Hyperactivation of L-type voltage-gated Ca²⁺ channels in Caenorhabditis elegans striated muscle can result from point mutations in the IS6 or the IIIS4 segment of the α1 subunit

Calcium ions play multiple roles in cell physiology, starting from the initiation of a one-cell growth during fertilization to its termination by apoptosis. Among the molecules involved in fine-tuning of Ca²⁺ homeostasis, voltage-gated Ca²⁺ channels are responsible for local and transient increases of intracellular Ca²⁺ concentration. Based on pharmacological and biophysical properties, voltage-gated Ca²⁺ channels are classified into three families: the L-type, the P/Q/R/N-type and the T-type channels (Catterall, 2011). L-type channels have been shown to play crucial roles in Ca²⁺ homeostasis of excitable cells. Dysfunction of these channels has been reported in several congenital disorders, including hypokalemic periodic paralysis (hypoPP), malignant hyperthermia or Timothy syndrome (Jurkat-Rott et al., 1994; Ptácek et al., 1994; Monnier et al., 1997; Splawski et al., 2004). The molecular bases of these channelopathies have been well described: mutations are located at multiple sites of the pore-forming and voltage-sensor regions of the α₁ subunit. The α₁ subunit is composed of four equivalent domains (named I to IV) containing six transmembrane segments (named S1 to S6) and linked to each other by large cytoplasmic loops (Catterall, 2011).

In hypoPP and malignant hyperthermia, most mutations target the arginine residues in the S4 voltage-sensor segments of the skeletal subunit CACNA1S (Jurkat-Rott et al., 1994; Ptácek et al., 1994; Monnier et al., 1997; Wang et al., 2005; Chabrier et al., 2008; Carpenter et al., 2009; Ke et al., 2009; Pirone et al., 2010; Toppin et al., 2010; Hirano et al., 2011; Li et al., 2012). In Timothy syndrome, two glycine residues in IS6 segment have been reported to be mutated in the cardiac and neuronal subunit CACNA1C (Splawski et al., 2004; Splawski et al., 2005). Understanding the functional effects of these L-type channel mutations is a prerequisite to develop new therapeutic strategies. However, the physiological consequences of the mutations are still poorly understood. Expressing these channels in heterologous expression systems remains challenging. Moreover, these channels interact with multiple cellular partners in vivo. Thus in vivo studies of mutant forms of L-type calcium channels are necessary. Until now, very few of these studies have been published. Cheng et al. have described that the inactivation kinetics of the current is strongly slowed in mice carrying the Timothy syndrome mutation G406R (Cheng et al., 2011). Wu et al. have shown that skeletal muscle fibers of mice carrying the hypoPP mutation R528H exhibit an anomalous leak current through the mutated S4 segments (Wu et al., 2012). Generation of such mutant mice is still very time consuming, so developing other in vivo approaches could be a fruitful way to understand the effects of the multiple voltage-gated Ca²⁺ channel α₁ subunit mutations.

The nematode Caenorhabditis elegans Maupas could be an invaluable source for α₁ subunit mutants. Indeed, over 150 mutations have already been reported in egl-19, the sole gene encoding an L-type α₁ voltage-gated Ca²⁺ channel subunit in C. elegans (www.wormbase.org) (Lee et al., 1997). Based on their genetic and phenotypic features, egl-19 mutants have been classified into three groups: the complete loss-of-function mutants that are lethal, the partial loss-of-function mutants that exhibit a slender morphology, and the gain-of-function mutants that are characterized by a bulky body (Lee et al., 1997). To date, the effects on currents have only been described for two partial loss-of-function mutations, n582 and ad1006. The n582 mutation changes an arginine residue to a histidine in segment IIIS4 of EGL-19. In the non-striated pharyngeal muscle cells and in the striated body wall muscle cells of n582 mutants, voltage-dependent currents activate more slowly than those of wild-types and their voltage dependency is shifted towards positive potentials (Jospin et al., 2002; Shtonda and Avery, 2005; Gao and Zhen, 2011). In body wall muscle cells of ad1006 mutants, voltage dependency is also shifted towards positive potentials and the current amplitude is decreased (Gao and Zhen, 2011). The functional effects of two gain-of-function mutations, n2368 and ad695, have also been partially investigated: action potentials have been monitored in pharyngeal and body wall muscle cells (Lee et al., 1997; Gao and Zhen, 2011; Liu et al., 2011), and Ca²⁺ transients have been measured in the pharynx (Kerr et al., 2000). However, these studies have led to conflicting results: Gao and Zhen have shown that action potentials of ad695 mutants are similar to those of wild-type (Gao and Zhen, 2011), whereas Liu et al. reported action potentials with longer duration and amplitude (Liu et al., 2011).

