Worms were cultivated on standard NGM agar plates seeded with E. coli OP50 at room temperature (~22°C). The Bristol N2 was used as wild-type strain, and other mutant strains and transgenic strains used in this study are listed in Supplementary Table S1.

Standard methods for molecular biology were used to construct plasmid DNAs. For the expression of the calcium indicator protein G-GECO1.2 (a kind gift from Takeshi Ishihara) or GCaMP6 (kind gift from Junichi Nakai), each coding sequence was inserted between the AgeI and EcoRI sites of the pPD95.79 vector (kind gift from Andy Fire). Then, the promoter region for cell-specific expression of the cDNAs was inserted between the SphI and XmaI sites of the resulting pPD95.79/G-GECO1.2 or pPD95.79/GCaMP6 plasmids. We used the following promoters for cell-specific expression: gcy-5 for ASER, gcy-7 for ASEL, and npr-9 for AIB. To generate plasmid DNAs for cell-specific UNC-13 expression, the full-length unc-13 cDNA fragment was amplified by overlapping PCR fusion using the following primers:

The resulting full-length UNC-13 cDNA fragment was inserted between the XmaI and KpnI sites of the pPD95.79 Vector. Then, each promoter sequence for ASEL or ASER was inserted between the SphI and XmaI sites of the pPD95.79/UNC-13 plasmid DNA. To generate a tetanus toxin expression plasmid, the TeTx::mCherry fragment from Pttx-3::TeTx::mCherry (a kind gift from Sreekanth Chalasani) was exchanged with the GCaMP6 sequence of the Pgcy-5::GCaMP6 plasmid DNA. For the expression of the glutamate-gated chloride channel glc-3, the glc-3 cDNA fragment was amplified by PCR fusion using the following primers:

The resulting full-length glc-3 cDNA was inserted between the BamHI and KpnI sites of the pPD95.79/Venus Vector. Then, the npr-9 promoter region was inserted between the SphI and BamHI sites of the pPD95.79/glc-3 plasmid DNA. For the expression of an AMPA-type ionotropic glutamate receptor glr-1, the glr-1::GFP fragment was excised by KpnI and EcoRV digestion from the Pttx-3::glr-1::GFP plasmid (a kind gift from Takaaki Hirotsu), and then inserted into the Pgcy-7::R-GECO1 plasmid DNA. Finally, the npr-9 promoter region was inserted between the KpnI and ApaI sites. For the rescue experiments of glc-3 or glr-1 mutants, Venus or GFP sequence was removed from each expression plasmid DNA by using in-Fusion reaction (Takara), respectively.

For the generation of transgenic animals, the resulting plasmid DNAs were injected into N2 (Bristol) or mutant animals using a standard microinjection method (Mello et al., 1991). Details of the strains used in this study are listed in Supplementary Table S1.

Calcium imaging was performed as described previously (Kuramochi and Doi, 2017). Adult transgenic worms were used for imaging. Worms were immobilized in a microfluidic device fabricated from polydimethylsiloxane (PDMS; Chronis et al., 2007). The microfluidic device was set on an inverted fluorescent microscope (Olympus IX71), and time-lapse images were captured (10 frames/s) using an ORCA-Flash 4.0 CCD camera (Hamamatsu Photonics) controlled by HCImage software (Hamamatsu Photonics). Recordings started within 5 min after removal from food. The following buffers for calcium imaging were used: 5 mM KPO₄ (pH 6.0), 1 mM CaCl₂, 1 mM MgSO₄, including 0 or 50 mM NaCl for the stimulation. All the buffers were adjusted to 350 mOsmol/L H₂O with glycerol (Oda et al., 2011). The patterns of salt stimulation were automated using the Perfusion Valve Controller System VC-6M (Warner Instruments) and Arduino microcontroller to control solenoid valves (Arduino SRL) with a pre-generated sequence. We used ΔF/F₀ to indicate fluorescence intensity change. F₀ was defined as the average fluorescence in a 5 s window before stimulation. After background subtraction, the total fluorescence intensity was measured from individual regions of interest (ROIs) in each neuron. An animal was imaged twice, with a 30 s interval between the first and second observation. For the comparison of ΔF/F₀ among genotypes, “response (ΔF/F₀)” was calculated using the average fluorescence in a 20 s window when NaCl concentration is decreased or in a 10 s window when NaCl concentration is increased.

