Behavioral Impact of Neurotransmitter-Activated G-Protein-Coupled Receptors: Muscarinic and GABAB Receptors Regulate Caenorhabditis elegans Locomotion (original) (raw)
Articles, Cellular/Molecular
Journal of Neuroscience 9 July 2008, 28 (28) 7104-7112; https://doi.org/10.1523/JNEUROSCI.0378-08.2008
Abstract
Neurotransmitter released from presynaptic terminals activates both ligand-gated ion channels (ionotropic receptors) and a variety of G-protein-coupled receptors (GPCRs). These neurotransmitter receptors are expressed on both presynaptic and postsynaptic cells. Thus, each neurotransmitter acts on multiple receptor classes, generating a large repertoire of physiological responses. The impact of many ionotropic receptors on neuronal activity and behavior has been clearly elucidated; however, much less is known about how neurotransmitter-gated GPCRs regulate neurons and circuits. In Caenorhabditis elegans, both acetylcholine (ACh) and GABA are released in the nerve cord and mediate fast neuromuscular excitation and inhibition during locomotion. Here we identify a muscarinic receptor (GAR-2) and the GABAB receptor dimer (GBB-1/2) that detect synaptically released ACh and GABA, respectively. Both GAR-2 and GBB-1/2 inhibited cholinergic motor neurons when ACh and GABA levels were enhanced. Loss of either GPCR resulted in movement defects, suggesting that these receptors are activated during locomotion. When the negative feedback provided by GAR-2 was replaced with positive feedback, animals became highly sensitive to ACh levels and locomotion was severely impaired. Thus, conserved GPCRs act in the nematode motor circuit to provide negative feedback and to regulate locomotory behaviors that underlie navigation.
Introduction
Divergence is a widespread and important theme in the biology of signaling: the divergence of one molecule to many receptor classes generates a large repertoire of responses to the release of a single chemical. All known transmitter molecules that mediate fast chemical synaptic transmission via ionotropic receptors also activate a variety of G-protein-coupled receptors (GPCRs), thereby extending neurotransmission into multiple intracellular signaling pathways. Ionotropic receptors and GPCRs have very distinct signaling properties. Fast synaptic transmission is mediated by ligand-gated ion channels, which have rapid kinetics (operating on timescales of a few milliseconds), and relatively low affinity for neurotransmitter ligands (typically KD ∼0.1–1 mm), which restricts activation to only those receptors that are clustered adjacent to presynaptic release sites. These properties endow synapses with the ability to produce the extremely fast, local signals that underlie fast synaptic transmission.
In contrast, neurotransmitter-gated GPCRs have properties that typically limit their function to slower, modulatory signals. Neurotransmitter-gated GPCRs regulate neuronal activity via a cascade of cytoplasmic second messengers, giving these signals intrinsically slow kinetics (operating on timescales of seconds to minutes), although in some cases, direct binding of the βγ subunit can modulate ion channels in tens to hundreds of milliseconds (Dittman and Regehr, 1997). GPCRs have relatively high affinity for neurotransmitter ligands (KD ∼0.1–1 μm), which permits activation of receptors that are relatively far removed for neurotransmitter release sites (Hille, 1992). Thus, signals produced by neurotransmitter-gated GPCRs act over longer timescales and larger spatial domains.
Although a great deal is known about the contribution of neurotransmitter-gated ion channels to fast synaptic transmission and behavior, much less is known about how neuro-transmitter-gated GPCRs regulate circuit activity and behavior. In mammals, extrasynaptic GABA inhibits neuronal activity via metabotropic GABA (GABAB) receptors (Isaacson et al., 1993; Dittman and Regehr, 1997), whereas extrasynaptic glutamate can both stimulate and inhibit neighboring neurons by activating ionotropic and metabotropic receptors (Mitchell and Silver, 2000; DiGregorio et al., 2002; Szapiro and Barbour, 2007). Thus, the same neurotransmitters that directly gate a postsynaptic ionotropic receptor can influence intracellular signaling in both the presynaptic and postsynaptic cells as well as neighboring neurons.
The nematode Caenorhabditis elegans provides a simple model organism for investigating the behavioral impact of neurotransmitter-gated GPCRs. Nematode propulsion arises from sinusoidal waves of muscle contraction and relaxation driven by acetylcholine (ACh) and GABA, respectively (Croll, 1975; Niebur and Erdos, 1993; Schuske et al., 2004). C. elegans locomotion is extensively regulated by G-protein signaling pathways, including antagonistic regulation of ACh secretion by Go and Gq (Hajdu-Cronin et al., 1999; Lackner et al., 1999; Miller et al., 1999; Nurrish et al., 1999; Robatzek and Thomas, 2000; Robatzek et al., 2001; Bastiani and Mendel, 2006). Although these studies suggest that Go and Gq play important roles in regulating locomotion, much less is known about the upstream GPCRs regulating these pathways and specifically about the role of neurotransmitter-gated GPCRs. In this study, we identified two motor neuron GPCRs that detect ACh and GABA and analyzed the locomotion behavior of mutants lacking these receptors.
