Silencing of spontaneous activity at α4β1/3δ GABA_A receptors in hippocampal granule cells reveals different ligand pharmacology

Abstract

Background and Purpose

The δ‐subunit‐containing GABA_A receptors, α₄β₁δ and α₄β₃δ, in dentate gyrus granule cells (DGGCs) are known to exhibit both spontaneous channel openings (i.e. constitutive activity) and agonist‐induced current. The functional implications of spontaneous gating are unclear. In this study, we tested the hypothesis that constitutively active α₄β_1/3δ receptors limit agonist efficacy.

Experimental Approach

Whole‐cell electrophysiological recordings of adult male rat and mouse hippocampal DGGCs were used to characterize known agonists and antagonists at δ‐subunit‐containing GABA_A receptors. To separate constitutive and agonist‐induced currents, different recording conditions were employed.

Key Results

Recordings at either 24°C or 34°C, including the PKC autoinhibitory peptide (19–36) intracellularly, removed spontaneous gating by GABA_A receptors. In the absence of spontaneous gating, DGGCs responded to the α₄β_1/3δ orthosteric agonist Thio‐THIP with a four‐fold increased efficacy relative to recording conditions favouring constitutive activity. Surprisingly, the neutral antagonist gabazine was unable to antagonize the current by Thio‐THIP. Furthermore, a current was elicited by gabazine alone only when the constitutive current was silenced (EC₅₀ 2.1 μM). The gabazine‐induced current was inhibited by picrotoxin, potentiated by DS2, completely absent in δ^−/− mice and reduced in β₁ ^−/− mice, but could not be replicated in human α₄β_1/3δ receptors expressed heterologously in HEK cells.

Conclusion and Implications

Kinase activity infers spontaneous gating in α₄β_1/3δ receptors in DGGCs. This significantly limits the efficacy of GABA_A agonists and has implications in pathologies involving aberrant excitability caused by phosphorylation (e.g. addiction and epilepsy). In such cases, the efficacy of δ‐preferring GABA_A ligands may be reduced.

Abbreviations

ACSF

artificial CSF

BIC

bicuculline

CaMKII

Ca²⁺/Calmodulin‐dependent protein kinase II

CP

capacitance

DGGC

dentate gyrus granule cell

DS2

4‐chloro‐N‐[2‐(2‐thienyl)imidazo[1,2a]pyridine‐3‐yl]benzamide

EDTA

2,2′,2″,2‴‐(Ethane‐1,2‐diyldinitrilo)tetraacetic acid

EGTA

Ethylene glycol‐bis(2‐aminoethylether)‐N,N,N′,N′‐tetraacetic acid

FLIPR

fluorescent imaging plate reader

FMP

FLIPR membrane potential

GBZ

gabazine

ICS

intracellular solution

IPSC

inhibitory post‐synaptic current

NMDG

N‐Methyl D‐Glucamine

PTX

picrotoxin

RS

series resistance

RT

room temperature

SEM

standard error of mean

SNP

single nucleotide polymorphism

TCD

tonic current density

TEA

tetraethylammonium

Thio‐THIP

4,5,6,7‐Tetrahydroisothiazolo[5,4‐c]pyridin‐3‐ol

THIP

4,5,6,7‐Tetrahydroisoxazolo[5,4‐c]pyridin‐3‐ol

TPMPA

(1,2,5,6‐Tetrahydropyridin‐4‐yl)methylphosphinic acid

TTX

tetrodotoxin

What is already known

• The α₄β_1/3δ GABA_A receptors exhibit both ligand‐mediated and spontaneous gating.

• Spontaneous gating by GABA_A receptor channels in neurons cannot be inhibited by gabazine.

What this study adds

• The spontaneous gating by GABA_A receptors in hippocampal granule cells is kinase‐dependent.

• Efficacy of GABA_A orthosteric agonists at δ‐subunit‐containing receptors depend on the level of constitutive activity.

What is the clinical significance

• The agonist efficacy at α₄β_1/3δ receptors in vivo is dependent on the intracellular kinase activity.

• This is relevant to pathologies such as addiction and epilepsy.

