A hyperpolarizing neuron recruits undocked innexin hemichannels to transmit neural information in Caenorhabditis elegans

We have previously shown that inx-4, a member of innexins, functions in the AFD thermosensory neurons to regulate TTX (32). In this study, we conducted a genome-wide survey of innexins, aiming to reveal the roles of the innexin family genes in the control of TTX (Fig. 1 and SI Appendix, Figs. S1 and S2). Of the 24 remaining innexin genes present in the C. elegans genome, we focused on the genes that are dispensable for viability and locomotion and are expressed in the nervous system (33–36). We examined the TTX behaviors of mutants for such innexin genes and found that all innexin mutants examined displayed the wild-type TTX phenotype (SI Appendix, Fig. S1). We also assessed the effect of AFD-specific overexpression of the innexin genes. Seven innexin genes—inx-1, inx-2, inx-4, inx-7, inx-10, inx-19, and unc-7—were reported to be expressed in AFD of the wild-type animals (37). We overexpressed each of these genes in AFD and observed that overexpression of unc-7 caused a thermophilic phenotype (Fig. 1B and SI Appendix, Fig. S2). While the wild-type animals cultivated at 20 °C stayed around the cultivation temperature, animals overexpressing unc-7 in AFD migrated toward a temperature region higher than the cultivation temperature (Fig. 1B). By contrast, animals overexpressing other innexin genes in AFD did not show abnormality in TTX (SI Appendix, Fig. S2).

Mutants harboring a presumptive null allele of unc-7, unc-7(e5), were uncoordinated (33), and hence, we could not test their TTX behavior. To address whether a loss of unc-7 function affects TTX, we attempted to generate strains lacking unc-7 only in AFD using the Cre/loxP system. We inserted two loxP sequences into the unc-7 locus and crossed with this unc-7(nj300) animal a strain carrying a single-copy insertion of the Cre gene fused with the AFD-specific gcy-8L promoter (Fig. 1C). This strain, hereafter referred to as unc-7(AFD KO) animals, displayed a cryophilic defect and migrated toward a lower temperature region than the wild-type animals did (Fig. 1D). Strains carrying unc-7(nj300) or the gcy-8Lp::Cre insertion alone did not show abnormality in TTX. The TTX defect of unc-7(AFD KO) animals was rescued by expressing unc-7 specifically in AFD (Fig. 1E). These results indicate that UNC-7 functions in AFD to regulate TTX.

A previous study showed that UNC-7 is required for presynaptic differentiation during synaptogenesis (38). This observation suggests that the TTX defect of unc-7(AFD KO) animals could be attributable to abnormalities of synaptogenesis of AFD. To assess whether UNC-7 is required for the development of the AFD thermosensory neurons, we examined the critical period of UNC-7 for the regulation of TTX by the Auxin Induced Degradation (AID) system (39). We expressed TIR1 in AFD of animals in which unc-7 was tagged with a degron sequence. While the animals cultivated with auxin throughout the development showed cryophilic defects, the animals cultivated without auxin did not show abnormality in TTX (SI Appendix, Fig. S3 A and B), indicating that the AID system successfully knocked down the UNC-7 activity. When the animals that had been cultivated in the absence of auxin for 2 d, at which most animals had grown to the second or third larval stage, were transferred onto plates with auxin, they displayed a cryophilic defect (SI Appendix, Fig. S3 C and D). Conversely, the animals treated with auxin only for the first 2 d of their development did not exhibit TTX defects. These results indicated that the activity of UNC-7 in AFD is required at or later than the third larval stage, the period at which AFD has connected to the majority of its postsynaptic neurons (40), suggesting that UNC-7 is required to function in the mature AFD neuron.

Innexins form two types of channels, gap junction channels and undocked hemichannels (41–43). To address whether UNC-7 acts as a gap junction channel or a hemichannel for the regulation of TTX, we utilized the cysteine mutants of UNC-7, UNC-7(Cysless; C173A, C191A, C377A, and C394A), and UNC-7(C191A). These cysteine residues were previously reported to be essential for the formation of gap junctions but were dispensable for hemichannel activity (44). We therefore asked whether the expression of unc-7(Cysless) or unc-7(C191A) in AFD could rescue the TTX defect of unc-7(AFD KO). However, we observed that these mutants of UNC-7 failed to localize to the axonal region of AFD unlike the wild-type UNC-7, and that the expression of these mutant unc-7 genes did not rescue the TTX defect of unc-7(AFD KO) (SI Appendix, Fig. S4). These observations suggested that the cysteine residues are required for the transport of UNC-7 in the AFD sensory neurons. Since the cysteine mutants of UNC-7 were reported to localize in the neuronal processes of the ventral nerve cord (44), UNC-7 might be transported by distinct mechanisms depending on the neuronal subtypes.

