To look for evidence of reproductive satiety in Drosophila, single males that had been isolated from females for 3–6 days were placed in food vials containing ∼25 virgin females and examined for mating behaviors (courtship or copulation) every 30 min. Initially, males spent ∼80% of their time engaged in mating behaviors (courtship and copulation), but the percentage gradually declined to ∼10% over 4 hr (Figures 1A and S1A, available online). During the assay, males mated an average of 3.3 ± 0.3 times with an average copulation duration of 24.6 ± 0.4 min that did not change with repeated matings. We use the term satiety to reflect the gradually diminishing propensity to engage in mating behaviors as the assay progresses. To quantify this change, we calculate a satiety index: the fraction by which mating behavior is reduced in the last 60 min relative to the first 60 min of the assay, which is 0.82 ± 0.02 for wild-type flies (Figures 1A and 1B; see Experimental Procedures). Robust reproductive satiety is seen at the level of both courtship (satiety index 0.96 ± 0.02; Figure 1C) and copulation (satiety index 0.82 ± 0.03; Figure 1D). Unlike other aspects of male mating behavior, satiety is not affected by the male’s prior social experience with other males (Dankert et al., 2009,Inagaki et al., 2014,Kim et al., 2012) (Figure S1B).

Figure 1Reproductive Satiety in Male Flies

Satiety is not the result of habituation to the presence of females, but requires copulation. Preventing copulation through the use of low-receptivity females (Nakayama et al., 1997), or males that have physical defects in external genitalia (Castrillon et al., 1993), resulted in high levels of courtship that were maintained throughout the satiety assay (Figures 1E and S1D). The reduction in mating behavior seen at the end of the assay was not altered by placing the satiated male with fresh virgin females (Figure 1F), but gradually recovered over the course of 3–4 days if the male was removed from females (Figure 1F). The reduction in mating drive in satiated males was also seen in standard courtship assays (one male and one female in a 15 min assay, used throughout this study when assaying for hyposexuality; see Experimental Procedures) (Figures 1G and 1H). When satiated males did court, they spent a comparable proportion of courtship time performing unilateral wing vibration (“singing”), but their courtship was often transient (Figures 1G and 1I). Satiety is unlikely to be a consequence of physical exhaustion, because locomotor activity was not changed by completion of the satiety assay (Figure 1J). We conclude that, similar to mammals (Beach and Jordan, 1956), male Drosophila exhibit a transient reproductive satiety induced by mating, a hallmark of sexual motivation.

Reproductive satiety parallels the male’s diminishing reproductive capacity after repeated matings. This reduced potency is evident in males that have recently completed the satiety assay, as their subsequent copulations result in few, if any, progeny (Figure 1K) (also seeCrickmore and Vosshall, 2013,Lefevre and Jonsson, 1962,Linklater et al., 2007). A physical basis of this diminishing fertility is found in the male’s ejaculatory bulb, which houses mature sperm and seminal fluid (Demerec, 1950). The volume of the bulb gradually decreases over the course of repeated matings, with kinetics redolent of the decline in mating behaviors (Figure 1L). By the end of the satiety assay, reproductive fluids are largely depleted and the lumen of the bulb has shrunken substantially, a presumptive cause of the impaired fertility of males in a state of reproductive satiety. The fullness of the bulb recovers after the satiety assay, though only to ∼75% after 3 days away from females (Figure 1M).