1. INTRODUCTION

The functional GABA_A receptor that comprises the orthosteric binding site for GABA is a pentamer typically composed of two α‐, two β‐ and one γ‐ or δ‐subunit. Sampled from a population of six α (1–6), three β (1–3), two γ (1–2) and one δ, the possible number of permutations is in the hundreds. However, assembly rules favouring distinct combinations allow for only a subset to exist in vivo (Martenson, Yamasaki, Chaudhury, Albrecht, & Tomita, 2017; Olsen & Sieghart, 2008). The anatomical localization is subunit‐specific: Typically, the γ‐subunit‐containing receptors mediate a synaptic, fast inhibitory post‐synaptic current (IPSC) and the δ‐ and the α_4–6‐subunit‐containing receptors mediate a slow extrasynaptic tonic current (Chandra et al., 2006; Farrant & Nusser, 2005). The magnitude of the tonic current is region‐ and neuron‐specific.The general view is that GABA diffusing from synapses or released into extrasynaptic areas and buffered by GABA uptake is the main determinant of GABA in the CSF and of the tonic current (reviewed by Glykys & Mody, 2007; Lee & Maguire, 2014; Overstreet‐Wadiche & McBain, 2015). However, several GABA_A receptor subtypes in recombinant systems, cultured hippocampal neurons and acute slices also display a degree of constitutive activity (Birnir, Everitt, Lim, & Gage, 2000; Hoestgaard‐Jensen et al., 2014; McCartney, Deeb, Henderson, & Hales, 2007; Othman et al., 2012; Tang, Hernandez, & Macdonald, 2010; Wlodarczyk et al., 2013). For example, the receptor subunit combinations α_1/4β₃γ_2(L) and α₄β_1/3δ are capable of both GABA‐activated and constitutive gating, the latter (α₄β₃δ) in a manner dependent on PKA activity (Hoestgaard‐Jensen et al., 2014; Jensen et al., 2013; Tang et al., 2010). Since the hippocampal dentate gyrus granule cells (DGGCs) in adult rats and mice exhibit high levels of co‐expressed α₄‐, β_1/3‐ and δ‐subunits (Sperk, Schwarzer, Tsunashima, Fuchs, & Sieghart, 1997), this suggests that α₄β_1/3δ GABA_A receptors are very probable inhabitants at peri‐ and extra‐synaptic loci in DGGCs (Chandra et al., 2006; Herd et al., 2008; Wei, Zhang, Peng, Houser, & Mody, 2003; Zhang, Wei, Mody, & Houser, 2007). A valuable tool for discriminating α₄βδ subtypes is the orthosteric agonist 4,5,6,7‐tetrahydroisothiazolo[5,4‐c]pyridin‐3‐ol (Thio‐THIP), which is highly selective for β_1/3 over β₂‐containing subtypes (Hoestgaard‐Jensen et al., 2014). By contrast, the closely related compound 4,5,6,7‐tetrahydroisoxazolo[5,4‐c]pyridin‐3‐ol (THIP) does not display this selectivity (used for reference in the present study). In a previous study, application of Thio‐THIP, paradoxically suggested a very limited expression of α₄β_1/3δ receptors in DGGCs and mainly at synaptic loci (Hoestgaard‐Jensen et al., 2014). Because constitutively active receptor populations of the α₄β₃δ subtype in recombinant systems simultaneously display reduced efficacy to low levels of GABA (Jensen et al., 2013; Tang et al., 2010), we hypothesize that the absent response to Thio‐THIP in DGGCs could be due to the presence of a constitutive current, providing a so‐called baseline‐floor effect of tonic inhibition (Tang et al., 2010). For these reasons, the present study was designed to separate the constitutive and agonist‐mediated gating in α₄β_1/3δ receptors in DGGCs using a systematic approach of different recording conditions and a carefully selected set of pharmacological tool compounds (Thio‐THIP and THIP), the neutral antagonist gabazine (GBZ), the inverse agonist bicuculline (BIC) and the channel blocker picrotoxin (PTX). Gabazine was previously shown to be inactive towards constitutively active receptors, that is, designated as a neutral antagonist. Thus, part of this work also sought to reassess the precise pharmacological action of gabazine.

2. METHODS

2.1. Chemical compounds

THIP and Thio‐THIP were synthesized at the Department of Drug Design and Pharmacology, University of Copenhagen, as previously described (Krehan et al., 2003). CGP54626, tetrodotoxin (TTX), gabazine, (1,2,5,6‐Tetrahydropyridin‐4‐yl)methylphosphinic acid (TPMPA), PKC autoinhibitory peptide (19–36), (4‐chloro‐N‐[2‐(2‐thienyl)imidazo[1,2a]pyridine‐3‐yl]benzamide (DS2) and GABA were obtained from Tocris Bioscience (Bristol, UK), while kynurenate, strychnine, atropine, picrotoxin and bicuculline‐methiodide were obtained from Sigma‐Aldrich (St. Louis, MO, USA).

2.2. Animal work and compliance with requirements for studies using animals in accordance with ARRIVE

Animal studies are reported in compliance with the ARRIVE guidelines (Kilkenny, Browne, Cuthill, Emerson, & Altman, 2010) and with the recommendations made by the British Journal of Pharmacology. The use of rat and mouse tissue for the present studies were necessary because the specific receptor pharmacology under study is dependent on neuron‐specific receptor‐interacting kinases and furthermore requires a functional synaptic circuitry, none of which can be faithfully reproduced in recombinant systems. All use and housing of animals described below were carried out in accordance with the European Communities Council Directive (86/609/EEC) for the care and use of laboratory animals and the Danish legislation regulating animal experiments. Adult male Sprague Dawley rats were obtained from Envigo and housed socially in cages holding up to six rats with ad libitum access to food and water in rooms maintained at 22–24°C, 55% humidity at a 12‐h/12‐h normal night/day cycle. Rats were housed for 7–21 days and aged 49–70 days at the time of experiment (weighing 230–300 g). Mice strains (described below) were bred at in‐house facilities and housed socially as described for rats. Only male mice at the age range of 60–120 days (weighing 25–35 g) were used. Cages were equipped with standard woodchip bedding, wood gnawing blocks, nestlets, cardboard tunnels (mice only) and red plastic huts. On the day of experiment, animals were terminated by decapitation (rats) or cervical dislocation (mice).

For mouse studies, δ and β₁ GABA_A receptor subunit null (germline) mutants were employed. A breeding pair of mice heterozygous for the δ deletion (Mihalek et al., 1999) of the strain C57BL/6J × 129Sv/SvJ was kindly provided by Dr. Delia Belelli, University of Dundee, Scotland and bred in‐house. Mice carrying deletions in the β₁ gene were generated and bred in‐house as described below. All mice were bred at the specific pathogen‐free animal facility at the University of Copenhagen. For both δ and β₁ deleted strains, heterozygotes were intercrossed to create littermates of all three genotypes (^−/−, ^+/+ and ^+/−) and monitored by PCR genotyping. Only ^−/− and ^+/+ male mice resulting from heterozygous breeding were used for the studies.

