7.1.1. Drosophila melanogaster as a Volatile Pheromone Model System

Nocturnal insect species like moths have developed extremely sensitive olfactory mechanisms to signal reproductive availability to appropriate mates in the absence of visual cues. Indeed, moths remain an attractive model system to study pheromone detection due to their relatively large size and amazing sensitivity to sex pheromones. However, the lack of potent genetic tools limits the use of these animals to dissect the molecular basis for pheromone transduction. Drosophila melanogaster has proven to be a valuable system to explore the molecular mechanisms mediating insect pheromone detection and to elucidate of the neuronal circuits underlying pheromone-induced behaviors. The tools developed from over a century of fly research allow us to produce single gene mutants and evaluate the contribution of that single gene on pheromone detection in vivo. Using flies, it is possible to identify gene products responsible for virtually any scorable phenotype. Genetic screens for mutants that are defective for pheromone responses have revealed mutations in several genes critical for this process (Jin et al. 2008). These mutations are mapped to the fly genome and fine mapping and sequence analysis identify the genes required for pheromone sensitivity. The advantage of such a forward genetic approach to find pheromone sensitivity factors is that it is unbiased. Any gene product required for any aspect of pheromone detection will be recovered, even if it encodes an unexpected factor. One major challenge for a forward genetic approach for pheromone-defective mutants is that lethal genes will not be recovered in a typical screen. Since we expect most or all genes involved exclusively in pheromone detection to be dispensable for life, this should not be a major problem. A second major challenge for a genetic screen is to elucidate the biological role for these important gene products in the pheromone detection process once they are identified. In addition to genes encoding factors that directly mediate signaling, a genetic screen will also recover mutations in transcription factors and other biosynthetic proteins required for proper production of signaling factors, since these will also cause pheromone insensitivity.

Flies offer additional advantages as a model to study pheromone biology. The fly genome is easily manipulated. Transgenic animals can be produced with single-copy transgenes, allowing the expression of modified proteins in the mutant background, essentially replacing the wild type protein with the mutant variant. This allows the evaluation of any defects under the most physiological relevant conditions possible.

The ability to manipulate the genome is also useful if one takes a reverse genetic approach. In this case, interesting genes may be identified in the genome or have an interesting homolog in another species. Using one of a plethora of tools, one can engineer mutants in virtually any Drosophila gene. Homologous recombination can be used to target individual genes and produce null mutants (Rong et al. 2002). In addition, the gene disruption project has produced thousands of fly stocks, each carrying a single transposable element that often disrupts expression of one or more genes at the integrations site (Bellen et al. 2004). Some elements encode flippase recognition target (FRT) sites, the recombination site for flippase (FLP) recombinase enzyme (Golic and Lindquist 1989). One can produce small deletions to remove a gene of interest by crossing two FRT strains with elements integrated on either side of the gene of interest, and inducing FLP. FLP will mediate mitotic recombination between the FRT sites on the two chromosomes, resulting in a chromosome lacking the DNA between the FRT sites. These approaches are the gold standard for analyzing the function of Drosophila genes, because they completely eliminate expression of the gene product of interest. Alternatively, transgenic RNA interference (RNAi) approaches are useful, and transgenic lines carrying RNAi constructs are available that target every gene in the genome. RNAi works well in Drosophila, with >95% knockdown of protein products achieved routinely (Kalidas and Smith 2002). However, like RNAi in vertebrate cells, knockdown is never complete and there may be enough protein produced to confer pheromone sensitivity. In such cases, a protein essential for pheromone detection might be missed in a leaky RNAi line.

Olfaction in flies can be probed using electrophysiology, behavior, cell biology, and biochemistry. While small, flies are amenable to electrophysiology. For olfaction, we perform single sensillum recordings (SSRs) that are extracellular recordings of the olfactory neurons within a single sensillum. Single sensillum electrophysiology allows monitoring of real-time neuronal activity in pheromone-sensitive neurons. Optogenetic approaches continue to improve and be applied to Drosophila pheromone biology. For example, transgenic expression of calcium-activated fluorescent proteins like GcAMP3 (Tian et al. 2009) emit light when they bind the calcium that floods the cytoplasm when neurons are activated. These reporters lack the exquisite time resolution of electrophysiology but have the advantage that the activity of many neurons can be monitored simultaneously. Finally, techniques are available in flies that allow us to trace the neuronal circuits underlying pheromone behaviors. Mosaic analysis with a repressible cell marker (MARCM) (Lee and Luo 2001) was developed by Liqun Luo and coworkers and utilizes FLP-mediated expression of a green fluorescent protein (GFP) reporter in small subsets of cells, producing an effect analogous to a fluorescent golgi stain (Lee and Luo 2001). Photoactivatable GFP is a GFP variant that fluoresces only after exposure to high-intensity 710-nm light (Patterson and Lippincott-Schwartz 2002). Thus, if the axonal target for a neuron in the pheromone detection circuit is known, the target can be illuminated, allowing the next order neuron in the circuit to be labeled and traced (Datta et al. 2008). These techniques allow us to follow the information flow from olfactory neurons activated by pheromone to the downstream circuits and determine how that information is delivered and processed by the central nervous system (see below) (Jefferis et al. 2007; Ruta et al. 2010).

