Muscarinic Long-Term Enhancement of Tonic and Phasic GABAA Inhibition in Rat CA1 Pyramidal Neurons

Acetylcholine (ACh) regulates network operation in the hippocampus by controlling excitation and inhibition in rat CA1 pyramidal neurons (PCs), the latter through gamma-aminobutyric acid type-A receptors (GABAARs). Although, the enhancing effects of ACh on GABAARs have been reported (Dominguez et al., 2014, 2015), its role in regulating tonic GABAA inhibition has not been explored in depth. Therefore, we aimed at determining the effects of the activation of ACh receptors on responses mediated by synaptic and extrasynaptic GABAARs. Here, we show that under blockade of ionotropic glutamate receptors ACh, acting through muscarinic type 1 receptors, paired with post-synaptic depolarization induced a long-term enhancement of tonic GABAA currents (tGABAA) and puff-evoked GABAA currents (pGABAA). ACh combined with depolarization also potentiated IPSCs (i.e., phasic inhibition) in the same PCs, without signs of interactions of synaptic responses with pGABAA and tGABAA, suggesting the contribution of two different GABAA receptor pools. The long-term enhancement of GABAA currents and IPSCs reduced the excitability of PCs, possibly regulating plasticity and learning in behaving animals.


INTRODUCTION
Acetylcholine (ACh) plays a fundamental role in the regulation of network operation in the hippocampus (Watanabe et al., 2006;Connelly et al., 2013;Dominguez et al., 2014Dominguez et al., , 2015. CA1 pyramidal cells (PCs) participate in circuits involved in cognition and spatial navigation, however, the underlying cellular mechanism by which ACh acts on CA1 networks have been insufficiently explored. In PCs ACh can induce a long-term potentiation (LTP) of excitatory synapses through post-synaptic mechanisms (Markram and Segal, 1990;Fernandez de Sevilla et al., 2008;Fernandez de Sevilla and Buno, 2010;Dennis et al., 2015). ACh can control inhibitory synapses in PCs both through presynaptic (Wu and Saggau, 1997;Alger, 2002;Kano et al., 2009) and post-synaptic mechanisms (Kittler and Moss, 2003;Bannai et al., 2009;Castillo et al., 2011;Luscher et al., 2011;Dominguez et al., 2014Dominguez et al., , 2015. In addition, via the control of network activity in the hippocampus ACh can regulate learning and memory (Hasselmo, 2006;Robinson et al., 2011). GABA A Rs elicit the tonic current (tGABA A ), which hyperpolarizes CA1 PCs, reduces network excitability (Semyanov et al., 2003;Semyanov et al., 2004), and regulates information processing (Mitchell and Silver, 2003;Chadderton et al., 2004) and behavior (Belelli et al., 2009;Brickley and Mody, 2012). We have reported that a longterm enhancement of IPSCs induced by ACh combined with depolarization (ACh+depolarization) was paralleled by a potentiation of tonic inhibition in PCs (Dominguez et al., 2015), but the LTP of tonic inhibition remains insufficiently explored. Therefore, the central aim of this work was to determine the effects of the activation of cholinergic receptors and membrane depolarization on responses resulting from the activation of synaptic and extrasynaptic GABA A Rs.
Here, we report that in immature rats, under blockade of glutamatergic ionotropic receptors, a brief pulse of ACh on the apical dendritic shaft while the PC was repeatedly depolarized during the experiment caused a durable increase in tonic GABA A current (tGABA A ), and a LTP of puff-evoked GABA A currents (pGABA A ), that we call pLTPextra. These effects were matched by a LTP of IPSCs that we have termed GABA A -LTP (Dominguez et al., 2014(Dominguez et al., , 2015. The parallel long-term enhancement of tonic and phasic inhibition caused a strong reduction of the excitability of PCs that possibly regulates network operation, plasticity and learning in behaving animals. The enhancement of both tonic and phasic inhibition followed similar time course and rules. However, we could not observe changes in IPSCs following GABA puff, suggesting that: (i) the long-term boost of tonic and phasic inhibition shared key mechanisms; (ii) GABA puffs and GABA released by inhibitory interneurons activated different GABA A R pools; (iii) GABA "spillover" did not play an important role in the effects of ACh+depolarization on synaptic responses.

