|Home | About | Journals | Submit | Contact Us | Français|
The elaborate courtship ritual of Drosophila males is dictated by neural circuitry established by the transcription factor Fruitless and triggered by sex-specific sensory cues. Deciphering the role of different stimuli in driving courtship behavior has been limited by the inability to selectively target appropriate sensory classes. Here, we identify two ion channel genes belonging to the degenerin/epithelial sodium channel family, ppk23 and ppk29, which are expressed in fruitless-positive neurons on the legs and are essential for courtship. Gene loss-of-function, cell inactivation and cell activation experiments demonstrate that these genes and neurons are necessary and sufficient to inhibit courtship toward males and promote courtship toward females. Moreover, these cells respond to cuticular hydrocarbons, with different cells selectively responding to male or female pheromones. These studies identify a large population of pheromone-sensing neurons and demonstrate the essential role of contact chemosensation in the early courtship steps of mate selection and courtship initiation.
Innate behaviors, from egg-rolling in geese to the honeybee waggle dance, are executed by genetically programmed neural circuits that are triggered by specific sensory cues. The Drosophila courtship ritual is comprised of a sequence of stereotyped behaviors that culminates in copulation, essential for propagation of the species. Courtship has emerged as a model for deciphering the neural basis of innate behavior because of its tight genetic control by the transcription factor Fruitless (Gill, 1963; Hall, 1978; Ito et al., 1996; Ryner et al., 1996). However, the sensory cues and neurons that initiate courtship behavior are only beginning to be elucidated.
Courtship involves a complex sequence of actions by the male. The male orients towards a female, chases her, taps her abdomen with his forelegs, plays a courtship song by extending and vibrating a single wing, contacts the ovipositor with his proboscis and finally mounts before copulation (Hall, 1994). This stereotyped behavior is dictated by the male-specific splice form of Fruitless, FruM (Goodwin et al., 2000; Ryner et al., 1996). fru mutants show reduced male-female courtship and enhanced male-male courtship (Ryner et al., 1996; Villella et al., 1997). Moreover, transgenic studies in which fruM was selectively expressed in females caused them to perform nearly all aspects of male courtship (Demir and Dickson, 2005; Manoli et al., 2005).
FruM is found in ~1500 neurons in the fly brain that mark neural circuitry for courtship behavior (Lee et al., 2000; Manoli et al., 2005; Stockinger et al., 2005). FruM labels a pathway of synaptically connected neurons that detect sex-specific olfactory cues (Datta et al., 2008; Ruta et al., 2010; Stockinger et al., 2005). In addition, five different classes of FruM neurons elicit courtship song, suggesting that they comprise male-specific song circuitry (von Philipsborn et al., 2011).
In addition to sex-specific neural circuitry governed by FruM, appropriate courtship requires that males sense cues that trigger courtship toward females and prevent non-productive courtship toward males. Long-chain cuticular hydrocarbons (CHCs) produced by oenocytes (oe) act as non-volatile pheromones to trigger sex-specific behavior (Billeter et al., 2009; Ferveur, 2005; Ferveur et al., 1997). Flies lacking oenocytes (oe-) fail to evoke appropriate courtship behavior in wild-type males, indicating that cuticular hydrocarbons are an essential sensory component for courtship (Billeter et al., 2009). Female CHCs are enriched for 7,11-heptacosadiene (7,11-HD) and 7,11-nonacosadiene (7,11-ND), compounds that stimulate male courtship (Ferveur, 2005; Jallon, 1984). In contrast, male CHCs are enriched for 7-tricosene (7T) and 7-pentacosene (7P) (Ferveur, 2005; Jallon, 1984). In addition, a volatile hydrocarbon not produced by oenocytes, cis-vaccenyl acetate (cVA), is enriched in males (Butterworth, 1969). These compounds represent only a small, studied fraction of the complete hydrocarbon profile of Drosophila (Everaerts et al., 2010; Yew et al., 2009).
