GABA_B receptor deficiency causes failure of neuronal homeostasis in hippocampal networks

Significance

How neuronal circuits maintain stable activity despite continuous environmental changes is one of the most intriguing questions in neuroscience. Previous studies proposed that deficits in homeostatic control systems may underlie common neurological symptoms in a variety of brain disorders. However, the key regulatory molecules that control homeostasis of central neural circuits remain obscure. We show here that basal activity of GABA_B receptors is required for firing rate homeostasis in hippocampal networks. We identified the principal mechanisms by which GABA_B receptors control homeostatic augmentation of synaptic strength to chronic neuronal silencing. We propose that deficits in GABA_B receptor signaling, associated with epilepsy and psychiatric disorders, may lead to aberrant brain activity by erasing homeostatic plasticity.

Abstract

Stabilization of neuronal activity by homeostatic control systems is fundamental for proper functioning of neural circuits. Failure in neuronal homeostasis has been hypothesized to underlie common pathophysiological mechanisms in a variety of brain disorders. However, the key molecules regulating homeostasis in central mammalian neural circuits remain obscure. Here, we show that selective inactivation of GABA_B, but not GABA_A, receptors impairs firing rate homeostasis by disrupting synaptic homeostatic plasticity in hippocampal networks. Pharmacological GABA_B receptor (GABA_BR) blockade or genetic deletion of the GB_1a receptor subunit disrupts homeostatic regulation of synaptic vesicle release. GABA_BRs mediate adaptive presynaptic enhancement to neuronal inactivity by two principle mechanisms: First, neuronal silencing promotes syntaxin-1 switch from a closed to an open conformation to accelerate soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complex assembly, and second, it boosts spike-evoked presynaptic calcium flux. In both cases, neuronal inactivity removes tonic block imposed by the presynaptic, GB_1a-containing receptors on syntaxin-1 opening and calcium entry to enhance probability of vesicle fusion. We identified the GB_1a intracellular domain essential for the presynaptic homeostatic response by tuning intermolecular interactions among the receptor, syntaxin-1, and the Ca_V2.2 channel. The presynaptic adaptations were accompanied by scaling of excitatory quantal amplitude via the postsynaptic, GB_1b-containing receptors. Thus, GABA_BRs sense chronic perturbations in GABA levels and transduce it to homeostatic changes in synaptic strength. Our results reveal a novel role for GABA_BR as a key regulator of population firing stability and propose that disruption of homeostatic synaptic plasticity may underlie seizure's persistence in the absence of functional GABA_BRs.

Neural circuits achieve an ongoing balance between plasticity and stability to enable adaptations to constantly changing environments while maintaining neuronal activity within a stable regime. Hebbian-like plasticity, reflected by persistent changes in synaptic and intrinsic properties, is crucial for refinement of neural circuits and information storage; however, alone it is unlikely to account for the stable functioning of neural networks (1). In the last 2 decades, major progress has been made toward understanding the homeostatic negative feedback systems underlying restoration of a baseline neuronal function after prolonged activity perturbations (2–4). Homeostatic processes may counteract the instability by adjusting intrinsic neuronal excitability, inhibition-to-excitation balance, and synaptic strength via postsynaptic or presynaptic modifications (5, 6) through a profound molecular reorganization of synaptic proteins (7, 8). These stabilizing mechanisms have been collectively termed homeostatic plasticity. Homeostatic mechanisms enable invariant firing rates and patterns of neural networks composed from intrinsically unstable activity patterns of individual neurons (9).

However, nervous systems are not always capable of maintaining constant output. Although some mutations, genetic knockouts, or pharmacologic perturbations induce a compensatory response that restores network firing properties around a predefined “set point” (10), the others remain uncompensated, or their compensation leads to pathological function (11). The inability of neural networks to compensate for a perturbation may result in epilepsy and various types of psychiatric disorders (12). Therefore, determining under which conditions activity-dependent regulation fails to compensate for a perturbation and identifying the key regulatory molecules of neuronal homeostasis is critical for understanding the function and malfunction of central neural circuits.

In this work, we explored the mechanisms underlying the failure in stabilizing hippocampal network activity by combining long-term extracellular spike recordings by multielectrode arrays (MEAs), intracellular patch-clamp recordings of synaptic responses, imaging of synaptic vesicle exocytosis, and calcium dynamics, together with FRET-based analysis of intermolecular interactions at individual synapses. Our results demonstrate that metabotropic, G protein-coupled receptors for GABA, GABA_BRs, are essential for firing rate homeostasis in hippocampal networks. We explored the mechanisms by which GABA_BRs gate homeostatic synaptic plasticity. Our study raises the possibility that persistence of epileptic seizures in GABA_BR-deficient mice (13–15) is directly linked to impairments in a homeostatic control system.

Results

GABA_BR Blockade Disrupts Firing Rate Homeostasis.

