Insect pheromones elicit stereotypic behaviors that are critical for survival and reproduction. Defining the relevant molecular mechanisms mediating pheromone signaling is an important step to manipulate pheromone-induced behaviors in pathogenic or agriculturally important pests. The only volatile pheromone identified in Drosophila is 11-cis-vaccenyl acetate (VA), a male-specific lipid that mediates aggregation behavior. VA activates a few dozen olfactory neurons located in T1 sensilla on the antenna of both male and female flies. Here, we identify a neuronal receptor required for VA sensitivity. We identified two mutants lacking functional T1 sensilla and show that the expression of the VA receptor is dramatically reduced or eliminated. Importantly, we show misexpression of this receptor in non-T1 neurons, normally insensitive to VA, confers pheromone sensitivity at physiologic concentrations. Sensitivity of T1 neurons to VA requires LUSH, an extracellular odorant-binding protein (OBP76a) present in the sensillum lymph bathing trichoid olfactory neuron dendrites. Here, we show LUSH are also required in non-T1 neurons misexpressing the receptor to respond to VA. These data provide new insight into the molecular components and neuronal basis of volatile pheromone perception.
Chemicals released by individuals to influence the behavior of other members of the same species are defined as pheromones (for review, see Dahanukar et al., 2005). Pheromones are widely used in the animal kingdom, eliciting stereotypic behaviors in both invertebrates and vertebrates (Bigiani et al., 2005; Howard and Blomquist, 2005; Vosshall, 2005). Pheromone communication is used extensively in insects, in which a variety of chemicals trigger such diverse behaviors as mating, recruitment, aggregation, and dispersal (for review, see Vander Meer et al., 1998). However, the molecular components required for pheromone sensitivity are mostly unknown.
The only volatile pheromone identified to date in Drosophila is 11-cis-vaccenyl acetate (VA). This male-specific lipid induces aggregation behavior in both male and female flies (Bartelt et al., 1985; Xu et al., 2005). VA detection is mediated by olfactory neurons located in the trichoid sensilla on the ventral–lateral surface of the antenna (Clyne et al., 1997). Trichoid sensilla fall into multiple functional groups, including T1, and T2a and T2b (Clyne et al., 1997; Couto et al., 2005; Xu et al., 2005). For simplicity, the T2 types will be referred to as “non-T1” here. T1 sensilla typically contain a single olfactory neuron that is exquisitely tuned to VA, and have spontaneous action potential firing rates of ∼1 spike per second (Clyne et al., 1997; Xu et al., 2005). T1 sensilla are clustered in the proximal portion of the trichoid sensilla expression zone (Xu et al., 2005) (see Fig. 1). Non-T1 sensilla as a group contain multiple neurons, have a combined spontaneous rate of ∼35 spikes per second, and are not influenced by VA. Non-T1 neurons are inhibited by high concentrations of various alcohols (Clyne et al., 1997; Xu et al., 2005).
Recent work indicates the molecular mechanisms underlying pheromone perception are unique. LUSH, an extracellular odorant-binding protein (OBP76a), is required for VA-induced activation of T1 neurons and pheromone-triggered behaviors (Xu et al., 2005). LUSH is expressed exclusively in trichoid sensilla (Kim et al., 1998; Shanbhag et al., 2001, 2005). LUSH is secreted by non-neuronal support cells into the fluid bathing the trichoid olfactory neuron dendrites, but is not expressed by the neurons themselves (Kim et al., 1998; Shanbhag et al., 2001). Mutants defective for LUSH lack pheromone-induced behavioral responses and T1 sensitivity. Therefore, the binding protein is required for pheromone detection.
The LUSH protein itself appears to specifically trigger action potentials in T1 neurons, producing the normal spontaneous activity observed in wild-type animals. Spontaneous activity is reduced 400-fold in lush mutants, and is restored when recombinant LUSH protein is infused directly to mutant T1 sensilla (Xu et al., 2005). No effect is observed on the spontaneous activity of non-T1 neurons. Therefore, there is something in T1 sensilla lacking in non-T1 sensilla responsible for VA sensitivity and spontaneous activity in the presence of LUSH protein. A likely candidate for such a factor is a receptor expressed by T1 neurons lacking in non-T1 neurons. In this paper, we report the identification of a receptor candidate expressed by T1 neurons required for VA sensitivity.
