Abstract
The heterotrimeric G protein Gq positively regulates neuronal activity and synaptic transmission. Previously, the Rho guanine nucleotide exchange factor Trio was identified as a direct effector of Gq that acts in parallel to the canonical Gq effector phospholipase C. Here we examine how Trio and Rho act to stimulate neuronal activity downstream of Gq in the nematode Caenorhabditis elegans. Through two forward genetic screens, we identify the cation channels NCA-1 and NCA-2, orthologs of mammalian NALCN, as downstream targets of the Gq/Rho pathway. By performing genetic epistasis analysis using dominant activating mutations and recessive loss-of-function mutations in the members of this pathway, we show that NCA-1 and NCA-2 act downstream of Gq in a linear pathway. Through cell-specific rescue experiments, we show that function of these channels in head acetylcholine neurons is sufficient for normal locomotion in C. elegans. Our results suggest that NCA-1 and NCA-2 are physiologically relevant targets of neuronal Gq-Rho signaling in C. elegans.
Introduction
Heterotrimeric G protein pathways play central roles in altering neuronal activity and synaptic transmission in response to experience or changes in the environment. Gq is one of the four types of heterotrimeric G proteins alpha subunits in animals (Wilkie et al. 1992) and is widely expressed in the mammalian brain (Wilkie et al. 1991) where it typically acts to stimulate neuronal activity and synaptic transmission (Krause et al. 2002; Gamper et al. 2004; Coulon et al. 2010). These roles are conserved in the nematode C. elegans. Unlike mammals which have four members of the Gq family, C. elegans has only a single Gqα (Brundage et al. 1996). In C. elegans, loss-of-function and gain-of-function mutants in the single Gqα gene egl-30 are viable but have strong neuronal phenotypes, affecting locomotion, egg-laying, and sensory behaviors (Brundage et al. 1996; Lackner et al. 1999; Bastiani et al. 2003; Matsuki et al. 2006; Esposito et al. 2010; Adachi et al. 2010). We aim to identify the signal transduction pathways and downstream targets by which Gq signaling alters neuronal activity.
In the canonical Gq pathway, Gq activates phospholipase Cβ (PLC) to cleave the lipid phosphatidylinositol 4,5,-bisphosphate (PIP2) into diacylglycerol (DAG) and inositol trisphosphate (IP3), each of which can act as a second messenger. This pathway operates in both worms and mammals, but in both systems, a number of PLC-independent effects of Gq have been described, indicating the existence of additional Gq signal transduction pathways (Lackner et al. 1999; Miller et al. 1999; Vogt et al. 2003; Bastiani et al. 2003; Sánchez-Fernández et al. 2014). Using a genetic screen for suppressors of activated Gq, we identified the Rho guanine nucleotide exchange factor (GEF) Trio as a direct effector of Gq in a second major Gq signal transduction pathway independent of the canonical PLC pathway (Williams et al. 2007). Biochemical and structural studies demonstrated that Gq directly binds and activates RhoGEF proteins in both worms and mammals, indicating that this new Gq pathway is conserved (Lutz et al. 2005, 2007; Williams et al. 2007).
Gq activation of the Trio RhoGEF leads to activation of the small GTP-binding protein Rho, a major cellular switch that affects a number of cellular processes, ranging from regulation of the cytoskeleton to transcription (Etienne-Manneville and Hall 2002; Jaffe and Hall 2005). In C. elegans neurons, Rho has been shown to regulate synaptic transmission downstream of the G12-class G protein GPA-12 via at least two pathways, one dependent on the diacylglycerol kinase DGK-1 and one independent of DGK-1 (McMullan et al. 2006; Hiley et al. 2006). Here we investigate what targets operate downstream of Rho in the Gq signaling pathway to regulate neuronal activity. Through two forward genetic screens, we identify the cation channels NCA-1 and NCA-2 (NALCN in mammals) as downstream targets of the Gq-Rho pathway. The NALCN channel is a relative of voltage-gated cation channels that has been suggested to be a sodium leak channel required for the propagation of neuronal excitation and the fidelity of synaptic transmission (Lu et al. 2007; Jospin et al. 2007; Yeh et al. 2008). However, there is controversy over the current carried by NALCN and whether it is indeed a sodium leak channel (Senatore et al. 2013; Senatore and Spafford 2013; Boone et al. 2014). It is also unclear how NALCN is gated and what pathways activate the channel. Two studies have shown that NALCN-dependent currents can be activated by G protein-coupled receptors, albeit independently of G proteins (Lu et al. 2009; Swayne et al. 2009), and another study showed that the NALCN leak current can be activated by low extracellular calcium via a G protein-dependent pathway (Lu et al. 2010). Our data presented here suggest that the worm NALCN orthologs NCA-1 and NCA-2 are activated by Rho acting downstream of Gq in a linear pathway.
Materials and Methods
Strains
Worm strains were cultured and maintained using standard methods (Brenner, 1974). A complete list of strains and mutations used is provided in the strain list (Table S1).
Isolation of suppressors of activated Gq
We performed an ENU mutagenesis to screen for suppressors of the hyperactive locomotion of an activated Gq mutant, egl-30(tg26) (Ailion et al. 2014). From approximately 47,000 mutagenized haploid genomes, we isolated 10 mutants that had a fainter phenotype when outcrossed away from the egl-30(tg26) mutation. By mapping and complementation testing, we assigned these ten mutants to three genes: three unc-79 mutants (yak37, yak61, and yak73), six unc-80 mutants (ox329, ox330, yak8, yak35, yak36, and yak56) and one nlf-1 mutant (ox327). Complementation tests of ox329 and ox330 were performed by crossing heterozygous mutant males to unc-79(e1068) and unc-13(n2813) unc-80(ox301) hermaphrodites. Complementation tests of yak alleles were performed by crossing heterozygous mutant males (m/+) to unc-79(e1068) and unc-80(ox330) mutant hermaphrodites. For all crosses, we assessed the fainting phenotype of at least five male cross-progeny by touching animals on the head and scoring whether animals fainted within five seconds. For the crosses to unc-13 unc-80, we also scored the fainting phenotype of at least ten Non-Unc-13 hermaphrodites similarly. Control crosses with wild-type males demonstrated that unc-79/+ and unc-80/+ heterozygous males and hermaphrodites are phenotypically wild-type and do not show fainting behavior, demonstrating that these mutants are fully recessive.