Materials and Methods
Strains.
Strains were maintained at 20°C as described by Brenner (1974). The following strains were used in this study: N2 Bristol, gar-2(ok520), gbb-1(tm1406), gbb-2(tm1165), goa-1(sa734), KP3473: nuIs155 [_Punc-17::Venus::GAR-2_], KP3457: nuEx1076 [_Punc-17::Venus::GAR-3_], KP3456: gar-2(ok520);nuEx1075 [Pgar-2::Venus::GAR-2_], KP3526: nuEx1072 [Pgar-2::GFP_];nuEx1067 [_Punc-25::dsRed2_], KP3447: nuEx1066 [_Pgbb-1::GFP_];nuEx1067, KP4488: Punc-25::Venus::GAR-2;gar-2(ok520), eri-1(mg366);lin-15B(n744).
The eri-1;lin-15B strain was used for feeding RNA interference (RNAi) experiments because neurons of wild-type animals are refractory to RNAi. This strain was developed to provide enhanced sensitivity to RNA interference in neurons (Sieburth et al., 2005).
Constructs and transgenes.
cDNA encoding the splice form GAR-2a was subcloned into the fire lab vector pPD49.26 flanked by _Nhe_I and _Kpn_I. An in-frame _Not_I site was introduced after the start codon, and YFPVenus was ligated to create an N-terminal fusion protein. The signal sequence from pat-3 β integrin was included on the 5′ end of YFPVenus to ensure proper receptor insertion. Tissue-specific expression of this fusion protein was determined by including ∼3 kb of 5′ regulatory sequence from unc-17 vesicular ACh transporter (VAChT) (for all cholinergic neurons), gar-2 (for rescue construct), or unc-25 glutamic acid decarboxylase (GAD) (for rescue in GABAergic neurons only). cDNA encoding the splice form GAR-3b was subcloned as described for GAR-2a. Transcriptional reporters for gar-2 and gbb-1 were made by subcloning 3 kb of 5′ regulatory sequence for each of these genes into the fire lab vector pPD95.75, which contains a cDNA encoding soluble green fluorescent protein (GFP). The unc-25 reporter construct was made by replacing GFP in pPD95.75 with DsRed2 (Clontech) and then subcloning 3 kb of 5′ regulatory sequence from the unc-25 gene.
Predictions for the gene structures of gbb-1 and gbb-2 were generated using GeneMark HMM routines (http://opal.biology.gatech.edu/GeneMark/eukhmm.cgi), and cDNAs were amplified and sequenced from a C. elegans cDNA library. Two splice variants for gbb-1 were identified (2.5 and 2.6 kb), whereas only one cDNA for gbb-2 was identified. GenBank accession numbers are as follows: gbb-1, EU29334; gbb-2, EU29335.
Fluorescence imaging.
Venus::GAR-2-expressing animals were mounted on agarose pads and viewed on a Zeiss Axiovert microscope, using an Olympus PlanApo 100×, numerical aperture 1.4 objective, as described previously (Dittman and Kaplan, 2006). Images were captured with a Hamamatsu ORCA digital camera, and line scans were analyzed with custom software in Igor Pro (Wavemetrics). Images of 500 nm fluorescein-conjugated beads (Invitrogen) were captured during each imaging session to provide a fluorescence standard for comparing absolute fluorescence levels between animals. Background signal (CCD dark current and slide autofluorescence) was subtracted before analysis.
Aldicarb and levamisole assays.
Sensitivity to aldicarb and levamisole was determined by analyzing the onset of paralysis after treatment with either 1 mm aldicarb (Chem Services) or 200 μm levamisole (Sigma) as described previously (Nurrish et al., 1999). Each assay included between two and four plates of at least 20 animals for each genotype being tested. The experimenter was blind to the genotypes. Each genotype was tested between 4 and 10 independent experiments, and paralysis curves were generated by averaging paralysis time courses for each plate.
Worm tracking and analysis.
For tracking measurements, worms were reared at 20°C and moved to room temperature 30 min before imaging. Young adult animals were picked to agar plates with no bacterial lawn (5–15 worms per plate). Imaging began 5–10 min after worms were removed from food. One minute digital videos of individual animals were captured at 8× magnification and 4 Hz frame rate on a Zeiss Discovery Stereomicroscope using Axiovision software. If the animal left the field of view, acquisition was paused while the plate was manually recentered, and, after a few seconds, acquisition was restarted. The center of mass was recorded for each animal on each video frame using object tracking software in Axiovision. The trajectories were then analyzed using custom software written in Igor Pro 5.0 (Wavemetrics). The raw data were filtered to remove artifactual discontinuities in the trajectory attributable to noise in the center-of-mass estimates. Speed and angle values were calculated and from frame-by-frame changes in position and angle (from −180° to +180°), with turns and reversals estimated as follows: turns were counted as angle changes between 50° and 160° (taking the absolute value of the angle change). Reversals were counted as angle changes between 170° and 180°. The accuracy of these criteria was assessed by visual inspection of the video. Averages for speed, turn number, and reversal number were determined for each animal and then pooled to form estimated values for each genotype. SE was based on the animal-to-animal variation in parameter values. Typically between 40 and 70 animals were assayed for each genotype. For _R_max distributions, the maximal Euclidean distance from the starting point within a 40 s time interval was recorded for each animal. The cumulative distributions of _R_max values were compared between genotypes using the Kolmogorov–Smirnov statistic.