Mice lacking functional GABA_BRs because of GB₁ or GB₂ subunit deletion display continuous spontaneous seizure activity (13–16). These findings are quite counterintuitive in light of a wide range of G protein-coupled receptors that mediate synaptic inhibition and might compensate for GABA_BR loss of function. Therefore, we hypothesized that functional GABA_BRs may play an essential role in neuronal homeostasis. To test this hypothesis, we examined homeostasis of mean firing rate of a neuronal population after chronic blockade of GABA_BRs. To chronically monitor neuronal activity in stable neural populations, we grew hippocampal cultures on MEAs for ∼3 wk. Each MEA contains 59 recording electrodes, with each electrode capable of recording the activity of several adjacent neurons (Fig. 1A). Spikes were detected and analyzed using principal component analysis to obtain well-separated single units that were consistent throughout at least 2 d of recording (SI Appendix, Methods). To assess how chronic inhibition of the basal GABA_BR activity affects mean firing rate of neural network, we measured spiking activity during a baseline recording period and for 2 d after application of CGP54626 (CGP), a selective GABA_BR antagonist. Fig. 1B illustrates raster plots during periods of baseline, 1 h, and 48 h after 1 μM CGP application in a single experiment. Indeed, CGP causes an acute increase of 67 ± 13% in the mean firing rate (Fig. 1C), confirming proconvulsive properties of GABA_BR antagonists in vivo (17). However, to our surprise, mean firing rate was not normalized during 2 d in the constant presence of CGP. After 2 d, firing rate remained 67 ± 18% higher in the presence of GABA_BR antagonist (P < 0.01; Fig. 1C). Notably, under control conditions, network spike rates were stable during the 2 d of recording (no treatment, P > 0.2 between 1 h and 48 h; Fig. 1D).

Fig. 1.

GABA_BR blockade disrupts firing rate homeostasis in hippocampal networks. (A, Left) Image of MEA dish. (Middle) Image of dissociated hippocampal culture plated on MEA. Black circles at the end of the black lines are the recording electrodes. (Right) Representative traces of recording from four MEA channels (a, b, c, and d). (B) Representative raster plot of MEA recording before and 1 and 48 h after application of the GABA_BR antagonist CGP (1 μM). (C–F) Mean firing rate of hippocampal neuronal cultures incubated with CGP (n = 4; C), no treatment (Cnt, n = 6; D), gabazine (GBZ, 30 μM, n = 4; E), and CGP+GBZ (n = 4; F) during 2 d of MEA recordings.

To confirm that the lack of firing rate homeostasis is specific to the GABA_BR blockade, we examined how chronic blockade of GABA_ARs affects the population firing rate in hippocampal networks. Application of GABA_AR antagonist gabazine (30 µM) caused a fast and pronounced increase in the population firing rate to 330 ± 32%, which gradually declined over the course of 2 d in the presence of gabazine (Fig. 1E), despite the constant presence of the antagonist. Washout of gabazine after 2 d caused a significant decrease in firing rate, indicating sustained activity of both gabazine and the GABA_ARs (SI Appendix, Fig. S1). Moreover, a GABA_BR agonist, baclofen, triggered a pronounced block of firing rate that was precisely restored to the baseline level after a period of 2 d (9). The observed compensatory responses to increase in spiking activity by GABA_AR antagonist or decrease in spiking activity by GABA_BR agonist confirm the idea that homeostatic mechanisms maintain stable circuit function by keeping neuronal firing rate around a “set point” (10, 18).

Given a 3.4-fold difference in the magnitude of the initial firing rate increase produced by GABA_AR versus GABA_BR blockade, it is plausible that the lack of firing rate renormalization in the presence of CGP was a result of its relatively weak effect on firing rate. If this is the case, concurrent blockade of GABA_ARs and GABA_BRs would result in a reversal of firing rate, as in the presence of the GABA_AR blocker alone. However, coapplication of gabazine and CGP triggered an initial increase in firing rate by 416 ± 61% that remained at 415 ± 63% for 2 d in the presence of the GABAR blockers (P < 0.001; Fig. 1F), suggesting selective GABA_BR blockade truly disrupts firing rate renormalization. Altogether, these results demonstrate that ongoing GABA_BR activity is required for firing rate homeostasis in hippocampal networks.

GB_1a-Containing GABA_BRs Mediate Homeostatic Increase in Evoked Vesicle Release.

What are the mechanisms underlying disruption of firing rate homeostasis by GABA_BR blockade? To address this question, we assessed the dependency of synaptic homeostatic responses that normally contribute to firing rate homeostasis on the GABA_BR function during activity changes. As a perturbation, we applied tetrodotoxin for 48 h (TTX_48h) to silence spiking activity, a classical paradigm in homeostasis research. First, we asked whether active GABA_BRs are required for inactivity-induced increase in spike-evoked vesicle exocytosis estimated by the FM1-43 method (19). To this end, the total pool of recycling vesicles was stained by maximal stimulation (600 stimuli at 10 Hz) and subsequently destained by 1 Hz stimulation (Fig. 2A and SI Appendix, Fig. S2A). TTX_48h induced a 1.4-fold increase in the destaining rate constant (measured as 1/τ_decay, whereas τ_decay is an exponential time course) and, thus, vesicle exocytosis (P < 0.001; Fig. 2 B and G). However, the adaptive enhancement of release probability (Pr) to inactivity was abolished when neurons were treated with TTX in the presence of CGP over the course of 48 h [(TTX + CGP)_48h; P > 0.4; Fig. 2 C and G]. Importantly, CGP application acutely increased FM destaining rate (P < 0.01; SI Appendix, Fig. S2B), which remained elevated for 48 h in the presence of CGP (CGP_48h; P < 0.01; Fig. 2 C and G), indicating that the expected compensatory reduction in Pr was impaired under the GABA_BR blockade. Acute application of CGP after TTX_48h treatment during FM destaining did not alter adaptive increase in vesicle exocytosis (SI Appendix, Fig. S2C). It is noteworthy that Pr was not saturated under GABA_BR blockade, as presynaptic homeostatic compensation by TTX_48h was normally expressed in high-Pr boutons under elevated extracellular Ca²⁺ levels (Fig. 2 D and G and SI Appendix, Fig. S2D). Moreover, GABA_AR blockade by gabazine for 48 h induced an adaptive reduction in Pr that was prevented by CGP coapplication (SI Appendix, Fig. S3). Thus, the failure in the homeostatic mechanisms is not associated with modulation of basal Pr. Taken together, these results indicate that basal GABA_BR activity is necessary to achieve a compensatory increase in spike-evoked synaptic vesicle exocytosis in hippocampal synapses.

