Neuronal GABAB receptors regulate calcium and potassium currents via G-protein-coupled mechanisms and play a critical role in long-term inhibition of synaptic transmission in the CNS. Recent studies have demonstrated that assembly of GABAB receptor GABABR1 and GABABR2 subunits into functional heterodimers is required for coupling to potassium channels in heterologous systems. However whether heterodimerization is required for the coupling of GABAB receptors to effector systems in neurons remains to be established. To address this issue, we have studied the coupling of recombinant GABAB receptors to endogenous Ca2+ channels in superior cervical ganglion (SCG) neurons using nuclear microinjection to introduce both sense and antisense expression constructs. Patch-clamp recording from neurons injected with both GABABR1a/1b and GABABR2 cDNAs or with GABABR2 alone produced marked baclofen-mediated inhibition of Ca2+ channel currents via a pertussis toxin-sensitive mechanism. The actions of baclofen were blocked by CGP62349, a specific GABAB antagonist, and were voltage dependent. Interestingly, SCGs were found to express abundantly GABABR1 but not GABABR2 at the protein level. To determine whether heterodimerization of GABABR1 and GABABR2 subunits was required for Ca2+inhibition, the GABABR2 expression construct was microinjected with a GABABR1 antisense construct. This resulted in a dramatic decrease in the levels of the endogenous GABABR1 protein and a marked reduction in the inhibitory effects of baclofen on Ca2+ currents. Therefore our results suggest that in neurons heteromeric assemblies of GABABR1 and GABABR2 are essential to mediate GABAergic inhibition of Ca2+ channel currents.
GABABreceptors play a critical role in long-term inhibition of synaptic transmission in the CNS (Bowery, 1993; Mott and Lewis, 1994). Inhibition is mainly achieved via modulation of neurotransmitter release from presynaptic terminals and hyperpolarization of postsynaptic membranes. Although activation of inwardly rectifier potassium channels (GIRKs) is thought to be responsible for a GABAB-mediated membrane hyperpolarization, inhibition of presynaptic N-type Ca2+channels is thought to be responsible for modulation of neurotransmitter release (Mott and Lewis, 1994; Kaupmann et al., 1998;Takahashi et al., 1998). The regulation of GABABreceptor function has been implicated in cognition enhancement and induction of long-term potentiation (Davies et al., 1991; Olpe et al., 1993). In addition, involvement of the GABABreceptor in a number of diseases of the CNS such as epilepsy, anxiety, depression, and cognitive deficits make it an attractive target for therapeutic agents (Bittiger et al., 1993; Kerr and Ong, 1995;Dichter, 1997).
The first GABAB receptors, GABABR1a and GABABR1b, were cloned from a mouse cortical and cerebellar cDNA library (Kaupmann et al., 1997). They exhibited a low affinity for agonists and did not couple efficiently to neuronal Ca2+channels when expressed heterologously in sympathetic neurons (Couve et al., 1998). Subcellular distribution of epitope-tagged GABABR1 demonstrated that recombinant receptors failed to reach the cell surface and were retained in intracellular compartments when expressed in a variety of neuronal and non-neuronal cell types. These results provided an explanation for the lack of function of recombinant GABABR1 receptors in coupling to other effector systems such as K+ channels (Couve et al., 1998). Recently GABABR2, a second receptor ∼35% identical to GABABR1, has been identified by several laboratories (Jones et al., 1998; Kaupmann et al., 1998; White et al., 1998; Kuner et al., 1999). Unexpectedly for a G-protein-coupled receptor, heteromeric assembly of GABABR1 and GABABR2 has been shown to be necessary for coupling to adenylyl cyclase and activation of GIRKs inXenopus oocytes and human embryonic kidney 293 (HEK293) cells (Jones et al., 1998; Kaupmann et al., 1998; White et al., 1998;Kuner et al., 1999). However, it has yet to be determined whether heterodimer formation is required for coupling to their effector systems (such as Ca2+ channels) in neurons.
Here we demonstrate that the heteromeric assembly of GABABR1 and GABABR2 receptor subunits is required to allow GABABreceptor-induced inhibition of Ca2+currents in rat superior cervical ganglion (SCG) neurons.