Since mating drive roughly tracks ejaculatory bulb volume, we asked if information about the volume of the bulb is used to gate the execution of sexual behaviors. To allow copulation but prevent depletion of reproductive fluids from the bulb, we performed a modified satiety assay wherein copulations were artificially terminated 5 min after initiation—before reproductive fluids are transferred (Crickmore and Vosshall, 2013,Gilchrist and Partridge, 2000,Tayler et al., 2012). As expected, the ejaculatory bulb remains full throughout this modified assay (Figure 2A). Unexpectedly, though, we find that truncated copulations trigger satiety to the same degree as normal copulations: the number of matings per male in the assay remained unchanged when copulations were disrupted at 5 min (Figure 2B), despite the cumulative time spent in copulation being dramatically reduced (Figure 2C). Normal satiety in males with truncated copulations is also seen in a courtship assay (Figure 2D). We then examined the consequences of artificially emptying the ejaculatory bulb without allowing the male to copulate. Thermogenetic activation of corazonin (crz) neurons causes repeated ejaculation, even in the absence of copulation (Tayler et al., 2012). Ejaculatory bulbs of males in which the warmth-sensitive cation channel TrpA1 (Hamada et al., 2008) was targeted to crz neurons (crz > TrpA1) were depleted after incubation at 30°C for 3 days, but this treatment had no effect on courtship behavior (Figure 2E). Moreover, recovery of mating drive after the satiety assay was unimpeded by artificially maintaining the empty status of the ejaculatory bulb (Figure 2E). The finding that ejaculatory bulb fullness is entirely dissociable from courtship vigor is consistent with the delayed recovery kinetics of bulb fullness compared to mating drive (Figure 1M). Together, these results argue against a direct role of sperm and seminal fluid accumulation or depletion in instructing mating drive. To ask whether reproductive organs have any role in the acute determination of mating drive, we surgically removed the posterior abdomen, which contains the internal and external reproductive organs. This surgery had no major detrimental effect on courtship in naive males, and did not stimulate courtship in satiated males (Figure 2F; Movie S1). These results show that though mating drive is calibrated to reflect reproductive capacity, the reproductive organs themselves have no acute influence on the male’s response to the presentation of a virgin female. Though sensory information from the genitalia may signal the onset of copulation, the resulting adjustments to and maintenance of mating drive are carried out within the CNS.

To identify the genes and neurons that control mating drive in male Drosophila, we examined over 1,500 genetic manipulations (see Experimental Procedures) in the satiety assay in search of animals that maintained high levels of mating behaviors after repeated copulations. The only hit obtained in this screen resulted from thermogenetic activation of dopaminergic neurons. When dopaminergic neurons were stimulated at the end of the satiety assay using tyrosine hydroxylase-Gal4 (TH > TrpA1; TH is an enzyme required for dopamine synthesis), satiated males showed a dramatic rebound in mating behaviors (Figure 3A). The rebound is seen in both copulation (58% decrease in satiety) and courtship (84% decrease in satiety) (Figures S2A and S2B). This finding parallels previous work showing that dopaminergic neurons of the ventral nerve cord (VNC) promote the male’s motivation to sustain individual copulation bouts (Crickmore and Vosshall, 2013). In that study, dopaminergic function was localized to the VNC due to the ability of TshGal80 (which inhibits the activity of Gal4 in most VNC neurons) to block Gal4-mediated phenotypes. In contrast, TshGal80 has no effect on the reversal of satiety seen in TH > TrpA1 males (Figures 3A, S2A, and S2B). Together, these findings reveal two functionally and anatomically distinct roles for dopamine in promoting mating drive in flies. The initial drive to mate is set by dopamine neurons of the brain, and once copulation is initiated, dopaminergic neurons of the VNC determine the persistence and duration of the mating.

To identify which brain dopaminergic neurons underlie mating drive, we thermogenetically activated four subsets of dopaminergic neurons using Gal4 lines derived from enhancer elements of the pale locus, which encodes TH (Liu et al., 2012). Stimulation of either of two subsets of TH neurons (TH-C’ and TH-D’; Figures 3B and 3C) resulted in partial recovery from satiety. This reversal of satiety is not a secondary effect of increased locomotion, as stimulating either population does not significantly impact basal activity (Figure S3A). The activity of neurons within these populations is necessary for mating drive, as adult-specific silencing of TH-C’ or TH-D’ neurons with the potassium channel Kir2.1 decreased courtship (Figure 3D). Dopamine itself is likely the mating drive signal, as lowering TH production in TH-C’ or TH-D’ neurons with pale-RNAi dramatically reduced courtship (Figure S4).

