Selective Activation of G Protein Subtypes in the Vomeronasal Organ upon Stimulation with Urine-derived Compounds*

Chemosensory neurons in the vomeronasal organ (VNO) detect pheromones related to social and reproductive behavior in most terrestrial vertebrates. Current evidence indicate that the chemoelectrical transduction process is mediated by G protein-coupled second messenger cascades. In the present study, attempts were made to identify the G protein subtypes which are activated upon stimulation with urinary pheromonal components. G protein-specific antibodies were employed to interfere specifically with inositol 1,3,4-trisphosphate formation induced by urinary stimuli and to immunoprecipitate Gα-subunits, activation dependently labeled with [α-32P]GTP azidoanilide. The results of both experimental approaches indicate that stimulation of female VNO membrane preparations with male urine samples induces activation of Gi as well as Go subtypes. Experiments using different fractions of urine revealed that upon stimulation with lipophilic volatile odorants, only Gi proteins were activated, whereas Go activation was elicited by α2u-globulin, a major urinary protein, which is a member of the lipocalin superfamily. Since each G protein subtype is stereotypically coexpressed with one of the two structurally different candidate pheromone receptors (V1R and V2R), the results provide the first experimental evidence that V1Rs coexpressed with Gimay be activated by lipophilic probably volatile odorants, whereas V2Rs coexpressed with Go seem to be specialized to interact with pheromonal components of proteinaceous nature.

Terrestrial vertebrates detect chemical signals via sensory neurons located in two anatomically distinct systems: the nasal olfactory epithelium (OE) 1

and the vomeronasal organ (VNO).
Whereas the main olfactory system is responsible for the "conventional" sense of smell, the VNO appears to specifically detect pheromones, thereby inducing a distinct social or sexual behavior or endocrine response (1)(2)(3). In rodents, the major source of phermones seems to be the urine, however, only few volatile (4) and non-volatile (5) urinary substances producing a definite endocrine or behavioral response have been identified. The best characterized non-volatile urinary components in male mouse urine are the "major urinary proteins" (6). Major urinary proteins have been suggested to be involved in puberty acceleration (5,7,8), an endocrinological effect prevented by lesions of the VNO or the accessory olfactory bulb (1). Major urinary proteins of mice and the rat equivalent ␣ 2u -globulin (9) belong to the superfamily of lipocalins, a structurally homologous but diverse family of extracellular proteins, characterized by their ability to bind small, principally hydrophobic molecules (10). Although recent structural studies have characterized the naturally bound volatile substances (11), major urinary proteins lacking the natural volatiles remain active (8) suggesting that proteins themselves act as pheromones.
The molecular mechanisms responsible for the detection of pheromones and the signal transduction processes in the VNO are not well understood. Initially, it has been suggested that the transduction mechanisms in the VNO would resemble those in the olfactory epithelium, but some of the specific molecular components involved in the transduction of odorants in receptor neurons of the OE, like G olf , adenylyl cylase (AC) subtype III, and the ␣-subunit of the cyclic nucleotide-gated cation channel could not be detected in the neuroepithelium of the VNO (12,13). Moreover, recent molecular biology advances implicate that neurons in the VNO utilize another set of receptor genes than the olfactory epithelium. Two novel and structurally unrelated multigene families of putative pheromone receptors have been identified (V1R and V2R) which are exclusively expressed in neurons of the VNO (14 -17). Although receptors of both families represent members of G proteincoupled receptors, the structural differences especially in the N-terminal region led to the concept that each receptor family may be tuned to recognize a distinct class of ligands.
Receptors of each family show a distinct nonoverlapping expression pattern within two major laminar zones. Whereas V1Rs are expressed in the apical half of the neuroepithelium, V2Rs are restricted to the basal region of the epithelium. An identical spatial expression pattern has also been observed for two types of G protein ␣-subunits. Whereas G i is expressed only in apical neurons, G o expression is restricted to neurons in the basal half of the VNO neuroepithelium (13,14,18). The stereotypical distribution of both receptor families with distinct G protein subtypes implies a linkage between both components.