MATERIALS AND METHODS
Neuron preparation and cDNA injection. Neuron isolation and injection procedures have been described previously in detail (Filippov et al., 1997, 1998; Couve et al., 1998). Briefly, single SCG neurons were dissociated from 15- to 19-d-old rats and plated on laminin-coated glass coverslips. Five hours after plating, neurons were microinjected into the nucleus with plasmids carrying cDNAs of the GABAB receptor subunits. Plasmids were stored at 4°C as 1 μg/μl stock solutions in 10 mm TRIS and 1 mm EDTA, pH 8, and injected at a final pipette concentration of 0.15–0.5 μg/μl. (Experimental data were indistinguishable over this range of concentrations for GABABR1a and GABABR1b.) Neurons were co-injected with cDNA for the “enhanced” S65T mutant of jellyfish green fluorescent protein (GFP) at concentrations of 0.01–0.5 μg/μl for later identification of the cells with successfully expressed cDNAs. For controls, neurons were injected with GFP cDNA alone or not injected at all. After injections, cells were incubated for 14–48 hr in a humidified incubator (5% CO2/95% O2) at 37°C. Injected neurons with successfully expressed cDNAs were identified as bright fluorescent cells using an inverted microscope (Diaphot 200; Nikon) equipped with an epifluorescent block N B2E (Nikon). Electrophysiological recordings were routinely made 16–20 hr after injection at room temperature (20°C). In some experiments, recordings were made 40 hr after injections. Where indicated, neurons were incubated for 12–16 hr with pertussis toxin (PTX; 0.5 μg/μl) added to the culture media.
Plasmids. A cytomegalovirus (CMV) promoter-based mammalian expression vector containing a myc-tagged version of the rat GABABR1a receptor has been described previously (Couve et al., 1998). The GABABR1a antisense cDNA was obtained by subcloning the sequence of the complete receptor in the opposite orientation with its 3′ end facing the CMV promoter. A GABABR2 receptor plasmid was obtained by subcloning an ∼3.5 kb XbaI containing the sequence of the rat GABABR2 receptor from pC1-Neo/GABABR2 (kindly provided by Dr. Bernhard Bettler) into a CMV promoter-based mammalian expression vector.
Antibodies. A guinea pig pan antibody directed against the C-terminal domain of GABABR1a was obtained from Chemicon International. A rabbit antibody specific for GABABR1a was raised against the peptide CHPPWEGGIRYRGLTRDQVK from the N-terminal domain of GABABR1a conjugated with keyhole limpet hemocyanin and was immunopurified using the same peptide. A rabbit antibody directed against the C-terminal domain of GABABR2 was kindly provided by Dr. Bernhard Bettler. MAP2 monoclonal antibody was obtained from Sigma (St. Louis, MO). Monoclonal protein disulfide isomerase (DPI) antibody was purchased from Stress Gen. Synaptophysin antibody was kindly provided by Dr. Dan Cutler. Texas Red- and fluorescein-conjugated anti-rabbit and anti-mouse secondary antibodies were purchased from Jackson ImmunoResearch (West Grove, PA).
Ca2+ channel current recording. Currents through voltage-gated Ca2+ channels were recorded using the conventional whole-cell patch-clamp method as described previously (Caulfield et al., 1994). Cells were superfused (20–25 ml/min) with a solution consisting of 120 mmtetraethylammonium chloride, 3 mm KCl, 1.5 mmMgCl2, 5 mmBaCl2, 10 mm HEPES, 11.1 mm glucose, and 0.5 μm tetrodotoxin. The pH was adjusted to 7.35 with NaOH. Patch electrodes (2–3 MΩ) were filled with a solution containing 110 mm CsCl, 3 mm MgCl2, 40 mm HEPES, 3 mm EGTA, 2 mm Na2ATP, and 0.5 mm Na2GTP, with pH adjusted to 7.4 with CsOH. Neurons were voltage-clamped using a discontinuous (“switching”) amplifier (Axoclamp 2B) sampling voltage at 6–8 kHz (50% duty cycle). Commands were generated via a Digidata 1200 interface using “Clamp 6” computer software (Axon Instruments, Foster City, CA). Ca2+ channel currents were routinely evoked every 20 sec with a 100 msec depolarizing rectangular test pulse to 0 mV from a holding potential of −90 mV. Currents were digitized and stored on a computer for later analysis using Clamp 6 software. Ca2+channel current amplitudes were measured isochronally 10 msec from the onset of the rectangular test pulse, i.e., near the peak of the control current. To eliminate leak currents, Co2+was substituted for Ba2+ in the external solution at the end of each experiment, blocking all Ca2+ channel currents, and the residual current was digitally subtracted from the corresponding currents in the Ba2+ solution. As reported previously (Filippov et al., 1997, 1998), currents were primarily N-type with negligible contribution by L-type channels.