To ask whether the mating drive dopaminergic neurons lie within the Fruitless circuitry, we introduced a Gal80 transgene whose expression pattern is indirectly controlled by the cis-regulatory elements of fruitless (see Figure 3B legend). Thermogenetic stimulation in these flies (TH-C’ > TrpA1;FruGal80 and TH-D’ > TrpA1;FruGal80) fails to reverse satiety (Figure 3B), arguing that mating drive dopamine neurons are Fruitless positive. In both TH-C’ and TH-D’, FruGal80 blocked Gal4 activity in a single neuron per brain hemisphere, and in both cases this neuron was located in the PAL dopaminergic cluster (Mao and Davis, 2009) (Figures 3E and S5). In adults, no TH-C’ or TH-D’ neurons stain with Fruitless antibodies (data not shown), but in both populations, an FLP recombinase transgene inserted into the fruitless locus (FruFLP) identifies single dopaminergic neurons belonging to the PAL dopaminergic cluster (Figure 3F) (Keleman et al., 2012, Mao and Davis, 2009). This neuron is called aSP4 (Keleman et al., 2012, Yu et al., 2010) and has been shown to impair courtship behavior when feminized in males (von Philipsborn et al., 2014). TH-F and TH-G do not label any neurons in the PAL cluster, including aSP4 (Liu et al., 2012). aSP4 projects to several dorsal brain regions (von Philipsborn et al., 2014), including the anterior of the superior medial protocerebrum (SMPa) (Ito et al., 2014) (Figure 3F).

While activation of aSP4 is necessary to reverse satiety in this system, it does not appear to be sufficient, as thermogenetic activation of VT02857-Gal4, which targets aSP4 (Keleman et al., 2012), did not increase mating behavior in satiated males (Figure 3G), even when two copies of the Gal4 were used to increase expression (data not shown). This suggests the existence of other dopaminergic neurons that work with aSP4 to promote mating drive. We therefore screened ∼30 Gal4 lines labeling dopaminergic neurons, looking for lines that could overcome satiety when activated together with aSP4. We identified NP5945-Gal4 (Kuo et al., 2015), which does not label aSP4 and did not reverse satiety when stimulated alone, but did increase mating behavior from satiated males when stimulated together with VT02857 (Figure 3G). Silencing either VT02857 or NP5945 neurons in adults using Kir2.1 decreased courtship behavior (Figure 3H), suggesting that the combined activity of these two neuronal populations is necessary to promote the normal accrual of mating drive. NP5945 labels non-Fruitless dopaminergic neurons in four clusters (PPL2ab, PPL2c, PPM2, and PPM3) and has projections to the SMPa (Figure 3I). Although we cannot currently isolate the neurons within NP5945 that promote mating drive together with aSP4, our results suggest the SMPa as a locus for the dopaminergic control of sexual motivation in males.

As a first step toward identifying the neuronal targets of dopamine signaling relevant to mating drive, we examined courtship in animals carrying mutations in the four Drosophila dopamine receptors. When tested at 3 days old, a combination of mutations in three of the receptors showed normal levels of courtship, whereas a deletion of the fourth, DopR2 (a D1-like receptor) (Han et al., 1996,Keleman et al., 2012), resulted in dramatically reduced courtship without affecting locomotor activity (Figures 4A and S3B). These data initially seemed to be in conflict with a previous study reporting normal levels of courtship in these same DopR2 mutants in the same genetic background (Keleman et al., 2012). Further examination of these mutants led us to a likely explanation for this discrepancy: we found that the mating drive of DopR2 animals gradually increased over 10–15 days (Figure 4B), as did, to a lesser extent, the mating drive of wild-type males (Figure S1C). This suggests that as mating drive increases with age, additional dopamine receptors gradually become capable of receiving the mating drive signal. DopR2 functions in neurons (as opposed to non-neuronal cells) to promote the timely accumulation of mating drive, as knockdown of DopR2 with the pan-neuronal elav-Gal4 driver phenocopied the deletion mutant (Figure 4C). Matings involving 3-day-old DopR2 mutant males were fully fertile (data not shown), so the behavioral phenotype is not a secondary consequence of sterility.

