Odorants selectively activate distinct G protein subtypes in olfactory cilia.

Chemoelectrical signal transduction in olfactory neurons appears to involve intracellular reaction cascades mediated by heterotrimeric GTP-binding proteins. In this study attempts were made to identify the G protein subtype(s) in olfactory cilia that are activated by the primary (odorant) signal. Antibodies directed against the alpha subunits of distinct G protein subtypes interfered specifically with second messenger reponses elicited by defined subsets of odorants; odor-induced cAMP-formation was attenuated by Galphas antibodies, whereas Galphao antibodies blocked odor-induced inositol 1,4, 5-trisphosphate (IP3) formation. Activation-dependent photolabeling of Galpha subunits with [alpha-32P]GTP azidoanilide followed by immunoprecipitation using subtype-specific antibodies enabled identification of particular individual G protein subtypes that were activated upon stimulation of isolated olfactory cilia by chemically distinct odorants. For example odorants that elicited a cAMP response resulted in labeling of a Galphas-like protein, whereas odorants that elicited an IP3 response led to the labeling of a Galphao-like protein. Since odorant-induced IP3 formation was also blocked by Gbeta antibodies, activation of olfactory phospholipase C might be mediated by betagamma subunits of a Go-like G protein. These results indicate that different subsets of odorants selectively trigger distinct reaction cascades and provide evidence for dual transduction pathways in olfactory signaling.

Chemoelectrical signal transduction in olfactory neurons appears to involve intracellular reaction cascades mediated by heterotrimeric GTP-binding proteins. In this study attempts were made to identify the G protein subtype(s) in olfactory cilia that are activated by the primary (odorant) signal. Antibodies directed against the ␣ subunits of distinct G protein subtypes interfered specifically with second messenger reponses elicited by defined subsets of odorants; odor-induced cAMP-formation was attenuated by G␣ s antibodies, whereas G␣ o antibodies blocked odor-induced inositol 1,4,5-trisphosphate (IP 3 ) formation. Activation-dependent photolabeling of G␣ subunits with [␣-32 P]GTP azidoanilide followed by immunoprecipitation using subtype-specific antibodies enabled identification of particular individual G protein subtypes that were activated upon stimulation of isolated olfactory cilia by chemically distinct odorants. For example odorants that elicited a cAMP response resulted in labeling of a G␣ s -like protein, whereas odorants that elicited an IP 3 response led to the labeling of a G␣ o -like protein. Since odorant-induced IP 3 formation was also blocked by G ␤ antibodies, activation of olfactory phospholipase C might be mediated by ␤␥ subunits of a G o -like G protein. These results indicate that different subsets of odorants selectively trigger distinct reaction cascades and provide evidence for dual transduction pathways in olfactory signaling.
Chemoelectrical signal transduction is considered to be mediated via intracellular reaction cascades triggered by G protein-coupled receptors (1). Biochemical studies over the last decade have revealed that odorants elicit the formation of either cAMP or IP 3 1 in olfactory preparations (2)(3)(4)(5)(6). Whereas the functional implications of the dual transduction pathways in the crustacean olfactory system are well established (7,8), in vertebrates the relative importance of the two pathways in olfactory signaling remains controversial (9,10). Heterotrimeric GTP-binding proteins play a key role in signal transduction processes, coupling activated receptors to the appropriate effector system. A variety of different G␣ subtypes have been identified in vertebrate olfactory epithelium including G s short , G il , G i2 , G i3 , G o , and G q (11)(12)(13)(14)(15)(16)(17). Even an olfactory-specific isoform of G s (G olf ) has been discovered (18). However, it is currently unclear how many and which type of G proteins are involved in olfactory signal transduction. To approach the question of which G protein subtype(s) may mediate the transduction processes in olfactory sensory cells, it is necessary to identify the G protein that is activated upon stimulation with distinct odor ligands. This can be accomplished by an activationdependent labeling procedure (19), in which receptor-activated G protein ␣ subunits are photolabeled using the hydrolysisresistant GTP-analogue [␣-32 P]GTP azidoanilide. Subsequent immunoprecipitation of G␣ subunits with subtype-specific antibodies permits identification of G protein subtypes that are labeled upon stimulation of olfactory cilia preparations with distinct odorants. The data indicate that cAMP-and IP 3 -inducing odorants result in labeling of different G protein subtypes.

