Monoamines differentially modulate neuropeptide release from distinct sites within a single neuron pair

Monoamines and neuropeptides often modulate the same behavior, but monoaminergic-peptidergic crosstalk remains poorly understood. In Caenorhabditis elegans, the adrenergic-like ligands, tyramine (TA) and octopamine (OA) require distinct subsets of neuropeptides in the two ASI sensory neurons to inhibit nociception. TA selectively increases the release of ASI neuropeptides encoded by nlp-14 or nlp-18 from either synaptic/perisynaptic regions of ASI axons or the ASI soma, respectively, and OA selectively increases the release of ASI neuropeptides encoded by nlp-9 asymmetrically, from only the synaptic/perisynaptic region of the right ASI axon. The predicted amino acid preprosequences of genes encoding either TA- or OA-dependent neuropeptides differed markedly. However, these distinct preprosequences were not sufficient to confer monoamine-specificity and additional N-terminal peptide-encoding sequence was required. Collectively, our results demonstrate that TA and OA specifically and differentially modulate the release of distinct subsets of neuropeptides from different subcellular sites within the ASIs, highlighting the complexity of monoaminergic/peptidergic modulation, even in animals with a relatively simple nervous system.


Introduction
Monoamines and neuropeptides interact extensively to modulate behavior in both vertebrates and invertebrates, and the disruption of monoaminergic/peptidergic crosstalk has been implicated in a range of neurological disorders, including anxiety, depression and drug abuse [1][2][3]. However, given the complexity of the mammalian central nervous system, the specifics of these interactions are often difficult to analyze. In contrast, Caenorhabditis elegans contains only 302 neurons, but still exhibits complex behaviors that are extensively modulated by both monoamines and neuropeptides, with many similarities to mammalian systems. For example, antidepressants, such as Prozac, opiates and cannabinoids all work through orthologous pathways in both systems, and the noradrenergic inhibition of pain/aversive responses is also similar, with α 2 -like adrenergic receptors inhibiting primary nociceptors and α 1 -like adrenergic PLOS  ASI specificity when coupled to sequence coding for GFP, it is possible that the additional sequence coding for the RNAi could alter this specificity. Similarly, the α 1 -adrenergic-like Gα qcoupled OA receptor, SER-6, and ASI neuropeptides encoded by nlp-6, -7 and -9 are required for the OA inhibition of the aversive response to higher 1-octanol concentrations. Again, in agreement, RNAi knockdown of nlp-9 abolished the OA inhibition of the aversive response (Student's t-test: srg-47::nlp-9 p = 0.0001; Fig 2B; [6]). Together, these data suggest that monoamine-dependent Gα q -coupled signaling in the ASIs differentially modulates ASI neuropeptide signaling and subsequently aversive behavior. Therefore, the present study was designed to build upon our previous findings to understand how TA and OA might alter neuropeptide signaling from the ASIs, focusing specifically on ASI neuropeptides encoded nlp-9, -14 and -18.