We also attempted to utilize another mutant form of unc-7, UNC-7(M121L) (45). A previous study showed that the gap junction activity of UNC-7 was required for the regulation of locomotion and that certain genomic constructs of UNC-7(M121L) did not rescue the locomotory defect of unc-7(e5), while others did, depending on the promoters used to drive unc-7(M121L) (45). We found that when an unc-7(M121L) cDNA fused with an AFD-specific promoter was introduced into unc-7(AFD KO) animals via an extrachromosomal array or a single-copy insertion, the animals no longer showed abnormality in TTX (SI Appendix, Fig. S5). However, our subsequent analysis indicated that the cDNA form of UNC-7(M121L) we used might retain some activity to form gap junctions (see below). These observations suggested that neither the cysteine mutant of UNC-7 nor UNC-7(M121L) could be utilized to address whether UNC-7 functions as a gap junction or a hemichannel to regulate TTX.

To address the type of channel that UNC-7 functions during TTX, we attempted to design a form of UNC-7 that would lose the ability to form gap junctions but retain the hemichannel activity. Since previous structural studies suggested that the formation of a gap junction is mediated through the interaction between the second extracellular loops (EL2) of the opposing hemichannels (46), we generated a chimeric form of UNC-7 whose coding sequence for the EL2 was replaced by that of mouse pannexin 1 (mPANX1), which belongs to the family of pannexin, the functional homolog of innexins (47–49) (Fig. 2A). The AFD-specific expression of this chimeric UNC-7 rescued the TTX defect of unc-7(AFD KO) animals, indicating that the activity of the chimeric UNC-7 is sufficient for regulating the TTX (Fig. 2B). By contrast, expression of a full-length mPANX1 specifically in AFD did not rescue the TTX defect of unc-7(AFD KO) (SI Appendix, Fig. S6).

To assess whether the chimeric UNC-7 retains hemichannel activity, we expressed the chimeric UNC-7 in Xenopus oocyte and confirmed that the chimeric UNC-7 exhibited voltage-dependent hemichannel currents comparable to that observed by the wild-type UNC-7 (Fig. 2C and SI Appendix, Fig. S7). To test whether the chimeric UNC-7 displays the gap junction activity, we asked whether the chimeric UNC-7 rescues the locomotion defect of unc-7(e5) mutants. The locomotory defect of unc-7(e5) animals is manifested in the uncoordinated propagation of the body-bending waves (13). To characterize the locomotion defect, we conducted curvature analyses of the animal posture during the locomotion (Fig. 2 D and E and SI Appendix, Fig. S8) and quantified the propagation of the body bends (Fig. 2F and SI Appendix, Fig. S9). These analyses captured the propagation of the body bends from the head to the tail in the wild-type animals but not in unc-7(e5) mutants (Fig. 2 E and F, SI Appendix, Figs. S8 A and B and S9 A and C, and Movies S1 and S2). We observed that the panneuronal expression of the wild-type UNC-7 restored to unc-7(e5) animals the continuous body-bending waves during forward movement (Fig. 2 E and F, SI Appendix, Figs. S8C and S9 A and C, and Movie S3), whereas the panneuronal expression of the chimeric UNC-7 could not (Fig. 2 E and F, SI Appendix, Figs. S8 D and E and S9 A and C, and Movies S4 and S5). To exclude the possibility that the lack of the rescuing activity of the chimeric UNC-7 is due to the absence of a functional copy of the chimeric UNC-7 in the extrachromosomal arrays, we transferred these arrays to an unc-7(AFD KO) background and confirmed that these transgenes rescued the TTX defect of unc-7(AFD KO) animals (SI Appendix, Fig. S10). These results indicate that the chimeric UNC-7 can not rescue the locomotion defects of unc-7(e5) and suggest that the chimeric UNC-7 can not form a gap junction in the locomotory circuit. We also observed that the chimeric UNC-7 loses the gap junction activity in AFD (see below). Together, these results indicated that the chimeric UNC-7 loses the ability to form gap junctions but retains the hemichannel activity and suggested that the hemichannel activity of UNC-7 in AFD is sufficient to regulate TTX.