L4 larvae were mounted on a 1.5% agarose pad with 20 mM sodium azide in M9 solution for anesthesia. Images were acquired on an inverted confocal microscope (Nikon A1, Nikon) with a 60× objective lens, and were analyzed by NIS-Elements C/NIS-Elements C-ER and ImageJ software, respectively. The Z-stack image was acquired from the whole animals expressing each fluorescent fusion proteins. Co-localization between GLR-1 or GLC-3 GFP fusion proteins and a presynaptic mCherry::RAB-3 fusion protein was quantified by counting the number of GFP pixels overlapping with mCherry signal in a single z-axis flame. The number of pixels in each frame of Z-scan was averaged in each animal and used for quantitative analyses.

All data, except for the ASE activity, did not show gaussian distribution based on the Shapiro–Wilk test. Thus, a non-parametric Wilcoxon rank sum test was used to evaluate the median difference in calcium response.

C. elegans detect NaCl concentration gradients via the ASEL and ASER sensory neurons and move to a preferential NaCl condition by using several behavioral strategies (Pierce-Shimomura et al., 1999; Iino and Yoshida, 2009). Both ASEL/R have synaptic connections to downstream first-layer interneurons including AIB neurons (Figure 1A). To understand the role of presynaptic excitatory/inhibitory switch in postsynaptic responses of AIB neurons, we recorded fluorescent changes of the genetically-encoded calcium indicator G-GECO1.2 (Zhao et al., 2011) that is specifically expressed in either pre- or postsynaptic neurons. Consistent with previous reports (Suzuki et al., 2008), the ASEL neuron did not show any response to a downstep in NaCl concentration (from 50 mM to 0 mM NaCl), whereas the ASER neuron showed a large, long-lasting response to the downstep (Figure 1B). Conversely, the ASEL neuron showed a fast calcium response to an upstep in NaCl (0 mM to 50 mM), and its response immediately decayed to the steady state level. The ASER calcium response decayed immediately after an upstep in NaCl concentration (Figure 1D). The AIB neurons showed a similar response pattern as the ASER: their responses rose slowly after a downstep in NaCl concentration, and after the peak level, slowly decayed during the exposure to 0 mM NaCl (Figure 1B). In response to an upstep in NaCl concentration, the AIB calcium level immediately decayed to the steady-state level (Figure 1D). Thus, as shown in previous report (Kunitomo et al., 2013; Wang et al., 2017), AIB neurons show similar neuronal responses to both up and down changes in NaCl concentration as the ASER neuron.

Sensory neurons, including AFD, ASE, ASH, ASJ, AWB, and AWC, are thought to function as salt-sensing neurons (Thiele et al., 2009; Zaslaver et al., 2015), and AIB interneurons receive synaptic inputs from many sensory neurons. To examine whether the AIB neuronal responses to changes in NaCl concentration are regulated by ASE neuronal activity specifically, we recorded the AIB response in che-1(p679) mutant worms that specifically lacks ASE neurons (Chang et al., 2003; Uchida et al., 2003). Here we found that the AIB neurons in che-1 mutants showed no significant calcium response to either a downstep or upstep in NaCl concentration (Figures 1B–E). These results indicate that ASE neurons strongly affect AIB activity in response to NaCl concentration changes; other sensory neurons probably have a weak or no effect on AIB activity as they could not compensate for a loss of ASE neurons.