Isolation of suppressors of activated Go
We first isolated suppressors of the activated Go mutant unc-109(n499) by building double mutants of unc-109(n499) with the activated Gq allele egl-30(tg26). Unlike unc-109(n499) homozygotes which are lethal, egl-30(tg26) unc-109(n499) homozygotes are viable, but paralyzed and sterile, indicating that activated Gq partially suppresses activated Go, consistent with the model that these two G proteins act antagonistically. We built a balanced heterozygote strain egl-30(tg26) unc-109(n499)/egl-30(tg26) unc-13(e51) gld-1(q126) which has an “unmotivated” phenotype in which the worms move infrequently and slowly. We mutagenized these animals with ENU and screened for F1 progeny that moved better. From a screen of approximately 16,000 mutagenized haploid genomes, we isolated two apparent unc-109 intragenic mutants, ox303 and ox304. ox303 is a strong unc-109 loss-of-function allele, as evidenced by the fact that egl-30(tg26) unc-109(n499 ox303)/egl-30(tg26) unc-13(e51) gld-1(q126) mutants resembled egl-30(tg26) mutants (i.e. hyperactive). Additionally, unc-109(n499 ox303) mutants themselves are hyperactive in an otherwise wild-type background. ox304, however, appears to be a partial loss-of-function mutant, because the egl-30(tg26) unc-109(n499 ox304)/egl-30(tg26) unc-13(e51) gld-1(q126) mutant moves better than the egl-30(tg26) unc-109(n499)/egl-30(tg26) unc-13(e51) gld-1(q126) parent strain, but is not hyperactive like the egl-30(tg26) strain. Also, unc-109(n499 ox304) homozygote animals on their own are viable but are almost paralyzed and show very little spontaneous movement, with a typical straight posture. However, when stimulated by transfer to a new plate, the unc-109(n499 ox304) mutant is surprisingly capable of coordinated movements. This strain was used for mapping and sequencing experiments that demonstrated that unc-109 is allelic to goa-1, encoding the worm Go ortholog (see below). The unc-109(n499 ox304) strain was also used as the starting point for a second screen to isolate extragenic suppressors of activated Go.
Previously, a screen for suppressors of activated goa-1 was performed using heat-shock induced expression of an activated goa-1 transgene (Hajdu-Cronin et al. 1999). This screen isolated many alleles of dgk-1, encoding diacyclglycerol kinase, but only a single allele of eat-16, encoding a regulator of G protein signaling (RGS) protein that negatively regulates Gq (Hajdu-Cronin et al. 1999), along with mutants in three genes needed for expression of the heat-shock goa-1 transgene (Hajdu-Cronin et al. 2004). Because of the strong bias of this screen for isolating alleles of dgk-1, we used the goa-1(n499 ox304) strain to perform a screen for suppressors of activated Go that did not involve overexpression of goa-1 or rely on expression of a heat-shocked induced goa-1 transgene. We performed ENU mutagenesis of goa-1(n499 ox304) and isolated F2 animals that were not paralyzed. From a screen of approximately 24,000 mutagenized haploid genomes, we isolated 17 suppressors, 9 with a relatively stronger suppression phenotype and 8 that were weaker. Of the 9 stronger suppressors, we isolated two alleles of eat-16 (ox359, ox360), three alleles of the BK type potassium channel slo-1 (ox357, ox358, ox368), one allele of the gap junction innexin unc-9 (ox353), one gain-of-function allele in the ion channel gene nca-1 (ox352), and two mutants that were mapped to chromosomal regions distinct from the other mutations listed above, but not further characterized (ox356, ox364). The Go suppressors we isolated are consistent with the established model that activation of GOA-1 activates the RGS EAT-16, which inhibits Gq EGL-30 (Hajdu-Cronin et al. 1999). Reduced Gq activity leads to reduced activation of the NCA cation channels, which can be reversed by an activating mutation in NCA-1. Previously, it has been shown that lack of the depolarizing NCA cation currents can be suppressed by a compensatory loss of the hyperpolarizing potassium current from SLO-1 (Kasap et al. 2017), or by loss of the gap junction proteins UNC-9 or UNC-7 (Sedensky and Meneely 1987; Morgan and Sedensky 1995; Bouhours et al. 2011). This is consistent with our isolation of slo-1 and unc-9 mutants as suppressors of activated Go, since activation of Go also leads to reduced NCA activity.
Mapping and cloning nlf-1(ox327)
We mapped the ox327 mutation using single nucleotide polymorphisms (SNPs) in the Hawaiian strain CB4856 as described (Davis et al., 2005). The ox327 mutation was mapped to an approximately 459 kb region on the left arm of the X chromosome between SNPs on cosmids F39H12 and C52B11 (SNPs F39H12[4] and pkP6101). This region included 74 predicted protein-coding genes. We injected cosmids spanning this region and found that injection of cosmid F55A4 rescued the ox327 mutant phenotype. We performed RNAi to the genes on this cosmid in the eri-1(mg366) lin-15(n744) strain that has enhanced RNAi and found that RNAi of the gene F55A4.2 caused a weak fainter phenotype. We sequenced F55A4.2 in the ox327 mutant and found a T to A transversion mutation in exon 1, leading to a premature stop codon at amino acid C59. We also rescued the ox327 mutant with a transgene carrying only F55A4.2, confirming the gene identification. We subsequently obtained a deletion allele tm3631 that has fainter and Gq suppression phenotypes indistinguishable from ox327. F55A4.2 was given the gene name nlf-1 (Xie et al. 2013).
