Neuronal control of suppression, initiation and completion of egg deposition in Drosophila melanogaster

Egg-laying in Drosophila is the product of post-mating physiological and behavioural changes that culminate in a stereotyped sequence of actions. While egg-laying behaviour has been mostly used as a system to understand the neuronal basis of decision making in the context of site selection, it harbours a great potential as a paradigm to uncover how, once a site is selected, the appropriate motor circuits are organized and activated to deposit an egg. To study this programme, we first describe the different stages of the egg-laying programme and the specific actions associated with each stage. Using a combination of neuronal activation and silencing experiments we characterize the role of three distinct neuronal populations in the abdominal ganglion with different contributions to the egg deposition motor elements. Specifically, we identify a subset of glutamatergic neurons and a subset of cholinergic neurons that promote the initiation and completion of egg expulsion respectively, while a subset of GABAergic neurons suppresses egg-laying. This study provides insight into the organization of neuronal circuits underlying complex motor behaviour.


Introduction 17
Motor execution of behavioural programmes must be tightly controlled so that the 18 appropriate sequences of actions is executed once the internal state and 19 environmental cues are considered. Egg-laying is executed by female fruit flies up to 20 80 times per day 1 . For each egg laid, the female follows an egg-laying motor 21 programme described as comprising a search-like period followed by egg deposition 22 and subsequent clean and rest 2 . Presumably, during the search period the fly 23 evaluates the environment in order to find the best site to deposit the egg. For flies 24 and the other oviparous animals when and where an egg is deposited has a profound 25 impact on the survival of the offspring. A female decides an egg laying site based on 26 substrate texture 3,4 , food availability 2,5,6 protection from climate elements 7,8 , predator 27 avoidance 9-16 , and disease avoidance [17][18][19][20] . The female will also evaluate information 28 from other flies [21][22][23][24][25] . In recent years a number of mechanosensory 3,4 , chemical 2,5-7,9-29 13,17,18,21 and visual 8, [14][15][16]23 cues that modulate egg-laying site selection have been 30 identified, creating a broad view of the external cues guiding egg-laying decisions. 31 Internal cues prompt the search period in the female 26 . Movement of the egg through 32 the reproductive system is in part controlled by octopamine, which contracts the 33 ovaries and relaxes the oviduct, and glutamate which contracts the oviduct [27][28][29] . 34 Sensory neurons expressing the mechanosensory channel Piezo at the oviduct detect 35 contraction/distention leading to a search for an egg-laying site 26 . While great progress 36 has been made in understanding site selection and its neuronal underpinnings, very 37 little is known regarding the consummatory component of egg-laying: egg deposition. 38

39
Here we addressed the neuronal basis of the egg deposition motor programme, one 40 of the phases of egg-laying, at the lower level of motor control. We began with a 41 6 ( Fig. 1h). The timing of abdominal contortions is very variable though abdominal 92 contortions generally increase a few seconds after egg expulsion and decrease in the 93 minute leading up to egg expulsion. It is likely that abdominal contortions correspond 94 to the moment of ovulation since the next phase, exploration, is activated by 95 ovulation 26 . As abdominal contortions subside, the exploration motor programme 96 initiates (Fig. 1i). Females contact the substrate by extending the proboscis and by 97 bending the abdomen to reach the surface with the extruded ovipositor. A small peak 98 of proboscis extension is also observed after egg expulsion indicating there may also 99 be an association of this behaviour with the post-egg expulsion motor sequence. Part 100 of these ovipositor contacts are accompanied by burrowing. These two behavioural 101 elements are the same as those observed in the egg deposition programme except 102 that they are not accompanied by egg pushing or followed by egg expulsion. An 103 overview of exploration and egg deposition phases together with the respective motor 104 elements is provided in Supplementary Video 1. Abdominal contortions phase is 105 represented in Supplementary Video 2. We next analysed the transition probabilities 106 between the different phases (Fig. 1j). This analysis revealed that egg deposition is 107 always followed by abdominal contortions. Abdominal contortions are either followed 108 by exploration or transitions directly to egg deposition with similar likelihood. Once the 109 female initiates exploration she does not return to abdominal contortions without 110 depositing an egg first. Similarly, once the female deposits an egg, she does not 111 explore without going through abdominal contortions first. In our quantification of 112 phase transitions, we did not consider proboscis extension, as it is not exclusive to the 113 exploration phase nor to egg-laying behaviour. Still, the results suggest that 114 exploration is optional in this setup. Here we described seven behavioural elements 7 and associated the elements with each phase, which sets the stage to analyse the 116 neuronal circuits of egg-laying. 117 118