Here we describe the functional effects of egl-19(n2368) and egl-19(ad695) gain-of-function (gf) mutations in striated body wall muscle cells of C. elegans. By phenotypic and pharmacological analysis, we show that the bulky phenotype of these gain-of-function mutants is correlated with decreased locomotion, partially caused by the hypercontraction of striated body muscles. With in situ patch-clamp recordings we provide evidence that the hypercontraction of the muscles is due to the hyperactivation of L-type channels: the voltage dependency is shifted towards more negative potentials and the inactivation is altered. We also demonstrate that the amplitude of the voltage-dependent Ca²⁺ current is decreased in n2368 mutants, partially due to the membrane mislocalization of the mutant channel.

As previously described (Lee et al., 1997), we found that egl-19(ad695) and egl-19(n2368) mutants were shorter than, but as wide as, wild-type animals (Fig. 1A). We quantified this size change by measuring the length and width of young adults. egl-19(ad695) and egl-19(n2368) animals were, respectively, 20 and 30% shorter than wild-type, but barely thinner (Fig. 1B). We used the width/length ratio as an indicator of body morphology (Spooner et al., 2012): a low ratio corresponds to a slender body, whereas a high ratio indicates a thickset morphology. The width/length ratios of both egl-19(gf) alleles are significantly higher, with egl-19(n2368) being greater than egl-19(ad695). Both strains have a stocky body shape.

We then tested whether the stocky morphology of egl-19(gf) was correlated with locomotion defects. Worm locomotion was observed in liquid where animals thrash vigorously, flexing and bending their bodies. Quantification of body bends showed that this form of locomotion was reduced in both egl-19(gf) mutants (Fig. 1C), but more strikingly for the n2368 strain (90% decreased). In addition, we observed that swimming of egl-19(gf) mutants was altered: these worms were not bending on either side like wild-type, but rather tended to curl up. We thus showed that besides morphological defects, egl-19(gf) worms showed quantitative and qualitative locomotion defects.