We obtained six independent nlf-1 cDNAs that were predicted to be full-length: yk1105g4, yk1159a2, yk1188d11, yk1279a1, yk1521f8, and yk1709b10. Restriction digests suggested that all six were of the same size. We sequenced yk1159a2 and yk1279a1 and both gave the same nlf-1 exon-intron structure, which differed from the gene-structure on Wormbase WS253 in several ways: nlf-1 is 4 bp shorter at the 3’ end of exon 5, and has a new 154 bp exon (now exon 6) not predicted on Wormbase. nlf-1 consists of 8 exons and is predicted to encode a protein of 438 amino acids (Figure 3A). This is identical to the gene structure reported independently (Xie et al. 2013). Both sequenced cDNAs had 5’UTRs of 64 bp, with yk1159a2 (but not yk1279a1) being trans-spliced to the SL1 splice leader, and 3’UTRs of 424 bp (yk1279a1) or 429 bp (yk1159a2). The yk1279a1 cDNA was mutation-free and was cloned into a Gateway entry vector for use in rescue experiments. The full-length sequence of the yk1279a1 nlf-1 cDNA was deposited in GenBank under accession # KX808524.
Mapping and cloning unc-109(n499)
unc-109 was shown to be allelic to goa-1. First, we performed SNP mapping of both unc-109(n499) and its intragenic revertant unc-109(n499 ox303), using the Hawaiian strain CB4856 as described (Davis et al. 2005). These experiments mapped unc-109 to an approximately 1 Mb region in the middle of chromosome I between SNPs on cosmids D2092 and T24B1 (SNPs CE1-15 and T24B1[1]). A good candidate in this region was goa-1. We confirmed that unc-109 was indeed goa-1 by sequencing the three unc-109 mutants: the gain-of-function allele n499 carries a point mutation that leads to an R179C missense mutation, affecting a conserved arginine residue shown to be important for the GTPase activity of G proteins (Coleman et al. 1994); the partial loss-of-function allele ox304 carries a point mutation leading to a W259R missense mutation; and the strong loss-of-function allele ox303 carries a one basepair deletion that leads to a premature stop 32 amino acids from the C-terminal.
Molecular biology and transgenes
A complete list of constructs is provided in the plasmid list (Table S2). Most of the constructs were made using the three slot multisite Gateway system (Invitrogen). For C3 transferase constructs, a promoter, an FRT-mCherry-FRT-GFP cassette (pWD178), and the C. botulinum C3 transferase-unc-54 3’UTR (cloned into a Gateway entry vector from plasmid QT#99) were combined into the pDEST R4-R3 destination vector. For nlf-1 tissue-specific rescue constructs, a promoter, the nlf-1 coding sequence (genomic DNA or cDNA), and a C-terminal GFP tag were cloned along with the unc-54 3’UTR into the pDEST R4-R3 destination vector. Promoters used were nlf-1p (5.7 kb upstream of the ATG), rab-3p (all neurons), unc-17p (acetylcholine neurons), unc-17Hp (head acetylcholine neurons) (Hammarlund et al. 2007), acr-2p (acetylcholine motor neurons), unc-17βp (acetylcholine motor neurons) (Charlie et al. 2006), and glr-1p (glutamate-receptor interneurons). Extrachromosomal arrays were made by standard transformation methods (Mello et al. 1991). Constructs of interest were injected at 10 ng/μl with marker and carrier DNAs added to make a final total concentration of at least 100 ng/μl. For most constructs, we isolated multiple independent insertions that behaved similarly. C3 transferase extrachromosomal arrays were integrated into the genome using X-ray irradiation (4000 rads). Integrated transgenes were mapped to chromosomes and outcrossed twice before further analysis.
Locomotion assays
We performed two different assays to measure locomotion. Body bend assays measured the rate of locomotion. Radial locomotion assays measured the radial distance animals moved from a point in a given unit of time, which provides a combined measurement of different aspects of the locomotion phenotype including the rate of locomotion, waveform, and frequency of reversals. Both types of assays were performed on 10 cm plates seeded with thin lawns of OP50 bacteria. These plates were prepared by seeding 1.5 ml of stationary phase OP50 bacteria to evenly cover the entire plate surface, and then growing the bacteria for two days at room temperature. Plates were stored at 4° for up to one month before being used. For body bend assays, first-day adult worms were picked to an assay plate, allowed to rest for 30 seconds, and then body bends were counted for one minute. A body bend was defined as the movement of the worm from maximum to minimum amplitude of the sine wave (Miller et al. 1999). To minimize variation, all animals in a body bend experiment were assayed on the same plate. For radial locomotion assays, five to eight first-day adults were picked together to the center of a plate to begin the assay (time 0). Positions of the worms were marked on the lid of the plate every ten minutes for up to forty minutes. Following the assay, the distance of each point to the center was measured. For most strains, radial distances did not increase after the first ten minutes of the assay and all data presented here are for the ten-minute time point. Analysis of the data at later time points leads to the same conclusions. For all locomotion assays, the experimenter was blind to the genotypes of the strains being assayed.
For Rho inhibition experiments (Figure 1), expression of C3 transferase (C3T) was induced by FLP-mediated recombination. Expression of FLP was induced by heat shock for 1 hr at 34°C, plates were returned to room temperature, and animals were scored for locomotion 4 hrs after the end of the heat shock period. For heat-shock induction of activated Rho (Figure 8), worms were heat shocked for 1 hr at 34°C, returned to room temperature, and scored for locomotion 2 hrs after the end of the heat-shock period.
Fainting assays
Backward fainting times were measured by touching a worm on the head with a worm pick to stimulate movement, and measuring the time to faint with a stopwatch. Forward fainting time was measured following a touch on the tail. Fainting was defined by an abrupt stop of movement along with a characteristic straightening of the head (Figure 6A). Alternatively, we touched worms on the head or tail with a pick and counted the number of body bends until the worm fainted. If a worm moved 10 body bends without fainting, we stopped the assay.
