Inhibition of dendritic Ca²⁺ spikes by GABA_B receptors in cortical pyramidal neurons is mediated by a direct G_i/o-βγ-subunit interaction with Ca_v1 channels

Abstract

Voltage-dependent calcium channels (VDCCs) serve a wide range of physiological functions and their activity is modulated by different neurotransmitter systems. GABAergic inhibition of VDCCs in neurons has an important impact in controlling transmitter release, neuronal plasticity, gene expression and neuronal excitability. We investigated the molecular signalling mechanisms by which GABA_B receptors inhibit calcium-mediated electrogenesis (Ca²⁺ spikes) in the distal apical dendrite of cortical layer 5 pyramidal neurons. Ca²⁺ spikes are the basis of coincidence detection and signal amplification of distal tuft synaptic inputs characteristic for the computational function of cortical pyramidal neurons. By combining dendritic whole-cell recordings with two-photon fluorescence Ca²⁺ imaging we found that all subtypes of VDCCs were present in the Ca²⁺ spike initiation zone, but that they contribute differently to the initiation and sustaining of dendritic Ca²⁺ spikes. Particularly, Ca_v1 VDCCs are the most abundant VDCC present in this dendritic compartment and they generated the sustained plateau potential characteristic for the Ca²⁺ spike. Activation of GABA_B receptors specifically inhibited Ca_v1 channels. This inhibition of L-type Ca²⁺ currents was transiently relieved by strong depolarization but did not depend on protein kinase activity. Therefore, our findings suggest a novel membrane-delimited interaction of the G_i/o-βγ-subunit with Ca_v1 channels identifying this mechanism as the general pathway of GABA_B receptor-mediated inhibition of VDCCs. Furthermore, the characterization of the contribution of the different VDCCs to the generation of the Ca²⁺ spike provides new insights into the molecular mechanism of dendritic computation.

Key points

• Voltage-dependent Ca²⁺ channels mediate a large repertoire of physiological actions, including the generation of dendritic spikes in neocortical pyramidal neurons; however, the type of Ca²⁺ channels involved in their generation remains unknown.

• We found that L-type Ca²⁺ currents generate the sustained plateau potential of the Ca²⁺ spike. GABA_B receptors inhibit Ca²⁺ spikes by specifically blocking dendritic L-type currents.

• This inhibition is mediated by a direct G_i/o-βγ-subunit interaction with the Ca_v1 channels.

• Protein kinases (protein kinase C and A) have an important influence on the generation and sustaining of dendritic Ca²⁺ spikes; however, their activity is not involved in the GABA_B-mediated inhibition of Ca²⁺ spikes.

• Inhibitory modulation of dendritic activity is important to understand the transformation of synaptic inputs into neuronal output activity. Our results shed light on the molecular mechanisms by which GABA acting via its GABA_B receptors can exert this inhibitory action.

Introduction

Voltage-dependent calcium channels (VDCCs) are a major source of calcium influx into neurons. They serve versatile roles in neuronal signaling, such as transmitter release, neuronal plasticity, gene expression and, very importantly, neuronal excitability (Tsien et al. 1988; Catterall, 2000; Catterall & Few, 2008). Different neurotransmitters regulate these fundamental physiological aspects by modulating the activity of VDCCs through their G-protein-coupled receptors (Catterall, 2000; Dolphin, 2003; Catterall & Few, 2008).

In particular, the neurotransmitter GABA acting through metabotropic GABA_B receptors can inhibit VDCCs (Dolphin & Scott, 1987) with important implications for neuronal function. Upon activation, GABA_B receptors inhibit presynaptic Ca_v 2.1 and 2.2 VDCCs (P/Q- and N-type Ca²⁺ currents, respectively), modulating synaptic release. This mechanism is exerted through a membrane-delimited pathway involving the βγ-subunit of a G_i/o protein (G_i/o-βγ; Kavalali et al. 1997; Bettler et al. 2004; Padgett & Slesinger, 2010). GABA_B receptor activation can also inhibit Ca_v1 VDCCs (L-type Ca²⁺ currents) in native neurons (Maguire et al. 1989; Marchetti et al. 1991; Scholz & Miller, 1991; Amico et al. 1995; Chalifoux & Carter, 2011), resulting in a reduction in somatic and dendritic Ca²⁺ influx. However, the mechanism by which this inhibition takes place is still unresolved. The possibility that this action is mediated by the G_i/o-βγ-subunit in a membrane-delimited fashion might a priori be discarded, as it has been shown that Ca_v1 VDCCs expressed in heterologous systems (Bourinet et al. 1996; Toth et al. 1996; Zhang et al. 1996) or in peripheral neurons (Plummer et al. 1989; Cox & Dunlap, 1992) can not be downregulated by neuromodulators that exert their physiological actions through this pathway. The remaining possibility would be that the GABA_B modulation of the Ca_v1 is exerted by the activation of cytosolic protein kinases (Catterall, 2000; Catterall & Few, 2008).

Here, we investigated this important functional link between GABA_B receptor activation and the inhibition of Ca_v1 channels. In the distal apical dendrite of neocortical pyramidal neurons, activation of GABA_B receptors leads to the inhibition of calcium-mediated electrogenesis, the so-called Ca²⁺ spike (Pérez-Garci et al. 2006; Breton & Stuart, 2012). These events are generated in a specific region of the apical dendrite and initial tuft dendrites about 600–900 μm from the pyramidal cell soma (Amitai et al. 1993; Yuste et al. 1994; Schiller et al. 1997; Larkum & Zhu, 2002; Larkum et al. 2009), suggesting a compartmentalization of the underlying VDCCs. Ca²⁺ spikes serve important functions in dendritic computation and can transiently switch action potential (AP) output from regular to burst firing mode (Larkum et al. 2001). They are suggested to act as coincidence detectors for correlating feedback inputs from higher cortical areas, preferentially innervating distal tuft inputs, with feedforward sensory information. Furthermore, Ca²⁺ spikes can amplify distal tuft synaptic inputs that are received in specific spatio-temporal sequences (Larkum et al. 1999b, 2009; Schaefer et al. 2003; Larkum & Nevian, 2008). Their actual influence on neuronal activity is tightly controlled by inhibitory inputs (Larkum et al. 1999b; Murayama et al. 2009; Gidon & Segev, 2012). GABA_B-mediated inhibition of Ca²⁺ spikes can occur by synaptic release of GABA from local interneurons (Pérez-Garci et al. 2006). However, very little is known about the composition of VDCCs in the Ca²⁺ spike initiation zone that contributes to the initiation and sustaining of the dendritic plateau potentials, as well as how GABA_B receptors interact with them to inhibit the Ca²⁺ spike. Given that Ca²⁺ spikes result in large elevations of intracellular Ca²⁺ concentration in this particular region and that they are strongly modulated by GABA_B receptor activation renders the apical dendrite of cortical layer 5 pyramidal neurons an ideal model system to study the interaction of GABA_B receptors with VDCCs in native central neurons.