In their initial study, Lee et al. (Lee et al., 1997) suggested that the morphological defect of egl-19(gf) worms was due to a high muscle tone. However, because egl-19 has been shown to be involved in worm development (Tam et al., 2000; Moghal et al., 2003; Bauer Huang et al., 2007), a developmental failure could also explain the shorter size of the gain-of-function mutants. To test whether egl-19(gf) mutants are indeed hypercontracted, we carried out a set of pharmacological tests, adapted from Petzold et al. (Petzold et al., 2011). We took advantage of the double innervation of C. elegans muscles, coming from GABA and acetylcholine motor neurons, to pharmacologically induce either hypercontraction or relaxation of wild-type and egl-19(gf) animals. Worms were exposed to 0.1 mmol l⁻¹ levamisole, a cholinergic agonist increasing muscle tone, for 30 min. As expected for a hypercontracting drug, the levamisole induced an increase of the width/length ratio for the three strains (Fig. 2A). To compare the drug responsiveness of each strain, the width/length ratio of each worm prior to treatment was subtracted from the ratio of the same worm after the 30 min treatment, and this variation was then normalized to the ratio prior to treatment, to give the levamisole-induced contraction percentage. If egl-19(gf) strains are already hypercontracted, they should contract less in response to levamisole. We observed that levamisole indeed affected the body morphology of wild-type animals more than that of egl-19(gf) mutants: after the levamisole treatment the width/length ratio varied by 24% in wild-types, whereas variations of 14 and 17% were observed in egl-19(ad695) and egl-19(n2368) mutants, respectively (Fig. 2B). Conversely, the aptitude of body wall muscles to relax was measured upon a 30 min treatment with 10 mmol l⁻¹ muscimol, a GABAergic agonist inducing muscle relaxation. As expected, a decrease of the width/length ratio after muscimol treatment was observed in the three strains (Fig. 2C). The width/length ratio of each worm after the 30 min treatment was subtracted from the ratio prior to treatment of the same worm, and this variation was then normalized to the ratio prior to treatment, to give the muscimol-induced relaxation percentage. If egl-19(gf) worms are hypercontracted, they should relax more than wild-type animals in the presence of muscimol. Muscimol was found to have a stronger effect on egl-19(gf) worms compared with wild-type: after the muscimol treatment, the width/length ratio varied by 5% in wild-type, whereas variations of 7 and 11% were observed in egl-19(ad695) and egl-19(n2368) mutants, respectively (Fig. 2D). The decrease of levamisole-induced contraction and the increase of muscimol-induced relaxation in both egl-19(gf) mutants are in agreement with a hypercontracted basal state.

Using the whole-cell configuration of the patch-clamp technique in the presence of K⁺ channel blockers, we investigated the properties of Ca²⁺ currents from body wall muscle cells of dissected wild-type, egl-19(ad695) and egl-19(n2368) worms. Muscle cells were depolarized by 200 ms steps from a holding potential of −60 mV. As previously described (Jospin et al., 2002), voltage-dependent Ca²⁺ currents recorded from wild-type animals displayed a biphasic appearance, with an early transient peak followed by a slowly inactivating component (Fig. 3A). This decay was altered in egl-19(n2368) and egl-19(ad695) worms: the early transient peak was difficult to distinguish in these animals (Fig. 3A,B). To confirm that inactivation was modified in egl-19(gf) worms, we analysed the inactivation kinetics. For each cell, we measured the R₂₀₀ parameter, defined as the current fraction still present at the end of the pulse. We found that this fraction was significantly increased by 25 and 66% in egl-19(ad695) and egl-19(n2368) mutants, respectively, compared with that of wild-type (Fig. 3C). We also determined the activation time constant τ and showed that it is increased about 2-fold in egl-19(n2368) worms (Fig. 3C).

We then established the current–voltage relationships at the maximal amplitude of the current (Fig. 3D). For each cell, individual current–voltage relationship was fitted with Eqn 1 (see Materials and methods) in order to obtain the Ca²⁺ current parameters G_max, V_rev, V_0.5 and k (Fig. 3E). The reversal potential V_rev was shifted toward negative potentials in egl-19(ad695) (50.2±1.4 mV) and egl-19(n2368) (40.0±3.0 mV) mutants, compared with that of wild-type (57.1±1.1 mV). This result suggests that voltage-gated Ca²⁺ channels might be less selective in egl-19(gf) animals than in wild-types. Moreover, the voltage dependency of the Ca²⁺ currents in egl-19(gf) mutants was altered: the half-activation potential V_0.5 was decreased by 6 and 16 mV, respectively, for ad695 and n2368 mutants compared with wild-type, and the steepness factor k was increased by 20% in n2368 mutants. These results show that the voltage-gated Ca²⁺ channels in both egl-19(gf) strains are hyperactive: they open at more negative potentials because of the voltage dependency shift, and their mean open time is longer because of the partial or complete lack of the inactivation properties. Interestingly, the maximal conductance G_max was significantly reduced in egl-19(n2368) mutants, by 60% compared with the wild-type. This result could be explained by lower levels of Ca²⁺ channels at the plasma membrane, and/or by intrinsic properties of the channel, such as modification of the pore.

