G Protein βγ Subunit-Mediated Presynaptic Inhibition: Regulation of Exocytotic Fusion Downstream of Ca2+ Entry

See allHide authors and affiliations

Science  13 Apr 2001:
Vol. 292, Issue 5515, pp. 293-297
DOI: 10.1126/science.1058803


The nervous system can modulate neurotransmitter release by neurotransmitter activation of heterotrimeric GTP–binding protein (G protein)–coupled receptors. We found that microinjection of G protein βγ subunits (Gβγ) mimics serotonin's inhibitory effect on neurotransmission. Release of free Gβγ was critical for this effect because a Gβγ scavenger blocked serotonin's effect. Gβγ had no effect on fast, action potential–evoked intracellular Ca2+ release that triggered neurotransmission. Inhibition of neurotransmitter release by serotonin was still seen after blockade of all classical Gβγ effector pathways. Thus, Gβγ blocked neurotransmitter release downstream of Ca2+ entry and may directly target the exocytotic fusion machinery at the presynaptic terminal.

A number of neurotransmitters have been shown to modulate release from presynaptic terminals (1, 2) through activation of a G protein–coupled receptor (GPCR) (3, 4). The two arms of activated G proteins, Gα and Gβγ, may exert their modulatory effect by regulating second messenger enzymes, ion channels, or other targets (5). The cellular and molecular mechanisms underlying GPCR-mediated presynaptic modulation remain largely undetermined because of the difficulties in manipulating and recording from presynaptic terminals. The reticulospinal/motoneuron synapse in the lamprey is one of the few vertebrate synapses that are experimentally accessible both pre- and postsynaptically, because of their large size. In this giant synapse, serotonin decreases the efficacy of synaptic transmission presynaptically (6, 7) through a GPCR.

We microinjected proteins into the presynaptic terminal and determined the effect of serotonin on the strength of synaptic transmission. We recorded from pre-and postsynaptic elements simultaneously (6,8), as indicated in Fig. 1A. Presynaptic axons were stimulated intracellularly to evoke excitatory postsynaptic currents (EPSCs) in the postsynaptic neurons. This resulted in mixed electrical and chemical EPSCs (Fig. 1B, part ii). Separation between the two components was clear with a fast invariable electrical component and a slower chemical component that exhibited quantal variation in amplitude (Fig. 1C, part i). The averaged postsynaptic responses before and after exposure to serotonin (30 μM) are shown in Fig. 1B, part ii. Application of serotonin reduced the amplitude of the chemical component to 19.8 ± 7.5% of the control's amplitude [four pairs; P < 0.01 (9)] (Fig. 1B, part ii). Serotonin-mediated depression was seen concurrently with application of serotonin (<1 min) without affecting the amplitude of the electrical component. This serotonin-mediated depression of synaptic transmission is presynaptic (6) and is independent of the modulation of Ca2+ entry into the presynaptic terminal (7).

Figure 1

Serotonin (5-HT) and presynaptic Gβγ inhibit neurotransmitter release. (A) (Left) Schematic diagram to show the recording arrangement in paired recordings between reticulospinal axons and ventral horn neurons. The axon is held under current clamp with a microelectrode, and the motoneuron is simultaneously held under voltage clamp with a patch electrode. 5-HT R, serotonin receptor. (Right) A hypothesized signaling cascade evoked by serotonin released onto the presynaptic cell. (B) Serotonin (30 μM) inhibits synaptic transmission at the giant reticulospinal synapse. Part i shows presynaptic recordings from a giant axon. The presynaptic action potential was evoked by a depolarizing current pulse (2 ms) through the recording microelectrode. Part ii shows the presynaptic action potential–evoked EPSCs in the postsynaptic motoneuron. The application of serotonin reduced the amplitude of the EPSC but had no effect on the membrane potential. These data represent the mean responses of 20 consecutive trials. (C) The pressure injection of Gβ1γ2 to the presynaptic axon had no effect on the membrane potential or evoked presynaptic action potential (part iii) but reduced the amplitude of the EPSC (part i). Part iv is the average of 20 consecutive traces; raw traces are also shown before (part i) and after (part ii) microinjection of Gβ1γ2. (D) Microinjection of the carrier vehicle has no effect on transmitter release (data are the average of 20 consecutive trials).