Ethical Approval and Animal Handling
Procedures of animal care and slice preparation approved by the CSIC followed the guidelines laid down by the European Council on the ethical use of animals (Directive 2010/63/EU) and with every effort made to minimize the suffering and number of animals.

Electrophysiology
Whole-cell voltage-and current-clamp recordings were from the soma of CA1 PCs (Figure 1A), using patch pipettes (4-8 M ) that contained in mM: 140 K-MeSO 4 , 10 HEPES, 10 KCl, 4 Na-ATP, 0.3 Na-GTP and 0.1 EGTA, buffered to pH 7.2-7.3 with KOH. A −65 mV chloride equilibrium potential was calculated with the intra-and extracellular solutions used. Neurons were only accepted if during the experiment the seal resistance was >1 G , the series resistance (7-14 M ) did not change >15%, and the holding current did not exceed 300 pA at −75 mV. The average resting Vm was −70.2 ± 3 mV (N = 127). In some experiments BAPTA (20 mM), a fast Ca 2+ chelator or heparin (5 mg/ml), which inhibits IP 3 Rs were added to the pipette solution.

Stimulation under Voltage-Clamp
Steps from −75 to 0 mV lasting 30 s were applied every 75 s during the experiment (Figure 1B), while GABA (500 µM), diluted in the control ACSF was repeatedly puffed through a fine tipped pipette every 5 or 10 s on the apical dendritic shaft of the patched PC (Figures 1A,B). In most experiments, stimulation at the stratum radiatum (SR) evoked single or pairs (50-100 ms delay) of inhibitory post-synaptic currents or potentials (IPSCs-IPSPs) either in isolation or 2.5 s after GABA puffs. Responses were recorded both at 0 and −75 mV. We analyzed the voltagedependence of pGABA A with I/V relationships that involved 500 ms duration 10 mV voltage control steps from −100 to +20 mV, applied every 10 s, combined with a puff of GABA. In all experiments following a 15-20 min control recording after attaining the whole-cell configuration, a single 100-300 ms pulse of ACh was applied by iontophoresis though a pipette loaded with ACh (1 M) dissolved in distilled water. The ACh pulse was aimed at the SR close to the base of the apical dendrite of the patched PC. To avoid spurious release of ACh the pipette was withdrawn. Stimulation and recording continued >1 h after the ACh pulse. ACh effects were essentially identical when the pulse was applied during brief interruptions of the depolarizing protocols (≈3 min) or during the protocols, and did not depend on the Vm or inhibitory activity (Dominguez et al., 2014(Dominguez et al., , 2015. In some cases voltage steps (as above) were applied in the absence of ACh ( Figure 1C).

Stimulation under Current-Clamp
To determine modifications in the excitability of PCs we estimated the changes in action potential (AP) firing evoked by 1 s duration depolarizing current pulses applied throughout the experiment every 5-10 s at twice the AP threshold intensity ( Figure 7A). In another group of experiments to estimate the effects on both the enhanced tGABA A and IPSPs on PC excitability we applied 500 ms duration current pulses every 10 s at twice the AP threshold intensity. Current pulses were coupled with paired-pulse stimulation (50-100 ms delay) of inhibitory inputs at the SR both during AP bursts and silent periods between bursts ( Figure 7E). In both experimental groups stimulation was transiently interrupted (≈3 min) after a 15-20 min control recording and the ACh pulse was applied ( Figure 7B). Under current-clamp the AP threshold was derived from the current intensity of a 1 s duration depolarizing current pulse just sufficient to bring the cell to AP generation when the PC was at the resting Vm.