Data were presented as means ± SEM as appropriate. Student's test (unpaired) was applied to determine statistical significance. The difference was considered significant if p ≤ 0.05.
Baclofen and norepinephrine were purchased from Sigma. PTX was obtained from Porton Products (Dorset, UK).
Protein extracts, immunoblots, and immunofluorescence.Immunofluorescence and immunoblots were performed as described previously (Couve et al., 1998). For preparation of brain membranes, fresh rat brains were homogenized in 10 vol of 5 mmTris-Cl, pH 7.4, 0.32 m sucrose, and a cocktail of protease inhibitors with 50 strokes using a glass and Teflon homogenizer. The suspension was centrifuged at 1000 × g for 15 min. The pellet was discarded, and the resulting supernatant was centrifuged at 17000 × g for 30 min. The pellet was washed twice in 50 mm Tris-Cl, pH 7.4, and resuspended at ∼5 mg/ml in the same buffer. Aliquots were frozen and stored at −20°C until used. Extracts from superior cervical ganglia were prepared by homogenizing ganglia in RIPA buffer containing 50 mm Tris-Cl, 1 mm EDTA, 2 mm EGTA, 10 mmNa+ pyrophosphate, 1 mm Na+orthovanadate, 50 mm NaF, 150 mm NaCl, 1% NP-40, 0.5% deoxychoic acid 0.1% SDS, and a cocktail of protease inhibitors. All extracts were stored at −20°C and mixed with SDS sample buffer before loading onto SDS-PAGE.
Heterologously expressed GABABR1 and GABABR2 or GABABR2 but not GABABR1 couple to N-type Ca2+ channels in sympathetic neurons
It has been reported previously that baclofen, a well characterized GABAB agonist, produced only a minor inhibition of Ca2+ channel current in neurons injected exclusively with GABABR1 cDNA, for greater than that (12.7 ± 3.2%) in control neurons injected with GFP cDNA (Couve et al., 1998). This lack of functional inhibition of Ca2+ currents in SCG is in agreement with the endoplasmic reticulum retention of GABABR1 after homomeric expression in neurons. In neurons injected with cDNAs for GABABR1a and GABABR2, bath application of 50 μm baclofen produced a substantial and reversible inhibition of Ca2+ channel current (recorded using Ba2+ as the charge carrier) (Fig. 1 A). This inhibition was accompanied by a current-onset slowing similar to that reported after activation of many other endogenous or heterologously expressed G-protein-coupled receptors in SCG neurons (Hille, 1994; Ikeda et al., 1995; Filippov et al., 1998, 1999). Bath application of 1 μm CGP62349, a specific antagonist of GABAB receptors (Billinton et al., 1999), blocked the effect of baclofen (Fig. 1 C,D). Interestingly, washing out of the agonist produced an initial over-recovery of the current followed by a slow return to the control level [as reported for some other receptors (see Meza et al., 1999)].
To study the requirement of heterodimer formation in the ability of recombinant GABAB receptors to couple to Ca2+ channels, GABABR2 cDNA, in the absence of GABABR1, was injected into SCG neurons. Surprisingly, Ca2+ channel currents were also inhibited in neurons injected exclusively with GABABR2. This effect was indistinguishable from that obtained after coexpression of GABABR1 and GABABR2 (Fig. 1 B). Thus, the mean inhibition of Ca2+ channel current (measured 10 msec after onset of the test pulse) by 50 μm baclofen was 56.1 ± 3.2% for neurons expressing recombinant GABABR1 and GABABR2 and 52.3 ± 4.8% for neurons expressing recombinant GABABR2 exclusively (Fig.1 D).
