βγ dimers derived from G_o and G_i proteins contribute different components of adrenergic inhibition of Ca²⁺ channels in rat sympathetic neurones

Abstract

1. Using perforated-patch recordings, we have examined the part played by endogenous G-protein subunits in the α₂-adrenoceptor-mediated inhibition of N-type Ca²⁺ currents in sympathetic neurones.

2. Two components of I_Ca inhibition by noradrenaline were recorded: a prominent, high affinity and voltage-dependent pertussis toxin (PTX)-sensitive pathway and a minor, low affinity and mostly voltage-insensitive PTX-resistant pathway.

3. PTX-sensitive inhibition was reduced by microinjection of antibodies against either Gα_oA,B or Gα_i1,2. The voltage-dependent fraction of inhibition was reduced by anti-Gα_o but not by anti-Gα_i antibody.

4. Antisense depletion of Gα_oA led to a marked reduction of noradrenaline-induced inhibition and voltage dependence. By contrast, Gα_i depletion attenuated noradrenergic modulation without affecting the voltage dependence.

5. Expression of the βγ-binding agents β-adrenergic receptor kinase 1 (C-terminus, βARK1_C-ter) or Gα_i1 with a Cys3 to Ser mutation partially prevented noradrenergic inhibition while α-transducin abolished it. Residual inhibition was mostly voltage independent in cells expressing βARK1_C-ter but was strongly reversed by depolarization in Gα_i1 Cys3Ser-expressing cells.

6. Expression of the PTX-resistant Gα_i1 Cys351Ile mutant in cells treated with PTX restored α₂-adrenoceptor inhibition. This restored inhibition was weakly reversed by depolarization. Both the degree and voltage dependence of inhibition were correlated with the level of expression of the Gα_i1 Cys351Ile subunit.

7. Our findings identify βγ dimers associated with Gα_oA and Gα_i as mediators of the PTX-sensitive α₂-adrenoceptor-mediated inhibition of N-type Ca²⁺ channels. Different βγ combinations may account for the differential voltage-dependent effects of G_o and G_i on I_Ca.

G-protein-coupled neurotransmitter receptors inhibit N-type (α_1B) Ca²⁺ channels in sympathetic neurones via two biochemically and biophysically distinct routes. The most commonly documented is that mediated by receptors that rapidly inhibit I_Ca through activation of pertussis toxin (PTX)-sensitive G-proteins (Hille, 1994). This form of inhibition occurs in a voltage-dependent manner (Bean, 1989) and is mediated by G-protein βγ dimers released from their association with Gα subunits when the latter are activated by the receptor (Ikeda, 1996; Herlitze et al. 1996; Delmas et al. 1998a,b). The other main form of inhibition, which is one order of magnitude slower, occurs in a voltage-independent manner; it results from the coupling of neurotransmitter receptors to PTX-resistant G-proteins and the activation of subsequent diffusible messenger cascades (Bernheim et al. 1991; Beech et al. 1992). In this case, it is the G-protein α subunit (e.g. Gα_q in the case of the M₁ muscarinic acetylcholine receptor inhibition), rather than the βγ complex, that appears to mediate inhibition (Delmas et al. 1998a).

α₂-Adrenoceptors represent a family of G-protein-coupled receptors that regulate effector function via activation of multiple members of G_i and G_o G-protein families (Limbird, 1988). In rat superior cervical ganglion (SCG) neurones, the PTX-sensitive G_o heterotrimer has been shown to participate in coupling α₂-adrenoceptor(s) to Ca²⁺ channel inhibition (Caulfield et al. 1994). This inhibition occurs through the fast and voltage-dependent, presumed Gβγ-mediated pathway (Herlitze et al. 1996; Ikeda, 1996; Delmas et al. 1998b). However, the generality of this latter statement is not yet entirely clear since G_i-type G-proteins are involved (along with G_o) in the PTX-sensitive adrenergic inhibition of N-type Ca²⁺ currents in chick sensory ganglion cells (Diversé-Pierluissi et al. 1995, 1997). Moreover, in this latter preparation, the roles of the α and βγ subunits seem to be reversed: the βγ subunits appear to mediate a voltage-insensitive inhibition (via stimulation of phospholipase C/protein kinase C), not a voltage-dependent inhibition like that found in SCG neurones, while GTP-γ-S-activated recombinant Gα_o produces a voltage-dependent inhibition.

Accordingly, in the present study, we have investigated further the nature of the G-proteins and G-protein subunits that mediate noradrenergic inhibition of N-type Ca²⁺ channels in SCG neurones. Our data indicate that both G_o- and G_i-type G-proteins couple α₂-adrenoceptor(s) to I_Ca with associated βγ subunits acting as the most probable final transducers. However, we find that inhibitions apparently mediated by G_oβγ and G_iβγ display differential sensitivities to voltage and to different Gβγ-sequestering agents, suggesting that the endogenous βγ subunits associated with these two proteins are not equivalent in identity and/or action.

METHODS

DNA plasmids

The cloning and specificity of plasmids generating antisense RNA to various G-protein α subunits were described previously (Abogadie et al. 1997; Delmas et al. 1998a; Haley et al. 1998). Antisense sequences of rat Gα_oA (clone 207-8) and Gα_q (clone C23-16, EMBL accession number Y17164) were subcloned into pCR3 expression vector (Invitrogen, NV Leek, The Netherlands). The antisense sequence of rat Gα_icommon (clone 50-2) was subcloned into pCR3.1. This clone corresponds to nucleotides 1045-1215 of Gα_i2 and shares approximately 80 % identity with Gα_i1 and Gα_i3. cDNA encoding the C-terminus of β-adrenergic receptor kinase 1 (βARK1 495-689) was subcloned in pCIN1 as described previously (Delmas et al. 1998b). The generation of PTX-resistant Gα_i1 subunits (Gα_i1 Cys351Ile and Gα_i1 Cys351Arg, mutated Gα_i1 subunits in which the cysteine 351 residue was replaced with isoleucine or arginine) was as detailed in Wise et al. (1997b) and Bahia et al. (1998). cDNAs encoding these mutants were subcloned into pCDNA3 (Invitrogen). Retinal Gα-transducin and the palmitoylation-negative Gα_i1 Cys3Ser mutant (Wise et al. 1997a) were subcloned into pCDNA3. Plasmids were propagated in either XL1-Blue or DH5αEscherichia coli and purified using Qiagen maxiprep columns (Hilden, Germany).