Our results suggest that ASE neurons are the main regulators of AIB activity during salt-chemotaxis. To confirm the role of each ASE neuron in the AIB response to changes in NaCl concentration, we analyzed the neurotransmission from the ASEL or the ASER neuron to AIBs. First, we monitored AIB responses to changes in NaCl concentration in the synaptic transmission-defective mutant unc-13 (Richmond et al., 1999). unc-13 encodes a protein required for vesicle priming at the presynapse and mutations in this gene cause severe defects in neurotransmitter release. The AIB neurons in unc-13(e312) mutants did not show any response to either a downstep or upstep in NaCl concentration (Figures 2A–F). Therefore, together with our findings in the che-1 mutant, these results suggest that synaptic transmission from ASE neurons is required to regulate AIB activity when NaCl concentrations change.

We further examined how the two distinct ASEL and ASER neuronal responses to changes in NaCl concentration cooperatively modulate AIB activity. To answer this question, we monitored AIB responses when only ASEL or ASER synaptic transmission was functional. To this aim, UNC-13 protein was specifically expressed in either ASEL or ASER neurons in the unc-13(e312) mutant background to recover synaptic transmission from one of these neurons. We then monitored AIB calcium responses in these cell-specific synaptic rescue worms. With regards to ASER function, we found that ASER activates AIBs only when the NaCl concentration is decreased. In worms in which transmission from the ASER was rescued by cell-specific expression of UNC-13 in the ASER neuron, we observed that the calcium response in AIB neurons increased just after the NaCl concentration is decreased (Figures 2A–C). However, a strong inactivation following an increase in NaCl concentration was not observed in this rescue animal. These results suggest that the ASER neuron may activate AIBs when NaCl decreases, but likely has no inhibitory effect on AIB activity when NaCl increases. To further confirm these results, we also monitored AIB activity in transgenic worms in which the tetanus toxin light chain from Clostridium tetani (TeTx) was expressed in a cell-specific manner to block synaptic transmission. TeTx expression reduces presynaptic vesicle release by cleaving synaptobrevin/VAMP protein, a core component required for synaptic vesicle fusion (Schiavo et al., 1992). The transgenic worms expressing TeTx specifically in the ASER elicited a significantly weaker calcium response in AIBs during a downstep of in NaCl concentration compared to control worms (Figures 2G–I). Therefore, we conclude that the ASER signal probably excites AIB neurons when the NaCl concentration decreases.

With regards to ASEL function, we found that ASEL signaling provides an inhibitory signal to inactivate AIB activity. AIB neurons were rapidly inactivated during an upstep in NaCl concentration, and this inactivation was also observed when the synaptic transmission from the ASEL neuron was specifically rescued upon expressing UNC-13 in the ASEL neuron of unc-13 mutants (Figures 2D–F). Conversely, AIB activation in response to a downstep in NaCl concentration was not observed in this transgenic worm, suggesting that the ASEL may not have an excitatory role in AIB activation. Therefore, we conclude that the ASEL signal likely inactivates AIBs in response to ASEL transient activity when NaCl concentrations increase.

Thus far, we have shown that ASEL and ASER activities have opposing effects on AIB activity in response to changes in NaCl concentration (Figures 1B,D). Both ASEL and ASER neurons connect to AIB neurons via chemical synapses and probably release glutamate as a neurotransmitter (Serrano-Saiz et al., 2013). To test how glutamate delivers both as an excitatory and inhibitory signal to AIBs, we monitored the AIB calcium response in the eat-4(ky5) mutant. eat-4 encodes a vesicular glutamate transporter in C. elegans, and mutations in this gene cause lack or decrease of glutamate in synaptic vesicles (Lee et al., 1999). We found that eat-4(ky5) mutants did not show a clear AIB response to either an increase or decrease in NaCl concentration (Figure 3). These results suggest that the glutamate signal from each ASE neuron likely generates both excitation and inhibition in AIBs.