Activity of OvAbg neurons promotes egg deposition 119
Organs in the abdominal region of the flies connect to the central nervous system at 120 the abdominal ganglion (Abg) of the ventral nerve cord (VNC). Since the reproductive 121 organs are located in the abdomen and many egg-laying behavioural elements involve 122 abdominal movement, this region is candidate to identify neurons involved in the 123 execution of egg-laying behaviours. Therefore, we performed an activation screen of 124 splitGal4 lines 32-34 selected based on Abg expression and from this screen we 125 selected a line, we will call OvAbg, for further investigation. Interestingly, the OvAbg 126 line encompasses regulatory fragments of the sex determination gene, doublesex 127 (dsx), widely known to control sex specific reproductive behaviours in flies [34][35][36][37][38] . The 128 OvAbg line anatomy reveals a group of neurons exclusively localized in the Abg (Fig.  129 2a). OvAbg neurons innervate other ganglia of the ventral nerve cord, the 130 suboesophageal zone (SEZ) (Fig. 2b), and the reproductive system, specifically in the 131 lateral oviducts and the distal uterus (Fig. 2c). Additionally, we found one OvAbg 132 projection in the seventh abdominal segment of the body wall muscle (Supplementary 133 Fig. 1a and b). To identify the polarity of OvAbg neurons we used synaptotagmin and 134 Denmark labelling which revealed presynaptic terminals in the SEZ and mixed 135 labelling of processes in the Abg and lateral oviducts ( Supplementary Fig. 1c-k). To 136 investigate the role of OvAbg neurons in egg-laying, we optogenetically activated them 137 using CsChrimson 39 with 6 stimuli of 10 seconds with a 20 second interval between 138 stimuli (Fig. 2d). The neuronal activation always and only leads the mated female to 139 assume an egg pushing posture (Fig. 2e, Supplementary Video 3). In contrast, the 8 control females, which have the same genetic background as OvAbg but no regulatory 141 element for splitAD 40 , never assume the egg pushing posture during light stimulation. 142 status affects the output of OvAbg neurons, we activated virgin OvAbg females. We 145 observed that virgin females, like mated females, assume an egg pushing posture in 146 all light ON periods (Fig. 2e). To test the contribution of the OvAbg projections in the 147 brain to the activation phenotype, we activated headless females. We observed that 148 similarly to intact OvAbg females, headless OvAbg females assume an egg pushing 149 posture at every light stimulation ( Fig. 2g and h), indicating that OvAbg brain 150 projections do not contribute to this activation phenotype. Next, we quantified egg 151 expulsion during the activation protocol. We observed that nearly half of the test flies 152 expels an egg during the activation protocol whereas control females do not expel 153 eggs (Fig. 2i). Most of the eggs are expelled during the first stimulation period (Fig.  154 2j). These results show that activity of OvAbg consistently leads to an egg pushing 155 posture but does not always lead to egg expulsion. Given that OvAbg neurons control 156 a female specific behaviour we asked whether these neurons are present in the male. 157 Indeed, there are male OvAbg neurons albeit fewer and with dimorphic projections 158 (Supplementary Fig. 1l-n). Activation of OvAbg neurons in males always triggers two 159 motor elements associated with copulation, abdomen curling and aedeagus extrusion, 160 suggesting an analogous role of OvAbg neurons in male reproduction (Supplementary 161 Fig. 1o and p). To further investigate the role of OvAbg activity in egg-laying we 162 performed silencing experiments using the inwardly rectifier potassium channel 163 Kir2.1 41 . For egg-laying experiments females were paired with males on apple juice 164 agar plates and monitored for two hours for copulation. The number of eggs laid by 165 mated females over 24 hours was nearly abolished by OvAbg silencing (Fig. 2k). 166 Notably, we observed that 88.7% test and 75% control flies mated showing that OvAbg 167 neurons are not involved in female receptivity ( Supplementary Fig. 1q). This result 168 together with the observation that flies survive and appear healthy with constitutive 169 silencing of OvAbg neurons indicates that they are specifically involved in egg-laying. 170 To ascertain that egg-laying defect does not result from defect in egg production or 171 ovulation, we dissected the reproductive system. We observed eggs jammed in the 172 lateral oviducts in all test flies together with an excess of mature eggs in the ovaries 173 We have shown that upon activation of OvAbg neurons a single egg deposition motor 184 element -egg pushing -is induced and that OvAbg silenced females do not lay eggs. 185 How do OvAbg silenced females behave? Do they perform all the behaviour elements 186 with the exception of egg pushing or are other motor elements are affected? To answer 187 these questions we used the anion channelrhodopsin GtACR1 42 for acute optogenetic 188 silencing of OvAbg neurons. We analysed 15 minutes of light stimulation as well as 189 10 minutes pre-and 5 minutes post-stimulation (Fig. 3a). Acute silencing of OvAbg 190 neurons blocked egg-laying; the number of eggs laid during the stimulation was 191 severely reduced and partially recovered in the post-stimulation period (Fig. 3b). 