Despite the diversity of pheromones, only a handful of receptors and cell types have been implicated in pheromone sensing. Both olfactory and gustatory neurons mediate pheromone detection, with olfactory neurons detecting volatile cues and gustatory neurons sensing contact-mediated cues. The best-characterized olfactory receptor-ligand pair is Or67d-cVA (reviewed in Vosshall, 2008). The gustatory receptor genes Gr32a and Gr33a encode putative pheromone receptors for contact mediated male-male repulsion, as mutants show enhanced courtship of beheaded males (Miyamoto and Amrein, 2008; Moon et al., 2009). Additionally, Gr68a, Gr39a and the ion channel gene, ppk25, have been implicated in male-female attraction (Bray and Amrein, 2003; Ejima and Griffith, 2008; Lin et al., 2005; Watanabe et al., 2011). The pheromones that these candidate receptors recognize and their relationship to FruM neurons have not been established.
The subtle phenotypes seen upon compromising small subsets of sensory neurons contrast with the dramatic defects observed when sex-specific circuitry or pheromone cues are absent. This suggests that additional sensory populations are required to evaluate potential mates. Contact-mediated recognition occurs via chemosensory bristles on the proboscis, internal mouthparts, legs, wing margins and ovipositor (Stocker, 1994). Most chemosensory bristles on the proboscis contain four neurons, three of which sense sugars, bitter compounds or water (Cameron et al., 2010; Chen et al., 2010; Thorne et al., 2004; Wang et al., 2004). However, the modality sensed by the fourth population of gustatory neurons and the functional and behavioral relevance of this cell type is not clear.
Here, we identify two ion channel genes, ppk23 and ppk29, that are co-localized in the fourth population of gustatory neurons on the proboscis and in FruM-positive leg neurons. Gene loss-of-function, cell inactivation and cell activation experiments demonstrate that these genes and neurons are essential for recognition of males and females at early courtship steps. These studies identify a large population of chemosensory neurons responding to cuticular hydrocarbons and demonstrate the essential role of contact chemosensation in mate selection and courtship initiation.
We previously performed a microarray-based screen for genes enriched in taste neurons (Cameron et al., 2010). Three taste-enriched genes are members of the degenerin/epithelial sodium channel (Deg/ENaC) family. ppk28 is expressed in gustatory neurons that mediate water taste detection (Cameron et al., 2010). The other two genes are ppk23 and CG13568. CG13568, which we name ppk29, contains 24% predicted amino acid identity to ppk23. Because Deg/ENaC channels are important for detection of a variety of stimuli including water, sodium, acids, mechanosensory stimuli and peptides (Mano and Driscoll, 1999), we examined whether ppk23 or ppk29 participates in gustatory detection.
To visualize expression of ppk23 and ppk29, we generated transgenic flies to drive expression of reporters under the control of putative promoters, using the Gal4/UAS system. ppk23-Gal4 and ppk29-Gal4 drove expression of GFP in neurons on the proboscis, all legs and wing margins in both sexes (Figure 1A, Figure S1A and not shown). Moreover, axons projected to the subesophageal ganglion (SOG), the primary taste relay (Figure 1A and Figure S1A). No expression of ppk23 or ppk29 was detected outside of gustatory sensory neurons.
Two-color in situ hybridization studies with ppk23 and ppk29 demonstrated that both genes were co-expressed in the same proboscis population (93% [92/99] ppk29 neurons expressed ppk23, 72% [92/128] ppk23 neurons contained ppk29) (Figure 1B). In contrast, ppk29-Gal4 labeled far fewer cells than ppk23-Gal4 per labellum (ppk29-Gal4=5+/-2, ppk23-Gal4=22+/-2, n=10, t-test P=e-21), suggesting that the ppk29-Gal4 line under-represents ppk29 expression, limiting its usefulness. However, ppk23-Gal4 faithfully recapitulated ppk23 endogenous expression (84/93 [90%] ppk23-Gal4 neurons expressed ppk23, 84/84 [100%] ppk23 neurons expressed ppk23-Gal4) (Figure S1B). These data show that ppk23 and ppk29 are largely co-expressed and that ppk23-Gal4 reproduces ppk23 expression.
