Neuroligin tuning of pharyngeal pumping reveals extrapharyngeal modulation of feeding in Caenorhabditis elegans

Complex cognitive function is underpinned by discrete microcircuits that are integrated to control behaviour in a manner appropriate for the environmental context (LeDoux, 2000; Barkus et al., 2010; O'Connor et al., 2010). Arguably one of the most important environmental stimuli for an animal is its food. This relationship has been subjected to extensive investigation for the model genetic animal Caenorhabditis elegans with a view to providing insight into the fundamental behavioural relationship of an organism with its food source (Shtonda and Avery, 2006; Greene et al., 2016; Thutupalli et al., 2017). For C. elegans, this food is bacteria, which it ingests by increasing its rate of pharyngeal pumping in a process controlled by pharyngeal motor neurons that direct coordinated cycles of pharyngeal radial muscle contraction and relaxation (Franks et al., 2006). Intrinsic control of the pharynx modulated by MC and M3 neurons is required for an efficient intake of bacteria (Avery, 1993a; Avery, 1993b). In addition, there is an extrinsic control pivoting on the single RIP neuron that connects the pharyngeal circuit to the central nervous system. Worms in which RIP has been ablated pump normally on food. This suggests that neuroanatomical connectivity is not essential for the regulation of feeding and that neurohormonal signalling is sufficient to sustain pharyngeal pumping in response to food (Albertson and Thomson, 1976; Dallière et al., 2016). These mechanisms provide a means whereby the rate of pharyngeal pumping is influenced by the presence of food (Horvitz et al., 1982; Rogers et al., 2001).

MC neurons act as pacemakers to control the rapid activation of the pumping rate in C. elegans (Avery and Horvitz, 1989; Raizen and Avery, 1994). This is facilitated by the motorneuron M3 releasing glutamate onto the muscle to facilitate its relaxation. This relaxation of the metacorpus and isthmus is critical in trapping the bacteria in the corpus (Albertson and Thomson, 1976). Electrophysiological recording of the pharyngeal neural network via electropharyngeogram (EPG), provides a measurement of fine features that underpin pumping (Avery et al., 1995). This allows recording of the electrophysiological signature reflecting the activity of both M3 and MC neurons and has identified neuronal signalling and neuromodulatory components of the pharyngeal microcircuit (Raizen and Avery, 1994; Dillon et al., 2009). Two serotonergic neurons play a key role in mediating the pharyngeal response to the availability of food: the pharyngeal secretory neuron NSM and the chemosensory neuron ASD (Li et al., 2012). The neuron NSM is embedded within the pharyngeal circuit and implicated in the acute response to newly encountered food (Iwanir et al., 2016). Although functional NSM neurons are required for bursts of fast pumping in the presence of food, serotonin production in HSN neurons is sufficient to trigger food-dependent pumping dynamics when serotonin production in NSM is perturbed. This suggests that the HSN neurons, located in the middle of the worm body and sending no processes into the pharynx, are involved in the regulation of feeding (Lee et al., 2017). Finally, the role played by ADF in feeding regulation is related to sensing fluctuations in the environment (Lee et al., 2017).

Food availability also modifies the worm's locomotory behaviour and induces high probability of dwelling. This behaviour promotes the worm’s residence on a patch of food. This dwelling behaviour is interspersed with roaming, a behaviour that allows the worms to explore their environment (Flavell et al., 2013). Thus, C. elegans exhibits a repertoire of behavioural responses to food that requires coordination of precisely organised neural circuits. How the upstream sensory inputs are integrated and modulated to bring about this coordinated behavioural response is important in understanding the systems-level control of foraging and feeding.

