Drosophila female-specific brain neuron elicits persistent position- and direction-selective male-like social behaviors

Latent neural circuitry in the female brain encoding male-like mating behaviors has been revealed in both mice and flies. In Drosophila, a key component of this circuitry consists of the doublesex-expressing pC1 neurons, which were deemed to exist in both sexes and function based on the amount of cells being activated. Here, we identify pC1-alpha, a female-specific subtype of pC1, as responsible for inducing persistent male-like social behaviors in females. We demonstrate that activation of a single pC1-alpha neuron is sufficient for such induction in a position- and direction-selective manner, and activity of pC1-alpha neurons as a whole is indispensable for maintaining normal sexual receptivity. These dual functions of pC1-alpha may require different neurotransmission, with acetylcholine specifically required for the former but not the latter. Our findings suggest that pC1-alpha may be the female counterpart of male P1 due to their shared similarities in morphology, lineage, and social promoting function.


Summary
Latent neural circuitry in the female brain encoding male-like mating behaviors has been revealed in both mice and flies. In Drosophila, a key component of this circuitry consists of the doublesex-expressing pC1 neurons, which were deemed to exist in both sexes and function based on the amount of cells being activated. Here, we identify pC1-alpha, a female-specific subtype of pC1, as responsible for inducing persistent male-like social behaviors in females. We demonstrate that activation of a single pC1-alpha neuron is sufficient for such induction in a position-and direction-selective manner, and activity of pC1alpha neurons as a whole is indispensable for maintaining normal sexual receptivity. These dual functions of pC1-alpha may require different neurotransmission, with acetylcholine specifically required

Introduction
The neural substrates of social behaviors, such as courtship and aggression, and their dimorphisms between the sexes have been studied extensively in Drosophila melanogaster (Anderson, 2016;Auer and Benton, 2016;Dickson, 2008). Like most unisexual species shaped by sexual selection, flies exhibit conventional sex roles: males compete with each other via aggression, but perform an elaborate ritual of courtship toward choosy females, who rarely fight and almost never court (Bateman, 1948;Trivers, 1972). Such sexually distinct social behaviors are neurally hardwired through the coordinated actions of the sex determination genes doublesex (dsx) and fruitless (fru) (Demir and Dickson, 2005;Kimura et al., 2005;Manoli et al., 2005;Rideout et al., 2010;Stockinger et al., 2005). Dsx/fru + P1 neurons promote an internal state that enhances both courtship and aggression (Bath et al., 2014;Hoopfer et al., 2015;Inagaki et al., 2014). Their presence in males and absence in females due to developmental apoptosis may explain the sexually dimorphic patterns of flies' social behaviors (Kimura et al., 2008). Yet, recent studies in both mice and flies have revealed the existence of latent neural circuitries that encode malelike mating behaviors in the female brain Rezaval et al., 2016;Wei et al., 2018), suggesting a conserved bisexual nature of brain organization and function . In flies, this circuitry consists of the dsx + pC1 cluster of neurons that are also essential for female sexual receptivity (Rezaval et al., 2016;Zhou et al., 2014). However, the pC1 neurons and its subsets were generally treated as more or less homogeneous and present in both sexes (Rezaval et al., 2016;Zhou et al., 2014), and their function in triggering male-like courtship behavior was deemed to depend on the amount of cells being activated (Rezaval et al., 2016). Here, we identify pC1-alpha, a female-specific subtype of the pC1 cluster, as responsible for inducing persistent male-like social behaviors in females.
We demonstrate that activation of a single pC1-alpha neuron is sufficient for such induction in a position-and direction-selective manner, and activity of pC1-alpha neurons as a whole is indispensable for maintaining normal female sexual receptivity. These dual functions of pC1-alpha may require different neurotransmission, with acetylcholine specifically required for the former but not the latter.
Our findings suggest that pC1-alpha may be the female counterpart of male P1 due to their shared similarities in morphology, lineage, and social promoting function, and may likely represent a novel cell type that arises specifically in the female brain from cooption of P1-like neural templates.