Intranuclear injection of plasmids

Plasmids were diluted into calcium-free Krebs solution (290 mosmol l⁻¹, pH 7.3) containing fluorescein isothiocyanate-conjugated dextran (FITC-dextran, 70 kDa, 0.5 %; Molecular Probes) to a final concentration of 10-600 μg ml⁻¹ and then centrifugated and filtered (0.2 μm) to remove particles. Injection electrodes were pulled with a one-stage pull using a Flaming-Brown horizontal puller (P-87, Sutter Instruments) and had a series resistance of 50-80 MΩ when loaded (2-3 μl) with the plasmid-containing solution. Microinjection was performed under fluorescence microscopy (Nikon Diaphot 300) with the assistance of an Axoclamp-2B amplifier (Axon Instruments). Contact of the electrode with the cell and impalement were detected by passing hyperpolarizing current into the electrode. Injection was achieved by applying a positive pressure to the micropipette solution through the side arm of the pipette holder. Pressure was gentle in order to minimize nuclear swelling. Cells were returned to the incubator after microinjection.

Loading of antibodies

Antibodies were diluted into modified Krebs solution (KCl based and Ca²⁺ free) containing 0.5 % FITC-dextran and pressure-injected into the cytosol of SCG neurones (Caulfield et al. 1994; Delmas et al. 1998a). The resistance of the injecting electrodes was 40-60 MΩ. Injection was best achieved by positioning the electrode tangentially to the nucleus. Injection of antibody-free buffer (containing FITC-dextran) served as a control. Following injection, cells were maintained in culture for 3-4 h and identified for recording by fluorescence microscopy. Antibodies used were immunoglobulin G fractions raised against carboxyl terminal domains of Gα_oA/B and Gα_i1/2 subunits (Goldsmith et al. 1987). The success of cytoplasmic antibody injections was routinely verified by immunostaining (see Fig. 3).

Figure 3. Antibody loading.

Anti-rabbit immunoreactivity in a rat SCG neurone microinjected with rabbit anti-Gα_o antibody. The neurone was fixed with acetone 3 h following cytoplasmic microinjection. Note the staining of distal neurites. Scale bar, 20 μm.

Cell culture

Sympathetic neurons were isolated from SCG of young rats (15-19 days old) as described previously (Delmas et al. 1998b). The rats were killed by exposure to a rising concentration of carbon dioxide, followed by decapitation, according to the Animals (Scientific Procedures) Act 1986. Cells were seeded at a density of 50 cells mm⁻² onto laminin-coated glass coverslips for immunocytochemistry. Glass slides were imprinted with squares for later location of injected cells.

Electrophysiological recordings

Calcium currents were measured using the amphotericin B perforated-patch method largely as described by Delmas et al. (1998a, b). Electrodes were fabricated from borosilicate glass capillaries (Clark Electromedical Instruments) with the use of a Flaming-Brown (P-87) micropipette puller (Sutter Instruments) and fire polished. Pipettes had a resistance of 1-2 MΩ when filled with the following intracellular solution (mM): CsCl, 30; caesium acetate, 110; MgCl₂, 1; and Hepes, 10 (pH 7.2-7.3 with CsOH; 295-300 mosmol l⁻¹) and 0.1 mg ml⁻¹ amphotericin B. The external solution consisted of (mM): NaCl, 130; KCl, 3; MgCl₂, 1; Hepes, 10; tetrodotoxin (TTX), 0.0005; CaCl₂, 2; and glucose, 11 (pH 7.3 with NaOH, ∼300 mosmol l⁻¹). Access resistances after permeabilization (10-20 min at 32-33°C) ranged between 6 and 10 MΩ. Cell membrane capacitance and series resistance compensations (∼80 %) were applied. For whole-cell (‘ruptured-patch’) recording, pipettes (1.5-2.5 MΩ) were filled with an intracellular solution consisting of (mM): CsCl, 130; MgCl₂, 1; BAPTA, 10-20; CaCl₂, 0.1-0.5; Na₂ATP, 2; Na₃GTP, 0.12; and Hepes 10 (pH 7.2-7.3 with CsOH). Recordings were obtained with an Axopatch 200A patch clamp amplifier (Axon Instruments), filtered at 2-5 kHz and corrected for leak and capacitive currents using the leak subtraction procedure (P/6) of pCLAMP 6 software (Axon Instruments). pCLAMP 6 software was used to collect and analyse the data. Calcium currents were recorded at 32-33°C in relatively small (17-40 pF) SCG neurones. Space clamp quality was assessed by examining activation and tail current time constants evoked by positive voltage pulses. Data are from cells which exhibited graded voltage-dependent current activation and tail current time constants < 1 ms at 0 mV. Calcium current-voltage relationships determined in the perforated-patch configuration displayed a similar voltage dependence to those obtained in the classical whole-cell recording configuration. Inward Ca²⁺ currents activated near -30 mV, reached a maximum amplitude near +5 mV and approached a zero current asymptote at about +50 mV (with internal Cs⁺). Approximately 90 % of high voltage-activated (HVA) Ca²⁺ currents appeared sensitive to ω-conotoxin GVIA (500 nM, n = 5) (in accordance with previous data: see Plummer et al. 1989). No significant rundown of I_Ca was observed using the perforated patch over > 1 h recording whereas rundown was relatively rapid (25-35 % in < 20 min) in whole-cell recording. The amplitude of I_Ca was measured isochronally 4 ms after the onset of a test pulse after subtracting the current remaining in the presence of 500 μM Cd²⁺. The voltage dependence of inhibition was examined using a three-pulse voltage protocol consisting of two test pulses (P1 and P2) to 0 mV separated by a conditioning prepulse to +90 mV (see Fig. 4). Facilitation was then determined as the P2/P1 ratio of the current amplitudes. The ratio of inhibition (IR) was defined as the ratio of inhibitions determined before and after the conditioning pulse (with 100 - 100/IR = percentage of voltage-dependent component). Data are expressed as the mean ±s.e.m. Student's unpaired t test and analysis of variance were applied to determine statistical significance. Differences were considered significant if P < 0.05.