Methods

Ethical approval

All experiments were approved by the veterinary office of the canton of Berne, Switzerland; and were carried out in compliance with The Journal of Physiology guidelines (Drummond, 2009).

Slice preparation

Acute cortical brain slices were obtained from Wistar rats (P28–P34). The animals were rapidly decapitated prior to brain extraction from the skull.

Parasagittal slices of the primary somatosensory cortex (300 μm thick) were cut with a vibrating microslicer on a block angled at 15 deg to the horizontal in ice-cold oxygenated artificial cerebral spinal fluid (ACSF) and then maintained at 37°C in ACSF for 15–120 min. Slices were perfused continuously with ACSF at 35 ± 2°C throughout the experiments. ACSF contained (in mm): NaCl, 125; NaHCO₃, 25; KCl, 2.5; NaH₂PO₄, 1.25; MgCl₂, 1; glucose, 25; CaCl₂, 2; pH 7.4.

Electrophysiology

Whole-cell recordings from somata and/or from dendrites were obtained with the aid of oblique illumination on a Nikon Eclipse E600FN or by infrared gradient contrast video microscopy on a Leica DMLFS microscope equipped with a 40× objective (Leica, HCX Apo, W40×, UVI, NA 0.8).

Somatic (4–6 MΩ) and dendritic (10–20 MΩ) recording pipettes were filled with an intracellular solution containing (in mm): potassium gluconate, 135; KCl, 7; Hepes, 10; Na₂-phosphocreatine, 10; Mg-ATP, 4; GTP, 0.3; 0.2% biocytin; pH 7.2 (with KOH); 291–293 mosmol l⁻¹. For double patch-clamp recordings, 10–20 μm Alexa 594 was added to the intracellular solution. For Ca²⁺ imaging, 300 μm Fluo-5F and 30 μm Alexa 594 were added to the internal solution. Dendritic Ca²⁺ spikes were evoked by injecting an EPSP-shaped current wave form: (double exponential shape: f(t) = (1 − e^−t/τ1)e^−t/τ2; where τ₁ = 4 ms and τ₂ = 10 ms; time to peak: 5 ms) via the dendritic pipette.

Two-photon calcium imaging

Two-photon-excited fluorescence was generated with a titanium-sapphire laser (Tsunami, Spectra-Physics) with laser pulses of 100 fs at a wavelength of 800 nm. The laser was coupled to a confocal scanning unit (TCS-SP2, Leica Microsystems) attached to an upright microscope (DMLFS, Leica), equipped with a 40× objective and non-descanned detectors. Red and green fluorescence signals were separated using a dichroic mirror (560DCXR, AHF) and corresponding bandpass filters (HQ520/25 and HQ585/40, AHF; Nevian & Sakmann, 2006). Raw fluorescence G(t) and R(t) was derived from line scans recorded at 800 Hz by integrating the fluorescence in regions of interest enclosing the fluorescent structures. Relative fluorescence changes were calculated as ΔG/R(t) = (G(t) − G₀)/R_ave, where G₀ is the mean resting green fluorescence during 50 ms before stimulation. The Ca²⁺-insensitive red fluorescence R_ave was calculated as the mean over the entire time course of the line scan recording (Nevian & Helmchen, 2007).

Pharmacology

Drugs were freshly prepared the same day of the experiment and consisted of (in μm): RS-baclofen, 10; nimodipine, 10; Gö6850 bisindolylmaleimide I, 1; KT 5720, 0.5; wortmannin, 0.2; Rp-cAMPS triethylammonium salt, 1000 (from TOCRIS Bioscience, Switzerland); Sp-cAMPS triethylammonium salt, 30 (from SIGMA-ALDRICH); SNX-482, 0.23; ω-agatoxin TK, 0.4; ω-conotoxin GVIA, 1 (from Alomone Labs, Israel).

For isolated dendritic L-type Ca²⁺ currents, the internal solution consisted of (in mm): caesium- methanesulphonate, 108; Hepes, 9; Na₂-phosphocreatine, 14; Mg-ATP, 4; Na-GTP, 0.3; 0.2% biocytin; pH 7.3 (with CsOH). The external solution included (in mm): TTX, 0.001; tetraethylammonium-chloride (TEA), 30; 4-aminopyridine (4-AP), 5; BaCl₂, 0.100; NiCl₂, 0.05 (Sigma-Aldrich); ω-conotoxin MVIIC, 0.0005 (from Alomone Labs).

When bath-applied, a minimum wait of 15 min was allowed to assess for the effect of the drug.

Local perfusion of baclofen (50 μm) onto the dendritic recording site was achieved by means of a puffing pipette (somatic pipette) located 50–100 μm lateral from the dendritic recording electrode. The range of area covered by the bolus expelled from the pipette was estimated to have a diameter of ∼180 μm, as measured by a test application of the fluorescent indicator Alexa 594 into a brain slice.

Data analysis

For each cell tested, the sweep showing the maximal effect of baclofen on the Ca²⁺ spike was used for analysis. The effectiveness of baclofen was quantified as the area underneath the control Ca²⁺ spike sensitive to the local application of baclofen to the apical tuft. This area was estimated by subtracting the baclofen-reduced Ca²⁺ spike from the control Ca²⁺ spike. For simplification, the area underneath the Ca²⁺ spike contributed by the back-propagating somatic AP was not considered in the analysis (although this may well have led to an underestimation of the physiological effect of activating GABA_B receptors in the apical tuft). Wilcoxon signed-rank tests for repeated measurements and Mann–Whitney U tests for independent samples were performed for statistical comparisons.

