The Voltage-Gated Anion Channels Encoded by clh-3 Regulate Egg Laying in C. elegans by Modulating Motor Neuron Excitability

CLC-2 is a hyperpolarization-activated, inwardly rectifying chloride channel. Although the properties of the CLC-2 channel have been well characterized, its function in vivo is not well understood. We have found that channels encoded by the Caenorhabditis elegans CLC-2 homolog clh-3 regulate the activity of the spontaneously active hermaphrodite-specific neurons (HSNs), which control the egg-laying behavior. We identified a gain-of-function mutation in clh-3 that increases channel activity. This mutation inhibits egg laying and inhibits HSN activity by decreasing its excitability. Conversely, loss-of-function mutations in clh-3 lead to misregulated egg laying and an increase in HSN excitability, indicating that these channels modulate egg laying by limiting HSN excitability. clh-3-encoded channels are not required for GABAA-receptor-mediated inhibition of the HSN. However, they require low intracellular chloride for HSN inhibition, indicating that they inhibit excitability directly by mediating chloride influx. This mechanism of CLH-3-dependent modulation may be conserved in other neurons in which the driving force favors chloride influx.


Introduction
Members of the CLC superfamily of anion transport proteins have been identified in all phyla. Two subfamilies of CLC proteins (defined by the vertebrate CLC-3/CLC-4/CLC-5 and CLC-6/CLC-7 genes) are composed of Cl Ϫ /H ϩ exchangers involved in acidification and chloride homeostasis in intracellular organelles. Members of the third subfamily (CLC-1/CLC-2/CLC-Ka/Kb) are voltagegated chloride channels localized primarily at the plasma membrane (Zifarelli and Pusch, 2007;Jentsch, 2008;Duran et al., 2010).
CLC-2 is a broadly expressed, hyperpolarization-activated, inwardly rectifying chloride channel. Although its physiological role is not well understood, CLC-2 is highly expressed in the nervous system and has been suggested to regulate neuronal ex-citability. Some studies have suggested that CLC-2 provides an efflux pathway for chloride that is necessary to maintain synaptic inhibition mediated by the GABA A receptor (GABA A R; Madison et al., 1986;Staley, 1994;Smith et al., 1995;Rinke et al., 2010). CLC-2 channels have also been proposed to control excitability directly by providing a background conductance that mediates chloride influx and modifies excitability by affecting the input resistance of the neuron (Rinke et al., 2010;Ratté and Prescott, 2011). Understanding the precise function of CLC-2 could have important implications for understanding how neuronal excitability is controlled.
The Caenorhabditis elegans homologs of CLC-2 are encoded by clh-3 (Schriever et al., 1999;Nehrke et al., 2000;. Like CLC-2, the clh-3-encoded channel variants CLH-3a and CLH-3b are swelling-and hyperpolarization-activated inwardly rectifying chloride channels. CLH-3b is expressed in C. elegans oocytes, where it is activated during meiotic maturation and plays a role in regulating ovulation (Rutledge et al., 2001). The in vivo role of the CLH-3a variant is not known. clh-3-encoded channels are also expressed in a single pair of neurons, the hermaphrodite-specific neurons (HSNs;Schriever et al., 1999;Nehrke et al., 2000), but their function there has not been characterized.
The HSNs play a central role in the egg-laying behavior of C. elegans and as such provide a useful system to investigate the function of clh-3-encoded channels in the nervous system in vivo. The HSNs excite directly both the vulval muscles and the VC motor neurons, which also regulate egg laying (Schafer, 2005;Zhang et al., 2008), by releasing acetylcholine, serotonin and multiple neuropeptides (Desai and Horvitz, 1989;Lickteig et al., 2001;Nathoo et al., 2001;Kim and Li, 2004). It is likely that the activity of the egg-laying circuit is initiated by the HSNs, because they exhibit spontaneous, rhythmic calcium oscillations that persist even when all synaptic inputs to the HSN are severed (Zhang et al., 2008). The molecular basis for this spontaneous activity and its modulation by external cues is not understood.
Here we elucidate the role of clh-3-encoded channels in the HSNs. We identify a gain-of-function mutation in clh-3 that activates the channels by altering their voltage dependence and inhibits egg laying by suppressing HSN activity. Conversely, loss of clh-3 leads to hyperactive egg laying, indicating that clh-3encoded channels modulate the activity of the HSNs. Optogenetic and genetic epistasis experiments suggest that clh-3-encoded channels reduce HSN excitability directly by mediating chloride influx.
Plasmids and transgenic strains. For overexpression of wild-type clh-3, the clh-3 gene including the 727 bp region upstream of the start codon of clh-3a was amplified from N2 genomic DNA in two overlapping fragments. The two fragments were injected into N2 (each at 15 ng/l), along with the coinjection marker pmyo-2::gfp (at 5 ng/l) and 1 kb ladder (at 125 ng/l) to create the array ljEx479. For overexpression of mutant clh-3, the same procedure was used except that clh-3 gene was amplified from genomic DNA isolated from egl-42(n995) mutants to create the array ljEx480. As a control, only the coinjection marker pmyo-2::gfp (at 5 ng/l) and 1 kb ladder (at 125 ng/l) were injected to create the array ljEx476.