To understand the origin of G_max decrease, we examined the membrane localization of voltage-gated Ca²⁺ channels in body wall muscles of wild-type, egl-19(ad695) and egl-19(n2368) worms. Because of the lack of specific antibodies, we decided to use a single-copy transgenic strain expressing a functional green fluorescent protein (GFP)-tagged voltage-gated Ca²⁺ channel to evaluate the Ca²⁺ channel localization. We generated an N-terminally GFP-tagged UNC-36 fusion protein because unc-36 encodes the sole α₂-δ auxiliary subunit of the L-type voltage-gated Ca²⁺ channel expressed in body wall muscles (Fig. 4A) (Lainé et al., 2011) and a recent study has also shown that the interaction between α₁ and α₂-δ subunits is strong and that they are intimately associated at the plasma membrane (Cassidy et al., 2014). The expression of GFP:UNC-36 in unc-36 mutants rescued the locomotion phenotype and the Ca²⁺ current defects observed in unc-36 mutants (Lainé et al., 2011; Zhan et al., 2014), so we assumed that the localization of the fusion protein matched that of endogenous channels. As previously described (Frøkjaer-Jensen et al., 2006), unc-36 staining was observed in most neuronal and muscle cells (Fig. 4B–E). The membrane fluorescence of striated muscle cells was imaged on the basal plasma membrane just above the hypodermis and showed a punctate pattern in wild-type animals (Fig. 5A). While this distribution was not affected in worms carrying the ad695 mutation, egl-19(n2368) worms showed irregular dots of fewer intensity surrounded by a diffuse staining (Fig. 5A,B). In addition, the mean fluorescence was reduced by 25% in egl-19(n2368) animals compared with wild-type, whereas it was not affected in egl-19(ad695) mutants (Fig. 5C). Finally, we imaged muscles at mid-height section and observed that the fluorescence matched the diamond-shaped outline of the cells, as expected considering the plasma membrane localization of voltage-gated Ca²⁺ channels. However, we noticed an accumulation of fluorescence at the tip of the cells in egl-19(n2368) animals, suggesting a submembrane accumulation of these channels (Fig. 5D). These results support the idea that the decrease of the maximal conductance in body wall muscles of egl-19(n2368) animals is at least in part due to a disrupted plasma membrane localization of the channel.

In this study, we provide a detailed characterization of two gain-of-function mutants in egl-19, the gene that encodes the α₁ subunit of L-type voltage-gated Ca²⁺ channels in C. elegans muscles. We showed that (1) the two variants n2368 and ad695 exhibit morphological and locomotion defects linked to a high muscle tone; (2) this high muscle tone is correlated to an increased sensitivity to the voltage of the channel and a decrease in the inactivation kinetics in both mutants; and (3) the L-type Ca²⁺ channel localization is altered in egl-19(n2368) mutants and the current conductance is decreased.

Why are the egl-19(gf) mutants hypercontracted? As in mammals, striated muscle contraction is dependent on action potentials firing (Liu et al., 2011; Liu et al., 2013); however, in C. elegans this firing involves voltage-gated Ca²⁺ channels and not voltage-gated Na⁺ channels (Jospin et al., 2002; Liu et al., 2011). Because of the increased sensitivity of the mutant channels to voltage, action potentials could be triggered by small depolarizations that would have no effect in wild-type. The increased propensity for action potentials would correlate with more frequent muscle contraction. Interestingly, the increase of muscle tone is more pronounced in egl-19(n2368) mutants, as is the modification of the voltage dependency. In egl-19(n2368) muscles, the shift of voltage dependency sets the current activation threshold around −30 mV instead of −20 mV in wild-type (Fig. 3D). Because the resting membrane potential of body muscle cells is typically around −25 mV in wild-type (Gao and Zhen, 2011; Liu et al., 2011) and is not different in egl-19(n2368) (our unpublished data), these results suggest that the voltage-gated Ca²⁺ channels are open at this potential in egl-19(n2368) muscles. Moreover, the slowing down of inactivation in egl-19(n2368) mutants might also increase the calcium flux into the cell. Thus despite the conductance decrease in this mutant, the calcium flux into the cell even at resting potential is likely to be sufficient to increase muscle tone. Alternatively, and non-exclusively, the hypercontraction could also have a neuronal origin: egl-19 is expressed in motor neurons (Lee et al., 1997) and its partial loss-of-function specifically in these cells decreases muscle excitability (Gao and Zhen, 2011).

