Serotonin and neuropeptides are both released by the HSN command neuron to initiate C. elegans egg laying

Neurons typically release both a small molecule neurotransmitter and one or more neuropeptides, but how these two types of signal from the same neuron might act together remains largely obscure. For example, serotonergic neurons in mammalian brain express the neuropeptide Substance P, but it is unclear how serotonin signaling might be modulated by a coreleased neuropeptide. We studied this issue in C. elegans, in which all serotonergic neurons express the neuropeptide NLP-3. The serotonergic Hermaphrodite Specific Neurons (HSNs) are command motor neurons within the egg-laying circuit that have previously been shown to release serotonin to initiate egg-laying behavior. We found that egg-laying defects in animals lacking serotonin were far milder than in animals lacking HSNs, suggesting that HSNs must release other signal(s) in addition to serotonin to stimulate egg laying. While null mutants for nlp-3 had only mild egg-laying defects, animals lacking both serotonin and NLP-3 had severe defects, like those of animals lacking HSNs. Optogenetic activation of HSNs induced egg laying in wild-type animals, or in mutant animals lacking either serotonin or NLP-3, but failed to induce egg laying in animals lacking both. We recorded calcium activity in the egg-laying muscles of animals lacking either serotonin, NLP-3, or both. The single mutants, and to a greater extent the double mutant, showed muscle activity that was uncoordinated and unable to expel eggs, such that the vm2 muscles cells that are direct postsynaptic targets of the HSN failed to contract simultaneously with other egg-laying muscle cells. Our results show that the HSN neurons use serotonin and the neuropeptide NLP-3 as partially redundant cotransmitters that together stimulate and coordinate activity of the target cells onto which they are released.

are command motor neurons within the egg-laying circuit that have previously 25 been shown to release serotonin to initiate egg-laying behavior. We found that 26 egg-laying defects in animals lacking serotonin were far milder than in animals 27 lacking HSNs, suggesting that HSNs must release other signal(s) in addition to 28 serotonin to stimulate egg laying. While null mutants for nlp-3 had only mild egg-29 laying defects, animals lacking both serotonin and NLP-3 had severe defects, like 30 those of animals lacking HSNs. Optogenetic activation of HSNs induced egg 31 laying in wild-type animals, or in mutant animals lacking either serotonin or NLP-32 3, but failed to induce egg laying in animals lacking both. We recorded calcium 33 activity in the egg-laying muscles of animals lacking either serotonin, NLP-3, or 34 both. The single mutants, and to a greater extent the double mutant, showed 35 muscle activity that was uncoordinated and unable to expel eggs, such that the Drugs that selectively manipulate serotonin signaling are widely used to treat 63 depression and other psychiatric disorders, yet these drugs are often ineffective, 64 and no specific molecular defects in serotonin signaling have been identified as 65 the cause of these disorders (1). This situation suggests there is more to 66 understand about the basic science of serotonin signaling that could help explain 67 the cause of many psychiatric disorders. One feature of serotonin signaling in the 68 mammalian brain that remains poorly understood is that serotonin neurons 69 appear to also release a specific neuropeptide, Substance P (2-6). Of the ~80 70 billion neurons in the human brain, only about 100,000 make serotonin: their cell 71 bodies are concentrated in the raphe nuclei of the brain stem, but extend axons 72 throughout the brain that release serotonin to influence many brain functions 73 (2,7,8). Several methods have been used to measure the proportion of serotonin 74 neurons that express Substance P in the various raphe subnuclei of human or rat 75 brain, with results suggesting that from 25% to nearly all serotonin neurons also 76 express substance P (2,5,9,10). The apparent co-release of serotonin and 77 Substance P from the same neurons is just one instance of the broad but poorly 78 antagonists showed that they, like selective serotonin reuptake inhibitors, can 82 have significant anti-depressant activity (15-17). One study in the brain stem 83 odor. This neuron coreleases a neuropeptide to activate a feedback loop to 105 dampen activity of the olfactory neuron on specific timescales (18). 