Figure 4. Both anti-Gα_o and anti-Gα_i antibodies reduce PTX-sensitive noradrenergic inhibition.

Left panels: calcium current inhibition induced by 1 μM noradrenaline in neurones cytoplasmically injected with either FITC-dextran (A) or antibodies raised against Gα_o (B) or Gα_i (C) subunits. Currents were elicited by the use of the three-pulse voltage protocol as illustrated. In B and C - and subsequent figures - the outward currents elicited by the +90 mV voltage pulses (at break) are omitted for clarity. Right panels: calcium current amplitude (•) and facilitation (○) plotted as a function of time for the corresponding cells shown in the left panels. All neurones were recorded using the perforated-patch method 3-4 h after cytoplasmic microinjection.

Immunocytochemistry

Immunocytochemistry was performed essentially as described previously (Abogadie et al. 1997). Briefly, following electrophysiological recordings, SCG neurones were fixed in acetone (10-20 min at room temperature). The cells were then incubated (1 h at room temperature) with polyclonal antibodies raised against Gα_o (sc-387, reactive with Gα_oA and Gα_oB, 1 : 1000 dilution; Santa Cruz Biotechnology, Santa Cruz, CA, USA), Gα_i3 (sc-262, reactive with Gα_i1, Gα_i2 and Gα_i3, 1 : 1000 dilution; Santa Cruz Biotechnology) and Gα_q (IQB, antiserum generated against a synthetic peptide corresponding to amino acids 119-134 of Gα_q, 1 : 1000 dilution; Milligan et al. 1993). Bound antibodies were detected using biotinylated Fab2 swine anti-rabbit IgG antibody (Dako, Denmark) conjugated with alkaline phosphatase (1 : 500 dilution). The specificity of the staining was assessed by competing out with the respective antigenic peptides (typically 10-fold excess) (Delmas et al. 1998a; Haley et al. 1998).

Drugs and chemicals

Cells were superfused at 10 ml min⁻¹ during recording. The solutions containing test agents were applied to neurones through a large-bore tube (∼1 mm i.d.) placed 2-3 mm away from the neurone under study. Noradrenaline (Sigma) was prepared daily from frozen stock solutions (10 mM). When used, Bordetellapertussis toxin (PTX, 1 μg ml⁻¹; Sigma) was added to the culture medium for 24-28 h (37°C, 5 % CO₂). All other chemical compounds were from Sigma, except oxotremorine-M (RBI).

RESULTS

PTX-sensitive and PTX-insensitive components of noradrenergic inhibition

A dose-response curve for the inhibition of high voltage-activated Ca²⁺ currents (HVA I_Ca) by noradrenaline recorded using the amphotericin B perforated-patch method is illustrated in Fig. 1 (○). Inhibition was detectable with 30 nM noradrenaline and reached maximum levels at concentrations > 1 μM. The data could be fitted by an equation of the form y =a/(1 + IC₅₀/[NA]) (where y is percentage inhibition, a is maximal percentage inhibition and [NA] is the concentration of noradrenaline), giving an IC₅₀ of 150 ± 50 nM (n = 7). This was significantly less than that obtained using whole-cell patch recording (IC₅₀= 500 ± 40 nM, n = 8; Fig. 1, ▵). In addition to the increase in sensitivity to noradrenaline, maximum suppression of I_Ca was also enhanced in perforated-patch recording (79 ± 4 % inhibition with 10 μM noradrenaline versus 66 ± 5 % in whole-cell recording). By contrast, the rate of desensitization of noradrenaline-mediated responses was similar under the two recording configurations (10-20 % within 2 min of exposure to 10 μM noradrenaline). Pretreatment with PTX reduced the noradrenaline-mediated inhibition of I_Ca by 73 ± 4 % in 11 out of 17 neurones and abolished the response in the remaining six cells. Following PTX treatment, inhibition of I_Ca required a higher concentration of noradrenaline (IC₅₀ 3.3 ± 0.3 μM; Fig. 1). Hence, in agreement with the observations made by Beech et al. (1992), Shapiro et al. (1994) and Zhou et al. (1997) using whole-cell recording, the action of noradrenaline on rat SCG neurones is mediated by at least two routes: a ubiquitously distributed, high affinity PTX-sensitive pathway and a rarer, low affinity PTX-resistant pathway.

Figure 1. Perforated-patch recording of α₂-adrenoceptor-mediated inhibition of HVA Ca²⁺ currents.

Concentration-inhibition curves for noradrenaline (NA) determined using the perforated-patch (circle) or whole-cell (triangle) methods, in SCG neurones pretreated (filled symbols) or not (open symbols) with PTX (1 μg ml⁻¹ for 24 h). Each point represents the mean ±s.e.m. of 4-9 cells. The data were fitted to the equation y =a/(1 + IC₅₀/[NA]) where y is percentage inhibition and a is maximal percentage inhibition. Calculated values for a and IC₅₀ were: ○, 79 % and 150 nM; ▵, 67 % and 500 nM; •, 27 % and 3.3 μM; and ▴, 22 % and 4.7 μM, respectively. Inset, superimposed HVA I_Ca traces recorded using the perforated-patch method, in the presence or absence of 1 μM noradrenaline.