Results
GAR-2/M2 mediates inhibition of cholinergic neurons when ACh levels are pharmacologically enhanced
The worm genome contains numerous conserved neurotransmitter-gated GPCRs, including three muscarinic acetylcholine receptors (mAChRs), four dopamine, three serotonin, and two glutamate receptors, and two predicted GABAB receptor subunits (Bargmann, 1998; Lee et al., 2000; Chase et al., 2004; Carre-Pierrat et al., 2006). We examined the effects of neuronal mAChRs on motor neuron function and locomotion using an M2 receptor GAR-2 deletion mutant [gar-2(ok520)] and RNA interference. The ok520 deletion removes 1.1 kb of the gene including most of the third intracellular loop, a region of the receptor that is essential for coupling to G-proteins (Kubo et al., 1988; Burstein et al., 1996) and consequently is likely to eliminate GAR-2 function. ACh levels are normally limited by synaptic AChEs so inhibiting cholinesterase activity will elevate ACh (Hartzell et al., 1975). To experimentally enhance ACh levels, we used the AChE inhibitor aldicarb. Aldicarb treatment induces paralysis with a characteristic time course attributable to enhanced muscle activation (Miller et al., 1996; Nurrish et al., 1999).
To test the effects of GAR-2 on cholinergic motor neuron synapses, we measured the time course of acute paralysis induced by aldicarb in gar-2 mutants. The gar-2 deletion mutation and gar-2 inactivation by RNAi both accelerated the time course of aldicarb-induced paralysis (Fig. 1A,B). Muscle sensitivity to the ionotropic cholinergic agonist levamisole was not altered in the gar-2 mutant (Fig. 1C), indicating that intrinsic muscle sensitivity to ACh was not affected by loss of GAR-2. Wild-type aldicarb sensitivity was restored in gar-2 mutants by expressing an N-terminal YFPVenus-tagged GAR-2 cDNA using the gar-2 promoter. These results suggest that GAR-2 regulates the activity of ventral cord motor neurons.
Figure 1.
gar-2 mutant animals are hypersensitive to the cholinesterase inhibitor aldicarb. A, Paralysis time course in 1 mm aldicarb for wild-type (open circles), gar-2 (filled squares), and cDNA rescue (filled circles). B, Paralysis time course in 1 mm aldicarb for the enhanced RNAi strain eri-1;lin-15B (see Materials and Methods). Animals were fed empty vector (open circles) or gar-2 RNA vector (filled squares). C, Paralysis time course in 0.2 mm levamisole for wild type (open circles) and gar-2 (filled circles). Data are mean ± SEM.
GAR-2 has been shown previously to be expressed throughout the nervous system, including ventral cord neurons, but not in muscle (Lee et al., 2000). Consistent with this study, we found that GAR-2 was expressed in some cholinergic motor neurons as well as GABAergic motor neurons, the two major types of ventral cord motor neurons (Fig. 2A). We examined subcellular localization of GAR-2 by expressing a YFP::GAR-2 fusion protein in cholinergic motor neurons and imaged the dorsal cord (Fig. 2Ad). The protein appeared to be diffusely distributed in axons with no obvious spatial relationship to presynaptic terminals. A similar pattern was seen using a C-terminal fusion protein GAR-2::YFP (data not shown).
Figure 2.
GAR-2/M2 is expressed in cholinergic and GABAergic motor neurons and requires goa-1 Gαo. A, Representative image of ventral cord motor neurons expressing soluble GFP driven by the gar-2 promoter (a) and soluble DsRed2 driven by the unc-25 GAD promoter (b). The two images are superimposed to illustrate neurons that coexpress GFP and DsRed2 (c). Asterisks mark GABA motor neurons expressing GAR-2. YFPVenus::GAR-2 expression in a dorsal motor axon (d). Scale bar: a–c, 10 μm; d, 3.3 μm. B, Paralysis time course in 1 mm aldicarb for wild type (open black circles), gar-2 (filled green squares), rescue:cDNA rescue under the gar-2 promoter (filled red circles), gar-2(GABA):cDNA rescue under the unc-25 GAD promoter (open purple squares), and gar-2(ACh):cDNA rescue under the unc-17 VAChT promoter (filled blue triangles). C, Paralysis time course in 1 mm aldicarb for wild type (filled black circles), goa-1 Gαo (filled red circles), g_ar-2(ACh)_:cDNA rescue under the unc-17 VAChT promoter (filled blue circles), and goa-1;gar-2(ACh) (filled green circles). Data are mean ± SEM.