Materials and Methods
Identification of mutants lacking T1 sensilla.
We screened a subset of the Zuker EMS mutant collection (Koundakjian et al., 2004) for abnormal responses to VA using single sensillum electrophysiological recordings (see below). We identified two mutants that lacked T1 sensilla, tod1 (T-one-deficient) mutants (Zuker no. 23-6129) and tot1 (T-one-transformed) mutants (Zuker no. 23-4626).
Single sensillum recordings.
Extracellular electrophysiological recordings were performed according to de Bruyne et al. (2001). Flies (2–7 d of age; males and females) were under a constant stream of charcoal-filtered air (36 ml/min; 22–25°C) to prevent any potential environmental odors from inducing activity during these studies. 11-cis-Vaccenyl acetate was diluted in paraffin oil and 1 μl was applied to filter paper and inserted in a Pasteur pipette, and air was passed over the filter and presented as the stimulus. The purity of the VA was confirmed both by nuclear magnetic resonance and by mass spectroscopy. Signals were amplified 1000× (USB-IDAC System; Syntech, Hilversum, The Netherlands) and fed into a computer via a 16-bit analog-digital converter and analyzed off-line with AUTOSPIKE software (USB-IDAC System; Syntech). Low cutoff filter setting was 200 Hz, and the high cutoff was 3 kHz. Action potentials were recorded by inserting a glass electrode in the base of a sensillum. Data analysis was performed according to de Bruyne et al. (2001). Signals were recorded for 20 or 30 s, starting 10 s before VA stimulation. Action potentials were counted 1 s before VA stimulation and for 1 s after VA stimulation. All recordings were performed from separate sensilla with a maximum of two sensilla recorded from any single fly.
RT-PCRs for Or gene products were performed using Trizol for RNA purification as described by the manufacturer (Invitrogen, Carlsbad, CA). The first strand of cDNA was synthesized from 1 μg of total RNA using reverse transcriptase (Ambion, Austin, TX) at 37°C for 1 h. Amplification was performed by PCR using Or67d specific primers, 5′-TGGTCACCTAATACTCACGGCTG-3′ and 5′-CGCAAGTTGTGGAAAGTTGAACG-3′. Or83b was amplified as a control for RNA for 40 cycles using specific primers of Or83b, 5′-GCCACCAAAATCAACGGAGTG-3′ and 5′-GAAGCAAACAAATCCAGGGAGAC-3′. RT-PCR products were separated and visualized on an agarose gel and ethidium bromide staining. Identical results were obtained in three independent experiments. Primer sequences and expected products for the complete set of Or genes we tested are presented in supplemental Table 1 (available at www.jneurosci.org as supplemental material).
Scanning electron microscopic analysis.
Wild-type, tot1 and tod1 mutants were prepared for scanning electron microscopic analysis according to (Shanbhag et al., 1999). Briefly, heads of w1118 , tod1 , and tot1 flies were removed with a razor blade, and unfixed specimens were quickly coated with 20 nm gold in a sputter coater. Specimens were viewed with JSM-840A model (JEOL, Tokyo, Japan) for scanning electron microscope (10 kV). Photographs of the anterior surface of several antennas from each genotype were used to quantify the numbers of sensilla.
RNA in situ hybridization.
Expression of Or67d in control, tot1 , and tod1 mutants were preformed as described previously (Smith et al., 1990), with modification for fluorescent probes as described by Couto et al. (2005). Antisense and sense riboprobes were transcribed from Or67d and Or83b cDNAs subcloned into PCR2.1-cloning vector (Invitrogen) using T7 RNA polymerase (Ambion). Or83b and Or67d antisense probes were incorporated with digoxigenin or fluorescein and detected with anti-DIG-POD (Roche, Indianapolis, IN) and anti-fluorescein-POD (Roche). Signals were amplified with the TSA-Plus Cyanine 5 system (PerkinElmer, Wellesley, MA) for Or83b or the TSA-Plus Fluorescence system (PerkinElmer) for Or67d (Fishilevich and Vosshall, 2005).