The onset rate for I_Ca inhibition exerted by noradrenaline in perforated-patch recordings was measured at saturating concentrations (30 μM) using the protocol shown in Fig. 2A. Four successive depolarizing steps to a test potential of -10 mV were applied to the cell from a holding potential of -70 mV. This stimulatory sequence was reiterated every 100 ms and caused very little cumulative inactivation of I_Ca; it was chosen to estimate the time constant of Cd²⁺ block, which was 40-80 ms. Flow of noradrenaline-containing solution was begun randomly and turned off after inhibition reached steady-state level. The onset rate of noradrenaline-induced inhibition was then obtained by subtracting the time constant of the action of Cd²⁺ from that of noradrenaline. Both the reduction of HVA I_Ca (Fig. 2B) and recovery (not shown) from noradrenaline inhibition developed exponentially. In cells not treated with PTX, time constants of noradrenaline inhibition were 380 ± 25 ms (n = 4) for the onset and 1.7 ± 0.5 s (n = 4) for the offset, whereas inhibition in PTX-treated cells displayed ‘slower’ time constants of 950 ± 40 ms (n = 4) and 4.5 ± 0.5 s (n = 3), respectively. For comparison, I_Ca inhibition through the slow M₁ muscarinic acetylcholine receptor (M₁ mAChR)-mediated pathway (in response to 10 μM oxotremorine-M in PTX-treated cells) developed with a 6 ± 0.5 s time constant (Fig. 2B, n = 3) and recovered (often incompletely) with a time constant of 22 ± 6 s (n = 3).

Figure 2. Time course of α₂-adrenoceptor-mediated inhibition of Ca²⁺ currents.

A, changes in peak Ca²⁺ currents induced by applying noradrenaline (30 μM) or cadmium (500 μM, applied at the end of the experiment; superimposed symbols, •). The horizontal bar indicates the time and duration of application of the drugs. Inset, superimposed current traces selected at the times indicated by the corresponding numbers in A. I_Ca was elicited by applying repetitive sequences consisting of 4 successive depolarizing steps to -10 mV. Cd²⁺-sensitive currents were subtracted off-line. B, time course of inhibition of HVA I_Ca by 10 μM noradrenaline: SCG neurones pretreated (•) or not (○) with PTX (1 μg ml⁻¹ for 24 h). Compare with the time course of inhibition induced by 10 μM oxotremorine-M (M₁ mAChR pathway, ▾) in a neurone incubated with PTX. Inhibition was normalized to the inhibition measured at steady state. Smooth curves are exponential fits to the data.

Neutralization of G_o- and G_i-type G-proteins differentially alters the PTX-sensitive inhibition

One micromolar noradrenaline was used to study the noradrenergic PTX-sensitive pathway in isolation (see Fig. 1). In confirmation, treatment with PTX (1 μg ml⁻¹ for 24 h) completely prevented the inhibitory effects of 1 μM noradrenaline (3 ± 1 %, n = 8). This inhibition exhibited the trademark features of voltage dependence as previously reported in whole-cell recordings (Bean, 1989; Beech et al. 1992), such that the inhibition was less at large that at moderate depolarizations. For example, noradrenaline showed a mean inhibition of 20 ± 1.5 % (n = 6) (cf. 15 and 30 % in Bean, 1989, and Beech et al. 1992, respectively) following a pulse to +120 mV compared with 61.1 ± 0.8 % (n = 20) at moderate depolarizations. Modulated HVA currents also showed slowing of activation kinetics and reversal of inhibition (‘facilitation’; Ikeda, 1991) following a conditioning membrane depolarization. Facilitation occurred only in the presence of noradrenaline, not in its absence, and was maximal (facilitation 1.9 ± 0.15, n = 16) at voltages near the point of peak Ca²⁺ current. In agreement with tail current recordings, reversal of inhibition using the three-pulse protocol was always incomplete, leaving part of the current still inhibited (24.1 ± 0.7 %, n = 16) following the intercalating depolarization. PTX-insensitive inhibition produced by higher concentrations of noradrenaline (10-30 μM; n = 6) was largely refractory to modulation by voltage (data not shown).

Previous experiments using microinjection of antisera against Gα subunits showed that G_o heterotrimers contributed to PTX-sensitive noradrenergic inhibition (McFadzean et al. 1989; Caulfield et al. 1994). However, it is not clear from these experiments which particular G-protein and G-protein subunit(s) is (are) responsible for voltage-dependent inhibition (see Introduction). To address this further, we injected antibodies (IgGs) against C-terminal decapeptide sequences of Gα_oA/B (anti-Gα_o) or Gα_i1/2 (anti-Gα_i) into the cytoplasm and examined the modulation of calcium currents 3-4 h later. To ensure antibody loading in recorded cells, the efficiency of the injection was assessed immunocytochemically. An example of a SCG neurone preinjected with Gα_o antibody is shown in Fig. 3 where the microinjected Gα antibody could be detected even in very distal processes. As reported previously (Delmas et al. 1998a), antibodies did not alter the kinetic properties of I_Ca; however, the injection per se slightly reduced the calcium current density. In agreement with the work by Caulfield et al. (1994), we found that the anti-Gα_o antibody significantly reduced noradrenergic inhibition from 58 ± 4 % (cells microinjected with antibody-free buffer) to 34.1 ± 3.3 % (Fig. 4B and Fig. 5A). We observed, however, that the anti-Gα_i antibody, though less effective, also attenuated noradrenergic inhibition (to 44.2 ± 2.1 %, n = 5) (Fig. 4C and Fig. 5A). The specificity of the effects of the antibodies was assessed on G_o-dependent M₄ mAChR inhibition of I_Ca (Delmas et al. 1998a) and G_q-dependent M₁ mAChR inhibition of M-type potassium currents (Haley et al. 1998). Anti-Gα_i antibody did not alter M₄ muscarinic modulation of calcium currents (Fig. 5A), nor did Gα_o antibody affect M₁ muscarinic inhibition of M-type K⁺ currents (see Delmas et al. 1998a).

