Abstract
Although primarily studied at the cell body, GABAB receptors (GABABRs) are abundant at spines and dendrites of cortical pyramidal neurons, where they are positioned to influence both synaptic and dendritic function. Here, we examine how GABABRs modulate calcium (Ca) signals evoked by action potentials (APs) in spines and dendrites of layer 2/3 pyramidal neurons in mouse prefrontal cortex. We first use two-photon microscopy to show that GABABRs inhibit AP Ca signals throughout the entire dendritic arbor of these neurons. We then use local pharmacology and GABA uncaging to show that dendritic GABABRs also decrease the input resistance, shorten the AP afterdepolarization, and generate inhibitory postsynaptic potentials. However, we find that these electrophysiological effects recorded at the cell body do not correlate with the inhibition of AP Ca signals measured in spines and dendrites. Instead, we use voltage-clamp recordings to show that GABABRs directly inhibit several subtypes of voltage-sensitive calcium channels (VSCCs) in both spines and dendrites. Given the importance of VSCC-mediated Ca signals for neuronal function, our results have implications for the functional role of dendritic GABABRs in the prefrontal cortex and throughout the brain.
Introduction
GABA is the major inhibitory neurotransmitter in the cortex, activating both ionotropic and metabotropic receptors (Sivilotti and Nistri, 1991). GABAB receptors (GABABRs) are G-protein-coupled receptors that modulate neuronal excitability (Bowery et al., 2002). GABABRs are highly expressed in the prefrontal cortex (PFC) (Margeta-Mitrovic et al., 1999) and regulate higher cognitive function (Mott and Lewis, 1991). They are also therapeutic targets for neuropsychiatric disorders, including schizophrenia and epilepsy (Bettler et al., 2004).
Although GABABRs have been primarily studied at the cell body, they are also abundant at spines and dendrites (Kulik et al., 2003), which receive many inhibitory inputs (Beaulieu and Somogyi, 1990). GABABRs open potassium (K) channels (Sodickson and Bean, 1996; Lüscher et al., 1997), also found at spines and dendrites (Chen and Johnston, 2005; Kulik et al., 2006). Activating K channels increases membrane conductance and evokes slow IPSPs (sIPSPs) (Newberry and Nicoll, 1984). These effects can suppress postsynaptic responses (Morrisett et al., 1991; Otmakhova and Lisman, 2004) and limit action potential (AP) backpropagation (Buzsáki et al., 1996; Tsubokawa and Ross, 1996; Leung and Peloquin, 2006).
GABABRs also inhibit many voltage-sensitive calcium (Ca) channel (VSCC) subtypes (Mintz and Bean, 1993; Pfrieger et al., 1994; Lambert and Wilson, 1996). VSCCs are the primary source of AP Ca signals in spines (Carter and Sabatini, 2004; Bloodgood and Sabatini, 2007) and dendrites (Magee and Johnston, 1995b; Markram et al., 1995). They also contribute to subthreshold synaptic responses (Magee et al., 1995; Bloodgood and Sabatini, 2007), suprathreshold dendritic spikes (Losonczy and Magee, 2006; Pérez-Garci et al., 2006; Larkum et al., 2007), and the induction of synaptic plasticity (Huang and Malenka, 1993; Zucker, 1999). By inhibiting spine and dendrite VSCCs, GABABRs may control many fundamental Ca-dependent neuronal processes.
The apical and basal dendrites of pyramidal neurons receive distinct inputs (Larkman, 1991; Petreanu et al., 2009) and have unique electrophysiological properties (Häusser et al., 2000; Spruston, 2008). In CA1 pyramidal neurons, GABABRs inhibit AP Ca signals in apical spines but not nearby dendrites or basal spines and dendrites (Sabatini and Svoboda, 2000). In cortical pyramidal neurons, this type of inhibition may prevent dendritic Ca spikes in the apical tufts (Pérez-Garci et al., 2006; Larkum et al., 2007). In layer 2/3 pyramidal neurons, GABABRs are also found in basal spines, where they strongly inhibit NMDA receptor Ca signals (Chalifoux and Carter, 2010). However, the extent to which GABABRs modulate AP Ca signals in spines and dendrites of these neurons remains unknown.
Here, we examine GABABR modulation in spines and dendrites of layer 2/3 pyramidal neurons in acute mouse PFC slices. We use two-photon microscopy to show widespread inhibition of AP Ca signals throughout the dendritic arbor. Although GABABRs also decrease the input resistance, shorten the AP afterdepolarization and generate sIPSPs, these effects do not cause the inhibition of AP Ca signals. Instead, we find direct inhibition of multiple VSCC subtypes in both spines and dendrites. Our results have important implications for the subcellular roles of GABABRs in the PFC and throughout the brain.
Materials and Methods
Preparation.
Recordings were made from layer 2/3 pyramidal neurons in the medial PFC of acute slices from male and female, postnatal day 21 (P21) to P28 C57BL/6 mice, as previously described (Chalifoux and Carter, 2010). Mice were anesthetized with a lethal dose of ketamine/xylazine and perfused intracardially with ice-cold external solution containing the following (in mm): 65 sucrose, 75 NaCl, 25 NaHCO3, 1.25 NaH2PO4, 25 glucose, 2.5 KCl, 1 CaCl2, 5 MgCl2, 0.4 Na-ascorbate, 3 Na-pyruvate (295–305 mOsm), bubbled with 95% O2/5% CO2. Coronal slices (300 μm thick) were cut in ice-cold external solution and transferred to artificial CSF (ACSF) containing the following (in mm): 119 NaCl, 25 NaHCO3, 1.25 NaH2PO4, 21 glucose, 2.5 KCl, 2 CaCl2, 1 MgCl2, 0.4 Na-ascorbate, 3 Na-pyruvate (295–305 mOsm), bubbled with 95% O2/5% CO2. After 30 min in ACSF at 35°C, slices were stored for ∼30 min at 24°C, after which experiments were conducted at 32–34°C.