Fig. 2.

Presynaptic homeostatic response is impaired by GABA_BR blockade or GB_1a deletion. (A) Experimental protocol for determining changes in synaptic vesicle exocytosis after prolonged inactivity by TTX_48h (0.5 µM, 48 h). (B–F) Representative FM destaining rate curves of 50 synapses incubated with/without TTX_48h in WT neurons (B), WT neurons treated by CGP_48h (1 µM CGP for 48 h; C), WT neurons grown in increased (2 mM) Ca²⁺ concentration (D), 1a^−/− neurons (E), and 1b^−/− neurons (F). (G) Effect of TTX_48h on average destaining rate constants in WT neurons (n = 687–701), WT neurons incubated with CGP_48h (n = 593–748), WT neurons in presence of high Ca²⁺ (n = 278–331), 1a^−/− neurons (n = 313–327), and 1b^−/− neurons (n = 303–313). (H) Schematic illustration of TTX_48h-induced Pr homeostatic regulation by neuronal inactivity via GABA_B1aRs.

Which isoform of the GABA_BRs mediates homeostatic increase in evoked vesicle release? GABA_BRs are obligatory heterodimers, requiring two homologous subunits, GB₁ and GB₂, for functional expression (20). In hippocampal synapses, the GB_1a isoform is predominantly expressed at glutamatergic presynaptic boutons, whereas GB_1b is predominantly expressed at spines (21). Thus, we examined whether the presynaptic homeostatic response is disrupted in 1a^−/− boutons lacking the GB_1a receptor subunit (21). The 1a^−/− boutons did not display a presynaptic response to activity blockade (Fig. 2 E and G). The deficits in presynaptic homeostatic plasticity were specific for the GB_1a isoform, as 1b^−/− boutons displayed a typical adaption to prolonged synaptic inactivity (Fig. 2 F and G). Notably, acute application of CGP increased evoked synaptic vesicle release in the wild-type and 1b^−/−, but not 1a^−/−, boutons (SI Appendix, Fig. S4), confirming that the GB_1a-containing receptors mediate inhibition of Pr by local GABA levels. Although our previous data demonstrated a correlation between inactivity-induced reduction in the GB_1a receptor activity and increase in Pr (22), our current results suggest that the basal GB_1a receptor activity is required for homeostatic Pr regulation (Fig. 2H).

In contrast to inactivity-induced regulation of spike-evoked vesicle exocytosis, neither acute (SI Appendix, Fig. S5) nor chronic (SI Appendix, Fig. S6) application of CGP affected miniature excitatory postsynaptic current (mEPSC) frequency. Moreover, TTX_48h alone or in the presence of CGP_48h did not significantly change mEPSC frequency (SI Appendix, Fig. S6). It is worth mentioning that TTX_48h reduced short-term synaptic facilitation during spike bursts measured by double-patch recordings (SI Appendix, Fig. S7), indicating an increase in Pr (23). Thus, the difference between spike-evoked and spontaneous vesicle release regulation observed under our experimental conditions reflects differential regulation of exocytosis during spontaneous and evoked synaptic activity (24). Although GABA_BR blockade did not affect regulation of mEPSC frequency, it impaired inactivity-induced increase in mEPSC amplitude, suggesting the postsynaptic GABA_BRs are involved in this regulation. Indeed, the effect of TTX_48h was absent in 1b^−/− neurons (SI Appendix, Fig. S8), suggesting the GB_1b-containg postsynaptic GABA_BRs play an important role in synaptic scaling.

Inactivity Promotes Syntaxin-1 Conformational Switch.

What are the molecular mechanisms underlying the homeostatic increase in Pr by presynaptic GABA_BRs? SNARE-complex assembly is initiated by a syntaxin-1 (Synt₁) switch from its closed conformation (in which the N-terminal H_abc domain of Synt₁ folds back onto its SNARE domain) into an open conformation (in which the SNARE domain becomes unmasked for SNARE-complex formation) (25). Rendering Synt₁ constitutively open induces an increase in Pr by enhancing SNARE-complex assembly per vesicle (26). However, activity-dependent mechanisms regulating Synt₁ conformational switch are not fully understood.