Methods
Antisera-Antisera against G protein subunits were obtained either after injection into rabbits of synthetic peptides representing subtypespecific regions of different subunits using procedures described previously (19,20), or from Santa Cruz Biotechnology (Santa Cruz, CA). In both cases, the peptide sequences used to raise the antisera are shown in Table I.
Isolation of Olfactory Cilia-Olfactory cilia preparations were obtained using the calcium-shock method (21,22). Briefly, after a short wash of the olfactory epithelium in ice-cold saline solution (120 mM NaCl, 5 mM KCl, 1.6 mM K 2 HPO, 25 mM NaHCO 3 , 7.5 mM glucose, pH 7.4), the tissue was subjected to Ringer's solution containing 10 mM calcium and gently stirred for 5 min at 4°C. Detached cilia were isolated by three sequential centrifugation steps for 5 min at 7,700 ϫ g. The supernatants were collected, and the resulting pellets were resuspended in Ringer's solution containing 10 mM CaCl 2 as described above. The cilia preparation was obtained after a final centrifugation step of all the pooled supernatants for 15 min at 27,000 ϫ g. The resulting pellet containing the cilia was resuspended in hypotonic buffer (10 mM Tris, 3 mM MgCl 2 , 2 mM EGTA, pH 7.4) and stored at Ϫ70°C. The yield of cilia was around 0.5 mg per rat.
Stimulation Experiments and Second Messenger Determination-To determine the influence of the subtype-specific G protein ␣ subunit antisera on the efficiency of odorant-induced second messenger re-sponses, isolated cilia were preincubated with the indicated dilutions of specific antisera and subsequently stimulated with an odorant mixture.
Stimulation experiments were performed at 37°C for 2 min in the presence of 1 mM isobutylmethylxanthine when cAMP formation was determined, or 10 mM LiCl when the IP 3 response was measured. Briefly, 205 l of reaction buffer (200 mM NaCl, 10 mM EGTA, 50 mM Mops, 2.5 mM MgCl 2 , 1 mM dithiothreitol, 0.05% sodium cholate, 1 mM ATP, and 2 M GTP, pH 7.4) including 12 nM free calcium calculated and adjusted as described elsewhere (23), with or without odorants was prewarmed at 37°C. The reaction was started by the addition of 30 l (0.4 -1 g/l) of olfactory cilia and stopped by addition of 7% ice-cold perchloric acid (100 l) prior to determining the concentration of cAMP (24) or IP 3 (25).
Photolabeling of Activated G Proteins-[␣-32 P]GTP azidoanilide was synthesized and purified (26). Frozen cilia preparations in hypotonic buffer were centrifuged (4 min, 12,000 ϫ g, 4°C) and resuspended in double concentrated labeling buffer (60 mM HEPES, 5 mM MgCl 2 , 200 mM NaCl, 200 M EDTA, pH 7.4) with different GDP concentrations. The protein concentration was adjusted to 2.6 g/l, and aliquots of 30 l were adapted to 37°C for 2 min with 20 l of [␣-32 P]GTP azidoanilide (1 Ci/tube). The reaction was started by adding 10 l of labeling buffer with or without odorants. After indicated time points, the reaction was terminated by cooling the samples to 4°C. Excess [␣-32 P]GTP azidoanilide was removed by centrifugation (4 min, 12,000 ϫ g, 4°C). The pellet was resuspended in labeling buffer containing 2 mM dithiothreitol, placed on a Parafilm-coated metal plate (4°C), and irradiated for 30 s with a 254-nm UV lamp (150 W, Vl-100 Grid-Tube, Herolab GmbH).