In all experiments, the ACSF contained 10 μm 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide (NBQX), 10 μm 3-((R)-2-carboxypiperazine-4-yl)-propyl-1-phosphonic acid [(R)-CPP], and 10 μm gabazine to block AMPARs, NMDARs, and GABAARs, respectively. In some experiments, GABABRs were activated by wash-in of 5 μm baclofen or blocked by wash-in of 2 μm (2S)-3-[[(1S)-1-(3,4-dichlorophenyl)ethyl]amino-2-hydroxypropyl]-(phenylmethyl)phosphinic acid (CGP-55845). For GABA uncaging experiments, the ACSF contained 10 μm RuBi-GABA. For voltage-clamp experiments, the ACSF also contained 1 μm TTX to block Na channels. In some of these experiments, the ACSF contained 2 mm BaCl2 to block K2P channels or 3 mm CsCl2 to block Kir channels. The mixture of VSCC blockers consisted of 10 μm mibefradil, 20 μm nimodipine, 0.3 μm SNX-482, and 1 μm ω-conotoxin-MVIIC. All chemicals were from Sigma-Aldrich or Tocris, with the exception of SNX-482 (Alomone Labs).
Physiology recordings.
Whole-cell recordings were obtained from layer 2/3 pyramidal neurons identified with infrared–differential interference contrast at 200–300 μm from the pial surface. Borosilicate recording pipettes (2–4 MΩ) were filled with one of two internal solutions. Current-clamp recordings used the following (in mm): 135 K-gluconate, 7 KCl, 10 HEPES, 10 Na-phosphocreatine, 4 Mg2-ATP, 0.4 Na-GTP, 290–295 mOsm, pH 7.35 with KOH. Voltage-clamp recordings used the following (in mm): 135 Cs-gluconate, 10 HEPES, 10 Na-phosphocreatine, 4 Mg2-ATP, 0.4 Na-GTP, 290–295 mOsm, pH 7.35 with CsOH. Solutions contained 200 μm Fluo-5F to monitor Ca levels and 20 μm Alexa Fluor 594 to image neuronal morphology. Neurons were filled via the patch electrode for at least 15–20 min (for proximal regions; <75 μm) or 25–30 min (for distal regions; >75 μm) before imaging. Dye concentrations were chosen to ensure that Ca signals were in the linear range of the indicators (Carter and Sabatini, 2004).
Recordings were made using a MultiClamp 700B amplifier, filtered at 5 kHz for current-clamp recordings and 2 kHz for voltage-clamp recordings, and sampled at 10 kHz. For current-clamp recordings, APs were triggered with brief (2 ms) current injections (1500–3000 pA) through the recording pipette. Input resistance (Rin) and series resistance (Rs) were calculated using long (300 ms) hyperpolarizing current injections (−30 pA). After wash-in of 5 μm baclofen, to counteract a small somatic hyperpolarization (−3.5 ± 2 mV; n = 5 cells), cells were held at −70 mV by injecting a small positive current (<80 pA), which was necessary to use the same current injection for triggering action potentials. In control experiments, we found that 5 μm baclofen inhibited the [Ca]bAP to the same degree in spines and dendrites when the soma was allowed to hyperpolarize (supplemental Fig. S1, available at www.jneurosci.org as supplemental material).
For baclofen puffing, a glass pipette with a 2–4 μm tip was bent using a microforge (Narishige) to allow focal application of baclofen at different depths of the slice without disturbing the whole-cell recording. The puffing pipette was filled with ACSF containing the following (in μm): 10 NBQX, 10 (R)-CPP, 10 gabazine, and 15 baclofen and was pressure-ejected (5–15 psi) using a Picospritzer. In separate experiments, tip size and pressure were empirically determined using ACSF with 20 μm Alexa 488 to visualize the spread of the ejected ACSF (see Fig. 4). The tip of the puffing pipette was placed 5–10 μm from the imaged spine–dendrite pair.
In voltage-clamp recordings, cells were held at −70 mV and VSCCs were opened by steps to +10 mV. The step duration ranged from 3 to 6 ms and was adjusted to produce baseline Ca signals similar to those evoked by action potentials in current-clamp recordings. Spine–dendrite pairs were chosen within 50 μm from the soma to minimize voltage-clamp errors. Rin and Rs were calculated using a 300 ms, 10 mV hyperpolarizing voltage step. For voltage-clamp recordings, data was discarded if the Rs changed by >15%.
For VSCC experiments, the selective blocker and then baclofen were sequentially added while recording from the same spine–dendrite pair. In control experiments, ACSF was substituted for the VSCC blocker. Drugs were allowed to wash-in for 5 min before acquiring data. For each recording condition, a total of 10 APs were collected at 0.07 Hz.
To predict the amount of modulation in the presence of blockers of VSCCs, we used the following: where λ represents percentage of [Ca]step remaining after inhibition by baclofen in the presence of a selective VSCC blocker; Δbac, percentage of [Ca]step inhibited by baclofen without the VSCC blocker; θ, fraction of the VSCC subtype that is inhibited by baclofen (0–1); Δvscc, percentage of [Ca]step eliminated by the VSCC blocker; Δr, percentage of [Ca]step contributed by SNX-resistant R-type VSCCs. The numerator, (Δbac − θΔvscc), represents the remaining percentage of VSCCs that can be modulated by baclofen. Because we cannot block the SNX-resistant R-type VSCCs, this numerator will never be less than Δr. The denominator, (1 − Δvscc), represents the percentage of remaining VSCCs after addition of the selective VSCC blocker. Full block of a channel subtype by baclofen (θ = 0.9–1) and no modulation of a channel subtype (θ = 0–0.1) were calculated and plotted as gray boxes in Figure 8.
Two-photon microscopy.
