olf is necessary for coupling D1 and A2a receptors to adenylyl cyclase in the striatum


Address correspondence and reprint requests to Denis Hervé, U536, Institut du Fer à Moulin, 17 rue du Fer à Moulin, Paris 75005, France. E-mail: denis.herve@infobiogen.fr


In the brain, dopamine and adenosine stimulate cyclic AMP (cAMP) production through D1 and A2a receptors, respectively. Using mutant mice deficient in the olfactory isoform of the stimulatory GTP-binding protein α subunit, Gαolf, we demonstrate here the obligatory role of this protein in the adenylyl cyclase responses to dopamine and adenosine in the caudate putamen. Responses to dopamine were also dramatically decreased in the nucleus accumbens but remained unaffected in the prefrontal cortex. Moreover, in the caudate putamen of mice heterozygous for the mutation, the amounts of Gαolf were half of the normal levels, and the efficacy of dopamine- and CGS 21680 A2 agonist-stimulated cAMP production was decreased. Together, these results identify Gαolf as a critical parameter in the responses to dopamine and adenosine in the basal ganglia.

Regulation of cyclic AMP (cAMP) levels in striatal neurones by stimulation of D1-like dopamine receptors plays a major role in the basal ganglia. Cyclic AMP-activated signalling pathways are involved in the acute effects of dopamine, as well as in the control of gene expression and synaptic plasticity, and in the long-lasting changes induced by repeated administration of drugs of abuse (see Greengard et al. 1999; Berke and Hyman 2000 for recent reviews). Adenosine A2a receptors are a second type of receptor positively coupled to adenylyl cyclase, which is highly expressed in the striatum (Premont et al. 1977; Schiffmann et al. 1991). These receptors have been proposed as a potential therapeutic target for treating psychosis and Parkinson's disease (Ledent et al. 1997; Svenningsson et al. 1999). Although sensitive techniques reveal some overlap (Surmeier et al. 1996), D1 or A2a receptors appear to be expressed abundantly in different populations of striatal cells with different peptide contents and different projection sites (Schiffmann and Vanderhaeghen 1993; Hersch et al. 1995; Le Moine and Bloch 1995; Svenningsson et al. 1999). In spite of its functional importance, the mechanism of coupling of D1 and A2a receptors to adenylyl cyclase is still poorly characterized.

Two highly homologous α subunits of heterotrimeric GTP-binding proteins (G proteins) are known to activate adenylyl cyclase: Gαs, the ubiquitous isoform, is found in most cell types, whereas Gαolf has been discovered in the olfactory epithelium, where it is necessary for olfaction (Belluscio et al. 1998; Jones and Reed 1989). Gαs is present in much lower quantities than Gαolf in the striatum (Drinnan et al. 1991; Herve et al. 1993). Although the observation of high Gαolf expression in striatal neurones bearing D1 and A2 receptors has led to the hypothesis that Gαolf may couple D1 and A2 receptors to adenylyl cyclase (Herve et al. 1993; Kull et al. 2000), direct proof for this role of Gαolf is lacking. The recent generation of a mutant mouse line with an invalidated olf gene provides a means by which to test this hypothesis directly (Belluscio et al. 1998), and the first aim of the present study was to demonstrate that the D1 and A2 receptor signalling are deficient in basal ganglia of these mice. In particular, deficient D1 signalling in the Gαolf knockout mice would explain their loss of behavioural responses to cocaine and a D1 agonist, as well as of cocaine-induced c-Fos expression in the caudate putamen (Zhuang et al. 2000).

We have found using homozygous Gαolf knockout mice that Gαolf is necessary for the stimulation of adenylyl cyclase by D1 and A2a receptors in the caudate putamen and the nucleus accumbens. In addition, the lower efficacy of the D1 and A2a receptors observed in the heterozygous Gαolf knockout mice indicated that Gαolf levels represent an important factor for determining the responsiveness of D1 and A2a receptors in vivo.

Materials and methods

Mutant mice

Pairs of heterozygous mice with a disrupted gene of Gαolf (Belluscio et al. 1998) and a hybrid 129 and C57Bl/6 genetic background were crossed. The genotype of progeny was determined by Southern blotting after a digestion of genomic DNA by HindIII using a Gαolf probe, corresponding to the 1.2-kb Pst/Kpn fragment of mouse olf gene located 0.5 kb upstream from the ATG codon translational start site (Belluscio et al. 1998).