Which cells express ppk23 and ppk29? Previous studies have identified three different taste cell populations in the proboscis, including sugar-sensing cells labeled by Gr64f (Dahanukar et al., 2007), bitter sensing cells labeled by Gr66a (Thorne et al., 2004; Wang et al., 2004) and water sensing cells labeled by Ppk28 (Cameron et al., 2010; Chen et al., 2010). Another population marked by ppk11-Gal4 has been proposed to mediate salt taste (Liu et al., 2003b); however, ppk11-Gal4 is found in support cells in the adult (Figure S1C). Co-expression studies revealed that ppk23 was not in sugar- or water-sensing cells (Figure 1B). A few Ppk23-positive cells were also Gr66a-positive (9 cells), but these represented only a small fraction of all Ppk23-positive (9/37) or Gr66a-positive cells (9/69) (Figure 1B). Two-color immunohistochemistry confirmed these results (Figure S1 DE). These analyses demonstrate that ppk23 and ppk29 mark an uncharacterized population of gustatory neurons in addition to a few bitter cells, suggesting that these neurons detect a novel taste modality.
Although ppk23-Gal4 is expressed in the same number of proboscis neurons in males and females (male=22+/-2 neurons, female=21+/-2 neurons, n=5/sex, t-test P=0.4), we found that it is expressed in twice as many leg neurons in males than females (first three tarsal segments of foreleg, male=26+/-1 neurons, female=14+/-0 neurons, n=4/sex, t-test P=0.004; Figure 1A). In addition, axonal projections from the forelegs are sexually dimorphic: male foreleg axons cross the ventral nerve cord midline whereas female axons do not (Figure 2A). Previous studies have shown that the transcription factor FruM is expressed in leg gustatory neurons and confers this male-specific axon projection pattern (Mellert et al., 2010), indicating that ppk23 might be co-expressed with FruM and mark sexually dimorphic sensory neurons.
Double-labeling experiments showed that ppk23-Gal4 and fruP1-LexA (marking FruM cells) were co-expressed in leg sensory neurons, with almost complete overlap in expression (97% overlap; 66/68 cells) (Figure 2B). In the legs, ppk23-Gal4 and fruP1-LexA mark more than one cell underneath a bristle, consistent with previous studies of fruP1-LexA (Mellert et al., 2010). Candidate pheromone receptors Gr32a and Gr68a target different cells (Figure 2C). In addition, markers for sugar, bitter and water did not co-label fruM-positive leg neurons (Figure 2C). In contrast to leg expression, ppk23-Gal4 and fruP1-LexA were not co-expressed in the proboscis. Instead, fruP1-LexA appears to label the mechanosensory neuron based on morphology and lack of marker co-expression (Figure S2) (Falk et al., 1976). Despite the differential proboscis labeling, the co-expression with fruP1-LexA in the legs raised the possibility that ppk23 might label sensory elements of the courtship circuit.
To test the role of ppk23 and ppk29 in sensory detection, we generated gene deletions by FLP-recombination target (FRT) mediated trans-recombination (Parks et al., 2004). Δppk23, Δppk29 flies showed no significant defects in detection of sugars, bitter compounds, salt or water using proboscis extension assays (Figure S3).
Because ppk23 and ppk29 are expressed in sexually dimorphic, fruM-positive leg neurons, we hypothesized that they might mediate pheromone detection during courtship behavior rather than food recognition. Mutant flies and isogenic w1118 controls were paired with wild-type males, wild-type virgin females and other mutant males in a courtship paradigm. Δppk23 males showed vigorous courtship toward wild-type target males as measured by the number of unilateral wing extensions (Figure 3A). Moreover, when 6-9 Δppk23 males were placed together in a chamber, they serially courted each other, forming long chains in which males followed each other and produced courtship song (Figure 3B). These dramatic defects were rescued by introduction of a ppk23 transgene under the control of the ppk23-Gal4 promoter into the mutant background. In contrast, Δppk29 males showed no aberrant courtship toward other males. These phenotypes suggest that ppk23 is required for the detection of an inhibitory signal present on males to prevent inappropriate male-male courtship.