Data Analysis
Data were analyzed with the pClamp programs (Molecular Devices, Chicago, IL, USA) and Excel (Microsoft, Redmond, WA, USA). Peak amplitudes of pGABA A s and IPSCs averaged over 5 min epochs were plotted versus time, expressed as a proportion of the baseline amplitude. Analysis of the spontaneous IPSCs ( S IPCSs) activity was performed with the pClamp software. Cumulative probabilities of amplitude and inter-event intervals of S IPCSs recorded during ≈5 min in control conditions and ≈5 min during the pLTPextra ≈40 min after the ACh pulse were computed. Statistically significant differences were established using the Kolmogorov-Smirnov test. Under voltage-clamp, shifts in the mean pre-ACh holding current (Ih) provided a measure of changes in tGABA A (Dominguez et al., 2014). Ih shifts were confirmed by the change in mean steady current after blocking GABA A Rs with P i TX (50 µM). To determine the temporal evolution of the peak amplitude of currents evoked by GABA puffs at all successive steps we averaged puff evoked current during each depolarizing step in six experiments and plotted them versus time, expressed as the proportion (%) of the mean value of puff evoked currents triggered during the first step (Figures 4A,B and see Figure 2 in Dominguez et al., 2015). Statistical analysis was performed using Student's two tail t-test and differences were considered statistically significant at * P < 0.05 level, * * P < 0.01, and * * * P < 0.001. Results are given as mean ± SEM (N = numbers of cells) and (n = number of averaged responses). There were no gender differences in our experiments.

RESULTS
In the Absence of ACh the Depolarization Protocol Did Not Modify GABA A Currents GABA B Rs are absent in rat PCs before postnatal day 22 (Nurse and Lacaille, 1999), but there are functional presynaptic GABA B Rs in the terminals of CA1 inhibitory interneurons in younger rats (Wu and Saggau, 1997;Dominguez et al., 2015). To avoid the possible activation of presynaptic GABA B Rs by the GABA puffed we usually recorded pGABA A , tGABA A , and IPSCs under blockade of GABA B Rs with CGP55845 (2 µM). Isolated pGABA A s had mean peak amplitudes of 575 ± 79 pA at 0 mV and of −294 ± 68 pA at −75 mV (N = 10) in the pre-ACh controls (insets in Figures 1D,E).
Under voltage-clamp a prolonged presentation of the depolarization protocol (>1 h) in the absence of the ACh pulse did not modify puff-evoked isolated GABA A currents recorded both at 0 and −75 mV (P > 0.05; N = 4; Figure 1C), indicating that repeated depolarization alone was unable to induce longterm changes in extrasynaptic GABA A currents. We have shown that the same protocol did not modify IPSCs in the absence of ACh (Dominguez et al., 2014).

ACh+depolarization Induced a Gradual Potentiation of pGABA A
After the ACh pulse (≈5 min) there was a gradual potentiation of pGABA A or pLTPextra, which in ≈40 min stabilized at mean peak values that were 237 ± 3% of the controls at 0 mV and of 251 ± 4% at −75 mV (P < 0.001; N = 10; Figures 1D,E). Therefore, pLTPextra had similar time-course and reached essentially identical values at 0 and −75 mV (P > 0.05 in both cases), suggesting that baseline pGABA A amplitude and the inward or outward Cl − flow did not contribute to the potentiating effects of ACh+depolarization. This was essentially identical to what occurred with IPSCs (Dominguez et al., 2015). An inward current that peaked at ≈30 s and gradually decayed to a steady state in ≈1 min typified the response evoked by the ACh pulse under voltage-clamp at 0 mV (see Figure 1F in Dominguez et al., 2014).