SDS-PAGE and Western Blot Analysis-Membrane preparations were prepared for SDS-PAGE as described previously (27), subjected to 12.5% acrylamide gel electrophoresis, and analyzed using the Laemmli buffer system (28). For Western blot analysis, the separated proteins were transferred onto nitrocellulose using a semidry blotting system (Pharmacia Biotech Inc.). The blot was stained with Ponceau S and stored at 4°C until use. For Western blot analysis, nonspecific binding sites were blocked with 5% nonfat milk powder (Naturaflor) in TBST (10 mM Tris, pH 8.0, 150 mM NaCl, and 0.05% Tween 20). The blots were incubated overnight with specific antibodies against the different G protein ␣ subunits diluted in TBST, containing 3% nonfat milk powder. After three washes with TBST, a horseradish peroxidase-conjugated goat anti-rabbit IgG (1:10,000 dilution in TBST with 3% milk powder) was applied, and the ECL system was used to monitor immunoreactivity.

RESULTS
The antisera used in this study were generated against synthetic peptides derived from sequence domains that are unique for a particular class of G protein ␣ subunits (see Table I).
Western blot analyses using specific antisera were performed to determine the relative distribution of distinct G␣ subunits within (a) a preparation of olfactory sensory cilia, (b) total olfactory epithelium, and (c) a cerebral cortex preparation (Fig. 1). The antiserum AS 348, which recognizes a decapeptide corresponding to the C terminus of G s as well as G olf (19), detected a single 44 -45-kDa band. The immunoreactive polypeptide was found to be highly enriched in olfactory cilia compared with the whole olfactory epithelium. The antibody C10, which recognized a C-terminal decapeptide of G␣ i-1 , G␣ i-2 ,

FIG. 1. Western blot analysis of different G protein subtypes in olfactory tissue.
To determine the relative distribution of different G protein ␣ subunits, proteins from isolated olfactory cilia (Ci) as well as membrane preparations from total olfactory epithelium (Oe) (25 g protein) were separated by SDS-PAGE, transferred to nitrocellulose, and subsequently probed with selective sequence-specific antibodies (AS 348 for G␣ s (1:1000), C-10 for G␣ i (1:2000), K-20 for G␣ o (1:3000), and AS 368 for G␣ q (1:2000). In addition, membrane fractions of cerebral cortex (Co) were assayed for G protein ␣ subunit expression. Immunoreactive polypeptides were visualized using the ECL system employing conjugated horseradish peroxidase goat anti-rabbit IgG as the second antibody. Note that G s and G o are enriched in olfactory cilia compared with preparations from whole epithelium, whereas immunoreactivity of G q and G i subtypes are not enriched in cilia preparations. Molecular masses (kDa) of marker proteins are indicated. revealed an immunoreactive band at 40 kDa that was enriched in the cilia (see Fig. 1). Antibodies specific for distinct G proteins have been used successfully in functional studies. For example specific inhibi- tion of (a) ligand-induced, ␣ subunit GTPase activity (29,30) and (b) ␣-phosphatidylinositol 4,5-diphosphate hydrolysis (31) have been described. Evidence for two second messenger pathways in olfactory signaling suggests that more than one G protein subtype may be involved in mediating olfactory transduction (8,5,32). Therefore attempts were made to determine if the utility of subtype-specific antibodies could be used as tools to identify G protein subtypes that are active in olfactory signaling cascades. Isolated olfactory cilia were pretreated with different concentrations of subtype-specific antibodies and subsequently stimulated with odorant mixtures, which elicit either cAMP (citralva, hedione, and eugenol) or IP 3 formation (lilial, lyral, and ethylvanillin).
The effect of increasing concentrations of the anti-G␣ s serum on odor-induced cAMP or IP 3 formation is shown in Fig. 2. Whereas odor-induced cAMP formation was blocked in a concentrationdependent manner reaching about 45% inhibition at a 1:100 dilution ( Fig. 2A), IP 3 formation elicited by appropriate odorants was not significantly affected (Fig. 2B).
Activation of phospholipase C␤ subtypes is mediated by members of either the pertussis toxin-insensitive G q family or by pertussis toxin-sensitive G i and G o proteins (33) . Fig. 3 shows the effect of increasing concentrations of G␣ i , G␣ q , and G␣ o antibodies on odor-induced second messenger responses. Pretreatment of cilia preparations with any of the three subtype-specific antibodies did not alter the responsiveness to the odorants citralva, hedione, and eugenol, known to induce a cAMP signal (Fig. 3A). The IP 3 response elicited by appropriate odorants was not affected by anti G␣ i serum (Fig. 3B). In contrast, antibodies against G␣ o significantly attenuated odorinduced IP 3 formation in a concentration-dependent manner; inhibition was 65% at a 1:250 dilution and more than 75% at a 1:100 dilution. G␣ q antibodies gave a significant inhibition only at the highest concentration (1:100).