Thermogenetic screen and MARCM identified two cell types for inducing male-like social behaviors
We performed an unbiased large-scale thermogenetic screen in flies (Bidaye et al., 2014), which utilized the Vienna Tiles (VT) collection of GAL4 driver lines (Tirian and Dickson, 2017) and a UAS-trpA1 transgene to target expression of the thermosensitive cation channel TrpA1 (Hamada et al., 2008) to random but defined subsets of neurons. VT-GAL4 UAS-trpA1 flies were slowly warmed to ~32°C to open the TrpA1 channel and visually examined for ectopic behaviors. We screened 3470 lines and identified six, in which specifically females showed male-like social behaviors toward other flies at elevated temperatures but not at the control temperature of ~25°C (Movie S1). These male-like behaviors included courtship and aggression by a TrpA1-expressing test fly toward a non-confronting and a confronting control or test fly, respectively (Movie S1). When multiple TrpA1-expressing test females were co-housed in the behavior arena, in addition to courtship and aggression, we also observed chaining behavior (Movie S1). The courtship-like behavior mainly consists of frequent orientation and following and occasional tapping and unilateral wing extension (Movie S2). Such courtship-like behaviors often elicited ovipositor extrusion from female targets ( Figure S1A), but we recorded no courtship song (Movie S2; Figure S1B and C). The most robust of these induced male-like social behaviors is the following behavior (including orientation), which is crucial for initiation and maintaining proximity for subsequent social actions. Therefore, we focused our analysis on the induced following behavior in the females of these lines.
The expression patterns of the six identified VT lines, examined with a membrane-tethered mCD8-GFP (green fluorescent protein) reporter, showed that a female-specific cell type might be present in all of them ( Figure 1A). This cell type, consisting of only two to three neurons in each hemisphere, morphologically resembles dsx + pC1 (Rezaval et al., 2016;Zhou et al., 2014). To achieve more restricted expression, we applied MARCM (mosaic analysis with a repressible cell marker) (Lee and Luo, 1999) on the sparsest VT2064-GAL4 line to generate mosaic females, in which expression of mCD8-GFP and TrpA1 were targeted to random subsets of neurons labeled in the line. We then monitored them for following behavior in warmed chambers and determined their exact expression patterns. We observed induced following behavior in 37 flies, expression analysis of which, in comparison to 34 flies that did not show the phenotype, revealed that induction of following behavior strongly correlates with expression in two cell types ( Figure 1B to D; Figure S1G). Cell type 1, located in the dorsal posterior brain, is the shared female-specific pC1-like neuron, whose neurites project first anterodorsally to arborize extensively toward the lateral and medial protocerebrum regions and then transversally to contralateral regions ( Figure 1E). Cell type 2 is a nearby local interneuron innervating the central complex ( Figure 1F). Their enrichment in females showing induced following behavior was unlikely due to biased MARCM labeling, as comparable numbers of cell types were labeled in non-following females ( Figure S1D). We conclude that activation of cell type 1 and/or 2 elicits male-like social behaviors in females. However, we hypothesize that only cell type 1 is responsible, because there are no MARCM flies that only label cell type 2 but show the behavior and not all six positive VT lines label cell type 2 ( Figure 1A, C and D; Figure   S1E). Cell type 2 may be linked to cell type 1 clonally, as they were always colabeled ipsilaterally ( Figure   S1F). (C) Expression patterns of MARCM females, showing following behavior (n=37, left) or no following (n=34, right) upon activation. Each row represents scores for a cell type, and each column a single fly.
Unilateral and bilateral labeling were indicated by red and purple colors, respectively.
(D) Labeling frequency for each cell type in MARCM females with and without induced following behavior. ****p<0.0001, Fisher's exact test.
(E and F) Registered single-neuron segmentations of cell type 1 (E) and 2 (F). Arrows indicate cell body locations.

Sparse genetic reagents confirmed that activation of cell type 1 elicited male-like social behaviors
From the six positive GAL4 lines, we generated corresponding split-GAL4 hemi-drivers to be combined pair-wise (Luan et al., 2006;Pfeiffer et al., 2010) and Hap1 lines using an alternative yeast-derived Hap1-  Figure 2F). Remarkably, induced following behavior persisted, even 15 min after returning to 25°C from 10-min activation at 31°C ( Figure 2G). Using VT8469.Hap1, we found that cell type 1 is colabeled by dsx-GAL4 ( Figure 2I), together with its cell body position, confirming that it indeed belongs to the dsx + pC1 cluster. Its morphology, revealed by these sparse drivers and single MARCM clones, overall resembles that of pC1 cluster, as previously defined by 71G01-LexA∩dsx-GAL4, but clearly differs from those revealed by multicolor flip-out analysis (Zhou et al., 2014). Its cell number (1.91±0. 20, n=16) is fewer than what has been reported for pC1 (8.0±2.0) or its 71G01-LexA∩dsx-GAL4 defined subset (6.6±0.3) (Zhou et al., 2014). We also confirmed that it was labeled by all the screen-identified GAL4 lines, by examination of derived split-GAL4 combinations ( Figure S2A to C) or the original positive GAL4 lines colabeled with VT8469.Hap1 ( Figure S2D). These data suggest that cell type 1 is likely a femalespecific stereotypic subtype of dsx + pC1. Therefore, we refer to it as pC1-alpha for its ability to induce male-like behaviors and conclude that the female-specific pC1-alpha is functionally responsible for inducing persistent male-like social behaviors in females. anti-GFP (green), and nc82 (magenta). A total of four pC1-alpha neurons per brain were co-labeled among the two drivers and dsx-GAL4. Another two neurons in the vicinity were only weakly labeled by Split-1-GAL4 (H, arrows).

