Identification of RL-TGR, a coreceptor involved in aversive chemical signaling

Abstract

Chemical signaling plays an important role in predator–prey interactions and feeding dynamics. Like other organisms that are sessile or slow moving, some marine sponges contain aversive compounds that defend these organisms from predation. We sought to identify and characterize a fish chemoreceptor that detects one of these compounds. Using expression cloning in Xenopus oocytes coexpressing the cystic fibrosis transmembrane conductance regulator (CFTR) chloride channel, the beta-2 adrenergic receptor (β₂AR), and fractions of a zebrafish cDNA library, we isolated a cDNA clone encoding receptor activity–modifying protein (RAMP)-like triterpene glycoside receptor (RL-TGR), a novel coreceptor involved in signaling in response to triterpene glycosides. This coreceptor appears to be structurally and functionally related to RAMPs, a family of coreceptors that physically associate with and modify the activity of G protein–coupled receptors (GPCRs). In membranes from formoside-responsive oocytes, RL-TGR was immunoprecipitated in an apparent complex with β₂AR. In HEK293 cells, coexpression of β₂AR induced the trafficking of RL-TGR from the cytoplasm to the plasma membrane. These results suggest that RL-TGR in the predatory fish physically associates with the β₂AR or another, more physiologically relevant GPCR and modifies its pharmacology to respond to triterpene glycosides found in sponges that serve as a potential food source for the fish. RL-TGR forms a coreceptor that responds to a chemical defense compound in the marine environment, and its discovery might lead the way to the identification of other receptors that mediate chemical defense signaling.

Sessile or slow-moving organisms, especially those in marine systems, commonly use chemical cues as a means of defense against predation (1, 2), but their mechanisms of action on potential predators are not well understood (3, 4). The cellular effects of only a few deterrent compounds in marine sponges have been investigated (5–7). Sheybani et al. (4) showed that defense compounds from the sea hare Aplysia californica were detected by the olfactory and gustatory systems of sea catfish, resulting in an electrophysiological response. Thus, deterrent compounds likely activate chemoreceptors, causing a cascade of events leading to the aversive behavioral response. However, despite the growing number of known marine chemical defenses (2), no receptor molecule that responds to these aversive compounds has yet been identified.

Genetic analysis of the fish gustatory system has shown that fish have receptor molecules homologous to those of the mammalian system (8), suggesting that the receptors are conserved as a mechanism through which an organism can distinguish between foods that are nutritious and those that are potentially harmful. Some organisms have other mechanisms for identifying harmful substances, such as the nociceptor pathway; for example, binding of capsaicin (a compound found in chili peppers) to its receptor results in the perception of heat and intense pain in mammals (9).

Little is known about which classes of receptors respond to chemical deterrents (3). Receptors involved in defensive signaling may be built as ion channels, G protein–coupled receptors (GPCRs), or accessory coreceptors that work in combination with another receptor. A few coreceptors that facilitate cell surface expression (10–12) and, in some cases, affect the pharmacology of associated receptors (13, 14) have been identified. For example, receptor activity–modifying proteins (RAMPs) affect the pharmacology of their GPCR partners (15). Despite the relatively low sequence similarity (∼30%) within this protein family (16), RAMPs have three common structural characteristics: a single predicted membrane-spanning domain, a short cytoplasmic domain, and a long extracellular domain (14). Coexpression of certain RAMPs with the calcitonin receptor (16) or other class B (17) and class C GPCRs (18) allows the formation of complexes of these membrane-associated receptors and produces novel binding sites for ligands, leading to unique signaling responses that are not present in cells that express either protein alone (16).

We previously demonstrated that deterrent compounds isolated from marine sponges are unpalatable to zebrafish (19). Furthermore, Xenopus oocytes expressing a zebrafish cDNA library, the cystic fibrosis transmembrane conductance regulator (CFTR), and the beta-2 adrenergic receptor (β₂AR) exhibited a receptor-mediated electrophysiological response to formoside, a triterpene glycoside deterrent found in the sponge Erylus formosus (19, 20). Using CFTR chloride channels to monitor activation of β₂AR, we found that the kinetics of the formoside-mediated response in oocytes expressing the cDNA library suggest that the zebrafish genome encodes a protein whose activation by formoside causes a G protein–signaling cascade culminating in activation of CFTR. We report here the functional isolation of a cDNA encoding RAMP-like triterpene glycoside receptor (RL-TGR), a protein that responds to formoside, along with its initial characterization. RL-TGR appears to be related to RAMP proteins and might function as a coreceptor underlying the aversive behavioral response to this chemical deterrent.