Misexpression of Or67d.
A cDNA encoding Or67d was obtained by reverse transcription, PCR amplification of RNA isolated from antenna of Drosophila w1118 , a strain known to be responsive to VA pheromone (our unpublished observation) (see Figs. 1, 4). This strain has four amino acid polymorphisms compared with the predicted Or67d gene product in the genome sequence (Adams et al., 2000), specifically S98C, D238E, W252F, and S268P. These polymorphisms were confirmed as true genomic polymorphisms by sequencing clones from independent RT-PCRs. The cDNA was cloned into pUAST (Brand and Perrimon, 1993), and transgenic animals were generated using standard techniques. Or67d was misexpressed in non-T1 neurons by crossing the UAS Or67d transgenic flies to flies expressing Gal4 under control of the neuron-specific promoter for the embryonic lethal abnormal vision (ELAV) gene (Koushika et al., 1996).
tod1 and tot1 mutants lack T1 sensilla
With the goal of identifying all genetic loci required for VA pheromone detection, we undertook a screen to recover mutants with abnormal electrophysiological responses to VA pheromone (see Materials and Methods). We identified two mutants out of 1200 lines that lack functional T1 sensilla. tod1 and tot1 have normal basiconic and non-T1 sensilla, but no T1 sensilla. tod1 and tot1 are both recessive and fail to complement one another, revealing these mutants lack T1 sensilla as a result of lesions in independent genes. Figure 1 A shows examples of single-unit recordings from the trichoid zone from tod1 mutants, tot1 mutants, and wild-type controls. In normal flies, there is a mixture of T1 and non-T1 subtypes in the trichoid zone, but the proximal part of this zone is enriched in T1 sensilla (Clyne et al., 1997; Xu et al., 2005). Morphologically, all trichoid sensilla are indistinguishable. However, T1 and non-T1 sensilla are clearly distinguishable by electrophysiology (Clyne et al., 1997; Xu et al., 2005). In a survey of over 2000 animals, recordings from random trichoid sensilla in the proximal zone identify VA-sensitive T1 sensilla 89% of the time and non-T1 11% of the time (Fig. 1 B) based on spontaneous activity rate and sensitivity to VA. VA could evoke activity in T1 neurons from 0.35 ± 0.14 spikes per second before stimulation to 36.89 ± 3.43 after stimulation. VA did not induce activity in non-T1 neurons (48.42 ± 3.93 before to 49.33 ± 4.13 after stimulation). Recordings from either tod1 mutants or tot1 mutants from the T1 zone always identified non-T1 sensilla. These sensilla have the classic characteristics of the non-T1 type, including lack of response to VA, more than one neuron present in the sensillum, and a high rate of spontaneous activity. We surveyed trichoid and large basiconic sensilla from across the antenna in tod1 mutants and tot1 mutants and could identify no VA-responsive neurons (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). Therefore, there does not appear to be a simple mislocalization of T1 sensilla to a different part of the antenna, but a complete loss of the T1 functional type.
tod1 mutants and tot1 mutants could lack T1 sensilla because the T1 sensilla are transformed into non-T1 sensilla during development. Alternatively, the T1 sensilla may simply no longer be present, perhaps because of death of T1 precursors. If the latter is true, there should be a reduction in the total number of trichoid sensilla in these mutants correlating with the number of T1 sensilla. Therefore, we performed scanning electron microscopic analysis of wild-type, tod1 , and tot1 mutants to determine whether there are fewer trichoid sensilla compared with the wild-type controls. Figure 2 shows that tod1 mutants have ∼30% fewer (40.8 ± 2.1) trichoid sensilla compared with controls (62 ± 1.7), whereas tot1 mutants (58.4 ± 1.9) have essentially the normal number of trichoids (Fig. 2 D). One possibility is that T1 precursors fail to develop T1 sensilla in tod1 mutants, resulting in a reduction in the total number of trichoid sensilla. In tot1 , in contrast, there is an overall increase in the non-T1 sensilla class at the expense of T1 class, whereas the total number of trichoid sensilla is equivalent to that of wild-type controls. This is consistent with a transformation of T1 into non-T1 sensilla instead of loss of T1 sensilla. The number of large basiconic sensilla is not different in any of the three genotypes (Fig. 2 E). A survey of large basiconic sensilla revealed no functional abnormalities in tod1 or tot1 mutants (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). Additional characterization of these mutants will be reported elsewhere (S. Mistry, T. S. Ha, and D. P. Smith, manuscript in preparation).