106 Here, we have applied the genetic toolbox described above to analyze the 107 functional consequences of cotransmission by serotonin and a neuropeptide 108 within a well-characterized small circuit of C. elegans. The C. elegans egg-laying 109 circuit contains three neuron types that release neurotransmitters onto egg-laying 110 muscles and each other to generate ~two minute active phases, during which 111 rhythmic circuit activity induces egg-laying behavior, that alternate with ~20 112 inactive phases, during which the circuit is largely silent and no eggs are laid 113 (20). The two serotonergic hermaphrodite specific neurons (HSNs) serve as the 114 command neurons (21) within this circuit in that 1) worms lacking HSNs are egg-115 laying defective (22); 2) optogenetic activation of HSNs is sufficient to induce 116 activity of the circuit that mimics a spontaneous active phase (23-25); 3) no other 117 cells in the circuit have these properties (25). We show here that the HSNs use a 118 combination of serotonin and a neuropeptide to induce the coordinated circuit 119 activity of egg-laying active phases. 120

121
The serotonergic HSN egg-laying neurons remain largely functional 122 without serotonin 123 The small circuit that initiates egg-laying behavior is schematized in Fig 1A. The 124 serotonergic Hermaphrodite-Specific Neurons (HSNs), along with the cholinergic 125 Ventral Cord type C neurons (VCs), synapse onto the type 2 vulval muscles 126 (vm2), which are electrically coupled by gap junctions to the type 1 vulval 127 muscles (vm1) and contract with them to expel eggs (20). Loss of the HSNs 128 results in a severe egg-laying defect: a mutation in egl-1 causes death of the 129 HSNs and results in animals that continue to make eggs but rarely lay them (26), 130 resulting in the striking phenotype of adult worms distended with accumulated 131 unlaid eggs (Fig 1B and 1C). Because addition of exogenous serotonin to worm 132 culture media is sufficient to induce egg laying, even in worms lacking HSNs (22), 133 it has been suggested that HSNs induce circuit activity simply by releasing 134 serotonin, sensitizing the egg-laying muscles to activation by the acetylcholine 135 released by other motorneurons of the circuit (25,27). 136 Contrary to this model, we found that animals lacking serotonin (Fig 1D)  137 had only mild egg-laying defects. The tph-1 gene encodes the serotonin 138 biosynthetic enzyme tryptophan hydroxylase, and animals with a tph-1 null 139 mutation have no serotonin detectable by anti-serotonin antibodies or HPLC 140 analysis (28,29). tph-1 mutant animals had only mild egg-laying defects (~18 141 unlaid eggs), appearing more similar to wild type (~12 unlaid eggs) than they did 142 to egl-1 mutants lacking HSNs (~47 unlaid eggs; Fig 1B-1D). This result, along 143 with previous pharmacological, genetic, and behavioral studies of the function of 144 serotonin in egg laying (30), are consistent with the idea that serotonin release 145 can only partially explain how the HSNs initiate egg laying. 146 To determine definitively if serotonin is required for the HSNs to stimulate 147 egg laying, we optogenetically stimulated the HSNs of animals either wild-type for 148 tph-1 or deleted for the tph-1 gene (Fig 1E). Animals with channelrhodopsin 149 (ChR2) expressed in the HSNs and that are wild-type for tph-1 have been shown 150 previously to lay eggs within a few seconds of exposure to blue light, but only if 151 the required ChR2 cofactor all-trans retinal (ATR) is supplied to the worms 152 (23,24). We found that upon optogenetic activation of HSNs, tph-1 mutant 153 animals laid a number of eggs statistically indistinguishable from the number laid 154 by control animals wild-type for tph-1 (Fig 1E) We found that overexpressing the nlp-3 neuropeptide gene resulted in a 173 dramatic increase in the frequency of egg-laying behavior. An increased rate of 174 egg-laying behavior results in an increased number of eggs being laid at early 175 stages of development, since the eggs have little time to develop inside the 176 mother before they are laid (38). More than 80% of the eggs laid by worms 177 carrying the high-copy nlp-3 transgene were laid at early stages of development, 178 compared to about 5% for control animals not overexpressing any neuropeptide 179 (Fig 2A). We saw no such phenotype for worms overexpressing any of the other 180 four neuropeptide genes (Fig 2A). 181 We obtained nlp-3 null mutant animals in which the nlp-3 gene is deleted. 