Results

Subcellular compartmentalization of VDCCs in the Ca²⁺ spike initiation zone

We studied the contribution of the different VDCCs to the generation of dendritic Ca²⁺ spikes as a first step to study their inhibition by GABA_B receptors. Dendritic whole-cell recordings were performed within the Ca²⁺ spike initiation zone of layer 5 neocortical pyramidal neurons around the main bifurcation of the apical dendrite (Fig. 1A). Dendrites were loaded with a Ca²⁺ indicator (Fluo-5F, 300 μm) and Alexa-594 (30 μm), and Ca²⁺ spikes were evoked by injecting an EPSP-shaped current with peak amplitudes of 800–1500 pA via the patch pipette into the dendrite. Ca²⁺ transients were measured in spines and adjacent dendritic shafts 50–150 μm distal from the recording pipette (Fig. 1B and C). Bath application of the Ca_v1 (L-type) VDCC blocker nimodipine (10 μm) resulted in an inhibition of the sustained depolarizing plateau that followed the initial depolarization of the Ca²⁺ spike and a large reduction of the dendritic Ca²⁺ transient by 55 ± 4% (n = 10; Fig. 1C and F). Ni²⁺ (50 μm), which blocks both Ca_v 3 (T-type) and Ca_v 2.3 (R-type) VDCCs prevented Ca²⁺ spike initiation in most cases, and abolished Ca²⁺ transients in both the spine and dendrite (Fig. 1D). Extra current (∼400 pA) injected into the dendrite recovered the Ca²⁺ spike, albeit with a significant decrease in the dendritic Ca²⁺ transient by 31 ± 8% (n = 7). SNX-482 (230 nm), a selective blocker of Ca_v 2.3 (R-type) VDCCs, had no effect on the initiation of the Ca²⁺ spike, but it resulted in a similar reduction of the dendritic Ca²⁺ transient (26 ± 1%; n = 3), indicating a large contribution of R-type conductances within the Ni²⁺-sensitive component. Ca_v 2.1 (P/Q-type) VDCCs, blocked by ω-agatoxin TK (400 nm), contributed 21 ± 1% (n = 4); and Ca_v 2.2 (N-type) VDCCs, blocked by ω-conotoxin GVIA (1 μm), contributed 20 ± 1% (n = 8) to the Ca²⁺ spike-evoked Ca²⁺ transients in the dendrite. The relative contribution of each VDCC subtype to the Ca²⁺ transients in spines was different (Fig. 1G). Ca_v1 VDCCs contributed 45 ± 4% to the Ca²⁺ transient in the spine, which was significantly less than in the parent dendrite (P < 0.05). Ca_v2.3 VDCCs contributed 34 ± 6% to the Ca²⁺ transient accounting for most of the combined R- and T-type contribution determined by application of Ni²⁺ (41 ± 7%). Ca_v2.1 VDCCs were found to contribute 21 ± 2% to the spine Ca²⁺ transient, whereas Ca_v2.2 VDCCs made no significant contribution (9 ± 5%).

Figure 1. Subcellular compartmentalized distribution of voltage-dependent calcium channels (VDCCs) and action of GABA_B receptors.

A, micrograph of a biocytin-filled L5 pyramidal neuron indicating the dendritic recording site. B, 2-photon fluorescence image (Alexa 594) of the recording site at the distal apical dendrite and a magnified view showing a spiny region in the apical tuft. Ca²⁺ imaging (Fluo-5) was measured by establishing a line scan (dashed line) covering a spine and its adjacent dendrite. C and D, dendritic Ca²⁺ spikes (top sweep) evoked by current injection via the recording pipette elicited Ca²⁺ transients in both the dendrite (middle panel) and in the spines (bottom panel; black sweeps). Coloured sweeps show representative examples of the action of different VDCCs blockers on the Ca²⁺ spikes and the resulting inhibition of Ca²⁺ fluorescence transients in both the dendrites and in the spines: nimodipine (10 μm, magenta traces), Ni²⁺ (50 μm, grey and blue sweeps, respectively, show recordings before and after injection of additional current via the recording pipette), SNX-482 (230 nm, light blue), ω-agatoxin TK (400 nm, green sweeps) and ω-conotoxin GVIA (1 μm, orange sweeps). E, baclofen added to the bath (10–30 μm) partially inhibited the Ca²⁺ spikes and their associated Ca²⁺ transients (red sweeps). F, summary bar plot showing the relative block of Ca²⁺ transients recorded in the dendrites for each drug tested. G, summary bar plot showing the relative block of Ca²⁺ transients recorded in the spines for each drug tested. Error bars indicate SEM. *P < 0.05, **P < 0.01.

GABA_B receptor activation inhibits Ca_v1 VDCCs

Ca²⁺ spikes initiated in the distal main apical dendrite of layer 5 pyramidal neurons are modulated by the local activation of GABA_B receptors inhibiting Ca²⁺ conductances (Pérez-Garci et al. 2006; Breton & Stuart, 2012). Therefore we assessed the effect of activating GABA_B receptors on the Ca²⁺ transients evoked by Ca²⁺ spikes. Bath application of the GABA_B agonist baclofen (10–30 μm) blocked the dendritic plateau potential that followed the upstroke of the Ca²⁺ spike and reduced the Ca²⁺ transients in dendrites by 59 ± 4% (P < 0.01; n = 9) and in spines by 52 ± 5% (P < 0.01; Fig. 1E–G). The effect of baclofen on the Ca²⁺ spike was similar to the effect of nimodipine in blocking the characteristic plateau potential and in the reduction of the Ca²⁺ transients associated to these events. These similarities suggested that GABA_B receptors might exert their modulatory effect on the Ca²⁺ spike by inhibiting Ca_v1 VDCCs.