2.3. Production of β₁ ^−/− mice

For the generation of mice with a targeted deletion of gabrb1, a Knockout First promoter driven mouse Gabrb1tm1a(KOMP)Wtsi was purchased from the KOMP repository (www.komp.org) and used to generate a full germline knockout (β₁ ^−/−). In the current study, only constitutive knockout mice were used. The conditional strain was simply a necessary intermediate step in the process of generating the full knockout. To this end, the obtained mouse was first crossed with a global Flp‐expressing mouse to generate a conditional loxP‐flanked (floxed) allele. Secondly, this mouse was crossed with a global Cre‐expressing mouse to generate mice with a constitutive knockout allele, β₁ ^+/‐Cre+. These were back‐crossed into the C57BL/6 background to remove the Cre and Flp and to produce a genetically pure offspring and carefully monitored by PCR analysis. The resulting mouse lacks exon 4 of the gabrb1 gene and instead contains one frt and one loxP site. After several steps of backcrossing to the C57BL/6J background strain, the generation equivalent level was determined by speed congenics using single nucleotide polymorphism (SNP) analysis technology (1450 SNP marker panel) (Taconic, Lille Skensved, Denmark). Of seven tested heterozygous males, all were found to be at least 99.72% identical to the C57BL/6J background, thus at least equivalent to the N8 generation. The resulting knockout, mice lacking either one or both deleted alleles (β₁ ^+/− or β₁ ^−/−), was viable and born at expected Mendelian ratios. The mice were normal in their development, fertile and did not exhibit obvious behavioural or physical phenotypes. Animals were genotyped using the REDExtract‐N‐Amp™ Tissue PCR Kit (Sigma‐Aldrich) according to the manufacturer's instructions. Primers were designed to discriminate knockout and wild‐type β₁ alleles (forward primer 5′‐GTATGGTCAGATGTCCTCAC‐3′ and reverse primer 5′‐CTGTCTTGCTAGCTTTGAG‐3′). The amplification was as follows (94°C 5 min initial; then 94°C 30 s, annealing at 57°C for 30 s, extension at 72°C for 1.5 min for 35 cycles) to generate fragments of 1,211 bp for wild‐type allele and 572 bp for the knockout allele. All mice used were genotyped before experiments. Expression at the protein level was also assessed with Western blot (see below).

2.4. Slice electrophysiology

At the day of experiment, a single rat or mouse was decapitated and the head immediately immersed in ice‐cold artificial CSF (ACSF, composition below) for 6–8 min before dissection of the brain. For dissection and slicing of the brain, ACSF contained (in mM): N‐Methyl D‐Glucamine (NMDG, 100), NaCl (26), KCl (2.5), CaCl₂ (1), MgCl₂ (3), NaHCO₃ (26), NaH₂PO₄ (1.25), D‐glucose (10), ascorbate (0.3), pyruvic acid (0.3) and kynurenic acid (1), adjusted to 305 ± 3 mOsm and aerated with carbogen (95% O₂/5% CO₂). The brain was glued to the base of a Leica VT1200 S vibratome and cut in 350‐μm‐thick horizontal sections. Slices were stored at 28–29°C in a carbogenated ACSF of composition as above but without kynurenate and NMDG replaced with equimolar NaCl. Whole‐cell patch clamp recordings of individual neurons in slices began after a 90‐min recovery period and slices were used up to 4 h thereafter. Recording ACSF were in composition of bulk chemicals similar to storage ACSF but contained in addition kynurenate (1 mM), CGP54626 (1 μM), atropine (1 μM), strychnine (1 μM) and TTX (0.5 μM). During recordings, slices were maintained in a chamber holding 2 ml of ACSF at a flow rate of 2.8–3.0 ml·min⁻¹ at a temperature of 23–24°C or 33–34°C. We used four different conditions for recording of the pharmacological effects of drugs efficacious at neuronal GABA_A receptors, which we refer to as conditions a, b, c and d. For condition a, recordings were made at 23–24°C using an intracellular solution (ICS) composed of (in mM) CsCl (135), NaCl (4), MgCl₂ (2), HEPES (10), Ethylene glycol‐bis(2‐aminoethylether)‐N,N,N′,N′‐tetraacetic acid (EGTA, 0.05), QX‐314 (5), tetraethylammonium chloride (TEA, 5), Mg‐ATP (2) and Na₂‐GTP (0.5). The pH of the intracellular solution was adjusted to 7.2 with CsOH and held an osmolarity of 290–295 mOsm. For condition b, recordings were made at 33–34°C using the same intracellular solution. For condition c, recordings were made at 33–34°C, using the same intracellular solution but including 10 μM of the auto‐inhibitory fragment (19–36) of PKC. For these experiments, an aliquot of the kinase inhibitor peptide was added to a prefiltered (0.22 μm) vial of intracellular solution before use. Condition d was performed also at 33–34°C, using intracellular solution of similar composition as in condition b, but in which the concentration of EGTA was increased from 0.05 to 10 mM and CsCl accordingly decreased to maintain osmolarity. Individual neurons were visualized in an IR video‐microscope and recorded using thin‐walled borosilicate pipettes (Sutter, outer/inner diameter 1.5/1.1 mm) with a 6‐ to 8‐MΩ resistance (conditions a–d). To ensure a good diffusion of the intracellular solution into the neuron, we used an 8‐ to 9‐min period between establishing the whole‐cell configuration and beginning recording of baseline activity. Recordings of phasic and compound induced currents consisted of a 3‐min baseline period, followed by a 6‐min drug period (Thio‐THIP, THIP and gabazine) and 3–5 min of bicuculline, gabazine or picrotoxin. Cell capacitance (CP) and series resistance (RS) were measured every 3–4 min throughout the recording and 80% compensated for series resistance. Cells were excluded from analysis if values for capacitance and series resistance deviated more than 30% from initial values determined at baseline. Whole‐cell recordings were made using a Multiclamp 700A patch clamp amplifier controlled by pClamp 9 software and digitized at 10 kHz with a digidata 1322 (all Molecular Devices) and filtered (8‐pole Bessell) at 3 kHz.