192 Analysis of the behavioural elements showed that, with the exception of grooming, all 193 egg deposition motor elements are abolished during stimulation and partially 194 recovered post-stimulation ( Fig. 3c-g). The expulsion of four eggs during the silencing 195 period was done without using most of the egg deposition motor programme 196 (Supplementary Video 4). The number of terminalia grooming bouts does not differ 197 from control during silencing (Fig. 3g). However, the time the female spent grooming 198 the terminalia is much larger in the test condition during silencing (Fig. 3h). OvAbg/Gad1 neurons were observed in the brain (Fig. 4b) or the reproductive system 226 (Fig. 4c). Silencing OvAbg/Gad1 had no effect on the number of eggs laid in 24h (Fig.  227 4d). Therefore, if OvAbg/Gad1 neurons contribute to egg-laying they may do so by 228 inhibiting egg-laying. To test this, we activated OvAbg/Gad1 neurons overnight (16h 229 activation). We observed that activation of OvAbg/Gad1 abolishes egg-laying (Fig.  230 4e). Dissection of the ovaries at the end of the experiment revealed that the eggs are 231 jammed at the lateral oviduct ( Fig. 4f and g). projections to the brain (Fig. 5b) and the reproductive system (Fig. 5c). Silencing 241 cholinergic OvAbg (OvAbg/Cha) neurons leads to a severe reduction in the number of 242 eggs laid (Fig. 5d) and a large fraction of the females display egg jamming in the lateral 243 oviduct (Fig. 5e). These results show a very similar phenotype to that observed when 244 silencing all OvAbg neurons ( Fig. 2j and k). Likewise, activation of OvAbg/Cha neurons 245 using the protocol shown in OvAbg/Cha females laid one egg (Fig. 5h), in contrast to less than half of OvAbg 250 females (Fig. 2h). Additionally, all eggs laid during the stimulation protocol by 251 OvAbg/Cha females were laid during the first stimulus (Fig. 5i), whereas egg-laying 252 timing by OvAbg females during the stimulation protocol was variable, with females 253 laying eggs in the third and fifth stimulus as well as in the interstimulus intervals (Fig.  254 2i). The results show that, upon activation, if and when an egg is laid is variable for 255 OvAbg, but not for OvAbg/Cha females. We also quantified, within the 10 second 256 stimulation bout, when the females expelled the egg. We found a striking difference 257 between OvAbg and OvAbg/Cha females (Fig. 5j) accompanying ovipositor contacts (Fig. 6d). Finally, we observed in 14% of 286 stimulations flies either extruded the ovipositor with straight abdomen or bent the 287 abdomen without extruding the ovipositor. Control flies did not display any of these 288 14 behaviours during light ON (Fig. 6d). Interestingly, we found that virgin and mated 289 OvAbg/VGlut females display different behavioural phenotypes upon activation. In 290 virgins, 66% of the stimulations do not evoke any behaviour (Fig. 6d). Ovipositor 291 extrusion or abdomen bending was observed in 18% of the stimulations. Ovipositor 292 contact was identified in only 15% of the stimulations and burrowing behaviour 293 displaying was residual (1% stimulations) (Fig. 6d). This result suggests that the 294 behavioural output of OvAbg/VGlut neurons is modulated by the mating status. In the 295 first section of this study, we characterized ovipositor contact and burrowing as motor 296 elements displayed by pregnant females in the exploration and egg deposition phases. 297 The phenotype of OvAbg/VGlut activation provides evidence for a role of OvAbg/VGlut 298 in controlling these behaviours, which could be specific to one of those phases, or be 299 implicated in both. To test this, we investigated in detail the egg-laying behaviour in 300 OvAbg/VGlut females silenced with the potassium channel Kir2.1 since a GtACR tool 301 that allows this type of genetic intersection is not available. We video-recorded 302 silenced flies for 15 minutes and annotated their behaviour, as well as that of the 303 control flies. Chronic silencing of OvAbg/VGlut neurons reduced, but did not abolish, 304 the number of eggs laid (Fig. 6f), suggesting that this neuronal subset is not necessary 305 for egg-laying in contrast to OvAbg/Cha group. Analysis of the behavioural elements 306 shows that OvAbg/VGlut silencing affects egg deposition behaviours up to egg 307 expulsion: ovipositor contact, burrowing and egg pushing (Fig. 6g, h