Wild-type males will court target males lacking oenocytes, the cells that produce cuticular hydrocarbons (Billeter et al., 2009). This attraction is due to an absence of inhibitory hydrocarbons on the target, unmasking an attractive olfactory cue (Wang et al., 2011). We tested whether Δppk23 male-male courtship relies on olfactory information and found that Δppk23 males lacking antennae showed a complete loss of male-male courting (Figure 3AB). This reinforces the model that Δppk23 males fail to detect inhibitory cuticular hydrocarbons on other males and instead detect an attractive cue via the olfactory system.
We tested if ppk23 and ppk29 are also important for courtship toward wild-type virgin females. In a thirty minute trial, both Δppk23 and Δppk29 males courted with reduced frequency, longer latency and often failed to court (Figure 3CD and Figure S4). Importantly, reintroduction of ppk23 into Δppk23 flies rescued the courtship defects. Similarly, reintroduction of ppk29 into Δppk29 flies using ppk23-Gal4 rescued the behavioral phenotypes. This argues that ppk29 is expressed in ppk23 cells. Together, these experiments suggest that both ppk23 and ppk29 are required for the detection of excitatory signals present on females during courtship.
As ppk23 and ppk29 are related members of the Deg/ENaC ion channel family, we wondered whether overexpression of one might compensate for the loss of the other. Introduction of UAS-ppk29 into Δppk23 using ppk23-Gal4 failed to rescue the courtship defects (Figure 3). Similarly, responses of Δppk29 males containing UAS-ppk23 and ppk23-Gal4 were identical to responses of Δppk29 males (Figure 3). Thus, the two genes have non-redundant functions in courtship.
To confirm and extend the mutant studies, we examined whether silencing ppk23 cells recapitulates the mutant phenotype. We expressed tetanus toxin light chain (UAS-TNT) in ppk23-Gal4 cells to block synaptic transmission (Sweeney et al., 1995) and examined courtship behavior. Males with silenced ppk23 cells increased single wing extensions to wild-type males (Figure 4A). They also increased courtship latency and decreased wing extensions toward wild-type females (Figure 4B and Figure S4). These findings are consistent with the Δppk23 studies.
Our expression studies indicated diversity in cell-types that contain ppk23. ppk23 is co-expressed with Gr66a in a few proboscis neurons and co-expressed with fruM in leg but not proboscis neurons. To decipher which neurons contribute to the courtship defects assayed, we inactivated subsets of ppk23 cells. lexAop-Gal80 transgenic flies were generated to inhibit expression of Gal4-dependent reporters in Gr66a-LexA cells or fruP1-LexA cells. ppk23-Gal4, UAS-TNT flies in which Gr66a cells expressed Gal80 showed male-male and male-female courtship defects similar to ppk23-Gal4, UAS-TNT flies (Figure 4AB). In contrast, ppk23-Gal4, UAS-TNT flies with fruM cells containing Gal80 showed no courtship to males and normal courtship to females, similar to wild-type controls (Figure 4AB). This argues that the ppk23, fruM-positive leg neurons are required for appropriate courtship and that the Gr66a cells do not significantly contribute.