To test this hypothesis, double patch-clamp recordings were performed from the soma and the distal apical dendrite (>700 μm from soma; Fig. 2) and Ca²⁺ spikes were evoked by dendritic current injection. The GABA_B agonist baclofen (50 μm) was applied by pressure ejection to the Ca²⁺ spike initiation zone using a third pipette located 50–100 μm lateral from the dendritic recording electrode to determine the effectiveness of GABA_B receptor inhibition on the Ca²⁺ spike after blockade of the various VDCCs. Under control conditions, dendritic depolarization evoked Ca²⁺ spikes that propagated towards the soma, evoking somatic APs. Subsequent transient activation of GABA_B receptors resulted in the inhibition of the depolarizing plateau that followed the initial upstroke of the Ca²⁺ spike and blocked somatic firing. The action of baclofen was completely reversed within 6 s after ceasing the pressure application. Block of Ca_v1 VDCCs either with nicardipine (5 μm; n = 6) or nimodipine (10 μm; n = 7) resulted in the inhibition of the plateau component of the Ca²⁺ spike leaving the fast initial depolarization intact. Under these conditions, baclofen reapplied locally to the apical dendrite did not cause further inhibition, indicating that blocking Ca_v1 VDCCs occluded the effect of baclofen (Fig. 2A). To test the possibility that the decrease in dendritic depolarization after blocking Ca_v1 channels affected the activation of other VDCCs that might be susceptible to modulation by GABA_B receptors, extra current (additional ∼400 pA) was injected in some experiments. Under these conditions somatic APs were reestablished; however, the effect of baclofen was still occluded (n = 5).

Figure 2. Activation of dendritic GABA_B receptors inhibits L-type Ca²⁺ conductances involved in the generation of Ca²⁺ spikes.

A–E, left columns: reconstruction of biocytin-filled L5 pyramidal neurons showing simultaneous somatic and distal (>700 μm) patch recordings, while a puff pipette expelled baclofen (50 μm) onto the apical tuft. Second columns: dendritic Ca²⁺ spikes (top, grey sweeps) were evoked by injecting an EPSP-shaped current waveform (middle, dashed sweeps) via the distal pipette. Dendritic spikes propagated towards the soma evoking APs (bottom, grey sweeps). Puffing baclofen on the dendrite shortened and reduced the amplitude of the Ca²⁺ spikes (red sweeps). The effect of baclofen was fully reversed after ceasing the local application of baclofen (blue sweeps). Third columns: control recordings and recordings obtained after bath application of different voltage-dependent calcium channels (VDCCs) blockers (black sweeps; A, nimodipne 10 μm; B, Ni²⁺ 50 μm; C, SNX-482, 230 nm; D, ω-agatoxin TK, 400 nm; E, ω-conotoxin GVIA, 1 μm). Fourth columns: reapplication of baclofen to the tuft in the presence of VDCCs blockers. In some cases larger current peaks were injected to the dendrite to reestablish the dendritic depolarization (see A–C). F, effectiveness of baclofen estimated as the area underneath the control Ca²⁺ spike sensitive to the local application of baclofen. Values obtained after blockade of VDCCs are normalized to those obtained before bath application of VDCCs blockers (see SI Materials and methods). Error bars indicate SEM. *P < 0.01.

As shown above, Ni²⁺ (50 μm) prevented the initiation of dendritic Ca²⁺ spikes near threshold current injection, but extra current (∼400 pA) injected into the dendrite reactivated the remaining VDCCs (Fig. 2B). Under these conditions, application of baclofen to the apical dendrite was still effective in inhibiting the Ca²⁺ spike (n = 5), indicating that the action of baclofen on the sustained component of the Ca²⁺ spike was not mediated by inhibition of Ni²⁺-sensitive (Ca_v3 and Ca_v2.3) VDCCs. Similarly, blocking Ca_v2.1, Ca_v2.2 and Ca_v2.3 VDCCs with ω-agatoxin TK (400 nm), ω-conotoxin GVIA (1 μm) and SNX-482 (230 nm), respectively, did not influence the effectiveness of baclofen to inhibit Ca²⁺ spikes (Fig. 2C–E). These results therefore showed that activation of GABA_B receptors in the distal apical dendrite exerted its inhibitory effect on dendritic Ca²⁺ spikes by specifically inhibiting Ca_v1 VDCCs (Fig. 2F).

After having identified the predominant effect of GABA_B-mediated inhibition of Ca²⁺ spikes to act on Ca_v1, we tested the action of baclofen on pharmacologically isolated L-type Ca²⁺ currents recorded from the distal apical dendrite directly (Fig. 3). Because voltage-clamp recordings in dendrites of neurons with an extensive dendritic arborization are limited by an incomplete space clamp of the dendritic membrane (Williams & Mitchell, 2008), we used this method only to phenomenologically assess the interaction of GABA_B receptors with L-type Ca²⁺ currents by comparing the relative effect of baclofen with control conditions. Dendritic patch-clamp recordings were performed with pipettes containing an internal Cs⁺-methanesulphonate-based solution. We first confirmed the proximity of the recording site to the dendritic Ca²⁺ spike initiation zone using current steps in current-clamp mode to evoke Ca²⁺ spikes in normal ACSF (Fig. 3A). The extracellular solution was then supplemented with a ‘cocktail’ for isolating L-type Ca²⁺ currents by blocking all other VDCCs, K⁺ channels and Na⁺ channels. In addition, Cs⁺ applied intracellularly effectively blocks I_h currents (Harris et al. 1994) and GIRK channels (Sodickson & Bean, 1996). L-type Ca²⁺ currents were then evoked in voltage-clamp mode by changing the dendritic command potential from −80 mV to 20 mV (Fig. 3B). Under these conditions, baclofen applied locally to the dendrite reversibly inhibited the L-type Ca²⁺ currents by 82 ± 12% (P < 0.01; n = 4; Fig. 3C). L-type Ca²⁺ currents subsequently recovered fully within about 3 s. Because the application of baclofen was spatially restricted (∼180 μm), the remaining ∼20% of the Ca²⁺ currents may be explained by L-type Ca²⁺ influx beyond the area affected. This suggested that under our experimental conditions the voltage-clamp step command still depolarized a stretch of dendrite beyond 180 μm sufficiently to activate L-type Ca²⁺ currents at even more distal sites. During the puff-application of baclofen we did not observe any change in the holding current required to clamp the dendrite at −80 mV, suggesting that baclofen did not activate or inactivate any potential leak conductances that would alter the spatial extend of the voltage-clamp. Subsequent bath application of nimodipine (10 μm) blocked the Ca²⁺ currents by 98.2 ± 8% (P < 0.01; n = 4; Fig. 3D), demonstrating the L-type nature of the recorded current. After having blocked L-type currents with nimodipine (and in the presence of the cocktail containing unspecific blockers for the common Na⁺, K⁺ and Ca²⁺ currents), a final application of baclofen was ineffective to evoke any detectable current corroborating the result that in this set of voltage-clamp experiments GABA_B activation did not modulate any other currents than L-type currents under our experimental conditions (data not shown).