The present study was aimed to characterize the G protein subtypes in the rat vomeronasal organ activated upon stimulation with male urine by performing photoaffinity labeling experiments. Furthermore, antibodies for different G protein subtypes were employed to interfere with the urine-induced IP 3 signaling. To evaluate whether there is any correlation between the chemical properties of applied urinary stimuli and the activation of distinct G protein subtypes, stimulation ex-periments were performed with different fractions of urine as well as with purified ␣ 2u -globulin.

Methods
Antisera-Antisera against G protein subunits were obtained after injection of synthetic peptides representing subtype-specific regions of different subunits into rabbits as described previously (20) or provided by Santa Cruz Biotechnology, Santa Cruz, CA. The peptide sequences used to raise the antisera are shown in Table I.
Urine Collection and Purification-Urine from male fertile rats (12-14 weeks old) was collected daily, pooled, centrifuged to remove cells (5 min, 5, 500 ϫ g), and stored as aliquots at Ϫ70°C until use. To extract hydrophobic volatile odorants, a 2-ml volume of pooled male urine was treated with 2 ml of dichloromethane; following separation of the organic and water phase by centrifugation (10 min, 6,000 ϫ g), both phases were collected and stored at Ϫ70°C.
Purification of ␣ 2u -Globulin-Urine was collected from fertile male rats, pooled, and stored immediately at Ϫ70°C. After thawing to 4°C, a 50-ml pool was centrifuged at 5,000 ϫ g for 30 min at 4°C, and the clear urine was dialyzed against 4 mM sodium phosphate buffer, pH 6.8. The purification of ␣ 2u -globulin was performed according to Holmquist et al. (21). Briefly, after addition of phenylmethylsulfonyl fluoride to a final concentration of 0.1 mM, the samples were applied to a 1.5 ϫ 10-cm hydroxyapatite column with a flow rate of 0.5 ml/min. The column was washed with 10 times column volumes of 4 mM sodium phosphate buffer, pH 6.8. Elution was performed using a linear 4 -400 mM sodium phosphate buffer gradient, pH 8.0. Total protein was evaluated by monitoring the A 280, corresponding fractions were pooled and analyzed by SDS-PAGE. Fractions containing the 18-kDa proteins were collected, dialyzed against 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 4 M GTP, 12 nM CaCl 2 , pH 7.4) and concentrated using Centricon-10 to a final protein concentration of 17 mg/ml.
Construction of Expression Plasmids-DNA-related methods were performed according to standard procedures (22). The coding sequences of ␣ 2u -globulins were amplified by polymerase chain reaction using cDNA obtained by reverse transcription of poly(A) ϩ RNA from male rat liver. A recombinant ␣ 2u -globulin with an N-terminal sequence corresponding to the endogenous protein was generated by using an extended 5Ј-oligonucleotide in the polymerase chain reaction that contained a BamHI site (underlined) and the sequence coding for an enterokinase cleavage site in-frame to the coding sequence: ␣ 2u -forward 5Ј-ggatccgtacgacgatgacgataaggaagaagctagttccacaagagg-3Ј; as an antisense primer oligonucleotide containing an HindIII site (underlined), the following oligonucleotide was used: ␣ 2u -reverse 5Ј-aaagcttcgtcaacctcaggcctggagacagcg-3Ј. The amplification products were cloned into the pGEM-T vector. All derived plasmid constructs were confirmed by DNA sequencing. For subcloning into the expression vector pTrcHis, the ␣ 2u -globulin sequences from pGEM-T-plasmids were cut out with BamHI and HindIII restriction enzymes and ligated in-frame into the corresponding vector sites. The expression constructs were transformed into Escherichia coli BL21 (DE3).
Growth of Cells and Protein Purification-100 ml of LB-medium with ampicillin (100 g/ml) was inoculated with an overnight culture in the same medium and grown at 37°C. Protein expression was induced at A 600 of 0.5-0.7 with 0.2 mM isopropyl-␤-D-thiogalactoside, and growth was continued for 3 h. Cells were harvested by centrifugation at 2,000 ϫ g for 15 min and resuspended in 20 ml of 50 mM sodium phosphate buffer, pH 8.0, 300 mM NaCl, 20 mM imidazole, 0.25 mg/ml lysozyme, and 1 mM phenylmethylsulfonyl fluoride and stored on ice for 1 h. For damage of spheroplastes, cells were frozen in liquid nitrogen and thawed at room temperature. After sonification with a tip sonicator for 2 min on ice, the soluble material was separated from cell debris and inclusion bodies by centrifugation at 30,000 ϫ g for 30 min. The supernatant was collected and used for affinity chromatography.