tod1 and tot1 mutants are both defective for expression of one member of the odorant receptor family
tod1 and tot1 mutants lack functional T1 sensilla. Therefore, T1-neuron specific gene products should be absent in these mutants. We reasoned that a neuronal receptor mediating VA responses might be a member of the odorant receptor family expressed specifically by T1 neurons. Indeed, Ors have been shown to specify odor specificity to olfactory neurons in a number of systems, including Drosophila (Sengupta et al., 1996; Malnic et al., 1999; Dobritsa et al., 2003; Hallem et al., 2004). Therefore, we screened candidate members of the Drosophila Or gene family for reduced expression in tod1 and tot1 mutants.
PCR primers were designed to all members of the Or gene family (Robertson et al., 2003) whose function or odor specificity had not been established previously (Stortkuhl, 2001; Dobritsa et al., 2003; Elmore et al., 2003; Hallem et al., 2004). In addition, at least one gustatory receptor (Gr) is expressed in the antenna (Scott et al., 2001), and several others including Gr10a, Gr32a, and Gr63a may also be expressed (K. Scott, personal communication). We generated primers to assay expression of these orphan receptors by RT-PCR (supplemental Table 1, available at www.jneurosci.org as supplemental material). Each primer pair spanned an intron to allow us to distinguish cDNA products from potential genomic DNA contamination. We found that a single receptor gene, Or67d, was absent or drastically reduced in tod1 and tot1 mutants compared with wild-type controls. Figure 3 shows RT-PCR analysis of Or67d from wild-type, tod1 , and tot1 mutant antenna. An Or67d spliced transcript is clearly present in wild-type antennas, but is reduced or absent in tod1 and tot1 mutants, even after 40 cycles of amplification (Fig. 3 B). Or83b (Vosshall et al., 1999), expressed in most olfactory receptor neurons (Vosshall et al., 1999; Elmore et al., 2003; Larsson et al., 2004) was present in all three samples, as were all other Or genes tested (data not shown). These results suggest Or67d expression is specifically reduced or eliminated in tod1 and tot1 mutants, and correlates with the loss of the T1 functional class in these genetically distinct mutants.
Having identified a candidate receptor correlating with the presence of T1 neurons, we set out to establish whether the expression pattern of Or67d in the antenna was consistent with the known T1 neuron distribution. We performed in situ hybridization using fluorescently labeled antisense RNA probes to Or67d to characterize expression of this putative receptor. Figure 3 C shows that antisense probes to Or67d specifically label cells on the ventral–lateral surface of the third antennal segment. Serial sections reveal the labeled cells are concentrated in the proximal T1 zone (our unpublished observation). These probes failed to identify similar positive cells in antenna tod1 or tot1 (3C) consistent with the functional loss of T1 sensilla. Therefore, Or67d expression correlates well with the known distribution of T1 sensilla in wild-type antenna and the absence of T1 sensilla in the mutants.
Or67d confers VA sensitivity to non-T1 neurons
Expression of Or67d in the T1 zone and its absence in tod1 and tot1 mutants is consistent with Or67d being the T1 VA receptor, but does not prove this receptor is responsible for VA sensitivity. For example, Or67d may have some functional role specific to T1 neurons that is unrelated to VA sensitivity. Alternatively, a subset of non-T1 class olfactory neurons may also be absent in tod1 and tot1 that specifically express Or67d. Therefore, to definitively prove Or67d mediates VA sensitivity, we misexpressed Or67d in olfactory neurons that normally do not express this receptor and are VA insensitive. Previous work has shown that coexpression of an extra odorant receptor in a Drosophila olfactory neuron results in an odor sensitivity profile that is the combination of the sensitivity of the individual receptors (Goldman et al., 2005). Therefore, we misexpressed Or67d in all neurons by driving Or67d expression with the pan-neuron promoter, ELAV. ELAV is a neuron-specific splicing factor expressed in all Drosophila neurons (Campos et al., 1985; Koushika et al., 1996). We confirmed that flies expressing Or67d under control of the ELAV promoter misexpress the Or67d transgene in olfactory neurons in the antenna (Fig. 3 C, bottom panel).