In this study, we have focused on the role of neuroligin in the regulation of this food-dependent behaviour. Neuroligin is a synaptic protein intimately involved in the molecular organization of discrete circuits (Calahorro, 2014; Bemben et al., 2015). In particular, in its absence, there is an impairment in the organization of the excitatory–inhibitory balance in neuronal circuits (Varoqueaux et al., 2006). nlg-1 is the C. elegans orthologue of human NLGN1; the neuroligin-1 protein is present at most synapses and has a conserved organization in its functional domains (Calahorro, 2014). C. elegans nlg-1 mutants show a range of sensory deficits (Calahorro et al., 2009; Hunter et al., 2010; Calahorro and Ruiz-Rubio, 2012). These investigations highlight the evolutionarily conserved role of neuroligin in the organisation of higher-level function, which underpins the association of neuroligin mutations with autism spectrum disorder in humans. Recent studies highlight the significance of neuroligin in the molecular organization of synapses; mutations in neuroligin impart synaptic dysfunction that, in turn, has consequences for discrete microcircuit function (Zhang et al., 2018; Polepali et al., 2017). This is particularly pertinent in autism spectrum disorders where the behavioural complexity underlying such disorders is related to ‘fractionable’ circuits (Williams and Bowler, 2014). Discrete disruption of individual circuits could be used to modify the function in downstream circuits not reliant on neuroligin, leading to an altered synaptic function and consequently inducing neuro-atypical behaviours (Rothwell et al., 2014; Bey et al., 2018). This principle is reinforced in animal models in which neuroligin mutation may selectively impact disease-related behaviour.

Indeed, C. elegans provide an excellent opportunity to delineate the role of neuroligin in the context of discrete functional microcircuit organization of sensory-driven behaviour. This is because C. elegans express one orthologue of neuroligin and can be investigated using behavioural outputs with well-established microcircuit dependence. In C. elegans, nlg-1 is selectively expressed in a subset of neurons. Here, we provide evidence that neuroligin functions in an extrapharyngeal circuit that coordinates a feeding response by modulating the cue promoting the pharyngeal function. This further supports its fundamental role in the integration of sensory information and co-ordination of activity via discrete microcircuits.

C. elegans strains were maintained under standard conditions (Brenner, 1974; Molin et al., 1999). Worms were synchronized by picking L4 larval stage to new plates 18 h prior to performing behavioural experiments. The strains used were: nlg-1 (ok259) X (6× outcrossed); nlg-1 (ok259) X, Ex [pPD95.77 (Pnlg-1::nlg-1 Δ#14); Pmyo-3::gfp]; nlg-1 (ok259) X, Ex [pPD95.77; Pmyo-3::gfp]; RM371, pha-1(e2123), sIs [Pnlg-1::yfp+pCI(pha-1)]; MT6308 eat-4 (ky5) III (2× outcrossed); CB156, unc-25 (e156) III (1× outcrossed); CB382, unc-49 (e382) III (2× outcrossed).

We identified neurons that express nlg-1 reporter constructs based on their cell position and co-labelling with mCherry, red fluorescent protein (RFP) or DsRed-based reporters. The co-labelled reporter transgenes were: (1) oyls 51 (also known as Psrh-142::rfp). A specific marker for ADF chemosensory neuron (Xu et al., 2015); (2) oyIs44, oyIs44 [Podr-1::rfp+lin-15(+)]. A specific marker for AWB, AWB odor sensory neurons (Sarafi-Reinach et al., 2001); (3) ofEx205 [pBHL98 (lin-15ab+); Ptrx-1::DsRed]. A specific marker for ASJ sensory neuron (Miranda-Vizuete et al., 2006); (4) dbEx719 [Pnpr-5::mCherry+Punc-122::gfp]. A specific marker for sensory neurons ADF, ASE, ASG, ASI, ASJ, ASK, AWA, AWB, IL2; interneurons AIA, AUA; phasmids sensory neurons PHA, PHB (Cohen et al., 2009); (5) sIs [Peat-4::mrfp]. A specific marker for neurons expressing the vesicular glutamate transporter eat-4 (Serrano-Saiz et al., 2013); (6) otIs151 [Pceh-36::rfp+rol-6 (su1006)]. A specific marker for chemosensory neurons AWC and ASE (Chang et al., 2003); (7) otIs181 [Pdat-1::mCherry+Pttx-3::mCherry]. Specific markers for both dopaminergic neurons (Pdat-1), ADE, PDE, CEP; and for the sensory interneuron AIY (Pttx-3) (Flames and Hobert, 2009); (8) vsIs108 [Pthp-1::rfp+Punc-13::gfp+lin-15(+)]. A specific marker for HSM and NSM serotonergic neurons (Tanis et al., 2008).

The genomic sequence of nlg-1 was used to design the following specific primers: nlg-1 forward, 5′-GGCATggatccCATTTATCTTCTTCTCC-3′ and nlg-1 reverse, 5′-GTAgaattcGTTAGACCTGTATCTCTTCC-3′. These were used to PCR amplify the coding region corresponding to the Δ#14 transcript, the dominant neuroligin isoform in the adult stage (Calahorro et al., 2015), from the cDNA clone yk1657a10 (from Yuji Kohara, National Institute of Genetics, Mishima, Japan).