Serotonin works via a GPCR in this synapse. After activation by a GPCR, the G protein dissociates into an activated GαGTP subunit (GTP, guanosine 5′-triphosphate) and a free Gβγ subunit (10). Which G protein subunit inhibits neurotransmitter release is not known. We thus injected proteins directly into the presynaptic terminal (11). Injection of vehicle had no significant effect (n = 3; EPSC amplitude changed to 104 ± 7% of the control's amplitude) on the EPSCs evoked in paired-cell recording (Fig. 1D). We injected Gβ1γ2 directly into the intracellular space of the presynaptic terminal through the microelectrode while simultaneously recording from the postsynaptic neuron. We chose the Gβ1γ2 subunit combination because it is widely distributed within the central nervous system (12). The averaged postsynaptic responses before and after Gβ1γ2 pressure injection are shown inFig. 1C, part iv. A number of individual EPSCs before and after Gβ1γ2 injection are shown in Fig. 1C, parts i and ii. Gβ1γ2 inhibited the chemical component of neurotransmission to 35 ± 23% of that of the control (n = 6; P < 0.01) while leaving the electrical component unaltered.

Gβγ can mimic serotonin's effect on synaptic transmission, but is it normally a part of the serotonin-mediated signaling cascade? A scavenger of Gβγ, the carboxyl terminus of the G protein–coupled receptor kinase 2 (ct-GRK2), was microinjected into the presynaptic axon (13). ct-GRK2 is a potent and specific Gβγ inhibitor (14). If the inhibition of neurotransmission by activation of serotonin receptors is via Gβγ, then ct-GRK2 should attenuate this inhibition. The addition of 30 μM serotonin to the superfusate resulted in a reduction of the chemical component of the EPSC (Fig. 2B). After serotonin was washed out, ct-GRK2 was microinjected into the presynaptic axon. Reapplication of serotonin (30 μM) had no effect on the amplitude of the evoked EPSC (n = 5) after microinjection of ct-GRK2 (Fig. 2C). Thus, the Gβγ scavenger completely occluded the ability of serotonin to inhibit chemical neurotransmission.

Figure 2

A Gβγ scavenger blocks the inhibitory effect of serotonin. (A) The presynaptic action potential evoked a postsynaptic EPSC that was reduced by serotonin (5-HT). (B) Control EPSCs are the mean of 12 consecutive stimuli from a typical ventral horn neuron. The presynaptic microinjection of ct-GRK2 results in an enhancement of the EPSC amplitude over control conditions recorded over the following 30 min, although this enhancement was not significant. (C) ct-GRK2 attenuates inhibition of neurotransmission by serotonin application.

We performed Ca2+ imaging experiments (15) to examine the effect of serotonin application or Gβ1γ2 microinjection on the fast Ca2+ transient elicited by presynaptic action potentials. We have previously demonstrated that high-affinity dyes [e.g., Oregon Green 488 BAPTA-1 (BAPTA, 1,2-bis(2-aminophenoxy)ethane-N,N,N,′N′-tetraacetic acid)] may be used to measure Ca2+ transients evoked after action potential activation of voltage-gated Ca2+ channels in giant axons (16). This allows us to examine serotonin or Gβγ modulation of voltage-gated Ca2+channels or modulation of other ion channels, which could, in turn, change the gating of Ca2+ channels during an action potential. The amplitude of these Ca2+ transients can be correlated with neurotransmitter release (7). Single action potentials were evoked in the presynaptic axons through a stimulating electrode (Fig. 3A). An action potential–mediated Ca2+ transient was observed, which results from the activation of Ni2+-sensitive voltage-operated Ca2+ channels in this preparation (7, 15). Serotonin (30 μM) slightly reduced the peak amplitude of this stimulus-evoked Ca2+ transient (the mean change was to 82.1 ± 3.9% of that of the control;P < 0.0001; n = 28) (Fig. 3B). We repeated this experiment with a low-affinity indicator [Fluo-4 dextran (dissociation constant K d ≅ 3 μM)] (17). A similar lack of effect of serotonin on the amplitude of the Ca2+ transient was observed (amplitude after serotonin application was 95% of the control's amplitude;n = 2) (18). Furthermore, serotonin has no effect on Ca2+ currents recorded from giant axons under whole-cell voltage-clamp conditions (7). The reduction in peak amplitude of the Ca2+ transient cannot account for the profound depression of the EPSCs elicited by application of serotonin.

