Candidate Taste Receptors in Drosophila

See allHide authors and affiliations

Science  10 Mar 2000:
Vol. 287, Issue 5459, pp. 1830-1834
DOI: 10.1126/science.287.5459.1830


Little is known about the molecular mechanisms of taste perception in animals, particularly the initial events of taste signaling. A large and diverse family of seven transmembrane domain proteins was identified from the Drosophila genome database with a computer algorithm that identifies proteins on the basis of structure. Eighteen of 19 genes examined were expressed in theDrosophila labellum, a gustatory organ of the proboscis. Expression was not detected in a variety of other tissues. The genes were not expressed in the labellum of a Drosophila mutant,pox-neuro70 , in which taste neurons are eliminated. Tissue specificity of expression of these genes, along with their structural similarity, supports the possibility that the family encodes a large and divergent family of taste receptors.

Although two putative mammalian taste receptors have recently been described (1), remarkably little is understood in general about taste receptors across species. A computer algorithm that seeks proteins with particular structural properties, as opposed to proteins with particular sequences, identified a large family of candidate odorant receptors from theDrosophila genomic database (2). Here, we report that further analysis of genes identified by this algorithm revealed one gene that defines a distinct large family of membrane proteins (3); 43 members of this family have been identified in the first 60% of the Drosophila genome that has been sequenced thus far (3). If the sequenced part of the genome is representative, then extrapolation suggests that the entire genome would encode on the order of 75 proteins, a figure comparable to our estimate of ∼100 candidate odorant receptors (2). The previously unidentified family of proteins shows no sequence similarities to any known odorant receptors or to any other known proteins. We have tentatively named this the gustatory receptor (GR) family, with each individual gene named according to its cytogenetic location in the genome. Thus, the GR59D.1 and GR59D.2 genes, which we abbreviate here as 59D.1 and 59D.2, refer to two family members located in cytogenetic region 59D on the second chromosome.

The amino acid sequences of 19 members of the GR family indicate the high degree of sequence divergence (Fig. 1). Sequence alignment revealed only one residue conserved among all members of the family shown and only 24 residues conserved among more than half of the genes shown. Fifteen of these conserved residues lie in the vicinity of the COOH-terminus. Amino acid identity between individual genes ranged from a maximum of 34% to <10%. By contrast, other features of the gene family show substantial conservation. The positions of a number of introns are conserved (Fig. 1), suggesting that the family originated from a common ancestral gene. Overall sequence length, ∼380 amino acids, is another common feature. All of the genes encode approximately seven predicted transmembrane domains, a feature characteristic of G protein–coupled receptors (GPCRs) (Fig. 2). All 43 of the predicted GR gene products were identified as GPCRs by an algorithm trained to distinguish between GPCRs and other multitransmembrane proteins (2, 4).

Figure 1

Amino acid sequence alignment of 19 GR proteins (25). Letters following protein designations identify alternative splicing products of individual genes. Residues conserved in >50% of the predicted proteins are shaded. The approximate locations of the seven predicted transmembrane domains are indicated. Intron-exon boundaries are shown with vertical lines. The sequences shown are the first 19 full-length proteins we identified. All DNA sequences are from the BDGP database (3).

Figure 2

Representative hydropathy plots of GR proteins. Hydrophobic peaks predicted by Kyte-Doolittle analysis appear above the center lines. The approximate positions of the seven putative transmembrane domains are indicated above the first hydropathy plot. Similar plots were obtained for all of the GR proteins.

The genes are widely dispersed in the genome, but at the same time, many are found in clusters. The two largest clusters each contain four genes; there are also several clusters of two or three genes. Genes within these clusters are closely spaced, with intergenic distances ranging from 150 to 450 base pairs (bp) in all cases for which the data are currently available. There is no rule specifying the orientation of genes within clusters, unlike the case of theDrosophila odorant receptors, in which genes within a cluster are in the same orientation in all clusters examined (2).