Behavioral assays. Egg-laying assays were performed essentially as described previously (Hart, 2006). For most assays, worms were picked at the early L4 stage, left at 20°C, and assayed 39 h afterward. Egg-retention and drug assays were performed in 96-well, flat-bottom plates. For eggretention assays, single worms were dissolved in 50 l of a 20% NaOCl solution in a well and the number of eggs that had been retained per worm was counted. For drug assays, single worms were allowed to lay eggs for 1 h in 50 l of M9 buffer, serotonin creatinine sulfate complex (at 7.5 mg/ml in M9 buffer), fluoxetine hydrochloride (at 0.5 mg/ml in M9), or imipramine (at 0.75 mg/ml in M9) and the number of eggs laid per worm per hour was determined. All genotypes were assayed on at least three different days. For determining egg-laying rates on and off food, worms were placed singly onto plates with or without food and left to lay eggs for an hour. The number of eggs laid per worm per hour was determined. For determining the stage of eggs laid, 30 worms were transferred to a plate and left to lay eggs for 30 min. The stages of the eggs laid were determined immediately afterward. The experiment was repeated until Ͼ100 eggs were examined per genotype. Muscimol assays were performed as described previously (Tanis et al., 2009). At least five pools of five animals were tested for each genotype on plates made with two different batches of muscimol.
Tracking assays. Worms were picked at the L4 stage, left at 22°C, and assayed 16 -18 h afterward. Tracking was performed on 35 mm, lowpeptone NGM plates that had been freshly seeded with 20 l of OP50 using Worm Tracker 2.0 (Yemini et al., 2013). The videos were captured at 5 fps and each worm was tracked for 6 consecutive hours. The frames during which egg-laying events took place were determined manually and the intervals between egg-laying events were analyzed as described previously (Waggoner et al., 1998).
Calcium imaging. HSN imaging was performed essentially as described previously (Zhang et al., 2008) using the integrated Cameleon ljIs19 [pcat-1::YC2.1] X with a few modifications. Briefly, adult worms were immobilized with 2-octyl cyanoacrylate glue (either Nexaband S/C veterinary glue or Dermabond topical skin adhesive) on an agarose pad (2% agarose in 10 mM HEPES, pH 7.1, 13 mOsm) and covered with the same HEPES buffer. A Zeiss Axioskop 2 upright microscope equipped with a Hamamatsu Orca R2 CCD camera, a Hamamatsu W-View emission image splitter, and a Uniblitz Shutter (Vincent Associates) was used to acquire the images with Metavue 7.6 (Molecular Devices). YFP and CFP fluorescence intensities were determined using the custom software Jmalyze (Shyn et al., 2003) and calcium spikes were detected and analyzed using the peak-finding software MiniAnalysis (Synaptosoft).
Channel rhodopsin experiments. Worms were transferred to retinal plates at the L4 stage, left at 20°C, and assayed 24 h later as adults. Retinal plates were made by seeding NGM plates with OP50 mixed with all-trans retinal. Then, 1000 l of OP50 culture was mixed with 4 l of a 100 mM all-trans retinal stock (in EtOH) and plates were seeded with 160 l. Control plates were made by substituting the all-trans retinal with EtOH. To stimulate the HSNs, worms expressing wzIs6 [pegl-6::ChR2::YFP] were illuminated for 5 s with 1 mW/mm 2 blue light (440 -460 nm) using LEDS (Luxeon III LXHL-PR09; Lumileds) controlled by a Mindstorms LEGO Intelligent NXT Brick.
After transfection, cells were incubated at 37°C for 24 -30 h. Approximately 2 h before patch-clamp experiments, cells were detached from growth plates by exposure to 0.25% trypsin containing 1 mM EDTA (Invitrogen) for 45 s. Detached cells were suspended in MEM, pelleted by centrifugation, resuspended in fresh MEM, and then plated onto poly-Llysine-coated coverslips. Plated coverslips were placed in a bath chamber mounted onto the stage of an inverted microscope. Cells were visualized by fluorescence and differential interference contrast (DIC) microscopy.
Patch electrodes were pulled from 1.5 mm outer diameter silanized borosilicate microhematocrit tubes; electrode resistance ranged from 4 to 8 M⍀. Currents were measured with an Axopatch 200B (Molecular Devices) patchclamp amplifier. Electrical connections to the patch-clamp amplifier were made using Ag/ AgCl wires and 3M KCl/agar bridges. Data acquisition and analysis were performed using pClamp 8 software (Molecular Devices). HEK cells were visualized continuously during patchclamp experiments by video-enhanced DIC microscopy. Cells that exhibited spontaneous swelling or shrinkage were discarded.
Quantification of CLH-3a and CLH-3b current properties. Because CLH-3a and CLH-3b channels are inwardly rectifying and active only at hyperpolarized voltages, we used current-to-voltage plots (Figs. 3B, 4 B, C) to estimate channel activation voltage, as we have described previously Falin et al., 2009). The kinetics of hyperpolarizationinduced activation of CLH-3a and CLH-3b are complex and vary with experimental conditions (Denton et al., 2005;He et al., 2006). To simplify presentation and interpretation of activation kinetics, time constants are not used. Instead, the time required for the whole-cell current to reach 50% activation when the membrane voltage is stepped from 0 mV to Ϫ100 mV for 1 s is quantified. This time is defined as the 50% rise time.