Results

Identification from a Zebrafish cDNA Library of a Clone Enabling a Response to Formoside.

As shown previously (19), oocytes that coexpress a whole zebrafish cDNA expression library along with CFTR and β₂AR exhibit an electrophysiological response to the application of formoside. In the present work, fractions of this cDNA library were transcribed into cRNA, microinjected into oocytes, and tested for an electrophysiological response (Fig. S1). Oocytes expressing fraction A responded to an application of 5 μM formoside (Fig. 1A) in a manner similar to that demonstrated by the entire library (19). Using expression cloning, we isolated a single clone (clone A9-f4-230) that enabled a functional response in oocytes similar to the formoside-mediated response of oocytes expressing the entire zebrafish cDNA library (Fig. 1B).

Fig. 1.

Identification and initial characterization of clone encoding a coreceptor for aversive compounds. (A and B) Traces in response to 1 μM isoproterenol (I) or 5 μM formoside (F) from oocytes coexpressing CFTR, β₂AR, and either cDNA library fraction A (n = 10; formoside response range, 0.05–0.2 μA) (A) or isolated full-length clone A9-f4-230 (n > 40; range, 0.1–3.7 μA) (B) (Scale bars: A, 0.5 μA, 2 min; A, Inset, 0.2 μA, 2 min; B, 0.5 μA, 2 min.) The inset in A shows a view of the same trace expanded in the amplitude dimension. (C and D) Electrophysiological responses to various compounds in oocytes expressing CFTR, β₂AR, and the full-length clone. (C) A mixture of ectyoplasides A and B (E) caused a response comparable to a formoside-induced response (n = 5; range, 0.1–0.4 μA) (Scale bar: 0.5 μA, 1 min.) (D) The ligand specificity of RL-TGR was investigated by exposing formoside-responsive oocytes to a variety of other compounds, including ectyoplasides A and B (10 μM), ouabain (5 μM), digoxin (5 μM), 17β-estradiol (5 μM), octanal (0.5 mM), cycloheximide (1.5 μM), and capsaicin (50 μM). Responses to these ligands were normalized to the response to 5 μM formoside in the same cell (n = 3–10). Error bars represent SD. *P < 0.05 compared with formoside in the same cell. (E) RT-PCR analysis of zebrafish tissues identified transcript in the head (H) and trunk (T). (F) RT-PCR analysis of bluehead wrasse head tissue.

To explore whether the active clone might encode a generalized receptor for chemical deterrents, we tested the specificity of this receptor by assaying other aversive compounds (Fig. 1 C and D). Oocytes expressing the full-length clone, β₂AR, and CFTR also responded to a mixture of ectyoplasides A and B, defensive triterpene glycoside compounds found in the marine sponge Ectyoplasia ferox. These cells did not respond to cycloheximide, which is perceived as bitter by humans, or to the odorant octanal. Formoside's activation of RL-TGR is specific to terpene glycoside chemical defenses; aversive compounds produced by nonterpenoid biosynthetic pathways, such as capsaicin and sceptrin (see ref. 19), did not affect this receptor. Not all terpene glycosides stimulated RL-TGR, however; ouabain and digoxin, plant cardiac glycosides derived from the terpenoid pathway, showed no effect. The ligand specificity of RL-TGR is not absolute; both the formoside and ectyoplaside groups of sponge chemical defenses activated this receptor. These compounds have similar, but not identical, triterpene carbon skeletons based on penasterol, differing in the presence of one methyl group and hydroxylation of the triterpene core. Formoside and ectyoplasides also differ in their degree of glycosylation; formoside is a tetrasaccharide [ara-(gal)-ara-gal], whereas the ectyoplaside triterpene is linked to three saccharides (gal-ara-gal). The minor activation of RL-TGR observed with 17β-estradiol, a highly derived nonglycosylated steroid that shares a triterpene biosynthetic precursor with sponge triterpene glycosides and plant cardiac glycosides, might have resulted from a different mode of receptor activation.