2.5. Analysis of patch clamp recordings in dentate gyrus granule cells (DGGCs)

The average holding current value was assessed as the centre value, xc, of a single Gaussian fit of the general expression $\mathrm{A}*{\mathrm{e}}^{-0,5*{\left(\frac{\mathrm{x}-\mathrm{xc}}{\mathrm{w}}\right)}^2}$ to single‐point trace values in a 2‐min period (1 min for final antagonist) sampled at a 100 ‐ms interval. The analysis window for baseline and drug periods ended immediately prior to beginning of bath perfusion of the ensuing drug. The tonic current density (TCD) was calculated as the difference in holding currents divided by the cell capacitance (pA·pF⁻¹) and all tonic current density D values are reported ± SD. Thus, an increase in the GABA_A receptor gating (e.g. by an agonist) results in a negative value for tonic current density. A positive value for tonic current density occurs when the current shuts. The noise was analysed in some cases and was calculated as the centre value of a Gaussian fit to values of SD of 2‐ms periods sampled at 100‐ms intervals. Endogenous tonic and compound induced currents, noise analysis and evoked events were analysed in Clampfit 9 and Origin (OriginLab, 2017). Miniature inhibitory post‐synaptic currents (mIPSCs) were analysed in Synaptosoft (6.07) and Origin as previously detailed (Hoestgaard‐Jensen et al., 2014). Synaptic events included in the average waveform were detected when exceeding a threshold 6* the signal SD (unit pA), determined in the drug period to obviate false detections due to the increased noise during Thio‐THIP application. The noise was also measured as the trace SD. The EC₅₀ value for gabazine was determined as the concentration of half‐maximal effect (x₀) in a fit to a logistic function of the general expression $=\left(\frac{A1-A2}{1+{\left(\frac{x}{x0}\right)}^p}\right)$.

2.6. Whole‐cell patch clamp electrophysiology on recombinant HEK‐293 cells

Whole‐cell patch clamp electrophysiology was performed on recombinant α4β1δ receptors expressed in HEK‐293 Flp‐In cells stably expressing the human GABA_A δ‐subunit (Falk‐Petersen et al., 2017); 48 h before performing, the assay cells were transfected using the polyfect transfections reagent (as described for the FLIPR membrane potential [FMP] assay) with a 1:1 ratio of human α₄ (pUNIV plasmid) and human β₁ (pUNIV plasmid) co‐transfected with GFP (1.4 mg + 1.4 mg + 1.2 mg) for positive selection of transfected cells. The day before, assay cell were seeded on PDL coated coverslips and on the day of the assay transferred to a chamber placed on the stage of an Axiovert 10 microscope (Zeiss, Germany), containing artificial balanced salt solution (ABBS) with the following composition (in mM): NaCl 140, KCl 3.5, Na₂HPO₄ 1.25, MgSO₄ 2, CaCl₂ 2, glucose 10 and HEPES 10; pH 7.35. The temperature was controlled by a TC‐344C temperature controller with a SH‐27B solution in‐line heater (Warner Instruments, USA) to 33–34°C measured in the chamber.

The cells were approached with micropipettes of 2.1‐ to 3.2‐MΩ resistance manufactured from 1.5 mm OD glass (World Precision Instruments, Sarasota, Florida) on a microelectrode puller, model PP‐830 (Narishige, Tokyo, Japan). The intrapipette solution resembled conditions b and d contained the following (in mM): CsCl (130 or 140), NaCl (4), MgCl₂ (2), EGTA (10 or 0.05), MgATP (2), Na₂‐GTP (0.5) and HEPES (10); pH 7.3. Cells were viewed at 200× magnification and cells containing GFP were visualized with UV light from an HBO 50 lamp (Zeiss, Germany). Recordings were made using the standard patch‐clamp techniques in the whole‐cell configuration and in voltage clamp mode, filtered (4‐pole Bessel) at 3 kHz and digitized at 10 kHz using an EPC‐9 amplifier controlled by Pulse Software (HEKA, Germany). A clamping potential of −60 mV was used, series resistance was 80% compensate and compounds were bath applied for 15 s with an application flow of 2 ml·min⁻¹. Data were analysed in IgorPro version 6.22A (HEKA, Germany).