Statistical analyses. All statistical analyses were performed using GraphPad Prism, except for permutation tests, which were performed using MATLAB (MathWorks). ANOVA with Dunnett's multiple-comparison test was used to analyze the data presented in Figures 1, D-F, 2, A-C, and 8B. The Mann-Whitney test was used in Figure 5, B and C. The 2 test was used in Figures

[ c l h -3 a ] [ c l h -3 a ( n 9 9 5 ) ] [ c l h -3 b ] [ c l h -3 b ( n 9 9 5 ) ] [ c l h -3 b ( s t o p n 9 9 5 ) ]
clh-3b B Figure 1. The n995 mutation is a gain-of-function mutation in clh-3. A, Gene structure of the clh-3 locus. clh-3 encodes two isoforms; the a isoform is shown on top and the b isoform on bottom. The n995/n996 mutation is a T to C substitution resulting in a V566A substitution in CLH-3a and a V495A substitution in CLH-3b. B, Alignment of membrane embedded ␣-helices O-R of 4 CLH-3a and CLH-3b with human (h), Cyanidioschyzon merolae (Cm), Arabidopsis thaliana (At), and Escherichia coli (Ec) CLC proteins. The n995 mutation substitutes an alanine for a highly conserved valine residue (highlighted) in the P-helix. C, DIC images of young adult worms all at the same age. Arrows indicate embryos; arrowheads indicate late-stage embryos (Ͼ400 cells). clh-3(n995) mutants and wild-type animals overexpressing clh-3(n995) engage in egg laying less frequently than the wild-type, which leads to the accumulation of twice as many eggs as the wild-type and clh-3(ok763) mutants, many of which reach Ͼ400 cells before they are laid. D-F, Quantification of number of eggs retained in the uterus of adults. Bars represent the means Ϯ SEM. ***p Ͻ 0.001 compared with N2 or as indicated; nϾ36 for each genotype or as indicated. D, clh-3(n995) is a dominant mutation that leads to the accumulation of eggs. E, Overexpression of both wildtype and mutant clh-3 leads to the accumulation of eggs. F, Expression of the clh-3b isoform, but not the clh-3a isoform, in the HSN leads to the accumulation of eggs. Introducing a stop codon early in the clh-3b coding sequence abrogates the effect, indicating that it is dependent on production of the CLH-3b protein. More than 10 independent transgenic lines were examined for each construct and one representative line is shown. n ϭ Ͼ 24 for each transgenic line.
tion test with 100,000 -1,000,000 permutations was used to analyze the data presented in Figures 6 and 7A.

egl-42(n995) mutants have a mutation in the CLC channel gene clh-3
The dominant n995 mutation was identified in a screen for mutations that disrupt the function of the HSNs (Desai and Horvitz, 1989). Because the main role of the HSNs is to promote egg laying via the release of serotonin, mutants lacking the HSNs or function of the HSNs are egg-laying defective (Egl-D). Whereas wildtype worms have a steady state of ϳ14 eggs in the uterus and lay the majority eggs before they reach the 100-cell stage, HSNdefective mutants can retain two to three times as many eggs as the wild-type, some of which develop to more than the 400-cell stage before they are laid (Figs. 1, 2). In addition, HSN-defective mutants lay eggs in response to exogenous serotonin, but, unlike the wild-type, they do not lay eggs in response to drugs that prevent the reuptake of endogenously released serotonin, such as imipramine and fluoxetine ( Fig. 2A).
Using a combination of three-point mapping and whole-genome sequencing, we determined that the n995 mutants carry a mutation in clh-3 (Fig. 1A), which encodes the hyperpolarization-activated, inwardly rectifying CLC-type chloride channel splice variants CLH-3a and CLH-3b. CLC channels are homodimers composed of two identical subunits, each containing 18 ␣-helices (designated A to Dutzler et al., 2002;Dutzler et al., 2003;Feng et al., 2010). Each subunit forms an independently gated pore that is opened and closed by movements of a conserved glutamate residue (Dutzler et al., 2003). A poorly understood regulatory mechanism termed "common gating" acts cooperatively on both pores and opens or closes them simultaneously, a process thought to depend on interface conformational changes (Duffield et al., 2003;Cederholm et al., 2010;Ma et al., 2011). The n995 mutation is a T to C substitution that results in a valine to alanine substitution (V566 for CLH-3a and V495 for CLH-3b) in a highly conserved residue in the P-helix ( Fig. 1B), which forms part of the subunit interface. We sequenced the clh-3 gene from the other allele of egl-42, n996, and found that it contains the same T to C substitution as the n995 allele.
Because the n995 mutation is dominant (Desai and Horvitz, 1989; Fig. 1D), we could not perform rescue experiments. Instead, to confirm that this substitution causes the Egl-D phenotype, we overexpressed the mutant version of clh-3 in the wild-type. Like the n995 mutation, this leads to an egg-laying-defective phenotype (Fig. 1C,E). It also leads to an HSNdefective phenotype because egg laying in wild-type animals overexpressing mutant clh-3 is stimulated by exogenous serotonin but less robustly by fluoxetine or imipramine ( Fig. 2A).
Overexpressing wild-type clh-3 also leads to a similar phenotype (Fig. 1E). Together, these observations suggest that the n995 mutation is a gain-of-function ( gf), hypermorphic mutation in clh-3. Consistent with the HSN-defective phenotype, clh-3 is expressed in only one neuron, the HSN (Schriever et al., 1999;Nehrke et al., 2000; and data not shown).