7.1.2. Drosophila Olfactory Anatomy

Flies have two paired olfactory organs located on the head; the antennae and the maxillary palps (Figure 7.1a, b). These organs are covered with hollow, hairlike cuticle structures called sensilla, each of which contains the dendrites of 1–4 olfactory neurons (Figure 7.1b, c). Thus the olfactory neurons are partitioned into discrete isolated units. Similar to human hair cells in the ear, the olfactory neuron dendrites within the sensilla are bathed in a potassium-rich fluid called the sensillum lymph. In Drosophila, there are four morphologically distinct classes of sensilla, the food-odorant detecting basiconic sensilla, the coeloconic sensilla, the trichoid sensilla, and the intermediate sensilla (Stocker 1994). The latter are so named because they are intermediate in morphology between the trichoid and basiconic sensilla. In most insects, pheromones are detected by neurons located within the trichoid sensilla. The olfactory neurons project dendrites into the hollow sensilla (Figure 7.1c), and their axons travel to the antennal lobes in the fly central nervous system where they synapse with projection neurons in neuronal clusters called glomeruli (Figure 7.1d).

FIGURE 7.1

Anatomy of the Drosophila olfactory system. (a) SEM of a Drosophila head. Arrow indicates the third segment of the antenna that contains the olfactory neurons. (b) Higher magnification of the antenna surface showing individual hairs or sensilla. Arrow (more...)

There are many parallels between insect and vertebrate olfactory processing (Hildebrand and Shepherd 1997). Like vertebrates, the olfactory neurons in insects tend to express a single tuning odorant receptor gene, and neurons expressing the same receptor innervate the same glomerulus in the brain (Figure 7.1d). Therefore, the odorant code established in the brain results from the pattern of glomeruli activated by an odor, which in turn, depends on the odorant sensitivity of individual odorant receptors. Neuronal activity in each glomerulus is relayed to higher processing centers by projection neurons (analogous to mitral cells in vertebrates). A total of 20 to 50 olfactory neurons expressing the same odorant receptor converge at the glomerulus and innervate a small number of dedicated projection neurons. Therefore, there is signal amplification and noise reduction occurring through convergence at the glomerulus (Bhandawat et al. 2007). In Drosophila, the projection neurons innervating most glomeruli activated by food odorants innervate distinct targets from the projection neurons activated by pheromones (Jefferis et al. 2007). Indeed, when analyzed in vivo, exposure of females to male fly volatiles activates a single glomerulus, DA1, corresponding to activation of the cVA-sensitive neurons (Masuyama et al. 2012). In contrast to food odorants that usually activate multiple odorant receptor types, cVA detection activates a single labeled line (Schlief and Wilson 2007).

In most insects, pheromone detection occurs in the trichoid sensilla neurons. Drosophila has three classes of trichoid sensilla, the at1, at3, and at4 classes (Figure 7.2). The at2 sensillum class, previously thought to be a member of the trichoid class (Couto et al. 2005) is actually an intermediate sensillum class we renamed ai2 (Ronderos and Smith, in press). The 50 or so at1 sensilla are specialized for cVA detection, and the single neuron located within these sensilla uniquely express the odorant receptor Or67d (Ha and Smith 2006; Kurtovic et al. 2007). The at3 and at4 sensilla each contain three neurons, but little is known about their odorant sensitivity or function. The projections of all seven classes of neurons have been mapped to the antennal lobe glomeruli (Figure 7.2) (Couto et al. 2005).

FIGURE 7.2

Drawing depicting the Drosophila trichoid sensilla. Three classes are present on the antenna, at1, at3, and at4. at1 sensilla contain a single neuron that expresses Or67d and is exquisitely tuned to cVA pheromone. at3 sensilla contain three neurons expressing (more...)