In addition, we assessed whether the panneuronal expression of UNC-7(M121L) could rescue the locomotion defect of unc-7(e5). We found that introduction of UNC-7(M121L) restored the propagation of body bending waves to unc-7(e5) animals (SI Appendix, Figs. S8F and S9A and Movie S6), suggesting that UNC-7(M121L) retains the ability to form gap junctions in the locomotory circuit.

A previous report indicated that the UNC-7 hemichannel is required for the sensory transduction of mechanosensation in C. elegans (50). Given this report, we assessed whether UNC-7 regulates temperature sensing in the AFD thermosensory neurons. The AFD neurons were previously shown to increase their calcium concentration in response to temperature warming, and this calcium response occurs around the cultivation temperature (23–27). We therefore monitored calcium dynamics of AFD in wild-type, unc-7(e5) and unc-7(AFD KO) animals and found that the calcium response of AFD was normal in unc-7 mutants (SI Appendix, Fig. S11). We did not detect significant differences in the maximum value of ratio change or response temperature of AFD between wild-type animals and unc-7 mutants (SI Appendix, Fig. S11 C and D). These results indicated that unc-7 is not required for the temperature-evoked calcium response in AFD and suggested that UNC-7 regulates a process downstream of calcium influx in AFD.

To identify neurons to which UNC-7 transmits temperature information from AFD, we conducted single-cell ablation experiments by using reconstituted caspases (18, 51). We individually ablated a subset of the neurons reported to be chemical or electrical synapse partners of AFD (20) and asked which neurons, when ablated, could suppress the effect of the AFD-specific overexpression of UNC-7. We observed that while overexpression of unc-7 affected TTX behaviors of animals lacking AIB (Anterior Interneuron B), AIZ (Anterior Interneuron Z), AVE (Anterior Ventral Process E), or AWC (Amphid Wing Neuron C), the ablation of the AIY interneuron or the RMD (Ring Motor Neuron D) motor neuron suppressed the defect caused by the unc-7 overexpression (Fig. 3). These results suggested that UNC-7 transmits temperature information to the AIY and RMD neurons. Since the AIY interneurons have been shown to play pivotal roles in TTX (16, 18, 52), we focused on the neural pathway between AFD and AIY and further investigated the mechanisms of the behavioral regulation by UNC-7 hemichannels. We note that the RMD head motor neurons (19) were previously implicated in TTX (18), implying the possibility that the temperature information transmitted to RMD via UNC-7 might also be important for the navigation toward cultivated temperature.

Previous studies showed that multiple behavioral components such as turns, reversals, and curves are important for the regulations of TTX behavior. Many of these behavioral components are regulated by the AFD-AIY neural pathway, and each of the behavioral components contributes to TTX to a different degree (17, 18). To address which behavioral components are regulated by UNC-7, we performed high-resolution behavioral analysis using the Multi Worm Tracking (MWT) system (18, 31, 53) (Fig. 4 and SI Appendix, Figs. S12 and S13). We cultivated animals at 20 °C and monitored their behaviors within the temperature range from 18.5 °C to 21.5 °C (Fig. 4A). Consistent with previous reports (18, 31), we observed that wild-type animals displayed curving bias toward warmer temperature when moving down the thermal gradient and curved toward colder temperature when moving up the thermal gradient. This bidirectional curving bias presumably drives the animals toward the cultivation temperature. unc-7(AFD KO) animals showed defects in the regulation of curve and, in particular, displayed lower curving bias than that of the wild-type animals when moving down the thermal gradient (Fig. 4 B–D). This curving defect of unc-7(AFD KO) was rescued by expressing the wild-type or the chimeric UNC-7 in AFD. These results suggested that UNC-7 hemichannels promote the curving bias toward warmer temperature when animals are moving down the temperature gradient away from the cultivation temperature, thereby promoting migration toward the cultivation temperature. We did not observe significant defects attributable to the loss of unc-7 in other behavioral components examined, including omega turn, reversal, shallow turn, or speed (SI Appendix, Fig. S12).