and i). Silenced 308
OvAbg/VGlut females displayed fewer bouts of each behaviour per egg, although we 309 still observed manipulated flies that perform these behaviours at levels comparable to 310 control. This reduction in the expression of motor elements that precede egg expulsion 311 results in abnormal initiation of the egg deposition bout (Supplementary Video 7). 312 Surprisingly, the post-egg expulsion behaviours -abdomen curling and grooming -313 were not reduced ( Fig. 6j and k). In fact, grooming was slightly increased which may 314 be associated with the deficient egg deposition (Fig. 6k). Indeed, the increased 315 grooming in test females is specifically associated with the egg deposition events, 316 rather than a generalised increase in grooming terminalia throughout the video 317 ( Supplementary Fig. S2). The contortions phase was not affected (Fig. 6l). Glutamate 318 has been shown to be involved in oviduct contractions 28  In summary, the activation and silencing results strongly suggest that glutamatergic 332 OvAbg neurons are involved in the initiation of the egg deposition motor sequence. 333 Although OvAbg/VGlut neurons are not necessary for egg-laying, egg-laying is less 334 efficient in OvAbg/VGlut silenced flies due to deficient egg deposition initiation. 335 Furthermore, OvAbg/VGlut neurons play a role in progeny survival as silencing 336 OvAbg/VGlut leads to fewer eggs buried in the substrate and, therefore, exposed to 337 predation and adverse climate conditions. The abdominal ganglion egg deposition circuit is poised to receive commands from 365 the brain. The brain is thought to be involved in adaptive behaviour based on 366 information collected from the environment and internal state. Thus, egg-laying site 367 selection is processed in the brain which communicates with the abdominal ganglion 368 for the execution of egg deposition. Indeed, many sensory inputs to the brain have Detailed behaviour. To analyse the motor elements associated with egg-laying 460 behaviour, females were collected soon after eclosion and housed in groups following 461 the egg-laying deprivation protocol described above. Aged 4-7 days females were 462 tested. Flies were gently aspirated into the egg-laying arena and behaviour was 463 recorded at 20 frames per second during 45 min for Canton S (Fig. 1) and during 15 464 min for OvAbg/VGlut silenced females (Fig. 6f-p). The same fly handling procedure 465 was performed for optogenetic experiments in which egg-laying motor elements were 466 analysed (for more detailed information, see the optogenetic stimulation section). 467 Optogenetic stimulation. For all experiments using CsChrimson, except in the 16h 468 egg-laying assay (Fig. 4e-g), the stimulation protocol included a 1 min baseline period 469 followed by 6 repetitions of 10 s red-light stimuli with a power of 4.40 mW/cm 2 and 20 470 s interval between stimuli. Fly behaviour was recorded at 20 frames per second, 471 except in the OvAbg line activation experiments (Fig. 2e-j) in which we used 15 frames 472 per second. In the OvAbg headless females' photoactivation (Fig. 2g), the head was 473 gently cut using dissection forceps (Dumont #55 Forceps, 11295-51) under CO2 474 anaesthesia. Flies were transferred to the egg-laying arena and allowed to recover 475 from this procedure for 5-10 min before photoactivation. In the 16h photoactivation 476 egg-laying assay (Fig. 4e-g), mated females were transferred to the apple agar plates 477 (described in the 24 h egg-laying assay section). The stimulation protocol included 478 constant red-light with a power ranging 4.19-4.85 mW/cm 2 during 16 h. At the end of 479 this period, the eggs were counted and the reproductive system was dissected to 480 measure egg jamming. For the silencing experiment using GtACR1 (Fig. 3), the 481 stimulation protocol included a pre-stimulation period that lasted for 10 min, followed 482 by constant green-light stimulation with a power of 5-6.23 mW/cm 2 during 15 min and 483 a post-stimulation period of 5 min. Videos were recorded at 20 frames per second. Quantification of behaviours. Data and statistical analysis were performed using 499 custom Python scripts in Fig. 1b-j, Fig. 3 and Fig. 6f-p. All the other analysis were 500 performed using Prism9 (GraphPad Software, La Jolla, CA). 501 The inter-egg expulsion intervals (Fig. 1d) were calculated as 502 Inter-egg expulsion intervals = first frame of egg expulsion bout -last frame of 503 previous egg expulsion bout (per fly) 504 505 The number of eggs per 5 minutes (Fig. 3b) was calculated as 506 # Eggs / 5 min = (sum # egg expulsions) / 5 minutes (per fly) 507 508 The number of behaviour bouts per 5 minutes ( Fig. 3c-g The mean bout duration ( Fig. 3h and j) was calculated as 515 mean bout duration = # behaviour frame duration (s) / total # behaviour bouts 516