7.1.3. Drosophila Odorant Receptor (Or) Family

The Drosophila Or family consists of 62 seven-transmembrane odorant receptors derived from 60 genes. These receptors have seven-transmembrane domains but are not G-protein-coupled receptors (GPCRs), as their topology is inverted in the membrane relative to classical GPCRs (Benton et al. 2006; Lundin et al. 2007). In contrast to vertebrate odorant receptors, the insect Ors are ligand-gated ion channels (Sato et al. 2008; Wicher et al. 2008). One putative receptor, Orco, is expressed in most olfactory neurons and functions as a coreceptor and forms a heterodimer with the tuning members of the Or family that are responsible for providing the odorant specificity (Larsson et al. 2004). Nearly all tuning odorant receptors have been mapped to specific functional classes of olfactory neurons (Couto et al. 2005; Vosshall et al. 2000). The tuning olfactory receptors appear necessary and sufficient to confer odorant sensitivity on basiconic neurons, and the tuning receptors present in basiconic sensilla are activated directly by food odorants. Mis-expression of an odorant receptor in a basiconic neuron lacking its normal tuning receptors converts the odorant specificity of that host neuron to that of the donated receptor (Hallem et al. 2004). By contrast, receptors normally found in trichoid sensillum neurons remain silent when expressed in basiconic neurons (Laughlin et al. 2008). Yet, misexpressing Or67d, normally expressed exclusively in the cVA sensing neurons, in other trichoid neurons confers robust cVA sensitivity that is nearly identical to that of at1 neurons. This suggests there are additional trichoid factors important for cVA detection that are not present in basiconic neurons. Therefore, pheromone signal transduction is distinct from food odorant sensing, requiring additional signaling factors (see below).

7.1.4. Drosophila Odorant Binding Proteins

Odorant binding proteins (OBPs) are small (~14 kD) proteins that are secreted into the sensillum lymph, bathing the olfactory neuron dendrites by nonneuronal support cells. In contrast to vertebrate OBPs (that are members of the lipocalin protein family), insects have evolved a novel OBP family. Fifty OBP genes are present in the Drosophila genome and share three features. All members studied to date are low-molecular weight (~14 kD), have six conserved cysteines, a signal sequence for secretion, and are expressed in chemosensory organs (Galindo and Smith 2001; Hekmat-Scafe et al. 2002; Robertson et al. 2003). The OBPs are expressed in stereotypic subsets of chemosensory sensilla, both in the olfactory and gustatory organs, and share little sequence homology, consistent with odorant-specific roles (Galindo and Smith 2001).

The first insect OBP was identified by Vogt and Riddiford (1981). These workers identified a pheromone-binding protein in the sensillum lymph of the pheromone-sensitive sensilla in male moths. Later, other OBP members were identified in moths that were expressed in food-sensing sensilla, and were called general odorant binding proteins (GOBPs). The first Drosophila OBPs were discovered in John Carlson’s lab and Mike Rosbash’s lab using molecular screens for genes specifically expressed in the antenna (McKenna et al. 1994; Pikielny et al. 1994). The completed Drosophila genome sequence revealed the large size of this gene family that is rivaled only by the chemoreceptor families (Galindo and Smith 2001; Hekmat-Scafe et al. 2002). The function of OBPs remains controversial. Because they are expressed by the support cells and not neurons and are secreted into the sensillum lymph, they must function upstream of any neuronal function in the odorant detection process. The role for one OBP, LUSH, is discussed below.

7.1.5. Gene Products Underlying cVA Pheromone Detection in Drosophila

Pheromone signal transduction mechanisms are distinct from those mediating detection of general odorants in both vertebrate and insect species (Chamero et al. 2012; Dulac and Axel 1995; Ha and Smith 2009; Ronderos and Smith 2009). cVA pheromone detection requires a set of gene products not required for olfaction to food odorants, yet does share a requirement for Orco. Orco, the Or coreceptor that makes up part of the odorant-gated ion channels, is required for cVA responses, as Orco² mutants lack cVA sensitivity (Figure 7.3). The requirement for Orco for cVA sensitivity logically implicated one of the 62 tuning receptors was likely required for cVA sensitivity as well. Indeed, one receptor, Or67d, is expressed exclusively in at1 neurons and mediates cVA responses (Ha and Smith 2006; Kurtovic et al. 2007).

FIGURE 7.3

LUSH, Orco, and Or67d are required for cVA responses and normal spontaneous activity. Sample traces from wild type (w¹¹¹⁸), Orco² mutants, lush¹ mutants, and Or67d^gal4 mutants. Arrow indicates application of 1% cVA for each trace.

Or67d was discovered by Hugh Robertson, scanning for odorant receptors in the Drosophila genome (Robertson et al. 2003). This receptor was first demonstrated to be the cVA tuning receptor for at1 neurons by Ha et al. (Ha and Smith 2006). Ha undertook a genetic screen for mutants defective for cVA detection (Jin et al. 2008). One gene identified was a transcription factor called Rotund. Mutants lacking rotund have defects specifying the cell fate of various olfactory neuron classes (Li et al. 2013). One consequence of this is that the at1 sensilla are transformed into at4 sensilla that normally do not express rotund. Therefore, there are no at1 sensilla on the antenna of rotund mutants. Ha then performed RT-PCR experiments looking for Or members expressed in wild type antenna, but not in rotund mutants. Or67d was absent in rotund mutant antenna but present in wild type by RT-PCR experiments. To prove Or67d actually confers cVA sensitivity, he misexpressed this receptor in all neurons using the pan-neuronal driver pELAV. Recordings from at4 sensilla containing neurons that are normally insensitive to cVA now responded robustly to cVA. cVA sensitivity in these at4 neurons also required LUSH, as cVA sensitivity was lost to basiconic neurons when the lush mutant was engineered into this stock (Ha and Smith 2006). However, Or67d, Orco and LUSH are not sufficient to confer cVA sensitivity to basiconic neurons (Figure 7.4) (Laughlin et al. 2008). Therefore there must be additional cVA sensitivity factors present in trichoid sensilla that are missing in basiconic sensilla.

