Presynaptic Gαo (GOA-1) signaling depresses command neuron excitability to allow for stretch-dependent modulation of egg-laying behavior in C. elegans

Caenorhabditis elegans egg laying is a two-state behavior modulated by sensory input. Feedback of egg accumulation in the uterus drives activity of the serotonergic HSN command neurons to promote the active state, but how aversive sensory stimuli signal to inhibit egg laying is not well understood. We find the Pertussis Toxin-sensitive G protein, Gαo, signals in HSN to inhibit circuit activity and prolong the inactive behavior state. Gαo signaling hyperpolarizes HSN, reducing Ca2+ activity and input into the postsynaptic vulval muscles. Loss of inhibitory Gαo signaling uncouples presynaptic HSN activity from a postsynaptic, stretch-dependent homeostat, causing precocious entry into the egg-laying active state. NLP-7 neuropeptides signal to reduce egg laying both by inhibiting HSN and by activating Gαo in cells other than HSN. Thus, Gαo integrates diverse signals to maintain a bi-stable state of electrical excitability that dynamically controls circuit activity and behavior output in response to a changing environment.


Introduction 47
A major goal of neuroscience is to understand how external and internal sensory signals 48 control the activity of neural circuits to drive changes in animal behavior. Such sensory 49 information triggers when a particular behavior should be initiated, for how long that behavior 50 state should be continued, and under what conditions that behavior should be terminated. For 51 example, hunger initiates searching behavior strategies to locate areas with food, and sensory 52 feedback of local food availability triggers the termination of searching and the initiation of there is no neural circuit in any organism for which we know precisely how signaling events drive 62 a serotonin-controlled behavior and how sensory input modulates these events. Small neural 63 circuits typically found in invertebrate model organisms combine anatomical simplicity with 64 uniquely powerful genetic and experimental accessibility, allowing for a complete understanding 65 of the molecular basis for a behavioral output (Marder, 2012). 66 The C. elegans female reproductive circuit is ideally suited to study how environmental 67 and internal sensory signals modulate decision making. The circuit is anatomically simple and 68 drives alternative egg-laying behavior states that are characterized by ~20 minute inactive 69 periods punctuated by ~2 minute active states in which ~4-6 eggs are laid in phase with the 70 animal's locomotion (Waggoner et al., 1998;Collins and Koelle, 2013;Collins et al., 2016). As 71 shown in Figure 1A, the circuit is comprised of two Hermaphrodite Specific Neurons (HSNs) that 72 function as command neurons to promote the active state (Waggoner et al., 1998;Emtage et al., 73 2012). Three locomotion motor neurons (VA7, VB6, and VD7) and six cholinergic Ventral C 74 neurons (VC1-6) synapse onto a set of egg-laying vulval muscles which contract to open the 75 vulva to release eggs from the uterus into the environment (White et al., 1986). HSNs release 76 serotonin and NLP-3 neuropeptides that signal to promote the active state of egg laying (Desai 77  laying, HSN-deficient animals will eventually enter active states with coordinated vulval muscle 90 Ca 2+ activity that allows efficient egg release (Collins et al., 2016). These results indicate that 91 while HSN activity is sufficient to induce circuit activity and behavior in adult animals, HSNs are 92 not strictly required. Other signals must initiate the egg-laying active state in the absence of 93

HSNs. 94
The major G protein, Gαo, mediates a large part of the modulatory signaling in the brain 120 (Jiang et al., 2001), but our understanding of the biochemical consequences of Gαo signaling in 121 vivo remain incomplete. Patient mutations in human GNAO1 have been identified that disrupt 122 Gαo plasma membrane localization and inhibition of voltage-gated Ca 2+ currents in response to 123 norepinephrine, with phenotypic consequences including epileptic encephalopathy (Nakamura 124 et al., 2013). Discovering the conserved mechanisms by which Gαo inhibits synaptic 125 transmission in simple neural circuits would inform the development of novel therapies for human 126 disorders where Gαo has an important modulatory role. C. elegans Gαo shares more than 80% 127 sequence identity with its corresponding mammalian ortholog, and knockout mutants show Gαo signaling itself affects egg-laying circuit activity and behavior has not been fully revealed. 