Figure 5. Anti-Gα antibodies alter the voltage dependence of inhibition.

A, scatter plot showing I_Ca inhibition by 1 μM noradrenaline (open symbols) or 10 μM oxotremorine-M (filled symbols) in individual neurones as indicated (Control, neurones microinjected with FITC-dextran). The horizontal bars indicate the mean values. I_Ca was evoked using the three-pulse voltage protocol as in Fig. 4 and inhibition was calculated from the baseline current in the presence of 500 μM Cd²⁺ as (1 - I_NA/I_control) × 100. * P = 0.015; ** P = 0.0012. Voltage-dependent M₄ mAChR inhibition was recorded in the whole-cell (patch-ruptured) mode using high BAPTA internal solution (see Methods). B, summary of facilitation (means and s.e.m.) observed in uninjected neurones or neurones injected with anti-Gα antibodies (Gα ab) in the presence (+) or absence (-) of 1 μM noradrenaline. n.s. (not significant), P = 0.42; ** P = 0.0039. C, mean inhibition ratio (and s.e.m.) under the different conditions indicated. Inhibition ratio was calculated as the ratio of inhibition before conditioning depolarization to inhibition after conditioning depolarization.

Noradrenaline-induced inhibition was then examined using the three-pulse voltage protocol in order to assay effects of antibodies on the voltage dependence of inhibition. As shown in Fig. 4A, strong depolarization in control cells restored most of the modulated I_Ca, in that inhibition decreased from 61 ± 2.5 to 22 ± 2 % (n = 16) following depolarization. In cells loaded with anti-Gα_o antibody, modulated I_Ca was largely resistant to facilitation, the inhibition after the conditioning step (23.1 ± 3 %) being little different from the overall inhibition (34 %, see above). Thus, the mean facilitation measured at 0 mV was 1.16 ± 0.17 (n = 6) compared with 1.88 ± 0.12 (n = 8) and 1.7 ± 0.2 (n = 3) in uninjected cells and cells injected with antibody-free buffer solution, respectively (Fig. 4B). By contrast, residual inhibition in cells loaded with anti-Gα_i antibody appeared strongly voltage dependent with a facilitation of 1.70 ± 0.2 (n = 5) (Fig. 4C and Fig. 5B). However, because the facilitation depends upon both basal modulation and the degree of inhibition, it does not provide a good estimate of the voltage-dependent action of neurotransmitters. Hence, we chose to take the inhibition ratio (IR, see Methods) as a much more appropriate index for voltage dependence. This clearly revealed that most (68 %) of the residual inhibition in the presence of the anti-Gα_o antibody was not reversed by depolarization whereas it was strongly reversed (by 82 %) in cells injected with the anti-Gα_i antibody (Fig. 5C).

Depletion of Gα_oA subunits, but not Gα_i, depresses the voltage-dependent fraction of the PTX-sensitive inhibition

Gα_oA and Gα_i were selectively depleted by expressing specific antisense RNAs. SCG neurones were injected intranuclearly with expression plasmids (400 μg ml⁻¹) designed to drive production of high levels of antisense RNA complementary to either the 3′ untranslated sequence of Gα_oA (the more abundant Gα_o isoform in SCG neurones) or a coding sequence common to Gα_i1-3 subunits. Expression of these antisense RNAs specifically decreased immunoreactivity for their targeted Gα proteins (Delmas et al. 1998a; Haley et al. 1998). Inhibition of calcium currents by 1 μM noradrenaline was then examined 2 days later when the effects of the antisense RNAs reach a plateau (Haley et al. 1998).

As reported previously (Delmas et al. 1998a), basal facilitation was high in neurones expressing Gα_oA antisense RNA (Fig. 6Ab). This effect was ascribed to the action of Gβγ dimers released from G_oA heterotrimers (as Gα_oA is depleted) and was not observed in cells expressing Gα_q or Gα_icommon antisense RNAs (Fig. 6Aa and Ac). Depletion of Gα_oA subunits reduced the noradrenergic inhibition from 61.1 ± 0.8 to 17 ± 4 % (n = 7). The residual inhibition seen in these cells was mostly voltage independent, with an inhibition ratio of 1.12 ± 0.12 (Fig. 6B). In cells expressing the Gα_icommon antisense RNA, inhibition was slightly reduced to 48.7 ± 2.8 % (n = 6) but, in contrast with the Gα_oA antisense RNA, was still strongly reversed by depolarization (IR = 4.8 ± 0.4) (Fig. 6B). The Gα_q antisense RNA had no significant effects on noradrenaline-induced inhibition.

Figure 6. Noradrenergic inhibition in cells depleted of either Gα_i or Gα_oA subunits.

A, superimposed Ca²⁺ current traces recorded in the absence and presence of noradrenaline (1 μM) in neurones preinjected intranuclearly with cDNA plasmids generating antisense RNA specific to Gα_q (a), Gα_oA (b) and Gα_i1-3 (c) subunits. Note the tonic facilitation (delineated by the dashed line) of basal I_Ca in the cell expressing Gα_oA antisense RNA (b). B, summary of calcium current inhibition (inhibition before (□) and after (▪) the conditioning depolarization) and ratio of inhibition (grey bars) in uninjected cells and cells expressing different Gα antisense RNAs. * P < 0.05; ** P < 0.005; *** P < 0.0001.