FIGURE 7.4

Expression of the known cVA sensitivity factors, LUSH, Or67d, and SNMP in the empty basiconic neuron system fails to restore full cVA responsiveness. (a) Single sensillum electrophysiological recordings using the empty neuron system expressing cVA sensitivity (more...)

7.1.6. LUSH OBP

LUSH is an OBP expressed exclusively in all trichoid sensilla in the antenna, but is also found in a subset of tarsal chemoreceptors on the forelegs (see below). The lush was identified in a transposable element-based enhancer-trapping screen designed to identify genes expressed in the antenna (Kim et al. 1998). A transposable element carrying the LacZ gene was mobilized and allowed to randomly integrate throughout the genome (Callahan and Thomas 1994). Among several thousand insertion lines, one expressed β-galactosidase exclusively in the ventral-lateral surface of each antenna (Figure 7.5a). Cloning the integration site of the transposon insertion from this line revealed it had integrated just downstream of a gene predicted to encode a new member of the invertebrate OBP family we called lush (Kim et al. 1998) that was abundantly expressed in the antenna but absent from heads and bodies. Immunofluorescence studies using antiserum against the recombinant LUSH protein revealed this OBP is expressed exclusively in trichoid sensilla in the same pattern as the LacZ in the enhancer-trap line (Kim et al. 1998). The next question was, what does LUSH do?

FIGURE 7.5

LacZ expression under control of the lush promoter. (a) Head of a transgenic fly expressing nuclear-localized β-galactosidase showing expression restricted to the support cells of the trichoid sensilla. (b) β-Galactosidase expression is (more...)

The role of OBPs in chemosensation was unknown at that time, and remains controversial. The work of Vogt and Riddiford indicated members of this family bind directly to odorant ligands (Vogt and Riddiford 1981). This led to two prominent models. In the first, the OBP was postulated to play a role in removing stray odorant molecules from the sensillum lymph so the animal is ready for the next odor plume. Indeed, to follow a concentration gradient, there must be some way to rapidly clear stray pheromone molecules. The second model suggested OBPs transport olfactory ligands through the sensillum lymph to the dendrites. We called this the carrier model. No mutants in any OBP gene had been previously described in any species to provide additional insight into OBP function to distinguish among these models. Having a transposable element integrated near the lush gene provided an exciting opportunity to generate the first OBP mutant. When excising, transposons can produce local deletions. Remobilizing the transposon from the integration site near the lush gene produced several hundred excision lines. One line lacked LUSH protein expression, confirmed by western blots from antennal extracts (Kim et al. 1998). Sequence analysis of the genomic region from this line revealed a 1569 base pair deletion removing the entire lush coding sequence. This mutant is called lush¹.

Initially, we tested general odorant detection in lush¹ mutants. Screening olfactory behavior in lush¹ mutants revealed these flies are more likely to enter traps containing high concentrations of short-chain alcohols, including ethanol. Since they showed affinity for high alcohol environments, the mutant was named lush (Kim et al. 1998). However, this attraction behavior (or perhaps lack of avoidance) only occurs at extremely high alcohol concentrations that are unlikely to be encountered in nature, so the significance of the alcohol phenotype is unknown. Structural studies with LUSH confirmed that LUSH protein binds directly to ethanol, and defined the first ethanol-binding site defined in a protein (Kruse et al. 2003). One possibility is that LUSH blocks access of ethanol molecules to neurons that mediate avoidance behavior to alcohol, but future work will be required to understand the behavior of lush¹ mutants to alcohol. What became clear is that lush¹ mutants are completely anosmic to physiological concentrations of the male-specific pheromone, cVA (Xu et al. 2005). This very dramatic effect on cVA detection reveals LUSH is required to facilitate activation of the at1 neurons in the presence of cVA pheromone. Since LUSH is required for activation of the cVA sensitive neurons, this eliminates the pheromone removal hypothesis in which the sole function of LUSH is to remove odorant molecules from the sensillum lymph.

A second phenotype that was entirely unexpected was that lush¹ mutants are associated with a profound loss of spontaneous activity (action potentials in the absence of cVA) in the at1 neurons. Normally, in the absence of cVA, at1 neurons spike at one action potential per second. The spontaneous activity rate of the at1 neurons in lush¹ mutants in the absence of cVA averages only one spike every 400 seconds. If LUSH functioned as a simple pheromone transporter, why would its absence alter the spontaneous activity rate of the at1 neurons in the absence of cVA? The effect on spontaneous activity is specific to at1 neurons and there is no impact on other trichoid neurons that suffer the same loss of LUSH from the sensillum lymph. Therefore it is not likely this defect results from an osmotic effect secondary to the loss of this abundant LUSH protein from the lymph. This finding indicates LUSH has a more intimate relationship with cVA signaling, perhaps as a partial agonist at the at1 neuronal receptors.