Figure 3

1γ2 has no effect on stimulus-evoked presynaptic Ca2+ transients. (A) Schematic to show the recording arrangement during Ca2+ imaging experiments. Axons were stimulated either through a microelectrode for Gβγ injections or extracellularly for experiments with serotonin (5-HT) application. Reticulospinal axons were labeled with the calcium-sensitive dye Oregon Green 488 BAPTA-1 [an example is shown in (C)]. (B) A single action potential was evoked in the axon by extracellular stimulation. This evoked a transient increase in axonal Ca2+, and localized areas of Ca2+ influx or “hot spots” are observed [e.g., part ii of (C)]. Data are normalized to prestimulus levels. The fluorescent response (ΔF/F) was integrated at each time point and plotted as a function of time (black trace, control data). The application of serotonin (30 μM) showed little effect on the amplitude of the evoked Ca2+ transient (blue trace). (C) Part i shows the axon filled with Oregon Green 488 BAPTA-1. Image data were taken from repeated scanning from the vertical white line at 500 Hz. The data were recorded from this axon with a microelectrode containing Gβ1γ2. In part ii, the transient increase in fluorescence is seen in response to an action potential evoked by intracellular current injection. Part iii shows the signal integrated from part ii before (black) and 30 min after (red) the intracellular injection of Gβγ. Data are shown with the same time scale as in part ii of both (C) and (D). (D) As seen in part i, microinjection of Gβ1γ2 had no effect on stimulus-evoked action potentials [the axon shown in (C), parts i and ii]. The action potential is shown before (black) and 30 min after (red) the injection of Gβγ into the presynaptic axon. Part ii shows the same data but on a time scale compressed to that of the image [(C), part ii]. The axon was stimulated intracellularly to evoke an action potential.

GPCRs modulate voltage-gated Ca2+ channels in the soma of cultured lamprey dorsal cells (19), as in other preparations (20–25). We thus performed Ca2+ imaging experiments as described above. Control Ca2+ transients were recorded (Fig. 3C, parts ii and iii), and then the presynaptic axons were loaded with Gβ1γ2. Microinjected Gβ1γ2 had no significant effect on evoked Ca2+ transients (mean change in Ca2+ transient amplitude was to 99.6 ± 0.25% of the control's amplitude after injection of Gβ1γ2; P < 0.01; n = 6) (Fig. 3C, part iii). Similarly, Gβ1γ2 injection had no effect on the amplitude or duration of the presynaptic action potential (Fig. 3D, parts i and ii). Therefore, presynaptic Gβγ does not control evoked Ca2+ entry through voltage-gated Ca2+channels and must mediate its inhibitory effect on neurotransmitter release downstream of Ca2+ entry.

There are many known Gβγ effectors, including adenylyl cyclase II (AC II), phosphatidylinositol 3-kinase (PI 3-kinase), phospholipase C–β2 (PLC-β2), and several tyrosine kinases (5). Any of these effectors could mediate the inhibition of neurotransmitter release. We wanted to block all known Gβγ effectors, as well as any possible Gβγ regulation of Ser/Thr kinases or phosphatases, and determine whether serotonin inhibition of synaptic transmission could still occur. Application of forskolin has no effect on the serotonin-mediated inhibition of the EPSC (7). Therefore, it is unlikely that Gβγ works via AC II. We have shown that Gβγ does not cause a change in the internal Ca2+concentration (7); thus, PLC-β2 is not a likely target. Presynaptic phosphatases were blocked by injection of 200 mM free phosphate through the presynaptic recording pipette. Gβγ activation of PI 3-kinase was blocked by wortmannin (1 μM), activation of tyrosine kinases was blocked by the wide-spectrum tyrosine kinase inhibitor genistein (10 μM), and activation of Ser/Thr kinases was blocked by staurosporine (1 μM). This cocktail of inhibitors did not occlude the ability of serotonin to inhibit neurotransmission (n = 4; serotonin reduced the evoked response to 23.5 ± 12% of that of the control) (Fig. 4B). These data suggest that Gβγ inhibition of neurotransmitter release is not mediated via classical Gβγ effectors or via other unknown effectors that may involve Ser/Thr or Tyr kinases or phosphatases. Thus, we considered the possibility that Gβγ may act directly on the exocytotic machinery that mediates vesicle fusion.