To assess whether chronic inactivity promotes Synt₁ opening, we used a recently developed intramolecular Synt_1a FRET probe (CFP-Synt_1a-YFP) that reports the closed-to-open transition as a decrease in FRET efficiency (27). The Synt_1a sensor contains CFP fluorophore inserted at the N terminus and YFP inserted after the SNARE motif (Fig. 3A) and is well expressed in processes of hippocampal neurons (Fig. 3B). The engineered Synt_1a FRET reporter has been shown to assemble into endogenous SNARE complexes and was able to reconstitute dense-core vesicle exocytosis in PC12 cells (27). Furthermore, we show that neurons transfected with the light chain of BoNT-C1 are not capable of synaptic vesicle recycling, even during strong stimulation (600 pulses at 20 Hz; Fig. 3C). However, coexpression of BoNT-C1, together with BoNT-C1-insensitive CFP-Synt_1aK253I-YFP mutant reporter, restores synaptic vesicle recycling to the level observed in wild-type neurons (Fig. 3C). These results strongly suggest the CFP-Synt_1a-YFP FRET reporter is functional, supporting synaptic vesicle exocytosis in hippocampal neurons.

Fig. 3.

Homeostatic mechanisms promote closed-to-open Synt_1a conformational switch by neuronal inactivity. (A) Schematic illustration of Synt_1a FRET construct (27). (B) Representative confocal image of a hippocampal neuron transfected with CFP-Synt_1a-YFP and zoom-in image of an axon. (C, Left) Synaptic vesicle recycling is blocked in neurons expressing BoNT-C1 light chain but is rescued by CFP-Synt_1a-K253I-YFP FRET probe. FM4-64 staining during 20-Hz stimulation (600 APs). (Right) Quantification of total presynaptic strength (S) in neurons expressing (1) CFP only (Cnt, n = 8) (2), BoNT-C1 (n = 12), and (3) BoNT-C1 + CFP-Synt_1a-K253I-YFP FRET probe (n = 9). (D) Pseudocolor-coded fluorescent images of CFP-Synt_1a-YFP protein expressing bouton before and after YFP photobleaching. (E) Synt_1a^Open reduces E_m (n = 30–38). (F) Summary of the TTX_48h effect in neurons expressing wild-type Synt_1a probe (n = 23–34) and Synt_1a^Open (n = 38–25). (G) Average destaining rate constants in Synt_1a^WT (n = 284), Synt_1a^WT treated by TTX_48h (n = 181), Synt_1a^Open (n = 270), and Synt_1a^Open treated by TTX_48h (n = 192). [Scale bars, 20 µm (B, 1) and 2 µm (B, 2 and 3, and D)].

To monitor Synt_1a conformational changes, we measured the steady-state FRET efficiency (E_m), using the acceptor photobleaching method at presynaptic boutons expressing CFP-Synt_1a-YFP (Fig. 3B, 2 and 3). High-magnification confocal images show an increase in CFP fluorescence after YFP photobleaching (Fig. 3D), indicating dequenching of the donor and the presence of FRET. On average, CFP-Synt_1a-YFP FRET efficiency across hippocampal boutons was 0.12 ± 0.02 (Fig. 3E). To validate that the probe reports closed-to-open transition as a decrease in FRET efficiency, we used the Synt_1a^Open FRET probe with L165E/L166E mutations, rendering Synt₁ in a constitutively open conformation (25). Indeed, Synt_1a^Open displayed 56% lower FRET efficiency in comparison with the wild-type Synt_1a probe (0.053 ± 0.01; P < 0.01; Fig. 3E).

Next, we asked whether Synt_1a conformation is homeostatically regulated by chronic neuronal inactivity to promote Pr augmentation. Indeed, TTX_48h induced a reduction in Synt_1a FRET to 0.05 ± 0.01 (P < 0.01; Fig. 3F). Notably, the reduction magnitude by TTX_48h was similar to that exhibited by Synt_1a^Open and was occluded by Synt_1a^Open. At the functional level, Synt_1a^Open increased the rate of vesicle exocytosis (P < 0.05; Fig. 3G), supporting results of earlier studies (26, 28). Most important, Synt_1a^Open occluded the effect of TTX_48h on Pr (P > 0.7; Fig. 3G), suggesting a functional importance of Synt_1a opening in compensatory Pr augmentation.

Removal of GABA_BR Block Mediates Inactivity-Induced Syntaxin-1 Opening.

To examine whether GABA_BR tone is required for inactivity-induced Synt_1a opening, we first asked whether Synt_1a conformation is regulated by basal GABA_BR activity. Acute application of GABA_BR antagonist CGP reduced mean FRET level to 0.048 ± 0.01 (P < 0.01; Fig. 4A). Application of GABA_BR agonist baclofen (10 μM) did not affect FRET (P > 0.05; Fig. 4A), indicating that basal GABA levels are sufficient to stabilize a closed Synt_1a conformation. Next, we asked whether GABA_BR blockade interferes with TTX_48h-induced Synt_1a opening. CGP_48h caused a reduction in Synt_1a FRET (P < 0.01; Fig. 4B) and occluded the effect of TTX_48h on Syn_1a conformational change (Fig. 4B). The effect of CGP_48h was mimicked by deletion of the GB_1a receptor subunit: Synt_1a FRET was twofold lower in 1a^−/− boutons and insensitive to chronic reduction in spiking activity by TTX_48h (Fig. 4B). Importantly, acute application of baclofen restored Synt_1a FRET in TTX_48h-treated neurons (Fig. 4C), suggesting prolonged silencing may promote Synt_1a opening by reducing the extracellular GABA levels (22).