TA-and OA-dependent neuropeptides appear to co-localize at individual ASI synapses
The ASIs contain 7-9 synapses in the distal portion of the axonal process based on serial electron microscopy reconstructions and confirmed by 7-9 fluorescent puncta found after ASI expression of synaptobrevin::GFP, a synaptic marker (Fig 3A; [22,23]). In addition, we previously demonstrated NLP-14::GFP fluorescence was observed in 7-9 discrete puncta along the distal portion of ASI axons that co-localized with the synaptic marker, RAB-3 ( Fig 3B circles; [11]). In the present study, we have confirmed this observation and demonstrated that fluorescence from NLP-18:: and NLP-9::GFP was also observed in 7-9 fluorescent puncta in the distal portions of ASI axons (Fig 3C and 3D circles). As observed previously for NLP-14::GFP, NLP-9::GFP also co-localized with a synaptic marker, mCHERRY::RAB-3, within these predicted synaptic regions at least at the level of fluorescence microscopy ( Fig 3E circles; [11]). However, since neuropeptide-transporting dense core vesicles (DCVs) often localize perisynaptically, the potential for a non-overlapping perisynaptic localization of the individual peptide-containing vesicles should not be ruled out [24,25]. NLP-14::GFP and NLP-9::mCHERRY also co- The two ASI sensory neurons integrate both monoaminergic and peptidergic signaling to modulate nociceptive behavior. The two ASI sensory neurons express Gα q -coupled TA, OA and 5-HT receptors that differentially modulate neuropeptide release and, in turn, differentially modulate aversive responses mediated primarily by the two ASH sensory neurons. For example: 1) the ASI OA receptor, SER-6, and neuropeptides encoded by nlp-6, -7 and -9 are essential for the OA inhibition of aversive responses to 100% 1-octanol [6], 2) the ASI TA receptor, TYRA-3, and neuropeptides encoded by nlp-1, -14 and -18 are essential to inhibit the 5-HT stimulation of aversive responses to 30% 1-octanol [11], and 3) the ASI 5-HT receptor, SER-1, appears to stimulate the release of neuropeptides encoded by nlp-24 that activate the Gα o -coupled ASI opiate receptor, NPR-17, to interfere with TYRA-3 signaling in response to 30% 1-octanol [5].
https://doi.org/10.1371/journal.pone.0196954.g001 localized at fluorescent puncta within predicted synaptic/perisynaptic regions of the ASIs, as anticipated, based on their co-localization with RAB-3 ( Fig 3F circles; [11]). Together, these data suggest that both TA-and OA-dependent neuropeptides co-localize to each of 7-9 synaptic/perisynaptic puncta within the distal portion of right and left ASI axons. Interestingly, although all of the animals expressing ASI::peptide::GFP exhibited 7-9 fluorescent synaptic/ perisynaptic puncta, the spacing of these puncta in either the right or left ASIs often varied significantly from animal to animal, making the correlation of individual fluorescent puncta with specific downstream synaptic partners problematic. Whether these differences result from preparation artifacts is unclear, but it is worth noting that most of what we know about synaptic connectivity in C. elegans is derived from very few animals and it would not be surprising if the physical relationship among ASI synapses varied from animal to animal. Additional peptide::GFP and mCHERRY::RAB-3 fluorescence was also observed in the soma and co-localized in the axon proximal to the cell body, where no synapses have been described by electron microscopy (Fig 3B-3F, arrowheads). Whether these axonal fluorescent puncta represent trafficking vesicles, additional synapses not observed by electron microscopy or sites of extra-synaptic release remains to be determined, but the localization of at least some of these puncta appeared to be consistent in ASIs expressing different peptide::GFPs. For example, fluorescence from NLP-9::, NLP-14:: and NLP-18::GFP was consistently observed in 3-4 distinct puncta in the proximal portion of the axon near the soma (Fig 3B-3D, arrowheads). Similar ASI fluorescent puncta have also been observed by others after ASI expression of synaptobrevin::GFP, another synaptic marker [22]. Interestingly, the ratio of GFP fluorescence in synaptic and predicted extra-synaptic puncta differed significantly in ASIs from animals expressing either the TA-dependent NLP-14::GFP or the OA-dependent NLP-9:: mCHERRY, suggesting that these TA-and OA-dependent neuropeptides may be trafficked in different vesicles or preferentially localized to different sites (Student's t-test: nlp-9 p = 0.0001; Fig 3G).