2.7. GABA_A receptor expression and FLIPR™ membrane potential Blue Assay on HEK‐293 cells

The FMP assay was performed on HEK‐293 Flp‐In™ cells expressing the human α₄β₁δ and α₄β₃δ GABA_A receptor as described previously (Falk‐Petersen et al., 2017). In brief, HEK‐293 Flp‐In cells stably expressing the human GABA_A δ‐subunit (HEK‐δ cells) were transfected in a 1:1 ratio (4 μg + 4 μg) of human α₄ (pUNIV plasmid) and human β₁ (pUNIV plasmid) and 2:0.01 ratio (8 μg + 0.04 μg) of α₄ to human β₃ (pcDNA3.1) using Polyfect Transfection Reagent (Qiagen, West Sussex, UK). The latter ratio was optimized in a previous study to limit the formation of β₃ homomers (Falk‐Petersen et al., 2017); 16–24 h after transfection, cells were plated into black poly‐D‐lysine coated clear bottom 96‐well plates (BD Biosciences, Bedford, MA, USA) at 50,000 cells per well. After 16–20 h, the media was aspirated, cells were washed in assay buffer and added 100 μl per well of FMP blue dye (0.5 mg·ml⁻¹) (Molecular Devices) (Sunnyvale, CA, USA) followed by incubation for 30 min in a humidified 5% CO₂ incubator at 37°C. Compound solutions were prepared in assay buffer in 5× which for antagonist testing contained a concentration of GABA corresponding to GABA EC₈₀. Compounds were added to a 96‐well ligand plate and incubated for 15 min at the desired recording temperature (either 25°C or 37°C) in the NOVOstar™ plate reader (BMG, LABTECH GmbH, Offenburg, Germany). The plate was read using an emission wavelength of 560 nm caused by excitation at 530 nm with the response detected as changes in fluorescent signal given in fluorescent units (ΔFU). The obtained data are based on five experiments from separate batches of transfected cells, each with three technical replicates. All potency determinations were based on at least five independent experiments and data from inactive compounds on five independent experiments. Data were analysed as relative changes in the fluorescent signal by taking the maximum compound induced peak/plateau signal and subtracting the baseline signal. Changes in the fluorescent signal caused by compound addition to the wells and artefacts were manually omitted from the analysis. Concentration–response curves were fitted using the four‐parameter concentration–response curve:

$$\mathit{Response}=\mathit{bottom}+\frac{\mathit{top}-\mathit{bottom}}{1+{10}^{\left(\left(\mathit{log}\left({\mathit{EC}}_{50}\right)-\left[A\right]\right)·{\mathit{n}}_{\mathit{H}}\right)}}$$

with bottom being the lower plateau response and top the upper. [A] is the logarithmic concentration of the compound and n _H the hill slope. Antagonists were fitted using the same equation giving IC₅₀ instead of EC₅₀ as for the agonist.

2.8. Western blots

After decapitation, the cortex was rapidly dissected out from at least four β₁ ^+/+ or β₁ ^−/− mice and tissues individually snap‐frozen on dry ice and stored at −80°C until use. Crude membranes were prepared from each sample by quickly thawing the brain tissue at 37°C and adding 5× w/v of 50‐mM Tris HCl, pH 7.4 supplemented with 2,2′,2″,2‴‐(Ethane‐1,2‐diyldinitrilo)tetraacetic acid (EDTA)‐free protease inhibitors (Roche). Tissue was homogenized using 2 × 1 mm zirkonium‐oxide beads using a Bullet Blender (Next Advance, NY, USA) for 30 s at max. speed. The protein concentration was determined using Bradford protein assay and 10 μg of each sample was added sample buffer and reduced using 100‐μM dithiotreitol and incubated for 15 min at 40°C and 15‐s sonication before loading onto a Mini‐Protean TGX gel (4%–20%, BioRad) and run for 40 min at 200 V. Protein gels were blotted onto PVDF membranes using the Trans‐Blot Turbo Transfer System (BioRad). Following blocking in 3% BSA in 1× Tris buffered saline‐Tween20 buffer for 60 min at room temperature (RT), blots were probed overnight at 4°C with anti‐β₁ (gift from the Medical University of Vienna, Center for Brain Research). The antibody was raised against an epitope (350–408) of the mouse β₁ protein and used at a concentration of 0.5 μg·ml⁻¹. Primary antibodies were detected with HRP‐conjugated goat anti‐rabbit antibody (#PI‐1000, RRID:AB_2336198, Vector Laboratories, USA) and visualized by chemiluminescence (Amersham ECL Prime (GE Healthcare Biosciences, Pittsburg, PA, USA) using FluorChem HD2 (Alpha Innotech, San Leandro, CA, USA). Protein levels were visualized using Image Studio Lite Ver 5.2. The immuno‐related procedures used comply with the recommendations made by the British Journal of Pharmacology (Alexander et al., 2018).

2.9. Experimental design and statistical analysis

The data and statistical analysis comply with the recommendations of the British Journal of Pharmacology on experimental design and analysis in pharmacology (Curtis et al., 2018). The experiments were designed to use groups of equal size, with the size depending on the type of experiment based on experience and reproducibility of the assays. In general, a group size of n = 10–15 was used for slice and patch‐clamp experiments and n ≥ 5 for FMP experiments, based on obtaining standard error of mean (SEM) values on the pooled EC/IC₅₀ values < 0.2. The exact group sizes used in the experiments are given in figure legends and tables and stated in the text in Section 3 in connection to the relevant data and analysis. For all group sizes less than 5, data are stated as preliminary with the exact n stated in the text. All given ns represent biological replicates and are the ns used when performing statistical analysis.

Hypothesis testing was performed either using one‐way ANOVA with Bonferroni post hoc test for multiple comparisons, or as Student's t‐tests for either pair‐wise or unpaired comparisons, depending upon the experimental design (specified in the figure legends) all with a threshold for statistical significance of P < 0.05 deemed before performing the analysis. In connection with each experiment, the statistical test employed along with the F value (when applicable), standard error and number of experiments are stated. Statistical tests were only performed for group sizes of n = 5 or more.

