Chemoreception is defined as the physiological response to a chemical stimulus. Depending on the spatial scale, a classical division exists between olfaction and taste chemoreception. Olfaction is involved in the detection of volatile molecules coming from long distances, while taste is a contact sense that allows detection of molecules at a short distance. Highly volatile hydrophobic molecules can be rapidly transported by air and, once they reach the living organism, activate olfactory receptors. On the contrary, hydrophilic molecules are less volatile and they most likely activate taste receptors when presented at a short distance. This definition might not be suitable for aquatic environments where solubility instead of volatility is the determinant factor for long-distance transport of molecules (Mollo et al., 2014).

One of the favorite model organisms for the study of olfaction and taste perception is the fly Drosophila melanogaster. In the last two decades, and due to its amazing repertoire of genetic tools, Drosophila has been at the leading front in the discovery of chemoreceptors and chemoreceptive neuronal pathways that account for the behavioral responses toward ecologically relevant chemicals. Even more, the extensive work done in Drosophila helps us to better understand the chemoreceptive systems of insects relevant for human's health, such as the mosquitos Anopheles gambiae and Aedes aegypti, dangerously efficient vectors of malaria and Dengue hemorrhagic fever.

Flies are able to perceive relevant chemical cues present in their food, in host plant, and those produced by conspecific. Attractive odors and tastants in the food can induce feeding, while toxic compounds present in food or produced by host plants trigger avoidance. Before activating the oviposition motor program, female flies carefully analyze the chemical composition of the substrate. Also, conspecific chemical cues are essential for aggregation, aggression, and courtship. All of these effects depend on proper detection of chemical cues at the level of olfactory and gustatory receptors present in dedicated structures.

Here we will review the extensive recent research focused on detection of ecologically relevant chemicals in flies and its behavioral consequences. Firstly, we will very briefly outline the olfactory and gustatory system of fly adults and larvae, giving more emphasis to the description of the different families of chemoreceptors. Secondly, we will present several examples of chemoreceptors involved in the detection of chemical signals that impact on behaviors relevant for fly survival, such as feeding, toxic compounds avoidance, and oviposition site and sexual partner selection. Finally, we will review and discuss the ecological relevance of specific chemicals and chemoreceptors in the context of the particular requirements of two stages of Drosophila life cycle, the larva and the adult fly.

The olfactory organs of the adult fly are located on the third antennal segment (also known as funiculus) and on the maxillary palps, where three different types of sensilla, the basiconic, the trichoid, and the coeloconic, harbor the olfactory sensory neurons (OSNs) (Figure 1A). In the OSNs, olfactory receptors directly contact their specific ligands. From the periphery, OSNs send axonal projections to specific glomeruli in the antennal lobe, the first olfactory relay center in the brain. In the antennal lobe, the odor signals are processed by local interneurons and projection neurons. Local interneurons connect different glomeruli mainly triggering later inhibition (Silbering and Galizia, 2007), and projection neurons transmit the olfactory trace to higher centers in the lateral horn and mushroom bodies (reviewed in Stocker, 1994; Laissue and Vosshall, 2008). Careful anatomical description of the olfactory system allowed building a near complete map of OSN's connectivity. OSNs expressing the same olfactory receptor project into the same unique glomerulus in the antennal lobe (Couto et al., 2005; Fishilevich and Vosshall, 2005; Goldman et al., 2005). In addition, OSNs harbored in different type of sensilla project into distinct regions of the antennal lobe, highlighting the level of topographic organization of the olfactory system (Couto et al., 2005).