138 Gαo could signal within the active state to reduce the probability of HSN burst firing, shortening 139 the duration of active states. Alternatively, Gαo may signal during the inactive state to reduce 140 HSN excitability and the probability of entering the egg-laying active state. Whether and how 141 such inhibitory signaling acts alongside the stretch-dependent homeostat is similarly unclear. 142 Gαo signaling in HSN has been found to inhibit tph-1 gene expression and serotonin biosynthesis 143 (Tanis et al., 2008), suggesting long-term changes in serotonin transmission might also 144 contribute to the dramatic egg-laying behavior phenotypes seen in Gαo signaling mutants. 145 Here we explore how Gαo signals to inhibit C. elegans egg-laying circuit activity and 146 behavior. Our data reveal that Go signaling reduces the electrical excitability of a command 147 neuron, allowing the circuit to execute a binary behavior decision upon the alignment of optimal 148 external and internal sensory conditions. 149 150

151
Gαo signaling inhibits egg-laying behavior in C. elegans. Animals with too much Go 152 signaling retain eggs in their uterus, while Go loss-of-function or null mutants retain fewer eggs 153 (Tanis et al., 2008). Embryos in such hyperactive egg-laying mutants also spend less time 154 developing in the uterus and are laid at earlier stages of development, typically fewer than eight 155 cells per embryo. Whether Go manipulations caused a change in the duration of the active state 156 (e.g. how frequently eggs are laid within an active state), duration of the inactive state (how 157 frequently animals enter an egg-laying active state), or both, remains unclear. To better 158 understand how inhibitory Gαo signaling contributes to the pattern of circuit activity that underlies 159 two-state behaviors, we analyzed the temporal pattern of egg laying during adult active states in 160 Go signaling mutants. 161 162 Reduced inhibitory Gαo signaling leads to premature egg laying and decreases the 163 duration of egg-laying inactive states 164 We find that Go signals to inhibit the onset of egg laying. We performed a 'time to first 165 egg' assay in wild-type animals and in mutants with too much or too little Gαo signaling. As 166 previously described, wild-type animals release their first embryo ~6-7 hours after becoming 167 adults (Ravi et al., 2018a). Animals bearing Go loss-of-function or null mutations laid their eggs 168 much earlier, 3-4 hours after becoming adults ( Figure 1B). n1134, a hypomorphic mutant 169 predicted to lack the conserved N-terminal myristoylation and palmitoylation sequence, and 170 sa734, an early stop mutant predicted to be a molecular null (Segalat et al., 1995;Robatzek and 171 Thomas, 2000), showed a similar precocious onset in egg laying ( Figure 1B). This phenotype 172 was shared in transgenic animals where Go function was inhibited just in HSNs through the 173 cell-specific expression of Pertussis Toxin (Tanis et al., 2008). Because the timing of this first 174 egg-laying event requires serotonin and HSN activity (Ravi et al., 2018a), these results suggest 175 that Gαo normally signals in HSN to inhibit neurotransmitter release and thereby delay the first 176 egg-laying active state ( Figure 1B). To test the effects of increased Gαo signaling, we analyzed 177 the behavior of egl-10(md176) mutants which lack the major RGS protein that terminates Gαo 178 signaling by promoting Gαo GTP hydrolysis (Koelle and Horvitz, 1996). egl-10(md176) mutants 179 showed a strong and significant delay in the onset of egg laying, laying their first egg ~15 hours 180 after reaching adulthood ( Figure 1B); this delay is similar to animals without HSNs (Ravi et al., 181 2018a). This delay in egg laying phenotype was shared in transgenic animals expressing the 182 constitutively active Gαo (Q205L) mutant specifically in the HSNs, consistent with Gαo signaling 183 in HSN acting to inhibit neurotransmitter release. 184 To understand how Gαo signaling controls the normal two-state pattern of egg laying, we 185 made long-term recordings of adults as they transitioned into and out of the active states in which 186 clusters of several eggs are typically laid. Intervals between egg-laying events were operationally 187 classified into two categories: intra-cluster intervals and inter-cluster intervals, as previously 188 described (Waggoner et al., 1998;Collins and Koelle, 2013;Banerjee et al., 2017;Zang et al., 189 2017;Chew et al., 2018). Intra-cluster intervals (< 4 minutes) are intervals between consecutive 190 egg laying events within a single active state. Inter-cluster intervals (> 4 minutes) are the 191 intervals between distinct active states, and thus provide us with a measure of the frequency of 192 egg-laying active states (Waggoner et al., 1998). Wild-type