For LUSH to function as a partial agonist in the absence of cVA, the LUSH protein itself must interact with neuronal receptors expressed in at1 neurons. One possibility is that LUSH can isomerize from an inactive state to an activated conformation capable of activating at1 neuronal receptors, and the role of the pheromone is to stabilize or induce the activated conformation in LUSH. If true, this model makes two testable predictions. First, there should be a cVA-dependent conformational change in LUSH that is distinct from the conformation induced by other ligands (such as ethanol) that can bind LUSH but do not mediate at1 activation. Second, it should be possible to engineer dominantly active mutant forms of LUSH that activate at1 neurons in the absence of cVA.

To explore the nature of the LUSH-cVA complex, the x-ray crystal structure of LUSH with and without cVA bound were solved (Laughlin et al. 2008). cVA is completely encapsulated within the LUSH-cVA structure. cVA binding induces a conformational change in LUSH, ejecting phenyalanine 121 out of the cVA binding pocket, leading to disruption of the salt bridge between aspartate 118 and lysine 87. This causes the C-terminal loop of LUSH to flip outward. Confirming that the conformational change is important for activation of at1 neurons, mutating the F121 to alanine should reduce the ability to induce the activated conformational shift. Indeed, LUSH^f121a, when infused through the recording pipette into lush¹ mutants, reduces cVA sensitivity 50-fold compared to infusing recombinant wild type LUSH. This mutation has no effect on cVA binding (Laughlin et al. 2008). Conversely, mutating phenylalanine 121 to the bulkier tryptophan (LUSH^f121w) makes at1 neurons neurons five times more sensitive to cVA when this protein is infused into the sensillum lymph! These findings support the idea that the conformational change is an important component of the at1 activating ability of the LUSH-cVA complex. Disruption of the salt bridge between K87 and D118 (LUSH^d118a) that helps maintain LUSH in an inactive state in the absence of cVA results in a protein that spontaneously activates at1 neurons when infused into these sensilla. at1 neuron activity is activated after a few minutes of LUSH^d118a infusion and peaks after approximately 20 minutes at an average of 20 spikes/sec (Laughlin et al. 2008). Infusion of LUSH^d118a into other sensilla has no effect on neuronal activity, confirming LUSH^d118a specifically activates receptors present on the at1 neurons but not in other olfactory neurons. Indeed, the crystal structure of LUSH^d118a is almost identical to that of cVA-bound LUSH, beautifully explaining the mechanism for the dominant activation by LUSH^d118a. While LUSH^d118a selectively activates at1 neurons in the absence of cVA, it does not completely mimic full activation achieved by high concentrations of cVA (20 spikes per second vs >50 spikes per second with cVA).

When transgenic lush¹ flies express lush^d118a under control of the lush promoter, the at1 activation effect is much weaker than when recombinant protein is infused directly into the sensillum lymph. There is only a twofold increase in spontaneous activity compared to wild type flies (Ronderos and Smith 2010). Despite this modest increase in activity of the at1 circuit, the transgenic flies act as if there is cVA present. Male flies exhibit delayed courtship, while females expressing lush^d118a exhibit enhanced courtship (Ronderos and Smith 2010). This is consistent with the sexually dimorphic behavioral effects of cVA (Kurtovic et al. 2007; Ronderos and Smith 2010). Can doubling the spontaneous activity rate affect behavior? It is known that even small increases in at1 neuronal activity are amplified because of convergence. Multiple at1 neurons innervate the DA1 glomerulus, increasing the likelihood of activating the projection neurons (Bhandawat et al. 2007). Therefore, small increases in at1 firing are sufficient to alter behavior. The relatively weak activation of the at1 neurons by transgenic lush^d118a was surprising, given that infusion of bacterially expressed LUSH^d118a activates at1 neurons strongly. One possibility is that the olfactory neurons desensitize to the presence of activated LUSH^D118A when expressed chronically as a transgene. Alternatively, there may be chaperone-like factors present in the secretory system of the support cells that prevent secretion of LUSH in the active conformation. Consistent with this notion, when LUSH^d118a is expressed as a transgene, it rescues cVA sensitivity (Ronderos and Smith 2010), while direct infusion does not (Laughlin et al. 2008). The fact that lush¹ mutants show defective spontaneous activity supports the idea that the binding protein is a partial agonist that activates at1 neuronal receptors in the absence of cVA. This is further supported by the mutational studies described above that functionally dissociate cVA from the conformational shifts in the LUSH protein structure that activate at1 neurons. Is the conformational activation model the whole story? Since dominant LUSH does not fully activate the at1 neurons, other protein interactions or perhaps cVA itself may be important for full at1 activation. Although the complete picture remains to be unraveled, what is clear at this point is that conformational activation of LUSH is part of the at1 activation mechanism.