Receptor Gene Is Expressed in Fish Heads.

Sequencing of the 1,199-bp insert of the full-length clone followed by BLAT/BLAST analysis revealed that the RL-TGR gene is located on zebrafish chromosome 12:3886126–3887324. Furthermore, the cDNA is identical to the genomic DNA, suggesting that the gene is intronless. For the encoded protein to be involved in a gustatory response to aversive compounds in potential prey, it would need to be expressed in the head. Consistent with this notion, RT-PCR analysis from zebrafish mRNA showed evidence of expression in the head, as well as in the trunk, suggesting that the gene also may be expressed in chemoreceptor cells on the body surface or internal organs (Fig. 1E). The functional clone from zebrafish was identified, taking advantage of the availability of high-quality cDNA libraries for this model genetic organism. However, because zebrafish are freshwater animals, it was important to ask whether RL-TGR is expressed in marine fish that co-occur with the marine sponge that contains formoside. As shown in Fig. 1F, using RT-PCR with primers designed to amplify a segment within the putative ORF of the functional clone, we also detected a transcript in RNA from heads of the bluehead wrasse (Thalassoma bifasciatum), a marine fish that co-occurs with the sponge E. formosus, a source of formoside (20). Comparison of the DNA sequence from the RT-PCR products from bluehead wrasse and the corresponding region of clone A9-f4-230 indicated 100% identity between these two divergent fishes (Fig. S2). Furthermore, a 41- bp region within this segment of the active clone is almost completely conserved in fish; it aligned to chromosomal DNA sequences from five fish species: zebrafish, stickleback, medaka, Tetraodon, and Fugu rubripes (Fig. S2). This same 41-bp segment is found only in fish species and is not found in any mammalian, reptilian, or amphibian genomes available through the National Center for Biotechnology Information (NCBI) and University of California Santa Cruz (UCSC), suggesting that it is a previously undescribed fish gene.

Predicted Protein Has Structural Similarity to RAMP Proteins.

We hypothesized that the formoside-responsive receptor is encoded by a 291-bp segment that encompasses the longest ORF within the active clone, even though a strong Kozak sequence is not apparent. To test this hypothesis, we subcloned the ORF into the pET-52b(+) vector to generate protein as a double-tagged fusion peptide (Fig. 2A). Oocytes injected with a cRNA transcript from this clone, CFTR, and β₂AR responded to formoside, confirming both the ORF and the translation frame (Fig. 2A). A BLAST analysis of the predicted peptide sequence found no homologs, indicating that clone A9-f4-230 encodes a previously undescribed protein. Surprisingly, the molecular weight of the native protein is predicted to be only ∼10 kDa, compared with ∼17 kDa for the tagged receptor. To explore whether formoside-responsive oocytes express the tagged receptor, we immunoprecipitated His-tagged protein from responsive oocytes. Subsequent immunoblot analysis against the Strep tag II showed bands at ∼34, ∼60, and ∼111 kDa (Fig. 2B). The band at ∼60 kDa detected in all lanes is from a nonspecific artifact. The band at ∼34 kDa likely represents homodimers of the 17-kDa tagged receptor. The band at ∼111 kDa likely represents a complex of proteins, possibly including the tagged receptor and β₂AR (∼47 kDa). These bands were detected in the membrane fraction, suggesting that the formoside receptor is membrane-associated. In addition, the 34- and 111-kDa bands were not present in the membrane fractions from uninjected oocytes or oocytes expressing only β₂AR and CFTR, indicating that they are specific to formoside-responsive oocytes. These results also indicate that oocytes injected with transcripts for the double-tagged protein, CFTR, and β₂AR produced full-length tagged receptor, given that these specific bands were detected only after immunoprecipitation for one tag and immunoblot analysis for the other tag.

Fig. 2.