TA and OA stimulate the release of distinct subsets of ASI neuropeptides
No methods have been described to measure acute neuropeptide release in C. elegans, but two indirect methods have been used previously to approximate release; treatment-dependent decreases in neuronal-specific peptide::GFP fluorescence or increases in fluorophore uptake by coelomocytes, scavenger cells located in the pseudocoelmic fluid [26][27][28][29]. Unfortunately, the coelomocyte uptake assay is not effective for all neuropeptides because uptake is dependent on the amount of neuropeptide released and the location of the neuropeptide-releasing cells [26]. Indeed, the coelomocyte uptake assay was not useful for measuring the release of ASI neuropeptides encoded by nlp-9, -14 or -18 in the present study, as coelomocyte uptake was low, inconsistent and different pairs of coelomocytes sometimes yielded different results. Therefore, we focused on monoamine-dependent decreases in fluorescence from GFP-tagged neuropeptides to approximate peptide release, using a monoamine-insensitive ASI-specific promoter, gpa-4. Worms were transferred to a monoamine-supplemented plate for 1 hr prior to imaging for all assays, and as noted in Fig 4A, TA and OA had no effect on ASI gpa-4::GFP expression. In addition, TA and OA also had no significant effect on gpa-4, nlp-14, -18 and -9 expression as assayed by qPCR, not surprisingly given the short time course of the experiment (Fig 4B).
Our behavioral data demonstrated that neuropeptides encoded by both nlp-14 and nlp-18 are required for the TA-inhibition of aversive responses (Fig 2; [11]). As predicted, TA significantly decreased total NLP-14::GFP fluorescence of nlp-14 null animals expressing ASI::NLP-14::GFP (One-way ANOVA: Total p = 0.020; Fig 4C). TA-dependent decreases in total ASI:: NLP-14::GFP fluorescence required the TA receptor TYRA-3, as TA had no effect in nlp-14; tyra-3 null animals lacking the key TA receptor essential for the TA inhibition of aversive responses (Fig 4C; [11]). In contrast, OA had no effect on ASI::NLP-14::GFP fluorescence ( Fig  4C). Together, these data demonstrate that decreases in ASI::NLP-14::GFP fluorescence were both TA-specific and TYRA-3-dependent. Importantly, TA, but not OA, also specifically decreased axonal GFP fluorescence in predicted synaptic/perisynaptic, but not extra-synaptic regions (One-way ANOVA: Axonal p = 0.006; Synaptic p = 0.006; Fig 4C). These TA-dependent decreases in ASI::NLP-14::GFP axonal fluorescence suggest either that TA reduces NLP-14::GFP expression and/or stimulates NLP-14::GFP release, but since TA had no effect on the ASI::GFP fluorescence driven by the gpa-4 promoter, we postulate that these decreases in peptide::GFP fluorescence were due to TA-dependent increases in synaptic/perisynaptic neuropeptide release. Indeed, neuropeptide release has been previously observed in vitro using C. elegans neurons. For example, Smith et al., observed stimulus-dependent decreases in FLP-17:: VENUS fluorescence from cultured BAG neurons [30]. Similarly, Zhou et al., employed total internal reflection fluorescence microscopy combined with capacitance measurements to observe DCV exocytosis in cultured C. elegans neurons [31].