Waveform quantification
To quantify the track waveform, first-day adult animals were placed on an OP50 plate and allowed to move forward for a few seconds. We then imaged each animal’s tracks using a Nikon SMZ18 microscope with the DS-L3 camera control system. Track pictures were taken at 40X and were processed using ImageJ. Period and 2X amplitude were measured using the line tool. For each worm, five period/amplitude ratios were averaged. Five individual worms were used per experiment.
The exaggerated waveform of egl-30(tg26) mutants is also characterized by the head of the worm occasionally crossing over the body, leading the worm to form a figure-eight shape (Figure 2A). This phenotype was quantified by placing animals on OP50 plates and allowing them to move forward, counting the number of times the head crossed over the body in one minute.
Imaging and image analysis
Worms were mounted on 2% agarose pads and anesthetized with sodium azide. Images were obtained using a Zeiss Pascal confocal microscope. For quantitative imaging of NCA-1::GFP and NCA-2::GFP (Figure S1), Z-stack projections of the nerve ring axons on one side of the animal were collected and quantified in ImageJ as described (Jospin et al. 2007). Dissecting microscope photographs of first-day adult worms were taken at 50X using a Nikon SMZ18 microscope equipped with a DS-L3 camera control system.
Statistics
P values were determined using GraphPad Prism 5.0d (GraphPad Software). Normally distributed data sets with multiple comparisons were analyzed by a one-way ANOVA followed by a Bonferroni or Tukey posthoc test to examine selected comparisons or by Dunnett’s test if all comparisons were to the wild type control. Non-normally distributed data sets with multiple comparisons were analyzed by a Kruskal-Wallis nonparametric ANOVA followed by Dunn’s test to examine selected comparisons. Pairwise data comparisons were analyzed by a two-tailed unpaired t test for normally distributed data or by a two-tailed Mann-Whitney test for non-normally distributed data.
Reagent and data availability
Strains and plasmids are shown in Table S1 and Table S2 and are available from the Caenorhabditis Genetics Center (CGC) or upon request. The full-length sequence of the yk1279a1 nlf-1 cDNA was deposited in GenBank under accession # KX808524. The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article and Supplemental Material.
Results
Inhibition of Rho suppresses activated Gq
Two pieces of data suggested that Gq may regulate locomotion in C. elegans through activation of the small G protein Rho. First, both the hyperactive locomotion and tightly wound body posture of the activated Gq mutant egl-30(tg26) are suppressed by loss-of-function mutations in the unc-73 RhoGEF Trio, an activator of Rho (Williams et al. 2007). Second, expression of activated Rho as a transgene causes worms to adopt a posture characterized by a tightly coiled, high-amplitude waveform (McMullan et al. 2006), reminiscent of the waveform of activated Gq worms (Bastiani et al. 2003; Ailion et al. 2014). To determine directly whether Gq signals through Rho, we tested whether Rho inhibition suppresses an activated Gq mutant.
In C. elegans, there is a single gene encoding Rho (rho-1). Loss of rho-1 causes numerous pleiotropic developmental phenotypes and embryonic lethality (Jantsch-Plunger et al. 2000), making it difficult to study Rho function using classical loss-of-function approaches. As an alternative method of inactivating Rho, we employed the Clostridium botulinum C3 transferase (C3T) which ADP-ribosylates Rho and has been widely used as a Rho inhibitor (Aktories et al. 2004). C3T has strong substrate specificity for Rho and has only weak activity towards other Rho family members like Rac and Cdc42 (Just et al. 1992). Though effects of C3T on targets other than Rho cannot be absolutely excluded, a study in C. elegans found that C3T, rho-1 RNAi, and expression of a dominant-negative form of Rho all caused similar defects in P cell migration (Spencer et al. 2001). To bypass the developmental roles and study Rho function in the adult nervous system, we expressed the C3T only in adult neurons by using the FLP recombinase/FRT system (Davis et al. 2008). In this system, temporal control of C3T is achieved through induction of FLP via heat-shock. FLP in turn promotes recombination between FRT sites to lead to expression of C3 transferase in specific neurons (Figure 1A). We expressed C3T in the following classes of neurons: all neurons (rab-3p), acetylcholine neurons (unc-17p), head acetylcholine neurons (unc-17Hp), and acetylcholine motor neurons (unc-17βp). mCherry fluorescence confirmed expression in the expected neurons and GFP expression was used to monitor induction of C3T following FLP-mediated recombination.
Inhibition of rho-1 in adult neurons caused a decreased locomotion rate (Figure 1B). This effect was greatest when Rho was inhibited in all neurons, but Rho inhibition in acetylcholine subclasses of neurons also led to slower locomotion. Thus, Rho acts in multiple classes of neurons to promote locomotion in adult worms. In the absence of heat-shock, all strains showed normal wild-type rates of locomotion and did not express any detectable GFP, indicating that these transgenes do not provide leaky expression of C3T in the absence of heat-shock. This confirms that Rho acts post-developmentally in mature neurons to regulate locomotion behavior (McMullan et al. 2006).
To determine whether Rho acts downstream of Gq signaling, we crossed the C3 transgenes into the background of the activated Gq mutant egl-30(tg26). Inhibition of rho-1 in all adult neurons strongly suppressed the high-amplitude waveform, body posture, and hyperactive locomotion of the activated Gq mutant (Figure 1, C-F). Inhibition of rho-1 in acetylcholine neurons suppressed the hyperactivity of activated Gq (Figure 1C), but only weakly suppressed the high-amplitude waveform (Figure 1, D-F). Thus, rho-1 exhibits genetic interactions consistent with a role in the Gq signaling pathway in both acetylcholine neurons and additional neurons.