For purification of His-tagged proteins, the cytosol was loaded onto a Ni-NTA agarose column equilibrated against washing buffer (50 mM sodium phosphate buffer, pH 8.0, 300 mM NaCl, and 20 mM imidazole) and washed with 5 column volumes of washing buffer. Subsequently, bound protein was eluted as single peak by increasing the concentration of imidazole to 250 mM. Fractions of 1 ml were collected and protein concentrations were monitored by A 280 ; corresponding fractions were pooled and dialyzed twice for 8 h against 50 mM Tris/HCl, pH 7.5.
Protein concentration was adjusted to 1 mg/ml with 50 mM Tris/HCl, pH 7.5, 0.1% Tween 20. Cleavage of fusion protein was performed by incubation with 1/40 (w/w) enterokinase for 16 h at 37°C. Subsequently, the NaCl concentration was increased to a final concentration of 500 mM, and agarose-coupled soybean trypsin inhibitor was added; after an incubation for 2 h at 4°C, soybean trypsin inhibitor bound enterokinase was precipitated by centrifugation (2,000 ϫ g for 10 min) and the supernatant was recovered.
Preparation of Vomeronasal Organ Membranes-Membrane fractions of the VNO were prepared as described previously (23). Briefly, VNOs removed from fertile male rats were washed twice in Ringer's solution (120 mM NaCl, 5 mM KCl, 1.6 mM K 2 HPO, 25 mM NaHCO 3 , 7.5 mM glucose, pH 7.4) to remove superficial blood and debris and subsequently frozen in liqiud nitrogen. Each pooled sample (30 -60 animals) was thawed on ice, minced, and subsequently subjected to Ringer's solution containing 10 mM calcium chloride; after gently stirring for 10 min at 4°C, debris was removed by centrifugation (10 min, 3,000 ϫ g), the resulting supernatant was collected and the pellet was resuspended again in Ringer's solution containing CaCl 2 as described above. The pooled supernatants containing the membrane fractions were centrifuged again for 30 min at 48,000 ϫ g and the resulting pellet was resuspended in aliquots (100 -500 g in 500 l) in hypotonic buffer (10 mM Tris/HCl, 3 mM MgCl 2 , 2 mM EGTA, pH 7.4) and stored at Ϫ70°C. Olfactory cilia of the OE were obtained using the calcium-shock method as described previously (24).
Stimulation Experiments and Second Messenger Determination-Stimulation experiments, performed with male rat urine, separated fractions of urine, isolated ␣ 2u -globulin of male urine, and recombinant ␣ 2u -globulin were performed as described previously (24).
Briefly, 300 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 4 M GTP, pH 7.4) including 12 nM free calcium calculated and adjusted as described (25), was prewarmed at 37°C with or without odorants. The reaction was started by the addition of 60 l of VNO membrane preparations, incubated for 2 min at 37°C and stopped by 7% ice-cold perchloric acid (200 l) before the concentration of IP 3 was determined according to Palmer et al. (26). To prevent degradation of IP 3 , stimulation experiments were performed in the presence of 10 mM LiCl (final concentration during incubation). To determine the influence of the subtype-specific G protein ␣-subunit antisera on the efficiency of urine-induced second messenger responses, membrane preparations of the VNO were preincubated with the indicated dilutions of the specific antisera and subsequently stimulated with urine or purified fractions of urine.