ASI neuropeptide proproteins required for either TA-or OA-dependent inhibition contain different leader sequences
The N-terminal sequence of neuropeptide preproproteins can be critical for efficient packaging and sorting into DCVs [34][35][36]. Therefore, we hypothesized that the ASI neuropeptides released by TA may localize to different DCVs than those released by OA and analyzed each of the predicted ASI neuropeptide preproproteins for clusters of charged amino acids similar to those involved in the trafficking of protachykinin. For example, charged N-terminal elements sort protachykinin into DCVs and neutralization of these charged elements dramatically reduce sorting efficiency [37]. Importantly, the identity of the N-terminal mature peptides encoded by nlp-1, -14, -18, -6, -7 and -9 have each been confirmed by mass spectrometry [21,38]. Each of the three TA-dependent preproproteins (encoded by nlp-1, -14 and -18) possessed dibasic cleavage sites upstream of the first N-terminal peptide, and after removal of the predicted signal peptide, each of the TA-dependent proproteins is predicted to contain a short, charged, N-terminal prosequence upstream of the first dibasic cleavage site (Fig 6A). In contrast, each of the three OA-dependent preproproteins (encoded by nlp-6, -7 and -9) lack dibasic cleavage sites upstream of the first N-terminal peptide ( Fig 6A). Presumably, the signal peptidase truncates these preproproteins immediately upstream of the first mature neuropeptide, leaving no charged prosequence, in contrast to proproteins encoded by nlp-1, -14 and -18.
To determine if these different preprosequences were involved in the differential action of TA and OA signaling, we constructed chimeric neuropeptide preproproteins by placing the predicted NLP-9 preprosequence upstream of the region coding for NLP-14 neuropeptides (ASI::NLP-9LS::NLP-14::GFP) and the predicted NLP-14 preprosequence upstream of the region coding for the NLP-9 neuropeptides (ASI::NLP-14LS::NLP-9::GFP) ( Fig 6B). Both chimeric neuropeptides localized to the 7-9 synaptic/perisynaptic axonal puncta observed for the native GFP-tagged neuropeptides, but TA or OA did not selectively decrease axonal fluorescence from either chimera (One-way ANOVA: Native NLP-14 p = 0.012 Fig 6C; Native NLP-9 p = 0.003 Fig 6D). Together, these data suggest that the differential effects of TA and OA on neuropeptide release do not lie exclusively within the different preprosequences. If all the information required for monoamine-specific neuropeptide release was contained within the preprosequence, we predict that the chimeric peptides would be differentially released in response to the appropriate monoamine-stimulus i.e. ASI::NLP-9LS::NLP-14::GFP should be sensitive to OA, not TA, and vice versa. However, our data did not reflect this, suggesting additional sequence within the peptide-encoding region are required. Alternatively, the chimeric peptides may not be properly processed and/or have altered folding, with either outcome inhibiting peptide release. Further investigation is required to understand how the N-terminal sequence functions in monoamine-specific neuropeptide release, as these properties may occur at the level of DCV packaging or downstream at the site of DCV localization/exocytosis. To further assess the importance of sequence downstream of the preprosequences, we generated a truncated ASI::NLP-14::GFP protein. The nlp-14 gene encodes seven predicted and five confirmed neuropeptides (Fig 7A; [21]). The truncated NLP-14  ::GFP, containing the complete N-terminal sequence, three predicted neuropeptides and two confirmed neuropeptides, mimicked the full length ASI::NLP-14::GFP and exhibited decreased axonal fluorescence following incubation in TA, but not OA (One-way ANOVA: Native NLP-14: p = 0.012; Fig  7C; NLP-14   Fig 7C). These data suggest that specific amino acid arrangements between the leader sequence and peptide-encoding regions are important for correct packaging, processing and/or monoamine sensitivity for release, i.e. in the case of NLP-14, the first 95 amino acids confer TA-dependent release and, in contrast, disruption of this sequence, as shown by the chimeric proproteins, abolishes neuropeptide release.

Discussion
The present study highlights the crosstalk between monoaminergic and peptidergic signaling in the modulation of C. elegans nociceptive behavior, and dissects differential monoaminedependent neuropeptide release from a pair of sensory neurons. The monoaminergic modulation of the aversive responses to 1-octanol is dependent on differential peptidergic signaling from the two ASIs, as outlined in Fig 1 [5,6,11]. The present study suggests that individual monoamines selectively stimulate the release of distinct subsets of ASI neuropeptides from unique subcellular sites, with TA-dependent neuropeptides encoded by nlp-14 and nlp-18 released at synaptic/perisynaptic sites or the soma, respectively, and OA-dependent neuropeptides encoded by nlp-9 released asymmetrically only from synaptic/perisynaptic sites on the right axon (Fig 8).