In contraposition to olfactory organs, taste organs are widely distributed in the adult body, with external gustatory centers on the proboscis's labellum, legs, wings, and female genitalia, and internal taste structures in the pharynx (Figure 1A). The labellum is the principal taste organ of the adult fly and it harbors two major types of sensilla, the taste bristles and taste pegs, wherein gustatory receptors expressed in gustatory receptor neurons (GRNs) directly detect tastant. In the pharynx, the labral sense organ (LSO), the ventral and dorsal cibarial sense organs (VCSO and DCSO), and a ventral and a dorsal row of “fish-trap” bristles allow taste detection after ingestion (reviewed in Stocker, 1994; Montell, 2009). From the taste organs located in the mouth parts, proboscis, and legs, GRNs transmit directly or through activation of interneurons the gustatory information to the subesophageal ganglion (SOG), a dedicated taste center in the brain (Wang et al., 2004). Some taste-like sensilla are also present on the genitalia and on the wing margin, but their precise role is still under investigation (Boll and Noll, 2002; Yanagawa et al., 2014). In the SOG, axonal projections coming from different peripheral tissues are segregated even if they contain the same receptor (Wang et al., 2004). Even more, bitter and sugar sensing neurons clearly segregate in the SOG, demonstrating that the first gustatory relay center displays a topographic and functional organization (Thorne et al., 2004; Wang et al., 2004).

The external chemosensory organs of the larvae are all located in the cephalic lobe with the exception of some putative taste organs in thoracic and abdominal segments (Dambly-chaudière and Ghysen, 1986; Scott et al., 2001) (Figure 1B). Larval olfactory structures are located in the dorsal organ, while external gustatory structures are mainly distributed between the terminal and ventral organs, and to a lesser extent, the dorsal organ. In addition, three internal pharyngeal organs, the dorsal, ventral, and posterior pharyngeal sense organs (DPS, VPS, and PPS, respectively) include mainly taste sensilla (Stocker, 2008). Similar to the case of the adult gustatory system, the larval SOG shows a certain topographic and functional organization although in the larvae there is no complete segregation between external and internal GRNs axonal projections (Colomb et al., 2007; Kwon et al., 2011). The dorsal organ is composed of the central “dome” that harbors the dendrites of the 21 larval OSNs and a few putative taste sensilla (Gerber and Stocker, 2007; Stocker, 2008). This small number of OSNs contrasts with the around 1300 OSNs that are present in adults. Despite these numeric differences, the adult and larval olfactory pathways share the same design (Stocker, 2009). Nonetheless, the larval olfactory system is not just a reduced version of the adult system because some olfactory receptors are only expressed in the larval stage (Stocker, 2008). Even more, the larval local interneurons in the antennal lobe do not keep resemblance with their adult counterparts and in the larva they tightly connect gustatory and olfactory centers (Thum et al., 2011).

Regarding the chemoreceptors, in flies more than 150 receptors are distributed in three principal families, the gustatory receptors (GRs), the odorant receptors (ORs), and the ionotropic receptors (IRs). In addition, some members of the TRP family and degenerin/epithelial sodium channel/pickpocket (ppk) channels as well as the insect orphan G-protein-coupled DmXR are either bona fide chemoreceptors or they are tightly involved in chemoreception in flies (Table 1).

In mammals, most chemoreceptors are classic seven-transmembrane G-protein-coupled receptors (GPCRs) (Chandrashekar et al., 2006; Spehr and Munger, 2009). In insects, GRs and ORs are also seven transmembrane domain proteins with a no amino acid sequence homology compared to mammalian ORs and GRs (Vosshall et al., 1999; Clyne, 2000; Scott et al., 2001); however, insect ORs have a topology opposite to mammalian GPCRs, with cytoplasmic N-termini and extracellular C-termini (Benton et al., 2006). A phylogenetic analysis indicated that insect OR family is an expanded lineage within the ancestral insect GR family (Robertson et al., 2003), highlighting a common evolutionary origin. In addition, insect IRs are ligand-gated ion channels involved in chemoreception and they belong to the superfamily of ionotropic glutamate receptors (Benton et al., 2009). Interestingly, a subgroup of IRs are expressed in the antenna and, contrary to what is seen in most of the ORs, they are highly conserved within insects, both in sequence and expression pattern, suggesting that antennal IRs might represent the ancestral olfactory receptor family in insects (Croset et al., 2010). Below, we will describe in more detail some characteristics of these families of chemoreceptors, separating between those involved in olfaction and the ones dedicated to taste perception.