The loss-of-function studies indicate that ppk23 cells are necessary to inhibit male-male courtship and promote male-female courtship. If ppk23 cells detect pheromones and actively mediate courtship behavior, then inducible activation of ppk23 cells may be able to drive courtship even in the absence of pheromonal cues. Courtship assays were performed with oe- flies as targets, as they lack the majority of cuticular hydrocarbons and wild-type males court both male and female oe- flies (Billeter et al., 2009). The temperature-gated cation channel dTRPA1 was expressed in ppk23-Gal4 neurons to conditionally activate these neurons upon thermal increases (Hamada et al., 2008). In these experiments, females were pierced through the head to prevent copulation and slow courtship (Gailey et al., 1984). At permissive temperature, ppk23-Gal4, UAS-dTRPA1 males and controls courted oe- males and females, as expected (Figure 4C). When dTRPA1 was activated at 30°C, ppk23-Gal4, UAS-dTRPA1 males reduced courtship toward oe- males and increased courtship toward oe- females (Figure 4C). When paired with oe+ flies (containing pheromones) at 30°C, ppk23-Gal4, UAS-dTRPA1 males did not court oe+ males but increased courtship toward oe+ females (Figure 4D). This suggests that endogenous male cues prevent male-male courtship, as expected. In addition, the enhanced attraction towards oe+ females suggests that endogenous female cues may be limiting and not maximally activate ppk23 cells.
These results demonstrate that activation of ppk23 cells drives appropriate courtship behavior. The observation that ppk23 cell activity inhibits male courtship yet triggers female courtship to flies lacking cuticular hydrocarbons implies that other sex-specific cues act in concert with ppk23 cell activation to select gender-appropriate behavior. Consistent with this, ppk23-Gal4, UAS-dTRPA1 and control males do not court without visual and olfactory inputs (Figure S4). This demonstrates the importance of other sensory cues in triggering courtship behavior. We hypothesize that these other cues assist the fly when all ppk23 cells are active, likely producing conflicting pheromone signals.
The courtship defects suggest that ppk23 and ppk29 may participate in detection of male inhibitory pheromones and female excitatory pheromones. To test the specificity of ppk23 cells in pheromone detection, behavioral approaches were used to examine the response of Δppk23 and Δppk29 males to cuticular hydrocarbons (Billeter et al., 2009; Wang et al., 2011). 7T and cVA are inhibitory compounds on males and 7,11-HD, and 7,11-ND are excitatory compounds on females (Ferveur, 2005; Jallon, 1984). 7P is abundant on males, with more complex roles in courtship (Ferveur, 2005; Jallon, 1984; Wang et al., 2011). Δppk23, Δppk29 and control males were paired with oe- males painted with 7T, 7P or cVA. Whereas controls reduced courtship to oe- males in the presence of these compounds as expected, responses of Δppk23 and Δppk29 were not significantly affected (Figure 5A). Courtship toward oe- males was due to an olfactory cues as loss of the antenna in control, Δppk23 or Δppk29 males abolished this behavior (Figure 5A). Additionally, controls increased courtship to oe- females painted with 7,11-HD, 7,11-ND or 7P, whereas Δppk23 and Δppk29 showed no change in behavior (Figure 5B). Thus, both Δppk23 and Δppk29 males behave as though they are blind to multiple pheromonal compounds.
Interestingly, both Δppk23 and Δppk29 showed decreased courtship to oe- females compared to control. This suggests that there are excitatory signals on oe- females that wild-type flies detect and mutants fail to recognize. Similarly, Δppk23 males appear to show increased courtship to oe- males, although this difference is not significant. This suggests that there are inhibitory signals on oe- males that wild-type flies detect and mutants fails to recognize. These signals may be residual pheromones due to incomplete ablation of oenocytes (Billeter et al., 2009) or non-oenocyte derived cues.
We monitored calcium increases by G-CaMP imaging to assess whether ppk23 cells directly sense pheromones. Single bristles were stimulated and fluoresecent changes of cells underneath the bristle were examined in response to mixes of five cuticular hydrocarbons (7P, 7T, cVA, 7,11-HD, 7,11-ND) dissolved in 10% ethanol. G-CaMP fluorescent increases were observed in male and female leg neurons to the pheromone mix but not to 10% EtOH alone (Figure 6A). Responses were dose-sensitive, with significant responses at 10 and 100 ng/μl. Additionally, bristles were stimulated with sucrose, salt and quinine (Figure 6B). Cells did not respond to sucrose or salt. 6/27 cells responded to quinine; the high variance led to statistically insignificant responses. The pheromone mix activated ppk23 cells from legs and proboscis for both sexes (Figure 6CD). These responses required ppk23 and ppk29, as Δppk23 and Δppk29 cells did not respond to the hydrocarbon mix. Moreover, re-introduction of ppk23 into Δppk23 rescued the G-CaMP response, arguing that loss of the response is due to loss of ppk23. The decreased sensitivity of proboscis neurons to the pheromone mix as compared to leg neurons suggests that high concentrations may be required to activate these cells and/or that other compounds may optimally activate the cells.