Gα proteins antibodies

A rabbit was immunized against a recombinant Gαolf fusion protein containing the full-length rat Gαolf protein and an additional sequence at the N-terminus side including six histidines and a factor X digestion site. The rabbit antiserum was purified by affinity on a Gαolf column and by adsorption on a Gαs column in order to obtain specific antibodies against Gαolf (SL48SP). Specific antibodies against Gαs (SL22AP) were raised against a Gαs-selective peptide and affinity-purified on a peptide column (Penit-Soria et al. 1997). The antibodies against Gαi1, Gαi2 and Gαo were mouse monoclonal antibodies from commercial sources (clones R4.5 and L5.6 and MAB3073).

Immunoblot analysis

Wild-type and mutant mice (2-month-old, age-matched) were killed by decapitation and their brains were immediately dissected out from the skull and frozen on dry ice. Microdiscs of tissue were punched out from frozen slices (500-µm thick) within the striatum using a stainless steel cylinder (1.4 mm diameter). Samples were homogenized in 1% SDS, equalized for their protein content and analysed by western blot as described previously (Herve et al. 1993). The antibody dilutions were 1/1000, 1/500, 1/500, 1/500 and 1/3000 for antibodies against Gαolf, Gαs, Gαi1, Gαi2 and Gαo, respectively. Antibodies were revealed by the peroxidase/chemiluminescence method and autoradiography. The antibodies were stripped one or two times to detect other antigens on the same membranes (Herve et al. 1993).

Adenylyl cyclase assays

Mouse brains were sectioned in 300-µm-thick slices in Ca2+-free artificial cerebrospinal fluid using Vibroslice apparatus. Tissue microdiscs were punched out from caudate putamen, nucleus accumbens and prefrontal cortex using a stainless steel cylinder and homogenized at 1 mg of protein/mL (2 mg/mL for the prefrontal cortex) in a buffer containing 2 mm Tris-maleate (pH 7.2), 2 mm EGTA and 300 mm sucrose, using a Potter-Elvehjem apparatus. Adenylyl cyclase activity was measured by the conversion of α-[32P]-ATP into cyclic [32P]-AMP as described previously (Premont et al. 1977; Herve et al. 1989). When A2 receptor-linked adenylyl cyclase was measured, papaverine (0.1 mm) was used to inhibit phosphodiesterases instead of theophylline which antagonizes adenosine receptors, and adenosine deaminase (0.4 U/mL) was added to eliminate endogenously produced adenosine from the incubation medium (Premont et al. 1977). Adenylyl cyclase activities were measured in the presence of vehicle or various concentrations of dopamine and CGS21680. The adenylyl cyclase activity was expressed in pmoles of cAMP produced per min and per mg protein.

3H]SCH23390 and [3H]CGS21680 binding

Series of three coronal brain sections (20 µm) containing the striatum were thaw-mounted onto SuperFrost microscope slides. The binding of [3H]SCH23390 and [3H]CGS21680 to tissue sections was performed as previously described in order to analyse D1 and A2 receptors, respectively (Savasta et al. 1988; Jarvis and Williams 1989). Tissue sections were incubated with 2.5 nm [3H]SCH23390 (91 Ci/mmol) or with 5 nm [3H]CGS21680 (47 Ci/mmol). Non-specific binding was determined from adjacent brain sections in the presence of 1 µm cold SCH23390 or 10 µm cold CGS21680. After 5 washes (2 min each) in ice-cold Tris buffer (50 mm, pH = 7.4), the brain sections were wiped off the slides using Watman GF/B glass fibre filters which were counted for radioactivity in a liquid scintillation counter. The specific binding was calculated as the difference between the total binding and the non-specific binding and was expressed in fmoles of bound ligand/section.

Results and Discussion

Striatal Gα protein levels in homozygous and heterozygous Gαolf knockout mice

In order to examine the role of Gαolf in the coupling of D1 or A2a receptors to adenylyl cyclase, we used mutant mice, homozygous or heterozygous, for a null mutation of the olf gene (Belluscio et al. 1998). In homozygous Gαolf knockout mice, the Gαolf protein was undetectable in the caudate putamen (Fig. 1). In heterozygous mutants, the striatal levels of the Gαolf protein were approximately half of those in wild-type littermates (52 ± 5%, n = 9, p < 0.01, Student's t-test see Fig. 1). This showed the lack of compensation for the haplo-insufficiency of the olf gene. In homozygous and heterozygous mice, no major change was observed in the levels of Gαi1, Gαi2 and Gαo proteins. The slight increase in Gαs protein levels seen in Gαolf (–/–) mice (about 40% above control levels) cannot compensate the absence of Gαolf because the levels of Gαolf are about 10-fold greater than those of Gαs in the striatum (Herve et al. 1993). The lack of alteration of several striatal markers confirms that the striatum of Gαolf (–/–) mice did not show obvious alterations of its gross anatomy (Belluscio et al. 1998).