To determine which clh-3 isoform may be acting in the HSN, we expressed individually cDNAs encoding either the a or the b isoform of the channel. We expressed the isoforms under the clh-3 promoter, which is specific to the HSN neuron, or the egl-6 promoter, which drives strong expression in the HSN and few , which directly stimulates the vulval muscles (VM) and promotes egg laying. In the wild-type, egg laying can be promoted by exogenous 5HT as well as imipramine (imi) and fluoxetine (fluo), which inhibit the reuptake of 5HT. In mutants lacking the HSN or HSN function, egg laying can only be promoted by exogenous 5HT (Trent et al., 1983). In clh-3(n995gf) mutants and wild-type worms overexpressing clh-3(n995gf), egg laying can be robustly stimulated by exogenous 5HT, but significantly less so by imipramine or fluoxetine, indicating a defect in HSN function. clh-3(ok763 and ok768) mutants lay more eggs in the presence of imipramine or fluoxetine than 5HT. Bars represent the means Ϯ SEM. n ϭϾ36 for each genotype and treatment. For statistical analyses, the effects of fluo and imi treatment are compared with the effect of 5HT treatment on that genotype. B, Comparison of egg-laying rates in the presence and absence of food. In the wild-type, egg laying is inhibited in the absence of food. Although clh-3(n995gf) mutants lay fewer eggs than the wild-type in the presence of food, clh-3(ok763 and ok768) mutants lay more eggs than the wild-type in the absence of food. Bars represent the means Ϯ SEM. n ϭ Ͼ 15 for each genotype on food and n ϭ Ͼ 36 off food. For statistical analyses, mutants are compared with the wild-type under the same condition. C, Comparison of egg laying in M9 buffer. In the wild-type, egg laying is inhibited by high osmolarity. Egg laying is similarly inhibited in clh-3 mutants. Bars represent the means Ϯ SEM. n ϭ Ͼ 22 for each genotype. For statistical analyses, mutants are compared with the wild-type. D, Quantification of eggs laid at early stages (at less than the nine-cell stage) and at late stages (at more than the 400-cell stage). clh-3(n995gf) mutants lay fewer eggs than the wild-type at early stages and more eggs than the wild-type at late stages; clh-3(ok763 and ok768) mutants lay more eggs than the wild-type at early stages. Bars represent the proportion Ϯ SE of the proportion. n ϭ Ͼ 100 eggs examined for each genotype. For all comparisons, *p Ͻ 0.05, **p Ͻ 0.001, and ***p Ͻ 0.001. other neurons (Ringstad and Horvitz, 2008; Fig. 1F ). We observed that only expression of the clh-3b isoform can lead to an Egl-D phenotype. Under its own promoter, expression of only the clh-3b cDNA harboring the n995 mutation confers an Egl-D phenotype, whereas under the egl-6 promoter, expression of both wild-type and mutant clh-3b confers an Egl-D phenotype. We examined Ͼ10 independent lines expressing both the wildtype and mutant versions of the clh-3a isoform under each promoter, including at higher concentrations, and none exhibited an Egl-D phenotype. These data strongly suggest that it is the CLH-3b channel that is active in the HSN.

n995gf mutation alters the properties of CLH-3a and CLH-3b and makes them hyperactive
Because the n995gf mutation appears to increase the activity of clh-3 by genetic criteria, we wanted to determine how the n995gf mutation affects the properties of the clh-3-encoded channels. To do this, we introduced the n995gf mutation into the CLH-3a and CLH-3b channel isoforms, expressed the channels in HEK293 cells, and quantified basic current properties (Figs. 3, 4). We first assessed the effect of the n995gf mutation on the CLH-3a splice variant (Fig. 3). The current-tovoltage relationships of wild-type and V566A CLH-3a were similar (Fig. 3 A, B). However, the activation voltage of the V566A channel was strongly and significantly ( p Ͻ 0.0001) depolarized compared with the wild-type (Fig. 3 B, C). The mean 50% rise time, a measure of the rate of hyperpolarization-induced channel activation, was also significantly reduced (mean Ϯ SE rise times for the wild-type and mutant channels were 5.8 Ϯ 0.3 ms, n ϭ 6, and 4.0 Ϯ 0.2 ms, n ϭ 8, respectively; p Ͻ 0.0002). These data indicate that the V566A mutant channels are more active than wild-type channels at less hyperpolarized membrane potentials.
Because the activity of CLH-3a channels is sensitive to predepolarization of the membrane, we assessed the effect of membrane depolarization on wild-type and V566A CLH-3a channels by stepping the membrane potential from 0 mV to ϩ60 mV for 1-6 s, followed by a 1 s pulse to Ϫ100 mV to activate the channels (Fig. 3 D, E). As shown previously (Denton et al., 2005), wild-type CLH-3a exhibits depolarization-induced potentiation; that is, increasing the time at which voltage is held ϩ60 mV results in greater hyperpolarization-induced channel activation at Ϫ100 mV. In contrast, membrane depolarization has little effect on the V566A mutant. Therefore, in addition to being more active at less hyperpolarized membrane potentials, the sensitivity of V566A mutant channels to membrane depolarization is strikingly decreased compared with wild-type CLH-3a channels.
We also assessed the effect of the mutation on the CLH-3b channel splice variant (Fig. 4). CLH-3b is inactivated by the Ste20 kinase GCK-3. GCK-3 reduces basal current amplitude, hyperpolarizes channel activation voltage, and slows voltagedependent channel activation Falin et al., 2009). When the V495A mutant was coexpressed with functional GCK-3, the activation voltage was hyperpolarized ϳ2-fold. However, compared with the wild-type, the effects of GCK-3 on V495A current amplitude and 50% rise time were strikingly reduced (Fig. 4). For example, GCK-3 significantly ( p Ͻ 0.001) inhibited current amplitude of wild-type CLH-3b at all test volt- WT CLH-3a Figure 3. Functional properties of wild-type and V566A CLH-3a channels. A, Representative current traces. B, Current-tovoltage relationships. *p Ͻ 0.01 compared with wild-type CLH-3a. Statistical analyses were performed only at voltages more hyperpolarized than the activation voltage of wild-type channels. C, Activation voltages. *p Ͻ 0.0001 compared with wild-type CLH-3a. D, Examples of whole-cell current traces of wild-type and V566A CLH-3a channels. Membrane potential was held at 0 mV and then stepped to ϩ60 mV for 1-6 s before currents were activated bya1shyperpolarizing step in membrane voltage to Ϫ100 mV. Cells were at 0 mV for 10 s after each hyperpolarizing voltage step. E, Time dependency of pre-depolarization at ϩ60 mV on peak wild-type and V566A current amplitude at Ϫ100 mV. Peak current amplitude was measured between 100 and 200 ms and between 850 and 950 ms after onset of the Ϫ100 mV test pulse in wild-type CLH-3a and the V566A mutant, respectively. *p Ͻ 0.01 compared with wild-type CLH-3a. Values shown in B, C, and E are means Ϯ SEM. n ϭ 5-8.