182 Unlike the egl-1 mutants lacking HSNs that accumulate ~47 unlaid eggs (Fig 1C), 183 nlp-3 null mutants accumulated only ~19 unlaid eggs (Fig 2B), and thus were 184 more similar to the wild type ( Fig 1B) or tph-1 mutant worms lacking serotonin 185 ( Fig 1D). However, when we made tph-1; nlp-3 double mutant so that the HSNs 186 lacked both serotonin and NLP-3 neuropeptides, the adult animals were 187 distended with ~42 unlaid eggs and thus showed a severe egg-laying defect 188 similar to that of egl-1 animals. We obtained a second, independent deletion 189 mutant for nlp-3 and observed the same mild defect in the single mutant and the 190 same severe defect in the double mutant with tph-1 (Fig 2D). To directly test what combination of transmitters the HSNs use to stimulate 205 egg laying, we generated animals that express ChR-2::YFP in the HSN neurons 206 ( Fig 3B) and that were wild-type for tph-1 and nlp-3 (controls), or that carried null 207 mutations in tph-1, nlp-3, or both. We then tested whether optogenetic stimulation 208 of the HSNs could induce egg laying. Both the control animals and null mutants 209 for tph-1 laid eggs readily upon ChR2 activation, with no statistically significant 210 differences in the number of eggs laid (Fig 3C), or in several other measures of 211 the egg-laying behavior induced (e.g. time to first egg laid, time to last egg laid, 212 Fig S2A-S2C). However, whereas all wild-type and tph-1 animals tested laid eggs 213 upon ChR2 activation, 7/20 nlp-3 mutant animals failed to lay any eggs, and the 214 13/20 that did lay eggs laid fewer on average than did the wild-type or tph-1 215 animals. No eggs were laid by any tph-1; nlp-3 double mutant animals ( Fig 3C). 216 Therefore, we conclude that the HSN neurons release both serotonin and NLP-3 217 peptides to stimulate egg laying, either signal alone is sufficient to stimulate at 218 least some egg laying, and when lacking both signals the HSNs have no 219 detectable ability to stimulate the behavior. To further investigate the relationship between serotonin and NLP-3 in activating 224 egg-laying behavior, we performed additional experiments to test if either of 225 these transmitters is required to allow the other to stimulate egg laying. For 226 serotonin stimulation of egg laying, we used a standard assay (22) in which 227 worms were placed in microtiter well containing plain buffer or buffer containing 228 serotonin, and the number of eggs laid in 60 minutes was counted. We saw, as 229 observed previously (39,42-44), that exogenous serotonin stimulates egg laying 230 in wild-type animals, but not in animals deleted for the serotonin receptor gene 231 ser-1 (Fig 4A). Null mutants for nlp-3 were stimulated by serotonin to lay eggs at 232 the same rate as were the wild-type controls, demonstrating that NLP-3 is not 233 required for serotonin to stimulate egg laying. 234 We used a converse experiment to test if NLP-3 could stimulate egg-235 laying in the absence of serotonin. We generated C. elegans transgenes that 236 overexpressed nlp-3 by containing multiple copies of nlp-3 genomic DNA, and 237 control transgenes that did not overexpress nlp-3. In a strain background wild-238 type for tph-1, we observed (Fig 4B), as we had seen previously in an analogous 239 experiment (Fig 2A), that overexpression of nlp-3 resulted in hyperactive egg However, in the wild-type less than 10% of the vulval muscle Ca 2 + transients 267 resulted in egg release, and even fewer successful egg-laying events occurred in 268 tph-1 mutants lacking serotonin (tph-1) or in nlp-3 mutants (30 eggs released 269 over three hours for the wild type, compared to 15 for tph-1 and 9 for nlp-3). 270 Activity in animals lacking both serotonin and NLP-3 neuropeptides (tph-1; nlp-3) 271 or lacking HSNs (egl-1) was actually more frequent than in the wild type, but very 272 rarely produced successful egg release (each genotype released just two eggs in 273 the three hours recorded). 274 To identify the differences between vulval muscle contractions that did or 275 did not release eggs, we adjusted how we collected images during Ca 2+ 276 recordings. Previously-published Ca 2+ imaging of the vulval muscles used 277 images focused at the center of the group of two vm1 and two vm2 muscles 278 found on either the left or right side of the animal, and the resulting images 279 showed Ca 2+ activity that was usually focused at the most dorsal tip of this group 280 of muscles, but that could not be assigned to individual muscle cells (25,46,47). The large majority of the muscle activity we observed in every genotype 287 examined occurred exclusively in one both of the vm1s imaged, with no 288 concurrent activity detected in the vm2s ( Fig 5C). In the wild-type, 12% of Ca2+ 289 transients involved both vm1 and both vm2 cells imaged, and we refer to such 290 events as "coordinated". We never observed an event in any genotype in which a 291 Ca 2+ transient occurred exclusively in vm2 cell(s) without accompanying activity 292 in vm1 cell(s). In the wild-type, coordinated vulval muscle contractions occurred 293 exclusively within active phases, the ~2 minute intervals during which eggs were 294 laid and that contained frequent vulval muscle transients ( Fig 5A). All 30 egg 295 release events observed in the wild type occurred during one of the 51 296 coordinated vulval muscle contractions we saw during the three hours of 297 recordings analyzed. Thus it appears that coordinated contraction of all the vulval 298 muscle cells is necessary for efficient egg release. 