Data are stated as mean ± SD except for recombinant experiments in HEK‐293 cells, in which mean potencies are stated as pEC₅₀/pIC₅₀ ± SEM. All concentration–response curves are presented using log‐transformed concentrations on the x‐axis to obtain normal data and sigmoidal curves.

2.10. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018) and are permanently archived in the Concise Guide to PHARMACOLOGY 2019/20 (Alexander et al., 2019).

3. RESULTS

We set out to determine the magnitude of the constitutive and the agonist‐induced current in dentate gyrus granule cell (DGGCs) in the presence of TTX under varying conditions of temperature, free intracellular Ca²⁺ concentration and PKC activity. We utilized four specific conditions, a (24°C and 50‐μM EGTA in the intracellular solution), b (34°C and 50‐μM EGTA in the intracellular solution), c (34°C including 10‐μM PKC 19–36 autoinhibitory peptide and 50‐μM EGTA in the intracellular solution) and d (34°C and 10‐mM EGTA in the intracellular solution) which aresummarized in Figure 1. We found that the magnitude of the endogenous tonic current after blocking GABA_A receptors with 20‐μM bicuculline was significantly larger in recording condition b, compared to conditions a, c and d with no difference between a, c and d (Figure 1a; Table 1). We also assessed the bicuculline sensitive tonic current density in δ^−/− mice by recordings made in conditions b and d (Figure 1c). In recording condition b, a tonic current density mean of 0.7 ± 0.3 was measured in δ^+/+ mice compared to 0.0 ± 0.4 in δ^−/− mice (n = 9/10). In recording condition d, the bicuculline‐induced tonic current density was 0.0 ± 0.4 in δ^+/+ and 0.1 ± 0.2 in δ^−/− mice (n = 10/11), indicating that the receptors responsible for the tonic current density in condition d are dependent on incorporation of the δ‐subunit. Because network levels of extracellular GABA were unlikely to be influenced by the composition of the intracellular solution in a single neuron, it appears plausible that the tonic current density in condition b (Figure 1a, b) is a result of predominantly constitutively active receptors, consistent with the idea that bicuculline is an inverse agonist (Birnir et al., 2000). It is important to stress that the absence of an endogenous tonic current in condition a, c and d in Figure 1a and throughout all experiments is due to the presence of TTX and the absence of exogenously added GABA in the ACSF (see Lee & Maguire, 2014, for review). Generally, for the measurement of the constitutive current, endogenous GABA inducing a tonus is a confounder, which is necessary to minimize.

FIGURE 1.

The magnitude of the GABA_A receptor TCD in adult rat and mouse dentate gyrus granule cells (DGGCs) depend on recording condition and expression of the δ‐subunit. The four recording conditions were a: 24°C and 50‐μM EGTA in the intracellular solution (ICS), b: 34°C and 50‐μM EGTA in the intracellular solution (ICS), c: 34°C including 10‐μM PKC 19–36 autoinhibitory peptide and 50 μM EGTA in the ICS and d: 34°C and 10‐mM EGTA in the ICS. (a) Representative recording traces showing that 20‐μM bicuculline (BIC) blocks IPSCs under all recording conditions but also blocks a tonic current in condition b, which is absent in condition a, c and d. (b) Summary of data from (a); see also Table 1. (c) Summary of TCD in DGGCs recorded in condition b in δ^+/+ and δ^−/− mice (n = 10 and 11, respectively). *P < 0.05, ANOVA/Bonferroni

TABLE 1.

Summary of the effect of bicuculline (BIC), Thio‐THIP, THIP and gabazine (GBZ) on the TCD in adult male rat dentate gyrus granule cells (DGGCs) recorded under four different conditions (a–d)

Temperature and ICS	Condition a	Condition b	Condition c	Condition d
	24°C	34°C	34°C, 50‐μM EGTA	34°C
	50‐μM EGTA	50‐μM EGTA	10‐μM PKC inh.	10‐mM EGTA
Drug	TCD, pA·pF−1 ± SD (no of cells)
20‐μM BIC (Figure 1)	0.0 ± 0.2 (12)	0.7 ± 0.2* (13)	0.1 ± 0.4 (14)	0.1 ± 0.3 (12)
100‐μM Thio‐THIP (Figure 2) + 100‐μM PTX	−1.2 ± 0.4 (12) 0.1 ± 0.4	−0.3 ± 0.3* (14) 0.9 ± 0.3*	−1.1 ± 0.3 (14) 0.1 ± 0.4	−1.2 ± 0.2 (12) −0.0 ± 0.2
2‐μM THIP (Figure 2) + 100‐μM PTX	−4.4 ± 0.7 (8) 0.1 ± 0.3	−2.4 ± 0.7* (8) 0.8 ± 0.4*	−5.5 ± 1.0 (8) −0.1 ± 0.3	−4.9 ± 1.2 (10) 0.1 ± 0.5
10‐μM X (Figure 3) + 100‐μM PTX	−1.0 ± 0.3 (10) 0.0 ± 0.4	−0.2 ± 0.2* (12) 0.9 ± 0.3*	−1.1 ± 0.3 (12) 0.2 ± 0.3	−1.2 ± 0.3 (12) 0.2 ± 0.2

Note: Reported values are the tonic current density (TCD), calculated as the ratio of the difference in holding current between baseline and drug period over cell capacitance. Statistical difference is determined between compound effects in the different conditions by ANOVA followed by Bonferroni mean comparison. ICS, intracellular solution; PTX, picrotoxin.