Muscle voltage-gated Ca²⁺ channels were visualized using a GFP-tagged α₂-δ/UNC-36 subunit and were found distributed at the boundary of body wall muscles and as a punctuated pattern at the basal membrane. This pattern is reminiscent of the one observed by Kim et al. (Kim et al., 2009) with a EGL-19-mCherry multi-copy array. These authors concluded that EGL-19 is localized near the dense bodies, which are the analogs of Z-discs coupled to the costamere adhesion complex (Lecroisey et al., 2007). Interestingly, flattened membrane vesicles are observed around the dense bodies between the plasma membrane and the myofilaments and they have been described as sarcoplasmic reticulum cisternae (Maryon et al., 1996). Further studies will be needed to determine whether voltage-gated Ca²⁺ channels are indeed in front of the sarcoplasmic reticulum.

ad695 mutation leads an alanine-to-valine change at residue 906 in the IIIS4 segment (supplementary material Fig. S1). The S4 segments contain four to seven repeated motifs of a positive charged residue followed by two hydrophobic residues that are critical components of the voltage sensor (Noda et al., 1984; Stühmer et al., 1989; García et al., 1997). Mutations in these charged residues have been identified in L-type channels from patients suffering from hypoPP (Fontaine et al., 1994; Jurkat-Rott et al., 1994; Ptácek et al., 1994; Wang et al., 2005; Chabrier et al., 2008; Li et al., 2012). One of these mutations has been shown to allow a Ca²⁺ leak through an aberrant conduction pathway, named the gating pore or the ω pore (Wu et al., 2012). Gating pore currents have been first recorded from voltage-gated K⁺ and Na⁺ channels (Starace and Bezanilla, 2004; Sokolov et al., 2005; Tombola et al., 2005). Depending on the nature of the mutations, these channels are activated at resting potentials or during depolarizations, and they are permeable either to protons or to several monovalent cations. In our study we described the functional effects of the substitution of a small hydrophobic residue for a larger one. Such a mutation has never been studied before in voltage-gated Ca²⁺ channels, but the role of hydrophobic residues in S4 have been analysed in Na⁺ and K⁺ voltage-gated channels (Auld et al., 1990; Lopez et al., 1991; McCormack et al., 1991; Bendahhou et al., 2007; Yang et al., 2007; Elliott et al., 2012; Perry et al., 2013). Bendahhou et al. have shown for Na⁺ voltage-gated channels that the size of the lateral chain can modify the half-activation potential and the inactivation kinetics (Bendahhou et al., 2007). These authors suggested that hydrophobic residues of S4 segment influence molecular interactions between S3 and S4 segments, and thus act on the voltage detection. The ad695 mutation leads to the replacement of an alanine by another hydrophobic residue, a valine, with a bigger lateral chain; this change could thus have a similar effect on voltage sensor as the one described for Na⁺ channels. Alternatively, Perry et al. have shown that mutations of the S4 hydrophobic residues in cardiac K⁺ voltage-gated channels alter the inactivation kinetics by modifying the S4 and S5 interactions (Perry et al., 2013). Besides the voltage dependency and the inactivation kinetic defects, L-type Ca²⁺ currents in egl-19(ad695) muscles reverse at more negative potentials, indicating a decrease of Ca²⁺ selectivity. Considering the localization of ad695 mutation in the S4 segment and the lesser selectivity for Ca²⁺ of gating pore currents, it is tempting to propose that Ca²⁺ channels in egl-19(ad695) muscle might support a gating pore current. Gating pore currents have never been associated with mutations other than the ones swapping arginine residues, so further studies will be needed to test this hypothesis.