Intracellular Ca imaging was performed with a custom microscope, as previously described (Carter and Sabatini, 2004; Carter et al., 2007). For two-photon laser-scanning microscopy (2PLSM), 810 nm light was used to excite Fluo-5F (green) and Alexa Fluor 594 (red), to monitor Ca signals and spine morphology, respectively. Line scans with current injection were interleaved with line scans with no current injection to detect any photobleaching. Reference frame scans were taken between each acquisition to correct for small spatial drift of the preparation over time. Baseline fluorescence was monitored and recordings were discarded if an increase was detected, which would indicate photodamage.
To quantify Ca signals, green and red fluorescence were collected during 500 Hz line scans across a spine–dendrite pair. Ca signals were quantified as changes in green fluorescence to red fluorescence (ΔG/R), normalized to the maximal green fluorescence to red fluorescence (Gsat/R), giving ΔG/Gsat. The value of Gsat/R was measured after each recording using a thin-walled pipette, containing the internal solution with a saturating concentration of Ca, which was positioned directly above the recorded cell and used at the same recording temperature (32–34°C). To create an internal solution with saturating Ca, 1 m CaCl2 was added at 1:25 to the internal.
GABA uncaging.
The back focal plane of the objective was filled with collimated 473 nm light from a DPSS laser (Laserglow Technologies) using a fiber optic cable (Thorlabs), planoconvex lenses (Edmund Optics), and a dichroic mirror (Chroma). Laser power was modulated by the variable current power supply and pulse duration was controlled with a TTL trigger. A fast shutter (Uniblitz) was placed before the photomultiplier tubes to protect them from the uncaging laser light while imaging. A telescope using two planoconvex lenses was used to focus the collected fluorescence through the shutter aperture.
Uncaging of 10 μm RuBi-GABA was evoked using 10–13 mW laser power for 10 ms on spines <25 μm from the surface of the slice. This power and duration was chosen because it elicited reliable sIPSPs at the soma and did not cause photodamage in the spine or dendrite after repeated uncaging (detected by increases in baseline fluorescence). In Figure 5, sIPSPs were measured at the soma as the uncaging point was moved laterally away from the soma. Because of radial branching of the basal dendrites, it was not possible to completely avoid uncaging on dendritic branches. In Figure 6, test trials consisted of an uncaging pulse followed with a 200 ms delay by a somatically evoked AP. These were interleaved with control trials with no uncaging pulse for comparison with baseline. Additional trials with no AP were used to subtract out any light-evoked phosphorescence and to obtain a stable baseline.
Data acquisition and analysis.
Image and physiology data were acquired using National Instruments boards and custom software written in Matlab (The MathWorks). Off-line analysis was performed using custom routines written in Igor Pro (Wavemetrics). Ca signal amplitudes were calculated over a 20 ms window, starting after the end of the stimulus (102–122 ms). sIPSP amplitudes were calculated over a 10 ms window around the peak. Two-photon images were treated with a 1.5 pixel radius Gaussian filter for display purposes. Average traces in Figure 8C were low-pass filtered for display purposes.
Electrophysiological and imaging data are reported in the text as median ± SE. The SE was calculated as the SD of the medians computed from 10,000 bootstrapped samples from the data. Traces in figures are the arithmetic mean ± SEM. There were occasionally small differences between the median and mean within a group, which explains some discrepancies between traces in figures and medians in the text and supplemental Table S1 (available at www.jneurosci.org as supplemental material). Summary data in figures is in box plot form, showing the median, interquartile range, 10–90% range (whiskers), and including the data from individual experiments (open circles). Significance was defined as p < 0.05 (*) and determined using the nonparametric two-tailed Wilcoxon–Mann–Whitney two-sample rank test or the two-tailed Wilcoxon signed rank test for paired data (when appropriate), neither of which make assumptions about the data distribution. Correlations are shown in scatter plots as straight lines and were calculated using a linear regression. These correlations are reported in figures as Spearman's rank correlation coefficient (ρ), which makes no assumption about distributions, but differed slightly in some cases from the linear regression (see Figs. 3, 4, 6).
Results
We used whole-cell recordings and two-photon imaging to investigate subcellular GABABR modulation in layer 2/3 pyramidal neurons of acute slices of mouse PFC (Fig. 1A). We found that somatic APs backpropagated throughout the dendritic arbor and triggered Ca signals ([Ca]bAP) in spines and dendrites (Fig. 1B). [Ca]bAP was mostly blocked by a mixture of VSCC blockers, indicating they were attributable to the opening of VSCCs (see Materials and Methods) (p = 0.008; n = 4 cells/8 spine–dendrite pairs) (Fig. 1C). The amplitude of [Ca]bAP in spines and dendrites decreased exponentially away from the soma, with similar length constants in apical dendrites (177 μm; n = 160) and basal dendrites (180 μm; n = 155) (Fig. 1D). In contrast to previous findings in the main apical dendrite of layer 2/3 pyramidal neurons in somatosensory cortex (Svoboda et al., 1997; Waters et al., 2003), we do not observe a peak dendritic Ca signal at 100 μm from the soma. Although signals were smaller in the distal dendrites, the ability to record [Ca]bAP throughout the dendrites allowed us to compare GABABR modulation in different subcellular compartments.
Apical dendrites
Previous results in CA1 pyramidal neurons indicate that GABABRs modulate [Ca]bAP only in apical oblique spines but not dendrites (Sabatini and Svoboda, 2000). We found that bath application of 5 μm baclofen, a selective activator of GABABRs, decreased [Ca]bAP in apical oblique spines (p = 0.007; n = 8/16) (Fig. 2A,B) (radial distance: median, 48 μm; range, 20–91 μm) (for full quantification of Ca imaging data, see supplemental Table S1, available at www.jneurosci.org as supplemental material). In contrast, wash-in of ACSF as a vehicle control had no effect on these signals (p = 0.85, n = 4/8; baclofen vs ACSF, p = 0.001). However, baclofen also decreased [Ca]bAP in apical oblique dendrites (p = 0.0005), whereas ACSF again had no effect (p = 0.20; baclofen vs ACSF, p = 0.0001). Modulation was not significantly different between spines and dendrites in these or any of the following experiments. Thus, in contrast to CA1 pyramidal neurons, GABABRs inhibit [Ca]bAP in apical oblique spines and dendrites of layer 2/3 pyramidal neurons.