Figure 3. Activation of dendritic GABA_B receptors inhibits dendritic L-type Ca²⁺ currents.

A, voltage recordings performed in current-clamp mode at a distal dendritic site (700 μm) immediately after seal rupture. The internal pipette solution included 108 mm Cs⁺. Dendritic Ca²⁺ spikes occurred with depolarizing current steps (lower panel). B, a depolarizing voltage command in voltage-clamp mode was applied to the dendrite in an external medium containing TTX (1 μm), tetraethylammounium-chloride (TEA; 30 mm), 4-aminopyridine (4-AP; 5 mm) and Ba²⁺ (100 μm). Voltage-dependent calcium channels (VDCCs) other than Ca_v1 were also blocked with Ni²⁺ (50 μm; T- and R-type) and ω-conotoxin MVIIC (500 nm; N- and P/Q-type). Under these conditions, inward L-type Ca²⁺ (after leak subtraction) currents were recorded (black sweep). Local application of the GABA_B agonist baclofen (50 μm) to the apical tuft inhibited the L-type Ca²⁺ currents (red sweep). The action of baclofen was reversed after 5 s ceasing the pressure application of baclofen (grey sweep). Subsequent bath application of nimodipine (10 μm) effectively blocked the L-type Ca²⁺ current (pink sweep). C, summary of the inhibitory effect of baclofen and nimodipine. Error bars indicate SEM. *P < 0.01.

These results show unequivocally that activation of GABA_B receptors in the distal apical dendrite directly inhibit Ca_v1 channels, which constitute the predominant conductance activated during the plateau potential of the dendritic Ca²⁺ spike.

Signalling mechanisms underlying the action of GABA_B inhibition on dendritic Ca²⁺ spikes

How is the observed interaction between GABA_B receptors and Ca_v1 channels mediated? GABA_B receptors act via a pertussis toxin-sensitive G_i/o protein (Bettler et al. 2004). They are known to inhibit Ca_v2 VDCCs (Kavalali et al. 1997; Bettler et al. 2004; Padgett & Slesinger, 2010) by a membrane-delimited pathway involving the βγ-subunit of the G_i/o protein. However, this form of inhibition has not been found for Ca_v1 VDCCs when studied in heterologous expression systems (Bourinet et al. 1996; Toth et al. 1996; Zhang et al. 1996) or in peripheral neurons (Plummer et al. 1989; Cox & Dunlap, 1992) so far. Alternatively, GABA_B receptors might inhibit Ca_v1 channels by cytosolic signalling pathways acting on protein kinases. Thus, we investigated the potential signalling mechanisms downstream of GABA_B receptor activation that led to the inhibition of Ca_v1 channels.

We first sought to test a direct interaction of the βγ-subunit with Ca_v1 channels. To this end we exploited the fact that the βγ-subunit inhibitory interaction with VDCCs is voltage dependent and can be transiently relieved by a relatively strong depolarization, or prepulse facilitation voltage command sequence (Bourinet et al. 1996; Herlitze et al. 1996; Ikeda, 1996; Toth et al. 1996; Zhang et al. 1996; Zamponi et al. 1997; De Waard et al. 2005). Therefore, we tested whether a prepulse voltage command was able to reverse the inhibitory effect of baclofen on isolated dendritic L-type Ca²⁺ currents (Fig. 4). Bath application of baclofen (10–30 μm) abolished the L-type Ca²⁺ current evoked by a voltage command from −80 mV to 20 mV (current integral compared with control; 2.3 ± 1%; P < 0.01; n = 4; Fig. 4B and C). Thereafter, a sequence of prepulse voltage commands consisting of 50 ms depolarizations from −80 mV to 70–100 mV preceding a 200 ms-long voltage step from −80 mV to 20 mV by 60 ms repeated five times every 5 s was applied to disrupt the potential βγ-subunit interaction with the Ca_v1 VDCCs. Indeed, after sessation of this prepulse stimulation protocol, we observed a partial recovery of the L-type Ca²⁺ current (57.5 ± 18%; Fig. 4B and C). This recovery lasted 10–30 s in the absence of the prepulse voltage commands, before baclofen exerted its inhibitory effect again (3.4 ± 2%; Fig. 4B and C). Addition of nimodipine to the bath solution maintained the complete block of the measured Ca²⁺ current as expected (3.1 ± 3%; Fig. 4B and C). However, in the presence of nimodipine, prepulse voltage commands were ineffective in rescuing the L-type Ca²⁺ current from GABA_B-mediated inhibition (5.8 ± 2.8%).

Figure 4. GABA_B inhibition of dendritic L-type Ca²⁺ currents is mediated by a direct G_i/o-βγ-subunit interaction.

A, experimental configuration of dendritic voltage-clamp experiments. B, pharmacologically isolated dendritic L-type Ca²⁺ currents (black trace) were evoked by injecting a test voltage command from −80 to 20 mV (test pulse; middle trace). Bath application of baclofen (10–30 μm) inhibited the L-type Ca²⁺ currents (red trace). A series of 5 prepulse voltage commands from −80 mV to 70 mV (50 ms duration) preceding a depolarization from −80 mV to 20 mV (200 ms) by 60 ms (prepulse; bottom middle trace), partially and transiently relieved the L-type Ca²⁺ current from its inhibition (blue trace). After recovery of the GABA_B-induced inhibition, nimodipine (10 μm) was supplemented to the bath (pink trace). In the presence of the Ca_v1 blocker, a new set of 5 prepulse voltage commands was ineffective to reverse the inhibition of L-type Ca²⁺ currents (cyan trace). The lower traces illustrate the sequence of test pulses and the prepulse protocol. C, summary of results. Error bars indicate SEM. *P < 0.01, relative to control; ♦P < 0.01, relative to baclofen inhibition (n = 4). D, bar graph showing the lack of variation in holding current values (at −80 mV) during the different experimental manipulations. E, sketch of the prepulse-induced relief of inhibition illustrating the interpretation of the current traces depicted in B.