GABAB-mediated inhibition of Ca2+current is strongly voltage dependent, suggesting involvement of the G-protein βγ subunit
Slowing of the current onset by baclofen in neurons with heterologously expressed GABABR1 and GABABR2 or GABABR2 indicated that GABAB receptor-mediated inhibition of Ca2+ channel current was voltage dependent and relieved during depolarization (cf. Bean, 1989), as reported previously for other G-protein-coupled receptors. Indeed, a double-pulse protocol demonstrated that inhibition of Ca2+ current produced by baclofen was greatly reduced after a 120 mV depolarizing prepulse (Fig.2 A). The prepulse also abolished the slowing of the current onset produced by baclofen. Identical results were obtained with both heterologously expressed GABABR1 and GABABR2 or GABABR2 receptors. Inhibition was reduced from 58.7 ± 2.4 to 15.2 ± 4.7% in neurons with heterologously expressed GABABR1 and GABABR2 and from 50.2 ± 3.8 to 14.1 ± 1.8% in neurons expressing GABABR2 (Fig.2 B). A similar effect was reported for the modulation of Ca2+ current by endogenous GABAB receptors in sensory neurons and in calyx nerve terminals of the mammalian auditory brainstem (Grassi and Lux, 1989; Isaacson, 1998). Similar voltage-dependent effects have been generally interpreted as produced by a direct interaction of G-protein βγ subunits with an α subunit of the Ca2+ channel (Dolphin, 1995;Herlitze et al., 1996; Ikeda, 1996; Jones and Elmslie, 1997; Delmas et al., 1998a,b).
A PTX-sensitive G-protein mediates Ca2+ current inhibition after activation of recombinant GABABR1 and GABABR2 or GABABR2 receptors
A strong voltage dependency of Ca2+channel current inhibition by baclofen in our experiments suggested that the recombinant GABAB receptors coupled to Ca2+ channels via PTX-sensitive G-proteins (for review, see Hille, 1994). Therefore, to determine the identity of the Gα-protein that mediates the Ca2+current inhibition in SCG neurons, responses to baclofen were determined after pretreatment with PTX. Indeed, after PTX pretreatment (500 ng/ml), baclofen no longer inhibited Ca2+ channel currents in neurons expressing recombinant GABABR1 and GABABR2 receptors. PTX sensitivity was identical when cells expressing recombinant GABABR2 were studied. These observations are summarized in Figure3 B. Taken together they clearly indicate that PTX pretreatment prevents Ca2+ channel current inhibition in neurons expressing recombinant GABABR1 and GABABR2 or GABABR2 alone (Fig. 3). These results strongly suggest that only PTX-sensitive G-proteins mediate Ca2+ current inhibition after activation of recombinant GABAB receptors. The complete disappearance of the baclofen response in the presence of PTX was unique to GABAB receptors because the same concentration of PTX was not capable of totally abolishing the response of endogenous α-adrenergic receptors to norepinephrine (Fig.3 A). The small PTX-insensitive effect of norepinephrine is in agreement with a minor involvement of PTX-insensitive G-proteins in adrenergic receptor-mediated function (Hille, 1994; Delmas et al., 1999).
Heteromeric assembly of GABABR1 with GABABR2 is required to couple effectively to Ca2+channels in neurons
The results presented above showing the strong inhibition of Ca2+ channel currents by GABABR2 cDNA alone were of interest because heterologous expression of GABABR2 is not sufficient to produce functional coupling to adenylyl cyclase or inward rectifier potassium channels in Xenopus oocytes or HEK293 cells (Jones et al., 1998; Kaupmann et al., 1998; White et al., 1998). One possible explanation is that SCG neurons express endogenous GABABR1 subunit that is not functionally coupled to Ca2+ channels but that, on expression of GABABR2, coassembles with recombinant GABABR2 to constitute a functional heteromeric receptor indistinguishable from a recombinant dimer. Several experiments were designed to test this hypothesis.