With the aim of identifying directly G protein subtypes that are activated upon odor stimulation, a photoaffinity labeling approach was employed using the photoreactive hydrolysis resistant [␣-32 P]GTP azidoanilide (26,34). Previous studies have shown that monitoring receptor-stimulated binding of GTP analogues require addition of exogenous GDP (19,35,36). Since G protein subtypes display different basal nucleotide exchange rates (37)(38)(39), it was necessary to determine the appropriate GDP concentration that allows visualization of odor-induced G protein labeling. In the first set of experiments, conditions were optimized toward an odorant-induced photolabeling of G␣ s proteins. Isolated olfactory cilia were incubated with [␣-32 P]GTP azidoanilide in the presence of different concentrations (0 -1 mm) of exogenous GDP, and incubation was continued for 20 s at 37°C upon application of a mixture of three odorant compounds (citralva, hedione, and eugenol, each 1 M). G␣ s subunits were immunoprecipitated using an antiserum directed against G␣ s subtypes, separated on SDS-PAGE, and the incorporated [␣-32 P]GTP azidoanilide label was determined by autoradiography. The results of a representative experiment (n ϭ 3) are shown in Fig. 4. The immunoprecipitate gave a single photolabeled band with an apparent molecular mass of 44 -45 kDa, a size identical to the molecular mass of the protein visualized in immunoblot experiments (see Fig. 1). However, the G s common antiserum AS 348 used to immuno- precipitate G␣ s subunits does not allow us to distinguish whether the G olf (44.7 kDa) or the G s short isoform (44.2 kDa), both of which are expressed in the olfactory system, is labeled upon odorant stimulation. Comparing the intensity of [␣-32 P]GTP labeling, it was clear that, at low GDP concentrations, photolabeling is similar under control conditions and in the presence of odorants. However, upon application of rather high GDP concentrations (1 mM), significantly enhanced labeling was detected in stimulated samples. This observation contrasts with studies on photolabeling of G␣ s in membrane preparations of human platelets, where agonist-induced labeling was detectable in the presence of 1 M GDP (19).
In view of the rapid kinetics of olfactory reaction cascades, time course experiments on agonist-induced labeling of G␣ s were performed, in which cilia preparations were photolabeled upon incubation with an odorant mixture (citralva, hedione, and eugenol) for different time intervals. Application of odorants elicited a rapid incorporation of the labeled GTP analogue (Fig. 5). The ratio of agonist-stimulated to basal photolabeling of the G␣ s -like protein was highest at short incubation times; the relative labeling was fully saturated after 10 s. In contrast, hormone-induced incorporation of [␣-32 P]GTP azidoanilide into G␣ s of human platelets has been shown to follow a very different time course (see Fig. 5). Maximal labeling is reached after about 10 min.
To determine the potency of individual odorants, photolabeling experiments were performed using different concentrations of citralva. As demonstrated by the autoradiogram in Fig. 6A, incorporation of [␣ 32 P]GTP azidoanilide into G␣ s -like protein increased in a concentration-dependent manner. In addition, it is clear that even very low doses (picomolar) of the odorant are sufficient to induce a significant labeling of G␣ s . The intensity of the labeling, evaluated densitometrically, was used to construct a concentration-response curve (Fig. 6B). In conformity with the results of many similar olfactory stimulation experiments, we did not detect saturation; an approximately halfmaximal activation at about 50 nM was estimated. These results are in line with previous experiments monitoring odorinduced second messenger responses (5).
Odorants showed different potencies when stimulation of adenylyl cyclase was examined (2). To explore whether less potent adenylyl cyclase activators also induce labeling of the G␣ s -like protein, cilia preparation were stimulated with the odorant eugenol, which shows only 47% of adenylyl cyclase activation compared with citralva. Stimulation of cilia preparations with 1.6 M eugenol caused a significant incorporation of [␣-32 P]GTP azidoanilide compared with samples incubated without odorant (Fig. 7A). Thus this procedure also allows for the determination of odorant-dependent G␣ s labeling by less potent adenylyl cyclase activators.