RESULTS
Based on a recent study indicating that treatment of vomeronasal organ slice preparations of rats with urine dilutions elicit an increased impulse frequency of sensory neurons which can be blocked by phospholipase C inhibitors (28), we set out to explore whether urine-induced IP 3 responses in membrane preparations of VNOs from female rat are mediated by a G protein-controlled reaction cascade. IP 3 responses induced upon stimulation with different concentrations of male urine (0.1 to 4%, v/v) were determined in the presence of GTP or GDP. The results in Fig. 1 demonstrate that male urine elicited a dose-dependent increase in the formation of IP 3 ; this response was only observed in the presence of GTP indicating that urine-induced phospholipase C activation in the VNO is mediated via G proteins.
Immunohistochemical studies (18,29,30) and in situ hybridization experiments (13) have demonstrated that different G proteins ␣-subunits (G s , G i2 , G o , and G q/11/13 ) are expressed in neurons of the VNO. Toward an identification of distinct G protein subtypes which are activated upon stimulation with urine, a photoaffinity labeling approach was employed using the photoreactive, hydrolysis resistant [␣-32 P]GTP azidoanilide (27,20). Female VNO membrane preparations were stimulated with two different concentrations of male urine in the presence of [␣-32 P]GTP azidoanilide, and subsequently distinct G protein ␣-subunits were immunoprecipitated (24) using subtypespecific antibodies for different G protein ␣-subtypes (Table I). Fig. 2 shows that urine addition significantly enhanced incorporation of [␣-32 P]GTP azidoanilide into G␣ i as well as G␣ o subtypes. Quantification of the labeling intensity indicates that stimulation with male urine (0.2%, v/v) induced a 57% increase in G␣ i labeling and a nearly 100% increase in the [␣-32 P]GTP azidoanilide incorporation into G␣ o compared with control samples. In contrast, proteins precipitated with an antiserum specific for G␣ s and G␣ olf (AS 348) as well as G␣ q subtypes (AS 368) show only very weak incorporation of [␣-32 P]GTP azidoanilide under control conditions, and no increase upon urine stimulation was registered (data not shown).
Antibodies for distinct G protein ␣-subtypes have been successfully used to selectively inhibit ligand-induced phosphatidylinositol bisphosphate hydrolysis (24,31). Accordingly, membrane preparations of female VNOs from rat were pretreated with different concentrations of subtype-specific antibodies and subsequently stimulated with a 2% (v/v) dilution of male urine. As demonstrated in Fig. 3A, preincubation with the G␣ i common antibody C-10 or with the G␣ o -specific antibody K-20 attenuated the urine-induced IP 3 response in a concentrationdependent manner; 50% inhibition was obtained at a 1:500 dilution of either antibody. In contrast, antibody AS 348 specific for G␣ s subtypes as well as AS 368 selective for the G␣ q subfamily did not affect the responsiveness significantly (Fig.  3B). These results indicate that the IP 3 response in female VNO preparations induced upon stimulation with a complex pheromonal mixture (male urine) is mediated via G i as well as G o proteins.
Functional and chemical analysis of urine from rodents have shown that pheromonal components are found in both the volatile as well as non-volatile fraction (5,32). The volatile components can be extracted from whole male urine using dichloromethane (8). Using this procedure, the dichloromethane-extractable urinary components as well as the remaining aqueous fraction were subsequently analyzed for their ability to induce IP 3 responses in VNO membrane preparations. As shown in Fig. 4, compounds extracted with dichloromethane induced a concentration-dependent increase in the level of IP 3 , indicating that this fraction contained active ligands; in contrast, the aqueous fraction did not induce any changes in the concentration of IP 3 compared with control samples (Fig.  4). Although it is unclear why the aqueous fraction is inactive it is conceivable that the proteins may be denaturated by dichloromethane.
In order to address the issue of which G protein subtypes are labeled upon stimulation with the volatile urinary components, additional photoaffinity labeling experiments were performed employing different concentrations of the dichloromethane extract (0.02-0.4%, v/v). As demonstrated in Fig. 5, the organic extract induced a dose-dependent incorporation of [␣-32 P]GTP azidoanilide into G␣ i , whereas a stimulus-induced labeling of G␣ o was not detectable (Fig. 5). These results indicate that G i proteins but not G o proteins are activated upon stimulation with volatile urinary components.