First, the expression of GABABR1 and GABABR2 was analyzed in SCG neurons via immunoblotting with specific antibodies against these proteins. A specific antibody was raised against a peptide in the N-terminal domain of the GABABR1a receptor, and expression of recombinant GABABR1 was used to test the specificity of this antisera. In agreement with previous observations, the antibody recognized GABABR1a as two specific bands in cell lysates derived from transfected COS cells. A lower band of ∼120 kDa represents the monomeric receptor, and an ∼250 kDa band presumably corresponds to receptor aggregates (Fig.4 A, lane 1). In contrast, a single ∼120 kDa band is detected in crude brain membranes in agreement with previous observations (Fig. 4 A, lane 2) (Benke et al., 1999). Interestingly, low levels of an ∼120 kDa band corresponding to the GABABR1a protein were detected in SCG extracts. Unexpectedly, an aggregate form of the GABABR1a protein was predominant in these cell lysates (Fig. 4 A, lane 3). This band matches the ∼250 kDa form of GABABR1a observed in transfected COS cells. The specificity of the GABABR1a antibody was demonstrated by competition with a GABABR1 peptide before immunoblotting (Fig. 4 A, lane 4).
Second, to determine the expression of GABABR2 in SCG, immunoblots with antibodies against GABABR2 were performed. A single band of ∼110 kDa was detected in crude brain membranes (Fig. 4 B, top). SCG extracts showed no detectable levels of GABABR2 (Fig.4 B, top). An immunoblot with anti-synaptophysin antibodies indicates that protein levels and the integrity of cell extracts were similar between brain and SCG preparations (Fig. 4 B, bottom). These observations were confirmed by detection of GABABR1 (see Fig. 6) but not GABABR2 (data not shown) by immunofluorescence in SCG neurons.
To investigate further the inhibition of Ca2+ current by GABAB receptor subunits in neurons, the effects of baclofen on expressed GABABR1 and GABABR2 or GABABR2 receptors were studied in detail using full dose–response curves. Baclofen inhibited Ca2+ currents more effectively and with higher potency in cells expressing GABABR1 and GABABR2 receptors than in cells expressing only heterologous GABABR2 receptors (Fig.5 A). The IC50 for inhibition was 0.09 ± 0.01 μm for GABABR1 and GABABR2 and approximately four times higher (0.38 ± 0.05 μm) for GABABR2. To confirm these observations using a different methodology, SCG neurons were injected either with GABABR2 cDNA or with an equal mixture of GABABR2 cDNA and GABABR1 antisense cDNA to reduce the expression of endogenous GABABR1 protein. As expected, immunofluorescence with GABABR1 antibodies revealed detectable levels of GABABR1 in all sympathetic neurons, and levels were reduced in cells injected with GFP and GABABR1 antisense cDNA (Fig. 5 C,top left, arrowhead). Furthermore, coexpression of the sense and antisense cDNAs of GABABR1 resulted in a significant decrease in expression of recombinant GABABR1 protein (data not shown). Control cells injected with GFP alone showed no difference in expression of GABABR1 (Fig. 5 C, bottom left, arrowhead). In accordance with these observations, inhibition of Ca2+ channel currents in neurons injected with GABABR1 antisense cDNA was significantly lower than inhibition in control neurons for all effective doses of baclofen (Fig. 5 B).
Taken together, the results presented above indicate that SCG neurons contain detectable levels of GABABR1a and suggest that responses to baclofen in neurons injected with GABABR2 result from activation of a heterodimer formed between endogenous GABABR1a and recombinant GABABR2 receptors. Interestingly, trace levels of GABABR2 were observed occasionally in SCG extracts. This might generate trace levels of endogenous GABABR1/R2 dimer and hence provide an explanation for the low response of control SCG neurons to baclofen (Fig. 1 D).
Endogenous GABABR1 is predominantly localized to the endoplasmic reticulum in SCG neurons
As reported previously, recombinant GABABR1 receptors are retained in an intracellular compartment when expressed in neuronal and non-neuronal cell types (Couve et al., 1998). To determine whether endogenous GABABR1 receptors failed to reach the cell surface in SCG neurons, immunofluorescence with GABABR1 antibodies was performed. The GABABR1 receptor was found mainly restricted to the cell body of SCG neurons (Fig.6 A, left). Detection of MAP2, a somatodendritic marker, shows a different distribution pattern with predominant staining in cell projections (Fig. 6 A). The pattern difference is still apparent after the levels of fluorescence have been equalized.