animals displayed a two-state pattern 193 of egg laying with multiple egg-laying events clustered within brief, ~2 minute active states about 194 every 20-30 minutes ( Figure 1C and Table 1). Animals with reduced inhibitory Gαo signaling 195 entered active states 2-3-fold more frequently, often laying single eggs during active states 196 separated by only ~12-13 minutes ( Figure 1C and Table 1 Table 1). Loss of 200 inhibitory Gαo signaling led to active states in which the 1-2 embryos in the uterus were laid 201 almost immediately after they were positioned next to the vulval opening. As a result, successive 202 egg-laying events were rate-limited by egg production, and the average intra-cluster intervals 203 were typically double that of wild-type animals ( Figure 1C, Figure Supplement 1, and Table 1). 204 In contrast, egl-10(md176) mutant animals and animals expressing the Gαo(Q205L) gain-of-205 function mutant in the HSNs had infrequent egg laying, lengthening the average inactive period 206 to 258 and 67 min, respectively ( Figure 1C, Supplemental Figure 1, and Table 1). Interestingly, 207 animals with too much Gαo signaling still laid eggs in clusters of multiple eggs (Table 1), 208 consistent with our results showing that a stretch-dependent homeostat can maintain the active 209 state even when neurotransmitter release from the HSN is inhibited (Collins et al., 2016;Ravi et 210 al., 2018a). These results show that Gαo signaling does not modulate patterns of egg laying 211 within active states. Instead, Gαo specifically acts to determine how frequently animals enter into 212 the egg-laying active state. In addition, these results suggest that Gαo signals to inhibit egg-213 laying behavior even under 'optimal' laboratory growth and culture conditions. 214 215 Gαo signaling inhibits HSN Ca 2+ activity to promote the inactive behavior state 216 To understand how Gαo signaling regulates HSN activity, we performed ratiometric Ca 2+ 217 imaging in our panel of Gαo signaling mutants. Animals bearing the goa-1(n1134) hypomorphic 218 or goa-1(sa734) null mutations that reduce inhibitory Gαo signaling showed a clear change in 219 HSN Ca 2+ activity from burst to more tonic firing (Figure 2A, Videos 1-3). Complete loss of 220 inhibitory Gαo signaling caused a significant increase in the frequency of HSN Ca 2+ transients 221 ( Figure 2B and 2C). We were surprised that the goa-1(n1134) mutants, which show strongly 222 hyperactive egg-laying behavior indistinguishable from that of goa-1(sa734) null mutants, 223 showed only a modest and insignificant increase in HSN Ca 2+ activity compared to wild-type 224 ( Figure 2C). The goa-1(n1134) hypomorphic mutant is expected to have residual Gαo signaling 225 activity in that its major defect is the absence of a proper membrane anchor sequence (Mumby 226 et al., 1990). These results suggest that the hyperactive egg-laying phenotypes observed in goa-227 1(n1134) mutants are separable from changes in presynaptic HSN Ca 2+ activity. Instead, these 228 behavioral effects may be a consequence of inhibitory Gαo signaling outside of HSN and/or 229 secondary changes in serotonin biosynthesis (Segalat et al., 1995;Tanis et al., 2008). 230 We next tested how increased inhibitory Gαo signaling affects HSN activity. Both egl-231 10(md176) mutants and transgenic animals expressing the activated GOA-1(Q205L) in HSNs 232 showed a significant and dramatic reduction in the frequency of HSN Ca 2+ transients, with single 233 HSN Ca 2+ transients occuring several minutes apart (Figure 2A and 2B). The rare egg-laying 234 events seen in animals with increased Gαo signaling were mostly associated with single HSN 235 Ca 2+ transients, not the multi-transient bursts seen in wild-type animals (Figure 2A and 2C). In 236 one egl-10(md176) animal, we observed one egg-laying event that was not accompanied by an 237 HSN Ca 2+ transient. This suggests that elevated Gαo signaling may effectively silence the HSNs, 238 and that, in this case, egg laying becomes HSN-independent. Consistent with this model, 239 complete silencing of HSNs in egl-10(md176) and egl-1(n986dm) mutants that lack HSNs show 240 similar defects in the timing of first egg laid (Ravi et al., 2018a). Alternatively (or additionally) Gαo 241 signaling may function to depress coordinated activity between the gap-junctioned, contralateral 242 HSNs, whose Ca 2+ activity we were unable to observe simultaneously because our confocal 243 imaging conditions only captures one HSN at a time.  