Both G_o and G_i signals are reduced by βγ-sequestering agents

The above experiments show that G_o and G_i heterotrimers are responsible for two biophysically distinct forms of inhibition: that mediated by G_o is voltage dependent whereas G_i-mediated inhibition is not. We next wanted to test whether βγ subunits are involved in these components. One approach to defining whether I_Ca modulation is transduced via Gβγ dimers is to express peptides containing Gβγ-binding domains (βγ-sequestering agents) thereby competing with the Ca²⁺ channel for free βγ dimers. There are inherent concerns with this strategy as Gβγ-sequestering agents may not bind all βγ dimers and may interfere with membrane receptor-G-protein interaction. To overcome this problem we over-expressed three distinct putative βγ scavengers (200 μg ml⁻¹): the carboxyl terminal domain of βARK1 (βARK1_C-ter), which can effectively neutralize released Gβγ (Koch et al. 1994; Delmas et al. 1998a,b; Stephens et al. 1998) but is thought to not bind all βγ subunits (Daaka et al. 1997); a mutationally modified (palmitoylation-negative) Cys3 to Ser Gα_i1 subunit, which binds Gβγ dimers but does not interact with the α_2A-adrenoceptor (Wise et al. 1997a); and retinal α-transducin. To assay specificity, the effects of these peptides were examined in parallel on G_q-mediated M₁ mAChR inhibition, which does not involve βγ subunits (Delmas et al. 1998a).

Gα_i1 Cys3Ser, βARK1_C-ter and α-transducin had no noticeable effect on basal calcium current properties. However, they strongly attenuated calcium current inhibition by 1 μM noradrenaline to 39.5 ± 1.5 % (n = 5), 29 ± 2 % (n = 6) and 3 % (n = 5), respectively. Residual inhibition was mostly voltage insensitive in cells expressing βARK1_C-ter (IR = 1.3 ± 0.2) whereas it was strongly reversed by depolarization in cells expressing Gα_i1 Cys3Ser (IR = 2.9 ± 0.3) (Fig. 7). βARK1_C-ter expression had no significant effects on M₁ mAChR-mediated inhibition (PTX-treated cells; see Delmas et al. 1998a) but Gα_i1 Cys3Ser and α-transducin slightly depressed M₁ mAChR responses by 17 ± 4 % (n = 4) and 34 ± 7 % (n = 3), respectively (data not shown).

Figure 7. Gβγ-binding agents.

Calcium current inhibition by noradrenaline (1 μM) in neurones expressing the C-terminal domain of βARK1 (A), a mutationally modified form of Gα_i1 (Gα_i1 Cys3Ser) (B) or α-transducin (C). The filled symbols indicate the inhibition ratio upon noradrenaline application. Insets show superimposed current traces collected at the indicated times (*). D, inhibition of I_Ca plotted against inhibition ratio in uninjected neurones (○) and in neurones expressing either the C-terminal domain of βARK1 (•) or Gα_i1 Cys3Ser (▴). Cells recorded 24 h after intranuclear injection. Plasmid concentration, 200 μg ml⁻¹.

Reconstitution of the α₂-adrenoceptor-G_i pathway by the use of PTX-resistant Gα_i mutants

To assess further the role of the G_i heterotrimer in mediating a component of noradrenergic inhibition, neurones were injected with cDNAs encoding PTX-resistant forms of Gα_i1 (Wise et al. 1997b; Bahia et al. 1998). Cells were subsequently treated with PTX to eliminate interactions of α₂-adrenoceptor(s) with the endogenous G_i/o population. As anticipated, there was no inhibition of I_Ca by 1 μM noradrenaline in control cells after PTX treatment (Fig. 8A). By contrast, in cells expressing the PTX-resistant Gα_i1 Cys351Ile mutant, noradrenaline depressed I_Ca by 39.2 ± 4 % (n = 10) (Fig. 8B). Expression of another PTX-resistant Gα_i1 mutant (Gα_i1 Cys351Arg) that could not be activated by the α_2A-adrenoceptor (Bahia et al. 1998) did not reconstitute noradrenergic inhibition (Fig. 8C). Interestingly, in most (7 out of 10) of the cells expressing Gα_i1 Cys351Ile, noradrenaline-mediated inhibition was only weakly reversed by depolarization, as shown by the relatively low inhibition ratio (1.24 ± 0.05; Fig. 8D). In the other three cells the inhibition ratios were 1.9, 2.1 and 2.3, so about half the inhibition in these particular cells was voltage sensitive (Fig. 8D) - i.e. similar to control cells. Interaction of the Gα_i1 Cys351Ile mutant with the α₂-adrenoceptor(s) appeared to be very specific since expression of this subunit had no significant effects on M₁ muscarinic inhibition of I_Ca (49 ± 4 % (n = 4) in control neurones and 56 ± 3 % (n = 5) in neurones expressing Gα_i1 Cys351Ile) and failed to reconstitute the fast M₄ mAChR inhibition in PTX-treated cells (n = 5) (Fig. 8B).

Figure 8. PTX-resistant Gα_i1 mutants rescue noradrenergic inhibition.

A-C, cells pretreated with 1 μg ml⁻¹ PTX for 24 h. Superimposed calcium current traces in the presence or absence of noradrenaline (1 μM) in an uninjected neurone (A) and in neurones expressing either Gα_i1 Cys351Ile (B) or Gα_i1 Cys351Arg (C) (100 μg ml⁻¹). Lower panels in A and B show the time course of noradrenaline and oxotremorine-M (1 μM) actions. Note the absence of fast muscarinic (M₄ mAChR) inhibition in B, inhibition being slow in onset. D, plot of inhibition of I_Ca in response to 1 μM noradrenaline against inhibition ratio in uninjected neurones (○, not treated with PTX) and in neurones expressing Gα_i1 Cys351Ile (•, treated with PTX).