During the course of the experiment, neither the tonic activation of GABA_B receptors by bath application of baclofen nor the prepulse voltage commands modified the holding current for a holding potential of −80 mV (Fig. 4D), arguing against the activation or inactivation of leak conductances that would compromise the extend of our voltage-clamp recordings and that would change the membrane length constant and therefore the number of activated Ca_v1 VDCCs. Thus, we conclude that our voltage-clamp conditions were stable and faithfully reflected the behaviour of the L-type current.

Not only was the prepulse protocol effective in recovering the L-type current, but also a 200 ms-long voltage step from −80 mV to 50–80 mV was able to reestablish the L-type Ca²⁺ currents, consistent with the voltage-dependence of the βγ-subunit interaction with VDCCs (Fig. S1; n = 3). After bath application of nimodipine the voltage steps applied did not uncover any additional currents demonstrating the L-type nature of the recorded currents. Even long (200 ms) current steps to +80 mV failed to activate any further currents (Fig. S1), arguing against the possibility that the activation of L-type currents at potentials of +20 mV resulted in a regenerative Ca²⁺-mediated event that escaped the voltage clamp and depolarized the membrane beyond the set level of +20 mV and thereby activated additional currents that required a higher threshold for activation.

These results demonstrated that the recovery from inhibition due to strong depolarization was indeed specific to the relief of the Ca_v1 VDCC. The most parsimonious explanation for the voltage dependency of GABA_B receptor-mediated inhibition of Ca_v1 channels is therefore a membrane-delimited pathway involving the G_i/o-βγ-subunit (Fig. 4E).

Next, we investigated if the GABA_B receptor-mediated inhibition of Ca²⁺ spikes could additionally be mediated by cytosolic signalling cascades. These pathways involve different enzymes that have VDCCs as their main targets: protein kinase C (PKC), cAMP-dependent protein kinases I and II (PKA), phosphatidylinositol 3-kinase (PI3K; Bettler et al. 2004; Deng et al. 2009) and phospholipase C (PLC; Sickmann et al. 2008).

We performed six pharmacological manipulations in order to test the influence of the activity state of these enzymes on the properties of dendritic Ca²⁺ spikes directly: pre-incubation of the slices with the inhibitor of PKC, bisindolylmaleimide I (1 μm); pre-incubation of the slices with the inhibitor of PKA, KT 5720 (500 nm); inclusion in the intracellular solution of the inhibitor of PKA, Rp-cAMPS (1 mm); inclusion in the intracellular solution of the activator of PKA, Sp-cAMPS (60–100 μm); inclusion in the intracellular solution of the PI3K inhibitor wortmannin (400 nm); and inclusion in the intracellular solution of the PLC inhibitor U73122 (1 μm). Of all these manipulations, only inhibition of PKC and inhibition of PKA affected dendritic Ca²⁺ spikes (Figs S2 and S3).

We next examined if these enzymes are activated by GABA_B receptors to inhibit dendritic Ca²⁺ spikes. Irreversible inhibition of PKC by bisindolylmaleimide I (1 μm) led to prolonged dendritic Ca²⁺ spikes (Fig. S2). This indicated that activation of PKC could be a signalling cascade to inhibit Ca²⁺ spikes. If GABA_B receptor activation upregulated the activity of PKC to inhibit dendritic Ca²⁺ spikes, pretreatment with bisindolylmaleimide I would occlude the effect of GABA_B activation with baclofen. However, we found that local application of baclofen to the apical tuft of neurons pretreated with bisindolylmaleimide I inhibited Ca²⁺ spikes to a similar extent as in untreated cells (untreated cells, 10.5 ± 1.6%; n = 10; bisindolylmaleimide I, 11.9 ± 0.2%; n = 3; P > 0.05; Fig. 5B and H).

Figure 5. GABA_B receptors inhibit Ca²⁺ spikes independent of protein kinases.

A, dendritic Ca²⁺ spike (black solid sweep) leading to somatic AP (black dashed sweep). Baclofen (50 μm; red sweeps) applied locally to the apical tuft induced inhibition of the Ca²⁺ spike. B–F, simultaneous double patch-clamp recording from the apical tuft and soma as in A, showing the baclofen-induced inhibition of Ca²⁺ spikes in cells pretreated with bisindolylmaleimide I (1 μm) to inhibit protein kinase C (PKC; B); Rp-cAMPS (1 mm) to inhibit cAMP-dependent protein kinases I and II (PKA; C); Sp-cAMPS (60–100 μm) to overactivate PKA (D); wortmannin (400 nm) to inhibit phosphatidylinositol 3-kinase (PI3K; E); and U73122 (1 μm) to inhibit phospholipase C (PLC; F). G, simultaneous double patch-clamp recording from the apical tuft and soma as in A, showing the baclofen-induced inhibition of Ca²⁺ spikes in the presence of Ba²⁺ (200 μm). H, summary of the inhibitory effect of baclofen (normalized to the area underneath the Ca²⁺ spike control) for each pharmacological condition. Error bars indicate SEM. *P < 0.05.

Irreversible inhibition of PKA with either KT 5720 or Rp-cAMPS decreased the duration or inhibited the generation of Ca²⁺ spikes (Figs S2 and S3). Under these conditions, the effect of baclofen could not be determined (Fig. 5C). However, the irreversible activation of PKA by adding Sp-cAMPS (60–100 μm) to the recording pipette did not occlude the effect of baclofen in inhibiting the Ca²⁺ spike (11.2 ± 3.4%, n = 4; P > 0.05; Fig. 5D and H). This shows that GABA_B-induced downregulation of the activity of PKA cannot explain the inhibitory action of baclofen. The inhibition of PI3K, a protein kinase known to be activated by G_i/o proteins (Salgado et al. 2007), was also ineffective in modifying the inhibitory effect of baclofen (9.4 ± 0.4%; n = 3; P > 0.05; Fig. 5E and H). Inhibition of PLC by U73122 has been reported to reduce the amplitude of GABA_B-mediated Kir3 responses in dissociated neocortical pyramidal neurons (Sickmann et al. 2008); however, intracellular application of U73122 did not modify the effect of baclofen on dendritic Ca²⁺ spikes, (11.8 ± 3.5%; n = 4; P > 0.05; Fig. 5F and H).