The 2.5 kb promoter region of nlg-1 was amplified using the Pnlg-1 forward 5′-TtctagaCATATTTTTGGGGAGGCTTTC-3′ and Pnlg-1 reverse 5′-GAAGGAGAAGAAGATAAATGggatccATGC-3′ primers from the C40C9 cosmid clone (from Sanger Institute, UK). The promoter sequence was cloned using the XbaI/BamHI site in a pDD95.77 vector where the GFP was removed. Finally, the nlg-1 Δ#14 sequence was fused to the nlg-1 promoter using the BamHI/EcoRI sites. The promoter was authenticated by sequencing using the following primers: 5′-AAGCTTGCATGCCTGCAGGTCGAC-3′, 5′-AAGCTTGCATGCCTGCAGGTCGAC-3′; and the following primers for nlg-1: 5′-AATGCAGACTGGAGAAACTTTG-3′, 5′-CTATTACCAGAGCAAGACGATG-3′, 5′-GCTTCTCTGGTTTCTCTTCTTATG-3′, 5′-CTGTTTCCTTTCCATTCTTGTGC-3′, 5′-AGAATGGAAAGGAAACAGAGCC-3′, 5′-GTGCGATGCGGATAGTAAGGG-3′.

nlg-1 (ok259) X 1-day-old adults were microinjected with nlg-1 plasmid (prepared as described above; 50 ng/μl) together with the ‘marker’ plasmid Pmyo-3::gfp (30 ng/μl), as previously described. For controls, nlg-1 (ok259) X animals were microinjected with ‘empty rescue’ plasmid and the marker plasmid Pmyo-3::gfp at the same concentration used for generating the rescue transgenic lines. Microinjection mixes were prepared using double-distilled water.

Feeding behaviour was visually scored by counting the number of pharyngeal pumps for 1 min using a binocular dissecting microscope (Nikon SMZ800; ×63). A single pharyngeal pump was defined as one contraction–relaxation cycle of the terminal bulb of the pharyngeal muscle. To count pharyngeal pumps on food, 1-day-old adult worms (L4+1) grown at 20°C, in standard NGM E. coli OP50 plates, were gently picked onto the middle of the bacterial lawn (OD₆₀₀ of 0.8 AU seeded the day before). After 10 min, the pharyngeal pumping was recorded using a hand counter. Five consecutive measurements (1 min each) were made and the mean of pharyngeal pumping rate for this time period was then calculated.

To count pharyngeal pumps off food, 1-day-old adult worms grown at 20°C, in standard NGM E. coli OP50 plates, were gently picked onto the middle of a 9-cm-diameter non-food plate and left for 5 min to clean themselves of attached bacteria. Animals were finally transferred onto a second clean non-food plate, where pumping rate was scored after 10 min. Five consecutive measurements (1 min each) were made and the mean of pharyngeal pumping rate for this time period was then calculated. During this experiment, worms dried out after migration off the edge of the agar plate were not counted.

NGM plates (9 cm) were poured the day before the assay and kept overnight at room temperature. These plates were seeded with a spot of 50 µl E. coli OP50 at an optical density of 0.8 AU (OD_600nm) displaced 2 cm from the edge of the plate and incubated overnight. The day before the assay 50–100 L4 animals were picked onto NGM E. coli OP50 plates and kept at 20°C for 18 h. These L4+1 worms were washed twice in M9 buffer to remove residual bacteria adhered to the body. The washed worms were added in a minimal volume 2 cm from the edge opposite the food patch. The number of animals reaching food was recorded every 10 min for 2 h. The cumulative number that reached the food spot was calculated.