ages but had no significant ( p Ͼ 0.2) effect on V495A current amplitude at Ϫ140 and Ϫ120 mV and only small effects at other test voltages (Fig. 4 B, C). GCK-3 increased the 50% rise time of wild-type CLH-3b nearly 35-fold from 2.3 to 79.5 ms, whereas the 50% rise time of the V495A mutant was increased only ϳ6fold from 1.2 to 7.4 ms (Fig. 4E). These results demonstrate that V495A channels are active at more depolarized membrane potentials, activate more rapidly in response to membrane hyperpolarization, and are less sensitive to GCK-3-mediated inhibitory phosphorylation events. Together, these analyses show that the mutant channels are more active than wild-type channels by several criteria, consistent with the mutation behaving as a gain-of-function hypermorph in vivo. In addition, this analysis reveals that the conserved CLH-3a V495/CLH-3b V566 residue regulates multiple properties of the channel, including its voltage dependence.

clh-3(gf) mutations inhibit the HSN whereas clh-3(lf) mutations hyperactivate the HSN
The hypermorphic mutation and overexpression of clh-3 leads to a defect in egg laying and serotonin release from the HSN, suggesting that clh-3-encoded channels inhibit the HSN. To test this directly, we introduced an integrated transgene that expresses the calcium sensor Cameleon into the HSNs (ljIs19 [pcat-1::YC2.1]; Zhang et al., 2008) and imaged HSN activity (Fig. 5). In the wildtype, under the permissive conditions we used for calcium imaging, the HSNs exhibit spontaneous, rhythmic calcium oscillations (Zhang et al., 2008;Fig. 5 A, B). We found that the clh-3(n995gf ) mutants have significantly decreased HSN activity, as indicated by a decrease in the average number of calcium spikes/min. Although the majority of the clh-3(n995gf ) mutants exhibit reduced activity, some of the mutants did exhibit a wild-type level of activity throughout the recording (Fig. 5 A, B).
These data indicate that the mutant channels inhibit the rhythmic activity of the HSN upstream of calcium entry, which is consistent with a role for clh-3-encoded channels in inhibiting the HSN. We next characterized the loss-offunction phenotype of clh-3 by examining how ok768 and ok763, two deletion alleles of clh-3 (Fig. 1A), affect egg-laying behavior. These mutants were not significantly different from the wild-type in terms of the number of eggs they retained (Fig.  1C,D) nor in their egg-laying rate in the presence of abundant food (Fig. 2B). In HSN calcium-imaging experiments, the clh-3(ok763 and ok768) mutants also exhibited spontaneous, rhythmic calcium oscillations at a frequency and magnitude comparable to wild-type (Fig. 5). This was also the case when we imaged HSN activity under more restrictive conditions, such as in the presence of hypertonic concentrations of salt (data not shown). These results suggest that clh-3 may not play a major role in controlling HSN activity under most conditions. However, in more sensitive egglaying assays, we did observe that the clh-3(ok768 and ok763) mutants had an egg-laying phenotype that was opposite to that of the clh-3(n995gf ) mutants. For example, when we examined the developmental stage at which the eggs were laid (which indicates how long the eggs remain in the uterus before being laid and thus serves as a measure of egg-laying rate), we found that the loss-offunction mutants lay significantly more eggs at early stages than A, Representative current traces. B, C, Current-to-voltage relationships. *p Ͻ 0.001 and **p Ͻ 0.01 compared with channels coexpressed with KD GCK-3. Statistical analyses were performed only at voltages more hyperpolarized than the activation voltage of channels coexpressed with GCK-3. D, Activation voltages. *p Ͻ 0.001 compared with wild-type CLH-3b coexpressed with KD GCK-3; **p Ͻ 0.0001 compared with channels expressed with KD GCK-3. E, The 50% rise times, defined as the time required for whole-cell current to reach 50% activation when membrane voltage is stepped from 0 mV to Ϫ100 mV for 1 s. *p Ͻ 0.001 compared with channels coexpressed with KD GCK-3; **p Ͻ 0.00001 compared with wild-type channels coexpressed with GCK-3. Values shown in B-E are means Ϯ SE. n ϭ 4 -6.
the wild-type, whereas the clh-3(n995gf ) mutant lays significantly fewer early-stage eggs and significantly more late-stage eggs than the wild-type (Fig. 2D). Similarly, in the drug assays, the clh-3(ok768 and ok763) mutants lay significantly more eggs in response to serotonin reuptake inhibitors than to exogenous serotonin ( Fig.  2A), whereas clh-3(n995gf ) mutants lay significantly more eggs in response to serotonin than to the reuptake inhibitors. Because the wild-type is similarly stimulated by serotonin and the inhibitors, this suggests that the HSNs are more active in the clh-3(ok768 and ok763) mutants and thereby release more serotonin. Together, these data suggest that the clh-3(gf ) mutations inhibit the HSN, whereas clh-3(lf ) mutations lead to hyperactivity in the HSN, which is consistent with a role for clh-3-encoded channels in inhibiting the HSN.