Figure 1.

Levels of G proteins in the caudate putamen of Gαolf knockout mice. Western blot analysis of caudate putamen homogenates from mice homozygous (–/–) or heterozygous (+/–) for a null mutation of the olf gene, and from their wild-type littermates (+/+). Blots were incubated with antibodies that recognize specifically Gαolf, Gαs, Gαi1, Gαi2 and Gαo, followed by chemiluminescence detection. The specific optical density of labeling with Gαolf, Gαs, Gαi1, Gαi2 and Gαo antibodies were 0.44, 0.32, 0.25, 0.14 and 0.31, respectively.

olf is essential for the coupling of D1 and A2a receptors to adenylyl cyclase in basal ganglia

The basal cAMP production in striatal membranes depends on Gαolf, as it was decreased in striatal samples from Gαolf (–/–) mice (− 58 ± 3%, p < 0.05, paired Student's t-test, n = 3) (Fig. 2). This basal activity has an unclear origin but could result from the background activity of receptors without bound ligands (Adie and Milligan 1994). The deficiency in Gαolf would uncouple these receptors from adenylyl cyclase. Alternatively, the lack of Gαolf could also down-regulate the levels and/or the activity of adenylyl cyclase in cells. It is noteworthy that basal adenylyl cyclase activity was dramatically reduced in mutant S49 lymphoma cells lacking Gαs (Harris et al. 1985).

Figure 2.

Adenylyl cyclase activity in the caudate putamen of Gαolf knockout mice. Effects of dopamine and an A2 agonist, CGS21680, on cAMP formation in caudate putamen membranes from Gαolf (–/–), Gαolf (+/–) and wild-type (+/+) mice. The D1 or A2 responses were determined by measuring the adenylyl cyclase activity in the absence or the presence of various concentrations of dopamine or CGS21680.

In Gαolf knockout animals, the stimulation of adenylyl cyclase activity induced by increasing concentrations of dopamine and an A2 agonist, CGS21680, was even more reduced than the basal activity (Fig. 2). The effects of dopamine were related to D1 receptor stimulation as similar results were obtained with a specific D1 agonist, 6-chloro-APB (SKF82958, data not shown). The deficient responses to dopamine and CGS21680 were not caused by an alteration of D1 and A2a receptors in the striatum of Gαolf knockout mice. In striatal sections of these mice, the specific binding of [3H]-SCH23390 and [3H]-CGS21680, ligands of D1 and A2 receptors, respectively, was found to be similar to that observed in sections of wild-type littermates (13.5 ± 0.7 and 12.8 ± 0.3 fmoles of [3H]-SCH23390/section for wild-type and Gαolf knockout mice, respectively; 2.4 ± 0.5 and 1.7 ± 0.3 fmoles of [3H]-CGS21680/section for wild-type and Gαolf knockout mice, respectively). These results were in agreement with a recent study showing a normal density of D1 receptors in Gαolf-deficient mice (Zhuang et al. 2000).

Previous studies had shown that the striatal cells bearing the D1 and A2a receptors expressed much more Gαolf than Gαs but could not exclude that small amounts of Gαs were sufficient to couple D1 or A2a receptors to adenylyl cyclase (Herve et al. 1993; Kull et al. 2000). Our present results demonstrate that the coupling of D1 and A2a receptors to adenylyl cyclase occurs predominantly through Gαolf in the caudate putamen. The small residual responses to dopamine and CGS21680 observed in Gαolf-deficient mice (Fig. 3) could correspond to a small fraction of D1 and A2a receptors linked to Gαs.

Figure 3.