Monoaminergic/Peptidergic crosstalk is extensive and modulates many key behaviors
The ASIs express Gα q -coupled monoamine receptors that selectively stimulate the release of different subsets of ASI neuropeptides to modulate aversive responses. These neuropeptides operate at multiple levels within the aversive circuit and are part of a large modulatory network that define behavioral state, such as nutritional status [5,6,11]. Monoaminergic/peptidergic interactions also modulate additional behaviors in C. elegans, including nociception, feeding, locomotion, egg-laying and mating [15,[39][40][41][42]. Monoaminergic/peptidergic signaling also interact to modulate behaviors in mammals, including fear, satiety and depression [2,[43][44][45]. For example, dysfunctional serotonergic and peptidergic signaling are associated with depression. Many antidepressants increase 5-HT by inhibiting 5-HT reuptake and neuropeptide Y (NPY) mimics selective 5-HT reuptake inhibitors in a rat model for depression [2,46,47]. In contrast, depression-like behaviors are attenuated in NPY 1 , -2 or -4 knockout mice, highlighting the complexity of the interactions [48][49][50]. Collectively, these studies demonstrate that monoamines and neuropeptides interact to modulate many of the same behaviors. In fact, wide-spread, complex crosstalk between these different signaling systems modulates the activity of individual neurons, entire neuronal circuits and, as we present here, interacting directly, with monoamines differentially regulating neuropeptide release from distinct sites within a single pair of neurons.

Both TA-and OA-dependent ASI neuropeptides co-localize with synaptic/ perisynaptic markers in the ASIs
Both TA-and OA-dependent neuropeptides co-localized with synaptic markers in the ASIs, although whether this localization is synaptic or perisynaptic is unclear. Indeed, neuropeptide- containing DCVs often localize perisynaptically [51]. For example, DCVs do not cluster at active zones in C. elegans cholinergic motor neurons, Aplysia motor neurons or the mammalian dorsal root ganglion [24,25,52,53]. Discrete neuropeptide-containing DCV populations have also been identified within the same cell. AtT20 cells expressing prothyrotropin-releasing hormone exhibit distinct DCV populations containing either N-or C-terminal peptides cleaved from the original prothyrotropin-releasing hormone [54]. In the present study, neuropeptide-positive puncta were also observed along ASI axons where synapses were not identified. Whether these extra-synaptic neuronal puncta are associated with DCV trafficking or novel sites of release requires further investigation; however, they do not appear to be sites of monoamine-dependent release.

TA and OA differentially modulate neuropeptide release from the ASIs
The monoamine selectivity of ASI neuropeptide release is dependent on the expression of different Gα q -coupled monoamine receptors. Since Gα q signaling increases IP 3 -mediated calcium (Ca 2+ ) release, we hypothesized that TA and OA stimulate local ASI Gα q signaling and increase Ca 2+ , stimulating the release of DCVs from distinct pools [55,56]. Together these observations suggest that ASI GPCRs are differentially localized and ASI Ca 2+ signaling compartmentalized. However, compartmentalized ASI GPCR signaling or Ca 2+ signaling have yet to be characterized directly. In contrast, localized Ca 2+ release has been observed in a variety of other systems, with G-protein-coupled receptors often in proximity to local Ca 2+ stores [57]. In C. elegans, a muscarinic-like G-protein-coupled receptor modulates local Ca 2+ dynamics in the RIA interneurons to coordinate locomotion [58]. In Drosophila, OA stimulates neuropeptide release by activating both a cAMP-dependent protein kinase and increasing Ca 2+ release from internal stores [59]. In mammalian retinal starburst amacrine cells, a localized Ca 2+ signal is also compartmentalized by mGluR2 signaling that restricts propagation [60,61]. Amacrine cells are similar to many C. elegans neurons in that they share both synaptic inputs and outputs along their neuronal processes. Similarly, many GPCRs are localized subcellularly. For example, the cannabinoid receptor, CB1R, and the somatostatin receptor, SSTR2, differentially localize to axonal and somatodendritic regions in mammalian cells, respectively [62]. Selectively modulating local Ca 2+ dynamics could expand the signaling capacity of individual neurons, an approach potentially significant in an organism like C. elegans with only 302 neurons.
Monoamines have been shown to modulate neuropeptide release in other C. elegans behaviors/processes. For example, 5-HT stimulates the release of flp-7 encoded peptides to decrease body fat and peptides encoded by daf-7 and -28 to facilitate exit from dauer [63][64][65]. Similarly, dopamine stimulates the release of nlp-12 encoded peptides to modulate local search behavior and inhibits the release of flp-8 encoded neuropeptides to modify copper aversion [29,66]. In Drosophila, OA stimulates the release of GFP-tagged atrial natriuretic factor [59]. In mammals, multiple 5-HT receptors also modulate neuropeptide release associated with feeding. 5-HT stimulates the release of α-melanocyte stimulating hormone to stimulate feeding and inhibits the release of the orexigenic peptides, NPY and agouti-related protein to inhibit feeding [45,[67][68][69].
Conversely, neuropeptides directly modulate monoamine release. For example, nlp-7, flp-10, -11 and -17 encoded neuropeptides inhibit 5-HT release to inhibit egg-laying and DAF-7 inhibits TA/OA release to promote satiety and quiescence [13,41,70,71]. In mammals, neuropeptide S inhibits the release of 5-HT and noradrenaline from mouse amygdala and prefrontal cortex synaptosomes, but stimulates the release of dopamine when administered intracerebroventricularly, highlighting the complex interaction between the two signaling systems [72][73][74]. Similarly, peptide YY  inhibits the release of dopamine and norepinephrine, but not 5-HT from isolated hypothalamic synaptosomes, suggesting that neuropeptides can confer specificity to monoamine release [75]. Collectively, these studies demonstrate that monoaminergic and peptidergic signaling often modulate the same behaviors, but also interact directly to modulate the others release.