Fig. 4.

Neuronal inactivity promotes Synt₁ opening by removal of GABA_BR-mediated block. (A) Acute application of CGP reduced CFP-Synt_1a-YFP E_m (n = 38), whereas 10 μM baclofen did not affect it (n = 88) in WT neurons. (B) Summary of the TTX_48h effect on CFP-Synt_1a-YFP E_m in the presence of CGP_48h in WT (n = 42–45) and in 1a^−/− (n = 40–80) neurons. (C) Baclofen (10 µM) rescued TTX_48h-induced FRET reduction (n = 25–40). (D) Representative confocal images of boutons cotransfected with GB_1a^Cer and Synt_1a^YFP. (Scale bar, 1 µm.) (E) Effect of TTX_48h on GB_1a^Cer/Synt_1a^YFP E_m (n = 48–57). CGP_48h increases GB_1a^Cer/Synt_1a^YFP E_m and abolishes TTX_48h-induced E_m changes (n = 41–69). (F) Schematic illustration of Synt₁ conformational switch regulation by neuronal inactivity via GABA_BRs.

Having established the necessity for GB_1a-containing GABA_BRs in the homeostatic conformational change of Synt_1a, we explored a possibility of Synt_1a and GB_1a interactions. To quantify activity-dependent changes in GB_1a–Synt_1a interactions at individual presynaptic sites, we used the FRET approach. To preserve the functionality of the Synt_1a FRET reporter, we replaced CFP by its nonfluorescent mutant CFP-W66A, and a Cerulean (Cer)-tagged GB_1a subunit at the C terminus (GB_1a^Cer) was used as a donor (Fig. 4D). Basal GB_1a^Cer/Synt_1a^YFP FRET efficiency was 0.04 ± 0.01 and underwent a 2.4-fold increase by TTX_48h (P < 0.01; Fig. 4E). Notably, blockade of GABA_BRs produced a similar effect (P < 0.01; Fig. 4E), indicating that GB_1a/Synt_1a interactions are regulated by basal GABA. Moreover, CGP_48h occluded TTX_48h-induced GB_1a^Cer/Synt_1a^YFP FRET changes (P = 0.9; Fig. 4E). To determine the existence of endogenous protein complexes containing Synt₁ and GABA_BRs, we performed coimmunoprecipitation experiments with solubilized mouse brain membranes. Anti-Synt₁ antibodies coprecipitated a significant amount of GB₁ proteins together with Synt₁ from WT, but not from full GB₁-KO mice (SI Appendix, Fig. S9). Taken together, these results indicate that Synt_1a conformational change constitutes a critical step in presynaptic homeostatic response, that basal GABA_BR activity maintains Synt_1a in a closed conformation (Fig. 4F, 1), and that prolonged inactivity removes a GABA_BR-imposed clamping of a closed Synt_1a conformation, allowing Synt_1a shift toward its open conformation (Fig. 4F, 2).

GB_1a Is Required for Homeostatic Increase in Presynaptic Calcium Flux.

Next, we examined whether the GB_1a receptor subunit controls homeostatic regulation of presynaptic Ca²⁺ transients (29), in addition to regulating Synt₁ conformational switch, to adapt Pr to chronic activity perturbations. Thus, we measured presynaptic Ca²⁺ transients evoked by 0.1-Hz stimulation, using high-affinity fluorescent calcium indicator Oregon Green 488 BAPTA-1 AM (Fig. 5A) at functional boutons identified by the FM4-64 marker. Although the size of action-potential dependent fluorescence transients (ΔF/F) varied between different boutons, presynaptic Ca²⁺ flux was significantly larger in TTX_48h-treated WT neurons (P < 0.0001; Fig. 5 A and C). Importantly, presynaptic Ca²⁺ flux was higher in boutons of 1a^−/− neurons (P < 0.0001; Fig. 5 B and C), occluding the effect of TTX_48h on further potentiation of calcium transients. It is noteworthy that Ca²⁺ transients were not saturated by single action potential in 1a^−/− boutons (SI Appendix, Fig. S10). Furthermore, acute application of CGP increased presynaptic Ca²⁺ flux that remained stable over the course of 2 d in the presence of CGP in WT (P < 0.001; Fig. 5 A and D), but not in 1a^−/− (P > 0.4; Fig. 5 B and D), neurons. This lack of compensation at the level of Ca²⁺ flux was specific for the GABA_BR deficit, as GABA_AR blockade by gabazine has been previously shown to trigger an adaptive reduction in presynaptic Ca²⁺ transients (29). These results indicate that GB_1a-containing GABA_BRs normally inhibit presynaptic Ca²⁺ flux evoked by single action potential and that removal of the GABA_BR-mediated block is essential for homeostatic potentiation of Ca²⁺ flux by chronic inactivity.

Fig. 5.