Genomic comparative studies highlight the rapid evolution of chemoreceptors both in number and identity (Robertson et al., 2003; Croset et al., 2010). This feature led to the hypothesis that changes at the level of chemosensory systems contribute to the diversification of behaviors (Cande et al., 2013). Evidence in favor of this hypothesis comes mainly from comparative studies of closely related Drosophila species with different behaviors, as it is the case of D. melanogaster and D. sechellia. D. melanogaster is a generalist species that can survive in several fruit substrates, while D. sechellia is a host-plant specialist. Interestingly, D. melanogaster has a complex olfactory system that allows detection of hundreds of fruit-derived odors; D. sechellia, on the contrary, has lost many chemoreceptors that are not relevant for its very specialized ecology (Stensmyr et al., 2003; Cande et al., 2013). Although a very provocative hypothesis, it is difficult to prove that mutations in chemoreceptor gene loci are important driving forces behind behavioral change. Nonetheless, there is no doubt that the current chemoreceptors allow detection of ecologically relevant chemicals present in the fruitfly's environment. In this section we will discuss the ecological relevance of specific chemoreceptors related to behaviors such as food searching and the analysis of its composition, avoidance of toxic or bitter compounds, oviposition site selection, and the search for a sexual partner (Table 2).

OSNs expressing either OR67d, OR47b, or IR84 are the only three OSNs that express a sex-specific transcript of the gene fruitless (FruM) (Stockinger et al., 2005; Grosjean et al., 2011), and this suggests their involvement in sex-specific behaviors such as courtship. These OSNs project respectively to DA1, VA1_v, and VL2a glomeruli, which are significantly larger in males. From these glomeruli, Fru⁺ projection neurons connect with the lateral horn (Kondoh et al., 2003; Stockinger et al., 2005; Grosjean et al., 2011). Regarding taste structures, males harbor more gustatory receptors in their legs compared to females, and sex determination factors like fruitless and doublesex are responsible for sexually dimorphic axonal pattern in these sensory neurons (Possidente and Murphey, 1989; Mellert et al., 2012; Yamamoto and Koganezawa, 2013; Koh et al., 2014) (Figure 2A). Fru⁺ gustatory neurons are present mainly in the dorsal labellum and foreleg tarsi (Stockinger et al., 2005). With the exception of IR84a that has been confirmed to respond to food odors (Grosjean et al., 2011), the rest of these sexually dimorphic receptors are annotated or predicted to detect sexual pheromones.

A sexual pheromone is defined as a chemical signal produced by the organism involved in the control of sexual behaviors. In Drosophila, the principal known sexual pheromones are the volatile cis-vaccenyl acetate (cVA) and the cuticular hydrocarbons 7-tricosene (7-T), 7,11-heptacosadiene (7,11-HD), and 7,11 nonacosadiene (7,11-ND). Briefly, cVA and 7-T are produced by males and they act as anti-aphrodisiac for other males (although cVA has several additional roles; see below) while 7,11-HD and 7,11-ND are female pheromones that promote courtship (Fernández and Kravitz, 2013; Gomez-Diaz and Benton, 2013). Anyhow, more recent and highly sensible methods of detection have demonstrated that all of these four pheromones are present in virgin socially naïve individuals of both sexes but in different quantities (Yew et al., 2009). cVA produced and stored in the ejaculatory bulb of males (Butterworth, 1969; Brieger and Butterworth, 1970) is transferred to females during copulation (Butterworth, 1969; Ejima et al., 2007). Together with cVA, an acetylated hydrocarbon named CH503 is also transferred to females during copulation leading to a prolonged inhibition of male courtship acting through an unknown sensory receptor (Yew et al., 2009) (Figures 2B–D and Table 2).