To ask whether the regulation of the curving bias by UNC-7 involves the AIY interneuron, we analyzed behavioral components of the AIY-ablated animals (Fig. 4 E–G and SI Appendix, Fig. S13). We found that the AIY-ablated animals showed the opposite curving bias to that observed in the wild type: They curved toward colder temperature when moving down the thermal gradient and toward warmer temperature when moving up the thermal gradient (Fig. 4E). Importantly, the curving bias of the AIY-ablated animals when moving down the thermal gradient was not affected by unc-7(AFD KO) (Fig. 4G). Together, these results indicate that UNC-7 regulates curving bias through the AIY interneuron while animals are moving toward colder temperature and suggest that UNC-7 mediates transmission of temperature information from AFD to AIY.

We also note that unc-7(AFD KO) appeared to slightly affect the curving bias of the AIY-ablated background when animals moved up the thermal gradient, especially in the entry direction around 45° (Fig. 4E), This result implied that UNC-7 might transmit temperature information to neurons besides AIY.

To assess whether UNC-7 regulates the neuronal activity of the AIY interneuron, we analyzed calcium dynamics of AFD and AIY in freely behaving animals using the microscope with an automated tracking system (Fig. 5). Since our behavioral component analysis suggested that UNC-7 transmits temperature information from AFD to AIY when animals were migrating down the temperature gradient, we subjected the animals to a cooling stimulus, where the temperature was decreasing away from the cultivation temperature: 20 °C (Fig. 5B). We observed that in response to this cooling stimulus, the wild-type AFD neurons showed decreases in the calcium concentration (Fig. 5C). This observation is consistent with the previous report that cooling hyperpolarizes the AFD neurons (25). We also found that the majority of the AIY neurons of the wild-type animals responded to the cooling stimulus by decreasing the calcium concentration. This result corresponds to our previous report that the AIY responses correlate with the valence of thermal stimuli, with stimuli associated with positive valence evoking excitatory responses and stimuli with negative valence inhibitory responses (31). When unc-7(AFD KO) animals were subjected to cooling stimuli, the activity of AFD was decreased similarly to those observed in the wild-type animals (Fig. 5C). By contrast, decreases in the AIY calcium concentration were attenuated in unc-7(AFD KO) animals when compared to those in the wild-type animals (Fig. 5 C and D). We found that the proportion of the fluorescent changes below the median value of the wild-type responses was significantly lower in unc-7(AFD KO) animals (Fig. 5D). This defect was rescued by AFD-specific expression of the wild-type UNC-7 (Fig. 5 C and D) or of the chimeric UNC-7 (Fig. 5 E and F). We also monitored the neuronal activities in animals subjected to warming stimuli and found that unc-7(AFD KO) did not affect the AIY activity upon warming (SI Appendix, Fig. S14). These results indicated that the hyperpolarizing AFD neuron inhibits the activity of the AIY neuron in response to cooling stimuli and that UNC-7 hemichannels regulate this process.