AMPA-type ionotropic glutamate receptors can act as excitatory postsynaptic receptors for odor-evoked responses in AIB neurons (Chalasani et al., 2007). As such, we questioned whether the same AMPA-type glutamate receptors also mediate AIB activity in response to changes in NaCl concentration. Here, we used glr-1mutants in which expression of the non-NMDA glutamate receptor is disrupted. In glr-1(n2461) mutants, the averaged AIB calcium response was weaker than that elicited in wild-type worms but higher than that in eat-4 mutant worms (Figures 3A–C). Although several studies have reported that AMPA-type excitatory receptors might modulate AIB activity via glutamate release (Chalasani et al., 2007; Piggott et al., 2011), we conclude that both GLR-1 and other receptors coordinately contribute to AIB neuronal excitation during exposure to decreases in NaCl concentration. This lower calcium responses in AIB neurons of glr-1 mutants was rescued by the AIB-specific expression of GLR-1 receptor, suggesting the cell-autonomous regulation of AIB neuronal activity by GLR-1 (Supplementary Figure S1). GLR-1 is not required for inactivation in response to increase of NaCl concentration because glr-1 mutant animals showed similar calcium responses with wild-type animals upon increase of NaCl concentration (Supplementary Figure S2).

The metabotropic G-protein-coupled glutamate receptors (encoded by three mgl genes mgl-1, mgl-2 and mgl-3) are thought to be expressed in AIB interneurons and act as excitatory postsynaptic receptors (Dillon et al., 2006). We next studied whether these metabotropic glutamate receptors contribute to AIB activity using mgl-1(tm1811), mgl-2(tm355), and their double mutant worms. Here, we found that the calcium responses in AIB neurons of the double mutants rose slowly to a peak following a downstep in NaCl concentration. On the other hand, the calcium responses in wild-type animals rose rapidly to a peak within 20 s after a downstep in NaCl concentration (Figures 3D,E). Furthermore, glr-1, mgl-1 and mgl-2 triple mutant animals showed significantly lower calcium responses than mgl-1; mgl-2 double mutant animals after a downstep in NaCl concentration (Figures 3D,E). These results suggest that both AMPA-type and metabotropic glutamate receptors are required for rapid AIB neuronal activity after a downstep in NaCl concentration.

As for inhibitory synaptic signaling via glutamate, a glutamate-gated chloride ion channel (encoded by four glc genes and two avr genes in C. elegans) can act as an inhibitory postsynaptic receptor (Cully et al., 1994; Hart et al., 1995; Maricq et al., 1995; Yates et al., 2003; Dillon et al., 2006). Mutations in glc-3, a subunit of the glutamate-gated chloride ion channel, can cause decreased activity of the glutamate-gated chloride ion channel (Cully et al., 1994). We hypothesized that the ASEL neuron may inhibit AIB activity via this glutamate-gated chloride ion channel on AIB neurites. To test this hypothesis, we examined the calcium response in AIB neurons of glc-3(ok321) mutants during an increase in NaCl concentration. We found that AIB responses in glc-3(ok321) mutants showed a slightly slower decay in activity than the responses in wild-type animals. However, AIB activity returned to baseline faster in glc-3(ok321) mutants than eat-4 mutants (where glutamate in synaptic vesicles is lost; Figures 3F–H). This slower inactivation was rescued by the AIB-specific expression of GLC-3, suggesting the cell-autonomous function of GLC-3 in AIB neurons (Supplementary Figure S1). These results suggest that the glutamate-gated chloride ion channel is required for AIB neuronal inactivation in response to an increase in NaCl concentration. The GLC-3 channel does not function for the fast excitation of AIB neurons when NaCl concentration is decreased (Supplementary Figure S2).