Mutations in NCA channel subunits suppress activated Gq
What acts downstream of Rho in the Gq signal transduction pathway? We screened for suppressors of activated Gq and found mutants in three categories: (1) the canonical Gq pathway (such as the PLC egl-8); (2) the RhoGEF unc-73 (Williams et al. 2007), and (3) genes that affect dense-core vesicle function (e.g. unc-31, rab-2, rund-1) (Ailion et al. 2014; Topalidou et al. 2016). Mutations in unc-73 strongly suppress the high-amplitude waveform of the activated Gq mutant (Figure 2, A and C), but mutations in the canonical Gq pathway or in genes that affect dense-core vesicle function only weakly suppress the high-amplitude waveform of the activated Gq mutant (Figure 2). Thus, strong suppression of the high-amplitude waveform of activated Gq may be a specific characteristic of mutations in the Rho pathway, and we hypothesized that downstream targets of Rho in this pathway would also suppress the high-amplitude waveform.
To identify possible downstream targets of Rho in the Gq pathway, we examined other mutants isolated from our screen for suppression of the high-amplitude waveform of activated Gq (Figure 2). When crossed away from the activating Gq mutation, several of these mutants had a “fainter” phenotype. Fainter mutants respond to a touch stimulus by moving away, but abruptly stop, that is “faint”, after only a few body bends. The fainter phenotype has been observed only in mutants that reduce the function of the NCA-1 and NCA-2 ion channels (Humphrey et al. 2007; Jospin et al. 2007; Yeh et al. 2008). We found that all our Gq suppressors that strongly suppressed the high-amplitude waveform and had a fainter phenotype were mutants in either unc-79 or unc-80, two genes required for function of the NCA channels (Humphrey et al. 2007; Jospin et al. 2007; Yeh et al. 2008). We also isolated a single mutant in the gene nlf-1 that also gave fainters after outcrossing away from the activated Gq mutation, but did not strongly suppress the high-amplitude waveform of the activated Gq mutant (Figure 2). unc-79 and unc-80 mutants have a strong fainter phenotype equivalent to that of a double mutant in nca-1 and nca-2, two genes that encode pore-forming subunits of the NCA channels in C. elegans (Humphrey et al. 2007; Jospin et al. 2007; Yeh et al. 2008). Like unc-79 or unc-80, an nca-1 nca-2 double mutant suppressed the activated Gq mutant. Additionally, nca-1 on its own partially suppressed activated Gq, but nca-2 did not (Figure 2). This suggests that although nca-1 and nca-2 are only redundantly required for normal worm locomotion, channels containing the NCA-1 pore-forming subunit have a larger role in transducing Gq signals than NCA-2 channels.
Cloning and characterization of nlf-1(ox327)
In addition to the previously known NCA channel subunits unc-79 and unc-80, we also isolated the ox327 mutant in a gene that had not been previously characterized at the time of our study. We cloned ox327 by single nucleotide polymorphism (SNP) mapping, RNAi and transgenic rescue experiments (see Materials and Methods), showing that it carries an early stop mutation in the gene nlf-1. We sequenced two nlf-1 cDNAs and found that its exon-intron structure differed from the gene structure predicted on Wormbase (Figure 3A, see Materials and Methods for details). nlf-1 was independently cloned by others (Xie et al. 2013).
nlf-1 encodes an endoplasmic reticulum-localized protein probably involved in proper assembly of the NCA channel, since nlf-1 mutants had reduced expression levels of GFP-tagged NCA-1 and NCA-2 (Xie et al. 2013). We also found that nlf-1 is required for normal axonal levels of both GFP-tagged NCA-1 and NCA-2 in the nerve ring (Figure S1). Additionally, an nlf-1 mutation suppressed both the coiled posture and slow locomotion of an activated nca-1 mutant (Figure 3, B-F), demonstrating that nlf-1 is important for NCA-1 function.
nlf-1 mutants have a weaker fainter phenotype than mutants of unc-79 or unc-80, or the nca-1 nca-2 double mutant (Figure 4, A and B). The nlf-1 fainting phenotype differs in two ways from those of the stronger fainting mutants. First, nlf-1 mutants take a longer time to faint following stimulation (Figure 4, A and B). Second, while the strong fainting mutants show a similarly strong fainting phenotype in either the forward or backward direction, nlf-1 mutants faint reliably in the backward direction but take much longer and have more variable fainting in the forward direction (Figure 4, A and B), suggesting that nlf-1 mutations cause a partial loss of function of the NCA channels. To determine how nlf-1 interacts with NCA mutants, we built nlf-1(ox327) double mutants with unc-79, unc-80, nca-1, and nca-2. The nlf-1 mutation did not enhance the unc-79 or unc-80 fainter phenotype, suggesting that these mutants act in the same pathway to control fainting (Figure 4, C and D). However, nca-1 strongly enhanced the nlf-1 fainter phenotype, but nca-2 did not significantly enhance nlf-1 (Figure 4, A and B). Neither nca-1 nor nca-2 single mutants have a fainter phenotype on their own (Humphrey et al. 2007), but the fact that an nca-1 nlf-1 double mutant has a strong fainter phenotype suggests that either nca-1 contributes more than nca-2 for normal locomotion or that nlf-1 specifically perturbs function of nca-2. Our data presented below are more consistent with the possibility that nca-1 contributes more than nca-2 to wild-type locomotion behavior.
We determined the cellular expression pattern of nlf-1 by fusing its promoter to GFP. nlf-1p::GFP was expressed in most or all neurons, but was not detected in other tissues (Figure 5A). This agrees with the expression pattern reported elsewhere (Xie et al. 2013). To determine the neuronal focus of the fainter phenotype, we performed rescue experiments in which we determined whether an nlf-1 mutant could be rescued by expression of a wild-type nlf-1 cDNA under the control of neuron-specific promoters. Expression of nlf-1(+) in all neurons (using the rab-3 promoter) or acetylcholine neurons (using the unc-17 promoter) fully rescued the nlf-1 mutant fainter phenotype (Figure 5B). Expression in acetylcholine motor neurons (using the acr-2 or unc-17β promoters) did not rescue the fainter phenotype, but expression driven by a head-specific derivative of the unc-17 promoter (unc-17Hp) fully rescued the fainter phenotype, indicating that the action of nlf-1 in head acetylcholine neurons is sufficient to prevent fainting (Figure 5B).