RL-TGR is responsible for the formoside-induced response. The largest ORF from the active clone was expressed as a Strep II/His–tagged fusion protein. (A) Representative response after the application of 1 μM isoproterenol (I) and 5 μM formoside (F) to oocytes expressing CFTR, β₂AR, and Strep II (S)/His (H)–tagged receptor fusion protein (see Inset) (n = 19; range, 0.05–2.2 μA). (Scale bar: 0.4 μA, 2 min.) (B) Western blot analysis showing Strep II/His–tagged receptor in formoside-responsive oocytes also expressing CFTR and β₂AR, immunoprecipitated with anti-His antibody and immunoblotted against Strep II. (C) Predicted peptide sequence of RL-TGR. Filled boxes denote extracellular cysteines. The predicted transmembrane region is underlined. Bracketed residues are encoded by the conserved DNA segment. (D) Transmembrane prediction plot (Left) and schematic of RL-TGR (Right). Red indicates the probability of being intracellular, blue indicates the probability of being extracellular, and gray bars represent transmembrane probability. Black-filled circles in (C) and (D) represent the putative PDZ-binding domain. (E) Confirmation of membrane topology. RL-TGR was expressed in HEK293 cells with a C-terminal His-tag (RL-TGR–His). Except where noted, immunofluorescence was performed in nonpermeabilized cells. (Panel 1) Cells expressing RL-TGR–His, probed for His tag. (Panel 2) Cells expressing RL-TGR–His, probed for actin. (Panel 3) To visualize actin, cells expressing RL-TGR–His were permeabilized before immunostaining for actin. (Panel 4) Cells transfected with empty vector, probed for His tag. (Panel 5) Cells expressing RL-TGR–His, probed with secondary antibody only. In each panel, the inset shows nuclear staining with DAPI in the same microscopic field as the main image.

The predicted primary sequence of the protein encoded by the active clone is shown in Fig. 2C. A transmembrane prediction program (TMHMM Server v. 2.0) predicted a single-pass transmembrane domain (Fig. 2D). The topology of the predicted protein, with a short intracellular amino terminal domain and a long extracellular carboxyl-terminal domain, is similar to that of RAMPs, which act as accessory proteins to many GPCRs (14, 21). However, the carboxy and amino terminals are oppositely oriented from all mammalian RAMPs described thus far; if other RAMP-like proteins also play roles in chemoreception, it will be interesting to explore whether they also exhibit inverted topology. Given the apparent similarity to RAMPs, we call this protein RAMP-like triterpene glycoside receptor (RL-TGR). All known RAMPs require physical interaction with a true receptor; consistent with this, RL-TGR has features suggesting its involvement in protein–protein interaction. A PDZ-binding domain, found in some RAMPs (22, 23), is predicted to be located in the cytoplasmic tail of RL-TGR and likely helps anchor it into a plasma membrane complex (Fig. 2D). The extracellular domain of RL-TGR has four cysteines, which also may be involved in protein–protein interactions, as suggested by the presence of extracellular cysteines in most known RAMPs.

To confirm the predicted topology of RL-TGR, we expressed the protein with a carboxyl-terminal His tag in HEK293 cells (Fig. 2E). In nonpermeabilized cells, the His tag was accessible to binding of an anti-His antibody. As a control, we show that under identical conditions, an antibody against actin could not reach that protein's intracellular location, confirming that the membrane was not permeabilized. Anti-actin staining was observed when the plasma membrane was permeabilized with 0.1% Triton X-100.

The observation that RL-TGR is expressed in the trunk as well as the head (Fig. 1E) suggests that RL-TGR also may be involved in chemoreception outside of the mouth, on the body surface or in organs known to express chemoreceptors, such as the gut (24), or that it might have additional roles in nonchemoreceptive tissues, similar to the various roles of known RAMP proteins (14).

RL-TGR Is Not a Ligand-Gated Ion Channel.

Oocytes expressing the full-length clone and β₂AR, but not CFTR, did not respond to formoside or isoproterenol, a β₂AR agonist (Fig. 3A), suggesting that the receptor itself did not directly cause the change in current and is not a ligand-gated ion channel. In cells expressing the full-length clone, β₂AR, and CFTR, reversal potentials of isoproterenol-activated CFTR current and formoside-activated current were the same [−29.1 ± 3.3 mV (n = 5) and −28.6 ± 5.3 mV (n = 7) respectively] (Fig. 3B), indicating that both responses reflect increased CFTR channel activity. RL-TGR responded to multiple applications of formoside with diminishing amplitude, not unlike the β₂AR-mediated responses to isoproterenol (Fig. 3 C and D). Taken together, these data suggest that formoside activates a receptor-mediated signaling cascade that culminates in the activation of CFTR.