OA asymmetrically stimulates neuropeptide release from the ASIs
OA asymmetrically modulates the release of nlp-9 encoded neuropeptides, with OA only stimulating release from the right ASIs. Indeed, the right and left ASIs exhibit different synaptic connectivities [23]. However, some/most neuropeptides and, indeed, most monoamines act at receptors distantly localized from their sites of release, bringing into question the potential significance of our observed spatial patterns and asymmetric peptide release, if this peptidergic signaling is extra-synaptic. Little is known about the localization of neuropeptides with their cognate receptors in C. elegans. Certainly, in C. elegans some/most neuropeptides and, indeed, most monoamines appear to signal extra-synaptically and act at receptors distantly localized from their sites of release, based on the localization of monoamine synthesizing neurons and monoamine receptors. Indeed, most monoaminergic neurons do not synapse directly onto neurons expressing most monoamine receptors [76][77][78]. For example, none of the monoamine-synthesizing neurons synapse directly on the ASIs and only the serotonergic ADFs synapse on the ASHs, even though these two pairs of sensory neurons express a number of 5-HT, TA, OA and dopamine receptors (available from: http://wormweb.org/neuralnet#c=ASI&m= 1). However, 1) some peptide receptors may be synaptic and focus more local, extra-synaptic, monoaminergic signaling or, more likely, 2) given the tight packaging of neurons in the C. elegans nerve ring, the site of neuropeptide or monoamine release may be relevant, even for extra-synaptically localized peptide/monoamine receptors. Indeed, peptide receptors have been identified in close proximity to sites of peptide release in other systems. For example, the tachykinin-related peptide 1 and its receptor, LTKR, arborize in close proximity in the cockroach brain [79]. In Drosophila, immunoreactivity-labeling of tachykinins was distinct and closely resembled that of their receptor, DTKR, suggesting a local site of activation [80]. In addition, the selective localization of neuropeptidases could further modify peptidergic signaling, especially if the gene encodes 1) a mix of receptor agonists that activate different receptors or 2) agonists/antagonists that have the potential to be differentially degraded, as we have suggested previously for C. elegans opiate signaling [5]. For example, nematode neprilysins are differentially localized and modulate the degradation of number of regulatory peptides [81,82].
No studies to date have demonstrated compartmentalization in the ASIs, but data from the RIAs and the functional asymmetry of other sensory neurons increases the plausibility of a similar situation for the ASIs. Indeed, asymmetry in both connectivity and function has been observed in other C. elegans neuron pairs. For example, the differential expression of guanylyl cyclases contributes to functional asymmetry in the two ASE sensory neurons [83][84][85][86]. Similarly, the two functionally distinct AWC sensory neurons asymmetrically express str-2, a putative olfactory receptor, specifying str-2-expressing cells as AWC ON and non-str-2-expressing cells as AWC OFF [87][88][89]. Collectively, these studies highlight the ability of individual pairs of neurons to differentially perceive and process sensory information.