To determine the subcellular distribution of GABABR1, SCG neurons were stained for GABABR1 and DPI, an endoplasmic reticulum marker. DPI is present in neuronal and non-neuronal cell types from the SCG preparation (Fig. 6 B). In contrast, GABABR1 is present exclusively in neurons, and it overlaps almost entirely with DPI in these cells. These observations strongly suggest that both proteins colocalize in the endoplasmic reticulum.
GABABR1 is localized to the plasma membrane after coexpression of GABABR2
To provide additional evidence of the formation of functional, cell surface heterodimers after expression of GABABR2, SCG neurons were microinjected with GABABR1 or GABABR2 independently or with GABABR1 and GABABR2 simultaneously. As reported previously (Couve et al., 1998), a myc-tagged version of GABABR1 is retained intracellularly in SCG neurons (Fig. 7, top). In contrast, a FLAG-tagged version of GABABR2 is targeted efficiently to the cell surface (Fig. 7, middle). Consistent with our electrophysiological findings, expression of GABABR2 changes the subcellular distribution of GABABR1 from the endoplasmic reticulum to the plasma membrane (Fig. 7, bottom). These observations support the idea that GABABR2 rescues GABABR1 from intracellular compartments and strongly suggest that heterodimer formation is a requirement for GABAB receptor function.
In this report we demonstrate that coexpressed GABABR1 and GABABR2 subunits produce a functional receptor in sympathetic neurons that couples effectively to N-type Ca2+channels. Thus, heteromeric assembly of GABABR1 and GABABR2 receptor subunits appears to be necessary for coupling to all three known effectors: Ca2+ channels, adenylyl cyclase, and inward rectifier K+ channels (Jones et al., 1998; Kaupmann et al., 1998; White et al., 1998). It might be noted, however, that the IC50 for inhibition of the Ca2+ current in sympathetic neurons (∼90 nm) was appreciably lower than that required to inhibit GIRK channels in non-neuronal cells, suggesting that expression of GABAB receptor dimers in their natural (neuronal) environment allows more effective ion channel coupling.
To the degree that the present experiments allow, the mechanism by which these exogenously expressed receptors inhibit neuronal Ca2+ channels appears identical to that used by native GABAB receptors present in other neurons. Thus, inhibition was strongly reduced by pertussistoxin, implying mediation by members of the Gi/Go family of G-proteins [probably Go (see Campbell et al., 1993;Menon-Johansson et al., 1993)]. Also, the strong voltage dependence of inhibition in sympathetic neurons accords with previous observations on sensory neurons (Grassi and Lux, 1989), suggesting that it was mediated by βγ subunits of the G-protein (Dolphin, 1995; Jones and Elmslie, 1997). Thus, our observations on the expression and functional coupling of recombinant GABAB receptors in sympathetic neurons are likely to be relevant to the properties of natively expressed receptors.
Immunoblot analysis demonstrated that the predominant form of GABABR1 in SCG corresponds to a large ∼250 kDa protein band. The nature of this high-molecular weight form of GABABR1 is not known, but it is unlikely to correspond to receptor dimers because homomeric interactions have not been shown to occur (Fritschy et al., 1999). A similar protein band is normally observed in other heterologous expression systems in which the GABABR1 receptor is not functional (Couve et al., 1998). We hypothesize that this structure corresponds to nonfunctional GABABR1 receptor aggregates retained in the endoplasmic reticulum.
In this context, our experiments strongly suggest that GABABR2 forms an essential subunit of the functional receptor and that it is required both for assembly and membrane insertion of the receptor. Thus, functional membrane receptors were generated not only by coexpression of recombinant GABABR1 and GABABR2 plasmids but also on expression of exogenous GABABR2 subunits alone. We attribute this to coassembly with the native GABABR1 subunits normally present in these sympathetic neurons, because reduction of GABABR1 protein by antisense expression also reduced the response of GABABR2-expressing cells to baclofen (Fig. 5). Because GABABR1 protein was predominantly confined to the endoplasmic reticulum, this also suggests that GABABR2 plays a crucial role in trafficking and membrane insertion of the dimeric receptor. The fact that GABABR2 is localized to the plasma membrane and traffics GABABR1 in heterologous cells (White et al., 1998; Martin et al., 1999) and in SCG neurons supports these observations.