In males, cVA acts as an anti-aphrodisiac that reduces courtship toward mated females or other males. cVA also modulates male-male aggression while increasing receptivity in females (Jallon, 1984; Ejima et al., 2007; Kurtovic et al., 2007; Wang and Anderson, 2010; Liu et al., 2011). At long ranges, cVA is described to function as an aggregation factor for males and females (Bartelt et al., 1985). Electrophysiological studies demonstrated that cVA is sensed through OSNs expressing OR67d and to a lesser extent, OR65a (Ha and Smith, 2006; Kurtovic et al., 2007; Van der Goes van Naters and Carlson, 2007). Although some discrepancies have been observed in different studies, the role of cVA on sexual behaviors seems to be mediated by OR67d activation (Ejima et al., 2007; Kurtovic et al., 2007). While acute promotion of aggression depends on OR67d, chronic exposure to cVA reduces aggression through OR65a activation (Wang and Anderson, 2010; Liu et al., 2011). Interestingly, both females and males express OR67d and OR65a, and these receptors respond equally to cVA in both sexes (Kurtovic et al., 2007; Van der Goes van Naters and Carlson, 2007). However, the neuronal circuit underlying OR67d is sexually dimorphic and, consequently, different neuronal cluster are activated in males and females (Datta et al., 2008; Ruta et al., 2010; Kohl et al., 2013). This sexual dimorphism in the neuronal circuit downstream of OR67d could be responsible for the different behaviors elicited by cVA in both sexes. In addition to the activation of OR67d and OR65a by cVA, uncharacterized fly odors activate OR47b and OR88a both in males and females, suggesting the presence of other volatile pheromones (Van der Goes van Naters and Carlson, 2007) (Figures 2B–D and Table 2).

7-T is a male hydrocarbon that inhibits courtship in other males (Antony and Jallon, 1982; Lacaille et al., 2007) and promotes male-male aggression by acting in the same neuronal pathway as cVA (Wang et al., 2011). Several receptors have been proposed for 7-T, in particular GR32a and GR33a (Miyamoto and Amrein, 2008; Moon et al., 2009). Males lacking GR32a display high courtship toward males and mated females, suggesting that GR32a could sense an anti-aphrodisiac signal produced by males and transferred to females during courtship (Miyamoto and Amrein, 2008) (Figures 2B–D). Moreover, GR32a prevents males to court with individuals from other species, contributing to the isolation barrier within the Drosophila genus (Fan et al., 2013) (Figure 2E). GR32a is present in the labellum and in the leg, but only in the leg GR32⁺ GRNs are surrounded by cells expressing OBP57, an odorant-binding protein implicated in the carrying of pheromones (Koganezawa et al., 2010). Although no sexual dimorphism is observed in GR32a sensory neurons, they seem to contact Fru⁺ neurons in the SOG that display sexually dimorphic dendritic arbors (Koganezawa et al., 2010; Fan et al., 2013) (Figure 2A). It would be interesting to study if these differences in the dendritic arbor determine different postsynaptic neuronal clusters that could activate different motor programs in males and females in response to GR32a activation. At the same time, GR33a, a key receptor in the detection of several aversive compounds, is also required to inhibit male-male courtship (Moon et al., 2009) and it is essential for the male preference for younger virgin females (Hu et al., 2015). GR33a and GR32a seem to be expressed in the same GRNs in the leg, suggesting that they might be part of the same heterodimeric receptor (Moon et al., 2009). In addition, 7-T appears to inhibit a male-male courtship-promoting-signaling pathway that is OR47b dependent (Wang et al., 2011) (Figures 2B–E and Table 2).

The production of 7,11 HD in females serves as an aphrodisiac for males of the same species (Antony and Jallon, 1982; Antony et al., 1985) and acts as a barrier to prevent interspecific courtship (Billeter et al., 2009). Males sense female pheromones, probably 7,11-HD, through GR68a expressed in their forelegs (Figures 2B,F and Table 2). GR68a is exclusively expressed in male forelegs (Figure 2A), and its expression depends on the sex determination factor doublesex (Bray and Amrein, 2003). In addition, GR39a may also be involved in female pheromones perception since male mutants for GR39a display reduced courtship toward wild-type females (Watanabe et al., 2011).