A Potentiation of IPSCs Accompanied pLTPextra
In some experiments, we recorded both IPSCs and pGABA A in the same PCs (see Materials and Methods) under blockade of GABA B Rs with CGP55845 (2 µM). Stimulation at the SR evoked outward IPSCs (258 ± 11 pA; N = 4) at 0 mV and inward IPSCs (−68 ± 8 pA; same cells) at −75 mV in the pre-ACh controls (inset in Figure 1F). Following the ACh pulse there was a gradual enhancement of the IPSC recorded at −75 mV that in ≈40 min reached a steady state mean peak value that was 319 ± 6% of the control or GABA A -LTP (P < 0.01; N = 4; Figure 1F). Therefore, the synaptic GABA A -LTP attained higher steady state values than pLTPextra (127 ± 5%; P < 0.05; N = 4). A similar IPSC enhancement that reached values of 298 ± 10% of the control value (P < 0.01; same sells) was recorded at 0 mV (data not shown, but see Figure 2S in Dominguez et al., 2014).
We also constructed plots of the mean peak pGABA A amplitude versus that of IPSCs to analyze the amplitude relationship between both responses in the same PCs. There was a linear correlation between the mean peak amplitudes of post-ACh pGABA A (abscissa) and IPSCs (ordinates) (slope 1.88; R = 0.97; N = 4; Figure 1G), indicating that IPSC increased more than pGABA A .
Both pLTPextra and the Enhanced tGABA A Were Blocked by P i TX To confirm the central contribution of GABA A Rs we tested the effects of P i TX (50 µM) applied following the ACh pulse when pLTPextra had reached values that were 209 ± 4% of the control (P < 0.001; N = 5). P i TX reduced pGABA A to values that were not significantly different from zero (P > 0.05; same cells; Figure 2A). These results suggest that an increased response of GABA A Rs generates pLTPextra. In addition, the magnitude of pLTPextra recorded at 0 mV in control ACSF 25 min after ACh (Figure 2A) was essentially identical in control solution and under CGP55845 ( Figure 1D; P > 0.05; N = 5 and N = 10, respectively), verifying that GABA B Rs did not contribute to the enhancing effects of ACh. Moreover, the mean amplitude of control pGABA A was essentially identical in control conditions and under CGP55845 (P < 0.05; same cells). The difference between the average control pre-ACh mean current and the average Ih associated with the pLTPextra provides a measure of the tonic GABA current (see Materials and Methods and Dominguez et al., 2014). Therefore, we tested for changes in tGABA A induced by the ACh+depolarization protocol. The mean Ih had negative values of −78 ± 3 pA at −75 mV in control conditions and changed to −102 ± 7 pA with the IPSC potentiation, indicating a negative Ih shift that was 136% of the control tGABA A values (P < 0.01; N = 5; Figure 2A, bottom traces and Figures 2B,C).
The mean peak amplitude of the spontaneous IPSCs ( S IPSCs) increased from 22 ± 8 mV to 37 ± 3 pA following ACh, indicating an increase that was 168% of the control (P < 0.05;  Figures 2B-D,G). In contrast, the mean S IPSCs frequency did not change after the ACh pulse (P > 0.05; Figure 2A, bottom traces and Figures  2B,C,E,G). These effects agree with the post-synaptic nature of the effects of ACh+depolarization. P i TX inhibited tGABA A and the S IPSCs activity, implying that GABA A Rs mediated both tonic and synaptic currents (Figure 2A, bottom traces).