Degree and voltage dependence of inhibition as a function of Gα_i1 plasmid concentration

To test whether the variability in the magnitude as well as the voltage dependence of G_i1 protein-mediated inhibition may result from different levels of expression of Gα_i1 subunits from cell to cell, we injected SCG neurones with various concentrations (10-600 μg ml⁻¹) of cDNA encoding Gα_i1 Cys351Ile. Figure 9A and B clearly shows that the degree of inhibition as well as the fraction of voltage-dependent inhibition increased as a function of the concentration of the Gα_i1 Cys351Ile plasmid. Increasing the plasmid concentration up to 600 μg ml⁻¹, however, tended to block the inhibition (13 ± 4 %, n = 4), probably due to the excess of Gα_i1 mutant with respect to activated Gβγ. The onset rate of the inhibition was also dependent upon Gα_i1 Cys351Ile plasmid concentration, with time constants of 2.47 ± 0.2 s (n = 3) and 0.67 ± 0.17 s (n = 4) at 10 and 300 μg ml⁻¹ plasmid concentration, respectively (Fig. 9D). Interestingly, the voltage-dependent fraction of G_i-mediated inhibition (with 300 μg ml⁻¹ Gα_i1 Cys351Ile plasmid) was largely prevented by co-expressing βARK1_C-ter (200 μg ml⁻¹) (Fig. 9C) while the time course of inhibition in these cells was not significantly altered (0.75 ± 0.2 s, n = 4) (Fig. 9D). Gα_i1 Cys351Ile-mediated inhibition was mostly prevented by co-expression of α-transducin (7 ± 2 %, n = 4).

Figure 9. Degree, voltage dependence and onset rate of inhibition as a function of Gα_i level.

A, degree of inhibition in response to 1 μM noradrenaline (□) and inhibition ratio (▪) in neurones preinjected with various concentrations (10-300 μg ml⁻¹ as indicated) of cDNA encoding Gα_i1 Cys351Ile. B, relationship between inhibition and inhibition ratio in individual cells preinjected with 10 (○), 30 (▿), 100 (▴) and 300 μg ml⁻¹ (♦) Gα_i1 Cys351Ile-encoding plasmids. C, the same as in B in cells expressing Gα_i1 Cys351Ile alone (300 μg ml⁻¹, ♦) or together with the C-terminal domain of βARK1 (200 μg ml⁻¹, ○). D, time course of inhibition in cells preinjected with Gα_i1 Cys351Ile-encoding plasmid alone (10 μg ml⁻¹, ▵; 30 μg ml⁻¹, ⋄; 300 μg ml⁻¹, ○) or together with βARK1_C-ter (300 μg ml⁻¹ Gα_i1 Cys351Ile + 200 μg ml⁻¹βARK1_C-ter, •). Continuous lines are mono-exponential fits to the data.

DISCUSSION

In agreement with previous studies (Beech et al. 1992; Shapiro et al. 1994; Zhou et al. 1997), but now using perforated-patch recording, we have detected two pathways leading to I_Ca inhibition in rat SCG neurones following noradrenergic stimulation: a predominant one mediated via PTX-sensitive G-proteins that responds to low concentrations of noradrenaline (IC₅₀∼150 nM); and one, more marginal, that needs higher concentrations of noradrenaline (IC₅₀∼3 μM) and is resistant to PTX. These two inhibitory pathways could also be differentiated kinetically, in that the PTX-resistant pathway had slower onset kinetics (τ∼1 s) than the PTX-sensitive one (τ∼400 ms). Zhou et al. (1997) have suggested that these two adrenergic signals may correspond to the contribution of low and high affinity receptors, although there is no direct evidence so far that a G-protein-coupled receptor(s) is actually involved in the ‘low affinity’ PTX-insensitive response. A possible candidate for mediating PTX-insensitive noradrenergic inhibition would be G_z; however, reconstituted G_z coupling to Ca²⁺ channels in SCG neurones appeared to be voltage dependent and as much as 10-fold slower in onset/offset (Jeong & Ikeda, 1998) than the PTX-insensitive component determined here.

Further dissection of the PTX-sensitive noradrenergic inhibition suggests that more than one G-protein is involved in coupling the receptor(s) to the channel. Microinjection of antibodies directed against the C-terminal decapeptide sequence of Gα subunits revealed that both G_o and G_i heterotrimers couple α₂-adrenoceptors to Ca²⁺ channels. The effects of the anti-Gα_i antibody cannot be ascribed to a non-specific action on G_o-type G-proteins since this antibody had no significant effects on M₄ mAChR inhibition of I_Ca which involves G_oA (Delmas et al. 1998a). Since the anti-Gα_o antibody was about twice as effective as the anti-Gα_i antibody, the G_o heterotrimer is likely to be the principle transducer (in agreement with the previous conclusion of Caulfield et al. 1994, from experiments using whole-cell recording). This inference is supported by the results obtained on expressing antisense RNAs to Gα_oA and Gα_i subunits, which decreased α₂-adrenoceptor inhibition with a similar potency ratio (G_o > G_i) to the antibodies.

The participation of multiple PTX-sensitive G-proteins in α₂-adrenoceptor modulation in rat SCG neurones is in agreement with a previous report showing that both G_o and G_i heterotrimers are involved in adrenergic inhibition of Ca²⁺ currents in chick ganglion neurones (Diversé-Pierluissi et al. 1995), although the mechanistic basis may differ (see below). Earlier biochemical studies using exogenously expressed α_2A-adrenoceptors have also shown that these receptors are tightly coupled to G_oA and G_i1-3 with little discrimination (Grassie & Milligan, 1995; Wise et al. 1997b). Whether the two G-proteins actually couple to the same α₂-adrenoceptor subtype in SCG neurones, or whether a different receptor subtype specifically interacts with each type of G-protein remains to be established.