Finally, selective blockade of Kir3 K⁺ channels by Ba²⁺ (Newberry & Nicoll, 1984) did not prevent the action of baclofen (control 10.5 ± 1.6 vs. 21.4 ± 1.4% in 200 μm Ba²⁺; Fig. 5G and H), ruling out that the inhibition of this conductance by GABA_B receptors was responsible for the inhibition of the Ca²⁺ spike. The increase in dendritic Ca²⁺ spikes under Ba²⁺ could be attributed to the increased permeability of VDCCs to Ba²⁺ versus Ca²⁺ ions and/or the increase in local input resistance due to blockade of K⁺ channels.

Taken together our results indicate that the inhibitory effect of GABA_B receptors on Ca²⁺ spikes is mediated predominantly by the downregulation of Ca_v1 VDCCs rather than the activation of Kir3 K⁺ currents. Moreover, this modulation did not depend on cytosolic signalling cascades commonly associated with GABA_B receptors, but is likely mediated by a direct interaction of the Ca_v1 VDCCs with the G_i/o-βγ-subunit.

Discussion

Inhibitory modulation of dendritic function is gaining much attention as a key mechanism to influence dendritic computation (Murayama et al. 2009; Lovett-Barron et al. 2012; Palmer et al. 2012). In particular, the influence of inhibition on non-linear dendritic integration by the modulation of voltage-dependent ion channels in dendrites is important in order to understand the transformation of synaptic inputs to AP output. This study revealed a direct interaction of the GABA_B receptor-activated G-protein βγ-subunit with Ca_v1 channels in the distal apical dendrite of cortical layer 5 pyramidal neurons. Furthermore, we demonstrated that this modulation of Ca_v1 channels is the molecular mechanism for the inhibition of Ca²⁺ spikes, which are regenerative dendritic depolarizations important for signal integration in cortical pyramidal neurons.

Differential roles of VDCCs in the generation of Ca²⁺ spikes

We found that all VDCCs are present in the distal Ca²⁺ spike initiation zone of cortical pyramidal neurons but with different relative contributions to the evoked Ca²⁺ transients. Similarly, it has been found that all VDCCs are present in the proximal stretch of the apical dendrites of various pyramidal cells and are activated by somatically evoked back-propagating APs (Jaffe et al. 1992; Yuste et al. 1994; Christie et al. 1995; Markram et al. 1995; Sabatini & Svoboda, 2000) or synaptic activity (Markram & Sakmann, 1994; Yuste et al. 1994; Magee et al. 1995; Magee & Johnston, 1995b; Bloodgood & Sabatini, 2007). Comparison of our results to the previous findings indicates that the VDCC subtypes are differentially distributed along the apical dendrite of layer 5 pyramidal neurons. The contribution of Ca_v1 channels to evoked Ca²⁺ transients is much larger (>50% contribution) in the distal parts of the apical dendrite, in which Ca²⁺ spikes can be generated as compared with more proximal parts (20% contribution; Markram et al. 1995). This suggests that the Ca²⁺ spike initiation zone is a special compartment in terms of ion channel composition (Larkum et al. 2009).

The use of non-specific high-VDCC blockers such as cadmium and cobalt has been shown to block the sustained plateau potential of dendritic Ca²⁺ spikes (Kim & Connors, 1993; Schiller et al. 1997; Larkum et al. 1999a, 2001; Larkum & Zhu, 2002). TTX and Ni²⁺ affect the initial fast component of Ca²⁺ spikes, pointing to Na⁺ and low-voltage-activated T-type Ca²⁺ currents as being responsible for the initiation of Ca²⁺ spikes (Larkum et al. 2001; Schaefer et al. 2003); however, the concentration of Ni²⁺ used in previous reports (>50 μm) is also known to block Ca_v2.3 channels (R-type; Soong et al. 1993; Randall & Tsien, 1997). Furthermore, Ca_v3 (T-type) conductances on their own can not account for the generation and duration of the Ca²⁺ spike (up to 50 ms; Larkum et al. 2001) given the fast inactivation kinetics (25 ms at 0 mV) and low percentage of available Ca_v3 channels at the dendritic resting membrane potential (<20% at −55 mV; Magee & Johnston, 1995a; Magee, 2008). We found that Ni²⁺ (50 μm) inhibited the generation of Ca²⁺ spikes to a large extent. The effect of Ni²⁺ was to increase the threshold for Ca²⁺ spike initiation because extra current injected to the distal dendrite reactivated Ca²⁺ spikes and their associated Ca²⁺ transients. Blockade of one fraction of the Ni²⁺-sensitive conductance (Ca_v2.3 channels) revealed a significant contribution of R-type conductances to the Ca²⁺ spike not previously reported. Our results are consistent with the apparent lack of T-type Ca²⁺ currents observed in nucleated patch-clamp recordings from somatosensory layer 5 pyramidal neurons (Almog & Korngreen, 2009).

Here we show that Ca_v1 (L-type) channels contribute predominantly to the sustained (>30 ms) dendritic plateau potential. It is this slow and long-lasting part of the dendritic Ca²⁺ spike that determines the number and pattern of somatic APs (Larkum et al. 2001). In this respect, Ca²⁺ spikes and Ca_v1 VDCCs are important for the computational properties of the thick tufted pyramidal neurons, such as performing coincidence detection and non-linear integration of synaptic inputs (Larkum et al. 1999b, 2009; Schaefer et al. 2003). All other VDCCs contribute to the Ca²⁺ signal, but they do have only a minor effect on the electrogenesis.