In the last few years, several studies have demonstrated the relevance of ppk channels, notably ppk23, ppk25, and ppk29, in sexual behavior. These 3 channels are expressed in Fru⁺ gustatory neurons of both sexes although males have around double the amount of ppk⁺ cells in the leg compared to females (Liu et al., 2012; Lu et al., 2012; Vijayan et al., 2014) (Figure 2A). Different subpopulations of ppk23⁺-FruM⁺ neurons in the leg respond to male and female pheromones, and both ppk23 and ppk29 are required for the pheromone-evoked effects on courtship (Lu et al., 2012; Thistle et al., 2012; Toda et al., 2012). Those responding to female pheromones express also ppk25, and this channel is necessary for the 7,11-HD effects of promoting courtship toward virgin females as well as for the stimulation of courtship by pheromones present on immature males and for normal female receptivity (Vijayan et al., 2014) (Figures 2B–D and Table 2). ppk25 is also expressed in olfactory neurons, but this expression is not relevant for courtship control (Starostina et al., 2012). Interestingly, similar responses at the level of ppk⁺ cells activation in response to male and female compounds were observed in both sexes (Thistle et al., 2012; Vijayan et al., 2014). This suggests, once again, that sexual dimorphism downstream of the activation of receptor neurons might be responsible for the different behaviors triggered in male and females in response to sexual pheromones. In effect, at least ppk23⁺ neurons display sexually dimorphic axonal projections, although the physiological consequences of this sexual dimorphism have not been studied (Lu et al., 2012; Thistle et al., 2012) (Figure 2A).

The studies presented here demonstrate the relevance of ppk23, ppk25, and ppk29 channels in sexual behavior, but a clear demonstration of their role as chemoreceptors is still lacking. Attempts to prove the direct requirement of ppk23 and ppk29 as pheromone receptors failed, suggesting that additional subunits may be required (Thistle et al., 2012). Consistent with this idea is the fact that ppk29 (also known as NOPE) forms a complex with ppk25 (Liu et al., 2012). Alternatively, ppk channels could be playing a fundamental role on pheromone-evoked responses, not as direct receptors but as a unique cellular component of gustatory neurons expressing a yet unknown chemoreceptor (Pikielny, 2012). Strikingly, a recent study demonstrated that ppk23 and ppk29 are also essential for the detection of a novel aggregation pheromone in D. melanogaster larvae, the (Z)-5 and (Z)-7-tetradecenoic acid (Mast et al., 2014). In this study the authors clearly demonstrate the relevance of ppk23 and ppk29 in the detection of the aggregation pheromone but, again, no direct proof of their role as chemoreceptors has been provided. Taking into account that ppk23-ppk29 are essential for the detection of signals of very different structure (long-chain fatty acids in the case of larval aggregation and hydrocarbons in the case of sexual behavior), it seems more reasonable that they don't act as direct chemoreceptors. Anyhow, more suitable experiments like analysis of response to pheromones using in vitro or in vivo ectopic expression of these proteins will help to clarify this matter.

In addition to ORs, GRs, and ppk channels, IRs play a relevant role in the control of sexual behaviors in Drosophila. IR52a, IR52c, and IR52d are present in the foreleg of both sexes although they are expressed in more cells in males (Figure 2A). IR52c and IR52d show complete, or nearly complete, coexpression in the foreleg suggesting that they are part of the same complex. IR52c⁺ neurons are activated by female compounds and form putative synapses with Fru⁺ neurons in the prothoracic ganglia. Interestingly, ectopic activation of IR52c⁺ neurons increase courtship while mutants lacking IR52c and IR52d display reduced courtship behavior and higher latency to copulate, suggesting a possible role on sexual pheromone detection (Koh et al., 2014) (Figure 2B and Table 2).

Last but not least, food odors, notably phenylacetaldehyde and phenylacetic acid, can also promote male courtship through IR84a-FruM⁺ OSNs. The VL2a FruM⁺ glomerulus is activated downstream of IR84a⁺ OSNs, and from this glomerulus, projection neurons send olfactory information to a pheromone-processing region of the lateral horn (Grosjean et al., 2011) (Figures 2B,D and Table 2). In this regard, we now understand that not only sexual pheromones but also compounds present in the environment, at least in the fly food, directly modulate sexual behavior in Drosophila, highlighting the impact of external cues in a key behavior for species survival.