As previously reported (Bean, 1989; Beech et al. 1992), reversal of α₂-adrenergic inhibition by depolarization is incomplete, leaving part of the current still modulated (by 10-30 %) even at very positive voltages. According to the two-state ‘willing-reluctant’ model (see Jones & Elmslie, 1997), this ‘voltage-insensitive’ component of inhibition may reflect low affinity binding of activated G-protein (e.g. βγ) to the open state of the Ca²⁺ channel, thus being phenomenologically identical to the voltage-dependent mechanism. Alternatively, this may result from an intrinsically distinct intracellular pathway like that described by Diversé-Pierluissi et al. (1997). The experiments presented here suggest that incomplete reversal of inhibition largely reflects two mechanistically different actions: a voltage-dependent action of G_o and a ‘voltage-independent’ (or less voltage-dependent) action of G_i. First, antibody neutralization of G_i heterotrimers and antisense depletion of Gα_i subunits both induced a substantial increase in the proportional amount of voltage-dependent inhibition, indicative of a major role for the G_o heterotrimer in this component. Conversely, neutralization of G_oA heterotrimers strongly reduced the voltage-dependent fraction of inhibition, suggesting that G_i mainly acts in a ‘voltage-resistant’ manner. Second, when activation of endogenous G-proteins was prevented with PTX, the reconstituted inhibition obtained on expressing a PTX-resistant form of Gα_i1 was clearly less voltage dependent than that normally seen when endogenous Gα_oA is allowed to participate in the inhibition (see Fig. 9).

There is now compelling evidence suggesting that the voltage-dependent inhibition of calcium currents produced by neurotransmitters results from the action of G-protein βγ subunits rather than the Gα subunits (Ikeda, 1996; Herlitze et al. 1996; Zamponi et al. 1997; Delmas et al. 1998a,b; Stephens et al. 1998; Zamponi & Snutch, 1998; see however Diversé-Pierluissi et al. 1997). The present results provide confirmatory evidence that this is true for G_o-mediated inhibition by noradrenaline in SCG neurones. However, they further suggest that the less voltage-sensitive G_i-mediated noradrenergic inhibition is also mediated by βγ subunits, and that the two components of inhibition may involve different βγ subunits. The obligatory role of Gβγ is demonstrated by the finding that expression of α-transducin virtually abolished inhibition, whether mediated through endogenous G-proteins or through exogenously expressed G_i, whereas it had only a minor effect on M₁ mAChR/Gα_q inhibition. We also found that the expression of βARK1_C-ter and Gα_i1 Cys3Ser differentially reduced the magnitude and the voltage dependence of inhibition. The greater attenuation of inhibition and the parallel reduction of the voltage-dependent fraction by βARK1_C-ter is compatible with the idea that it preferentially suppresses the predominant voltage-dependent G_oβγ pathway. Conversely, the Gα_i1 Cys3Ser mutant had the smallest effects on overall inhibition but enhanced the voltage-dependent fraction of inhibition, consistent with an action mainly directed against the ‘voltage-resistant’ G_iβγ pathway. The simplest interpretation of this is that different βγ pairs bind to Gα_o and Gα_i respectively, and that the two βγ pairs, when released, act in different ways on the Ca²⁺ channels. The latter inference is supported by our previous observations showing that only βγ dimers liberated from Gα_oA (following antisense depletion), not from Gα_i, induce (tonic) voltage-dependent inhibition (Delmas et al. 1998a; see also Fig. 6A). The participation of different βγ pairs is also substantiated by the finding that Gα_i1 Cys351Ile-mediated inhibition could be resolved into a voltage-dependent component sensitive to both βARK1_C-ter and α-transducin and a ‘voltage-independent’ component only sensitive to α-transducin. The increasing amount of (βARK1_C-ter-sensitive) voltage-dependent inhibition with increasing Gα_i1 Cys351Ile expression may be explained by the forced interaction of excess Gα_i1 with some ‘inappropriate’βγ dimers: under more physiological conditions Gα_i seems to couple preferentially to βARK1_C-ter-insensitive βγ dimers.

Such a specificity between α and βγ pairs has been documented in earlier studies using antisense oligonucleotides. For example, in the rat pituitary cell line GH₃, the M₄ mAChR couples to the G-protein trimer consisting of α_oAβ₃γ₄, the somatostatin receptor couples to the trimer α_oBβ₁γ₃ and the galanin receptor preferentially couples to the trimer α_oAβ₂γ₂ (Schneider et al. 1997). However, because of the difficulty of distinguishing the different β and γ isoforms immunocytochemically we did not attempt to define further the subunit composition of G_o and G_i heterotrimers using antisense strategies. Nevertheless, since βARK1 is known to bind Gβ₁ and Gβ₂ subunits (Daaka et al. 1997), the sensitivity of the voltage-dependent G_o-mediated inhibition to βARK1_C-ter suggests βγ dimers containing β₁ and/or β₂ as possible transducers of this component of inhibition. This is supported by the recent observation that voltage-dependent inhibition of Ca²⁺ currents is replicated by exogenous expression of Gβ₁ and Gβ₂ subunits but not Gβ₃ and Gβ₄ subunits (García et al. 1998).

Voltage-dependent inhibition of neuronal N-type (α_1B) and P/Q-type (α_1A) Ca²⁺ channels is thought to result from the direct binding of some Gβγ dimers to the pore-forming α₁ subunit. At least three putative sites of interaction have been identified: two of them, including the QXXER sequence common to many Gβγ-binding proteins, are located in the intracellular loop that connects transmembrane domains I and II of the Ca²⁺ channel α₁ subunit (De Waard et al. 1997; Zamponi et al. 1997), and a third is located in the C-terminal domain (Qin et al. 1997). Since voltage-dependent and ‘voltage-independent’βγ-mediated inhibitions were both rapid, with indistinguishable onset rates, it seems likely that both resulted from a direct action of the different βγ dimers on the Ca²⁺ channel. The question then arises whether these biophysically distinct inhibitions involve different Gβγ binding sites or whether the same Gβγ binding site(s) recognizes different βγ combinations varying in their binding affinity and voltage dependence of their dissociation rates - i.e. whether Ca²⁺ channels exhibit different binding affinities for defined βγ dimers.

In conclusion, our data indicate that the βγ dimers derived from G_oA or G_i heterotrimers are not equally effective in promoting voltage-dependent inhibition. They therefore favour the view that the intrinsic composition of the G-protein βγ complex (see also García et al. 1998) plays an important role in defining the voltage-dependent characteristics of Ca²⁺ current inhibition by membrane receptors.