The extracellular electrophysiological activity in the pharyngeal network was measured using the microfluidic device Neurochip as previously described (Hu et al., 2013). L4+1 worms were loaded into the NeuroChip in Dent's saline (in mmol l⁻¹: 10 D-glucose, 140 NaCl, 1 MgCl₂, 3 CaCl₂, 6 KCl, and 10 HEPES; pH 7.4) supplemented with 0.01% BSA (w/v). Extracellular voltage recordings were made in ‘bridge’ mode, and the extracellular potential was set to 0 mV using the voltage offset immediately prior to recording. Data were acquired using Axoscope (Axon Instruments) and stored for subsequent offline analysis. EPG traces were annotated with AutoEPG software (Dillon et al., 2009). Statistical analysis was performed using one-way ANOVA with Bonferroni multiple comparisons post-test. Recordings were made in both Dent's saline alone or when loaded into the device in Dent's saline and 5 mmol l⁻¹ 5-hydroxytryptamine (5HT; serotonin creatinine sulfate monohydrate used for electrophysiology experiments was obtained from Sigma-Aldrich). EPGs were recorded for 3–5 min from when the animal was trapped in the recording microchannel, conditions under which the 5 mmol l⁻¹ serotonin induced pumping (Hu et al., 2013). A single pump EPG recording corresponds to the coordinated contraction and relaxation of the pharynx. Recordings were analysed offline using a signal peak detection and measurements made to define: frequency; average duration of single EPGs; number of P waves per EPG, regulated by activity of the inhibitory motorneuron M3; shape of the EPG using the R to E ratio, closely associated with muscle depolarization (contraction) at the beginning of a pump and repolarization (relaxation) at the end of a pump; pattern of activity, under neural control.

A Nikon Eclipse (E800) microscope was used to image fluorescence from cells expressing GFP or RFP proteins under cell-specific promoters (see above). Worms were placed in 0.5 μl M9 buffer containing 10 mmol l⁻¹ levamisole on a thin 2% agarose pad. Immobilized worms were covered with a 24×24 mm cover slip, and viewed with 40× or 63× objective magnification for no more than 10 min after addition of levamisole. At least 10 independent worms per strain were analysed. The position of cell bodies relative to neuronal ganglia, as well as the dominant visible neuronal process were imaged by both DIC optical and epifluorescence microscopy. Images were acquired through a Hamamatsu Photonics camera software, and were cropped to size, assembled, and processed using Abode PhotoShop^® (Adobe Systems) and ImageJ (NIH) software.

A stock solution of 2 mg ml⁻¹ DiI in dimethyl formamide was prepared, and then a 1:200 dilution in M9 was used for staining. Animals were transferred into a microtitre well containing 150 µl stain and incubated for 2–3 h in the dark at room temperature. Using a pipette, the animals were transferred to a fresh plate and left to crawl on the bacterial lawn for about 1 h to destain. Finally, animals were immobilized on an agar pad with sodium azide and visualized by fluorescence microscopy.

Around 10 L4+1 animals were transferred into a 3 cm Petri dish containing 2 ml M9 supplemented with BSA and incubated for 1 h at 4°C to immobilize the animals. Then, animals were observed under a normal dissection microscope (40× magnification). Dissection was performed as previously described (Franks et al., 2009). The isolated pharynxes were pipetted onto a 2% agarose pad and covered with a cover slip prior to experimental observation. A total of five isolated pharynxes were checked.

Data are expressed as means±s.e.m. and were analysed using GraphPad Prism version 7.00 (GraphPad Software Inc.) using an unpaired Student's t-test, one-way ANOVA or two-way ANOVA, as indicated in figure legends. For ANOVA, the data were subjected to post hoc analysis using the method indicated in the figure legends. Significance level was set at P<0.05.

In a comparison of pharyngeal pumping with the N2 wild type (244±7 pumps min⁻¹), the nlg-1 mutant worms showed a reduction of ≈26% (180±12 pumps min⁻¹) (Fig. 1A). This was selective to the on-food context, as the pharyngeal pumping rate of nlg-1 mutants and wild-type animals was not significantly different 15 min after transferring to a non-food plate (39±4 min⁻¹ for wild type and 38±4 min⁻¹ for nlg-1; n=18 animals for each strain; P>0.05; data not shown). This food context-dependent reduction in pharyngeal pumping in nlg-1 was rescued by expressing a cDNA encoding the nlg-1 Δ#14 isoform from the nlg-1 promoter. The transcript Δ#14 is the dominant isoform in the adult stage (Calahorro et al., 2015). This fully rescued the pharyngeal deficit in nlg-1 mutants (Fig. 1A). These results indicate a selective role for NLG-1-dependent signalling in maintaining a high rate of pharyngeal pumping in the presence of food. It is unlikely that this deficit can be explained by an inability of nlg-1 worms to sense food as they show chemotaxis towards a point source of bacteria at the same rate as wild-type worms do (Fig. 1B).