Figure 3C). These results suggest that unidentified neurotransmitters and/or 252 neuropeptides signal even under 'optimal' steady-state growth conditions to activate HSN 253 receptors and Gαo, to reduce cell excitability, allowing the observed two-state pattern of HSN 254 activity and egg-laying behavior. Importantly, these results show that Gαo signals cell-255 autonomously in HSN to inhibit Ca 2+ activity. Such changes in cell excitability by Gαo signaling 256 are expected to precede presynaptic UNC-13 localization  and/or long-term 257 changes in serotonin biosynthesis (Tanis et al. 2008). 258 We have previously shown that burst Ca 2+ activity in the command HSN neurons is 259 initiated and sustained by a stretch-dependent homeostat. In chemically or genetically sterilized 260 animals, burst Ca 2+ activity in HSN is largely eliminated (Ravi et al., 2018a). As such, we were 261 surprised to observe high frequency Ca 2+ transients in Gαo signaling mutants because these 262 animals typically retain few (1 to 3) eggs in the uterus at steady state, conditions that normally 263 eliminate HSN burst firing. We hypothesized that the stretch-dependent homeostat was not 264 required to promote HSN Ca 2+ activity in Gαo signaling mutants. To test this, we chemically 265 To test how changes in inhibitory Gαo signaling affect the postsynaptic vulval muscles, we 278 recorded Ca 2+ activity in the vulval muscles of goa-1(n1134) mutant and Pertussis Toxin 279 expressing transgenic animals. Active states were operationally defined as beginning one 280 minute before the laying of the first egg and concluding one minute after the last egg-laying event. 281 As shown in Figure 4, inactive state vulval muscle Ca 2+ twitching activity is slightly increased in 282 goa-1(n1134) mutants but is dramatically increased in transgenic animals expressing Pertussis 283 Toxin in the presynaptic HSN neurons, confirming an increase in neurotransmitter release from 284 the HSNs. Surprisingly, egg-laying active state Ca 2+ activity in goa-1(n1134) mutants was not 285 significantly different from that seen in wild-type control animals ( Figure  presynaptic HSN Ca 2+ activity in these animals ( Figure 3C). Vulval muscle Ca 2+ activity after 294 FUDR treatment was still higher in animals expressing Pertussis Toxin in HSNs compared to 295 similarly treated wild-type animals ( Figure 4C and 4D). This result suggests that vulval muscle 296 activity remains dependent on egg accumulation and/or germline activity even when HSN activity 297 is dramatically increased. However, because these animals lay eggs almost as soon as they are 298 made, the degree of stretch necessary to induce the active state must be markedly reduced. 299 300

Gαo signaling modulates the HSN resting membrane potential 301
Reduction of inhibitory Gαo signaling strongly increased HSN Ca 2+ activity and burst firing, 302 prompting us to investigate whether Gαo signaling modulates HSN electrical excitability. We 303 recorded the resting membrane potential of the HSN neurons in animals with altered Gαo 304 signaling using the whole-cell patch clamp method ( Figure 5A), as described (Yue et al., 2018). 305 depolarized resting potentials (-17.9 mV) compared to wild-type animals (-21.1 mV), but this 307 difference was not statistically significant ( Figure 5B). In contrast, the resting membrane potential 308 of HSNs in egl-10(md176) Gαo RGS protein mutant animals with a global increase in Gαo 309 signaling (Koelle and Horvitz, 1996)

Inhibition of egg laying by Gαo is not replicated by elevated βγ expression 318
Receptor activation of Gi/o heterotrimers releases βγ subunits which have previously 319 been shown to bind to activate specific K + channels and inhibit Ca 2+ channels (Reuveny et al., 320 1994;Herlitze et al., 1996). To test if over-expression of βγ subunit in HSN would similarly inhibit 321 egg laying, we transgenically overexpressed the C. elegans Gβ protein and Gγ protein subunits 322 GPB-1 and GPC-2 under the tph-1 promoter along with GFP. We did not observe any significant 323 differences in steady-state egg accumulation Figure 5C. The number of eggs stored in-utero in 324 these animals (13.0±1.1) was comparable to wildtype animals (15.7±1.2) and less than egl-325 10(md176) mutant animals (44.53±2.3). These results suggest that Gαo signals to inhibit HSN 326 activity and egg laying via effectors distinct from simple titration or release of βγ subunits. 327 328

Egg-laying behavior is dysregulated in cAMP and cGMP signaling mutants 329