^* P < 0.05.

To proceed, two cases were considered: If extracellular levels of GABA mediated the tonic current density in condition b, GABA_A receptors could be internalized in conditions a, c and d, thus not in the capacity to respond to the extracellular levels of GABA. If, on the other hand, the tonic current density in condition b was due to constitutive activity, then channels must be either shut or internalized in conditions a, c and d. In either case, the effect of bath‐applied orthosteric agonists should reveal if the recording conditions had created a functional difference in the number of receptors available for activation. We therefore tested the δ‐subunit selective agonists Thio‐THIP (100 μM) and THIP (2 μM) in conditions a through d (Figure 2). In order to compare with effects of gabazine (below), for these and the following experiments, we used picrotoxin rather than bicuculline to block agonist‐induced currents.

FIGURE 2.

The effect of two GABA agonists, THIP (δ‐subunit preferring) and Thio‐THIP (α₄β_1/3δ selective) on the TCD is dependent on temperature, intracellular Ca²⁺‐EGTA chelation and PKC activity. The four conditions a–d were a: 24°C and 50‐μM EGTA in the intracellular solution (ICS), b: 34°C and 50‐μM EGTA in the ICS, c: 34°C including 10‐μM PKC 19–36 autoinhibitory peptide and 50‐μM EGTA in the ICS and d: 34°C and 10‐mM EGTA in the ICS. (a) Full recording traces of voltage clamped (−70 mV) DGGCs upon bath application of 100‐μM Thio‐THIP followed by picrotoxin (PTX). Thio‐THIP induces an ~4× larger TCD in conditions a, c and d than condition b. Note the presence of a positive endogenous TCD in condition b only. (b) Traces of mIPSCs display the Thio‐THIP induced reduction of the average mIPSC in all recording conditions. (c) Summary barplot of TCD values induced by Thio‐THIP (see Table 1 for individual n, *P < 0.05 ANOVA/Bonferroni). (d,e) Summary of peak and decay time of the average mIPSCs (*P < 0.05 paired t‐test). (f,g) Representative traces and barplot showing the effect of THIP in recording conditions a through d (see Table 1 for individual n, *P < 0.05 ANOVA/Bonferroni). (h) The TCD induced by Thio‐THIP in adult δ^+/+ and δ^−/− mice and the effect on the average mIPSC peak in the same recordings (n = 9 and 10, respectively, *P < 0.05, paired Student's t‐test for same‐cell measurements, unpaired for different cells)

These experiments revealed that the recording condition greatly influenced the magnitude of tonic current density induced by δ‐subunit selective agonists in DGGCs. For 100‐μM Thio‐THIP, the tonic current density was significantly less in condition b compared to a, c and d (Figure 2 a, c,f–h, Table 1). Similarly to the effect of bicuculline, picrotoxin administered after Thio‐THIP resulted in a tonic current density near 0 for conditions a, c and d, but 0.9 ± 0.3 for condition b (Table 1). The change in noise (only conditions b and d for Thio‐THIP analysed) was not significant from baseline to Thio‐THIP in recording condition b (mean pA ± SD) 1.61 ± 0.36 and 1.65 ± 0.39, respectively (n = 14), but was significantly increased in condition d from 1.72 ± 0.34 during baseline to 2.2 ± 0.4 in Thio‐THIP (n = 12). We did not detect a difference in baseline noise in recording condition b vs. d. In a different set of experiments (cells not used for measurement of tonic current density), we calculated the reversal potential for the averaged mIPSCs in a 20‐s window/step before and after Thio‐THIP at incremental voltage shits (20 mV/step −70 to +30 mV and found a reversal of −2.2 ± 4.0 in control and 0.2 ± 4.2 mV in 100‐μM Thio‐THIP (not significant, n = 6). The tonic current density observed for THIP (2 μM, Figure 2f, g) displayed a similar recording condition‐dependent effect as Thio‐THIP, being nearly twice as large in conditions a, c and d compared to b (Table 1). Because of the significantly increased response to orthosteric agonists in conditions a, c and d vs. b, we conclude that a net receptor internalization could not be the reason for the absent constitutive current under conditions a, c and d. While the tonic current density induced by δ‐subunit specific orthosteric agonists was highly dependent on recording condition (Figure 2a, c–g), the peak of the averaged non‐contaminated mIPSC was, however, similarly diminished by Thio‐THIP in all conditions (Figure 2b, d, e). For Thio‐THIP (Figure 2d), the peak mIPSC was significantly reduced from (mean pA ± SD) −20.0 ± 6.3 to −15.8 ± 6.2 (condition; n = 12), in condition b from −35.6 ± 8.8 to −30.4 ± 9.0 (n = 14), in condition c from −38.7 ± 7.6 to −34.1 ± 7.6 (n = 14) and in condition d from −34.6 ± 8.7 to −29.1 ± 7.4 (n = 12). The mIPSC peak was significantly different in condition a to conditions b, c and d. The selectivity of Thio‐THIP for δ‐subunit‐containing receptors in mediating effects on tonic current density and mIPSC peak was assessed in δ^−/− mice (Figure 2h). Recording from DGGCs in condition d, Thio‐THIP (100 μM) induced a tonic current density of −0.8 ± 0.2, which significantly reverted to 0.2 ± 0.3 after picrotoxin in δ^+/+ mice (n = 9), whereas the response to Thio‐THIP in δ^−/− mice was −0.1 ± 0.2, which was not changed by picrotoxin (0.2 ± 0.2, n = 10 ). The response to Thio‐THIP was significantly different between the two genotypes. Similarly, the Thio‐THIP‐induced reduction in mIPSC peak, observed in rat DGGCs (Figure 2b, d), was also seen in δ^+/+ mice (−41.4 ± 9.0 pA in baseline vs. −37.4 ± 9.0 pA in Thio‐THIP, n = 9), but absent in δ^−/− mice (−37.1 ± 7.6 in baseline vs. −37.2 ± 7.5 in Thio‐THIP, n = 10).