There Were No Changes in IPSCs Following GABA Puffs
The parallel increase in pGABA A , tGABA A , and IPSCs could suggest that an increased "ambient" GABA resulting from "spillover" and the puffed GABA caused the enhancement of currents and IPSPs. An increased number of GABA A Rs could take place in synaptic and also possibly in extrasynaptic sites, thus contributing to the result of ACh+depolarization. Therefore, we performed experiments in the same PCs in which IPSCs were evoked both in isolation and following GABA-puffs at delays of 2.5 s and in control and post-ACh (≈40 min) conditions (Figure 3). Both paired and isolated synaptic stimulation was at 0.1 s. In control pre-ACh conditions IPSC amplitudes were essentially identical when preceded or not by GABA puffs (Figures 3A,B). In addition, there were no statistically significant differences between both groups when data from different experiments was pooled (168.6 ± 17 pA pre-ACh and 173.7 ± 22 pA post-ACh, respectively, P > 0.05; N = 6; Figure 3C). In post-ACh conditions (≈40 min) IPSC amplitudes were larger but were not modified by the GABA puffs (Figures 3D,E), and there were no statistically significant differences between both groups when data from different experiments was pooled (336.7 ± 30 pA pre-and 353.4 ± 21 pA post-ACh, respectively; P > 0.05; N = 6; Figure 3F).
The absence of detectable interactions between pGABA A and IPSCs, suggests that two different receptor pools (i.e., extrasynaptic and synaptic) were activated by puffed and released GABA. These results could also suggest that the long-term enhancement of tonic and phasic inhibition shared key mechanisms and that GABA "spillover" did not play an dominant role in the effects of ACh+depolarization on IPSCs.

Following the ACh Pulse pGABA A Rapidly Increased during Depolarizing Steps
We analyzed the temporal evolution of the peak amplitude of pGABA A at successive 0 mV steps (see Materials and Methods). Control pre-ACh pGABA A s did not change during steps and were not enhanced by successive current steps. In contrast, following the ACh pulse there was a rapid enhancement of pGABA A during 0 mV steps that gradually increased in successive steps leading to a pLTPextra (Figures 4A,B). Therefore, the potentiation process involved the rapid buildup with repeated depolarization of the machinery that gradually developed to finally stabilize with the potentiation.

Endocannabinoids Did Not Contribute to pLTPextra
The activity of extrasynaptic GABA A Rs can also be enhanced by cannabinoids in a CB 1 R-independent manner (Golovko et al., 2015). Moreover, a robust hyperpolarization mediated by an increased K + conductance, which can be blocked by the type 1 endocannabinoid receptor (CB 1 R) antagonist AM-251, has also been shown (Bacci et al., 2004). Therefore, we tested if pLTPextra was modified by blockade of CB 1 R with AM-251 (2 µM). In these conditions pLTPextra was essentially identical to that induced in control ACSF both at 0 and −75 mV (compare Figures 4C,D  with Figures 1D,E), suggesting that endocannabinoids were not contributing to the effects of ACh+depolarization in our experimental conditions.

The pGABA A Decay Time Increased during the pLTPextra
We have shown that a increased decay time of IPSCs paralleled the synaptic GABA A -LTP (Dominguez et al., 2014), accordingly a increased decay time of pGABA A could also accompany pLTPextra. The decay of pGABA A was well-fitted by a single exponential. The decay time (tau) of pGABA A gradually changed from the pre-ACh 84 ± 10 s to reach steady state values of 184 ± 9 s or a 219 ± 10% increase ≈40 min after ACh (P < 0.001; N = 10; Figure 5A). We plotted the peak pGABA A amplitude (taken from Figure 2A) versus pGABA A tau, expressed as a proportion of the control pre-ACh pGABA A tau. The plot revealed a linear correlation (R = 0.98; N = 10) between the mean peak amplitudes and tau of post-ACh pGABA A ( Figure 5B). Therefore, pLTPextra involved a gradual increase in the contribution of extrasynaptic GABA A Rs with a slower rate of desensitization than naïve receptors. Note also the outward shift in holding current following the ACh pulse ( Figure 5A, insets 1 and 2).