To confirm this observation, the effects of G␣ i as well as G␣ o antibodies on dichloromethane extract-induced IP 3 signaling

Selective Activation of Distinct G Proteins in the VNO
were analyzed. VNO preparations were pretreated with different concentrations of each of the two subtype-specific antibodies and subsequently stimulated with a 2% (v/v) dilution of the organic urinary extract. Whereas the dichloromethane extract induced IP 3 formation was only slightly affected upon pretreatment with high concentrations of the G␣ o antibody (1:250 dilution), G␣ i antibodies caused already a strong inhibition at very low antibody concentrations: at a 1:2000 dilution of the G␣ i antibody, the dichloromethane extract-induced IP 3 response was blocked by 70% (Fig. 6).
Previous electrophysiological experiments have demonstrated that the vertebrate VNO system is not only selective to a narrow range of pheromonal components, but in addition is sensitive to various general odorants normally known to be reorganized by the olfactory epithelium (33). In order to address the question which G protein subtypes are activated upon stimulation with volatile odorants, amyl acetate and men-thone, which elicit electrical responses in turtle VNO neurons (33), were applied in photolabeling experiments. Membrane preparations of rat VNOs were stimulated with different con- centrations of the two general odorants and subsequently G␣ i as well as G␣ o subtypes were precipitated by means of subtypespecific antibodies. Application of menthone (Fig. 7) as well as amyl acetate (not shown) induced a concentration-dependent labeling of G␣ i , however, neither menthone (Fig. 7) nor amyl acetate (not shown) induced a significant incorporation of the GTP analog into G␣ o .
The observation that neither organic components from male urine nor general volatile odorants induced G o activation led to the idea that urinary proteins may be ligand candidates for G o stimulation. In urine of fertile male rats, androgen-dependent ␣ 2u -globulin, a member of the lipocalin superfamily (10), is the quantitatively major protein (9). Therefore, ␣ 2u -globulin was purified from male urine as described by Holmquist et al. (21) and assessed for its capability to induce IP 3 formation. The results of these experiments (Fig. 8) indicate that purified ␣ 2u -globulin induced a very strong concentration-dependent increase in the concentration of IP 3 .
Since the procedure for isolating native ␣ 2u -globulin, such as gel permeation and dialysis purification steps, are not sufficient to separate noncovalently bound volatile urinary compounds (34), recombinant ␣ 2u -globulin, heterologously expressed in E. coli, was employed. As shown in Fig. 8A, application of recombinant ␣ 2u -globulin also induced an elevated IP 3 response; however, the efficacy was considerably smaller compared with native purified urinary protein. Nevertheless, the ␣ 2u -globulin-induced second messenger response is specific since stimulation of olfactory cilia from the OE did not show any changes in IP 3 concentration (Fig. 8B). To characterize the G protein subtype involved in recombinant ␣ 2u -globulininduced IP 3 signaling, the effect of subtype-specific antibodies on IP 3 formation was analyzed. As documented in Fig. 9, G␣ i antibodies did not alter the responsiveness to ␣ 2u -globulin, even at rather high concentrations. In contrast, antibodies against G␣ o significantly attenuate IP 3 formation in a concentration-dependent manner. This observation was confirmed in photoaffinity labeling experiments; stimulating VNO membrane preparations with 1 M recombinant ␣ 2u -globulin led to a significant incorporation of [␣-32 P]GTP azidoanilide into G␣ o , whereas labeling of G␣ i subtypes was not changed (Fig. 10). DISCUSSION In the present study, G protein subtypes involved in urineinduced IP 3 formation in VNO preparations were characterized by two different experimental approaches. Subtype-specific G protein antibodies have been used as tools to selectively inhibit urine-induced IP 3 responses and to immunoprecipitate G protein subtypes photoaffinity labeled with the hydrolysis resistant GTP analogue [␣-32 P]GTP azidoanilide. The results of both experimental approaches indicate that whole male urine in-duces activation of G i as well as G o subtypes, suggesting a role for both G protein subtypes in transducing pheromonal responses mediated by urinary components. Activation of G i and G o subtypes in the VNO appears to be realized by two structurally different classes of odorants. Whereas G i activation was only observed upon stimulation with lipophilic, possibly volatile components, G o activation was elicited by ␣ 2u -globulin, a major urinary protein, which is a member of the lipocalin superfamily. Previous studies have shown that lipocalin proteins, like ␣ 2u -globulin and mouse urinary proteins, share the ability to bind volatile pheromones (34), suggesting that both proteins act as carrier proteins for hydrophobic ligands; however, the results of the present study suggest that ␣ 2u -globulin is not merely a carrier of volatile pheromones but is able to induce receptor-mediated G o activation. This observation is in line with previous behavioral studies indicating that members of the lipocalin superfamily may by themselves be active in VNO-mediated pheromonal responses. It has been demonstrated that mouse urinary proteins separated from volatile ligands, either by extraction or competitive displacement, were still active in bioassays (8). Furthermore, recombinant aphrodisin, a lipocalin-like protein from the female golden hamster vaginal discharge, expressed in E. coli induced a specific copulatory response in male golden hamster (35). This observation indicates that lipocalins can act as phermones by themselves. However, a functional role for small ligands cannot be excluded, since recombinant aphrodisin only gained its full activity when it was supplemented with organic extracts of the vaginal discharge (35), suggesting that ligands may modulate the activity of lipocalins by altering its conformation.