Figure 4

Blockade of kinases and phosphatases does not prevent the inhibition of neurotransmission by serotonin (5-HT). Paired recordings between a reticulospinal axon and a motoneuron are shown. (A) Presynaptic stimulation evoked an action potential. This was not altered by the application of free phosphate to block phosphatases (200 mM KPO4 ) through the presynaptic recording electrode and the application of 1 μM wortmannin, 10 μM genistein, and 10 μM staurosporin to the superfusate to block PI 3-kinase, tyrosine kinases, and Ser/Thr kinases, respectively (control not shown). Subsequent addition of serotonin also did not alter the action potential. (B) The presynaptic action potential results in a postsynaptic EPSC with an electrical and chemical component. The application of phosphatase and kinase inhibitors neither altered the response (parts i and ii) nor prevented serotonin-mediated inhibition of neurotransmitter release (part ii). Data are the mean of 10 consecutive trials.

We conducted binding assays with purified Gβ1γ2 and glutathioneS-transferase (GST) fusion proteins syntaxin1A (amino acids 1 through 265), solubleN-ethylmaleimide–sensitive factor attachment protein (SNAP)–25B, and vesicle-associated membrane protein (VAMP) II (amino acids 1 through 94) (26). Gβ1γ2 binding to syntaxin (27) and SNAP-25 was detected while binding to ternary soluble N-ethylmaleimide–sensitive factor attachment protein receptor (SNARE) complexes consisting of syntaxin, SNAP-25, and VAMP was markedly enhanced in comparison to either syntaxin or SNAP-25 binding alone.

Various GPCRs and G proteins modulate synaptic neurotransmitter secretion at several sites either upstream from, at, or downstream from Ca2+ entry mechanisms. In neuronal cell bodies, in transfected cell lines, and in Xenopus oocytes (28–31), Gβγ directly activates a class of G protein–coupled inwardly rectifying K+ channels that would result in inhibited Ca2+ entry. GPCR-mediated inhibition of neurotransmitter release via inhibition of voltage-gated Ca2+ channels has also been demonstrated at one presynaptic terminal (32, 33). The ability of G proteins to mediate inhibition of synaptic neurotransmitter release at a step beyond Ca2+ entry was first demonstrated in the neuromuscular junction (34, 35). Spontaneous exocytotic events can be detected by recording miniature postsynaptic currents (mPSCs). mPSCs are modulated by many GPCRs (7, 36, 37). Biochemical studies also support the idea that G protein activation inhibits neurotransmitter or hormone release downstream from Ca2+entry, because the inhibitory effects of GPCR activation are preserved after cell permeabilization (38).

Our results show that Gβγ, and not Gα, is the active G protein subunit that mediates the Ca2+-independent pathway of a GPCR's presynaptic inhibition. We demonstrated that Gβγ inhibits neurotransmitter release downstream of Ca2+ entry mechanisms. By blocking endogenous Gβγ activity, we also showed that activation of a GPCR requires free Gβγ to inhibit neurotransmitter release, because a Gβγ scavenger blocked the serotonin inhibition. Moreover, the serotonin inhibition persisted in the presence of agents that block classical Gβγ effector pathways, indicating that Gβγ may affect transmitter release mechanisms directly. Gβγ binds SNARE proteins syntaxin (27), SNAP-25, and the ternary SNARE complex, suggesting that Gβγ could directly target the fusion machinery to inhibit vesicular release. The mechanism underlying GPCR-mediated presynaptic inhibition at a site distal to Ca2+ entry may involve a direct interaction between Gβγ and the core fusion machinery.

  • * To whom correspondence should be addressed. E-mail: sta{at} and heidi.hamm{at} (H.E.H.)

  • Present address: Department of Pharmacology, 442 Robinson Research Building, Vanderbilt University Medical Center, Nashville, TN 37232, USA.


Stay Connected to Science

Navigate This Article