Extremely high cVA levels can directly activate at1 neurons in lush¹ mutants, demonstrating a LUSH-independent at1 activation mechanism. This is consistent with the idea that pheromone detection evolved from a system in which cVA originally activated at1 receptors directly. The LUSH cVA pheromone binding protein may have evolved initially as a cVA carrier or capturing factor to increase sensitivity and was later incorporated into the at1 signaling mechanism. It will be interesting to determine if mutations in the pheromone-binding OBPs from other species have a similar defect as is observed in lush¹ mutants.

7.1.7. SNMP

SNMP was first identified in moths as an abundant 67-kD protein expressed on the dendrites of a subset of olfactory neurons (Rogers et al. 1997). However, the function of this two-transmembrane-spanning protein was not clear. SNMP is a member of the CD36 protein family. Many CD36 members are scavenger receptors or play a role in the transmembrane transport of lipoprotein complexes (reviewed in Vogt et al. 2009). For example, in humans, CD36 is required for uptake of oxidized cholesterol by macrophages resulting in conversion of macrophages into foam cells—an important step in the formation of atherosclerotic plaques in arteries (Collot-Teixeira et al. 2007). In Drosophila, other CD36 homologs are important for recognition and removal of dead cells (Franc et al. 1996) and bacteria (Philips et al. 2005), absorption of vitamin A from the gut (Gu et al. 2004; Kiefer et al. 2002), and transfer from the hemolymph into the retina (Wang et al. 2007). Mutants defective for the Drosophila homolog of moth SNMP were independently produced and studied in the Smith lab (Jin et al. 2008) and the Vosshall lab (Benton et al. 2007). The Vosshall group produced an SNMP mutant using homologous recombination, following up on the findings of Rogers et al. showing SNMP is restricted to dendrites of the pheromone sensitive neurons (Rogers et al. 1997; Rogers et al. 2001a; Rogers et al. 2001b). Tal Soo Ha and Xin Jin identified SNMP mutants as one of several mutants defective for cVA detection recovered in a genetic screen (Jin et al. 2008). Mutants lacking SNMP are insensitive to cVA (Benton et al. 2007; Jin et al. 2008). However, unlike virtually all other mutants defective for cVA detection studied to date, Snmp mutants show increased, not decreased, spontaneous activity in the at1 neurons. The increased spontaneous firing rate, around 20 spikes per second, is similar to that observed when dominant lush^d118a is infused into the lymph of at1 sensilla (Laughlin et al. 2008). SNMP protein is expressed in subsets of olfactory neurons and support cells, but cVA sensitivity specifically requires SNMP expression in the at1 neurons (Benton et al. 2007; Jin et al. 2008). SNMP functions on the surface of the at1 neuron dendrites, the site of pheromone signal transduction, as infusion of anti-SNMP antiserum into the sensillum lymph mimics the mutant phenotype (Jin et al. 2008). The Vosshall group also showed fluorescence energy transfer (FRET) between SNMP and Orco, indicating these proteins are in close proximity to each other (Benton et al. 2007). This data is consistent with SNMP acting as a component of the neuronal cVA-activated receptors.

One model for SNMP function in pheromone detection is that it mediates the transfer of pheromone, directly or from the binding protein, to the Or67d/Orco complex (Benton et al. 2007; Rogers et al. 2001). While this model would explain the loss of cVA sensitivity at low concentrations, several lines of evidence are not consistent with this model. First, the loss of a pheromone transfer protein would not be expected to affect the spontaneous activity in the at1 neurons. Snmp mutants have higher than normal spontaneous activity. Second, if SNMP transfers cVA, it should be possible to activate the at1 receptors with high cVA concentrations in the absence of SNMP. No amount of cVA is capable of activating at1 neurons in Snmp mutants (Benton et al. 2007; Jin et al. 2008). Third, SNMP appears to be required for dominant LUSH^d118a to activate at1 neurons. While infusion of recombinant LUSH^d118a into the sensillum lymph of wild type at1 neurons activates at1 neurons, LUSH^D118A fails to activate at1 neurons from Snmp mutants (Jin et al. 2008). Either cVA and LUSH activate at1 neurons through SNMP, or the neurons lacking SNMP are simply not capable of further activation over their higher endogenous spontaneous rate. The latter seems less likely given that wild type at1 neurons can fire at over 50 spikes per second when stimulated by high levels of cVA. Therefore, it seems most likely that SNMP functions as a negative regulator of Or67d/Orco complex and its inhibition is relived by interactions with activated LUSH accounting for partial activation. Since dominant LUSH (and Snmp mutants) only produce partial activation, there may be role for cVA itself in full activation. This would also be consistent with cVA acting as a weak agonist in the absence of LUSH. Figure 7.6 shows a current model for cVA signal transduction in at1 neurons.