Homeostatic increase in presynaptic calcium flux is impaired by GB_1a deletion. (A and B) TTX_48h increased spike-dependent presynaptic Ca²⁺ entry in boutons of WT, but not in 1a^−/− neurons. (Top) Representative images of Ca²⁺ transients (average of seven traces) evoked by 0.2-Hz stimulation during a 500-Hz line scan in boutons under control, or TTX_48h, CGP_acute, and CGP_48h conditions in WT (A) and 1a^−/− (B) neurons. (Scale bar, 100 ms.) (Bottom) Averaged Ca²⁺ transients, quantified as ΔF/F, in control (n = 20), TTX_48h (n = 15), CGP_acute (n = 29), and CGP_48h (n = 37) conditions in WT (A) and 1a^−/− (B, n = 20, 21, 23, and 66 for Cnt, TTX_48h, CGP_acute, and CGP_48h, respectively) neurons. (C) Summary of TTX_48h effect (percentage of control) on Ca²⁺ transients in boutons of WT versus 1a^−/− neurons (the same data as in A and B). Note that Ca²⁺ transients were higher in 1a^−/− boutons than in WT ones. (D) Summary of CGP_acute and CGP_48h effects (percentage of control) on Ca²⁺ transients in WT versus 1a^−/− boutons (the same data as in A and B).

GB_1aPCT Mediates Presynaptic Homeostatic Response.

Next, we searched for the molecular domain in the GB_1a protein that mediates the presynaptic homeostatic response. In our previous work, we identified the proximal C-terminal domain of the GB_1a protein (R857-S877; GB_1aPCT; Fig. 6A) as essential for the compartmentalization of the presynaptic signaling complex of GABA_BRs, Gβγ G protein subunits, and Ca_V2.2 channels in hippocampal boutons (30). Interestingly, deletion of GB_1aPCT domain specifically impaired Ca_V2.2/Gβγ interaction and function, leaving Gα_i/o-dependent signaling unaltered.

Fig. 6.

The GB_1aPCT domain is required for presynaptic homeostatic response. (A) Schematics show GB_1aWT and GB_1aΔPCT constructs. 7TM, seven-transmembrane domain; CC, coiled-coiled domain; DCT, distal C-terminal domain; LBD, ligand-binding domain; PCT, proximal C-terminal domain; SD, two sushi domains. (B and C) Representative FM destaining rate curves of 50 synapses pretreated with/without TTX_48h in 1a^−/− neurons transfected with GB_1aWT (B) or GB_1aΔPCT (C). (D) Expression of GB_1aWT (n = 609–642), but not of GB_1aΔPCT (n = 598–721), restores presynaptic homeostatic adaptation in 1a^−/− neurons. (E and F) TTX_48h did not alter spike-dependent presynaptic Ca²⁺ entry in boutons expressing GB_1aΔPCT in 1a^−/− neurons. Representative images of Ca²⁺ transients (average of seven traces) evoked by 0.2-Hz stimulation in boutons of control and TTX_48h-treated GB_1aΔPCT-expressing boutons (E). Averaged Ca²⁺ transients in control (n = 44) and TTX_48h (n = 37) conditions (F). (G) Summary of TTX_48h effect on Ca²⁺ transients in boutons expressing GB_1aΔPCT protein (the same data as in F). (H) Effect of TTX_48h on GB_1a^YFP/Ca_V2.2^CFP E_m (n = 40–88). Deletion of PCT domain abolishes TTX_48h-induced change in GB_1a^YFP/Ca_V2.2^CFP E_m (n = 44–55).

First, we assessed the functional role of the GB_1aPCT domain in the homeostatic increase of Pr by comparing the effect of TTX_48h on FM4-64 destaining rates in 1a^−/− neurons transfected with GB_1aWT-CFP versus GB_1aΔPCT-CFP proteins. Expression of the GB_1aWT-CFP protein rescued inactivity-induced potentiation of vesicle exocytosis in 1a^−/− neurons (P < 0.01; Fig. 6 B and D in comparison with Fig. 2G). In contrast, TTX_48h-induced presynaptic enhancement remained impaired in boutons expressing GB_1aΔPCT-CFP (P > 0.8; Fig. 6 C and D). Moreover, deletion of the GB_1aPCT domain abolished inactivity-induced increase in presynaptic calcium transients (P > 0.8; Fig. 6 E–G). Thus, the GB_1aPCT domain is necessary for presynaptic adaptations to prolonged inactivity in hippocampal networks.

Finally, we examined whether inactivity induces conformational GABA_BR/Ca_V2.2 changes, and if so, whether the GB_1aPCT domain is involved in this homeostatic regulation. We monitored FRET efficiency between the YFP-tagged GB_1a receptor subunit at the C terminus and the CFP-tagged α1 subunit of the Ca_V2.2 channel at the N terminus (GB_1a^YFP/Ca_V2.2^CFP). TTX_48h induced an increase in GB_1a^YFP/Ca_V2.2^CFP FRET (P < 0.05; Fig. 6H), indicating that chronic neuronal silencing alters GB_1a/Ca_V2.2 interactions. Importantly, deletion of the GB_1aPCT domain disrupted TTX_48h-induced homeostatic GB_1a^YFP/Ca_V2.2^CFP changes (P = 0.49; Fig. 6H). Moreover, GB_1aPCT deletion occluded TTX_48h-induced homeostatic GB_1a^Cer/GB₂^Cit and GB_1a^YFP/Synt_1a^CFP changes (SI Appendix, Fig. S11). A physical interaction between endogenous GABA_BRs and Ca_V2.2 was revealed by coimmunoprecipitation experiments (SI Appendix, Fig. S9), confirming previous proteomic data (31). These results suggest that homeostatic mechanisms modulate GB_1a/Ca_V2.2 and GB_1a/Synt_1a interactions in the GABA_BR presynaptic signaling complex via the GB_1aPCT domain.