The TA-and OA-dependent ASI neuropeptides contain different predicted preprosequences
Neuropeptides are cleaved from preproproteins that generally encode multiple peptides, with packaging and trafficking information often located within the N-terminus of the preproprotein [38,90,91]. For example, efficient DCV packaging was observed when the first 26 N-terminal amino acids from proopiomelanocortin were fused to a reporter [92]. Indeed, the preprosequences predicted for the TA-and OA-dependent ASI neuropeptide-encoding genes differed significantly, with the TA-dependent preprosequences containing a predicted charged leader upstream of the first peptide and the OA-dependent preprosequences cleaved immediately upstream of the first peptide. However, these different preprosequences were not sufficient to confer monoamine-specificity, as chimeric NLP-9 and NLP-14 preproproteins with swapped preprosequences lost all monoamine-specificity. In contrast, the monoamine-specificity of the NLP-14 preproprotein was maintained after substantial C-terminal truncation. Similarly, proneuropeptide Y truncated at the C-termini was efficiently localized to DCVs and secreted [93]. Collectively, our data suggest that N-terminal signal sequence is not sufficient to confer monoamine-selectivity to neuropeptide release and that additional N-terminal peptideencoding sequence is required.
In summary, the present study highlights the complexity of monoaminergic/peptidergic interactions within a single neuron pair and demonstrates that the adrenergic-like ligands, TA and OA stimulate the release of distinct subsets of neuropeptides from the ASIs to modulate aversive behavior. In addition, the sites associated with the release of neuropeptides encoded by nlp-14, -18 and -9 within the ASIs also appear to be distinct, with TA-dependent neuropeptides encoded by nlp-14 and nlp-18 released at synaptic/perisynaptic sites or the ASI soma, respectively, and OA-dependent neuropeptides encoded by nlp-9 released asymmetrically only from synaptic/perisynaptic sites on the right axon.

Nematode strains
General culture and handling of strains were maintained as previously outlined [94]. The following alleles were used: wild-type N2 (Bristol strain), nlp-9 (tm3572), nlp-14 (tm1880), nlp-18 (ok1557) and tyra-3 (ok325). All strains were obtained from the Caenorhabditis Genetics Center (University of Minnesota, Minneapolis, MN) except nlp-9 (tm3572) and nlp- 14 (tm1880) which was received from the National Bio-Resources Project (Tokyo Women's Medical University, Tokyo, Japan). Animals containing double mutations were generated using standard genetic techniques and confirmed by PCR. All mutants were backcrossed with wild-type N2 strain at least four times.

Behavioral assays
Octanol avoidance assays were performed as described previously [11,98]. Briefly, L4 animals were picked 24 hrs prior to assay and kept at 16˚C on standard nematode growth medium (NGM) seeded with Escherichia coli strain OP50. NGM plates were prepared daily by adding 5-HT (4 mM) and/or TA/OA (4 mM) to liquid media before pouring. A hair (Loew-Cornell 9000 Kolinsky 8 paintbrush) was fused to a P200 pipette tip and dipped in either dilute (30% v/ v in 100% ethanol) or 100% 1-octanol and placed in front of an animal exhibiting forward locomotion. The time taken to initiate backward locomotion was scored for each animal and averaged. Assays were terminated after 20 sec as wild-type animals are known to spontaneously reverse~20 sec [98]. In all assays, animals were transferred to intermediate (non-seeded NGM) plates and left for 30 sec, then transferred to assay plates and left for 30 min before being scored for 1-octanol avoidance. In all assays, 27-38 worms were examined for each strain and condition. All strains were blinded and assays repeated in triplicate. Statistical analysis was performed using individual responses ± SE and Student's t test.

