (Kubo et al., 1993).
Go signaling and IRK-1 channels regulate the excitability of the HSN neurons
In vertebrates, inward rectifier potassium channels regulate the excitability of neurons (Lüscher et al., 1997). Because of the strong genetic interaction between irk-1 and egl-6, we hypothesized that EGL-6 signaling inhibits the excitability of HSN neurons in a manner that depends on IRK-1 channels. To test this hypothesis, we generated transgenic animals that express ChR2 (Nagel et al., 2005) using the egl-6 promoter. This transgene allowed us to deliver defined excitatory stimuli to the HSNs. The egl-6 promoter drives transgene expression in a small number of cells including the HSNs (Ringstad and Horvitz, 2008). To ensure that behavioral responses to photostimulation could be attributed to activation of the HSN neurons, we used structured illumination to functionally map the site of ChR2 action in Promegl-6::ChR2-expressing animals. Blue light illumination of the midbody, where the HSN cell bodies are located, robustly evoked egg-laying behavior (Fig. 5A,B). By contrast, animals illuminated with a pattern of blue light that excluded the midbody laid eggs at a low frequency that was indistinguishable from the response of animals grown in the absence of all-trans retinal, an essential cofactor for ChR2.
We next measured the effects of ChR2-mediated stimulation of HSNs of unrestrained wild-type animals. We exposed transgenic adult hermaphrodites to varying intensities of blue light. Blue light robustly evoked egg laying by transgenic animals, which executed an average of four bouts of egg laying in a 30 s period of exposure to the highest light intensities used (Fig. 6A). The behavioral response of transgenic animals saturated at high light intensities, and half-maximal responses were evoked by ∼70 W/m2 of blue light. We also measured latency to first behavioral response as a function of light intensity. At the highest light intensities used, 50% of the stimulated animals showed a behavioral response within 4 s (Fig. 6B). As light intensities were reduced, the times to first response increased nonlinearly (Fig. 6B,C). Animals exposed to blue light with irradiance of 29 W/m2, which is ∼30–50 times lower than that commonly used to stimulate C. elegans neurons (Guo et al., 2009; Leifer et al., 2011; Lindsay et al., 2011), still laid eggs within the 60 s observation period (Fig. 6B,C). Because the latency to first behavioral response was highly sensitive to changes in the intensity of light stimuli, we reasoned that this metric would also be highly sensitive to changes in HSN physiology and in subsequent experiments measured latencies to first behavioral response to a 29 W/m2 light stimulus. Importantly, under these conditions the egg-laying behavior evoked by whole-body illumination of unrestrained transgenic animals required HSN neurons: transgenic egl-1(gf) mutants, which lack HSNs, failed to demonstrate light-evoked egg laying (Fig. 6D).
Having established a method for exciting the HSN neurons in freely behaving animals, we next tested whether Go signaling controls the sensitivity of the HSNs to ChR2-mediated stimulation. egl-10 encodes an RGS family GTPase-activating protein (GAP) that accelerates hydrolysis of GTP by Goα and thereby antagonizes Go signaling (Koelle and Horvitz, 1996). egl-10 mutants carrying a Promegl-6::ChR2 transgene did not lay eggs in response to a photostimulus that reliably evoked egg-laying behavior when applied to wild-type transgenic animals (Fig. 6E), indicating that globally increasing Go signaling reduced the excitability of the HSN neurons. We next measured the effect of activating Go signaling downstream of the EGL-6 GPCR by testing the behavioral responses of transgenic egl-6(gf) mutants to photostimulation. Like egl-10 mutants, transgenic egl-6(gf) mutants had reduced behavioral responses to photostimulation of HSN neurons (Fig. 6F). Deletion of irk-1 significantly restored the response of egl-6(gf) mutants to excitatory input (Fig. 6F). irk-1 mutants themselves showed a small but significant decrease in latency to behavioral response (Fig. 6G).
Go signaling that couples the EGL-6 neuropeptide receptor to IRK-1 channels is independent of other GPCR-coupled Go signaling
The EGL-6 GPCR is one of several Go-dependent receptors expressed by the HSN neurons. We asked whether IRK-1 channels are generally required for inhibition of HSN neurons by Go signaling, or whether IRK-1 channels are specifically required for signaling downstream of EGL-6 to inhibit HSNs. To address this question, we first tested whether deletion of irk-1 could efficiently suppress the egg-laying defects of egl-10 mutants, which have upregulated Go signaling (Koelle and Horvitz, 1996). Surprisingly, irk-1 only weakly suppressed the egg-laying defect of egl-10(lf) mutants even though egl-10(lf) mutants were less defective in egg laying than were egl-6(gf) mutants (Fig. 7A,B).
We next tested whether IRK-1 channels are required for a different Go signaling pathway that inhibits the HSNs: signaling downstream of the orphan Go-coupled receptor EGL-47 (Moresco and Koelle, 2004; Tanis et al., 2009). We tested whether the egl-47(gf) phenotype requires IRK-1 channels. Although EGL-47 strictly requires Go to inhibit HSNs (Tanis et al., 2009), loss of IRK-1 channels did not suppress the egg-laying defects of egl-47(gf) mutants (Fig. 7C). These data suggest that, in the HSNs, there exists both Go signaling that strongly requires IRK-1 channels (downstream of EGL-6) and Go signaling that is independent of IRK-1 channels (downstream of EGL-47).