Discussion

The ability of neuronal circuits to stabilize their firing properties in the face of environmental or genetic changes is critical for normal neuronal functioning. Despite extensive research on a wide repertoire of possible homeostatic mechanisms, the key regulators of firing rate homeostasis in mammalian central neural circuits remain obscure. In this work, we identified the GABA_BR as a key homeostatic signaling molecule stabilizing mean firing rate in hippocampal networks. GABA_BRs enable inactivity-induced homeostatic increase in synaptic strength by three principle mechanisms: promoting synatxin-1 conformational switch to enhance SNARE-complex assembly, augmenting presynaptic Ca²⁺ flux to promote spike-evoked vesicle exocytosis, and increasing the quantal excitatory amplitude. Thus, GABA_BRs, in addition to modulation of short-term (32, 33) and long-term, Hebbian (21) synaptic plasticity, are essential for maintaining firing stability of neural circuits.

GABA_BRs and Synaptic Homeostatic Plasticity.

Homeostatic regulation of synaptic strength represents a basic mechanism of neuronal adaptation to constant changes in ongoing activity levels. Strong evidence exists on homeostatic augmentation of Pr and readily releasable pool size in response to prolong inactivity in hippocampal neurons (19, 22, 34–42). These homeostatic adaptations are associated with modulation of presynaptic Ca²⁺ flux (29) and remodeling of a large number of proteins in presynaptic cytomatrix (7). Recent studies have identified the mechanisms underlying presynaptic homeostatic signaling in Drosophila neuromuscular junction, implicating epithelial sodium leak channels (43) and endostatin (44) as homeostatic regulators of the presynaptic Ca_V2 channels (for review, see ref. 5). However, the critical molecules controlling presynaptic homeostasis in mammalian central synapses are not fully understood.

In this work, we show that chronic neuronal silencing induces an adaptive increase in evoked basal vesicle release through GABA_BRs by removing tonic inhibition of Synt₁ conformational switch and of presynaptic Ca²⁺ flux. These results are important for several reasons. First, they reveal a crucial role for GABA_BRs in presynaptic homeostasis. Taking into account a wide variety of G protein-coupled receptors that mediate presynaptic inhibition, the requirement for the GABA_BR tone in presynaptic homeostatic response is particularly striking. Second, they demonstrate, for the first time, the role of the extracellular GABA in determining Synt₁ conformation via the presynaptic GB_1a-containing GABA_BRs. Either genetic GB_1a deletion or pharmacological GABA_BR blockade stabilizes an open Synt₁ conformation in analogy to mutations rendering Synt₁ constitutively open, occluding adaptive response to neuronal silencing. Notably, addition of GABA_BR agonist after prolonged inactivity stabilizes a closed Synt₁ conformation, suggesting reduction in local GABA levels induces an adaptive Synt_1a response. Third, in addition to the well-known role of GABA_BRs in the presynaptic Ca²⁺ flux inhibition, at a rapid timescale of minutes (22, 45), our work revealed the necessity for basal GABA_BR activity in presynaptic adaptations of Ca²⁺ transients to chronic activity perturbations at extended timescales of days. Deletion of the GB_1aPCT domain blocks presynaptic homeostatic plasticity by disrupting GABA-mediated conformational changes within the presynaptic GB_1a/Ca_V2.2/Synt₁ signaling complex. Thus, endogenous molecular “brakes” imposed by GABA_BRs on Ca_V2.2 channels and SNARE complex assembly are essential for presynaptic homeostasis in hippocampal neurons.

It is important to note that in the present study, chronic inactivity by TTX induced a compensatory increase in mEPSC amplitude via the postsynaptic GB_1b-containing receptors without affecting mEPSC frequency. Given a pronounced effect of TTX_48h on spike-evoked synaptic vesicle exocytosis, these results suggest complete blockade of spikes does not trigger compensatory changes in spontaneous vesicle release. In previous studies, mEPSC frequency was found to be immune to chronic TTX treatment in some cases (18, 41), while being up-regulated by AMPA receptor blockade (36, 41), either by suppression of neuronal excitability (38) or by increase in the GABA_BR-mediated inhibition (9). Thus, the induction of presynaptic homeostatic changes may require minimal spiking activity (41).

GABA_BR-Mediated Neuronal Homeostasis and Brain Disorders.

