Adjusting reproductive behavior to nutrient availability is a common feature in many animals. Females of many animal species decrease their sexual receptivity to male courtship in response to food deprivation (Kauffman and Rissman, 2004; Pierce et al., 2007; Franssen et al., 2008; Lebreton et al., 2015). In the fruit fly Drosophila melanogaster, feeding, and mating are strongly interconnected. Mating modifies food preference in females (Carvalho et al., 2006; Walker et al., 2015) while feeding regulates their mating behavior (Lebreton et al., 2015, 2016). The molecular mechanisms regulating feeding behavior after mating are fairly well described and involve the transfer of a male component during copulation called Sex Peptide (Carvalho et al., 2006; Ribeiro and Dickson, 2010; Walker et al., 2015). On the other hand, the mechanisms by which feeding and starvation regulate sexual receptivity remain unknown.

One potential system that mechanistically links these two behaviors is insulin signaling as it not only regulates feeding (Broughton et al., 2005; Slaidina et al., 2009; Lebreton et al., 2014) but also various aspects of reproduction such as female attractiveness, egg-laying and remating rate (Yang et al., 2008; Wigby et al., 2011; Kuo et al., 2012). Although, insulin signaling is not necessary for virgin females to be receptive when fed (Wigby et al., 2011; Sakai et al., 2014; Lebreton et al., 2015; Watanabe and Sakai, 2015), so far no study tested the role of this signaling pathway on the sexual behavior of females undergoing a period of starvation. Considering the conserved effect of starvation on sexual receptivity on the one hand, and the fact that insulin signaling both responds to nutrition and regulates reproductive behavior on the other hand, we expected insulin signaling to modulate female sexual receptivity during starvation.

Eight insulin-like peptides (DILP1-8) have been characterized in D. melanogaster while only one insulin receptor is known (InR; Nässel et al., 2013, 2015). DILP2, 3, and 5 are produced in specific cells in the pars intercerebralis called insulin-producing cells (IPCs). In addition, DILP5 is expressed in follicle cells of ovaries and principal cells in renal tubules (Ikeya et al., 2002). DILP6 is, on the other hand, produced by adipose cells (Slaidina et al., 2009). In contrast to other DILPs, DILP7, and DILP8 have been suggested to be more related to relaxin peptides than insulin peptides (Yang et al., 2008; Grönke et al., 2010; Garelli et al., 2015). Whereas, it is unknown whether DILP7 acts through the same InR, it has recently been shown that DILP8 acts via a specific relaxin receptor (Garelli et al., 2015). In addition, two DILPs are expressed almost exclusively during larval stages: DILP1 and DILP4 (Nässel et al., 2015). These different DILPs seem to interact to regulate the fly behavior and metabolism (Grönke et al., 2010; Kannan and Fridell, 2013; Nässel et al., 2013).

DILPs regulate the activity of neuronal circuitries to match behaviors with nutritional status (Wu et al., 2005a,b). Several neuronal networks have been shown to be involved in female sexual receptivity. For instance, neurons expressing the fruitless (fru) gene are necessary for females to be sexually receptive (Demir and Dickson, 2005; Kvitsiani and Dickson, 2006). Similarly, a subset of neurons expressing doublesex (dsx) regulates receptivity, independently of fru (Zhou et al., 2014). Both fru and dsx encode for a transcription factor that are spliced differently in males and females and account for sexually dimorphic traits (Siwicki and Kravitz, 2009). fru is expressed in some pheromone-sensing olfactory sensory neurons (OSNs) expressing the odorant receptors Or67d and Or47b (Stockinger et al., 2005). Or67d detects the male aphrodisiac pheromone cis-vaccenyl acetate (cVA) while Or47b responds to methyl laurate, a compound produced by flies of both sexes (Dweck et al., 2015). Both OSNs expressing Or67d and Or47b have been shown to modulate female receptivity (Kurtovic et al., 2007; Sakurai et al., 2013). In addition to these neurons, an early study found a group of cells in the dorsal anterior brain to be necessary and sufficient to induce receptivity (Tompkins and Hall, 1983), though these cells have never been precisely characterized. Interestingly, these cells seem to be different from those required to perform courtship in males (Tompkins and Hall, 1983), suggesting that they are probably not fru-positive. In young virgin females, acquisition of sexual receptivity is paralleled by the growth of the ovaries and the corpora allata, endocrine glands producing the juvenile hormone (JH, Manning, 1967). Both JH and the corpora allata have been shown to be involved in this switch in female receptivity (Manning, 1967; Ringo et al., 1991). Interestingly, the corpora allata activity is modulated by the insulin pathway (Tu et al., 2005; Belgacem and Martin, 2007). However, whether or not insulin acts on these structures to regulate the female sexual receptivity is unknown.

In conclusion, despite the fact that insulin is known to both regulate reproductive behavior and respond to food intake/deprivation, its effect in the starvation-induced reduction in female receptivity has not yet been established. We hypothesize that, during starvation, insulin signaling acts on specific neuronal networks to reduce sexual receptivity in order to match nutrition and reproduction. In the present study, we first carefully analyzed the effect of starvation in female mating behavior in Drosophila. We then tested the role of insulin by analyzing the behavior of females deficient for seven different DILPs. Finally, in an attempt to identify some structures on which insulin signaling acts to regulate sexual behavior, we knocked down InR in various brain parts known to be involved in mating behavior or behavioral modulation.