Several odorants have been shown not to induce a detectable cAMP signal but rather the formation of IP 3 (32). However, it might be possible that adenylyl cyclase activation was not detected due to the insufficient sensitivity of the method used. By photoaffinity labeling of G␣ subunits, we were able to detect odor-induced activation of G␣ s at very low odorant concentrations (see Fig. 6). Therefore, we examined the labeling of G␣ s by individual phospholipase C-stimulating odorants, i.e. lyral, isovaleric acid, and pyrrolidine. The results depicted in Fig. 7, B and C, show that application of citralva at different concentrations induced a concentration-dependent increase in G␣ s labeling, whereas neither low (16 nM) (Fig. 7), nor high odor concentrations (1.6 M) of isovaleric acid and pyrrolidine (Fig.  7C) or lyral (Fig. 7A) affected labeling of G␣ s . Different G protein types are known to link receptors to phospholipase C (33). To evaluate which G protein subtype might be involved in odor-induced IP 3 formation, photolabeling studies were performed with a stimulating odorant mixture (lilial, lyral, and ethylvanillin, each 1 M) followed by immunoprecipitation with subtype-specific antibodies for G␣ q /G␣ 11 / G␣ 14 types (AS 348) (40), G␣ i (AS 266), and G␣ o isoforms (AS 6) (34). In all cases the different antibodies precipitated photolabeled proteins with molecular masses identical to those found in the immunoblotting experiment (see Fig. 1). However, proteins precipitated with G␣ q or with G␣ i antibodies did not show any increase in [␣-32 P]GTP azidoanilide incorporation upon odorant stimulation, neither in the presence of a low GDP concentration (not shown) nor in the presence of high GDP levels (Fig. 8, 500 M GDP). Nevertheless, for G␣ o the results were different. Whereas at low concentrations of GDP (0 -100 M) no differences were detected in photoaffinity labeling of G␣ o when compared with control samples (data not shown), at high exogenous GDP concentrations (500 M; see Fig. 8), an odorant-induced increase in [␣-32 P]GTP azidoanilide incorporation was detected in proteins precipitated with G␣ o antibodies. This indicates that the "IP 3 odors" activating G␣ o -like protein may have a similarly high nucleotide exchange rate as the G␣ s -like G protein, which is labeled upon application of "cAMP-odors" (see Fig. 4).
The potency of odorants in activating the G o -like protein is demonstrated in Fig. 9A; as shown for G␣ s activation (see Fig.  6). Very low odor concentrations were sufficient to induce significant labeling of G␣ o . A densitometric evaluation of the photoaffinity labeled G␣ o -like protein is presented in Fig. 9B; the concentration-response curve revealed an apparent halfmaximal labeling of G o at odorant concentrations of about 200 nM.
To evaluate the specificity of G␣ o labeling, we analyzed the effects of individual odorants that have been shown to activate phospholipase C. Stimulation with the "IP 3 -odorants" lyral ( Fig. 10A; 1.6 M) or isovaleric acid (Fig. 10, B, 16 nM, and C, 1.6 M) induced an enhanced photolabeling of G␣ o . Analyzing the intensity of the labeling densitometrically, it was clear that isovaleric acid induced a concentration-dependent increase in [␣-32 P]GTP azidoanilide incorporation. In contrast, the application of high odor concentrations of the cAMP compound eugenol did affect G o labeling (see Fig. 10A; 1.6 M); even stimulation with high concentrations of the very potent adenylyl cyclase activators citralva or hedione failed to induce a significant incorporation of the GTP analogue (see Fig. 10C, 1.6 M). As G i and G o subtypes usually activate phospholipase C through their ␤␥ subunits (33), experiments were performed to explore whether the ␤␥ subunit of the identified G o -like G protein is mediating phospholipase C activation. Isolated olfactory cilia were pretreated with an antiserum selective for the N-terminal sequence common to all members of G␤ subunits; subsequently, odor-induced second messenger responses were determined. Whereas odor-induced cAMP formation was not affected (Fig. 11B), the odor-induced IP 3 response was attenuated by G␤-antibodies; at a 1:250 antibody dilution, the odorinduced IP 3 signal was reduced to 30% (Fig. 11A). DISCUSSION The present study shows that subtype-specific antibodies attenuate odor-induced second messenger responses and immunoprecipitate activation-dependent photolabeled G␣ subunits. Both