Within the Fruitless circuitry, much attention has focused on a locus composed of ∼40 neurons, called P1, that is required for mating behavior (von Philipsborn et al., 2011). Stimulation of a subset of P1 neurons (labeled by the Gal4 line R71G01, here referred to as P1-B in reference to the Baker lab where it was first characterized) has been shown to drive courtship behaviors toward moving, fly-sized inanimate objects (Kohatsu and Yamamoto, 2015,Pan et al., 2012). In our hands, activation of P1-B neurons drives robust courtship behavior toward even stationary rubber band pieces—if they are painted black (Figure 5A; Movie S2). This phenotype is distinct from activation of the entire fruitless circuit, which drives courtship in the absence of a target (Figure S6A), and from activation of dopamine neurons, which does not cause courtship of inanimate objects (Figure 5A).

Based on anatomy and the Fruitless wiring diagram, the Dickson lab suggested that the aSP4 dopamine neuron may connect to P1 neurons (von Philipsborn et al., 2014,Yu et al., 2010). We found that activation of P1-B neurons drives robust courtship in satiated males (Figure 5B), as well as in 3-day-old males that lack DopR2 (Figure 5C). Activating P1 neurons therefore bypasses the requirement for state-dependent dopaminergic input, indicating that they are functionally downstream of the dopamine mating drive signal.

Are P1 neurons the direct targets of dopaminergic signaling? As noted by the Dickson lab (von Philipsborn et al., 2014,Yu et al., 2010), co-labeling P1-B neurons and dopaminergic neurons (Figure 5D) shows intermingling of neuronal processes between these two populations. We detected a strong GRASP (GFP reconstitution across synaptic partners) (Feinberg et al., 2008) signal between the two populations, indicating potential sites of synaptic connectivity (Figures 5E and S6B). Much of the GRASP signal was located in the SMPa, a region that is targeted by the aSP4 dopaminergic neuron (Figure 3F) as well as the NP5945 neurons (Figure 3I). When we knocked DopR2 down in P1 neurons, we saw a profound reduction of courtship behavior that recovered with age (Figure 5G), phenocopying the DopR2 mutant. This is strong evidence that P1 neurons receive the dopaminergic mating drive signal.

The P1-B Gal4 line labels some cells outside of the sexually dimorphic P1 cluster (Pan et al., 2012). When FruGal80 is introduced, P1-B labeling within the P1 cluster is greatly reduced, but expression outside the cluster is not obviously affected (Figure 5F). Knocking down DopR2 in these remaining neurons had no effect on courtship behavior (Figure 5H), confirming that the Fruitless-positive P1 neurons receive and interpret the dopamine signal. To ask whether additional neurons receive the mating drive signal through DopR2, we blocked Gal4 activity in P1-B neurons (see Figure 5I legend) while knocking down DopR2 in all other neurons. Blocking pan-neuronal DopR2 knockdown in P1-B neurons restored courtship to normal levels (Figure 5I), arguing that the dopaminergic mating drive signal is received exclusively by P1.

The presumptive synaptic contacts between dopaminergic and P1 neurons in the SMPa region led us to monitor the activity of dopamine neurons in this area in males of varying reproductive states. We drove the calcium sensor GCaMP6s (Chen et al., 2013) in dopaminergic neurons and found that fluorescence levels in the SMPa tracked the male’s level of satiety. Similar to mating drive, fluorescence levels declined gradually over the course of the assay, to the point that they were reduced by ∼80% after 4.5 hr (Figures 6A and 6B). Also similar to mating drive, dopaminergic activity in the SMPa gradually recovered from the satiety assay over the course of 3 days of isolation from females (Figure 6B). The aSP4 neuron accounts for much of the dopaminergic activity in the SMPa, as FruGal80 blocks ∼65% of the fluorescence in naive animals (Figure 6C). Fluorescence did not noticeably change with mating history in an unrelated region, the optic lobe (Figure 6A), or in a neighboring neuropil, the medial lobe of the mushroom body (Figure 6B). Nor were SMPa fluorescence changes observed in the calcium-insensitive fluorescent protein tdTomato (p > 0.95) or with immunostaining of fixed TH > GCaMP6s brains (p > 0.92) (Figures S7A and S7B). The mating-history-dependent GCaMP6s fluorescence change in SMPa was also detected using VT02857-Gal4, which labels aSP4 and a few other dopaminergic neurons (Figure 6D) (Keleman et al., 2012,Yu et al., 2010), as well as NP5945-Gal4, which labels neurons in four dopaminergic clusters (Figures 6E and 3I) (Kuo et al., 2015). These results show that dopaminergic activity in the vicinity of the connection to P1 neurons serves as a neuronal correlate of male mating drive.