It is tempting to speculate that many distinct neurologic and psychiatric disorders with different etiologies share common dysfunctions in pathways related to homeostatic plasticity. However, the molecular mechanisms by which defective homeostatic signaling may lead to common disease pathophysiology remain to be determined. Only a few molecular links, such as the schizophrenia-associated gene dysbindin, have been established between homeostatic system impairments and brain dysfunctions (46). Our study demonstrates that ongoing GABA_BR activity is essential for population firing rate homeostasis in hippocampal networks. This may explain why aberrant neuronal activity remains uncompensated in mice lacking functional GABA_BRs as a result of deletion of either the GB₁ (13) or GB₂ (15) receptor subunit. Interestingly, 1a^−/−, but not 1b^−/−, mice exhibit spontaneous epileptiform activity (16), suggesting the presynaptic GB_1a-containing receptors may play a more prominent role in firing rate homeostasis. Strikingly, our results show that homeostatic plasticity is impaired in synaptic networks displaying enhanced ongoing synaptic Ca²⁺ flux because of removal of the GABA_BR-mediated block. Thus, Ca_V2 channel gain of function may be as detrimental for neuronal homeostasis as Ca_V2 loss of function (47, 48), indicating that an optimal level of ongoing synaptic Ca²⁺ flux may be essential for homeostatic regulation. It remains to be seen whether the loss of functional GABA_BRs, associated with epilepsy and a wide range of psychiatric disorders (49), contributes to pathophysiology shared by these disorders through erasing critical pathways in homeostatic control systems.

Materials and Methods

Hippocampal Cell Culture.

Primary cultures of CA3-CA1 hippocampal neurons were prepared from WT, 1a^−/−, and 1b^−/− (BALB/c background) mice (21) on postnatal days 0–2, as described (50). The experiments were performed in mature (14–28 d in vitro) cultures. All animal experiments were approved by the Tel Aviv University Committee on Animal Care.

MEA Preparation and Recordings.

Cultures were plated on MEA plates containing 59 TiN recording and one internal reference electrodes [Multi Channel Systems (MCS)]. Electrodes are 30 µm in diameter and spaced 500 µm apart. Data were acquired using a MEA1060-Inv-BC-Standard amplifier (MCS) with frequency limits of 5,000 Hz and a sampling rate of 10 kHz per electrode mounted on an Olympus IX71 inverted microscope. Recordings were carried out under constant 37 °C and 5% CO₂ conditions, identical to incubator conditions. Spike sorting and analysis are described in ref. 9 and in SI Appendix, Spike Sorting and Data Analysis.

Molecular Biology.

GB_1aWT-, GB_1aΔPCT-, GB₂-, and Ca_V2.2-tagged proteins used throughout the study were constructed as described before (30). Synt_1a (CSYS-5RK), Synt_1a^Open, and Synt_1aK253I are as described in ref. 27. W66A point mutation was introduced to silence YFP in the YFP-Synt_1a-CFP construct (Fig. 4 D and E and SI Appendix, Fig. S11). BoNT/C1α-51-IRES-EGFP was designed and then generated by ProGen Israel, Protein and Gene Engineering Company. Transient cDNA transfections have been performed using Lipofectamine-2000 reagents, and neurons were typically imaged 18–48 h after transfection.

Estimation of Synaptic Vesicle Release Using FM Dyes.

Activity-dependent FM1-43 or FM4-64 styryl dyes have been used to estimate synaptic vesicle exocytosis and presynaptic plasticity, as described (22). The experiments were conducted at room temperature in extracellular Tyrode solution containing (in mM): NaCl, 145; KCl, 3; glucose, 15; Hepes, 10; MgCl₂, 1.2; CaCl₂, 1.2, with pH adjusted to 7.4 with NaOH. The extracellular medium contained nonselective antagonist of ionotropic glutamate receptors (kynurenic acid, 0.5 mM) to block recurrent neuronal activity. Synaptic vesicles were loaded with 15 μM FM4-64 in all of the experiments with GFP/CFP/YFP transfection, whereas 10 μM FM1-43 was used in all of the nontransfected neurons.

Detecting Presynaptic Calcium Transients.

Fluorescent calcium indicator Oregon Green 488 BAPTA-1 AM was dissolved in DMSO to yield a concentration of 1 mM. For cell loading, cultures were incubated at 37 °C for 30 min with 3 μM of this solution diluted in a standard extracellular solution. Imaging was performed using FV1000 Olympus confocal microscopes, under 488 nm (excitation) and 510–570 nm (emission), using 500-Hz line scanning.

FRET Imaging and Analysis.

Intensity-based FRET imaging was carried as described before (22, 30). Donor dequenching resulting from the desensitized acceptor was measured from Cer/CFP emission (460–500 nm) before and after the acceptor (YFP) photobleaching. Mean FRET efficiency, E_m, was then calculated using the equation E_m = 1 − I_DA/I_D, where I_DA is the peak of donor emission in the presence of the acceptor and I_D is the peak after acceptor photobleaching.

Statistical Analysis.

Error bars shown in the figures represent SEM. Where applicable, the number of experiments (cultures) or the number of boutons is defined by n. All the experiments were repeated at least in three different batches of cultures. One-way ANOVA analysis with post hoc Bonferroni’s was used to compare several conditions. Student’s unpaired, two-tailed t test has been used in the experiments in which two different populations of synapses were compared. Student’s paired, two-tailed t test has been used in the experiments where before and after treatments were compared at the same population of synapses. *P < 0.05; **P < 0.01; ***P < 0.001; ns, nonsignificant.