The main apical dendrite emerges from the cell body, projects to layer 1, and also receives excitatory input onto dendritic spines (Larkman, 1991). Similar to the apical obliques, we found that baclofen significantly decreased [Ca]bAP in main apical spines (p = 0.005, n = 7/11; ACSF, p = 0.27, n = 8/12; baclofen vs ACSF, p = 0.02) and dendrites (p = 0.001; ACSF, p = 0.02; baclofen vs ACSF, p = 0.0001) (Fig. 2B; supplemental Fig. S1, available at www.jneurosci.org as supplemental material) (radial distance: median, 73 μm; range, 17–125 μm). The small but significant decrease in the dendritic [Ca]bAP during ACSF controls may reflect gradual washout of cellular factors and was not seen in other compartments.
The apical tuft is distal to the bifurcation of the main apical dendrite and receives a distinct complement of synaptic inputs (Petreanu et al., 2009). Previous work indicates that distal GABABRs can inhibit the generation of dendritic Ca spikes (Pérez-Garci et al., 2006), but it is unknown whether the inhibition occurs in spines, dendrites, or both compartments. We found that baclofen significantly reduced [Ca]bAP in apical tuft spines (p = 0.03, n = 7/14; ACSF, p = 0.49, n = 9/17; baclofen vs ACSF, p = 0.03) and dendrites (p = 0.0005; ACSF, p = 0.49; baclofen vs ACSF, p = 0.006) (Fig. 2B; supplemental Fig. S1, available at www.jneurosci.org as supplemental material) (radial distance: median, 128 μm; range, 83–211 μm). Together, our results indicate that GABABR modulation of [Ca]bAP is widespread across apical spines and dendrites of layer 2/3 pyramidal neurons.
Basal dendrites
The basal dendrites of pyramidal neurons receive the majority of synaptic contacts (Larkman, 1991) and may integrate inputs differently than apical dendrites (Schiller et al., 2000). Previous studies in CA1 pyramidal neurons show that GABABRs do not modulate [Ca]bAP in basal spines or dendrites (Sabatini and Svoboda, 2000). We initially recorded [Ca]bAP in basal spines and dendrites proximal to the cell body (radial distance: median, 43 μm; range, 20–75 μm). We found that bath application of 5 μm baclofen significantly decreased [Ca]bAP in proximal basal spines (p = 0.0002, n = 5/10; ACSF, p = 0.70, n = 7/14; baclofen vs ACSF, p = 0.0001) and dendrites (p = 0.0001; ACSF, p = 0.06; baclofen vs ACSF, p = 0.0002) (Fig. 2C,D). These results contrast with CA1 pyramidal neurons and indicate that GABABR modulation of [Ca]bAP is not restricted to apical compartments.
The distal basal dendrites have different properties of synaptic plasticity and integration compared with proximal regions (Schiller et al., 2000; Gordon et al., 2006; Major et al., 2008). We found that baclofen also decreased [Ca]bAP in distal basal spines (p = 0.003, n = 8/15; ACSF, p = 0.77, n = 7/14; baclofen vs ACSF, p = 0.02) and dendrites (p = 0.0005; ACSF, p = 0.85; baclofen vs ACSF, p = 0.001) (Fig. 2D; supplemental Fig. S1, available at www.jneurosci.org as supplemental material) (radial distance; median, 130 μm; range, 99–150 μm). Together, these findings show that GABABR activation decreases [Ca]bAP in spines and dendrites across the entire dendritic arbor of layer 2/3 pyramidal neurons.
Modulation of excitability
GABABRs often activate inwardly rectifying K channels (Gähwiler and Brown, 1985; Sodickson and Bean, 1996; Lüscher et al., 1997) found in the spines and dendrites of pyramidal neurons (Takigawa and Alzheimer, 1999; Chen and Johnston, 2005; Kulik et al., 2006). Opening these channels can generate a shunting conductance that changes passive membrane properties. We found that bath application of 5 μm baclofen decreased the input resistance (Rin) recorded at the soma (p = 1 × 10−11, n = 41 cells; ACSF, p = 0.98, n = 34 cells; baclofen vs ACSF, p = 1 × 10−10) (Fig. 3A,B). Baclofen also decreased the width at 10% amplitude of the AP (width10%), which is a measure of the afterdepolarization (ADP) (p = 9 × 10−12; ACSF, p = 0.03; baclofen vs ACSF, p = 2 × 10−10) (Fig. 3A,B). However, baclofen had only a small effect on AP amplitude, which was not significant compared with ACSF controls (p = 0.03; ACSF, p = 0.90; baclofen vs ACSF, p = 0.06) (Fig. 3B). Moreover, baclofen had no effect on AP half-width (width50%) (p = 0.49; ACSF, p = 0.30; baclofen vs ACSF, p = 0.76) (Fig. 3B). The change in ADP positively correlated with the change in Rin (p < 0.01) (Fig. 3C), suggesting that changes in the AP waveform after GABABR activation reflect changes in passive membrane properties (Fernandez and White, 2009).