An Increased Contribution of Voltage-Sensitive GABA A Rs with Boosted GABA Sensitivity Underlies pLTPextra, Effects that Required a Cytosolic Ca 2+ Rise
We have shown that an increased slope conductance and strong outward rectification of IPSCs typified the synaptic GABA A -LTP (Dominguez et al., 2014). Since GABA A -LTP and pLTPextra share important properties, pGABA A could show an increased GABA-and a voltage-sensitivity. Therefore, we calculated I/V relationships of pGABA A , which revealed that the control pre-ACh I/V relationship was linear with a small average slope ( Figure 5C). In contrast, ≈40 min post-ACh the I/V plot showed an increased slope conductance and a strong outward rectification of pGABA A > −40 mV ( Figure 5C). Importantly, the ACh challenge did not cause changes in the reversal potential of pGABA A . We next tested BAPTA-loading (20 mM in the pipette solution), which blocked the increase in voltage-and GABA-sensitivity of pGABA A induced by ACh ( Figure 5D). The above results taken together suggest that a Ca 2+ -induced increase in the contribution of slow desensitizing voltage-sensitive extrasynaptic GABA A Rs with boosted GABA affinity caused pLTPextra as well as the synaptic GABA A -LTP (see Figure 6 in Dominguez et al., 2015).

M1-mAChRs and Ca 2+ Are Required to Induce the Potentiation
In CA1 pyramidal neurons depolarization coupled with M1-mAChR activation can induce a robust cytosolic Ca 2+ signal, which can regulate inhibition through pre-and postsynaptic mechanisms (Dominguez et al., 2014(Dominguez et al., , 2015. The ACh+depolarization protocol can increase intracellular Ca 2+ both through Ca 2+ release from IP 3 -sensitive intracellular stores and influx across L-type voltage-gated calcium channels (VGCCs) (Watanabe et al., 2006;Fernandez de Sevilla et al., 2008;Fernandez de Sevilla and Buno, 2010). Accordingly, we tested the effects of inhibiting Ca 2+ release from IP 3 -sensitive stores by loading the PC with heparin (5 mg/ml in the pipette solution). Inhibition of IP 3 Rs prevented pLTPextra and post-ACh pGABA A amplitudes reached values that were 118 ± 10% of the control (P > 0.05; N = 4; Figure 6A, HEPA). We also tested the effects of blocking L-type VGCC with nimodipine. Nimodipine (10 µM) inhibited pLTPextra, stabilizing post-ACh pGABA A amplitudes at values that were 97 ± 6% of the control (P > 0.001; N = 6; Figure 6A, NIMO). Finally, we examined the effects of BAPTA-loading to inhibit the cytosolic Ca 2+ rise. BAPTA-loading (20 mM in the pipette solution) blocked pLTPextra and pGABA A amplitudes reached values that were 89 ± 5% of the control (P > 0.001; N = 5; Figures 6A,B, BAPTA). We also plotted the temporal evolution of pGABA A amplitudes under BAPTA-loading ( Figure 6B).