Our results suggest that glutamate released by the ASER neuron generates an excitatory response in AIB neurons, probably via AMPA-type and metabotropic G-protein-coupled glutamate receptors. Conversely, glutamate released by the ASEL neuron may cause AIB inhibition via the glutamate-gated chloride ion channel. Because both neurons use glutamate as a transmitter, we wondered whether each synaptic site (or receptor) is randomly located on the AIB neurites or any positional arrangements exist. To answer this question, we analyzed the localization patterns of both the AMPA-type glutamate receptor GLR-1 and the glutamate-gated chloride ion channels GLC-3 on the postsynaptic AIB neurite. Because the process of each AIB neuron runs around the nerve ring from ventral to dorsal at the side of cell body and comes back to ventral at opposite side, we divided the neurite into two regions; “proximal region” which is the neurite from the cell body to dorsal midline, and “distal region” from dorsal midline to the tip of neurite (Figure 4A). In the transgenic animals expressing a GLR-1::GFP fusion protein specifically in the AIB, the GFP signal seemed to be localized uniformly across the neurite due to many small puncta (Figure 4B and see below for further observation). In the transgenic worms expressing a GLC-3::Venus in the AIB, however, the Venus signal was observed as several size of puncta (Figure 4C). Furthermore, we found that the GLR-1 fusion protein predominantly localized to the proximal region of the AIB cell body (Figures 4D,E), whereas the GLC-3 fusion protein was predominantly localized distal to the AIB cell body (Figures 4F,G). Thus, the localization patterns of GLR-1::GFP and GLC-3::Venus were strikingly different on the AIB neurite.

Although the AMPA-type glutamate receptor and the glutamate-gated chloride ion channel seem to be localized at distinct regions on the AIB neurite, we could not confirm whether each receptor really localizes at distinct regions on the neurite due to the localization of many fluorescent puncta on the neurite. We also could not confirm whether those localizations really corresponded to their specific functions as inhibitory (with ASEL) or excitatory (with ASER) synapses. To investigate these questions, we first observed the co-localization of a presynaptic vesicle-associated protein, RAB-3, on the ASEL or ASER and the postsynaptic GLR-1 on the AIB (Figures 5A,B). The positions of mCherry::RAB-3 on the ASEL and GLR-1::GFP on the AIB neurite were not well-correlated with each other, only a few GFP puncta overlapped with the mCherry signal on the ASEL neurite (Figures 5C,E,F). On the other hand, the positions of that on the ASER and that on the AIB neurite closely localized each other (Figures 5D–F). Especially, many synapses of the ASER localize to the proximal region of the AIB neurite (Figure 5E). The overlap between ASER and AIB was significantly larger than that between ASEL and AIB (Figure 5F). These results suggest that the AMPA-type glutamate receptor GLR-1 on the AIB predominantly localize on the proximal neurite to form synapses with the ASER and can receive excitatory glutamate signals from it.

Contrary to the ASER-to-AIB excitatory signal, our calcium imaging data showed that ASEL glutamatergic signaling inhibits AIB activity (Figure 3). We thus hypothesized that the ASEL and AIB may form inhibitory synapses, and that both the presynaptic and postsynaptic components of the synapse can closely localize with each other. To prove this, we observed the localization pattern of mCherry::RAB-3 on ASEL and ASER neurites and GLC-3::Venus on AIB neurites. We found that the puncta of mCherry::RAB-3 on the ASEs were mainly facing the proximal region of the AIB neurons whereas the puncta of GLC-3::Venus were localized strongly on the distal region of AIB neurons (Figures 4G, 6A,B). These results suggest that the presynaptic sites labeled by mCherry::RAB-3 on ASE neurons and postsynaptic GLC-3 accumulation on AIB neurons are distantly located, compared to the sites of RAB-3 on the ASE neurons and postsynaptic GLR-1 on the AIB (Figures 5A,B). So, we found that only a small number of mCherry::RAB-3 puncta on the ASEL was closely aligned with the GLC-3::Venus on the AIB (Figures 6C,E). On the other hand, mCherry::RAB-3 on the ASER was distinct from the distribution of GLC-3::Venus on the AIB, few overlap was observed (Figures 6D,E). The averaged overlap between ASEL and AIB was significantly larger than those between ASER and AIB (Figures 6E,F). These results suggest that a few but clear synapses are formed between the ASEL and AIB, and at those synapses, the glutamate-gated chloride ion channel GLC-3 on the AIB receives inhibitory glutamate signals from the ASEL.