By changing the somatic Rin and ADP, GABABRs could influence backpropagation of APs into the dendrites and consequently attenuate [Ca]bAP (Buzsáki et al., 1996; Tsubokawa and Ross, 1996; Leung and Peloquin, 2006). However, we observed no correlation between the change in either spine or dendrite [Ca]bAP and the change in Rin after baclofen (spine, p > 0.05; dendrite, p > 0.05; n = 80 spine–dendrite pairs) (Fig. 3D). Similarly, we observed no correlation between the change in spine or dendrite [Ca]bAP and the change in ADP after baclofen (spine, p > 0.05; dendrite, p > 0.05) (Fig. 3D). These findings suggest that modulation in spines and dendrites is not attributable to changes in either passive membrane properties or AP waveform.
Local modulation
The lack of correlation between changes in [Ca]bAP and either Rin or ADP suggests that GABABR modulation may occur locally in spines and dendrites. We next tested this possibility by puffing 15 μm baclofen from a pipette placed adjacent (Δx = 5–10 μm) to either apical tuft or distal basal dendrites (see Materials and Methods) (Fig. 4A). We found that puffing baclofen continued to decrease [Ca]bAP in apical tuft spines (p = 0.004, n = 5/10; ACSF, p = 0.57, n = 6/11; baclofen vs ACSF, p = 0.0002) and dendrites (p = 0.008; ACSF, p = 0.04; baclofen vs ACSF, p = 0.008) (Fig. 4B,C) (radial distance; median, 153 μm; range, 95–200 μm). Similar results were found in distal basal spines (p = 0.0002, n = 10/15; ACSF, p = 0.53, n = 7/16; baclofen vs ACSF, p = 0.04) and dendrites (p = 0.0002; ACSF, p = 0.98; baclofen vs ACSF, p = 0.0001) (Fig. 4B,C) (radial distance; median, 113 μm, range, 90–142 μm). These results indicate that GABABR modulation occurs locally in distal spines and dendrites and is unlikely attributable to failure of AP backpropagation.
In these puffing experiments, we found that local application of baclofen also decreased Rin and ADP recorded at the soma (Fig. 4D,E; supplemental Fig. S2, available at www.jneurosci.org as supplemental material). These findings are consistent with the dendritic localization of inwardly rectifying K channels (Takigawa and Alzheimer, 1999; Chen and Johnston, 2005; Kulik et al., 2006). Interestingly, the changes in Rin were larger for the apical dendrites (apical, n = 5 cells/5 dendrites; basal, n = 10 cells/10 dendrites; apical vs basal, p = 0.004) (Fig. 4D; supplemental Fig. S2, available at www.jneurosci.org as supplemental material). A similar difference was found for changes in ADP (apical vs basal, p = 0.04) (Fig. 4E; supplemental Fig. S2, available at www.jneurosci.org as supplemental material). These findings suggest that the apical dendrites may have more GABABRs, more K channels, greater coupling between the two, or greater electrical access to the soma. However, there was again no correlation between the change in [Ca]bAP and the change in either Rin (spine, p > 0.05; dendrite, p > 0.05; n = 25 spine–dendrite pairs) (Fig. 4D) or ADP (spine, p > 0.05; dendrite, p > 0.05; n = 25 spine–dendrite pairs) (Fig. 4E). Together, these results suggest a minor role for dendritic GABABR-activated K channels in local inhibition of [Ca]bAP.
GABA uncaging
In our previous experiments, we examined GABABR modulation of [Ca]bAP in response to tonic application of baclofen. We next used GABA uncaging to determine whether GABABR modulation occurs in response to a brief exposure of the endogenous neurotransmitter. The compound RuBi-GABA is excited at blue wavelengths and has previously been used to study GABAAR activation (Rial Verde et al., 2008). When fast neurotransmission was blocked, uncaging of 10 μm RuBi-GABA with a spot of 473 nm light centered at the soma produced a sIPSP that peaked at 146 ± 13 ms after the stimulus (n = 5 cells) (Fig. 5A) (see Materials and Methods). This sIPSP decreased in amplitude with a length constant of 34 μm as the uncaging spot was moved laterally away from the soma (n = 4 cells) (Fig. 5B) and was blocked by wash-in of 2 μm CGP-55845, a selective GABABR antagonist (p = 0.03; n = 5 cells) (Fig. 5C). In control experiments, we confirmed that bath application of baclofen continued to inhibit [Ca]bAP in spines and dendrites (supplemental Fig. S3, available at www.jneurosci.org as supplemental material), indicating that RuBi-GABA does not interfere with GABABR modulation. These findings show that GABA uncaging is a useful approach for studying transient GABABR modulation in spines and dendrites.
To examine uncaging-evoked GABABR modulation, we centered the laser spot on the imaged spines and dendrites, which we briefly illuminated for 10 ms, and evoked an action potential after a 200 ms delay (see Materials and Methods). We found that GABA uncaging reduced [Ca]bAP in apical oblique spines (p = 0.003; n = 9/16) and dendrites (p = 3 × 10−5) (Fig. 6A,B). This effect was prevented by 2 μm CGP-55845, confirming that it was GABABR mediated (spine, p = 0.11, n = 7/13; ACSF vs CGP, p = 0.002; dendrite, p = 0.31; ACSF vs CGP, p = 4 × 10−5). Similar results were found in basal spines (p = 0.03, n = 8/12; CGP, p = 0.16, n = 4/14; ACSF vs CGP, p = 0.01) and dendrites (p = 0.001; CGP, p = 0.22; ACSF vs CGP, p = 0.01) (Fig. 6A,B). These results demonstrate that GABABR inhibition of [Ca]bAP can occur in response to a short exposure of GABA at spines and dendrites.
Similar to the changes in Rin and ADP found in our puffing experiments, we observed larger sIPSPs when uncaging in apical versus basal dendrites (apical, n = 9 cells, 16 dendrites; basal, n = 8 cells, 12 dendrites; apical vs basal, p = 0.02) (Fig. 6C; supplemental Fig. S3, available at www.jneurosci.org as supplemental material). These results support the idea that apical GABABRs have a greater impact on somatic membrane potential. However, we also observed no correlation between the change in [Ca]bAP and the amplitude of the sIPSP measured at the soma (spine, p > 0.05; dendrite, p > 0.05; n = 28 spine–dendrite pairs) (Fig. 6D). Together, these results further suggest that GABABR modulation of K channels plays a limited role in the inhibition of [Ca]bAP.