To determine which motor neurons require GAR-2, we expressed gar-2 cDNA constructs in subsets of cholinergic or GABAergic motor neurons of gar-2 mutants (Fig. 2B). When GAR-2 was expressed in all cholinergic motor neurons using the VAChT unc-17 promoter, transgenic animals became strongly resistant to aldicarb [_gar-2(ACh)_] (Fig. 2B). In contrast, no rescue was observed when GAR-2 was expressed in GABAergic motor neurons using the unc-25 GAD promoter [_gar-2(GABA)_] (Fig. 2B). These results suggest that GAR-2 mediates feedback inhibition of cholinergic motor neurons when ACh levels are elevated by aldicarb treatment.
GOA-1 is required for GAR-2/M2-mediated feedback inhibition
GAR-2 is predicted to be a M2/4 subtype mAChR (36% identity, 60% similarity to rat M2), which couples to the Gαi/o class of G-proteins (Lanzafame et al., 2003). In animals lacking the Gαo subunit GOA-1, aldicarb-induced paralysis is greatly accelerated, consistent with a loss of feedback inhibition by Gαo-coupled GPCRs (Fig. 2C) (Hajdu-Cronin et al., 1999; Miller et al., 1999; Nurrish et al., 1999). The aldicarb resistance caused by gar-2(ACh) was eliminated in goa-1 mutants (Fig. 2C), indicating that GAR-2 inhibition requires Gαo.
Ectopic expression of GAR-3/M3 switches the sign of feedback regulation
Excess GAR-2 in all cholinergic neurons likely exaggerates Gαo-mediated inhibition when ACh is increased by aldicarb treatment. In contrast, the Gαq pathway enhances transmitter secretion in a variety of systems, including cholinergic motor neurons (Hajdu-Cronin et al., 1999; Lackner et al., 1999; Miller et al., 1999; Bauer et al., 2007). We reasoned that we could switch the sign of feedback regulation by ectopically expressing a muscarinic receptor that couples to Gαq in cholinergic motor neurons. To test this idea, we ectopically expressed the worm M1/3 type muscarinic receptor (GAR-3) in cholinergic neurons. GAR-3 normally activates the Gαq pathway in pharyngeal muscle (Steger and Avery, 2004; You et al., 2006). If GAR-3 activation in cholinergic motor neurons leads to enhanced ACh release, we would expect that the resulting positive feedback would severely disrupt normal locomotion. In contrast, excess GAR-2 expression would be expected to have little effect under conditions in which activation by excess ACh is minimal. We monitored the effects of changing muscarinic signaling on body posture as well as aldicarb sensitivity. As a measure of resting motor neuron drive onto body wall muscles, we calculated the curvature along the A-P axis of quiescent animals, in which basal curvature is maintained by the steady-state balance of muscle contraction on both sides of the body. The average A-P curvature was unaffected by either removal or overexpression of GAR-2, whereas overexpression of GAR-3 [_gar-3(ACh)_] caused resting curvature to double (122 ± 23% increase; p < 0.01) (Fig. 3A,B). As a second assay, we compared the time course of paralysis when ACh levels were increased with aldicarb. In contrast to the delayed paralysis observed in gar-2(ACh) animals, paralysis was greatly accelerated in gar-3(ACh) animals (Fig. 3C), further supporting the notion that gar-3(ACh) animals are highly sensitive to increases in ambient ACh.
Figure 3.
Ectopic GAR-3 mAChR3 in cholinergic motor neurons enhances ACh release. A, Representative images of wild-type, gar-2 mutant, gar-2(ACh), and gar-3(ACh) animals with overlaid trace of A–P axis used for measuring body posture curvature. Scale bar, 300 μm. B, A–P curvature averaged over length of animal for the four genotypes shown above (for calculation of curvature, see Materials and Methods). **p < 0.01 by Student's t test. C, Paralysis time course in 1 mm aldicarb for wild-type (black), gar-2 (red), gar-2(ACh):cDNA rescue under the unc-17 VAChT promoter (blue), and gar-3(ACh):gar-3 mAChR3 cDNA expression under the unc-17 VAChT promoter (green). Data are mean ± SEM.