To test whether ppk23/ppk29 are sufficient to confer responses to pheromones, we misexpressed ppk23/ppk29 in ppk28 cells in a Δppk28 background. These cells are water-sensing gustatory neurons but show no water responses in the absence of ppk28 and serve as “empty” gustatory neurons (Cameron et al., 2010). Misexpression of ppk23/ppk29 failed to confer responses to the pheromone mix upon stimulation of leg or proboscis bristles (Figure 6E). The same ppk23/ ppk29 constructs rescued mutant behavior and G-CaMP responses in the Δppk23 background, arguing that the constructs are functional. These studies suggest that additional components are required for pheromone sensitivity, although expression levels, folding or localization may limit the ability to assess function.
We next tested the response of ppk23 leg neurons to individual compounds. Several chemosensory bristles on the leg are innervated by two ppk23-positive cells, providing the opportunity to monitor the response of both cells upon stimulation at the bristle tip. The two cells underneath a bristle showed remarkable specificity toward pheromones. One cell responded best to 7,11-HD and 7,11-ND and the other cell responded to 7P, 7T and cVA, with heterogeneity in the response to these compounds (Figure 7). Cells from males or females showed similar response profiles. Grouping one of the two cells under a bristle as “female-sensing” and the other as “male-sensing” based on maximal responses revealed a clear segregation of sex-specific responses (Figure 7BC). The G-CaMP imaging experiments argue that the ppk23 cell population recognizes both male and female pheromones but that individual cells are tuned to a few compounds. Notably, cells generally responded to compounds from males or females but not both, arguing for sex-specific responses. Thus, the ppk23 cell population likely represents the majority of contact chemoreceptors for pheromones on the legs involved in male-male and male-female recognition.
In this study, we identify two ion channels, ppk23 and ppk29, selectively expressed in uncharacterized contact chemosensory neurons and show that these genes and neurons are essential for inhibiting inappropriate courtship toward males and promoting courtship toward females. ppk23 cells respond to either male or female pheromones and Δppk23 cells do not respond. Several important findings emerge from this work: (1) Ppk ion channels are critical for pheromone detection in sensory cells; (2) pheromone detection by contact chemoreceptors is essential for early courtship steps; (3) There are dedicated cells for pheromone detection that are distinct from sugar, bitter or water cells; (4) both males and females have gender-selective cells: one population responds to hydrocarbons produced by males and a different population responds to hydrocarbons produced by females. This work provides insight into the detection of non-volatile pheromones.
Ion channels of the Deg/ENaC family have been implicated in the detection of salts, acids, water, mechanosensory stimuli and peptides (Mano and Driscoll, 1999). In Drosophila, there are approximately 30 members of this family (Adams et al., 1998; Liu et al., 2003a). The founding member, pickpocket (Ppk) is thought to sense noxious mechanosensory stimuli (Zhong et al., 2010), Ppk11 and Ppk19 mediate salt taste detection in larvae (Liu et al., 2003b) and Ppk28 mediates water taste detection (Cameron et al., 2010; Chen et al., 2010). However, the majority of this family remains to be characterized.
Here, we identify ppk23 and ppk29 as co-expressed in uncharacterized neurons on the proboscis and in fruM-positive chemosensory neurons on the leg. These two genes play critical roles in courtship behavior. Δppk23 males increase courtship toward males and both Δppk23 and Δppk29 males decrease courtship toward females. Behaviorally, Δppk23 and Δppk29 fail to respond to individual male and female hydrocarbons. Moreover, ppk23 cells respond to pheromone mixes whereas Δppk23 and Δppk29 cells do not. These studies argue that both ppk23 and ppk29 are essential for the recognition of both male and female pheromones.