Direct modulation
The lack of correlation between K channel activation and changes in [Ca]bAP suggest that GABABR activation directly modulates VSCCs in spines and dendrites. We next performed voltage-clamp recordings to test this idea, using internal cesium (Cs) to block K channels and external TTX to block Na channels (see Materials and Methods) (Sabatini et al., 2002). Ca signals were elicited by brief steps from −70 to +10 mV and measured in apical oblique and proximal basal dendrites (Fig. 7A) (see Materials and Methods). These step-evoked Ca signals ([Ca]step) had similar amplitudes and kinetics to [Ca]bAP recorded in current clamp and allowed us to explore VSCC modulation without confounding effects caused by K channel activation.
In voltage-clamp recordings, we found that wash-in of 5 μm baclofen reduced [Ca]step in apical oblique spines (p = 0.004, n = 7/9; ACSF, p = 1, n = 5/6; baclofen vs ACSF, p = 0.004) and dendrites (p = 0.002; ACSF, p = 1; baclofen vs ACSF, p = 0.005) (Fig. 7A,B). Baclofen also inhibited [Ca]step in proximal basal spines (p = 0.002, n = 10/10; ACSF, p = 0.7, n = 18/18; baclofen vs ACSF, p = 4 × 10−4) and dendrites (p = 0.002; ACSF, p = 0.06; baclofen vs ACSF, p = 0.005) (Fig. 7A,B). Internal Cs blocked most of the change in Rin after baclofen (p = 1 × 10−4, n = 17 cells; ACSF, p = 1, n = 19 cells; baclofen vs ACSF, p = 0.02), and the remaining current was inward, indicating that K channel activation is prevented (Torrecilla et al., 2002) (Fig. 7C). Similar results were found in both 3 mm external Cs to block any residual Kir channels (Sodickson and Bean, 1996) and 2 mm external barium (Ba) to block leak K channels (K2P) (Deng et al., 2009) (supplemental Fig. S4, available at www.jneurosci.org as supplemental material). These findings suggest that GABABRs may directly inhibit VSCCs to suppress Ca signals in spines and dendrites, without involving the activation of dendritic K channels.
Voltage-sensitive Ca channels
We next examined which VSCCs are present in spines and dendrites, focusing on the basal compartment, which has not previously been examined at this level. We found that wash-in of the VSCC mixture greatly reduced [Ca]step in both spines and dendrites (spine, p = 0.03; mixture vs ACSF, p = 0.0004; dendrite, p = 0.03; mixture vs ACSF, p = 0.0004; n = 12/16) (Fig. 8A,B). We attribute the remaining Ca signal to SNX-insensitive R-type channels (Newcomb et al., 1998; Tottene et al., 2000; Sochivko et al., 2002, 2003). Wash-in of 10 μm mibefradil, a T-type VSCC blocker, greatly reduced [Ca]step in both spines and dendrites (spine, p = 0.004; mibefradil vs ACSF, p = 0.002; dendrite, p = 0.002; mibefradil vs ACSF, p = 0.0002; n = 9/10) (Fig. 8B; supplemental Fig. S5, available at www.jneurosci.org as supplemental material). Wash-in of 20 μm nimodipine, an L-type VSCC blocker, also decreased [Ca]step in both spines and dendrites (spine, p = 0.02; nimodipine vs ACSF, p = 0.046; dendrite, p = 0.008; nimodipine vs ACSF, p = 0.0004; n = 8/8) (Fig. 8B; supplemental Fig. S5, available at www.jneurosci.org as supplemental material). Wash-in of 0.3 μm SNX, an R-type VSCC blocker, again reduced [Ca]step in both spines and dendrites (spine, p = 0.02; SNX vs ACSF, p = 0.038; dendrite, p = 0.008; SNX vs ACSF, p = 0.02; n = 9/9) (Fig. 8B; supplemental Fig. S5, available at www.jneurosci.org as supplemental material). We found that mibefradil did not block inhibition by either nimodipine or SNX, indicating that it is selective for T-type VSCCs (supplemental Fig. S5, available at www.jneurosci.org as supplemental material). Finally, wash-in of 1 μm ω-conotoxin-MVIIC, a P/Q/N-type VSCC blocker, decreased [Ca]step in dendrites but not spines (spine, p = 0.98; MVIIC vs ACSF, p = 0.73; dendrite, p = 6 × 10−5; MVIIC vs ACSF, p = 0.002; n = 15/15) (Fig. 8B; supplemental Fig. S5, available at www.jneurosci.org as supplemental material). Thus, a combination of L-, R-, and T-type channels contribute to [Ca]step in both spines and dendrites, whereas N/P/Q-type channels contribute only in the dendrites. These results are similar to other neurons in which a wide variety of VSCCs contribute to Ca signals in spines and dendrites (Carter and Sabatini, 2004; Bloodgood and Sabatini, 2007).
Modulation of VSCC subtypes
We last sought to determine which VSCCs are modulated by GABABR activation in proximal spines and dendrites. Previous results using recordings at the cell body indicate that GABABRs can couple to a variety of VSCCs (Holz et al., 1986; Scholz and Miller, 1991; Mintz and Bean, 1993; Pfrieger et al., 1994; Guyon and Leresche, 1995; Lambert and Wilson, 1996). However, studies in CA1 pyramidal neurons indicate that these receptors primarily target R-type VSCCs in apical spines (Sabatini and Svoboda, 2000). We found that bath application of 5 μm baclofen continued to cause inhibition of spines and dendrites in the VSCC mixture, suggesting that SNX-insensitive R-type VSCCs are indeed modulated (spine, p = 0.02; mixture vs ACSF, p = 0.003; dendrite, p = 0.02; mixture vs ACSF, p = 0.0001; n = 6/6) (Fig. 8C,D; supplemental Fig. S6, available at www.jneurosci.org as supplemental material). However, because these channels contribute only a small fraction of the total Ca signal, these results suggest that other channels must also be targeted.