GABAB receptors mediate heterosynaptic feedback when ACh levels are enhanced
During a body bend, ACh depolarizes muscles on one side of the body, whereas GABA release hyperpolarizes the antagonist muscles on the contralateral side (Schuske et al., 2004). GABAergic motor neurons are driven exclusively by excitatory input by cholinergic motor neurons. As a consequence of this connectivity, conditions that enhance ACh release will cause a parallel increase in GABA release. Thus, in addition to the fast synaptic inhibition of muscle activity by ionotropic GABA receptors, metabotropic GABA receptors may detect elevated GABA levels and produce longer-term modulation of neuronal or muscle activity. Consistent with this idea, two genes predicted to encode GABAB receptor subunits were identified in a large-scale RNA interference screen for aldicarb hypersensitivity(Vashlishan et al., 2008). gbb-1 (Y41G9A.4) encodes the worm GABABR1 subunit (40% identity, 63% similarity to rat GBR1), and gbb-2 (zk180.1/2) encodes the worm GABABR2 subunit (31% identity, 56% similarity to rat GBR2). Knockdown of these two gene products by RNAi accelerated paralysis in aldicarb similar to RNAi of gar-2 (data not shown).
To confirm these RNAi results, we examined the effects of two deletion mutants that are predicted to eliminate GABAB receptor function. The gbb-1(tm1406) mutant contains a 1281 bp deletion that removes the ligand binding site for GABA on the N terminus of the receptor (Fig. 4A) (Bettler et al., 2004). The gbb-2(tm1165) mutant contains a 364 bp deletion that removes transmembrane regions 2, 3, and 4, including the second intracellular (i2) loop that is required for activation Gα subunits (Bettler et al., 2004) (Fig. 4A). GABAB receptors function as obligate heterodimers in which GABA binding is mediated by the GABABR1 subunit, and G-protein coupling is mediated by the GABABR2 subunit (Bettler et al., 2004); consequently, both deletion alleles are predicted to eliminate GABAB receptor function. As shown in Figure 4B, both deletion mutants resulted in nearly identical acceleration of paralysis in aldicarb, similar to loss of GAR-2 receptor function. This hypersensitivity to aldicarb was not likely attributable to changes in muscle sensitivity to ACh because gbb-1 and gbb-2 mutants were not hypersensitive to the cholinergic agonist levamisole (Fig. 4C). We examined interactions between GAR-2, GBB-1, and GBB-2 by measuring paralysis rates in aldicarb for double mutants. The rate of aldicarb-induced paralysis of the gbb-1;gbb-2 double knock-out was indistinguishable from that observed in either single mutant (Fig. 4D), consistent with GBB-1 and GBB-2 functioning together as a heterodimer. However, we found a significant degree of additivity in the gar-2;gbb-2 double knock-out at early time points in the aldicarb paralysis time course (Fig. 4D), suggesting that both mAChR and GABAB receptors contribute to Gαo signaling when ACh is elevated.
Figure 4.
GABAB receptor mutants are hypersensitive to aldicarb. A, Protein domain structure for GBB-1 and GBB-2. Domains indicated are signal peptide (s.p.), amino acid binding domain (GABA-Binding), heptahelical transmembrane region (7 TMR), and coiled-coil domain (C–C). Percentages are amino acid sequence identity to orthologous rat GABAB receptor domains. Parenthetical percentages are amino acid identity within the second intracellular (i2) loop. Deletions for knock-out alleles are indicated. B, C, Paralysis time course in 1 mm aldicarb (B) and 0.2 mm levamisole (C) for wild type (open black circles), gbb-1 (open blue circles), and gbb-2 (open red triangles). D, Percentage of animals paralyzed on 1 mm aldicarb at 50 min (black) and 70 min (gray) for wild type, single, and double mutants as indicated. E, Representative image of ventral cord motor neurons expressing soluble GFP driven by the gbb-1 promoter (a) and soluble DsRed2 driven by a GABAergic neuron-specific promoter (b). The two images are superimposed to illustrate neurons do not coexpress GFP and DsRed2 (c). Scale bar, 5 μm. Data are mean ± SEM. *p < 0.01 compared with gar-2, gbb-1, and gbb-2 by Student's t test.
A transcriptional reporter containing the gbb-1 promoter driving expression of GFP was broadly expressed in the nervous system, including cholinergic but not GABAergic motor neurons or muscle (Fig. 4E). Because gbb-1/2 mutants were hypersensitive to ACh accumulation and had wild-type levamisole sensitivity, cholinergic expression of GABAB receptors suggests that they detect GABA released from neighboring GABAergic motor neurons. These GABAergic neurons predominantly form synapses onto muscle so the GABA that activates cholinergic GABAB receptors is probably extrasynaptic in origin.