Experiments in which the RIP neuron is ablated indicate there may be important extrapharyngeal determinants that drive the pharyngeal microcircuit via volume transmission. However, these RIP ablations may also highlight a clear potential for intrinsic signalling pathways embedded within the pharyngeal nervous system to act as regulators of the food-induced pumping (Dallière et al., 2016). Insight into the contribution of distinct signalling pathways to pharyngeal regulation can be obtained from analysis of an electrophysiological recording called the electropharyngeogram or EPG (Avery et al., 1995; Cook et al., 2006; Franks et al., 2006), and by comparing the EPG waveform between wild-type and mutants that are defective in the signalling pathways of interest. Therefore, we made EPG recordings to provide insight into the neural mechanisms underpinning the nlg-1-dependent pharyngeal behaviour. Pump rate and pump duration are interlinked in that the pump rate cannot be higher than allowed by the pump duration i.e. another contraction–relaxation cycle cannot begin until the pharynx has fully relaxed. Nonetheless, within these constraints, it is possible to have distinct changes in the pattern of activity. Each EPG waveform represents the activity of the pharyngeal neuromuscular circuit during a single contraction–relaxation cycle of the pharyngeal muscle that underlies a pharyngeal pump and carries information about the activity of specific neurons that regulate pump frequency, pump duration and the discrete functional organisation of each pharyngeal pump (Franks et al., 2006). The typical EPG harbours potentials that relate to the MC synaptic activity (‘e’), the synchronous contraction of the muscle (‘E’), the inhibitory signals from M3 (‘P’), and the rapid relaxation of the muscle (‘R’) and the repolarization of the terminal bulb (‘r’) (Fig. 2Ai).

In the absence of 5-hydroxytryptamine (5-HT or serotonin), the frequency of EPGs is low for both N2 wild type and nlg-1 (N2, 6.19±1.65 pumps min⁻¹; nlg-1, 4.18±1.32 pumps min⁻¹; P>0.05), reflecting the fact that this is an ‘off-food’ pumping rate and therefore feeding rate is down-regulated (Dallière et al., 2016). This is consistent with the low pump rate seen in animals off food. However, despite the similarity in EPG frequency between N2 and nlg-1, the EPG analysis revealed discrete changes in the EPG waveform when N2 and nlg-1 mutants were compared: the nlg-1 mutant exhibited a marked increase in pump duration (Fig. 2A,B) and an absence of transient potentials called P waves that are dependent on the activity of the inhibitory glutamatergic neuron M3 (Avery, 1993b) (Fig. 2C). We next analysed the EPG waveform in the presence of 5-HT. This is a well-characterized positive regulator of pharyngeal pumping which is engaged in the presence of food (Niacaris and Avery, 2003). In the presence of 5-HT, EPG rate was elevated in both N2 and nlg-1 (Fig. 2B). Despite this 5-HT-mediated increase in pumping, the nlg-1 EPG waveform showed a significant reduction in the number of P waves, as well as an increased pump duration relative to recordings made from 5-HT-treated N2 pharynxes (Fig. 2A). The loss of these key signatures was rescued in the strain expressing the nlg-1 rescue transgene from the nlg-1 promoter (Fig. 2B,C).

To gain further insight into the neural basis of the effect of nlg-1 on the EPG waveform, we made a systematic comparison of EPGs in nlg-1 compared with mutants of known regulators of EPGs. It is well established that the absence of P waves provides a readout of activity of the inhibitory glutamatergic motorneuron M3 (Avery, 1993b), which regulates the duration of the pharyngeal contraction–relaxation cycle by accelerating muscle repolarisation. The loss of glutamate release from M3, as observed in the mutant eat-4 (Raizen and Avery, 1994), results in fewer P waves and increased pump duration (Raizen and Avery, 1994). nlg-1 mutants show a similar EPG signature to eat-4 mutants (Fig. 3). This is consistent with an involvement of glutamate signalling that is intrinsic to the pharyngeal circuit, as the increased pump duration and loss of P waves is characteristic of a role for the glutamatergic pharyngeal neuron M3.