As shown in Figure 5D  by Gαo has not been previously reported. We find that animals carrying gsa-1(ce81) gain-of-338 function mutations predicted to increase Gαs signaling accumulate fewer eggs compared to wild-339 type animals ( Figure 5D, middle). Because a reduction in steady-state egg accumulation could 340 result from indirect effects on egg production or brood size, we examined the developmental age 341 of embryos laid. Loss of inhibitory Gαo signaling causes embryos to be laid previously, before 342 they reach the 8-cell stage ( Figure 5D whether Gαo and Protein Kinase G regulate egg laying in a shared pathway, we performed a 357 genetic epistasis experiment. goa-1(sa734); egl-4(n479) double null mutants accumulate very 358 few eggs ( Figure 5D, middle), resembling the goa-1(sa734) null mutant. However, the low brood 359 size of the goa-1(sa734) mutant could prevent accurate measurement of these animal's egg-360 laying defects. To address this, we measured the stage of eggs laid. Loss of the EGL-4 Protein 361 Kinase G strongly and significantly suppressed the hyperactive egg-laying behavior of Gαo null 362 mutants ( Figure 5D, bottom). goa-1(sa734); egl-4(n479) mutants laid 33% of their embryos at 363 early stages compared to 88% for the goa-1(sa734) single mutant. The eggs laid by these double 364 mutants were at wild-type stages of development, not at late stages typically observed from egl-365 4(n479) single mutants (Trent et al., 1983). These results are consistent with Gαo acting 366 upstream or parallel to cGMP and/or Protein Kinase G signaling to regulate egg-laying behavior. 6A). To test how NLP-7 signals through Gαo to inhibit HSN activity and egg laying, we crossed 425 NLP-7 over-expressing transgenes into goa-1 mutant animals and evaluated their egg-laying 426 behavior phenotypes. goa-1(n1134) loss-of-function and goa-1(sa734) null mutants showed a 427 mild and strong suppression of the egg-laying defect of NLP-7 overexpressing animals with 428 animals storing 23.3±2.5 and 9.3±1.8 eggs, respectively ( Figure 6A). To confirm that Gαo was 429 required for NLP-7 inhibition of egg laying, we measured the stage of embryos laid by these 430 animals. goa-1(sa734) null mutant animals over-expressing NLP-7 laid ~100% of their embryos 431 at early stages, and this was not significantly different from 98% of embryos laid at early stages 432 by goa-1(sa734) single mutant animals ( Figure 6B). Together, these results strongly suggest 433 that NLP-7 neuropeptides signal to activate Gαo and inhibit egg laying. 434 Since the HSNs appear to be the principal sites of inhibitory Gαo signaling, we tested how 435 NLP-7 over-expression affects HSN Ca 2+ activity. As expected, over-expression of NLP-7 436 strongly inhibited HSN Ca 2+ activity ( Figure 6C  still show hyperactive egg-laying behavior (Segalat et al., 1995), suggesting that Gαo signals to 514 inhibit neurotransmitter release in cells other than HSN to regulate egg laying. We find that NLP-515 7 over-expression largely silences HSN Ca 2+ activity and blocks egg-laying behavior, consistent 516 with previous results (Banerjee et al., 2017). Our data further show that NLP-7 inhibition of egg 517 laying requires Go function, but that loss of Go does not rescue NLP-7 inhibition of HSN Ca 2+ 518 activity. This suggests that NLP-7 signals to inhibit HSN and egg laying via distinct pathways. 519 NLP-7 is predicted to be processed into four distinct peptides, and previous work has shown that 520 Several models for how Go signals to regulate neurotransmitter release have been 535

NLP
proposed, and our work is consistent with Go acting to inhibit multiple G protein effector 536 pathways instead of within a single, dedicated pathway. A major target of Gi/o family of G 537 proteins include inward rectifying K + channels thought to be activated by release of  subunits 538 (Hille, 1994). Previous work has shown the IRK-1 K + channel is expressed in HSN and is 539 required for inhibition of egg laying by the Go-coupled EGL-6 neuropeptide receptor (Emtage 540 et al., 2012). We do not observe behavior phenotypes upon over-expression of  in HSN, 541 suggesting Go may signal to inhibit HSN activity via direct effectors of the G subunit. We find 542 that mutations which increase cAMP and cGMP signaling cause hyperactive egg-laying behavior 543 phenotypes that resemble loss of inhibitory Go signaling. Such phenotypes would be consistent 544 with a model where Go signals to inhibit cAMP production and/or activate cGMP-specific 545 phosphodiesterases. Protein Kinase G signaling has been shown to regulate the expression of 546 a secreted protein in the uterine epithelium whose levels correlate with egg-laying rate (Hao et  Caenorhabditis elegans hermaphrodites were maintained at 20°C on Nematode Growth Medium 591 (NGM) agar plates with E. coli OP50 as a source of food as described (Brenner, 1974). For 592 assays involving young adults, animals were age-matched based on the timing of completion of 593 the L4 larval molt. All assays involving adult animals were performed using age-matched adult 594 hermaphrodites 20-40 hours past the late L4 stage.