CLH-3 is required in the HSNs for normal regulation of egg-laying behavior
The egg-laying behavior is modulated by a number of environmental conditions; in particular, hypertonic concentrations of salt or the absence of food strongly inhibit egg laying in wild-type animals. Like the wild-type, clh-3 mutants showed strong inhibition of egg laying in the hypertonic solution M9 (Fig. 2C). In contrast, the clh-3(ok768 and ok763) mutants lay eggs at high rates in the absence of food, whereas the clh-3(n995gf ) mutant laid eggs at low rates even in the presence of food (Fig.  2B). This suggests that clh-3 may modulate HSN activity in response to specific environmental cues, such as the presence of food. Because the HSNs control the temporal pattern of egg laying, we also examined how the clh-3 mutants affect the timing of the egg-laying behavior. In the presence of a thin bacterial lawn, worms alternate between an active egg-laying state, during which eggs are laid in clusters, and an inactive state, during which eggs are retained (Waggoner et al., 1998;Zhou et al., 1998; Fig. 6A). Both the onset of the active phase and egg laying within the active phase model as Poisson processes, with time constants in wildtype animals of ϳ20 min and 1 min, respectively. We found that in the clh-3(lf ) mutants, the intracluster time constants were significantly decreased compared with the wild-type, whereas in the clh-3(n995gf ) mutant, they were significantly increased ( Fig.  6B-D; Table 1). These results suggest that CLH-3 acts in the HSN to inhibit its activity mainly during the active phase of egg laying. Interestingly, the clh-3(n995gf ) phenotypes contrast with mutants that completely lack HSN neurons. As described previously (Waggoner et al., 1998), egl-1 mutants have dramatically increased intercluster time constants and decreased intracluster time constants (Fig. 6B; Table 1). This suggests that the clh-3(gf ) mutation alters the timing of HSN activity rather than completely inhibiting it.

clh-3 regulates HSN excitability
How might wild-type clh-3 inhibit HSN activity? Because clh-3 encodes chloride channels, we reasoned that it might affect HSN activity by affecting HSN excitability. To test this, we crossed the clh-3 mutants with an integrated transgenic line that expresses Channelrhodopsin-2 (ChR2), the blue-lightactivated cation channel (Nagel et al., 2005), in the HSNs (wzIs6 [pegl-6::ChR2]; Leifer et al., 2011;Emtage et al., 2012). In wildtype worms, egg laying is robustly stimulated by ChR2 activation (Leifer et al., 2011; Fig. 7). The magnitude of the response, as indicated by both the percentage of stimulations resulting in egglaying events and the number of eggs laid per stimulation, is dependent on both the strength and duration of the light stimulus (Fig. 7A). The response is also completely dependent on the addition of all-trans retinal, the cofactor for ChR2, to the plates (Fig. 7B), as well as the presence of the HSNs (Fig. 7D). We observed that the clh-3(n995gf ) mutant laid significantly fewer eggs per stimulation than the wild-type and blue light stimulation elicited an egg-laying event significantly less frequently in mutant animals than in wild-type. Conversely, the clh-3(ok768 and  (Zhang et al., 2008). Most clh-3(n995gf) mutants have greatly reduced HSN activity, whereas some retain wild-type levels. B, C, Quantification of calcium spike parameters. The average number of spikes/minute and average percentage ratio change of each spike was determined for each recording. Each point represents the value obtained from one recording obtained from one animal. n ϭϾ13 recordings, which corresponds to 145 spikes for clh-3(n995) and Ͼ 750 spikes for the other genotypes. Mutants were compared with the wild-type assayed at the same time. The clh-3(n995gf) mutation significantly reduces the spike frequency. None of the mutations alter spike parameters such as ratio change (as shown), area of spike, rise time, or decay time (data not shown).
ok763) mutants laid significantly more eggs than the wild-type and blue light stimulation elicited egg-laying events, including the laying of multiple eggs, more frequently than for the wildtype (Fig. 7C,D). Together, these data support a role for the clh-3-encoded channels in inhibiting HSN excitability: increased channel activity inhibits HSN excitability whereas loss of the channel promotes HSN excitability.

CLH-3-encoded channels regulate HSN excitability directly via an inward chloride current
How might clh-3-encoded channels regulate HSN excitability? One possibility is that the channels affect intracellular chloride concentrations ([Cl Ϫ ] i ) and thereby affect excitability via the GABA A R. The effect of GABA A R activation can vary from excitatory to inhibitory depending on [Cl Ϫ ] i (Farrant and Kaila, 2007); when intracellular levels of chloride are low, GABA A R activation leads to a hyperpolarizing inward chloride current that is inhibitory; when intracellular levels of chloride are high, GABA A R activation leads to chloride efflux, which can be excitatory (Fig. 8A). It has been suggested that CLC-2 channels might inhibit neuronal excitability indirectly by mediating chloride efflux necessary for GABA A Rmediated inhibition (Staley, 1994;Smith et al., 1995;Rinke et al., 2010). To determine whether the clh-3-encoded channels act in this way, we tested the mutants' response to the GABA A R agonist muscimol.