Feedback regulation by muscarinic and GABAB receptors regulates locomotion
Thus far, our results implicate GAR-2 and GBB-1/2 in the regulation of cholinergic motor neuron function when levels of ACh and GABA are increased by aldicarb treatment. To determine whether this feedback inhibition normally plays a role in locomotion, we analyzed the movement of actively foraging animals in the absence of aldicarb. Figure 5A shows representative trajectories of wild type, gar-2 mutant, gar-2(ACh), and gar-3(ACh) worms over a 1 min interval. These trajectories followed the center of mass for each animal as it explored an agar plate. Wild-type and gar-2 mutants had similar patterns of exploratory behavior consisting of long runs of forward locomotion interrupted by brief reversals and large-angle turns as has been described previously (Pierce-Shimomura et al., 1999; Hills et al., 2004). In contrast, gar-2(ACh) and gar-3(ACh) animals showed aberrant trajectories and explored a more restricted area in a given time interval. We quantified locomotory behavior using three parameters: speed, turning frequency, and reversal frequency. Speed of locomotion was slightly but significantly accelerated in gar-2 mutants (9.2 ± 3.1% increase; p < 0.01) and retarded in gar-2(ACh) animals (25.4 ± 2.7% decrease; p < 0.01), suggesting that GAR-2 regulates the speed of locomotion (Fig. 5B, top). When GAR-3 was ectopically expressed in cholinergic motor neurons, the locomotory speed was greatly reduced (74.6 ± 3.4% decrease; p < 0.01).
Figure 5.
Effects of muscarinic receptors on locomotion. A, Representative trajectories superimposed for 10 animals of each genotype as indicated. The starting points for each of the trajectories were aligned for clarity (red circles). B, Top, Average frame-to-frame speed of the center-of-mass in micrometers per second for the genotypes shown above as well as gar-2 cDNA rescue. Middle, Average frequency of large-angle forward turns defined as angles >50° but <160°. Bottom, Average frequency of direction reversals defined as angles near 180°. C, Diagram showing the trajectory of an animal during a 40 s interval. The maximal Euclidean distance from the starting point (red circle) is defined as _R_max. Scale bar, 1 mm. D, Cumulative distributions of R_max during a 40 s interval for wild-type (black), gar-2 mutant (red), and cDNA rescue (blue). Data are mean ± SEM. **p < 0.01 by Student's t test. ††_p < 0.01 by Kolmogorov–Smirnov test.
In addition to GAR-2 receptor effects on speed, direction changes were also subject to GAR-2 regulation. Turning frequency was defined as the number of large-angle (between 50° and 160°) forward turns per minute, whereas reversals were scored as 180 ± 10° changes in direction. Removal of GAR-2 decreased the frequency of turns in half (60.8 ± 8.3% decrease; p < 0.01), whereas excess GAR-2 did not affect turning frequency (Fig. 5B, middle). Ectopic expression of GAR-3, however, dramatically increased the turning rate (approximately eightfold increase; p < 0.01) to such an extent that animals coiled back on themselves and were unable to maintain forward progress (see image in Fig. 3A). The rate of reversals was also reduced by half in gar-2 mutants (52.5 ± 8.7% decrease; p < 0.01), whereas excess GAR-2 doubled the reversal rate (103.1 ± 28.3% increase; p < 0.01) and ectopic GAR-3 caused a modest increase (45.9 ± 23.5%; p < 0.05) (Fig. 5B, bottom). The maximum net distance an animal could travel in a given time interval is determined by the speed as well as turning and reversal rates. We calculated this maximum distance (_R_max) as a measure of the net velocity of an animal over an extended period of time. A schematic of the maximal distance measure is shown in Figure 5C. Within a 40 s interval, the most distant point from the starting location (_R_max) was determined for each animal, and a cumulative histogram of _R_max was generated. The median distance reached by wild-type animals (n = 67) was ∼3.5 mm, whereas gar-2 mutants (n = 72) reached a median distance of 4.3 mm, corresponding to a 25% increase (p < 0.01) in the net distance covered by gar-2 mutants (Fig. 5D). All effects on locomotory behavior could be rescued by Venus::GAR-2 expression under the endogenous promoter (Fig. 5, rescue). Thus, GAR-2 regulates how often animals change direction and the extent of the area they explore in a given time.
We next analyzed the role of GABAB receptors in regulating locomotion. The effects of removing GABAB receptor modulation on locomotion were similar to those observed in gar-2 mutants (Fig. 6). Speed was accelerated significantly (20 ± 3.5% increase; p < 0.01), whereas turn and reversal frequency were decreased (41.4 ± 10.1 and 42.5 ± 13.4%, respectively; p < 0.05). The maximal distance covered in 40 s was significantly increased in both double knock-outs (wild-type median, 3.5 mm; gar-2;gbb-2 median, 4.1 mm; gbb-1;gbb-2 median, 4.4 mm). In contrast to the additivity observed in the time course of aldicarb paralysis, no additivity was observed on locomotory changes. Thus, the same receptors that provide negative feedback under conditions of elevated transmitter also regulate locomotory behavior. The observations described above suggest that ACh and GABA released during normal foraging behavior modulate aspects of locomotion and navigation, in addition to their roles in fast excitatory and inhibitory transmission.
Figure 6.