To determine which other VSCCs were affected by GABABRs, we used bath application of 5 μm baclofen in the presence of the different blockers. For comparison, we calculated the range of inhibition expected if a given VSCC was unaffected (0–10% change) or completely blocked (90–100% change) by GABABRs (see Materials and Methods) (Eq. 1). We found that inhibition was significantly enhanced in 10 μm mibefradil compared with ACSF controls (spine, p = 0.002; mibefradil vs ACSF, p = 0.004; dendrite, p = 0.002; mibefradil vs ACSF, p = 0.01; n = 8/10) (Fig. 8D; supplemental Fig. S6, available at www.jneurosci.org as supplemental material), suggesting that T-type VSCCs are minimally modulated. In contrast, inhibition remained unchanged in either 20 μm nimodipine (spine, p = 0.002; nimodipine vs ACSF, p = 0.22; dendrite, p = 0.002; nimodipine vs ACSF, p = 0.48, n = 10/10) (Fig. 8D; supplemental Fig. S6, available at www.jneurosci.org as supplemental material) or 0.3 μm SNX (spine, p = 0.004; SNX vs ACSF, p = 0.18; dendrite, p = 0.004; SNX vs ACSF, p = 0.78, n = 9/9) (Fig. 8D; supplemental Fig. S6, available at www.jneurosci.org as supplemental material), suggesting partial modulation of L- and R-type VSCCs. Finally, inhibition was slightly decreased in 1 μm ω-conotoxin-MVIIC (spine, p = 0.004; MVIIC vs ACSF, p = 0.60; dendrite, p = 0.004; MVIIC vs ACSF, p = 0.31, n = 9/9) (Fig. 8D; supplemental Fig. S6, available at www.jneurosci.org as supplemental material), suggesting that P/Q/N-type VSCCs are fully modulated. Together, these results indicate that GABABR modulation is more complex than previously thought and targets multiple subtypes of VSCCs in spines and dendrites of layer 2/3 pyramidal neurons.
Discussion
We have shown that GABABR modulation of AP Ca signals is widespread throughout the dendritic arbor of layer 2/3 pyramidal neurons in the mouse PFC. Although GABABRs activate K channels to increase membrane conductance, change the AP waveform, and generate sIPSPs, we found no correlation between these effects and inhibition of Ca signals. Instead, by using voltage-clamp recordings that block K channels, we found that GABABRs directly inhibit a variety of VSCC subtypes in both spines and dendrites. Our results have important implications for how GABABRs shape subcellular Ca signals to influence Ca-dependent neuronal function in the PFC and throughout the brain.
Modulation of AP Ca signals is widespread
Modulation via GABABRs has been primarily studied using electrophysiological recordings at the cell body. However, anatomical studies indicate that GABABRs are abundant in the spines and dendrites of pyramidal neurons (Kulik et al., 2003, 2006). In the apical tuft, dendritic GABABRs potently inhibit Ca spikes mediated by VSCCs (Pérez-Garci et al., 2006). In basal spines, GABABRs inhibit postsynaptic NMDA receptor Ca signals (Chalifoux and Carter, 2010). Here, we found that GABABRs also inhibit AP Ca signals in spines and dendrites across the entire dendritic arbor. Our results contrast with CA1 pyramidal neurons, where GABABRs only inhibit these Ca signals in apical spines (Sabatini and Svoboda, 2000). The widespread modulation of AP Ca signals in the PFC may reflect the fundamental importance of GABABRs in modulating synaptic and dendritic function in these pyramidal neurons.
When using bath application of baclofen, it is difficult to determine whether somatic or dendritic GABABRs are responsible for modulating AP Ca signals. For example, any change in the membrane conductance or AP waveform could limit backpropagation and reduce Ca signals in spines and dendrites. We used two complementary approaches to show that GABABR modulation occurs locally in spines and dendrites. First, puffing baclofen tonically activated GABABRs and revealed that inhibition occurs at distal locations. Second, uncaging of RuBi-GABA rapidly activated GABABRs and confirmed that inhibition also occurs at proximal locations. In both cases, we observed prominent modulation of AP Ca signals in both spines and their parent dendrites. Our findings indicate that modulation of AP Ca signals occurs locally and is attributable to activation of GABABRs found throughout the dendritic arbor.
K channel activation does not explain inhibition
GABABRs can influence excitability by activating a variety of K channels (Gähwiler and Brown, 1985; Sodickson and Bean, 1996; Lüscher et al., 1997; Deng et al., 2009). We found that bath application of baclofen decreased Rin and shortened the ADP measured at the soma. Similar effects were seen with puffing of baclofen, suggesting they reflect the activation of dendritic GABABRs. These effects of K channel activation could reduce backpropagation and indirectly decrease Ca signals in spines and dendrites (Buzsáki et al., 1996; Tsubokawa and Ross, 1996; Leung and Peloquin, 2006). However, we observed no significant correlations between changes in AP Ca signals and changes in either Rin or ADP. Moreover, apical puffing caused larger changes in Rin and ADP compared with basal puffing, but similar changes in AP Ca signals were found in both compartments. These results suggest that changes in membrane conductance and AP waveform caused by K channel activation cannot explain the inhibition of AP Ca signals.