An unusual form of alternative splicing occurs in at least two chromosomal locations. Four large exons in cytogenetic region 39D each contain sequences specifying six predicted transmembrane domains, followed by three small exons that together specify a putative seventh transmembrane domain and the COOH-terminus (Fig. 3). Reverse transcription–polymerase chain reaction (RT-PCR) analysis revealed that each of the four large exons is spliced to the smaller exons, thereby generating four predicted seven transmembrane domain proteins. These four proteins are thus distinct through the first six transmembrane domains and identical in the seventh and in the COOH-terminal sequences. Likewise, in cytogenetic region 23A, there are two large exons, each of which specifies six transmembrane domains and is spliced to two small exons that together encode a seventh transmembrane domain and the COOH-terminus (Fig. 3). Thus, the gene in region 23A encodes two related proteins. This pattern of splicing, in which alternative large 5′ exons encoding most of the protein are joined to common short 3′ exons encoding only a small portion of the protein, is unusual among genes encoding GPCRs and proteins in general. This pattern of splicing provides a mechanism at a single locus for generating products that exhibit a pattern observed for this family in general: extreme diversity among all sequences of the proteins except in a small region in the vicinity of the COOH-terminus.

Figure 3

Genomic organization of the 39D.2 and 23A.1 loci. In the 39D.2 locus, the gray boxes labeled a through d represent four large 5′ exons, each of which can be spliced individually to the three 3′ exons (indicated in black) to produce alternative transcripts encoding four different proteins. All the exons of the 39D.2 locus are located in an intron of another gene, which is in the opposite orientation and whose exons are represented by white boxes. This other gene appears to encode a basic helix-loop-helix transcription factor expressed during embryogenesis. In the 23A.1 locus, the gray boxes labeled a and b represent two alternative large 5′ exons, either of which can be spliced to the two small 3′ exons (indicated in black) to produce transcripts encoding two different proteins.

To assess the tissue specificity of expression, we performed RT-PCR with primers that span introns in the coding regions. Of the 19 transcripts tested, 18 were expressed in the labellum (Fig. 4 and Table 1), the major gustatory organ of the fly (5–8). Moreover, for most of these genes, expression was labellum-specific in that only 1 of the 19 yielded amplification products from heads depleted of taste organs and only 2 showed expression in the thorax, which contains the thoracic nervous system but no characterized taste sensilla. Likewise, expression in several other tissues, including the abdomen, wings, and legs, was limited to a small fraction of genes (Table 1).

Figure 4

Tissue specificity of expression of 32D.1 in the labellum. Shown is a gel photograph of an RT-PCR experiment with primers spanning an intron in 32D.1. The size of the predicted PCR product from cDNA is 372 bp; any remaining genomic DNA would generate a product of 559 bp. A cDNA band is observed in the labellum lane only. In addition, 32D.1 is not expressed in the labellum of thepoxn70 mutant. Positive controls are described in (26).

Table 1

Tissue-specific expression of GR genes. RT-PCR was performed from RNA extracted from the indicated tissues (26). All primer pairs spanned introns. Positive controls are described in (26).

View this table:

To further analyze gene expression by in situ hybridization, we used 12 GR transcripts as probes. Each probe was used individually and in mixtures of multiple probes. Sequences encompassing all, or nearly all, of each transcript were used, and several diverse methods of signal amplification and detection were used, with a variety of experimental conditions (9). None of the genes showed detectable expression in any tissue, including the taste organs. As positive controls, the pheromone-binding protein-related protein-2gene (pbprp-2), which may encode a carrier of hydrophobic molecules (10), showed hybridization in taste sensilla on the labellum, and the Drosophila olfactory receptor gene 22A.2 (DOR22A.2) (2) hybridized to olfactory sensilla on the antenna. The simplest interpretation of these results is that expression levels of the GR genes are exceedingly low. Consistent with this interpretation is the fact that no expressed sequence tags have been identified for any of the 43 GR transcripts.

To further analyze the tissue specificity of GR expression, we performed a microdissection experiment in which the labral sense organ (LSO) (7, 8), a small taste organ that lines the pharynx, was surgically excised from each of 50 animals. The LSO consists of a very limited number of cells and is highly enriched in taste neurons; it does not, for example, contain muscle cells. By RT-PCR amplification, we detected the expression of seven GR transcripts in this taste organ (Fig. 5). These results indicate that expression of the GR family extends to include at least one additional taste organ besides the labellum. The data are also fully consistent with the notion that the GR genes are expressed in taste neurons.