The difference in behavior of Δppk23 and Δppk29 to males argues that the two genes have partially non-overlapping functions. The most parsimonious explanation is that Δppk29 males, unlike Δppk23 males, retain some ability to sense male inhibitory compounds. This suggests an underlying difference in expression or function of the two genes. The precise extent of co-expression is difficult to determine given the weak expression of the ppk29-Gal4 line.
Do Ppk23 and Ppk29 detect pheromones, transduce pheromone signals or more indirectly influence pheromone detection by setting the membrane potential? Our current studies do not address this, but their selective expression in a subpopulation of chemosensory neurons and the inability of the two genes to cross-rescue argues for a specific function. As ppk23 cells are heterogeneous in their response to individual hydrocarbons, it is unlikely that ppk23 alone provides response specificity toward pheromones. Moreover, misexpression attempts in which ppk23 and ppk29 were expressed in “empty” gustatory neurons did not confer responses to pheromones. Interpretations of the misexpression experiments are limited as expression levels, folding or localization may all impact function.
What components might be upstream of Ppk23/Ppk29 in pheromone detection? The two candidate pheromone receptors for which Gal4 lines have been generated, Gr32a and Gr68a, are not localized to ppk23 neurons. ppk25 has previously been implicated in male-female recognition but its expression has not been resolved (Ben-Shahar et al., 2010; Lin et al., 2005). One possibility is that response specificity could be achieved either by the heteromultimeric composition of Ppk channels or by accessory binding proteins such as CheB42a, previously implicated in pheromone detection (Park et al., 2006). Alternatively, unidentified molecules may provide specificity.
Although the precise role of Ppks in pheromone detection remains to be examined, the demonstration that ppk23 and ppk29 are expressed in sexually dimorphic, fruM-positive leg neurons and are essential for responses to male and female pheromones provides a strong foundation for future studies.
Courtship behavior is comprised of a series of behavioral subprograms that are executed in response to visual, auditory, olfactory and contact chemosensory cues (Greenspan and Ferveur, 2000; Hall, 1994). Teasing out the role of different sensory cues in driving courtship behavior has been limited by the ability to selectively target different classes of sensory neurons. Here, we identify a large population of pheromone-sensing neurons on the legs that co-express ppk23 and fruM, allowing us to selectively manipulate the contact chemosensory component of the fruitless circuit.
Contact chemoreceptors have largely been proposed to function in later stages of courtship during the foreleg tapping and proboscis contact steps, as these involve chemosensory organs (Bray and Amrein, 2003; Ferveur, 2005; Greenspan, 1995). In this study, we find that flies lacking ppk23 or ppk23 cell activity show behavioral defects at early steps in the courtship process: they court males and delay or fail to initiate courtship of females. Recent studies on males with only a few residual chemosensory bristles found that they increased courtship to males and decreased courtship to females under dark conditions (Krstic et al., 2009), in line with our results. The defects in courtship initiation do not preclude a role in later steps such as foreleg tapping or proboscis contact but complicate evaluation. The early defects in sex discrimination and initiation of courtship argue that contact chemoreceptors participate in male-female recognition prior to physical contact with other flies.
How does recognition occur at a distance? Pheromones that activate ppk23 cells could potentially be volatile or sprayed by wings and deposited by legs, leaving lipid trails. As cuticular hydrocarbons have low volatility (Antony and Jallon, 1982), it is more likely that ppk23-positive cells sense deposited lipids. An interesting possibility is that transient pheromone trails may guide a male to a female, similar to pheromone trails that recruit ants to a food source.