Neuropeptide fluorescence assay
For all neuropeptide fluorescence assays, transgenic L4 animals expressing either: Pgpa-4::gfp, gpa-4::nlp-9::gfp,gpa-4::nlp-14::gfp,gpa-4::nlp-18::gfp,gpa-4::nlp-14[leader sequence]::nlp-9::gfp, gpa-4::nlp-9[leader sequence]::nlp-14::gfp or gpa-4::nlp-14  ::gfp were picked 24 hrs prior to assaying and kept at 16˚C. NGM plates containing either TA or OA (10 mM) were prepared daily. Animals were incubated on NGM plates seeded with E. coli strain OP50 or transferred to assay plates for 1 hr prior to imaging. Images (0.4 μm Z-stacks) were captured using Olympus Fluoview FV1000 software and fluorescence was quantified using Volocity Software Version 4.0.0 Build 314 (Improvision Ltd). To quantify neuropeptide fluorescence, both the right and left ASIs were tightly cropped and divided into the soma and axon. All images were then blinded and fluorescence was filtered by intensity. The first 10 blinded images were used to generate an average threshold separating signal from noise, which was applied to all images within the experiment. Object intensities were summed for each image and objects < 3 voxels were regarded as non-specific background and excluded. The summed fluorescence of each treatment was normalized to the summed fluorescence of untreated (control) animals. All imaging was performed on an Olympus IX81 inverted confocal microscope and animals were immobilized in 20 mM sodium azide on 3% agarose pads. Statistical analysis was based on individual responses to treatment with n values given where appropriate. At least two plates per condition were used in all experiments.

RNA isolation and cDNA synthesis
RNA was isolated as outlined previously [99] with minor revisions. Briefly, wild-type animals were synchronized using alkaline hypochlorite and arrested at L1 before being moved to fresh NGM plates seeded with E. coli strain OP50 and maintained at 16˚C. Young adults were then washed in M9 and transferred to either fresh NGM plates seeded with E. coli strain OP50 (untreated population), fresh non-seeded NGM plates (control population), fresh OA (10 mM) or fresh TA (10 mM) NGM plates for 1 hr. Following incubation, animals were then washed in M9 and suspended overnight at -80˚C in Trizol™ (Invitrogen™). Animals were frozen and lysed via freeze-thaw cracking in liquid nitrogen. RNA was isolated using standard Trizol™ techniques and assessed for quality/quantity on an agarose gel and using Nanodrop™ ND-1000 Spectrophotometer. cDNA synthesis was performed using the SuperScript™ III First-Strand Synthesis system (ThermoFisher Scientific).

Quantitative PCR
qPCR studies were conducted using SsoFast™ EvaGreen1 Supermix (Bio-Rad) and Bio-Rad CFX96™ Real-Time PCR System. qPCR reactions were set up as per manufacturer's instructions. cDNA pools from either untreated, control, OA (10 mM) or TA (10 mM) populations were diluted to 50 ng/μl and 2 μl added for reaction set up. At least three separate pools of cDNA were analyzed. Cycling conditions were used as per the manufacturer's instructions. All results were normalized to the reference gene act-1 and relative expression determined using the Livak ΔCt method [100]. Data are presented as individual responses ± SE and analyzed using One-way ANOVA.

Statistics
All statistics were performed using SPSS software v24 (IBM). All data was initially assessed with either a Student's t-test or a One-way ANOVA and when appropriate, post-hoc analysis was carried out using a Dunnett or Dunnett T3 test.
Supporting information S1 File. Data supporting the underlying findings described within the manuscript. (XLSX)