When we compare a Drosophila larva with an adult, differences become much more obvious than similarities. Although sharing the same genome and developmental program, the larval and adult stages of holometabolous insects contrast strikingly in regards to general anatomy, behaviors displayed, and lifestyles or niches occupied. In the nervous system, the differences between larvae and adults rise as a consequence of the extensive apoptosis and neuronal remodeling occurring during the metamorphosis (Truman, 1990). The case of chemoreceptive structures is not an exception. The external taste organs of the larva, the terminal and ventral organs, undergo apoptosis during the metamorphosis and are then completely replaced by adult structures. While the main olfactory organ, the dorsal organ, does not disappear during metamorphosis, the olfactory system undergoes critical neuronal changes, e.g., neuronal migration, proliferation, and development of progenitor cells, dendritic pruning and extension, and axonal remodeling, among other processes (Gerber and Stocker, 2007; Rodrigues and Hummel, 2008). In this regard, taking into account the extensive remodeling of the chemoreceptive structures along the life cycle of the fly, it does not come as a surprise that larvae and adults can trigger very different behaviors in response to the same stimuli or that they may even use different receptors to detect the same compounds. Nonetheless, one can still wonder why invest so much energy in developing two strikingly different chemoreception systems in such a short time. Are these differences a consequence of developmental constrains or do they reflect an adaptation to different niches occupied along the life? Below, we will present the cases of sugar, carboxylic acid, and salt detection in adult and larvae as interesting examples of stage-specific chemoreceptors involved in the responses to ecologically relevant compounds. We will further discuss how these differences in the system and the behavior could be interpreted in the context of stage-specific needs.

Sugar detection is an interesting example of the differences between larval and adult taste systems. In adults, responses to common sugars, like glucose, sucrose, and maltose, may include up to eight gustatory receptors, i.e., GR5a, GR61a, and GR64a-f (Dahanukar et al., 2007; Jiao et al., 2007, 2008; Slone et al., 2007; Fujii et al., 2015). Fructose is detected by GR43a, a narrowly tuned receptor expressed in taste organs as well as in the central nervous system (Table 2 and Figure 3A). After a sugar-rich diet, the levels of glucose and trehalose in hemolymph do not increase significantly while fructose levels increase between 3 and 10 times. In line with this, GR43a located in the brain acts as a nutrient sensor, assessing the levels of fructose in the hemolymph (Miyamoto et al., 2012). Even though larvae detect and behaviorally respond to sugar, none of the sweet receptors GR5a, GR61a, and GR64a-f are expressed in this stage (Kwon et al., 2011). Alternatively, larvae only express GR43a in internal taste neurons and in the brain. Larvae display an immediate attraction to fructose and sucrose (disaccharide of fructose and glucose) and a delayed preference toward glucose and trehalose (disaccharide of glucose). Surprisingly, all of these attractive responses depend on GR43a; fructose's attraction depends on GR43a expressed in internal taste neurons while glucose's attraction requires GR43a in the brain (Mishra et al., 2013). These differences in dynamics and receptor localization, suggest that larvae sense fructose or fructose-containing disaccharides directly on internal pharyngeal taste neurons while detection of non-fructose sugars relies on their conversion to fructose post-ingestion, elevation of fructose levels in the hemolymph, and subsequent sensing in the brain (Mishra et al., 2013) (Table 2 and Figure 3A). This apparently inefficient sugar sensing setup provides a simple system that satisfies the larval needs. In order to grow, larvae need to constantly incorporate nutrients and, since their mobility is reduced, looking for the perfect sugar-source could result in great energy costs. It is the adult female fly who carefully analyzes the composition of the medium before choosing an oviposition site and, in doing so, it seems to look for a suitable substrate that will provide with the minimal nutritional requirements for the larvae to grow (Joseph et al., 2009; Schwartz et al., 2012). In addition, fructose and sucrose are present in most fruits, suggesting that having only a rapid fructose detection system may be sufficient on most ecologically relevant substrates. In this sense, larvae generally don't need to search for sugars but they can simply start eating and then evaluate the nutritional content by fast activation of pharyngeal receptors or slower activation of brain receptors. Adult flies, on the contrary, display more complex behaviors that involve the exploration of more heterogeneous environments. In these new environments, flies not only need to evaluate the substrates for the presence of sugars but also the quality of those sugars, since non-fructose sugars may be present in higher proportion. In this sense, a more complex taste system allowing rapid detection of a huge variety of sugars appears as a more suitable setup than the simple version of the larvae.