Fig. 3.

Formoside induces receptor-mediated activation of CFTR via interaction with a GPCR. (A) Oocytes expressing the full-length clone and β₂AR, but not CFTR, did not respond to 5 μM formoside (F) or 1 μM isoproterenol (I) (n = 5). (Scale bar: 0.5 μA, 1 min.) (B) Current-voltage plot for isoproterenol-induced (red) and formoside-induced (blue) responses in oocytes expressing CFTR, β₂AR, and the full-length clone. (C and D) Multiple applications of 5 μM formoside to oocytes expressing CFTR, β₂AR, and clone A9-f4-230 (C) caused repeatable electrophysiological responses similar to the receptor-mediated responses to 1 μM isoproterenol (D) (n = 5) [Scale bars: C, 1.0 μA, 2 min; D, 0.5 μA, 2 min.) (E and F) Response to formoside requires a functional interaction between RL-TGR and a GPCR. Only cells expressing either β₂AR (E) (n = 5) or the rat aldehyde receptor OR-I7 (F) (n = 7) along with RL-TGR and CFTR responded robustly to formoside. Representative traces are provided in Fig. S3. Error bars represent SE. (G) Indirect immunofluorescence of heterologously expressed Strep II–RL-TGR with (Left) and without (Middle) β₂AR in HEK293 cells probed with anti–Strep II (green). Nuclei were stained with DAPI (blue). (Right) A confocal overlay image of Strep II–RL-TGR (green) in HEK293 cells. Individual image panels are shown in Fig. S4.

RL-TGR Requires Coexpression of a GPCR to Respond to Formoside.

We hypothesized that if RL-TGR functions similarly to RAMPs, then it would require coexpression of a GPCR to respond to formoside. Indeed, robust electrophysiological responses to formoside in oocytes required coexpression of the full-length clone plus CFTR along with β₂AR; cells not coexpressing an exogenous GPCR exhibited a minimal response to formoside (Fig. 3E and Fig. S3A). Moreover, cells that were capable of responding to formoside lost this ability in the presence of 0.5 μM propranolol, an inverse agonist against β₂AR (n = 5) (Fig. S3B). These results suggest that RL-TGR functions as a coreceptor that may bind formoside directly, but requires interaction with a true GPCR to achieve signaling. This notion is consistent with the observation of only a minimal intracellular domain, likely too small to enable signaling on its own, in the RL-TGR clone.

Because it is unlikely that β₂AR is expressed in every cell in the zebrafish gustatory system, we asked whether other G_αS-coupled GPCRs also might interact with RL-TGR to form a functional formoside receptor. Interestingly, we also found a robust response to formoside in cells expressing CFTR, RL-TGR, and the rat aldehyde receptor OR-I7 (Fig. 3F and Fig. S3C). Furthermore, indirect immunofluorescence studies of HEK293 cells transfected to overexpress RL-TGR with and without β₂AR suggested that RL-TGR is localized to the plasma membrane efficiently only when β₂AR is also overexpressed above the endogenous level in these cells (Fig. 3G and Fig. S4). This indicates that a GPCR is involved in both the trafficking of RL-TGR and the functional response to ligand, further supporting our hypothesis that RL-TGR is a RAMP-like coreceptor that forms a signaling complex with a GPCR to detect formoside.

Discussion

This study presents the functional identification and initial characterization of RL-TGR, an accessory protein that responds to an aversive chemical defense molecule found in potential prey of fish. Despite the low amino acid sequence homology between RL-TGR and members of the RAMP family, this 10-kDa protein has both structural and functional similarities to this class of proteins. The trafficking of RL-TGR and subsequent signaling in response to ligand requires interaction with GPCRs such as β₂AR or OR-I7 (Fig. 4). These functional data, combined with the structural parallels to known RAMPs, support our hypothesis that RL-TGR is an accessory protein related to the RAMP family. The oppositely oriented structural topology of RL-TGR compared with mammalian RAMPs suggests that these coreceptors might have evolved separately through convergent evolution.