GABA_B modulation of VDCCs

Spines from apical dendrites are biochemically independent compartments, well isolated from their parent dendrites (Muller & Connor, 1991; Yuste & Denk, 1995; Svoboda et al. 1996; Sabatini & Svoboda, 2000) that contain VDCCs (Sabatini & Svoboda, 2000; Sabatini et al. 2001; Bloodgood & Sabatini, 2007). Evidence from hippocampal pyramidal neurons has shown a differential compartmentalization of VDCCs between shafts and spines where Ca_v2.3 channels form functional microdomains within the spine (Sabatini & Svoboda, 2000; Bloodgood & Sabatini, 2007). In particular, GABA_B receptor activation has been shown to inhibit Ca²⁺ transients evoked by somatic back-propagating APs only in the spines of CA1 pyramidal neurons, by inhibiting Ca_v2.3 channels (Sabatini & Svoboda, 2000). Our findings show that activation of GABA_B receptors or blockade of Ca_v1 VDCCs both strongly inhibited Ca²⁺ transients evoked by Ca²⁺ spikes in the distal apical dendritic shaft as well as the neighbouring spines. This result points to a fundamental difference between cortical and hippocampal pyramidal neurons in terms of Ca²⁺ signalling and its modulation by GABA_B receptors. In agreement with our results, recent work suggests that GABA_B receptor activation inhibits Ca²⁺ transients in the dendritic shafts of L2/3 cortical pyramidal neurons, evoked by back-propagating APs. This effect was only partially occluded by nimodipine, indicating a GABA_B-mediated inhibition of Ca_v1 VDCCs in addition to other VDCCs in these cells (Chalifoux & Carter, 2011). In the distal dendrites of L5 pyramidal neurons, only the blockade of Ca_v1 VDCCs occluded the postsynaptic actions of GABA_B receptors. Furthermore, local activation of GABA_B receptors inhibited isolated L-type Ca²⁺ currents recorded directly at the site of generation of Ca²⁺ spikes, showing unequivocally that GABA_B receptors exert their inhibitory action on dendritic Ca²⁺ spikes by inhibiting L-type Ca²⁺ conductances. Thus, the functional coupling of GABA_B receptors to VDCCs depends on cell type and/or the subcellular compartment.

GABA_B receptors are localized along the apical dendrite of pyramidal neurons. At the subcellular level they are present in both dendritic spines and their parent dendritic shafts (Kulik et al. 2003). At the spine level, GABA_B receptors form clusters together with Kir3.2 inwardly rectifying K⁺ channels, whereas in the shaft both proteins are segregated (Kulik et al. 2006). This raises the possibility that GABA_B receptors in the shaft might be co-localized to Ca_v1 channels, although this has to be determined at the electron-microscopic level. Such a degree of co-localization is consistent with a membrane-delimited coupling between these proteins via a G_i/o protein (Karschin, 1999; Bettler et al. 2004; Pinard et al. 2010). Although this form of regulation has been shown for GABA_B–Kir3.2 and GABA_B–Ca_v2 interactions, it has never been demonstrated for GABA_B–Ca_v1. Previous evidence already suggested an inhibitory role of GABA_B receptors on L-type Ca²⁺ conductances even though the mechanism for this interaction remained elusive (Maguire et al. 1989; Marchetti et al. 1991; Scholz & Miller, 1991; Amico et al. 1995; Chalifoux & Carter, 2011). Our results now show for the first time that GABA_B receptors can exert this modulatory action on dendritic Ca_v1 channels (L-type currents) via a membrane-delimited pathway. Three lines of evidence support this view. Firstly, GABA_B-mediated inhibition persisted after pharmacological manipulation of different cytosolic protein kinases known to be associated with the activation of G_i/o proteins, namely PKA, PKC and PIK3 (Bettler et al. 2004; Salgado et al. 2007; Deng et al. 2009). Secondly, prepulse facilitation relieved the GABA_B-mediated inhibition of dendritic L-type Ca²⁺ currents and, thirdly, L-type Ca²⁺ currents could be recovered by strong depolarizing voltage commands. The prepulse facilitation protocol has been shown to be effective for reversing the G_i/o-βγ-mediated inhibition of Ca_v2, but so far not for Ca_v1 VDCCs (Bourinet et al. 1996; Herlitze et al. 1996; Ikeda, 1996; Toth et al. 1996; Zhang et al. 1996; Zamponi et al. 1997; De Waard et al. 2005). Strong depolarizing voltages have been assumed to promote the uncoupling of the G_i/o-βγ-dimer from the βγ-binding site located in the intracellular loop I–II of the pore-forming α-subunit of the channel (De Waard et al. 2005). Indeed, this loop has been shown to act as a voltage sensor (Sandoz et al. 2004), and its movement during membrane depolarization might promote the dissociation. In addition, the recovery of L-type currents at higher depolarizations is consistent with a model by which the G_i/o-βγ complex shifts the activation curve of Ca_v1 VDCCs to more depolarized values, indicating their transition to a ‘reluctant state’ (Fig. 4D; De Waard et al. 2005).

The G_i/o-βγ-mediated regulation of VDCCs has mainly been tested in heterologous expression systems. This raises the possibility that native central neurons possess preserved Ca_v1 VDCCs susceptible to regulation through this pathway (Bourinet et al. 1996). Alternatively, G_i/o-βγ-mediated modulation might require additional factors like accessory proteins that were present in the specialized region of the neuron we investigated, but which were absent in the heterologous expression systems previously used. Our findings expand the repertoire of interactions between G-protein-coupled receptors and VDCCs opening new vistas in neuromodulation. Furthermore, our result implies that the G_i/o-βγ-subunit interaction with VDCCs is a common mechanism of action by which GABA_B receptors inhibit VDCCs.

Glossary

4-AP

4-aminopyridine

ACSF

artificial cerebral spinal fluid

AP

action potential

G_i/o-βγ

βγ-subunit of a G_i/o protein

PI3K

phosphatidylinositol 3-kinase

PKA

cAMP-dependent protein kinases I and II

PKC

protein kinase C

PLC

phospholipase C

TEA

tetraethylammonium-chloride

VDCCs

voltage-dependent calcium channels

Author contributions

E.P.-G., M.E.L. and T.N.: designed the experiments; E.P.-G. and T.N.: performed and analysed the experiments; E.P.-G., M.E.L. and T.N. wrote the manuscript. All authors approved the final version for publication.