Maximal dopamine and CGS21680 responses on adenylyl cyclase activity in various brain regions. These responses correspond to the difference between the activity obtained in the presence and in the absence of 10−4 m dopamine or 3 × 10−6 m CGS21680. The basal activity (in pmoles cAMP formed/min/mg protein) was 112 ± 29, 59 ± 16 and 42 ± 14 in the caudate putamen for Gαolf (+/+), (+/–) and (–/–) mice, respectively, 44 ± 12, 32 ± 8 and 37 ± 6 in the nucleus accumbens for Gαolf (+/+), (+/–) and (–/–) mice, respectively, and 48 ± 19, 35 ± 12 and 62 ± 11 in the prefrontal cortex for Gαolf (+/+), (+/–) and (–/–) mice, respectively. Statistical analysis was performed with anova (p < 0.04) followed by a paired Student's t-test (*p < 0.05, significantly different from wild-type littermates).

In the nucleus accumbens, in which the expression pattern of Gαolf and Gαs is similar to that observed in the caudate putamen, dopamine responses were also dramatically reduced in homozygous Gαolf knockout mice (Fig. 3). In contrast, in the prefrontal cortex in which Gαolf does not appear to be expressed, no significant variation of dopamine-stimulated cAMP production was observed (Fig. 3). Thus, the coupling of dopamine receptors to adenylyl cyclase depends on Gαolf in the dorsal striatum and in the nucleus accumbens, but not in the prefrontal cortex. This demonstrates that there is no obligatory association between D1 receptors and Gαolf, and also that alterations in Gαolf or Gαs properties would have different consequences on the mesostriatal and mesocortical dopaminergic pathways. Recently, the behavioural responses to cocaine and a D1 agonist, as well as the cocaine-induced c-Fos expression, were found to be abolished in Gαolf-knockout mice (Zhuang et al. 2000). Our results support strongly that these effects are a result of the impairment of D1 receptor signalling in the caudate putamen and in the nucleus accumbens.

olf levels are a limiting factor in D1 and A2a responses

In striatal samples from Gαolf (+/–) mice, the basal adenylyl cyclase activity was significantly reduced (− 45 ± 6%, n = 5, p < 0.05, paired Student's t-test). Moreover, the stimulation of cAMP production by dopamine or CGS21680 was intermediate between those observed in wild-type mice and those in homozygous mutant mice (Fig. 2). In heterozygous animals, it is important to note that there was no change in the levels of D1 and A2a receptors measured by [3H]-SCH23390 and [3H]-CGS21680 binding, respectively (data not shown).

Our results in heterozygous Gαolf mutant mice demonstrate that when the concentrations of Gαolf were reduced to half of the normal levels, an overall reduction in D1- and A2-stimulated cAMP production was observed. Therefore, it can be predicted that the reduction in Gαolf levels that could occur in physiological or pathological circumstances will impair the dopamine and adenosine neurotransmission in the basal ganglia. Such a situation exists in the substantia nigra in which Gαolf levels are 5-fold lower by comparison with those in the caudate putamen (Herve et al. 1993). As expected, dopamine is 5-fold less efficient to raise cAMP production despite the presence of concentrations of D1 receptors and adenylyl cyclase similar to those in the caudate putamen (Herve et al. 1993).

Several groups have reported an increase in dopamine-stimulated cAMP production in the striatum of animals bearing lesions of dopamine neurones, a model of Parkinson's disease (Herve et al. 1989). This observation could not be explained by an increase in D1 receptors, which were unchanged (Savasta et al. 1988). However, the lesion of dopamine neurones increased Gαolf levels (Mishra et al. 1974; Herve et al. 1993; Penit-Soria et al. 1997), suggesting that elevated cellular Gαolf concentrations account for the enhanced dopamine-stimulated cAMP formation. As increased (lesion of dopamine neurones) and reduced (substantia nigra and heterozygous Gαolf mutant mice) Gαolf levels affect adenylyl cyclase responses in the opposite direction, it is tempting to propose that Gαolf concentrations are a limiting factor for D1 signalling in the caudate putamen.

Altogether, these data demonstrate that Gαolf is critical for basal, as well as D1- and A2a-stimulated adenylyl cyclase activities in the striatum. Moreover, the results in heterozygous mice which show that there is a proportional decrease in Gαolf protein levels, and in basal and stimulated adenylyl cyclase activity, suggest that the availability of Gαolf is a limiting factor for cAMP production in these cerebral regions.


We thank J. Glowinski for his valuable support of this work, L. Belluscio and R. Axel, for kindly allowing the use of Gαolf knockout mice, and R. Urbe and H. Ibrahim for their help in breeding the animals. This work has been supported by Institut National de la Santé et de la Recherche Médicale and by a grant from MILDT addiction program.