Figure 5

GR gene expression in microdissected labral sense organs (LSOs). The red areas show the four major taste organs of the Drosophila head: the LSO, the dorsal cibarial sense organ (DCSO), the ventral cibarial sense organ (VCSO), and the labellum. The gel track shows an amplification product from RNA extracted from 50 LSOs, amplified with primers N23A.3J and N23A.2D from two exons of gene 23A.1. Specifically, one primer is from the large exon 23A.1a (Fig. 3), and the other is from the first common exon at the 3′ end. The amplification product is 430 bp, which is the expected length for a cDNA product; any remaining genomic DNA would generate a product of 1598 bp. The primer pair did not amplify a product from nongustatory tissue (Table 1). The following transcripts were detected in the LSO: 22B.1, 23A.1a, 23A.1b, 32D.1, 39D.2c, 43C.1, and 58A.2.

To confirm the gene expression in taste receptor neurons, we used aDrosophila mutant, pox-neuro70 (poxn70 ), in which chemosensory bristles are transformed into mechanosensory bristles (11–14). Specifically, inpoxn70 , which behaves as a null mutation with respect to adult chemosensory organs, chemosensory bristles are transformed into mechanosensory bristles with respect to various morphological and developmental criteria. In particular, most chemosensory bristles in wild-type Drosophila are innervated by five neurons: four chemosensory neurons and one mechanosensory neuron. In contrast, wild-type mechanosensory bristles contain a single mechanosensory neuron. In chemosensory bristles transformed to mechanosensory bristles by poxn70 (11), the number of neurons is reduced from five to one. We predicted that if the GR family is in fact expressed in the chemosensory neurons of taste sensilla, their expression would likely be eliminated in the poxn70 mutant. Consistent with this prediction, 18 of 19 GR transcripts examined were not expressed in the labellum of the poxn70 mutant (Table 1 and Fig. 4). These results indicate that the GR gene family is expressed in labellar chemosensory neurons.

The large size of this protein family likely reflects the diversity of compounds that flies can detect. The labellar hairs of larger flies are not only sensitive to a variety of simple and compound sugars (15), but also to a wide variety of other molecules, such as amino acids (16). Behavioral studies have shown thatDrosophila are sensitive to quinine (17), which is perceived by humans as bitter, and other insects have been shown to be sensitive to an array of structurally diverse bitter compounds. Moreover, an individual insect taste receptor cell can respond to a broad range of structurally heterogeneous alkaloids and other bitter molecules (18, 19). The extreme diversity of these receptors may not only reflect diversity among the ligands that they bind, but also diversity in the signal transduction components with which they interact. For example, the lack of conserved intracellular regions suggests the possibility that, during the evolution of this sensory modality, multiple G proteins arose, each interacting with a different subset of receptors. Finally, it seems likely that the Drosophila genome encodes taste receptors in addition to those of the GR family. Although we have detected expression in the labellum and the LSO, few if any family members are expressed in the leg or wing chemosensory hairs (Table 1), some of which are morphologically similar to labellar taste hairs (7). The Drosophila olfactory system also contains more than one organ, the antenna and maxillary palp, which respond to all, or nearly all, of the same odorants and which derive from the same imaginal discs (20). However, most individual members of the DOR gene family are expressed in one or the other but not in both olfactory organs (2, 21). Perhaps the distinction among taste receptor genes is even more extreme in the gustatory system, whose organs derive from different imaginal discs. For example, the legs may express a completely distinct family of genes or a subfamily whose similarities to the present family are sufficiently tenuous as to place it slightly beyond the boundaries that define the GR family.

  • * These authors contributed equally to this work.

  • Present address: Department of Biochemistry and Biophysics, University of California at San Francisco, Parnassus Avenue, Box 0448, San Francisco, CA 94143, USA.

  • To whom correspondence should be addressed. E-mail: john.carlson{at}


View Abstract

Navigate This Article