Fig. 4.

Proposed schematic of the coreceptor/GPCR complex. We hypothesize that RL-TGR, like known RAMPs, forms a complex with a GPCR to cooperatively bind ligand. The ligand-bound complex activates a signaling cascade through the GPCR's cognate G protein, resulting in the activation of signaling pathways that regulate ion channels, leading to the aversive behavior.

RAMP proteins are a family of accessory proteins that affect the localization and pharmacology of GPCRs (14, 21). They are expressed fairly ubiquitously across tissues, suggesting that they play diverse roles, including some that remain undefined (25). In a unique mechanism, these single-pass transmembrane receptors act as chaperones that associate with GPCR families to bind ligands that these GPCRs cannot bind alone, producing non-native signaling responses (14, 15). Like known RAMP proteins (26), the extracellular tail of RL-TGR has several cysteines that might be involved in protein–protein interactions with the extracellular tail of a GPCR, in addition to a possible PDZ- binding domain on the cytoplasmic N-terminal tail that might help anchor the transmembrane protein to the plasma membrane in a macromolecular signaling complex via interactions with scaffolding proteins (27). Although a PDZ- binding domain that is not at the extreme C terminus is uncommon, there are examples of these noncanonical binding domains (27), such as the domains involved in the well-studied interaction between neuronal nitric oxide synthase and syntrophin (28).

Because RL-TGR is a small RAMP-like protein, it likely does not have direct signaling capabilities. Instead, the electrophysiological response might occur because the extracellular tail of this accessory protein facilitates binding of triterpene glycosides in cooperation with a GPCR (in the case of our experimental approach, either β₂AR or OR-I7), causing a conformational change in this GPCR, resulting in the activation of its cognate G protein (Fig. 4). Thus, triterpene glycoside–mediated signaling via RL-TGR takes advantage of the G protein activation mechanism provided by the associated GPCR. Although for the experiments reported here we used β₂AR and OR-I7 as representative G_αS-coupled GPCRs, these receptors likely do not coexpress with RL-TGR endogenously in zebrafish and do not form a receptor complex that responds to triterpene glycososides. In fact, an endogenous GPCR may have a higher affinity for interacting with RL-TGR, allowing for more robust responses to triterpene glycosides than we were able to detect in our experiments. The identity of an endogenous GPCR whose activity is modified by RL-TGR remains unclear, and this merits further investigation. It will be interesting to examine whether RL-TGR is expressed in olfactory or gustatory cells along with GPCRs that are known to function in chemical sensing (3, 29). Furthermore, future studies should investigate whether an RL-TGR/GPCR complex underlies the aversive behavioral response in zebrafish, as well as in fish that co-occur with marine sponges containing chemical defense compounds. Triterpene glycosides are found in a number of marine and terrestrial organisms as putative defenses (30–32); therefore, there likely is a conserved mechanism for detecting these compounds among predators and herbivores that encounter these potential food items.

It is interesting to note that RAMPs seem to be common GPCR regulators in numerous tissue types and are found in many organisms (14, 16, 25, 33), suggesting that they have a conserved purpose. Furthermore, a single RAMP has the capability to detect multiple types of ligands with substantial specificity, depending on the GPCR with which it is associated at any given time (15). Such a mechanism for detecting harmful compounds would be evolutionarily advantageous, because an organism would not need to generate a specific full-complement receptor with both ligand-binding and signaling capabilities for every possible compound with which it might come in contact. Using this flexible signaling mechanism, a vast number of specific compounds could be detected with a small number of true GPCRs, which physically combine with an RL-TGR–like peptide in limitless permutations to form specific receptors, allowing an organism to easily detect and avoid potentially harmful compounds with as little energy expenditure as possible. Furthermore, because RL-TGR seems to be specifically expressed in fish, this coreceptor in fish might have coevolved with the triterpene glycoside compounds found in their potential prey organisms. Triterpene glycosides are closely related to steroidal glycosides and saponins, which are found in a variety of marine and terrestrial organisms (20, 30, 31, 34–43). Thus, RL-TGR and related receptors might serve as detectors of these classes of compounds, and other RAMP-like coreceptors might have evolved as detectors of other chemical defenses. Because RAMP homologs are expressed in diverse organisms, such as fish and mammals (33), and RL-TGR appears to be conserved across freshwater and marine fish species, these species might have evolved the same or homologous RAMP-like coreceptors for the detection of chemical defenses; that is, these coreceptors might have evolved through divergent evolution as a protection mechanism for predators, herbivorous/browsing animals, and other vertebrate consumers.