The Enhanced tGABA A and IPSPs Reduced the Excitability of PCs
The tonic GABA current can play key roles in regulating network excitability (Bai et al., 2001;Semyanov et al., 2004), information processing (Mitchell and Silver, 2003;Chadderton et al., 2004) and behavior (Pavlov et al., 2009;Houston et al., 2012). Therefore, we performed current-clamp experiments to determine modifications in excitability of PCs induced by the enhanced tGABA A and IPSPs. We first depolarized PCs with 1 s duration current pulses applied every 10 s at twice the AP threshold during ≈10 min that triggered repetitive AP firing ( Figure 7A). We interrupted the stimulation (≈3 min) and applied the ACh pulse that transiently depolarized the PC and evoked repetitive spiking ( Figure 7B). Current pulse stimulation was resumed and the firing rate gradually decreased to stabilize ≈40 min later (Figure 7C), suggesting a gradual decrease in the excitability of the PCs. The ACh pulse induced a mean decay in firing rate from the control 35 ± 8 APs −1 to stabilize at 16 ± 9 APs −1 ≈40 min later (or a 48% decrease from the control; P < 0.05, N = 6; Figure 7D).
We next investigated the effects of both tGABA A and IPSCs on AP responses evoked by depolarizing current pulses. We depolarized PCs with 500 ms duration current pulses applied every 5-10 s at twice the AP threshold and simultaneously stimulated SCs to evoke pairs of IPSPs (see Materials and Methods; Figures 7E,F). We transiently interrupted the stimulation (≈3 min) and applied the ACh pulse that briefly depolarized the PC and evoked repetitive spiking (as above). Following the ACh pulse (≈40 min) there was a reduction in spike rate during depolarization from 45 ± 6 APs −1 to 18 ± 5 APs −1 (or a 39% decrease from control values; P < 0.01; N = 4; see Materials and Methods). The ACh pulse also increased IPSP amplitude (from 22 ± 11 to 48 ± 8 mV; P < 0.01; N = 4) or 211% of the control and delayed post-IPSP spikes from 22 ± 5 to 35 ± 10 ms (P < 0.01; N = 6; Figures 7E-G) or a 156% increase from control values.
An interpretation of the above results is that ACh+depolarization reduced the excitability of PCs through both an increased tGABA A and IPSPs. The resting membrane potential (green interrupted lines in Figure 7) hyperpolarized by 8 ± 3 mV and the AP after-hyperpolarization (AHP) (red interrupted lines in Figure 7) increased 15 ± 8 mV during ≈40 min following the ACh pulse (P < 0.05; N = 10; Figure 7). The difference between the resting potential and the AHP provide a rough estimate of the depolarization attained during the current pulses, suggesting that more depolarization was required to reach AP threshold after the ACh+depolarization protocol.
Phasic inhibition follows from activation of low affinity synaptic GABA A Rs by brief release of high concentrations of GABA by exocytosis of presynaptic vesicles into the synaptic cleft (Farrant and Nusser, 2005). GABA A Rs mediating the two inhibitory modalities normally exhibit differences in subunit composition, GABA affinity and subcellular localization. However, ACh+depolarization induced a profound transformation that ended up with GABA A Rs displaying similar properties in extra-and synaptic compartments (Dominguez et al., 2014, 2015 andsee Results). Accordingly, our results could suggest that the same intracellular mechanisms operate to increase the number of GABA A R of the same subtypes at synaptic and also possibly extrasynaptic sites. Importantly, GABA puffs increased ambient GABA, but did not modify IPSCs in control and potentiated conditions (Figure 3). In the controls there is a substantial difference in the GABA affinity of extra-and synaptic receptors (several orders of magnitude; Farrant and Nusser, 2005;Patel et al., 2016). Indeed, GABA increases to millimolar concentrations at the synaptic cleft to activate post-synaptic GABA A Rs, but only nanomolar concentrations are sufficient to activate extrasynaptic receptors during tonic inhibition (Santhakumar et al., 2006;Patel et al., 2016). Consequently, ambient GABA would readily activate extrasynaptic but not synaptic GABA A Rs because high ambient GABA concentrations, not usually attained in vitro, would be required to activate synaptic GABA A Rs.
In contrast, potentiated pGABA A and IPSCs display similar GABA affinity, outward rectification and decay kinetics (see Results andDominguez et al., 2014, 2015), suggesting the presence of GABA A Rs with essentially identical biophysical properties and possibly similar subunit composition in both extra-and synaptic sites. However, although the increased ambient GABA activated extrasynaptic GABA A Rs, there was no detectable effect of GABA puffs on IPSCs. These results suggest that even with a significant increase in ambient GABA, the transmitter did not influence synaptic receptors during IPSCs in our experimental conditions.