Discussion
An ionic mechanism for the modulation of HSN excitability by the neuropeptide receptor EGL-6
Our data indicate that Go signaling downstream of the neuropeptide receptor EGL-6 inhibits serotonergic neurons of the C. elegans egg-laying system via a mechanism that requires IRK-1 potassium channels. Although deletion of irk-1 only slightly changes the function of the egg-laying circuit, the same mutation strongly suppressed inhibitory effects of EGL-6 signaling both on egg-laying behavior and in assays that measure responses of the egg-laying system to direct stimulation of HSNs. This is consistent with a model in which IRK-1 channels have low constitutive activity that is increased by activation of EGL-6. That IRK-1 channels have constitutive activity was clearly demonstrated by the appearance of macroscopic potassium currents in X. laevis oocytes expressing IRK-1 in the absence of EGL-6 or its associated G-proteins. Related vertebrate channels of the Kcnj family demonstrate constitutive activity both in vivo and in heterologous expression systems.
Regulation of ion channels is an established mechanism for the modulation of neuronal activity by GPCRs. Opening IRK-1 channels might decrease the excitability of the HSNs in two ways. First, IRK-1 channels might hyperpolarize the HSNs as they did Xenopus oocytes. Second, opening IRK-1 channels might decrease the input resistance of the HSNs, making the membrane potential less sensitive to depolarizing synaptic currents. C. elegans neurons have extremely high input resistances (Goodman et al., 1998) and are readily depolarized or hyperpolarized by synaptic currents. Indeed, we observed that ChR2-mediated stimulation of the HSN neurons could effectively evoke egg-laying behavior even when low light intensities were used, perhaps reflecting how sensitive HSNs are to such depolarizing currents. Opening IRK-1 channels could therefore both bring the resting potential of the HSNs away from a threshold required for activating voltage-sensitive membrane conductances and decrease the ability of excitatory inputs to depolarize the HSNs.
We also observed that behavioral responses to HSN photostimulation rapidly saturated, indicating that the egg-laying system does not respond in a graded manner to HSN stimulation. This nonlinearity could arise in the HSNs, which might express ion channels capable of generating plateau potentials such as those observed in C. elegans sensory neurons and interneurons (Goodman et al., 1998; Mellem et al., 2008; Ramot et al., 2008). Plateau potentials allow neurons to stably maintain more than one transmembrane potential, which can correspond to discrete functional states (i.e., “on” and “off”). If HSN neurons express plateau potentials, inhibitory neuropeptides that activate EGL-6, and subsequently activate IRK-1 channels, might regulate such state transitions to favor hyperpolarized states. Alternately, the nonlinearity of behavioral responses to increasing HSN stimulation might arise in the vulval muscle, which might integrate synaptic input from the HSNs until a threshold for contraction is reached. In either model, small changes in HSN membrane conductances caused by activation of IRK-1 channels could cause large changes in the function of the egg-laying circuit. It is interesting to speculate that regulation of IRK-1 channels by neuropeptide signaling is integrated with regulation of other ion channels to finely tune the activity of the egg-laying system in response to changes in the environment or internal state.
IRK-1 channels might be homologous to vertebrate G-protein-gated channels
IRK-1 is an ion channel subunit similar to vertebrate Kcnj3, Kcnj5, Kcnj6, and Kcnj9, which are subunits of G-protein-regulated potassium channels. Does IRK-1 also form channels that are directly regulated by G-protein subunits? That eliminating irk-1 function did not enhance the effect of a goa-1 mutation on EGL-6 signaling is consistent with a model in which IRK-1 channels are downstream of Go, although we cannot preclude the possibility that IRK-1 channels act in parallel to Go. The primary sequence of IRK-1 suggests that IRK-1 channels might be directly regulated by Go subunits. Sequences forming a pair of β strands called the βL-βM loop are both required for the activation of vertebrate Kcnj channels by G-protein βγ subunits and form part of the interaction surface between KCNJ channels and Gβγ. These sequences are partly conserved between vertebrate G-protein-gated Kcnj channels and IRK-1 (Fig. 3B), suggesting that like Kcnj channels IRK-1 might bind Gβγ subunits and undergo the specific conformational changes that accompany Gβγ binding (Whorton and MacKinnon, 2011).