FIGURE 7.6

Model for cVA detection in Drosophila at1 neurons. Left, inactive LUSH (open squares) does not activate at1 neuronal receptors, composed of Or67d (blue), Orco (orange), and SNMP (purple). SNMP is postulated to maintain the at1 receptors in the off state. (more...)

7.2.1. Courtship

In Drosophila, like all animals, behaviors are elicited in response to sensory inputs. Courtship is a well-characterized set of stereotypical behavior patterns that occurs prior to mating. Males actively court females through a series of ritual behaviors that ultimately conclude in copulation (reviewed in Dickson 2008; Greenspan and Ferveur 2000; Hall 1994; Manoli et al. 2006; Pan et al. 2011; Vosshall 2008). Progression through courtship requires interactions between partners that are mediated through the olfactory, visual, auditory, tactile, and gustatory senses to insure mating is directed toward an appropriate partner (Amrein 2004; Dickson 2008). cVA activates at1 neurons equally in both male and female fruit flies. This pheromone plays a role in courtship behavior in both sexes but induces different behaviors in each sex. What is the role of cVA in courtship, and how can cVA trigger different behaviors in males and females?

The first clue that cVA is important for courtship came from the work of Jallon who noted an antiaphrodisiac effect of cVA (Jallon et al. 1981). The effect of cVA on courtship became clearer when Or67d mutants were studied. Males lacking Or67d display increased courtship toward other males, compared to wild type controls (Kurtovic et al. 2007). Females lacking this receptor show prolonged latency to copulation (Kurtovic et al. 2007). The opposite behaviors are observed for each sex when at1 neurons are activated by LUSH^d118a, indicating cVA is an aphrodisiac for females but an antiaphrodisiac for males (Ronderos and Smith 2010). Are at1 neurons the sole source of cVA detection? Studies reported that the odorant receptor Or65a, normally expressed by neurons in the at4 sensilla (Couto et al. 2005) could be activated by high cVA concentrations when this receptor was misexpressed in basiconic neurons (van der Goes van Naters and Carlson 2007). Furthermore, expressing tetanus toxin under control of an Or65a promoter, which blocks synaptic transmission in these neurons, disrupts cVA-induced suppression of courtship, while tetanus toxin had no effect when expressed by Or67d promoter (Ejima et al. 2007). However, the Or65a promoter used in this work also drives tetanus toxin in other brain neurons, which could affect behavior independently of activity in the Or65a neurons (Ejima et al. 2007). Regardless, it is possible that activation of multiple neuronal circuits, triggered through different odorant receptors, underlies cVA-induced behavior.

To address whether cVA acts through Or67d-expressing at1 neurons alone or through additional circuits to modulate courtship behaviors, David Ronderos studied transgenic flies expressing the dominant LUSH^D118A in trichoid sensilla in lush¹ mutants. By expressing LUSH^d118a as a transgene under control of the lush promoter, he can specifically activate at1 neurons in the absence of cVA (Ronderos and Smith 2010), since dominant LUSH^d118a only activates at1 neurons (Laughlin et al. 2008). Dominant LUSH^d118a induced suppression of courtship in males and increased receptivity to courtship advances in females (Ronderos and Smith 2010) (Figure 7.7). Importantly, the LUSH^d118a behavioral effects on both sexes was absent when Or67d was removed, confirming these effects are mediated through at1 neurons. These findings are consistent with the receptor mutant studies by Kurtovic et al. (2007) and indicate at1 neurons are necessary and sufficient for cVA-induced courtship behavior.

FIGURE 7.7

Sexually dimorphic cVA-triggered mating behaviors are mediated specifically by T1 neurons. Wild type male and female pairings result in a courtship index of ~30% (left). cVA exposure suppresses male courtship behavior in wild type males to a level indistinguishable (more...)

How can cVA cause different behavior in males and females? As previously discussed, there is no difference in cVA sensitivity or responses at the at1 neuron level between males and females. Therefore the differences must be downstream in the circuit. Using photoactivatable GFP, Datta et al. illuminated the DA1 glomerulus in flies broadly expressing the GFP construct, resulting in activation of GFP only in the illuminated neurons (Datta et al. 2008). They were able to trace the DA1 projection neurons (PNs), whose dendrites were labeled in this fashion, and examine the synaptic connections these cells made. Remarkably, when they looked at the structure of these PNs, they noted a clear difference in males and females. Males have a unique axonal branch in the lateral horn that innervated cells that are not innervated in females, and this branch was observed in all males examined (Datta et al. 2008). We do not know if this axon branch is responsible for any of the sexually dimorphic behavior induced by cVA, but future studies may find ways to cleave this branch and evaluate the effects on courtship. Richard Axel’s group has now extended the cVA circuit to four neurons connected by three synapses using photoactivatable GFP and illuminating the axonal targets of each group of neurons (Ruta et al. 2010). The last neuron innervates the ganglia in the ventral nerve cord. Interestingly, these downstream neurons are also sexually dimorphic and express the sex determination transcription factor fruitless (Fru) (Yu et al. 2010). Understanding precisely how these circuits operate to produce behavior is still unknown, but a wiring diagram of the circuit is certainly a good roadmap. Perhaps computer modeling of the circuit will lead to new insight.