A main result of the present study is the observation that both G protein subtypes, activated by different classes of urinary stimuli, led to the activation of phospholipase C and the generation of IP 3 . Thus, it is the question whether or not there is a physiological significance of having two populations of VNO neurons with distinct activated G protein subtypes that activate the same downstream effector. Activation of phospholipase C by pertussis toxin-sensitive G protein subtypes, like G i and G o , appears to be mediated by ␤␥-subunits of the trimeric G protein (36). Therefore one might speculate whether the simultaneously released ␣-subunits of G i and G o may affect pathways not monitored in this study, suggesting a bifunctional action of these G proteins. G␣ o -subunits are well known to modulate the activity of Ca 2ϩ channels (37); thus a G␣ o - regulated channel may contribute to govern the membrane potential in this subset of neurons. In the context of a bifunctional role for G␣ i -subunits, it is interesting to note that vomeronasal neurons express adenylyl cyclase subtype II (13). G␣ isubunits inhibit AC II (38), thus, it is conceivable that activation of G i by volatile pheromonal components may not only lead to phospholipase C activation mediated by ␤␥-subunits, but in addition to G␣ i controlled inhibition of AC II. Interestingly, an attenuation of cAMP levels in VNO preparations upon stimulation with the volatile pheromonal components dehydro-exo-brevicomin and 2-(sec-butyl)-4.5-dihydrothiazole has recently been reported (39).
The stereotypical coexpression pattern of the two families of candidate pheromone receptors (14 -17) with either G i or G o implies a functional linkage between both signal transduction elements. Based on characteristic structural features of each receptor family, it has been suggested that the two receptor classes may recognize structurally different chemical stimuli (40). One could imagine that V1Rs, characterized by short N-terminal domain and interconnecting loops, like ORs (14), detect low molecular weight volatiles, whereas V2Rs, having an extremely large highly variable N-terminal domain, may recognize high molecular weight nonvolatile pheromones. Alternatively, as V2Rs-related G protein-coupled receptors, like metabotrophic glutamate receptors (41) and the extracellular calcium sensing receptors (42) recognize very small ligands by means of the large N-terminal domain (43), it appears conceivable that V2Rs may bind small pheromonal components via a "venus flytrap" mechanism as suggested for the metabotrophic glutamate receptors, whereas V1Rs, resembling ORs in the olfactory epithelium, may recognize volatile odorants presented by members of the lipocalin family, e.g. odorant-binding proteins (44). The result of the present study demonstrating that G i coexpressed with V1Rs is activated by volatile ligands, whereas G o , coexpressed with V2Rs is activated upon stimulation with the protein pheromone ␣ 2u -globulin, provide the first experimental evidence that each of the two VNO-receptor families is indeed activated by a distinct class of ligands. The emerging concept that V1Rs are activated by small volatile compounds, whereas V2Rs are specialized for interacting with protein pheromones can now be tested in experimental studies analyzing heterologously expressed receptors.