The decreased basal activity of dopaminergic projections in the SMPa suggests that these neurons may become less excitable following repeated matings. To test this idea directly, we drove the red-light-sensitive ion channel CsChrimson (Hoopfer et al., 2015,Klapoetke et al., 2014) together with GCaMP6s in dopaminergic neurons and used two-photon microscopy to image optogenetically induced calcium transients in the SMPa. At all stimulation lengths, dopaminergic SMPa fluorescence peaks were lower in satiated brains than in naive brains, or in brains tested 3 days after completion of the satiety assay (Figures 6F and 6G). Fluorescence decay kinetics did not change with mating history (Figure 6H). Mating therefore decreases the excitability and/or calcium influx of SMPa-projecting dopaminergic neurons while leaving calcium buffering properties unchanged. These long-lasting changes suggest a molecular mechanism acting either on the inputs to dopamine neurons, or within the dopamine neurons themselves. The resulting dopamine tone in the SMPa serves as a neuronal correlate of mating drive, gating the activation of P1 neurons in response to sensory stimulation from females.

Courtship assays were carried out and videotaped in cylindrical courtship chambers (10 mm diameter and 3 mm height) at 23°C and ambient humidity. Unless otherwise stated, a male fly (3–6 days old) was paired with a w¹¹¹⁸ virgin female fly in each chamber. The courtship index was manually scored as the fraction of time during which the male fly is engaged in mating behaviors (courtship and copulation) within 5 min once the male fly started courting. In this paper, we manually scored courtship assays to keep consistent with the courtship scoring in the satiety assays, whose 3D environment and large number of animals complicate automated scoring. A bout of courtship was scored as initiated when the male oriented toward the female, began tracking her, and unilaterally extended his wing to sing to her. A bout was scored as terminated if the male did not sing in the subsequent 30 s or turned away from the female. If the male fly did not court within the first 15 min of each assay, the courtship index was scored as 0. Using these criteria, two experimenters (M.A.C. and S.X.Z.) routinely gave near-identical courtship indices in pilot assays. The same criteria were applied when producing the ethograms in Figure 1G, where percent time singing during courtship was calculated as time_spent_singing/total_time_spent_courting × 100%. In thermogenetic experiments, the courtship assays were performed first at 23°C and then again at the higher temperature. For experiments involving tubGal80^ts (Figures 3D and 3H), male flies were moved to 30°C after eclosion (Figure 3D) or 1 day after puparium formation (Figure 3H) and kept there (isolated from females) until just before the assay, which took place at 23°C. For experiments involving DopR2 mutants or RNAi knockdown, only 3-day-old males were used to avoid the confounding effects of age. The DopR2^attp animals had previously been introgressed into a wild-type Canton-S background (Keleman et al., 2012), which served as the wild-type background control in our behavioral experiments. To further ensure background consistency, we re-crossed the mutant into the Canton-S background three more times, with no effect on courtship. We therefore combined the courtship data from before and after the second introgression.

Locomotion assays were carried out in the cylindrical chambers (30 mm diameter and 3 mm height) at 23°C and ambient humidity. Canton-S males (3–6 days old), half of which had just completed the satiety assay, were individually gently aspirated into the chambers and allowed to explore the chambers for 1 min before their locomotion was recorded with a standard video camera at 30 fps for 10 min. Video segmentation and tracking were carried out using FIJI (Schindelin et al., 2012) with the MTrack2 plugin (http://valelab.ucsf.edu/∼nstuurman/ijplugins/MTrack2.html).