In the wild-type, muscimol inhibits the HSN, indicating that GABA A R activation normally leads to chloride influx in the HSN. When pathways of chloride efflux are lost, such as in kcc-2 and abts-1 mutants, which lack the chloride-extruding K ϩ -Cl Ϫ cotransporter (KCC2) and the Na ϩ -driven anion exchanger (NDAE), respectively, the HSNs become resistant to the inhibitory effects of muscimol (Tanis et al., 2009;Bellemer et al., 2011). In contrast to these known Cl Ϫ -effluxdefective mutants, clh-3(ok763 and ok768) mutants are as inhibited by muscimol as the wild-type (Fig. 8A). These findings indicate that clh-3 is not required for GABA A R-mediated inhibition and do not support a requirement for these channels in reducing chloride concentrations in the HSN. Therefore, the main mechanism by which clh-3-encoded channels inhibit the HSN is unlikely to be via the GABA A R. Another possibility is that the channels affect excitability directly by mediating chloride influx (Madison et al., 1986;Rinke et al., 2010;Ratté and Prescott, 2011). If this were the case, we reasoned that if we were to reverse the driving force on which the clh-3-mediated inhibition depends, the inhibitory effects of clh-3 on HSN activity would be suppressed. To test this, we turned again to the mutants of the chloride extruders kcc-2 and abts-1, which presumably have increased intracellular chloride levels, and crossed them with the clh-3(n995gf ) mutant. We found that both of these mutations completely suppressed the egg-layingdefective phenotypes of the clh-3(n995gf ) mutant (Fig. 8 B, C). Whereas the clh-3(n995gf ) mutants retain significantly more eggs Animals alternate between active periods during which eggs are laid and inactive periods during which no eggs are laid (Waggoner et al., 1998). B-F, Histograms of intervals between egg-laying events. Intervals are plotted on a natural log scale on the x-axis and relative frequencies are plotted on the y-axis. The graphs show a bimodal distribution, with the first peak corresponding to the intervals between events in the active periods (the intracluster intervals) and the second peak corresponding to the intervals between active periods (the intercluster intervals). Intervals Ͻ400 s were considered intracluster intervals and intervals Ͼ 400 s were considered intercluster intervals. Vertical lines indicate median intervals for each peak; dashed lines indicate the median values for the wild-type; solid lines indicate the median values for the mutant in each panel. The egl-1 mutation both shortens the intracluster intervals ( p ϭ 0.00048) and lengthens the intercluster intervals ( p Ͻ 0.000001). clh-3 mutations mainly affect the intracluster intervals: in clh-3(n995gf) mutants, the intervals between egg-laying events in the cluster are increased ( p ϭ 0.00142), whereas in clh-3(ok768 and ok763) mutants, the intervals are decreased ( p ϭ 0.01663 and p ϭ 0.00062). For the long intervals, the p-values for D-F are p ϭ 0.00093, p ϭ 0.0702, and p ϭ 0.29785, respectively. Sample sizes and other quantifications are shown in Table 1.
Because this measure of egg laying can be dependent on brood size, which is reduced in these mutants (data not shown), we also looked at the stages of eggs laid by the mutants, a brood-size-independent measure of egg laying. As described previously, the clh-3(n995gf ) mutant lays eggs at later stage than the wild-type (Fig. 2D); we found that the kcc-2 and abts-1 mutations completely suppress this phenotype (Fig. 8C). Therefore, increasing [Cl Ϫ ] i appears to decrease or even reverse the driving force for chloride flow through clh-3-encoded channels, indicating that the driving force normally promotes chloride influx (Fig. 8D). Consistent with this, we also found that the clh-3(n995gf ) mutants are partially resistant to low concentrations of muscimol. For example, although egg laying in the wild-type worms is already maximally (98%) inhibited by 10 M muscimol, the clh-3(n995gf) mutants are only partially (65%) inhibited at this concentration. This supports a role for clh-3 channels in chloride influx, because muscimol resistance is brought about by increasing [Cl Ϫ ] i . Together, these observations suggest that clh-3-encoded channels regulate HSN excitability directly via an outward current (chloride influx).

Discussion
Here, we have identified a role for clh-3 in the C. elegans nervous system. We have shown that the voltage-dependent chloride channels encoded by clh-3 regulate egg laying in C. elegans by controlling the excitability of the HSN neuron. Loss of the channels increases egg laying, whereas increasing CLH-3b channel activity and channel overexpression decreases HSN activity and inhibits egg laying; therefore, the normal function of the clh-3-encoded channels is to inhibit HSN function. Optogenetic experiments indicate that this occurs through an inhibition of HSN excitability.
Our results suggest a mechanism by which CLC-2/CLH-3a/b channels could affect neuronal excitability. Many previous studies have favored the possibility of an indirect role by which CLC-2 provides a chloride efflux pathway that helps maintain the chlo-ride gradient required for GABA A -receptor-mediated inhibition (Madison et al., 1986;Staley, 1994;Smith et al., 1995;Rinke et al., 2010). However, we have shown clearly here that clh-3 is not required for GABA A R-mediated inhibition of the HSN. Although the HSN is inhibited by the GABA A R agonist muscimol, and this effect requires both the GABA A receptor and the chloride extruders KCC-2 and ABTS-1 (Bellemer et al., 2011), it does not require clh-3-encoded channels (Fig. 8A). Therefore, the CLH-3 channels are not necessary for chloride efflux from the HSN and their effect on HSN excitability is unlikely to be via a modulation of the response to GABA A R activation. Channel rhodopsin experiments reveal that CLH-3 inhibits HSN excitability. A, The egg-laying response in the wildtype is dependent on both the intensity of the light stimulus and the duration of the light stimulus. Bars represent means Ϯ SEM. B, The response to 5 s, 1 mW/mm 2 blue light stimulation is dependent on the presence all-trans retinal, the cofactor for CHR2. Bars represent means Ϯ SEM. More than nine stimulations were tested for each condition and genotype. C, Histogram of eggs laid per 5 s, 1 mW/mm 2 blue light stimulation. Number of eggs laid per stimulation is plotted on the x-axis and the relative frequency is plotted on the y-axis. In clh-3(n995gf) mutants, stimulations more frequently do not lead to egg-laying events, whereas in clh-3(ok768 and ok763) mutants, stimulations more frequently do lead to egg-laying events, including the laying of more than one egg. n ϭ Ͼ 42 stimulations/genotype. D, Quantification of number of eggs laid/stimulation in the presence of all-trans retinal. Bars represent means Ϯ SEM. n ϭ 18 stimulations for egl-1 and Ͼ 42 stimulations for the other genotypes. The response to blue light stimulation requires the HSN because egl-1 mutants do not respond. clh-3(n995gf) mutants lay fewer eggs per stimulation, whereas clh-3(ok768 and ok763) mutants lay more eggs. **p Ͻ 0.01; ***p Ͻ 0.001. Analysis of the temporal pattern of egg-laying events obtained from tracking data. The number of hours of video analyzed (hours obs.), number of egg-laying events observed (events obs.), and number of total eggs laid per hour are indicated. As described in Waggoner et al. (1998), p is the probability that, after a given egg-laying event, another egg will be laid before the animal enters the inactive state. 