In mammals, pheromones are detected by the vomeronasal organ and the olfactory epithelium, allowing animals to respond to diverse chemical cues signifying potential predators or mates (reviewed in Stowers and Marton, 2005). In Drosophila, subsets of olfactory and gustatory neurons are responsible for pheromone detection. In the olfactory system, three of ~50 glomeruli are fruM-positive (Manoli et al., 2005; Stockinger et al., 2005), suggesting that they comprise a pheromone-specific subsystem. In the gustatory system, fruM and ppk23 are co-expressed in sensory neurons that are essential to promote courtship toward females and prevent inappropriate courtship toward males. These cells are distinct from those expressing markers for sugar, bitter or water-sensing cells, indicating that they form a pheromone-specific subsystem of the gustatory system. In the legs, these cells selectively respond to male or female pheromones, suggesting that there are sex-specific sensory cells. This argues that dedicated neurons for pheromone detection act as labeled lines in males to inhibit male-male courtship and initiate male-female courtship. The role of these sex-specific cells in females will be interesting to explore. Thus, the contact chemosensory system contains different cell populations tuned to sugars, toxins, water or other flies, extracting the essence for life from the environment.
P element-mediated transformations of w1118 were performed using standard techniques (Genetic Services Inc.) Lines used: UAS-TNT (Sweeney et al., 1995); UAS-mCD8GFP (Lee and Luo, 1999); UAS-CD2GFP and lexAop-CD2GFP (Lai and Lee, 2006); UAS-TRPA1 (Hamada et al., 2008); Gr64f-Gal4 (Dahanukar et al., 2007), Gr66a-Ires-GFP (Wang et al., 2004), fruP1-LexA (Mellert et al., 2010), ppk28-Gal4 (Cameron et al., 2010), Gr5a-Gal4 (Chyb et al., 2003), Gr68a-Gal4 (Bray and Amrein, 2003), Gr66a-Gal4 and Gr32a-Gal4 (Scott et al., 2001).
Transgenes were generated from PCR amplification and cloning into pCaSpeR-AUG-Gal4 (Vosshall et al., 2000), pUAST, pLOT or LexA-pCaSpeR. Constructs were verified by sequencing. See Supplemental Experimental Procedures for details.
Double label in situ hybridization experiments were performed as described (Fishilevich and Vosshall, 2005).
Staining was performed as described (Wang et al., 2004).
oe- male and female target flies were generated as previously described (Billeter et al., 2009). oe- females were pierced through the head to prevent copulation (Gailey et al., 1984). 0.2μg of pheromone dissolved in hexane (7,11-HD, 7,11-ND, 7T and 7P) or ethanol (cVA) was applied to filter paper and evaporated. 8-16 oe- flies were gently vortexed with the filter paper twice for 20 seconds, roamed for ~30 minutes, then were transferred to a fresh vial 24 hrs prior to courtship assays.
Flies expressing UAS-GCaMP3 and ppk23-Gal4 were immobilized and forelegs tethered using parafilm. Pheromone mixes of 7,11-HD, 7,11-ND, 7T, 7P and cVA in 10% EtOH were applied to single bristles for 30s. 1M sucrose, 1mM quinine, and 1M NaCl were delivered followed by 10% ethanol alone. Responses were recorded on a 3i Spinning Disk fixed-stage confocal. The maximum change in fluorescence (ΔF/F) equals the peak intensity change divided by the average intensity four seconds prior to stimulation. See Supplemental Experimental Procedures for additional details.
The Scott lab provided helpful comments and discussion. Dr. Gautam Agarwal provided MATLAB advice and Dr. Seth Ament performed the ANOVA analysis. Dr. Charles Zuker provided UAS-GCaMP3 flies. RT is supported by a predoctoral NSF fellowship; PC was supported by a predoctoral NRSA fellowship, F31DC009389. This research was supported by a grant from the NIDCD 1R01DC009470. KS is an Early Career Scientist of the Howard Hughes Medical Institute.
Author contributionsR.T. performed most experiments, including Gal4 expression studies, calcium imaging and behavioral assays and co-wrote the paper. P.C. generated Δppk23 and Δppk29, Gal4 and UAS transgenic animals, made isogenic lines and carried out expression studies. A.G. assisted R.T. with behavior. L.D. generated ppk28-LexA and contributed to in situ experiments. K.S. supervised the project, co-wrote the manuscript and performed expression studies.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.