GABAB receptors modulate locomotion. A–C, Center-of-mass speed (A), large-angle turn frequency (B), and reversal frequency (C) for wild-type, gbb-1;gbb-2 double knock-out, gar-2, and gar-2;gbb-2 double knock-out animals. D, Cumulative distributions of R_max for wild-type (thin line), gbb-1;gbb-2 double knock-out (thick black line), and gar-2;gbb-2 double knock-out (thick gray line). Data are mean ± SEM. *p < 0.05, **p < 0.01 compared with wild type by Student's t test. †_p < 0.05, ††p < 0.01 by Kolmogorov–Smirnov test.
Discussion
Here we use C. elegans as a model to identify two neuronal GPCRs that detect transmitter ligands and to test their impact on a behavioral circuit. First, we identify GAR-2/M2 and GBB-1/2 GABABR as the GPCRs that detect ACh and GABA, respectively, in the ventral cord. The M2 receptor GAR-2 is expressed in motor neurons and mediates negative feedback via the Gαo pathway during periods of increased ACh release. Second, C. elegans uses a GABAB receptor dimer in cholinergic motor neurons that mediates heterosynaptic inhibition, likely through the Gαo pathway. Third, both GAR-2/M2 and GABAB receptors promote initiation of direction changes during locomotion. Ectopic expression of GAR-2 in all cholinergic neurons bestows profound negative feedback when ACh levels are elevated, and transgenic animals exhibit opposite locomotory phenotypes as loss of GAR-2 including slower speed and greater frequency of reversals. Finally, adding positive feedback through ectopic expression of the Gαq-coupled muscarinic receptor GAR-3/M3 severely disrupts locomotion by promoting exaggerated body bends.
Site of action: synaptic versus extrasynaptic activation of GAR-2 and GBB-1/2
In this study, we found that cholinergic motor neurons expressed muscarinic and GABAB receptors. In principle, these receptors could be strictly localized to postsynaptic sites, in which case, the behavioral effects described here were attributable to disruption of a select group of synaptic inputs with a major postsynaptic metabotropic component. Alternatively, these receptors could be acting as detectors of extrasynaptic ACh and GABA accumulating through ongoing synaptic activity. We favor the latter possibility for several reasons. First, the affinity and signaling kinetics of metabotropic receptors are not suited to the rapid spikes of neurotransmitter concentration that peak at hundreds of micromolar to millimolar levels and decay within milliseconds. In contrast, neurotransmitter levels can increase above the micromolar level outside of the synaptic cleft via spillover and summate when multiple synapses are concurrently active (Isaacson et al., 1993; Dittman and Regehr, 1997; DiGregorio et al., 2002). This extrasynaptic source of neurotransmitter would take significantly longer to clear below micromolar levels attributable to the larger volume (Carter and Regehr, 2000; DiGregorio et al., 2002), making it ideal for activation of slow, high-affinity receptors.
Second, when subcellular localization of metabotropic receptors has been studied using immunoelectron microscopy and fluorescently tagged GPCRs, there is often a significant population of perisynaptic and nonsynaptic receptors (Liu et al., 1998; Fritschy et al., 1999). We did not observe enrichment of GAR-2 within neuronal processes, nor did GAR-2 appear to be excluded from synaptic regions, further suggesting that these receptors are not strictly localized. Furthermore, given the high affinity of GAR-2 receptors for ACh, receptors found at the synapse would be expected to be constitutively liganded and desensitized by the high local concentrations of ACh achieved in the synaptic cleft.
Third, many GPCRs detect neurotransmitter that is not synaptically released onto the neurons in which they are expressed. For instance, GABAB receptors are commonly found near presynaptic terminals of glutamatergic neurons that do not receive direct axo-axonal synapses from GABAergic interneurons in the hippocampus and cerebellum. Cholinergic motor neurons expressed the GBB-1 GABAB receptor subunit, although no GABAergic to cholinergic neuron synapses have been described in C. elegans (White et al., 1976). Indeed, cholinergic neurons also express dopamine receptors, although they do not receive direct dopaminergic inputs (Chase et al., 2004). For these reasons, we postulate that GAR-2 and GBB-1/2 act to inhibit cholinergic activity in response to increases in ambient extrasynaptic ACh and GABA.
Conservation of muscarinic and GABAB receptors in nematodes
Presynaptic M2 muscarinic receptors are known to modulate synaptic transmission at both neuromuscular junctions and central synapses in mammals by signaling through Gαo (Haddad and Rousell, 1998; Gomeza et al., 1999; Zhang et al., 2007). Conservation of this pathway in nematodes suggests that GAR-2 participates in an ancient negative modulatory axis to regulate neuronal function.