We compared the VA sensitivity of wild-type animals and those misexpressing Or67d in non-T1 sensilla. Neurons in wild-type non-T1 sensilla are insensitive to VA (Fig. 4, top left panel). However, animals misexpressing Or67d in all neurons have non-T1 neurons that are highly responsive to VA (Fig. 4, top right panel). Indeed, dose–response analysis reveals these neurons are nearly as sensitive to VA as wild-type T1 neurons (Fig. 4, graph). By all other criteria, these neurons are non-T1 and not T1 neurons. They display high spontaneous activity and contain multiple neurons, and their distribution is typical of the non-T1 functional class. The only difference we can observe in these neurons compared with wild-type controls is VA sensitivity. Conferring VA sensitivity on non-T1 neurons by expressing Or67d receptors demonstrates that this receptor is both necessary and sufficient to confer VA sensitivity on non-T1 neurons.
LUSH is also required for non-T1 neurons misexpressing Or67d to respond to VA
LUSH protein is required in the T1 sensillum lymph for T1 neurons to be sensitive to VA (Xu et al., 2005). Non-T1 sensilla also express LUSH protein in the sensillum lymph (Kim et al., 1998; Shanbhag et al., 2001, 2005). Is LUSH also required for sensitivity of non-T1 neurons misexpressing Or67d? We crossed the lush1 mutation into the stock misexpressing Or67d in all neurons. Figure 4 shows that LUSH protein is critical for non-T1 neurons to respond to VA as well, because when the lush1 mutation is crossed into the misexpressing flies, VA sensitivity is lost in non-T1 neurons (Fig. 4, bottom right panel). This clearly demonstrates that LUSH is required in the extracellular space in order for non-T1 neurons misexpressing Or67d to be responsive to VA. Therefore, both the receptor Or67d and a specific extracellular binding protein, LUSH, are required for VA sensitivity.
Pheromones drive innate behaviors in insects. VA pheromone functions to attract other individuals to a location, most likely at food sources (Vander Meer et al., 1986). We previously showed the extracellular protein LUSH is required for VA pheromone sensitivity, both for T1 neuron responses and for behavioral responses to pheromone (Xu et al., 2005). In this paper, we identified an odorant receptor expressed by trichoid neurons that mediates VA pheromone sensitivity. Or67d expression is defective in two independent mutants that lack functional T1 sensilla, and this receptor is necessary and sufficient to confer VA sensitivity on non-T1 neurons that are normally insensitive to this pheromone. Together, these data provide strong proof that Or67 is the receptor, or a component of the receptor, that mediates VA sensitivity in olfactory neurons in T1 sensilla.
Although there are no Or67d mutants available to show directly that Or67d mediates VA behavior, this is almost certainly true. LUSH OBP is expressed exclusively in trichoid sensilla (Kim et al., 1998; Shanbhag et al., 2001, 2005), and lush1 mutants are completely defective for behavioral responses to VA pheromone (Xu et al., 2005). Therefore, VA-induced behavior is mediated through trichoid sensilla. Because T1 neurons are the only trichoid neurons that respond to VA (Clyne et al., 1997; Xu et al., 2005), and Or67d mediates VA sensitivity in these trichoid neurons, it is reasonable to deduce that the behavioral responses to VA pheromone are mediated by Or67d receptors. The responses of tod1 or tot1 mutants to VA should also be defective, because they both lack T1 sensilla. Unfortunately, these mutants are heavily mutagenized, and loss of VA-induced behavior is masked by the general locomotor defects of these animals.