In the above rescue experiments for the TTX defect of unc-7(AFD KO) animals, expression of the chimeric UNC-7 in AFD rescued the defect properly (Fig. 2B), whereas that of the wild-type UNC-7 resulted in a weak thermophilic phenotype (Fig. 1E). We also tested the effects of these UNC-7 transgenes in a wild-type genetic background and found that the wild-type UNC-7 caused a thermophilic defect while the chimeric UNC-7 did not (Fig. 6A). These results led us to hypothesize that the thermophilic defect caused by the AFD-specific overexpression of the wild-type UNC-7 was brought by excessive activity of the gap junction. Because our cell-ablation analysis also indicated that the effect of the wild-type UNC-7 overexpression is mediated by the AIY interneuron (Fig. 3C), we examined whether innexins expressed in AIY are required for the effect of the UNC-7 overexpression in AFD. We tested mutants for a subset of the innexin genes expressed in AIY, inx-1, inx-2, inx-19, and unc-9, and injected into these mutants a plasmid that expresses the wild-type UNC-7 in AFD. Animals mutant for unc-9 display an uncoordinated phenotype (33), which precludes us from analyzing their TTX behaviors. We therefore constructed animals lacking unc-9 specifically in AIY by the Cre/loxP system. These animals, hereafter referred to as unc-9(AIY KO), displayed a normal TTX behavior (SI Appendix, Fig. S15A). We observed that an AFD overexpression of the wild-type UNC-7 in inx-1, inx-2, and inx-19 mutants caused thermophilic phenotypes (SI Appendix, Fig. S15 B–D). By contrast, introduction of the UNC-7 construct into unc-9(AIY KO) animals did not cause a thermophilic behavior. Instead, unc-9(AIY KO) animals overexpressing the wild-type UNC-7 displayed a weak cryophilic phenotype (Fig. 6B). To exclude the possibility that unc-9(AIY KO) animals generated an unusual UNC-7 transgene, we transferred into unc-9(AIY KO) animals another unc-7 transgene that had rendered the wild-type animals a thermophilic defect (Fig. 3) and found that unc-9(AIY KO) animals carrying this transgene displayed cryophilc behaviors (SI Appendix, Fig. S15 E and F). These observations suggest that the wild-type UNC-7, when overexpressed in AFD, can form excessive gap junctions with UNC-9 in AIY and alter the TTX behavior. That the overexpression of the chimeric UNC-7 did not alter TTX indicates that the chimeric UNC-7 loses the ability to form gap junctions in AFD.

To further corroborate this finding, we evaluated the effect of UNC-7 overexpression on the neural activity of AIY (Fig. 6 C and D). We observed that the wild-type animals overexpressing the wild-type UNC-7 in AFD showed a stronger inhibition of the AIY activity upon cooling stimulus compared to their control animals without the UNC-7 overexpression. This inhibitory effect on the AIY activity was abolished in the chimeric UNC-7. Furthermore, unc-9(AIY KO) mutation completely suppressed the inhibition of the AIY activity mediated by the wild-type UNC-7 overexpression (Fig. 6 E and F). These results indicated that the wild-type UNC-7, when overexpressed in AFD, can form gap junctions with UNC-9 in AIY and that the chimeric UNC-7 loses the ability to form gap junctions in AFD. Together, these results strongly support that the hemichannel activity of UNC-7 in AFD is sufficient for the regulation of TTX.

Since the AIY responses in unc-9(AIY KO) animals expressing the wild-type UNC-7 were indistinguishable from those of the wild-type animals without UNC-7 overexpression, the cryophilic phenotype caused by the overexpression of UNC-7 in unc-9(AIY KO) animals might be due to excessive or ectopic gap junctions between AFD and neurons other than AIY.

Additionally, we observed that the AFD-specific overexpression of UNC-7(M121L) in a wild-type genetic background did not cause a thermophilic defect (Fig. 6A). We also evaluated the effect of UNC-7(M121L) overexpression on the neuronal activity of AIY (SI Appendix, Fig. S16 and Fig. 6 C and D). The AFD-specific expression of UNC-7(M121L) rescued the defects of the AIY response in unc-7(AFD KO) animals (SI Appendix, Fig. S16). We also observed that this transgene, when introduced into a wild-type background, caused a stronger inhibition of the AIY neuronal activity compared to the animals without the transgene, but the extent of the inhibitory effect by UNC-7(M121L) was smaller than that caused by the wild-type UNC-7 (Fig. 6 C and D). Together, these results suggest that the gap-junction activity of UNC-7(M121L) in AFD might be partially compromised.

We next address whether inhibition of the AIY activity by UNC-7 is mediated by controlling the synaptic transmission from AFD. The AFD-AIY synaptic transmission was shown to be brought about by two kinds of neurotransmitters, neuropeptides for excitatory signaling and glutamates for inhibitory signaling (28, 29). The glutamatergic transmission from AFD requires eat-4, which encodes a vesicular glutamate transporter (VGLUT) that transports glutamates into synaptic vesicles (29, 54). To assess whether UNC-7 functions through controlling synaptic release of glutamates, we conducted the epistasis analysis between unc-7 and eat-4. We generated animals lacking eat-4 specifically in AFD by the Cre/loxP system and found that eat-4(AFD KO) animals displayed thermophilic defects (SI Appendix, Fig. S17A), indicating that EAT-4 acts in AFD to regulate TTX behavior. Importantly, the unc-7(AFD KO) mutation affected TTX in animals lacking eat-4 in AFD (Fig. 7A). These results indicated that UNC-7 acts in parallel to EAT-4 to regulate TTX.