(per fly) 517
Where the behaviour frame duration is given by the last frame subtracted to the first 518 frame of each behaviour bout converted to seconds. 519 520 The proportion of eggs not buried (Fig. 6p) was calculated as 521

Proportion of eggs not buried = (sum # eggs not buried) / total # eggs (per fly) 522
Egg expulsion bouts in which it was not possible to determine whether the egg is 523 buried or not were excluded from this analysis. 524

525
The percentage of behaviour displayed by flies during photoactivation ( Fig. 2e-g, Fig.  526 5f) was calculated as 527 % Behaviour = (sum # stimuli with behaviour) / total # stimuli 528 The percentage of females laying eggs ( Fig. 2i and Fig 5h) was calculated as 529 % Females laying eggs = (sum # females that laid eggs) / total # of females 530 531 The percentage of eggs laid during the photoactivation protocol ( Fig. 2j and Fig. 5i) 532 was calculated as 533 % Eggs = (sum # eggs laid on each stimulus or ISI) / total # eggs 534 535 The latency for egg expulsion (Fig. 5j) was calculated as 536 Latency egg expulsion = first frame corresponding to egg expulsion eventfirst 537 frame corresponding to the stimulus initiation (per fly) 538 The product was converted to seconds. 539

540
The percentage of eggs jammed (Fig. 2l, Fig. 4f and Fig. 5e) was calculated as 541 % Eggs jammed = (sum # reproductive systems with eggs in the oviducts) / 542 total # reproductive systems 543 544 To investigate the egg-laying phases transition probability (Fig.1j) 2) proboscis extension behaviour is not exclusive to egg-laying, being also displayed 552 25 in other behavioural contexts (ex: feeding); 3) burrowing behaviour is displayed along 553 with ovipositor contact but with lower frequency (see Fig. 1i). We calculated for the 554 total video duration (45 min), the likelihood of all transitions between phases by using 555 a first order Markov Chain analysis 51 . Phases transitions were identified as changes 556 in the selected behavioural patterns for each phase. 557 To calculate the probability of the egg-laying behaviours around egg expulsion (Fig.  558 1f-i and Supplementary Fig. 2a and b) we aligned all the egg expulsion events of all 559 flies and we counted how many behaviours were occurring in each of the 1200 frames 560 preceding and following the end of egg expulsion. We then normalized the counts over 561 the total number of egg expulsions. To represent the egg expulsion event, the last 562 frame for each event was selected. We excluded from the analysis egg expulsion 563 events that are less than 1200 frames from the start or end of the video. The P value is provided in comparison with the control and indicated as * for p < 0.05, 578 ** for p < 0.01, *** for p < 0.001, **** for p < 0.0001, and 'ns' for non-significant (p ≥ 579 0.05).