Previously, it was reported that expression of nlf-1 in premotor interneurons is sufficient to rescue the nlf-1 mutant fainter phenotype (Xie et al. 2013). However, we found that expression of nlf-1 in premotor interneurons using the glr-1 promoter (glr-1p) had only a weak effect on the nlf-1 mutant fainter phenotype and did not fully restore wild-type locomotion (Figure 5C). Fainting behavior was difficult to score in the glr-1 promoter-rescued nlf-1 mutant animals because they had sluggish movement and stopped frequently, though generally not with the suddenness and characteristic posture typical of fainters. Though we could blindly score the rescue of fainting in these animals by eye, we saw only weak rescue of the nlf-1 mutant by glr-1 promoter expression in our quantitative fainting assays and the effect was not statistically significant (Figure 5C). When we instead measured fainting as the percentage of animals that fainted within ten body bends, we did see a marginally significant rescue by glr-1 promoter expression (backward fainting: nlf-1 = 92%, nlf-1; glr-1p::nlf-1(+) = 68%, P=0.0738, Fisher’s exact test; forward fainting: nlf-1 = 80%, nlf-1; glr-1p::nlf-1(+) = 48%, P=0.0378, Fisher’s exact test). We may be underestimating the rescue of the fainting phenotype by glr-1 promoter expression due to the difficulty distinguishing the frequent pausing from true fainting. Nevertheless, rescue of the nlf-1 mutant is clearly stronger by expression in head acetylcholine neurons using the unc-17H promoter (Figure 5B). Though our data seem to contradict the previous study reporting that nlf-1 acts in premotor interneurons (Xie et al. 2013), there are several possible explanations. First, like our data, the data in the previous study in fact showed only partial rescue of fainting behavior by expression in premotor interneurons (Xie et al. 2013). Second, the premotor interneuron promoter combination used in the previous study (nmr-1p + sra-11p) leads to expression in several other head interneurons that may contribute to the phenotype. Third, it is possible that rescue is sensitive to expression level and that different levels of expression were achieved in the two studies, leading to different levels of rescue. We conclude that NLF-1 acts in head acetylcholine neurons, including the premotor interneurons, to promote sustained locomotion in the worm. Consistent with this, the premotor command interneurons have recently been shown to use acetylcholine as a neurotransmitter (Pereira et al. 2015).
Mutations in NCA channel subunits suppress activated Rho
To determine whether NCA mutants act downstream of Rho, we took advantage of an activated Rho mutant (G14V) expressed specifically in the acetylcholine neurons (McMullan et al. 2006). Like an activated Gq mutant, this activated Rho mutant has an exaggerated waveform. We built double mutants of the activated Rho mutant with mutations in unc-79, unc-80, nlf-1 and mutations in the NCA channel genes nca-1 and nca-2. The nca-1 and nca-2 genes are redundant for the fainter phenotype because neither mutant has a fainter phenotype individually, but the double mutant has a fainter phenotype indistinguishable from unc-79 and unc-80 mutants (Humphrey et al. 2007; Jospin et al. 2007; Yeh et al. 2008). We found that the high-amplitude waveform caused by activated Rho was strongly suppressed in unc-80 mutants, as well as in nca-1 nca-2 double mutants (Figure 6). In both cases, the resulting double or triple mutants had a fainter phenotype like the unc-80 or nca-1 nca-2 mutants on their own. By contrast, the high-amplitude waveform of activated Rho was incompletely suppressed by a mutation in nlf-1, consistent with the nlf-1 mutation causing only a partial loss of NCA channel function (Figure 6). Reciprocally, activated Rho suppressed the weak fainter phenotype of an nlf-1 mutant, again because Rho can act on NCA even in the absence of nlf-1. Additionally, a mutation in nca-1 also partially suppressed the high-amplitude waveform of activated Rho, but a mutation in nca-2 did not suppress the high-amplitude waveform (Figure 6). Thus, channels containing the NCA-1 pore-forming subunit have a larger role than NCA-2 channels in transducing Rho signals, similar to the interaction of NCA-1 and Gq signaling.
Because an activated Rho mutant has slow locomotion (Figure 7A) and unc-79, unc-80, and nca-1 nca-2 double mutants also have slow locomotion, it is difficult to determine whether these NCA channel mutants suppress the locomotion phenotype of activated Rho in addition to its waveform. We performed radial locomotion assays that provide a combined measurement of several aspects of the locomotion phenotype, including the rate of movement, frequency of reversals, and amplitude of the waveform (see Materials and Methods). By these assays, mutations in unc-80 or nca-1 nca-2 lead to only small increases in the radial distance traveled by an activated Rho mutant and an nca-2 mutant had no effect (Figure 7B). However, a mutation in nlf-1 much more strongly increased the radial distance traveled by an activated Rho mutant. Because the nlf-1 mutant is not as slow on its own, we could also directly assay its effect on the rate of locomotion of an activated Rho mutant by counting the number of body bends per minute. An nlf-1 mutation strongly increased the rate of locomotion of an activated Rho mutant (Figure 7A). In fact, the nlf-1 double mutant with activated Rho had a faster rate of locomotion than either activated Rho or nlf-1 on its own, similar to the effect of nlf-1 on the locomotion of an activated nca-1 mutant (Figure 3E). Additionally, a mutation in nca-1 strongly increased the locomotion rate of the activated Rho mutant, but a mutation in nca-2 had no effect (Figure 7A), further supporting the idea that NCA channels consisting of the NCA-1 subunit act downstream of Gq and Rho in this pathway.