We also tested the possibility that UNC-7 is involved in the peptidergic signaling within AFD. Previous studies indicated that unc-31, which encodes the calcium-dependent activator protein essential for secreting neuropeptides, is required for the peptidergic signaling between AFD and AIY (28, 55). We tested the epistasis between unc-7 and unc-31 and found that while unc-31(AFD KO) did not display a TTX defect (SI Appendix, Fig. S17B), animals lacking both unc-7 and unc-31 showed a cryophilic phenotype stronger than that of unc-7(AFD KO) or unc-31(AFD KO) animals (Fig. 7B). These results suggested that UNC-31 plays at least a minor role in AFD to regulate TTX and that UNC-7 acts in parallel to UNC-31. Furthermore, we examined TTX behaviors of single mutants for a panel of the neuropeptide genes that are expressed in AFD (56, 57) and found that none of the mutants tested displayed a TTX defect (SI Appendix, Fig. S18). These results implied that UNC-7 functions to regulate TTX independently of the synaptic vesicle exocytosis.

C. elegans animals were cultivated at 20 °C on nematode growth media (NGM) plates seeded with Escherichia coli OP50 bacteria (33). N2 (Bristol) was used as the wild-type strain. Germline transformation was performed by microinjection as previously described (65). Genome editing was performed by the CRISPR Cas9 system as previously described (66, 67). We used the unc-7(isoform a) cDNA sequence for expression of the wild-type unc-7(+), chimeric unc-7, and unc-7(M121L). All strains used in this study are shown in SI Appendix, Table S1.

TTX assays were performed as previously described (68). A linear thermal gradient from 17 °C to 23 °C was formed on a TTX assay plate with a temperature steepness of approximately 0.5 °C/cm. The center of the plate was set at 20 °C. Animals cultivated at 20 °C were placed on the center of a TTX assay plate and were allowed to freely behave for 1 h. We divided the TTX assay plate into 8 sections along the temperature gradient, and the number of animals in each section was counted. We calculated TTX indexes using the formula shown in Fig. 1A.

Curvature analysis was performed essentially as described (69). Animals’ movements were tracked and recorded by using the MWT system (53). The MWT system extracted the coordinates along the body segment that equally divides the midline of the body. A one-dimensional median filter was applied to the coordinates of each point to reduce noise effects. Each segment angle was calculated with the MATLAB program and converted into a spatiotemporal color map (x-axis = time and y-axis = head–tail axis). Angles were assigned ascending numbers (1~9) along from head to tail. Dorsal and ventral bending is represented by blue and red color, respectively. Cross-correlation functions between time series of angle 2 and that of posteior angles (angle 2 ~ angle 8) were also calculated with the MATLAB program. Each cross-correlation value was normalized so that the autocorrelation value of angle 2 at lag 0 (s) is 1. We did not analyze angle 1 and angle 9 because of the noises in the coordinate acquisition of the end segment points. To quantify the propagation of the body bends, we defined and compared “cross-correlation ranges”, which were calculated by the following formula: the maximum cross-correlation value (angle 2 vs. angle 3, lag > 0)—the minimum cross-correlation value (angle 2 vs. angle 3, lag > 0). The value of cross-correlation range would become high if the body bending wave propagates along the head-to-tail axis, giving large absolute values in both positive and negative cross-correlations.

Whole-cell voltage-clamp recording of Xenopus oocytes was performed as previously described (70). The wild-type and chimeric unc-7 genes were cloned into pGEM-HeFx plasmids, and the cRNA was generated from each plasmid by using an RNA preparation kit (mMessage mMachine T7 Transcription Kit; Invitrogen) according to the manufacturer's protocol. The oocytes were collected from Xenopus laevis and then treated with collagenase solution, which contains 2 mg/mL collagenase type I (Gibco) dissolved in OR2 buffer (82.5 mM NaCl, 2 mM KCl, 1 mM MgCl₂, and 5 mM HEPES [adjusted to pH 7.5 with NaOH]) at 18 °C for 1.5 h. Forty nanograms of unc-7 cRNA was coinjected with 10 ng of antisense oligonucleotide DNA for Xenopus Cx38 into oocytes. The oocytes for negative controls were injected with the antisense DNA only. The oocytes were incubated at 18 °C for 2 to 3 d in ND96 buffer (93.5 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 2 mM MgCl₂, and 5 mM HEPES [adjusted to pH 7.5 with NaOH]). Hemichannel currents were recorded by using the iTEV 90 Multi-Electrode Clamp Amplifier (HEKA Elektronik). The oocytes placed in ND96 buffer without CaCl₂ were clamped at −30 mV initially and then subjected to 10 s voltage steps from −50 mV to +50 mV in 20 mV increments.