Another interesting example of chemoreception dimorphism in the fly's life cycle is the case of carboxylic acids perception. While adult flies are strongly repulsed by acidity or high carboxylic acid concentrations (Fuyama, 1976; Ai et al., 2010), larvae display clear attraction (Monte et al., 1989; Kreher et al., 2008; Khurana and Siddiqi, 2013). The chemoreceptors relevant for the attractive responses in larvae have not been identified yet, and only weak activation of some ORs in response to carboxylic acids has been observed (Kreher et al., 2005, 2008). In the case of adults, carboxylic acids are detected through olfaction and taste. In the olfactory system, protons are directly detected by the complex IR64-IR8a (Ai et al., 2010, 2013), and different carboxylic acids trigger strong electrophysiological responses in IR8a⁺ neurons and mild responses in OR neurons (Hallem and Carlson, 2006; Abuin et al., 2011; Silbering et al., 2011). Nevertheless, efforts to elucidate the role of IR8a on carboxylic acid-triggered behavioral avoidance produced contradictory results (Silbering et al., 2011; Ai et al., 2013). In the gustatory system, it was recently shown that high concentrations of carboxylic acids activate a subset of bitter neurons while they inhibit the activity of sweet neurons (Charlu et al., 2013). At the same time, another study demonstrated that carboxylic acids suppress bitter neuron activity when presented in dietary relevant concentrations (Chen and Amrein, 2014). Interestingly, normally aversive bitter/sugar mixtures are rendered more appealing with the addition of moderate concentrations of carboxylic acids (Chen and Amrein, 2014). The identity of the carboxylic acid receptors in the taste organ has not been revealed yet, but they seem to be different from the H⁺ sensor IR64a and the bitter receptors GR33a and Painless (Charlu et al., 2013) (Table 2 and Figure 3B).

Several carboxylic acids are normally present in fly food as fermentation products of yeast and bacteria (Bridges and Mattice, 1939; Idstein et al., 1985; Moat et al., 2002) so, in addition to a simple pH indicator, detection of high concentration of carboxylic acids may serve also as indication of rotten fruit and of the presence of yeasts. This would be important because a previously processed substrate such as rotten fruit could be easier for larvae to feed on; plus, yeasts are the typical source of important nutrients for the larvae, such as proteins and some carbohydrates (Lee et al., 2008; Schwarz et al., 2014). On the other hand, extremely acid solutions are very toxic for adult flies and can result in high mortality in the population (Chakir et al., 1993). The resistance to high concentration of carboxylic acids observed in larvae could represent a tolerance product of the long exposure to low pH media. Consistent with this idea is the fact that female flies normally lay eggs in rotting fruit (Atkinson and Shorrocks, 1977; Markow, 1988), suggesting that larvae are exposed to low pH media throughout their development (Figure 3B). In adults, tolerance to high acetic acid concentrations has already been described for a geographic population. Although still unclear, this increased tolerance could be a consequence of a more efficient detoxification system (Chakir et al., 1993). It would be interesting to test if larvae also have a more efficient detoxification system that allows them to tolerate long exposures to high concentrations of carboxylic acids present in their environment.

Animals in general present bimodal responses to salts: low concentrations of salt trigger attraction while high concentrations, repulsion. This feature reflects the dual effect of salt in the organism: moderate levels of salt are necessary to control electrolyte homeostasis, neuronal activity, and muscle contraction while high levels have deleterious effects as dehydration and hypertension (Liman et al., 2014). In Drosophila, larvae and adult also display the same bimodal responses to salts (Miyakawa, 1981; Balakrishnan and Rodrigues, 1991) but they detect salt apparently through different mechanisms. Attractive responses to low-salt concentrations in larvae require the ENac channels ppk11 and ppk19 (Liu et al., 2003b) in taste neurons. Interestingly, ENac channels are also involved in the low-salt responses in mammals (Chandrashekar et al., 2010), albeit there is no consistent proof that ppk11 and ppk19 in fly larvae or ENac channels in mammals are direct receptors for low-salt solutions (Liu et al., 2003b; Chandrashekar et al., 2010). In adult flies, a recent paper described IR76b as a Na⁺-permeable channel essential for low-salt responses. Furthermore, the authors have clearly demonstrated the existence of two populations of GRNs, one displaying a stronger response to high-salt concentrations and the other one, expressing IR76b, displaying a stronger response to low-salt concentrations. IR76b⁺ taste neurons constitute a new class of GRN specifically tuned to low-salt detection (Zhang et al., 2013a). Interestingly, IR76b is also present in some adult antennal coeloconic OSNs where it might act as a coreceptor (Silbering et al., 2011). In addition to IR76b, ppk11 and ppk19 might also play a role in low-salt detection in adults (Liu et al., 2003b). Coimmunostaining analysis would help to elucidate if ppk11, ppk19, and IR76b are all part of the same detection system or if they constitute two parallel pathways. Moreover, future experiments should analyze whether IR76b is also required in larvae for low-salt detection (Table 2 and Figure 3C).