Rho regulates worm locomotion independently of effects on development, as demonstrated by the fact that heat-shock induction of an activated Rho transgene in adults leads to a high-amplitude waveform similar to that seen in worms that express activated Rho in acetylcholine neurons (McMullan et al. 2006). Consistent with the idea that NCA-1 acts downstream of Rho, mutations in the fainter genes unc-80 and nlf-1 suppress the high-amplitude waveform phenotype of heat-shock induced activated Rho (Figure 8, A-C). Additionally, nlf-1 also suppresses the locomotion defect of heat-shock induced activated Rho (Figure 8D). Thus, Rho regulates worm locomotion via the NCA channels by acting in a non-developmental pathway that operates in adult neurons.
A dominant NCA-1 mutation suppresses activated Go
In C. elegans, the Gq pathway is opposed by signaling through the inhibitory Go protein GOA-1 (Hajdu-Cronin et al. 1999; Miller et al. 1999). Thus, loss-of-function mutants in goa-1 are hyperactive and have a high-amplitude waveform, similar to the gain-of-function Gq mutant egl-30(tg26). We found that the uncloned dominant mutant unc-109(n499) which is paralyzed and resembles loss-of-function mutants in egl-30 (Park and Horvitz 1986) carries an activatin 1mutation in goa-1 (see Materials and Methods). The goa-1(n499) mutant is paralyzed as a heterozygote, and is lethal as a homozygote (Park and Horvitz 1986). We performed a screen for suppressors of goa-1(n499) and isolated a partial intragenic suppressor, goa-1(n499 ox304) (Materials and Methods). goa-1(n499 ox304) homozygote animals are viable but are almost paralyzed and show very little spontaneous movement, with a straight posture and low-amplitude waveform (Figure 9, C and D). However, when stimulated, the goa-1(n499 ox304) mutant is capable of slow coordinated movements (Figure 9, A and B).
We mutagenized the goa-1(n499 ox304) strain and screened for animals that were not paralyzed or did not have a straight posture. Among the suppressors, we isolated a mutant (ox352) that was found to be a dominant gain-of-function mutation in the nca-1 gene (Bend et al. 2016). The nca-1(ox352) mutant has a coiled posture and high-amplitude waveform reminiscent of the activated Gq and activated Rho mutants (Figure 3B). Furthermore, nca-1(ox352) suppresses the straight waveform of the activated Go mutant goa-1(n499 ox304) (Figure 9, C and D). Thus, activation of NCA-1 suppresses activated Go whereas loss of NCA-1 suppresses activated Gq, both consistent with the model that Go inhibits Gq and that NCA-1 is a downstream effector of the Gq pathway.
Discussion
In this study, we identify the NCA-1 and NCA-2 ion channels as downstream effectors of the heterotrimeric G proteins Go and Gq. Previously, it has been demonstrated that activation of Goα GOA-1 activates the RGS protein EAT-16, which in turn inhibits Gqα (Hajdu-Cronin et al. 1999; Miller et al. 1999). Loss of goa-1 leads to hyperactive worms (Mendel et al. 1995; Ségalat et al. 1995). Here we identified an activated mutation in goa-1 that causes animals to be paralyzed, closely resembling null mutations in the Gqα gene egl-30 (Brundage et al. 1996). Suppressors of activated Go include loss-of-function mutations in the RGS EAT-16, and gain-of-function mutations in the NCA-1 channel, suggesting that activated Go inactivates Gq, and could thereby indirectly inactivate the NCA-1 cation channel.
Further genetic epistasis data indicate that Gq activates the RhoGEF Trio and the small GTPase Rho. Rho then acts via an unknown mechanism to activate the NCA-1 and NCA-2 ion channels, which are required for normal neuronal activity and synaptic transmission in C. elegans (Jospin et al. 2007; Yeh et al. 2008; Xie et al. 2013; Gao et al. 2015). Thus, this work identifies a new genetic pathway from Gq to an ion channel that regulates neuronal excitability and synaptic release: Gq → RhoGEF → Rho → NCA channels (Figure 10). The NCA channels have not been previously identified as effectors of the Gq pathway, so this pathway may give insight into how the NCA channels are activated or regulated.
Mutations that eliminate the NCA channels, such as unc-80 or the nca-1 nca-2 double mutant, suppress the locomotion and body posture phenotypes of an activated Rho mutant (Figures 6-8), suggesting that NCA channels act downstream of Rho in a linear pathway. Additionally, channels composed of the pore-forming subunit NCA-1 may be the main target of Gq-Rho signaling. First, loss-of-function mutations in nca-1 alone, but not nca-2, partially suppress activated Gq and activated Rho mutants. Second, activating gain-of-function mutants of nca-1 have phenotypes reminiscent of activated Gq and activated Rho mutants, but no activated mutants in nca-2 have been isolated. Third, loss-of-function mutations in nca-1, but not nca-2, enhance the weaker fainting phenotype of an nlf-1 mutant which does not fully eliminate function of the NCA channels. Together these data suggest that although either NCA-1 or NCA-2 activity is sufficient for wild-type locomotion, NCA-1 is likely to be the main target of G protein regulation.
We characterized three locomotory behaviors in this manuscript: locomotion rate, waveform, and fainting. The Gq-Rho-NCA pathway regulates all three behaviors, but our genetic epistasis and cell-specific rescue experiments suggest that these behaviors are differentially regulated and involve at least partially distinct sets of neurons. First, the hyperactive locomotion and high-amplitude waveform phenotypes of the activated Gq mutant are genetically separable, since they are differentially suppressed by mutations in the PLCβ pathway and the Rho-NCA pathway. Mutations in the Rho-NCA pathway suppress both the locomotion rate and high-amplitude waveform of activated Gq whereas mutations in the PLCβ pathway suppress the locomotion rate but only very weakly suppress the high-amplitude waveform of activated Gq. Thus, Gq acts through both the PLCβ pathway and the Rho-NCA pathway to regulate locomotion rate, but primarily through the Rho-NCA pathway to regulate the waveform of the animal. Second, we found that NLF-1 activity in head acetylcholine neurons is sufficient to fully rescue the fainting phenotype of an nlf-1 mutant. However, inhibition of Rho in the head acetylcholine neurons did not suppress the high-amplitude waveform of the activated Gq mutant, but inhibition of Rho in all neurons did suppress. This suggests that Rho does not act solely in head acetylcholine neurons to regulate the waveform. Thus, the Gq-Rho-NCA pathway acts in at least two different classes of neurons to regulate fainting and waveform.