7.2.2. Aggregation

In addition to effects on courtship, cVA also modulates aggregation and aggression behavior. The combination of cVA and food odorants provide a potent cue that fruit flies find extremely attractive (Bartelt et al. 1985). cVA is also transferred to females upon mating, both by contact with male cuticle and within the seminal fluid (Butterworth 1969). The latter results in cVA deposition on the food substrate when eggs are laid. Both males and females are attracted to this odorant cocktail, making this a nice mechanism to aggregate both sexes to the same location where they will be close enough to promote courtship. Additionally, a member of the ionotropic odorant receptor family, IR84a, is activated by the fruit odorants phenylacetic acid and phenylacetaldehyde, and is expressed in olfactory neurons that activate a FRU-positive circuit that enhances male courtship behaviors (Grosjean et al. 2011). These mechanisms promote reproductive behavior at appropriate sites for feeding and egg laying.

7.2.3. Aggression

Male aggression is a social behavior that is manifested in almost all species of the animal kingdom. Males display aggressive behaviors to establish dominance, to compete for a limited food resource, and for the chance to mate with receptive females. In Drosophila, aggressive behavior was first observed in 1915 by Sturtevant and has since been characterized in terms of behavior traits. Flies exhibit characteristic aggressive behaviors during male–male interactions (Dahanukar and Ray 2011; Dickson 2008). These behaviors include (1) approaching, when a male fly will lower his body and advance towards the second fly, (2) wing threats, when one fly will quickly raise its wing towards another, (3) lunging, when the aggressive fly will throw himself on the other fly, (4) boxing, when flies will raise up on their hind legs and hit each other using their forelegs, and finally, (5) tussling, when both flies fall over each other, holding, kicking, and chasing each other (Chen et al. 2002; Dow and von Schilcher 1975; Skrzipek et al. 1979; Zwarts et al. 2012). Aggression assays are usually measured by placing pairs of flies in a chamber with a resource, either food or a female, and videotaping behavior and counting the number of times the aggressor threatens or lunges, or boxing episodes (Certel and Kravitz 2012).

Fruitless expression was shown to underlie dominance behavior in male fruit flies (Vrontou et al. 2006) and the cVA circuit contains several Fru-expressing neurons (Vrontou et al. 2006). In a series of elegant experiments cVA was shown to directly promote aggression between two male flies when present in the observation chambers (Liu et al. 2011; Wang and Anderson 2010; Wang et al. 2011). Activating at1 neurons by expressing an activated cation channel under the Or67d promoter that causes chronic depolarization of at1 can recapitulate the aggression phenotype in the absence of cVA (Wang and Anderson 2010). The effect of cVA on aggression is dose-dependent (Wang and Anderson 2010). At lower concentrations of cVA, the pheromone promotes aggregation (Bartelt et al. 1985; Wang and Anderson 2010). When the number of flies increases, the cVA concentration increases and may produce increased male–male aggression that disperses the flies. Thus, cVA may regulate population density at food resources. Whether this model is correct and how the relative concentration of cVA causes a change in behavior is unknown. However, this could easily be tested using cVA-insensitive mutants.

Interestingly, cVA fails to promote aggression in Gr32a null males (Wang et al. 2011). Gr32 is a taste receptor expressed in the legs and is thought to be stimulated by the male-specific cuticle hydrocarbon, 7-(z)-tricosene (Wang et al. 2011). Perhaps dual inputs from both the cVA and Gr32 circuits are required to promote aggressive behavior. This could account for aggressive behaviors being restricted to males and only when they are in direct contact (Wang et al. 2011). It will be interesting to determine how and where the Gr32a and Or67d circuits interact to promote aggression.

Serotonin, octopamine, and the product of the white gene have been implicated in modulating aggression (Certel et al. 2010; Dierick and Greenspan 2007; Hoyer et al. 2008; McDermott et al. 2009; Zhou and Rao 2008). A cytochrome P450 homolog, Cyp6a20, has also been suggested to have a role in modulating aggression in Drosophila (Wang et al. 2008). Cyp6a20 is expressed by support cells in the pheromone-sensitive trichoid sensilla (van der Goes van Naters and Carlson 2007). These support cells also express LUSH (Kim et al. 1998). Since Cyp6a20 is not secreted into the sensillum lymph, it remains to be determined exactly how this factor influences aggressive behavior.