Regarding high-salt detection, the receptors are still elusive, but ppk11 and ppk19 may contribute to the aversive responses in both adults and larvae (Liu et al., 2003b; Alves et al., 2014). A recent study identified Serrano (SANO), a cytoplasmic protein expressed in bitter neurons, as an essential molecule for the behavioral aversion to high-salt concentrations in larvae. Moreover, inactivation of SANO⁺ bitter neurons triggers an attractive response to high-salt concentrations in larvae (Alves et al., 2014). This strongly suggests that, as it is the case in adults, two neuronal groups are simultaneously activated in response to both low and high-salt concentration, but it is the outcome between these two populations what will determine if there is attraction or repulsion (Figure 3C). Again, the requirement of SANO for high-salt detection in adults was not analyzed. Complementary studies are needed to clearly define whether larvae and adult salt-detection systems are conserved or not.

Our current knowledge of chemosensory perception in D. melanogaster is growing very fast with the identification and ongoing characterization of the different receptor families. However, there is still a lot to do to clearly understand how chemicals are detected, and how this information is processed at the periphery and in the brain to lead to a specific behavioral response. For example, most studies focus on identifying potential ligands for a specific chemoreceptor by using single odor stimulation, a case far removed from the complexity of the natural environmental conditions to which flies are normally exposed. In the natural environment odors are generally present in complex mixtures and it is from these blends that flies need to extract the most relevant signals to behave accordingly. Some putative mechanisms for how the olfactory system decodes relatively complex odorant mixtures have been proposed (Silbering and Galizia, 2007) but this still remains a very important open question.

The complexity of sensing and decoding chemical mixtures is also true for taste perception. It has been shown in a recent study (Chen and Amrein, 2014) that the presence of carboxylic acids in a mixture can modulate bitter and sweet perception. Several questions arise from this observation, could the activation of a specific neuron sensitive to acids potentiate the activity of the neighboring sugar sensing neurons in the peripheral nervous system? Is this possible mechanism shared by all chemosensory neurons? Or could it be specific to some neurons and sensory modalities? Moreover, could acids also inhibit bitter sensing neurons as suggested by this recent work (Chen and Amrein, 2014)?

Concerning the integration of the chemosensory stimuli in the brain the picture is still incomplete. Even though the olfactory system is better described than the gustatory system their neuronal networks are still under characterization. What are the exact connections between the different centers in the brain? Moreover, the precise and complete network from the detection of a chemical at the periphery to the muscle cells that lead to a behavioral output is still partially described. Some recent studies on cVA detection have started to decipher this network (Kohl et al., 2013), and have shown that this cVA circuit seems to be interconnected with other sensory modalities such as hearing (Zhou et al., 2014) which highlights the importance of the connectivity between modalities.

D. melanogaster is a powerful genetic model and we owe it most of our current knowledge on the molecular basis of chemoreception in insects. Nonetheless, it would be interesting to compare how chemical perception is processed in other Drosophila species that have a highly specialized living substrate and to analyze differences and similarities between them. Through these comparative studies we could follow evolutionary traces and study if specific sensory systems have been selected to ensure species survival. The comparison with the chemosensory systems of more distant insects such as mosquitos and bees would also be of great value for the management of species that impact deeply on human health and agriculture.