299 Mutants lacking serotonin, NLP-3, or both continued to show vm1 Ca2+ 300 transients at a rate similar to or even greater than seen in the wild-type, but a 301 decreased portion of these events were accompanied by vm2 Ca2+ transients to 302 produce coordinated events (Fig 5C). This decrease in coordinated events was 303 modest in tph-1 and nlp-3 single mutants, but severe in the tph-1; nlp-3 double 304 mutant and in egl-1 animals lacking HSNs. In the mutants, as in the wild-type, 305 egg release occurred almost only during coordinated events that included both 306 vm1 and vm2 activity ( Fig S3): we only observed one exceptional egg-release 307 event that occurred in an egl-1 animal during a vm1-only contraction. Thus the 308 loss of vm2 activity in the mutants correlated strongly with the loss of successful 309 egg laying. Thus, while vm1 calcium activity occurs in each genotype we 310 observed, only wild-type animals, in which the HSN neurons were able to release 311 both serotonin and NLP-3 neuropeptides onto vm2 cells, were able to frequently 312 trigger coordinated activity in both vm1 and vm2 and efficiently lay eggs. Our results show that serotonin and NLP-3 released by the HSNs induces 318 activity of the vm2 muscle cells and coordinates their activity with that of the vm1 319 muscles to productively release eggs. In wild-type animals, the egg-laying circuit 320 alternates between ~20 minute inactive states during which no eggs are laid, and side of the vulva would efficiently electrically couple these cells so that they 336 would contract as unit. In this study, we increased the spatial resolution of our 337 Ca 2+ imaging and found that this assumption was incorrect. We observed that the 338 vm twitch Ca 2+ transients seen in the inactive state occur in some or all of the 339 all of which may affect circuit function. The genetic approaches for analyzing co-501 transmission described in this work provides a useful complement to 502 electrophysiological studies, as they permit us to manipulate endogenous 503 signaling molecules with mutations and transgenes, to record circuit activity using 504 genetically-encoded calcium indicators, and to manipulate neural activity using 505 optogenetics, all within intact, freely-behaving animals. We are aware of just one 506 previous study that focused on co-transmission using this combination of genetic 507 approaches (18). In this pioneering study, an odor was shown to cause a C. 508 elegans sensory neuron to release glutamate to act via ionotropic receptors on 509 specific interneurons that further regulate a complex and incompletely 510 understood motor circuit to evoke a behavioral response to the odor. The same 511 sensory neuron also releases a neuropeptide that acts via a G protein coupled 512 receptor on a different interneuron to cause it to in turn release a second 513 neuropeptide back onto the sensory neuron, limiting activity of the sensory 514 neuron and the timescale of the behavioral response to the odor. 515 Our studies of co-transmission focus on the C. elegans egg-laying circuit 516 because its anatomical simplicity holds the promise that all the cells and 517 signaling events that control this circuit can be defined, something that has not 518 yet been accomplished for any neural circuit. We discovered that serotonin and 519 NLP-3 peptides released from the HSN command neurons have parallel and 520 partially redundant effects to activate coordinated, rhythmic contraction of the 521 egg-laying muscles. This finding may be analogous to results of some previous 522 studies of co-transmission, in which the two co-released signals act convergently 523 to increase activity the same target cells. The most relevant such example is in 524 the mammalian brain respiratory circuit, where co-release of serotonin and the 525 neuropeptide Substance P have parallel effects promoting rhythmic circuit activity 526 (10). It will be interesting to determine just how mechanistically analogous these 527 two cases of serotonin/neuropeptide co-transmission actually are, and whether 528 the action of serotonin within the C. elegans egg-laying circuit will provide a 529 model for the detailed workings of serotonin within neural circuits of the human 530 brain. 531

C. elegans strains 534
C. elegans strains were cultured at 20°C on NGM agar plates with E. coli strain 535 OP50 as a food source (66). All strains were derived from the Bristol N2 wild-type 536 strain. Genetic crosses and generation of transgenic strains were by standard 537 methods (67,68). A list of strains, mutants, and transgenes used in this study can 538 be found in Table 1.