The n2368 mutation causes a substitution of glycine with arginine at residue 365 of the IS6 segment (Lee et al., 1997). This glycine residue is the second glycine of a GX₉GX₃G motif located in the lower third part of IS6, a residue conserved in all high voltage-activated calcium channels (supplementary material Fig. S1). Residue substitution experiments associated with crystal structure analysis extrapolated from K⁺ channels have demonstrated that S6 segments play a central role in voltage-dependent channel gating (Zhou et al., 2001; Hohaus et al., 2005; Long et al., 2005; Zhen et al., 2005; Beyl et al., 2007; Long et al., 2007; Kudrnac et al., 2009). In particular, glycine residues of the GX₉GX₃G motif are involved in activation and inactivation mechanisms: mutations of these residues lead to a shift of the half-activation potential and a decrease of inactivation speed (Splawski et al., 2004; Splawski et al., 2005; Raybaud et al., 2006; Barrett and Tsien, 2008; Cens et al., 2008; Depil et al., 2011; Yazawa et al., 2011). Mutations of the second and third glycine have been identified in patients suffering from Timothy syndrome, a multisystem disorder mainly characterized by cardiac arrhythmias, autism and syndactyly (Splawski et al., 2004; Splawski et al., 2005). The n2368 mutation replaces the second glycine by an arginine. Such a change has not been described before; however, the effects of replacement of this glycine by methionine, asparagine, alanine, serine or tryptophane have been analysed in rabbit L-type channels expressed in heterologous systems (Depil et al., 2011). Depending on the mutations, the half-activation potential was shifted towards more negative potentials, as we observed for egl-19(n2368), or towards more positive ones. In all cases, inactivation kinetics of these mutant channels was dramatically reduced. Finally, the replacement of glycine by arginine in the n2368 mutation creates a RX₂S motif, a consensus sequence for phosphorylation by Ca²⁺/calmodulin-dependent protein kinase type II (CaMKII) (Kennelly and Krebs, 1991). Mutations of the third glycine have led to similar changes in some patients with Timothy syndrome. In heterologous expression systems, Erxleben et al. have shown that L-type channels carrying this kind of mutation exhibit an increase of the mean open time associated with a decrease of unitary conductance, both dependent on CaMKII activity (Erxleben et al., 2006). Thus CaMKII could also play a role in egl-19(n2368) mutants by modulating the properties of the mutant channels.

What is the origin of the Ca²⁺ current conductance decrease in egl-19(n2368) muscles? This decrease can be caused by a reduction of the channel unitary conductance, as observed by Erxleben et al. (Erxleben et al., 2006) with a similar mutation, and/or by a decrease of the channel density at the cell surface. Our data showed that the channel localization is altered in egl-19(n2368) mutants; so the n2368 mutation in the IS6 segment seems at least to impair channel trafficking and/or recycling. The IS6 segment has never been shown to be involved in channel trafficking before; however, the n2368 mutation might have an effect beyond the IS6 segment. The I–II loop downstream of IS6 is crucial for proper targeting of the channel upon β subunit binding (Pragnell et al., 1994). Structure–function studies have shown that β binding at the alpha interaction domain (AID) of I–II loop induces a coil-to-helix conformation of the proximal linker, strengthening the connection between the pore and the β subunit/I–II loop complex (Opatowsky et al., 2004; Van Petegem et al., 2004; Arias et al., 2005; Findeisen and Minor, 2009; Almagor et al., 2012). Proximal linker structure integrity is thought to be essential for β-dependent modulation of α₁ gating but its role in the localization of the channel has been less investigated (Findeisen and Minor, 2009; Gonzalez-Gutierrez et al., 2010; Almagor et al., 2012). So according to these data, the n2368 mutation in IS6 could possibly disrupt the proximal linker integrity, decreasing β modulation of channel inactivation and α₁ trafficking. Alternatively the mislocalization of the channel may be a consequence of the mutated calcium channel hyperactivation. Ca²⁺ influx through L-type Ca²⁺ channels has been shown to couple membrane excitation to transcription (D'Arco and Dolphin, 2012). In particular, the distal C-terminal domain of L-type channels may act as a transcriptional regulator. The cleavage of this domain is thought to be Ca²⁺ dependent, and the resulting transcription factor has been shown to modulate the transcription of several genes, including the gene encoding the L-type α₁ subunit itself (Gomez-Ospina et al., 2006; Schroder et al., 2009). The increased calcium flux in egl-19(n2368) muscles, even at resting potentials, could decrease egl-19 transcription level or modify the expression of proteins involved in Ca²⁺ channel trafficking or recycling.