Single pC1-alpha neurons induce position-and direction-selective male-like following with directional neural signals
During the VT2064.GAL4 MARCM experiment, we noticed that one of the mosaic females specifically labeled a single pC1-alpha neuron in the brain ( Figure 3A) and directed its induced male-like following behavior only toward target flies moving in a specific direction (Fly 1; Figure 3A; Movie S4). To confirm this observation, we performed a second MARCM experiment using flies carrying VT8469.Hap1 HBS-GAL4 transgenes, effectively converted into a GAL4 driver. We obtained another four flies that labeled one or two ipsilateral pC1-alpha neurons (Fly 2 to 5; Figure 3B). In these flies, male-like following was often discontinuous, fragmented, and only observed when target flies moved across their field of view in contralateral-to-ipsilateral directions: Fly 1 with left pC1-alpha activated followed mainly toward flies moving in front of her from right to left, while Fly 2 to 5, which labeled the right pC1-alpha, showed following behavior toward flies moving from left to right ( Figure 3C, D and F; Movie S4). This was in direct contrast to flies with bilateral pC1-alpha neurons activated, who showed longer bout duration and pursued target flies not only moving across their field of view from both directions, but also those directly in front of them ( Figure 3D and G; Movie S4). We used MateBook (Ribeiro et al., 2018) to extract positional and angular information of both test and target flies ( Figure 3E to G) and manually annotated video recordings for frames where test flies showed following behavior (Movie S4). We compiled frames of about 20 following bouts from each of the five MARCM flies (with direction for Fly 1 reversed), together with those from five Split-1 UAS-trpA1 flies, and examined the distribution of distance and relative angles Θ 01 and Θ Heading between test and target flies ( Figure 3E). Distance was similarly biased toward small values for both MARCM and Split-1 flies, but angles Θ 01 and Θ Heading were specifically biased to the ipsilateral side for MARCM flies ( Figure 3H) and to both sides for Split-1 UAS-trpA1 flies ( Figure 3I).
We conclude from these data that activation of a single female pC1-alpha neuron is sufficient to induce male-like following behavior in a position-and direction-selective manner with a bias toward the ipsilateral side.
Such ipsilateral bias led us to examine the information flow in the complex arborizations of pC1-alpha neurons. Expression of a somato-dendritic marker DenMark (Nicolai et al., 2010) and a presynaptic marker synaptotagmin-GFP (Zhang et al., 2002) in all pC1-alpha neurons showed an almost complete overlap of the two markers ( Figure 3J). However, when expressed in only one or two ipsilateral cells in mosaic females generated by MARCM, the two markers partially separated. Namely an exclusively dendritic ipsilateral ventral and an exclusively synaptic contralateral dorsal region, while the ipsilateral dorsal region displayed both markers ( Figure 3K, n=8). Such a pattern may suggest an ipsilateral-tocontralateral and ventral-to-dorsal flow of information, which may be modulated in between.

pC1-alpha requires acetylcholine for following behavior induction but not normal sexual receptivity
We subsequently examined which neurotransmitter pC1-alpha neurons used for induction of male-like social behaviors. Antibody staining against choline acetyltransferase (ChAT), which catalyzes the biosynthesis of the neurotransmitter acetylcholine (ACh), showed co-localized signals with pC1-alpha neurons, suggesting that they are cholinergic ( Figure 4A). In addition, when we knocked down ChAT specifically in pC1-alpha neurons, by expressing two different RNAi lines with Split-1-GAL4, induction of male-like behavior was significantly reduced or abolished ( Figure 4B). We conclude that pC1-alpha neurons induce male-like following behavior via ACh.
Since dsx + pC1 neurons play an important role in female sexual receptivity (Rezaval et al., 2016;Zhou et al., 2014), we asked whether activity of the specific pC1-alpha subtype is essential for normal female receptivity, and if so, whether ACh is required for its function. We first inactivated pC1-alpha neurons by combining Split-1 to 4-GAL4 with a UAS-TNT transgene, which encodes an inhibitor of synaptic transmission (Sweeney et al., 1995). We observed a dramatic reduction of copulation in comparison with the controls, suggesting an essential role of pC1-alpha in maintaining normal female receptivity ( Figure 4C). However, when we expressed the two ChAT RNAi lines in pC1-alpha neurons, female receptivity remained comparable to controls ( Figure 4D), suggesting a potential co-transmission mechanism (Kupfermann, 1991;Nusbaum et al., 2001). However, it is also possible that our RNAi lines may have reduced ChAT expression in pC1-alpha neurons to a level that could block the induction of male-like behavior, which requires stronger activation, but leave intact female receptivity functions, which may require weaker activation. (D) RNAi knock-down of ChAT had no effect on female receptivity. Data are represented as mean ± SD. n.s., p>0.05 when compared to control at 10min, Chi-square test.