GABAB receptor subunits have been predicted by sequence homology after sequencing of the worm genome (Duthey et al., 2002). Here, we confirm that C. elegans expresses two gene products that share a high degree of homology to vertebrate GABAB receptors in both the GABA-binding domain and the heptahelical transmembrane region. In vertebrates, GABAB receptors are heterodimers made up of a GABABR1 subunit that binds GABA and a GABABR2 subunit that binds the C terminus of Gαo through its i2 loop (Duthey et al., 2002; Bettler et al., 2004; Thuault et al., 2004). Interestingly, the GABABR1 subunit GBB-1 GABA-binding domain has substantially greater amino acid sequence identity with its rodent ortholog than GBB-2 compared with rodent GABABR2 (55 vs 31%) (Fig. 4A), whereas GBB-2 shares higher amino acid identity with its ortholog in the heptahelical region, especially in the i2 loop (58 vs 25% for GBB-2 and GBB-1, respectively). Thus, phylogenetic conservation of these domains reflects their function in GABA binding and Gαo signaling.
Determinants of locomotory behaviors in C. elegans
Neurotransmitter-activated GPCRs have been implicated in many functions of the mammalian sensory and motor systems. For instance, rodent knock-outs of M2 and GABABR subunits have been associated with specific locomotory alterations (Gomeza et al., 1999; Schuler et al., 2001; Zhang et al., 2002; Gassmann et al., 2004). C. elegans possesses simple sensory and motor circuits that are amenable to precise genetic changes in GPCR function. These nematodes navigate their environment through long runs of forward locomotion interrupted by brief reversals and large-angle (omega turn) body bends that alter direction (Croll, 1975). Many of the neurons that initiate or control these stereotyped behaviors have been identified (Pierce-Shimomura et al., 1999; Hills et al., 2004; Gray et al., 2005). Because loss of either GAR-2 or GBB-1/2 decreases reversal and omega turn frequencies and overexpression of GAR-2 increases reversal frequency, it is possible that these inhibitory receptors act in neurons that normally suppress turns and reversals. The gustatory sensory neuron ASI, the interneuron AIY, as well as a head motor neuron RIM have all been reported to inhibit reversals and omega turns (Gray et al., 2005). Thus, these neurons would be candidates for targets of inhibitory GPCR modulation. RIM in particular is cholinergic and therefore may use homosynaptic negative feedback to regulate its suppression of reversals. Additional studies elaborating on the specific neurons that endogenously express GAR-2 will reveal how ACh signaling in the navigation circuits use GAR-2 to trigger direction changes.
GPCRs generally modulate neuronal activity on a timescale of hundreds of milliseconds to many seconds. Because the body bends that drive locomotion occur at a rate of about one per second, it is unlikely that transmitter release and activation of GAR-2 and GBB-1/GBB-2 will modulate locomotion within a body bend. Instead, these pathways may set the basal tone of inhibition. In addition, the GPCR mutant data suggest that the transition between behaviors (i.e., reversal of direction) may itself be modulated. These transitions occur at a frequency of one to five per minute so there is ample time for neurotransmitter to accumulate and activate the Gαo pathway.
An in vivo analysis of feedback regulation
Neurotransmitter spillover has been proposed to mediate homosynaptic and heterosynaptic depression and, consequently, to play an important role in activity-dependent modulation of circuit function (Dittman and Regehr, 1997; Mitchell and Silver, 2000). Here we describe a possible role for extrasynaptic ACh and GABA in regulating a locomotory circuit via metabotropic receptors. In particular, we focused on a negative feedback loop created by the inhibitory action of ACh and GABA on cholinergic neurons. Negative feedback typically provides a stabilizing force that can contribute to the fine tuning of a network. Exchanging a negative feedback loop for a positive feedback would likely abolish this stability because small perturbations would be greatly amplified. In the case of exchanging an M2 receptor for an M3 receptor, small increases in extrasynaptic ACh would now drive additional ACh release by coupling to a stimulatory pathway. Even under resting conditions, transgenic animals expressing the M3 receptor showed severe dysfunction of coordinated locomotion. Body bend angles were greatly increased, compromising the ability of the animal to generate and propagate the sinusoidal waves necessary for normal locomotion.
In addition to the homosynaptic negative feedback provided by cholinergic GAR-2 receptors, GABA motor neurons also expressed GAR-2. Although GABAergic rescue of GAR-2 did not restore wild-type aldicarb sensitivity, it is possible that heterosynaptic effects of ACh on GABA neurons may be important for navigational behaviors.
Our results extend C. elegans as a genetic model to dissect the functional impact of GPCR-mediated feedback regulation on a behavioral circuit. Future studies will help elucidate the particular neurons involved in modulating locomotory behavior as well as the spatial and temporal domains over which the GAR-2 and GBB-1/2 receptors sense their ligands and modulate the nervous system.
Footnotes
- This work was supported by National Institutes of Health Grant GM54728 (J.M.K.) and a Damon Runyon CRF postdoctoral fellowship (J.S.D.). We thank the C. elegans Genetic Stock Center and S. Mitani for strains. We also thank members of the Kaplan laboratory for comments on this manuscript.
- Correspondence should be addressed to Joshua M. Kaplan, Department of Molecular Biology, Massachusetts General Hospital, Simches Research Building, 7th Floor, 185 Cambridge Street, Boston, MA 02114. kaplan{at}molbio.mgh.harvard.edu