Fly brains were removed in PBS or Schneider’s medium (ThermoFisher, 21720-024) and immediately fixed in PBST and 4% paraformaldehyde at room temperature for 20 min. After washing with PBST three times (20 min each), the brains were incubated with primary antibodies (diluted in PBST) at 4°C for 48 hr. Following a PBST wash (three times for 20 min), the brains were incubated with secondary antibodies (diluted in PBST) at 4°C for 48 hr. After three more PBST washes (20 min each), the brains were mounted with VectaShield (Vector Labs) on glass slides using standard procedures. Confocal sections were acquired using an Olympus Fluoview 1000 microscope or an Olympus Fluoview 1200 microscope at 3–5 μm intervals, and maximum projections of image stacks were obtained in FIJI (Schindelin et al., 2012).

Fly brains were dissected from 3- to 8-day-old males in external saline (Wilson and Laurent, 2005) and mounted, anterior side down, onto the base of a glass-bottom petri dish (MatTek, P35G-1.5-20-C) housing 4 mL static external saline. To stabilize the samples, optic lobes were adhered to the coverslip with UV glue (Kemxert, KOA 300) solidified with a 5 mw purple laser pointer (Oxlasers, OX-B005). To image the SMPa, four to seven confocal sections spanning this region were acquired using an Olympus Fluoview 1000 microscope at 3 μm intervals, and maximum projections of image stacks were obtained in FIJI (Schindelin et al., 2012). After one of us (S.X.Z.) took the images, another of us (D.R., blind to the treatment) selected regions of interest (ROIs) of ipsilateral SMPa, mushroom body medial lobe, and background (near the antennal lobe, where dopamine projections are sparse) using the tdTomato channel (Figures 6B, 6D, and 6E) or the GCaMP6s channel (Figure 6C). ROI sizes were kept roughly the same between samples. Average pixel fluorescence in each ROI was calculated with a custom-written program MATLAB (Mathworks). Normalized fluorescence was calculated as (GCaMP6s_SMPa − GcaMP6s_background)/(tdTomato_SMPa − tdTomato_background). The same formula was used for the mushroom body medial lobe as well. The program is available upon request.

Two-photon laser-scanning microscopy was performed using a custom microscope as previously described (Carter and Sabatini, 2004). Shortly after eclosion, flies were transferred to vials containing rehydrated potato flakes (Carolina Biological Supply, 173200) and 100 μL of all-trans-retinal stock solution (Sigma, R2500; 35 mM in ethanol). All experiments were conducted with adult male flies within 3–6 days after eclosion. Prior to imaging, fly brains were carefully dissected under low-light conditions and mounted, posterior side down, onto a coated petri dish (Thermo Scientific, 150318) containing 3 mL HL3.1 saline solution (Feng et al., 2004). GCaMP6s was excited at 920 nm and 11.8 mW at the sample, and light was collected in 128 × 128 frames at 256 ms per frame. A basal level of retinal-dependent CsChrimson activation was observed at 920 nm (and at all wavelengths tried between 860 and 940 nm) that typically plateaued within the first 20 frames (∼5 s). At frame #24, a pulse of red light (655 nm) was delivered using an LED (Luxeon Star LEDs, LXM3-PD01-0350), which was driven by a 720 mW BuckPuck (Luxdrive, 3021-D-E-700), and collimated with a lens (Carclo, 13193) ∼3 cm away from the sample. No fluorescence data were collected during stimulation, in order to protect the photomultiplier tube. For each experimental series, light pulses of different widths were delivered in increasing, decreasing, or random order. Fluorescence data were analyzed in MATLAB using a semi-automated program modified from a published independent component analysis algorithm (Mukamel et al., 2009). ecay half-life was estimated by fitting the fluorescence decay data with a single-exponential function. All time constants measured in Figure 6H are longer than that of GCaMP6s itself (<1.2 s in Drosophila) (Chen et al., 2013). The program is available upon request.