Although GABA is not known to be a neurotransmitter within the pharyngeal microcircuit (Franks et al., 2006), GABAergic signalling has been reported to modulate pumping (Dallière et al., 2016). Recent observations show a nlg-1 dependence of GABAergic signalling in both C. elegans (Maro et al., 2015; Tu et al., 2015) and mammals (Budreck and Scheiffele, 2007; Fu and Vicini, 2009). Motivated by these findings, we next compared the EPG waveform for the GABAergic signalling mutants unc-25 (e156) and unc-49 (e382) with nlg-1. Surprisingly, given that neither GABA nor GABA receptors are present in the pharyngeal neurons, the EPG phenotype is even more marked in unc-25 and unc-49 (Fig. 3A,B) than in eat-4, with a very extended pump duration and absence of P spikes. This therefore identifies a previously unreported GABAergic extrapharyngeal regulation of pharyngeal physiology. Moreover, the similarity in the EPG phenotype between the GABAergic mutants and nlg-1 suggests neuroligin signalling exerts its organization of pharyngeal function at an extrapharygneal locus rather than at a site intrinsic to the pharyngeal circuitry. Therefore, we next investigated the expression pattern of nlg-1 in the pharyngeal and extrapharyngeal circuits.

nlg-1 is widely expressed in the nervous system of C. elegans, including the head region (Hunter et al., 2010). Although detailed expression has been ascribed to AIY, DAs, VAs, PVD, URA and an undefined subset of cholinergic and GABAeregic neurons (Hunter et al., 2010; Hu et al., 2012; Maro et al., 2015; Tu et al., 2015), there is no definitive information on expression in the pharyngeal microcircuit. Therefore, an important first step was to resolve nlg-1 expression in the pharyngeal nervous system. This consists of 20 neurons that are embedded within the pharyngeal muscle and separated from the central nervous system of the worm by a basal lamina (Albertson and Thomson, 1976). We dissected and isolated pharynxes from two nlg-1 transcriptional reporter strains and established that there is no nlg-1 expression in the pharynx or in its associated microcircuit (Fig. 4A). Therefore, neuroligin exerts its regulation of pharyngeal behaviour through extrapharyngeal circuits.

To place nlg-1 in the context of neural circuits that link food-related sensory inputs to pharyngeal behaviour (Fig. 1A), we conducted detailed mapping of its expression pattern. Approximately 20 neurons express neuroligin in the anterior ganglia. We focused our attention on extrapharyngeal neurons that are known to regulate food-related behaviours (Table 1). To facilitate this, we carried out colocalization experiments using arrays expressing RFP in specific sensory neurons (Table 1). ADF is a sensory neuron that provides a serotonergic drive to increase pumping rate on food (Song et al., 2013). We found no co-expression of neuroligin in ADF (Fig. 4B) and no expression in major sensory neuron classes, using specific markers for the following neurons: ASJ, AWB, AWC and ASE neurons (Fig. 4, Fig. S1). However, although neuroligin is not expressed in NSM, we found expression in HSN neurons in the vulva area (Fig. 4B). Finally, to further investigate expression in a subset of sensory neurons we tried to identify neuroligin expression in the DiI-labelled amphid neurons: ADL, ASH, ASI, ASJ, ASK, AWB. Again, we found no neuroligin expression (Fig. 5A).

However, nlg-1 is expressed in a subset of eat-4 glutamatergic neuron classes (Fig. 5, Fig. S1). In this context, it is noteworthy that glutamatergic signals are required for both foraging and feeding behaviours (Hills et al., 2004; Dallière et al., 2016). In addition, neuroligin expression was localized specifically in a subset of dopaminergic sensory neurons, ADEs (Fig. 5B), which also are involved in food-dependent behaviours (Hills et al., 2004), as well as in the bilateral interneurons AIY (Fig. 5B), that function to extend food-seeking periods (Shtonda and Avery, 2006). Finally, by positional identification of cell bodies, we found expression of nlg-1 in the sensory neurons ALA and URX, as well as the interneuron AVJ (Fig. 5A).