The above results suggest that although GABA can flow in and out of the synaptic cleft the effects of outward GABA flow are clear-cut but those of inward flow are absent or unimportant. In contrast, when ambient GABA is significantly enhanced, such as high frequency stimulation of inhibitory inputs, increased interneuron activity, epileptic activity, and abnormal function of the GABA uptake, the massive increase in ambient GABA may modify synaptic responses (Barbour and Hausser, 1997;Farrant and Nusser, 2005;Glykys and Mody, 2007). However, the effects of an abnormally high concentration of GABA in the synaptic cleft can also reduce release through blockade of presynaptic Ca 2+ channels via activation of presynaptic GABA B Rs (Wu and Saggau, 1997). Indeed, increasing ambient GABA by blocking neuronal GABA uptake can induce a strong GABA B R triggered presynaptic inhibition without signs of enhanced post-synaptic GABA A R activity (Dominguez et al., 2015).
Taken together our present and previous results (Dominguez et al., 2014(Dominguez et al., , 2015, suggest that following the ACh pulse both an increased ambient GABA and number of slow desensitizing high-affinity voltage-sensitive GABA A Rs can occur. The increase in GABA A R number is likely to occur through the rapid lateral transit and clustering leading to enhanced responses (Kittler and Moss, 2003;Semyanov et al., 2004;Bannai et al., 2009;Pavlov et al., 2009;Ransom et al., 2010;Luscher et al., 2011;Brickley and Mody, 2012;Ransom et al., 2013;Dominguez et al., 2014Dominguez et al., , 2015. Interestingly, the dynamic lateral mobility of GABA A Rs can be enhanced by neuronal hyperactivity and operate in the 10s-of-milliseconds time range (Bannai et al., 2009; Dominguez et al., 2015), thus providing an exceptionally rapid negative feedback through the control of GABA A R number (Gaiarsa et al., 2002;Petrini et al., 2004;Luscher et al., 2011).
The ACh+depolarization protocol can trigger vigorous Ca 2+ signals because the M1-mAChR-mediated blockade of K + conductance raises the membrane resistance making the PC electrically compact (Benardo and Prince, 1982), boosting the depolarization-induced Ca 2+ influx through L-type VGCC. In addition, activation of M1-mAChRs can induce Ca 2+ release from IP 3 -sensitive stores (Watanabe et al., 2006;Fernandez de Sevilla and Buno, 2010). The strong cytosolic Ca 2+ signal can trigger a rapid increase in the number of GABA A Rs at the membrane, which is critically dependent on Ca 2+ influx through L-type VGCCs (Saliba et al., 2012).
It has been shown that the activity of extrasynaptic GABA A Rs can also be enhanced by cannabinoids in a CB 1 R-independent manner in neocortical pyramidal neurons (Golovko et al., 2015). Moreover, a prolonged CB 1 R-dependenet hyperpolarization mediated by an increased K + conductance has been demonstrated in neocortical inhibitory interneurons (Bacci et al., 2004). These unconventional effects mediated by the release of endogenous cannabinoids, which could regulate synaptic strength and excitability, were not functional in our experimental conditions. Changes in the Cl − concentration gradient caused by Cl − flux through activated GABA A Rs may globally modify GABA A -mediated activity (Woodin et al., 2003;Raimondo et al., 2012). However, the GABA A -LTP, which was induced in essentially identical experimental conditions and shares key mechanisms with the pLTPextra and the increase tGABA A , was unaffected by the Cl − driving force, the Cl − concentration gradient and K + conductance block (Dominguez et al., 2014). In addition, the reversal potential of pGABA A did not change in the present conditions, suggesting that the effects of ACh+depolarization do not involve changes in the Cl − concentration gradient.
We show that pGABA A inhibition displays both an increased slope conductance and a strong outward rectification (Dominguez et al., 2014) and thus exerts a stronger inhibition on excitatory inputs that depolarize the PC close to AP threshold, while it barely affects subthreshold inputs (Pavlov et al., 2009). Moreover, both the slope conductance and the rectification increase in function of time and the degree of PC activation (Dominguez et al., 2014), suggesting an homeostatic feedback role in the control of excitability (Mody, 2005). These effects could have a strong influence on network operation by maintaining the activity of the network within functional limits and could be a target for the treatment of hyperexcitable states.

AUTHOR CONTRIBUTIONS
SD, DF, and WB implemented conception and design of research; SD performed experiments; SD, DF, and WB analyzed data and interpreted results; SD, DF, and WB prepared figures; DF and WB wrote manuscript; SD, DF, and WB edited, revised, and approved final version of manuscript.