G-proteins also regulate ion channels through second messenger signaling. For example, muscarinic acetylcholine receptors in vertebrate neurons alter cell excitability through the modulation of voltage-sensitive potassium channels in the Kcnq family, which are gated by the phospholipid phosphatidylinositol (4,5) bisphosphate (PIP2), the levels of which are controlled by G-protein signaling (Suh and Hille, 2008). Like Kcnq channels, Kcnj channels are regulated by PIP2 (Huang et al., 1998). A recent crystal structure of PIP2-bound Kcnj6 revealed that binding of PIP2 induces a conformational change in the channel that is distinct from and synergizes with the proposed effect of Gβγ binding, and also identified residues that mediate interactions with PIP2 (Whorton and MacKinnon, 2011). In Kcnj6 channels, PIP2 is coordinated by lysine residues, which are conserved between Kcnj channels and IRK-1. Thus, the primary sequence of IRK-1 channel subunits suggests two mechanisms that might couple these channels to G-protein signaling pathways: direct interaction with G-protein subunits, and regulation by PIP2.
Go signaling in the HSN neurons might be functionally compartmentalized
Our results show that not all Go-dependent inhibitory pathways in the HSNs strongly depend on IRK-1 channels. The egg-laying defect of egl-10 mutants, which are mutant for a GAP that acts on Goα (Koelle and Horvitz, 1996), was only weakly suppressed by irk-1 deletion, indicating that a significant amount of Go signaling in the egg-laying system does not require IRK-1 channels. Inhibition of HSNs by the orphan receptor EGL-47 did not require IRK-1 channels at all. While EGL-47 is a divergent member of the GPCR family (Moresco and Koelle, 2004) and might function in a manner different from EGL-6, which is a rhodopsin-like GPCR (Ringstad and Horvitz, 2008), both receptors share a requirement for Go to inhibit HSNs, suggesting that both are Go-coupled receptors (Moresco and Koelle, 2004). We cannot rule out by genetic arguments, however, that EGL-47 signaling and Go signaling function in parallel to inhibit the HSNs, thereby explaining the lack of a requirement for IRK-1 channels in EGL-47 signaling. It is also possible that some mutations studied affect components of the egg-laying system other than the HSNs, while the function of EGL-6 and IRK-1 channels is specific to the HSNs. We note, however, that EGL-10 expression is enriched in the HSNs (Koelle and Horvitz, 1996) and the HSNs have been shown to be the focus of Go signaling in the egg-laying system (Tanis et al., 2008). It is therefore likely that the effects of mutating Go signaling pathways specifically alter HSN function.
Another explanation for the ability of irk-1 mutation to suppress the effects of activating one Go-dependent receptor signaling pathway but not another is that Go signaling in the HSN neurons is functionally compartmentalized. How could such functional compartmentalization of Go signaling occur? In vertebrate neurons, a given Gα subunit can associate with many combinations of β and γ subunits. One determinant of the functional relationship between a GPCR and its effectors is the composition of receptor-associated heterotrimeric G-proteins. C. elegans, however, expresses only one Goα subunit, and the HSN neurons are only known to express one of the two β subunits and one of the two γ subunits encoded by the C. elegans genome. Therefore, the specific association of IRK-1 channels with EGL-6 is unlikely to be mediated by diversity in Go molecules in the HSNs.
In vertebrate neurons, GPCRs couple to distinct effectors depending on where in the cell they are located. For example, metabotropic Gi/o-coupled GABA receptors couple to Kcnj family potassium channels postsynaptically to regulate action potential generation (Lüscher et al., 1997), but act on different effectors presynaptically to regulate release probability. We found no evidence that the cell surface expression of either the neuropeptide receptor EGL-6 or IRK-1 channels was restricted to a particular part of the HSN: we observed both proteins on the cell soma and the neurite, including regions of the neurite that are enriched for presynapses. It is possible that IRK-1 channels associate with EGL-6 in microdomains that are distributed over the cell surface. Vertebrate Kcnj channels associate with G-protein-coupled GABAB receptors (Fowler et al., 2007), suggesting the existence of a preformed signaling complex. In some cells, GPCRs and KCNJ channels associate with a GAP that acts on Gi/o (Labouèbe et al., 2007), indicating that the organization of GPCR signaling complexes differs between cell types and that such complexes can include both positive and negative regulators of Kcnj channels. In the C. elegans HSN neurons, the neuropeptide receptor EGL-6 might function in a signaling complex similar to that described for the vertebrate GABAB receptor. Such a complex would account for the specific functional association that we observed between EGL-6 and IRK-1 channels. Also, discrete GPCR signaling complexes that match a receptor to an effector would allow C. elegans neurons to generate diverse responses to activation of different GPCRs in the absence of molecular diversity of G-protein subunits. Mechanisms that mediate the specific functional association of EGL-6 with IRK-1 channels might exist in vertebrate neurons and contribute to the diversity and complexity of their responses to GPCR activation.
Footnotes
This work was supported by NIH Grant R01-GM024663 (H.R.H.) and NIH Grant R01-GM098320 (N.R.). N.R. is a Pew Scholar in the Biomedical Sciences. Some strains were provided by the Caenorhabditis Genetics Center, which is funded by the National Center for Research Resources. We thank Alexander Gottschalk for providing a ChR2 expression construct and Kara Zang for comments on this manuscript.
The authors declare no competing financial interests.
- Correspondence should be addressed to Niels Ringstad at the above address. niels.ringstad{at}med.nyu.edu