What are the neuronal circuits that mediate VA-induced behaviors? Our identification of Or67d as a component of the neuronal VA receptor is consistent with expression studies identifying Or67d receptor expression in one class of trichoid sensillum (Couto et al., 2005). A class of trichoid sensilla was identified that expressed Or67d and contained a single neuron (Couto et al., 2005). Or67d-expressing neurons have been shown to innervate two glomeruli in the antennal lobes, DA1 and VA6 (Couto et al., 2005; Fishilevich and Vosshall, 2005). We speculate, therefore, that higher brain functions are alerted to of the presence of VA pheromone through activity in these glomeruli. Using genetically encoded fluorescent monitors of neuronal activation (Nagai et al., 2001; Pologruto et al., 2004), it should now be possible to elucidate the neuronal circuits in higher brain centers that are triggered by VA pheromone.
Why does VA pheromone detection require both a specific receptor and a binding protein? One possible explanation is specificity. Most odorant receptors in Drosophila appear to be broadly tuned (Clyne et al., 1999; Dobritsa et al., 2003; Hallem et al., 2004). Broad tuning of receptors presumably allows Drosophila to detect a wide range of biologically relevant chemicals in the environment with a limited repertoire of 62 different receptor proteins. In contrast, pheromone detection should be highly specific so that inappropriate cues are not misconstrued as pheromones. Perhaps Or67d is specifically tuned to VA but requires LUSH to transfer VA through the aqueous lymph. Alternatively, Or67d may be more broadly tuned, and LUSH OBP restricts the diversity of ligands that are presented to the receptor by only interacting with VA. A third possibility is that LUSH protein undergoes a unique conformational change after VA binding and this conformation of LUSH triggers Or67d activation. Consistent with this last model, folded, but not unfolded LUSH protein alone appears to be a partial agonist of T1 neurons (Xu et al., 2005). Additionally, x-ray crystal structures show unique conformational changes in LUSH after binding alcohols (Kruse et al., 2003) and VA (J. D. Laughlin, T. S. Ha, D. P. Smith, and D. N. M. Jones, manuscript in preparation). Now that these components have been identified, future work will focus on defining the biochemical relationship between LUSH, VA, and Or67d.
This work extends the evidence that extracellular LUSH protein is critical for VA detection. Non-T1 neurons misexpressing Or67d are only sensitive to VA when LUSH protein is present. Therefore, LUSH is required for VA detection by Or67d-expressing neurons. Detection of an unknown contact pheromone involved in courtship behavior in Drosophila is mediated by taste receptor Gr68a (Bray and Amrein, 2003). It will be of interest to determine whether a binding protein mediates this behavior as well. Artificial stress results in release of CO2 from adult fruit flies and may act like a pheromone (Suh et al., 2004). Recent work in the moth Bombyx mori has uncovered several putative sex pheromone receptors for the female sex pheromone bombykol (Krieger et al., 2005; Nakagawa et al., 2005). B. mori Or1 (BmOR1) and BmOR3 are expressed preferentially in male antenna, in neurons known to respond to female pheromone (Krieger et al., 2005). Heterologous expression of BmOR1 in Xenopus laevis oocytes confers sensitivity to bombykol at high pheromone concentrations (Nakagawa et al., 2005). The poor sensitivity of moth pheromone receptors expressed in oocytes may result from relatively poor expression or from missing factors that contribute to sensitivity (like extracellular pheromone binding proteins). Misexpression of Drosophila Ors in “empty” olfactory receptor neurons (lacking endogenous receptors) has been shown to confer odor response profiles that are very similar to the endogenous neurons expressing those receptors, indicating this is a more physiologic approach (Hallem et al., 2004). Using a similar approach, we conferred VA sensitivity on non-T1 neurons at pheromone concentrations that are close to the threshold levels that activate T1 neurons. This supports the idea that Or67d in combination with LUSH protein confers sensitivity to physiological relevant concentrations of VA in vivo.
Defining the molecular basis for VA pheromone perception in Drosophila provides an important step in understanding the molecular basis of pheromone-induced behavior in an animal model system amenable to genetic analysis. If pheromone detection mechanisms in Drosophila prove similar to those in other insects, this information may have practical application to control of insect pests and disease vectors in the future.
This work was supported by National Institutes of Health Grant R01-DC02539. We thank PingXi Xu for assistance with Or primers, Linda Hannigan for help with electron microscopy, and Charles Zuker and members of his laboratory for access to their mutant collection.
- Correspondence should be addressed to Dean P. Smith, Department of Pharmacology and Center for Basic Neuroscience, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390-9111.