Drosophila females need nutrients to produce eggs and a nutrient rich substrate to lay their eggs (Drummond-Barbosa and Spradling, 2001; Becher et al., 2012). When food is scarce it would therefore be beneficial for flies to decrease their sexual behavior and to focus on food searching instead. On the other hand, female flies can store sperm and use it several days later when conditions are suitable (Qazi et al., 2003). It could therefore be optimal for females to remain receptive for short periods of food deprivation. Several insulin peptides produced in specific spatiotemporal patterns acting through one single receptor enables a fine-scale regulation of behaviors in response to changes in physiology. The expression of the different dilps is differentially affected by food quality or food deprivation (Bai et al., 2012; Whitaker et al., 2014; Post and Tatar, 2016). For instance, both starvation and dietary restriction reduce the expression of dilp5 but increase the expression of dilp6, while the expression of dilp2 is not affected by either condition (Bai et al., 2012; Whitaker et al., 2014). Our results suggest that DILP5 might be involved in the decrease of receptivity during non-feeding stages. Indeed, dilp5 mutant females were less affected by starvation than other dilp mutants. The effect of the lack of DILP5 was no longer observed in the simultaneous absence of DILP2 and DILP3. Although, we cannot completely rule out background mutation effects, this suggests that DILP5 might interact with other DILPs to finely tune female sexual receptivity.

Insulin is known to act on the olfactory system to modulate odor sensitivity after feeding (Root et al., 2011). Moreover, normal InR expression in Or67d-expressing (Fruitless-positive) OSNs is necessary for fed females to be attracted to a blend of food odors and cVA (Lebreton et al., 2015), a pheromone promoting sexual receptivity (Kurtovic et al., 2007). Our results suggest that insulin signaling in Fruitless-positive neurons, and more specifically in Or67d OSNs may decrease sexual receptivity during starvation.

Fruitless-positive cells other than pheromone-sensing neurons can also be involved. We found different Fruitless-positive cells in the protocerebrum that strongly express InR. First of all, a large number of Kenyon cells in the calyx of the mushroom bodies express both Fruitless and the insulin receptor. Additionally, we found one pair of neurons with somata located in the anterior dorso-lateral protocerebrum. We could not trace any processes from these somata and do thus not know what neuropils they innervate. However, the fact that InR immunostaining was observed in Fruitless neurons, most of which were Kenyon cells, corroborate our behavioral results. Indeed, the sexual receptivity of females in which insulin signaling was knocked down in the mushroom bodies was not affected by starvation. Interestingly, the mushroom bodies are not required for virgin females to be receptive (Neckameyer, 1998), suggesting that these structures may regulate the activity of neuronal networks inducing sexual receptivity. However, this result must be take with caution, given the fact that the Gal4 line we used to target the mushroom bodies also drives expression to some extent in other brain tissues (Aso et al., 2009). Further experiments will be necessary to confirm that the mushroom bodies are indeed responsible for this effect.

Insulin signaling not only modulates neuronal activity in adults but also shapes neuronal networks during development (Song et al., 2003). The effects we observed in our study may therefore be the consequence of a developmental defect of specific neuronal circuitry rather than a direct effect of insulin on these neurons during starvation. However, Fruitless-positive neurons being required for females to be receptive (Kvitsiani and Dickson, 2006), we would expect fed females to be unreceptive if the disruption of insulin signaling had altered the connectivity of these neurons during development, which was not the case. This suggests that insulin acts on these neurons during adult stage to modulate sexual receptivity. This is different for the mushroom bodies, which are not necessary for females to be receptive (Neckameyer, 1998). Knocking down InR specifically during development or specifically in adults will be necessary to disentangle these two possible modes of action of insulin.

In contrast with Fruitless neurons and the mushroom bodies, we did not observe any effect of the corpora allata in the insulin-dependent control of sexual receptivity, whereas these structures have been linked to the development of receptivity in virgin females (Manning, 1966, 1967). This result should however be taken with caution, considering the behavioral variability displayed by the different transgenic lines, which would have prevented us from observing subtle changes. Nonetheless, our results suggest that the structures that generate behaviors (such as the corpora allata) and those modulating these behaviors (for example the mushroom bodies) can be different and the underlying mechanisms uncoupled.

Taken together, Drosophila flies adjust their sexual behavior to match their nutritional state. Together with other hormonal pathways (Lebreton et al., 2016), insulin regulates some aspects of sexual activity (Wigby et al., 2011; Kuo et al., 2012; Sakai et al., 2014; Watanabe and Sakai, 2015), both after food intake (Lebreton et al., 2015) and after a period of starvation. Our results suggest that specific insulin peptides regulate female receptivity, possibly by acting on pheromone perception at the periphery or directly in the central nervous system. Indeed, the mushroom bodies probably play a major role in the insulin-dependent effect of starvation on female sexual receptivity. The next step will be to untangle the specific neuronal circuitry involved.

All the experiments described in the manuscript were performed with laboratory-reared insects. No recommendations from an ethical committee were required.

SL and PW conceived the project. SL designed, performed and analyzed the behavioral experiments. MC performed the immunostaining and the acquisition of confocal microscopy images. SL wrote the first draft of the manuscript. All authors contributed to the final version of the manuscript.