We generated unc-7(nj291) strain, in which a degron sequence was inserted at the unc-7 locus, and introduced into this strain a transgene that drives an AFD-specific expression of TIR1, an auxin-dependent E3 ubiquitin ligase. We added 1mM auxin indole-3-acetic acid (IAA) into NGM and TTX plates to allow TIR1 to bind degron-tagged UNC-7 and promote their ubiquitylation, leading to the degradation of degron-tagged UNC-7 by the proteasome specifically in AFD (39).

Calcium imaging recordings of the AFD neurons in immobilized animals were performed as previously described (26). Animals expressing calcium probe GCaMP6m together with TagRFP in AFD (67) were cultivated at 20 °C and were immobilized on 10% agarose pads by polystyrene beads. We subjected animals to stepwise temperature warming and recorded fluorescence images of the AFD cell body by EM-CCD camera with 400 ms pulsed illumination every 1 s. The fluorescence intensities were acquired from these images using MetaMorph software, and the ratio of fluorescence intensity (GCaMP3/TagRFP) was calculated. The ratio change was calculated by the following formula: (ratio—baseline ratio)/baseline ratio, where the baseline ratio value was the mean of the ratio values during the first 10 s before the beginning of the temperature change.

Calcium imaging of the AFD neurons and the AIY neurons in freely behaving animals was performed as previously described (31, 71). Animals expressing calcium probe YCX simultaneously in the AFD nuclei and the AIY neurons were cultivated at 20 °C and allowed to freely behave on 2% agarose pads. We subjected animals to warming or cooling stimuli and tracked their movement. We recorded fluorescence images of AFD and AIY under the epifluorescent microscope with SOLA LED light engine as a light source with 30 msec pulsed illumination every 1 s for 30 s.

The fluorescence intensities from these images were analyzed using the MATLAB program (71), and the ratio of fluorescence intensity (GCaMP3/TagRFP) was calculated. The intensities of AFD were acquired from its nuclei and those of AIY were acquired from a part of the axonal region shown in Fig. 5A. The ratio of fluorescence intensity (YFP/CFP) was standardized to 0 to 1 range, and the standardized ratio change was calculated by the following formula: (ratio—baseline ratio)/baseline ratio, where the baseline ratio value was the mean of the ratio values during first 10 s before the beginning of the temperature change.

MWT assays were performed as previously described (18, 31). Animals cultivated at 20 °C were placed on the TTX assay plate, and their behaviors were recorded using a CMOS camera (8 bits, 4,096 × 3,072 pixels; CSC12M25BMP19-01B, Toshiba-Teli) for 1 h. The MWT system extracted the coordinates of individual animal’s centroids and spines from the recording video. Using these data, the behavioral components of the worms on the temperature gradient within the range of 18.5 °C to 21.5 °C were analyzed by the MATLAB program previously described (18).

The Brunner–Munzel test, which does not require the assumption of normality and homogeneity of variance, was utilized to analyze differences in TTX indexes, voltage-clamp currents (Fig. 2C), cross-correlation ranges (SI Appendix, Fig. S9C), and AFD calcium response (SI Appendix, Fig. S11 C and D). To avoid type I errors, P-values in multiple comparisons were adjusted based on Bonferroni correction. Chi-square tests with Bonferroni correction were performed for multiple comparisons of the percentage of a categorical AIY response (Fig. 5 D and F and SI Appendix, Figs. S14C and S16B). The significance level in all statistics tests is 0.05.

A.N., M.W., A.O., I.M., and S.N. designed research; A.N., M.W., R.Y., H.K., and S.N. performed research; A.N., H.J.M., and S.N. analyzed data; and A.N., I.M., and S.N. wrote the paper.

The authors declare no competing interest.