The C. elegans neuroligin loss-of-function mutant is unable to maintain the same high rate of pharyngeal pumping in the presence of food compared with the wild type. This suggests that neuroligin is required in the circuitry that regulates feeding behaviour in response to sensory cues arising from bacteria. The activity of the pharynx is regulated by local and hormonal excitatory and inhibitory systems that converge to fine-tune food intake in a context-dependent manner (Dillon et al., 2015, 2016; Dallière et al., 2016). Insight into pharyngeal regulation came from a study in which it was shown that laser ablation of all pharyngeal neurons does not completely abolish pharyngeal pumping (Avery and Horvitz, 1989). It has been proposed that there is an intrinsic myogenic rhythm that is modulated by the pharyngeal nervous system. This system is embedded underneath the pharyngeal basal membrane and is connected to the extrapharyngeal network via the RIP neurons (Bhatla et al., 2015). This provides a neural pathway for regulation of feeding behaviour by the extrapharyngeal nervous system (Trojanowski et al., 2016) in response to sensory cues arising from food (Bhatla et al., 2015). However, ablation of RIP does not remove all food-dependent pharyngeal behaviours, suggesting that other pathways, either intrinsic to the pharynx (e.g. sensing of bacteria in the pharyngeal lumen) or arising through extrapharyngeal neurohormonal signals, are important (Dallière et al., 2016). The involvement of neuroligin in the first mechanism is unlikely as there is no evidence for nlg-1 expression in the pharynx. This suggests neuroligin is required in an extrapharyngeal circuit for robust up-regulation of pharyngeal pumping in the presence of food.

Our data suggest that neuroligin-dependent extrapharyngeal processing of the food cue modifies the manner in which the pharynx responds to the presence of food (Fig. 6) and that it is required to decrease the duration of a pharyngeal pump, most likely by increasing activity of the glutamatergic neuron M3. An inability to execute this control in the presence of food is consistent with the observation that nlg-1 is unable to sustain a high level of pumping in the presence of food as a long pump duration is incompatible with fast pumping. We considered where in the extrapharyngeal circuitry neuroligin may mediate this effect by mapping the expression pattern of nlg-1. It may function in circuits detecting food-related stimuli from the environment through sensory neurons such us ALA and URX and those processing information through interneurons such us AIY and AUA (Coates and de Bono, 2002; Chalasani et al., 2007). In addition, our data show a co-expression of neuroligin in extrapharyngeal glutamatergic neurons and this mirrors neuroligin control of glutamate transmission in mammals (Graf et al., 2004; Budreck et al., 2013). Neuroligin is also expressed in the ADE dopaminergic neurons and this could indicate a deficit in the dopaminergic contribution to food-related behaviours in nlg-1 (Hills et al., 2004). In this context, it is interesting to note that in mice neuroligin-3 modulates inhibition onto dopaminergic neurons (Rothwell et al., 2014).

We observed nlg-1 expression in ALA and AVJ which are GABA-positive cells (Gendrel et al., 2016). This goes hand in hand with the electrophysiological data showing that GABA signalling-deficient mutants phenocopy the neuroligin pharyngeal EPG phenotype. This indirectly suggests a pivotal role for regulation of GABAergic signalling by neuroligin, which is required to sustain a high level of feeding in the presence of food. A role for neuroligin in controlling GABAergic signalling in the context of a food-driven behaviour has parallels with recent evidence showing neuroligin organizes C. elegans GABAergic postsynapses (Maro et al., 2015; Tong et al., 2015; Tu et al., 2015) and the neuroligin dependence of GABA cellular signalling at the body wall neuromuscular junction (Tu et al., 2015).

In conclusion, we provide evidence for the altered processing of food cues in a neuroligin-deficient mutant, nlg-1, which results in an inability to maintain a sustained level of feeding in the presence of an abundant food source. This altered processing of sensory cues imparted by neuroligin deficiency is interesting in the broader context of the role of neuroligin in human autism spectrum disorder, which features dysfunctional sensory processing (Beker et al., 2018). In addition, the data highlight how molecular determinants when restricted to remote circuits can impact distal circuits that do not express the key molecular determinant but have significant consequence for the prime output of that circuit. This distribution of cause and effect will be an important consideration when investigating the behavioural consequence of genetic determinants in the context of interacting microcircuits.

Our study delivers new insight into the way in which neuroligin is required for context-dependent behavioural responses (Calahorro et al., 2009; Hunter et al., 2010; Calahorro and Ruiz-Rubio, 2012). We suggest a model in which neuroligin functions in the extrapharyngeal nervous system and links sensory detection of food to feeding behaviour through communication between sensory neurons and interneurons impacting in downstream pharyngeal circuits. Thus, neuroligin is required to organize the circuit(s) that integrates food sensory cues to generate an appropriate behavioural response.

We thank James Rand for kindly sharing strains and unpublished data; Antonio Miranda-Vizuete for sharing the trx-1::DsRed strain. Additional strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440).