mate to LX1836 for Fig 3C " Serotonin receptor 1 deletion All animals used in optogenetic assay were also mutant for the lite-1 gene to 567 eliminate an endogenous response of C. elegans to blue light. The wzIs30 568 transgene was homozygous for the experiment shown in Fig 1E, but we noticed 569 that the homozygous transgene caused developmental defects in the HSNs of 570 some animals (Fig S1) that resulted in these animals being egg-laying defective. 571 Therefore, for the experiment in Fig 1E, we examined the animals prior to 572 optogenetic stimulation and discarded the small percentage of animals that were 573 visibly egg-laying defective. The experiment shown in Fig 3C was carried out  574 such that all animals were wzIs30/+ heterozygotes, which we found had 575 morphologically normal HSNs (Fig S1). First we constructed the strains indicated 576 in Table 1 that were homozygous for wzIs30 and also homozygous for the other 577 mutations required by the experiment. We generated males of each of these 578 strains, and mated them to corresponding strains that were genetically identical 579 except that they lacked wzIs30. The cross progeny, identified by the presence of 580 YFP-labeling, thus were heterozygous for wzIs30 but homozygous for all other 581 mutations used in the experiment. were established for each injection, and the early stage egg assay (76) was 618 carried out on 50 eggs per line (250 eggs total per condition tested). One 619 representative line for each condition is listed in Table 1. 620 621

Ratiometric Calcium Imaging 622
Freely-behaving animals were mounted between a glass coverslip and chunked 623 section of an NGM plate for imaging as described (25,46,47) and recorded with a 624 20X Plan-Apochromat objective (0.8 NA) using a Zeiss LSM 710 Duo LIVE head 625 set to record two channels. Recordings were collected at 20 fps at 256 x 256 626 pixel, 16 bit resolution, for 1 hour. The stage and focus were adjusted manually to 627 keep the egg-laying system in view and focused during recording periods. Care 628 was taken to find a lateral focus that included as much of the vm1s and vm2s as 629 possible. Ratiometric analysis for Ca 2+ recordings was performed in Volocity 630 (version 5, PerkinElmer). A ratio channel was calculated from GCaMP5 (GFP) 631 and mCherry fluorescence channels. Volocity was also used to identify the vulval 632 muscles using size and intensity parameters that varied over a small range 633 based on individual animals. Any misidentified objects were manually excluded 634 prior to final analysis. The lowest 10% of the GCaMP5/mCherry ratio values were 635 averaged to establish a ΔR/R baseline using a custom Matlab script. This script 636 also identifies the peak of a transient based on identifying a change in 637 prominence that was typically 0.25 ΔR/R over the preceding second, but this was 638 adjusted based on the smoothness of the data for individual animals. With the 639 experimenter blinded to the genotype of the animals being scored, video of each 640 peak was observed in the ratio channel to determine whether the indicated 641 activity was restricted to vm1 or present in both vm1 and vm2, and whether an 642 egg was laid. We scored a transient as vm1-only if it was clear in the ratio 643 channel that there was a difference of more than 50% of maximum activity 644 between the vm1s and the adjacent regions where vm2 cells were located. 645 646

Statistical Methods 647
Statistical analyses were performed using GraphPad Prism for Mac OS X v. 7.0a. 648 95% confidence intervals were determined and 1-or 2-way ANOVA with multiple 649 comparisons were performed to determine statistical significance. For egg stage 650 assays, we used the Wilson-Brown method for determining the 95% confidence 651 intervals for binomial data. 652  nlp-3 mutants accumulated significantly more eggs than did the wild type, and 927 significantly fewer than did the tph-1; nlp-3 double mutant strains (p≤0.05, 928