Now that we have characterized the effects of egl-19(gf) mutations on L-type voltage-dependent currents, we are planning to use these mutants to identify new partners of the L-type Ca²⁺ channel using genetic screens. In cardiomyocytes of mice, Cheng et al. have shown that downregulation of the anchoring protein AKAP150 restores normal gating of L-type Ca²⁺ channels carrying a Timothy syndrome mutation (Cheng et al., 2011). We believe that identification of new partners of this channel could lead to novel therapeutics targets for treating pathologies involving L-type channels, such as hypoPP or Timothy syndrome.

All C. elegans strains were grown at 20°C according to the method of Brenner (Brenner, 1974). N2 Bristol, DA695 egl-19(ad695) IV, MT6129 egl-19(n2368) IV, CB251 unc-36(e251) III and EG4322 [ttTi5605 II, unc-119(ed9) III] strains were provided by the Caenorhabditis Genetics Center (CGC, Twin Cities, MN, USA).

The targeting vector (pHZ025) used to generate the krSi3(II) line was built as follows. A plasmid containing the unc-36 cDNA was provided by C.I. Bargmann (Saheki and Bargmann, 2009). This plasmid was digested by KpnI and EcoRI to remove egfp from the UNC-36 C-terminus, treated with T4 polymerase and circularized by self-ligation, resulting in the plasmid pHZ001. Next, egfp was fused in frame after the signal-peptide coding sequence in the unc-36 cDNA by PCR fusion. First, egfp was amplified using oHZ07 5′-TGGCCAGCTAGCAAAGGAGAA-3′ and oHZ019 5′-TTCTTTTATGCTCTCCTTATTAAAGTCCATGCCATGTGTAATCCC-3′. Next, the unc-36 cDNA was amplified using oHZ021 5′-ATGGCATGGACTTTAATAAGGA-3′ and oHZ08 5′-AAAAGATCTTTTGTTTCTTCACGATTT-3′. Both fragments were then fused by PCR fusion using oHZ07 and oHZ08 and the resulting fragment was inserted into a MscI/BglII-digested pHZ001. A 2 kb region upstream of the unc-36 open reading frame was amplified from genomic DNA using oHZ09 5′-AAAAAAAAGCGGCCGCAAAATGCATGCAGTAATTAGTGTCC-3′ and oHZ010 5′-TGGCCACTTATTAAAACTGCTTGT-3′, and this sequence was inserted into the previous plasmid after NotI and MscI digestion, resulting in plasmid pHZ009. Finally, the transgene sequence (Punc-36::egfp::unc-36::3′UTR unc-54) was released from pHZ009 using NotI and ApaI, blunted using T4 polymerase and inserted into the pCFJ151 vector backbone, resulting in pHZ025.