It is not clear whether Rho activation of the NCA channels is direct or indirect. Gq directly interacts with and activates the RhoGEF and Rho, as shown by the crystal structure of a complex between Gqα, RhoGEF, and Rho (Lutz et al. 2007). Rho is known to have many possible effectors and actions in cells (Etienne-Manneville and Hall 2002; Jaffe and Hall 2005). In C. elegans neurons, Rho has been previously shown to regulate synaptic transmission via at least two pathways, one involving a direct interaction of Rho with the DAG kinase DGK-1 and one that is DGK-1 independent (McMullan et al. 2006). A candidate for the link between Rho and NCA is the type I phosphatidylinositol 4-phosphate 5-kinase (PIP5K) that synthesizes the lipid phosphatidylinositol 4,5,-bisphosphate (PIP2). PIP5K is an intriguing candidate for several reasons. First, activation of Gq in mammalian cells has been shown to stimulate the membrane localization and activity of PIP5K via a mechanism that depends on Rho (Chatah and Abrams 2001; Weernink et al. 2004). Second, in C. elegans, mutations eliminating either the NCA channels (nca-1 nca-2 double mutant) or their accessory subunits (unc-79 or unc-80) suppress mutants in the PIP2 phosphatase synaptojanin (unc-26) and also suppress phenotypes caused by overexpression of the PIP5K gene ppk-1 (Jospin et al. 2007). Loss of a PIP2 phosphatase or overexpression of a PIP5K are both predicted to increase levels of PIP2. Because the loss of NCA channels suppresses the effects of too much PIP2, it is possible that excessive PIP2 leads to overactivation of NCA channels and that PIP2 might be part of the normal activation mechanism. There are numerous examples of ion channels that are regulated or gated by phosphoinositide lipids such as PIP2 (Balla 2013), though PIP2 has not been shown to directly regulate NCA/NALCN.
The NCA/NALCN ion channel was discovered originally by bioinformatic sequence analyses (Lee et al. 1999; Littleton and Ganetzky 2000). It is conserved among all metazoan animals and is evolutionarily-related to the family of voltage-gated sodium and calcium channels (Liebeskind et al. 2012), forming a new branch in this super-family. Although the cellular role of the NCA/NALCN channel and how it is gated are not well understood, NALCN and its orthologs are expressed broadly in the nervous system in a number of organisms (Lee et al. 1999; Lear et al. 2005; Humphrey et al. 2007; Lu et al. 2007; Jospin et al. 2007; Yeh et al. 2008; Lu and Feng 2011; Lutas et al. 2016). Moreover, mutations in this channel or its auxiliary subunits lead to defects in rhythmic behaviors in multiple organisms (Lear et al. 2005, 2013; Lu et al. 2007; Jospin et al. 2007; Yeh et al. 2008; Pierce-Shimomura et al. 2008; Xie et al. 2013; Funato et al. 2016). Thus, NCA/NALCN is likely to play an important role in controlling membrane excitability. Additionally, NALCN currents have been reported to be activated by two different G protein-coupled receptors (GPCRs), the muscarinic acetylcholine receptor and the substance P receptor, albeit in a G protein-independent fashion (Lu et al. 2009; Swayne et al. 2009), and by low extracellular calcium via a G protein-dependent pathway (Lu et al. 2010). The latter study further showed that expression of an activated Gq mutant inhibited the NALCN sodium leak current, suggesting that high extracellular calcium tonically inhibits NALCN via a Gq-dependent pathway and that low extracellular calcium activates NALCN by relieving this inhibition (Lu et al. 2010). By contrast, we find that Gq activates the NCA channels in C. elegans, and we show that the NCA channels are physiologically relevant targets of a Gq signaling pathway that acts through Rho.
In the last few years, both recessive and dominant human diseases characterized by a range of neurological symptoms including hypotonia, intellectual disability, and seizures have been shown to be caused by mutations in either NALCN or UNC80 (Köroğlu et al. 2013; Al-Sayed et al. 2013; Chong et al. 2015; Aoyagi et al. 2015; Shamseldin et al. 2016; Stray-Pedersen et al. 2016; Gal et al. 2016; Fukai et al. 2016; Karakaya et al. 2016; Perez et al. 2016; Bend et al. 2016; Lozic et al. 2016; Valkanas et al. 2016; Wang et al. 2016; Vivero et al. 2017). Notably, dominant disease-causing mutations in the NALCN channel were modeled in worms and resemble either dominant activated or loss-of-function NCA mutants such as the ones we use in this study (Aoyagi et al. 2015; Bend et al. 2016). Human mutations in other components of the pathway we have described may cause similar clinical phenotypes.
Acknowledgments
We thank Shohei Mitani for the nlf-1(tm3631) mutant; Steve Nurrish for worm strains and plasmids carrying activated Rho or C3 transferase; Ken Miller for a plasmid with the unc-17β promoter; Wayne Davis for FLP/FRT plasmids; Yuji Kohara for cDNA clones; the Sanger Center for cosmids; Brooke Jarvie, Jill Hoyt, and Michelle Giarmarco for the isolation of unc-79 and unc-80 mutations in the Gq suppressor screen; and Dana Miller for the use of her microscope and camera to take worm photographs. Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). M.A. is an Ellison Medical Foundation New Scholar. E.M.J. is an Investigator of the Howard Hughes Medical Institute. This work was supported by NIH grants R00 MH082109 to M.A and R01 NS034307 to E.M.J.