approaches indicate that a G s -like protein mediates odor-induced cAMP, whereas a G o -like protein controls odorinduced formation of IP 3 . Thus, we have demonstrated that different subsets of odorants selectively activate one of the two G proteins. This finding is consistent with previous biochemical studies indicating that odorants elicit either a cAMP or an IP 3 response in olfactory cilia preparations. Therefore, these results provide further evidence that the phenomenon of dual transduction pathways in olfactory signaling, which is well established for the lobster (8), is also found in vertebrates. However, it is presently unclear how these biochemical results can be reconciled with the observation that transgenic mice lacking a functional cyclic nucleotide-activated cation channel displayed general anosmia (10). The G s common antiserum AS 348 used to immunoprecipitate G ␣s subunits did not allow us to distinguish whether G olf (44.7 kDa) or the olfactory G s short isoform (44.2 kDa) is labeled upon odorant stimulation. Although it appears likely that G olf is involved in the cAMP pathway (41), the identity of the G s subtype photolabeled upon odor stimulation remains to be determined. Phospholipases of the ␤-type are regulated either by the ␣ subunits of the G q family or by the ␤␥ subunits of trimeric G proteins. Although the G proteins that release the activating ␤␥ subunits are not identified, there is evidence, that they are subtypes of the pertussis toxin-sensitive G i /G o family (33). The involvement of G o proteins in phospholipase C regulation was first observed in experiments on Xenopus oocytes demonstrating that G o proteins specifically enhance the Cl Ϫ current elicited by muscarinic receptors via IP 3 and Ca 2ϩ (42)(43)(44). The observation that olfactory phospholipase C activation occurs through ␤␥ subunits of a G o -like G protein is thus of particular interest and in line with previous studies indicating that odorant-induced IP 3 formation in rat olfactory cilia is mediated by a pertussis toxinsensitive G protein (5). Although a firm identification of the G o -like protein in olfactory cilia was not possible in this study, future investigations using more specific antibodies may allow us determine whether one of the two previously identified G o isoforms (45,46) or a novel G␣ subtype, which shares epitopes with G o protein, is active in olfactory neurons. In this context it is interesting to note that G o proteins are also thought to be involved in mammalian pheromone signaling by chemosensory neurons of the vomeronasal organ (47,48). The observation that two different G proteins are active in chemosensory neurons of the rat finds its parallel in the nematode Caenorhabditis elegans. Recent findings have demonstrated that two different G proteins encoded by gpa-2 and gpa-3 (49,50) are expressed in the ciliated chemosensory amphid neurons and are involved in pheromone detection (51). Interestingly, one of the two subtypes (GPA-3) comprises a conserved cysteine residue near the carboxyl terminus, which is considered to be a substrate for the pertussis toxin-catalyzed attachment of an ADP-ribose moiety to G o and G i proteins (52).
The rapid kinetics of ligand-induced GTP incorporation in olfactory cilia, when compared with endocrine cells (Fig. 5), suggest a particularly efficient interaction between the signaling molecules, notably the activated receptors and G proteins. This is reminiscent of the high speed activation described for the rhodopsin/transducin system (53), which may be based on fast lateral diffusion rates due to a special lipid composition of the membrane (54). Alternatively, response time may not be diffusion-limited, but rather elements of the transduction machinery may be organized as architectually and spatially distinct ultramicrodomains. Such "transducisomes" have recently been described for the fly photoreceptors (55). Caveolae, specialized microdomains in the plasma membrane, appear to be another compartmental basis for a rapid and efficient coupling of transmembrane signaling events (56,57). Preliminary studies revealed that caveolin, a characteristic integral membrane protein that acts as an oligomeric docking site for distinct proteins of signaling cascades, is indeed present in olfactory sensory cilia, 2 suggesting that such transduction centers might provide a basis for the rapid kinetics of second messenger signaling in olfaction.