By activating K channels, GABABRs also generate sIPSPs (Newberry and Nicoll, 1984; Andrade et al., 1986; Dutar and Nicoll, 1988). In our bath application experiments, we corrected for a small and tonic hyperpolarization by injecting an inward current at the soma. However, we continued to observe pronounced inhibition of AP Ca signals when cells were allowed to hyperpolarize (supplemental Fig. S1, available at www.jneurosci.org as supplemental material). In our uncaging experiments, we observed a GABABR sIPSP that peaked at ∼150 ms. Given the spatial resolution of our uncaging, this sIPSP is likely attributable to activation of dendritic GABABRs. However, we observed no significant correlation between the change in AP Ca signals and the amplitude of this sIPSP. Moreover, basal uncaging generated smaller sIPSPs compared with apical uncaging, but similar changes in AP Ca signals were found in both compartments. Thus, although a local hyperpolarization occurs during the sIPSP, it may not be sufficient to influence the activation of voltage-sensitive Ca channels in spines and dendrites. In summary, although our results show that dendritic GABABRs activate K channels to decrease Rin, shorten the ADP, and cause sIPSPs, these effects are not responsible for inhibition of AP Ca signals. Instead, we conclude that GABABRs must directly modulate VSCCs to locally inhibit these Ca signals in spines and dendrites.
GABABRs modulate multiple VSCC subtypes in spines and dendrites
GABABRs can potentially inhibit a variety of VSCCs to influence neuronal function (Holz et al., 1986; Scholz and Miller, 1991; Mintz and Bean, 1993; Pfrieger et al., 1994; Guyon and Leresche, 1995; Lambert and Wilson, 1996). We examined modulation of VSCCs using voltage-clamp recordings, which allowed us to circumvent indirect influences caused by modulation of K channels (Sabatini et al., 2002). As with our current-clamp experiments, we observed prominent inhibition of step-evoked Ca signals in apical and basal spines and dendrites. We determined that these Ca signals are mediated by a variety of VSCCs, as observed in other neurons in different brain regions (Carter and Sabatini, 2004; Bloodgood and Sabatini, 2007). Modulation persisted in the presence of a mixture to block all VSCCs, consistent with modulation of residual SNX-resistant R-type channels (Sabatini and Svoboda, 2000). However, because these VSCCs contribute a small fraction of the total step-evoked Ca signal, these results also indicate that other VSCCs must also be modulated.
Identifying which VSCCs are modulated by GABABRs is complicated by the presence of multiple potential targets. If GABABRs completely block a single VSCC, including the appropriate channel blocker should eliminate inhibition. However, if GABABRs target many VSCCs, including this blocker should only cause a shift in the amount of inhibition. For example, GABABRs normally inhibit ∼40% of the total VSCCs in spines, and L-type currents make up ∼25% of the total VSCCs in spines. If all of the L-type channels are inhibited by GABABRs, we expect that baclofen should have less effect in the presence of nimodipine (20% = (40 − (1 * 25))/(100 − 25)) (see Materials and Methods) (Eq. 1). In contrast, if none of the L-type channels are inhibited, we expect more modulation [53% = (40 − (0 * 25))/(100 − 25)]. Finally, if only one-half of the L-type channel population is inhibited, we expect similar modulation to controls [37% = (40 − (0.5*25))/(100 − 25)]. Thus, the shift of inhibition also depends on the extent to which a given population of VSCCs is modulated by GABABRs.
With this conceptual framework, our results imply that GABABRs modulate multiple VSCC subtypes in spines and dendrites. For example, less inhibition of dendritic Ca signals occurs in the presence of ω-conotoxin-MVIIC, suggesting that N/P/Q-type VSCCs are fully modulated. However, because these channels contribute only ∼20% of the dendrite Ca signal and none of the spine Ca signal, other VSCCs must also be targeted. More inhibition of spine and dendrite Ca signals occurs in the presence of mibefradil, suggesting that low-threshold T-type VSCCs are minimally modulated. Again, these results suggest that other VSCCs must also be targeted to account for the ∼40% total modulation. Finally, intermediate inhibition of spine and dendrite Ca signals occurs in the presence of nimodipine or SNX, suggesting that a subpopulation of both L- and R-type VSCCs is modulated. Together, our results indicate that GABABR modulation of AP Ca signals is more complex than previously thought, with a single class of receptors able to inhibit multiple VSCC subtypes in both spines and dendrites.
Functional consequences
Extensive GABABR modulation of VSCCs could greatly impact synaptic transmission, integration, and plasticity. For example, GABABRs could attenuate subthreshold EPSPs by inhibiting spine and dendrite VSCCs (Magee and Johnston, 1995a). Alternatively, if these VSCCs participate in local feedback loops with K channels, GABABRs could actually enhance both EPSPs and synaptic Ca signals (Bloodgood and Sabatini, 2007). Our results also suggest that GABABRs could inhibit VSCC-mediated dendritic spikes throughout the dendritic arbor, as previously demonstrated in the apical tufts of layer 2/3 and layer 5 pyramidal neurons (Pérez-Garci et al., 2006; Larkum et al., 2007). Finally, because Ca influx via VSCCs is an important intracellular messenger, GABABRs could help control synaptic plasticity (Grover and Teyler, 1990; Huang and Malenka, 1993; Zucker, 1999). Given the widespread importance of Ca signals mediated by VSCCs, our results have important implications for synaptic and dendritic function in the PFC and throughout the brain.
Footnotes
- Received August 31, 2010.
- Revision received January 5, 2011.
- Accepted January 12, 2011.
This work was supported by National Institutes of Health (NIH) Grant F30MH087409 (J.R.C.) and The Klingenstein Fund and NIH Grant 1R01MH085974 (A.G.C.). We thank members of the Carter Laboratory for helpful discussions and comments on this manuscript.
The authors declare no competing financial interests.
- Correspondence should be addressed to Adam G. Carter, Center for Neural Science, New York University, 4 Washington Place, New York, NY 10003. adam.carter{at}nyu.edu
- Copyright © 2011 the authors 0270-6474/11/314221-12$15.00/0