Discussion
The dsx + pC1 neurons, previously reported to constitute the latent courtship circuit in the brain of Drosophila females, were characterized as a functional unit that is sex-common and cell-number dependent in eliciting male-like courtship (Rezaval et al., 2016). Here, we identify the exact cell type that is responsible for triggering such male-like social behavior in females and show its female specificity and sufficiency in recapitulating the overall phenotypes observed with the pC1 cluster. Our results reveal that pC1 neurons, although all expressing dsx, are likely rather heterogeneous, and it is the cell identity rather than the cell number that determines their function. In our case, activation of a single pC1-alpha neuron and inactivation of the whole pC1-alpha cluster are sufficient to induce male-like social behaviors and suppress sexual receptivity, respectively.
Our findings suggest that activation of pC1-alpha neurons triggers a visually guided following behavior that encourages overall social interactions, rather than just male-like courtship behavior alone. The outcome of such increased social interactions may be either courtship or aggression, depending on social feedback (Movie S1 and S3). When a female is paired with a target fly, visual sensory inputs of the target are constantly processed, but normally do not trigger male-like following behavior in the female.

Its induction by pC1-alpha activation suggests a link between motion detection and locomotion
circuitries, which is likely gated by pC1-alpha activity. Motion of small objects, such as a fly, is detected by the LC10 visual projection neurons to direct courtship behavior in males and likely male-like social behaviors in females as well (Ribeiro et al., 2018). Given that LC10 neurons are not selective to movement direction (Ribeiro et al., 2018), the position-and direction-selectivity of unilateral pC1-alpha activation likely emerges downstream of LC10, possibly by potentiation of downstream descending neurons that lead to turning and forward walking. It would be interesting to identify the exact location where visual information is integrated with pC1-alpha gating signals.
The activity of pC1-alpha likely promotes an internal state of arousal or motivation, due to its induction of persistent social behaviors in females paired with a target and the lack thereof in isolated females.
This, in addition to its morphology, dsx expression and sex-specificity, is highly reminiscent of male P1 (Anderson, 2016). P1 likely also shares the same DM4 lineage with pC1-alpha, based on their homologous morphology ( Figure S4A to D) (Ren et al., 2016). However, unlike P1 which fates to die and is faithfully rescued by blocking apoptosis in females (Kimura et al., 2008), pC1-alpha may never exist in males, as blocking apoptosis throughout DM4 lineage development failed to rescue male dsx + neurons with the characteristic medial ventral neurites of pC1-alpha ( Figure 3A, B, J and K, Figure S2E to G, double arrowheads) (Ren et al., 2016). Thus, we speculate that pC1-alpha likely represents a femalespecific "de novo" P1-like cell type, which may be a result of nature's tinkering (Jacob, 1977). We further speculate that such neuronal organization may allow sufficient flexibility to encode new mating strategies during evolution (Donegan and Ewing, 1980), including the extreme cases of sex role reversal, where females compete for male mates (Kokko et al., 2006).
In summary, our results demonstrate that activation of a single female-specific dsx + pC1-alpha neuron is sufficient to induce male-like social behaviors in a position-and direction-selective manner, and activity of pC1-alpha neurons as a whole is indispensable for maintaining normal sexual receptivity. Cholinergic neurotransmission of pC1-alpha neurons is specifically required for male-like behavior induction, but not sexual receptivity regulation. Our findings suggest a link between male-and female-specific higher-order neurons that both promote a persistent internal state to enhance social behaviors, identifying pC1-alpha as P1's female counterpart likely co-opted from P1-like templates to regulate female sexual receptivity.
Therefore, our study provides additional circuit hypotheses that can be tested in further studies.

Declaration of Interest
The authors declare no competing interests.

Behavioral assay and analysis
For activation with TrpA1, flies were raised at 22°C and collected shortly after eclosion, then aged in groups of 15-20 in separate vials for 10-14 days at 22°C before testing. During testing, flies were paired with whitewild type flies or TrpA1-expressing test flies and assayed for following behavior in 10mm chamber for 10min under different temperatures. Ambient temperatures were adjusted and maintained constant, using a heating glass ceiling and a power supply with a feedback controller. Percentage of time spent in following within a 10min assay, the following index, was used as a measure of following behavior. Videos were either automatically tracked using Matebook (Ribeiro et al., 2018) (Figure 2C to F, 4B) or manually analyzed ( Figure 2G, 3C and D, 4C and D), or both ( Figure 3F to I).

MARCM (mosaic analysis with a repressible cell marker)
Flies carrying appropriated transgenes were crossed and progenies from the crosses were heat shocked at late embryonic and early larval stages for 1-1.