1 and 2 are the rate constants for the active and inactive egg-laying states, respectively. 1/1 represents the intracluster time constant (in seconds) and 1/p2 represents the intercluster time constant (in seconds). . CLH-3 is not required for chloride efflux and mediates chloride influx. A, The effect of GABA A R activation can vary from inhibitory to excitatory. When intracellular levels of chloride are low, GABA A R activation leads to chloride influx, which inhibits via hyperpolarization or shunting inhibition; when intracellular levels of chloride are high, GABA A R activation leads to chloride efflux, which can becomeexcitatory.InC.elegans,theGABA A RagonistmuscimolcanbeusedtodetermineifintracellularchloridelevelsarealteredintheHSN (Tanis et al., 2009). In the wild-type, muscimol inhibits the HSN via activation of the GABA A R UNC-49. Mutants defective in chloride extrusion have elevated intracellular chloride levels and are resistant to the inhibitory effects of muscimol. In contrast to the chlorideextrudingtransporterskcc-2andabts-1,clh-3mutantsarenotresistanttomuscimol,indicatingthattheyarenotrequiredforchlorideefflux from the HSN. Bars represent the means Ϯ SEM of Ͼ 5 experiments per genotype. B, C, Genetic interactions between clh-3(n995) and chloride-extruding transporters kcc-2 and abts-1. B, kcc-2 and abts-1 mutants retain fewer eggs than the wild-type. These mutations are epistatictoclh-3(n995),whichleadstoanaccumulationofmoreeggsthanthewild-type.BarsrepresentmeansϮSEM.nϭϾ36foreach genotype. For statistical analyses, single mutants are compared with wild-type and the double mutants are compared with clh-3(n995) unlessotherwiseindicated.C,Quantificationofeggslaidatearlystages(atlessthanthenine-cellstage)andatlatestages(atmorethanthe 400-cell stage). kcc-2 and abts-1 mutations completely suppress the egg-laying phenotypes of clh-3(n995) mutants such that the double mutants are now indistinguishable from the single mutants, with the exception of abts-1; clh-3, which lays slightly more late stage eggs thanabts-1.BarsrepresenttheproportionϮSE.nϭϾ120eggsexaminedforeachgenotype.Forstatisticalanalyses,thesinglemutants are compared with wild-type (for all comparisons, the single mutants were indistinguishable from wild-type) and the double mutants are compared with clh-3(n995) unless otherwise indicated. For all comparisons, *p Ͻ 0.05 and ***p Ͻ 0.001. D, Model for the role of CLH-3 channels. CLH-3 normally participates in a background conductance that mediates chloride influx in the HSN. In the clh-3(lf) mutants, this conductance is absent, which makes the HSNs more excitable and leads to a hyperactive egg-laying phenotype. In the clh-3(gf) mutants, the activity of the CLH-3 channel is altered in such a way that it leads to an increase in the chloride current through the channel, which inhibits the excitability of the HSN and leads to an egg-laying-defective phenotype. Mutation of the chloride extruders kcc-2 and abts-1 increases [Cl] i , reverses the driving force for chloride flow through the channel, reverses the effects of increasing the activity of the channel, and thereby restores egg laying. This indicates that the driving force normally promotes chloride influx through the CLH-3 channels.

N 2 c l h -3 k c c -2 c l h -3 ; k c c -2 a b t s -1 a b t s -1 ; c l h -
pattern of egg laying within the active phase and the laying of early-stage eggs. clh-3 is also required for the modulation of egg laying by food. Unlike wild-type animals, which strongly reduce their rate of egg laying in the absence of food, clh-3 loss-of-function mutants lay eggs at high rates both on and off food (Fig. 2B). This suggests that these channels may be required to modulate HSN activity in specific contexts, such as in response to food. Consistent with this idea, a pathway involving the CO 2 -sensing BAG neurons, neuropeptides encoded by flp-17, a G o -coupled neuropeptide receptor (EGL-6), and an inwardly rectifying potassium channel (IRK-1; Ringstad and Horvitz, 2008;Emtage et al., 2012) also negatively regulates egg laying through the HSN. However, this pathway affects the latency of response to ChR2-mediated activation of the HSN (Emtage et al., 2012), whereas the clh-3-encoded channels affect the magnitude (and not the latency) of the response (Fig. 7 and data not shown). Possibly, the EGL-6/IRK-1 pathway and the clh-3 pathway are required to modulate different properties of the HSN or its activity in response to different cues.
One way the clh-3 pathway could be activated under particular conditions is by phosphorylation of the channels. In oocytes, the activity of CLH-3b is strongly inhibited by phosphorylation by the GCK-3 kinase (Denton et al., 2005) and activated by the GLC-7␣/␤ serine/threonine phosphatases (Rutledge et al., 2002). The activity of CLC-2 is also regulated by phosphorylation (Rutledge et al., 2002) and by PKC (Madison et al., 1986). In the future, identifying the signaling pathways upstream of clh-3 in the HSN might provide a window into the contexts in which this family of channels acts to modulate neural activity.
Why are the HSNs the only neurons in C. elegans that express clh-3? The HSNs are unique among C. elegans neurons in that they exhibit spontaneous rhythmic activity that is generated cell autonomously (Zhang et al., 2008). Although CLH-3 channels are not necessary for this rhythmic activity, strong activation of the channel can suppress it (as in the n995gf mutant). It is possible that these channels contribute to the properties of the neuron that allow for spontaneous rhythmicity, for example, by modulating the membrane conductance. In the future, it would be interesting to investigate whether these channels also modulate spontaneous neural activity in other organisms.