The identification of RL-TGR represents the discovery of a coreceptor that responds to marine chemical defense compounds. This accessory protein might have evolved as a flexible mechanism through which organisms can detect and avoid potentially harmful compounds. The discovery of RL-TGR is significant not only because it defines a chemoreceptor–ligand pair in a field where few of these interactions are known, but also because the gene encoding RL-TGR is the first identified that encodes a coreceptor that responds to a chemical defense compound. This finding may lead the way for the identification of many other receptors that mediate chemical defense signaling in both marine and terrestrial environments, because this protein has the potential to represent the first of an entire family of coreceptors that respond to aversive compounds. The further study of RL-TGR and related coreceptors should deepen our understanding of the molecular mechanisms of chemical defenses and their effects on predator–prey interactions.

Methods

Isolation of Chemoreceptor Gene.

We previously showed that the rat aldehyde receptor OR-I7, which couples to G_olf, is capable of activating a Gα_S-mediated signaling cascade, resulting in the opening of heterologously expressed CFTR ion channels in the plasma membrane of Xenopus oocytes expressing this channel and thereby changing the current, as measured by two-electrode voltage clamping (TEVC) (19). Furthermore, we showed that application of formoside to oocytes coexpressing a whole zebrafish cDNA library, β₂AR, and CFTR causes a CFTR-like electrophysiological response (19). The electrophysiological response for CFTR channels activated by a chemoreceptor-mediated signaling cascade is a slow, broad change in current that slowly returns to baseline. Consequently, we hypothesized that we could use this bioassay to isolate a cDNA encoding a formoside-responsive receptor.

The bioassay-guided fractionation technique used in the present work makes use of the aforementioned electrophysiological bioassay to separate zebrafish cDNA library clones by sequentially fractionating the pools of clones that induce a positive response (Fig. S1). In brief, pools of library clones were linearized with Pac I and transcribed into cRNA. These pools of cRNA were microinjected into X. laevis oocytes along with cRNA encoding CFTR and β₂AR. Oocytes were tested via the electrophysiological bioassay to probe for an increase in CFTR current in response to formoside. The cDNAs contained in library fractions corresponding to the oocytes that responded to the application of formoside were subfractionated, mini-prepped as a pool, in vitro transcribed into cRNA, microinjected into oocytes, and tested via the bioassay. This iterative process continued until one active clone was isolated and sequenced in both the forward and reverse directions with T7 promoter and T7 terminator universal primers, respectively.

Electrophysiology.

X. laevis oocytes were isolated from adult females and prepared as described previously (44, 45). Various combinations of library transcript (2.5–10 ng), CFTR transcript (1.25–5 ng), β₂AR transcript (0.5–2 ng), and OR-I7 transcript (0.5–2 ng) were microinjected into stage V oocytes. After incubation for 48–96 h in L-15 media (Invitrogen) at 17 °C, the oocytes were tested by TEVC using a GeneClamp 500 amplifier (Axon Instruments). The recording solution was ND96 (96 mM NaCl, 1 mM MgCl₂, 2 mM KCl, 5 mM Hepes (pH 7.50)] with 1.8 mM CaCl₂ added. Oocytes were treated with deterrent compounds dissolved in ND96 buffer and a minimal amount of solvent (ethanol, DMSO, or water), usually ∼0.01% of the final concentration. Electrophysiological responses were detected by TEVC, indicating that the expressed receptor was activated by the compound, inducing a signal cascade that resulted in a change in current. Whole oocyte currents were recorded at V_M = −60 mV. Application of vehicle alone did not cause a change in current.

Statistical Methods.

A paired Student t test was performed to compare electrophysiological responses between formoside and other compounds in the same oocyte. Error is expressed as SEM unless noted otherwise.