Cultured SCG neurons respond poorly to GABABagonists and are generally thought to be devoid of GABAB receptors. However, GABAB responses may appear as a result of GABABR2 expression during pathology, in certain stages of development, or under specific and local stimuli. Our findings together with the fact that GABABagonists have been shown to modify superior cervical ganglia after prolonged applications (Parducz et al., 1990), that SCG neurons contain GABAA receptors (Brown et al., 1979), and that SCG neurons are innervated by GABAergic terminals (Kasa et al., 1988) certainly suggest that GABAB receptors have the potential to function in sympathetic neurons. Nevertheless, the conditions under which GABAB receptors may be physiologically expressed remain to be determined. If this hypothesis is correct, other neuronal types might use a similar regulatory mechanism, in which expression of the GABABR2 subunit or other accessory protein determines the formation of functional receptors by trafficking existent GABABR1 from the endoplasmic reticulum to the plasma membrane.
In situ hybridization studies have revealed certain areas in the CNS where expression of GABABR1 is much greater than GABABR2 and others where GABABR1 is present and GABABR2 is not (Jones et al., 1998;Kaupmann et al., 1998; Martin et al., 1999). These observations have led to the suggestion that GABABR1 can function as a monomer or with accessory proteins other than GABABR2 in some areas of the CNS (Marshall et al., 1999). The fact that baclofen produced a small inhibition of the Ca2+ current in sympathetic neurons in the absence of exogenous GABABR2 subunits might appear to offer some support for this view. However, because trace levels of endogenous GABABR2 protein were occasionally detected in SCG extracts, we cannot exclude the possibility that the small effects of baclofen on uninjected cells might have resulted from activation of a small number of endogenous GABABR1/R2 dimers.
The question also arises whether GABABR2 subunits might form functional receptors independent of GABABR1 subunits. GABABR2 has been reported to couple negatively to adenylyl cyclase in transfected COS cells (Martin et al., 1999) and HEK293 cells (Kuner et al., 1999). However, the effectiveness of this coupling is under question because the baclofen concentrations used in these experiments were very high (300–500 μm; compared, for example, with an IC50 of ∼90 nm for Ca2+ current inhibition mediated by GABABR1 and GABABR2 in the present experiments), and GABABR2 does not couple to the inward rectifier GIRK channels coexpressed in Xenopusoocytes or HEK293 cells (Jones et al., 1998; Kaupmann et al., 1998;White et al., 1998; Kuner et al., 1999). Our experiments with GABABR1 antisense cDNA injections in SCG neurons confirm that responses apparently mediated by exogenous GABABR2 receptors are primarily (and possibly entirely) caused by formation of GABABR1/GABABR2 dimers with endogenous GABABR1 subunits. Nevertheless, more conclusive evidence is necessary to exclude the possibility that either GABABR1 or GABABR2 might be able to function independently. New subunit-specific agonists and definitive binding studies are necessary to resolve this issue.
Notwithstanding such considerations, the principal point emerging from the present experiments is that efficient coupling of GABAB receptors to Ca2+ channels in sympathetic neurons requires the heteromeric assembly of both GABABR1 and GABABR2 subunits and therefore that this is likely to be true for those endogenous GABAB receptors responsible for Ca2+ current inhibition (and resultant presynaptic inhibition) in other parts of the mammalian nervous system.
This work was supported by The Wellcome Trust and the Medical Research Council. We would like to acknowledge Drs. Dan Cuttler and Bernhard Bettler for providing synaptophysin and GABABR2 antibodies, respectively.
A.K.F and A.C contributed equally to this work.
Correspondence should be sent to Dr. Stephen J. Moss, Medical Research Council Laboratory of Molecular Cell Biology, University College London, Gower Street, London WC1E 6BT, United Kingdom. E-mail:.