We followed the direct-insertion MosSCI protocol (Frøkjaer-Jensen et al., 2008) to generate krSi3. In brief, we micro-injected the gonads of EG4322 worms with a DNA mix containing: the targeting vector pHZ025 at 50 ng μl⁻¹, pJL43.1 at 50 ng μl⁻¹ (Mos transposases expressed under germline constitutive promoter Pglh-2), and transformation markers Pmyo3::gfp at 10 ng μl⁻¹ and Prab3::gfp at 10 ng μl⁻¹. MosSCI events were selected by screening for the restoration of wild-type locomotion and loss of the transformation markers. The krSi3 insertion was validated by PCR using oCF418 5′-TCTGGCTCTGCTTCTTCGTT-3′ and oCF419 5′-CAATTCATCCCGGTTTCTGT-3′ (Frøkjaer-Jensen et al., 2008). krSi3 was subsequently crossed into the unc-36(e251) null mutant, resulting in EN5403 krSi3[Punc-36::egfp::unc-36, cb-unc-119 (+)] II; unc-36(e251) III. EN5403 was further crossed with egl-19(ad695) and egl-19(n2368) to generate KAG200 and KAG214, respectively. All enzymes were obtained from Thermo Fisher Scientific (Waltham, MA, USA).

For morphological analysis, 20–30 L4 larvae were placed on freshly seeded nematode growth medium (NGM) plates at 15°C. The following day, individual worms were transferred into a small drop of M9 buffer and imaged on an Olympus IMT2 (Hamburg, Germany) microscope. The animal width and length were determined using ImageJ software (NIH Image, Bethesda, MD, USA): length was measured using a segmented-line spline fit from the tip of the nose to the region where the tail is 10 μm wide, and width was measured just posterior to the vulva. In pharmacological assays, 10 young adult worms were imaged after a 30 min incubation with 0.5 mmol l⁻¹ levamisole or 10 mmol l⁻¹ muscimol. For each animal the ratio width/length was determined before and after the drug treatment and the percentage of variation was calculated.

For locomotion tests, 10–20 L4 larvae were placed on freshly seeded NGM plates at 15°C. The following day, individual animals were transferred into a microtiter well (Corning Inc., Corning, NY, USA) containing 50 μl of M9 buffer on top of 2% agar in M9. After a 2 min recovery period, thrashes were counted for 1 min. A thrash was defined by Miller et al. (Miller et al., 1996) as a change in the direction of bending at the mid-body. All chemical were obtained from Sigma-Aldrich (St Louis, MO, USA).

Microdissection of C. elegans and electrophysiological methods were performed as described previously by Lainé et al. (Lainé et al., 2011). Membrane currents were recorded in the whole-cell configuration using an RK-400 patch-clamp amplifier (Bio-Logic, Claix, France). Acquisition and command voltage were controlled using pCLAMP9 software driving a 1322A Digidata (Molecular Devices, Sunnyvale, CA, USA). Data were analysed and graphed using pCLAMP9 and Origin (OriginLab, Northampton, MA, USA) software. The resistance of recording pipettes ranged between 2.5 and 3 MΩ. Recordings were performed after 1 min dialysis only on cells exhibiting resistances above 800 MΩ. Capacitance and resistance were not compensated.

All experiments were done at room temperature.

Young adults carrying krSi3 were transferred into a drop of M9 containing 20 mmol l⁻¹ sodium azide on a top of an agarose pad. Animals were imaged on an inverted LSM 5 Exciter laser scanning confocal microscope (Zeiss, Oberkochen, Germany) equipped with a ×63 oil immersion objective, ×1.4 numerical aperture. GFP was excited at 488 nm, and the emission from 505 to 600 nm was collected. Muscle cells were imaged in the head region of the animals and were analysed with the software ImageJ (NIH Image, Bethesda, MD, USA). Mean fluorescence of basal plasma membrane was quantified from an 18 μm² area in the middle region of the cells.

We thank the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health (NIH) Office of Research Infrastructure Programs (P40 OD010440) for strains, C. I. Bargmann for cDNA, M. Peters and T. Boulin for critical reading of the manuscript, and K. Gieseler and B. Allard for sharing facilities and scientific inputs.