It has been commonly observed that gabazine cannot close GABA_A receptor channels displaying constitutive activity (McCartney et al., 2007; Wlodarczyk et al., 2013). However, the recognized ability to minimize the constitutive gating through the recording condition now offered a new and untried possibility for probing the antagonist effects of gabazine. If gabazine is a neutral antagonist at constitutively active receptors, it follows that we should expect a tonic current density near zero for gabazine in the conditions under which the constitutive gating is shut. However, in these conditions, gabazine (10 μM), although completely blocking all mIPSCs, also induced a tonic current density, that is, an apparent receptor activation in conditions a, c and d but not in condition b (see Table 1 for n in individual experiments). The block of the mIPSCs by gabazine was complete in <20 s from onset and coincident with onset of the tonic current density, which continued to develop for at least 2 min (Figure 3). A concentration–response relation of gabazine for this effect recorded in condition d gave an EC₅₀ of 2.1 μM (five concentrations in the range 10‐nM to 50‐μM gabazine tested, Figure 3d, one concentration per cell). However, if the constitutive current in condition b was first shut by 20‐μM bicuculline before perfusion of gabazine (10 μM), we found that gabazine now induced a tonic current density of −1.3 ± 0.4, which again could be shut by 100‐μM picrotoxin (Figure 3b, as originally observed by Wlodarczyk et al., 2013). The difference in tonic current density produced by gabazine (10 μM) in condition b + 20‐μM bicuculline and condition c or d (gabazine alone, from Table 1) was not significant.

FIGURE 3.

Gabazine (GBZ) induces a δ‐subunit‐dependent TCD in adult rat and mice dentate gyrus granule cells (DGGCs) under recording conditions phenomenologically characterized by the absence of a constitutive current. The four conditions a–d were a: 24°C and 50‐μM EGTA in the intracellular solution (ICS), b: 34°C and 50‐μM EGTA in the ICS, c: 34°C including 10‐μM PKC 19–36 autoinhibitory peptide and 50‐μM EGTA in the ICS, and d: 34°C and 10‐mM EGTA in the ICS. (a) Recording traces displaying the effect of GBZ and picrotoxin (PTX) in recording conditions a–d. GBZ quickly blocks mIPSCs but also induces a TCD in conditions a, c and d, which is reversed by PTX. (b) Shutting of the constitutive current by bicuculline (BIC; 20 μM) in recording condition b is reversed by GBZ (10 μM). (c,d) Concentration–response relation of GBZ in blocking mIPSCs and inducing a TCD in recording condition d. (e) Summary of effects of GBZ to induce a TCD in recording conditions a–d, incl. recording condition b in which the constitutive current is shut by BIC. (f,g) The δ‐subunit selective positive modulator DS2 potentiates the effect of GBZ in producing a PTX‐sensitive current. (h,i) Recordings of DGGCs from adult δ^+/+ and δ^−/− mice demonstrate that the effect of GBZ inducing a TCD is dependent on expression of the δ‐subunit. For panel (i), n = 7 for each genotype *P < 0.05, Student's t‐test ANOVA/Bonferroni

To assess if this apparent agonistic effect of gabazine was dependent on GABA_A receptors incorporating the δ‐subunit, we first tested gabazine (1 μM) in the presence of the δ‐subunit positive modulator DS2 (1 μM). In recording condition d, bath application of DS2 (1 μM) and subsequent (3 min later) administration of gabazine (1 μM) increased tonic current density to −1.1 ± 0.3 (n = 8). This was significantly larger than the tonic current density for application of 1‐μM DS2 (−0.2 ± 0.1, n = 5) or 1‐μM gabazine (−0.3 ± 0.5, n = 8) alone. Further, the effect of gabazine (10 μM) recorded under condition d in δ^−/− mice was 0.0 ± 0.1, but −0.9 ± 0.2 in δ^+/+ mice (Figure 3i), indicating that the effects of gabazine in producing a tonic current density was indeed dependent on the presence of the δ‐subunit. Finally, we assessed whether the effect of gabazine in producing a tonic current density could be mediated through homomeric GABA receptor ρ1 subunit and tested 10‐μM TPMPA in condition d, which was ineffective in blocking the effects of gabazine. However, despite these clear effects of gabazine in inducing a current in receptors incorporating the δ‐subunit with at least partial contribution from the β₁‐subunit (below, Figure 5), we could not demonstrate such effects of gabazine in HEK‐293 cells expressing human recombinant α₄β_1/3δ receptors (Figure 4). Using a recently characterized assay employing voltage‐sensitive fluorescence emitting dyes (Falk‐Petersen et al., 2017), the effect of GABA, gabazine and bicuculline was characterized at human α₄β₁δ receptors at 37°C and 25°C. At 37°C, GABA displayed an EC₅₀ value of 0.28 μM (pEC₅₀ = 6.56 ± 0.10, n = 5) (Figure 4a) and gabazine displayed an IC₅₀ value of 0.34 μM (pIC₅₀ = 6.46 ± 0.14, n = 5) against a concentration of GABA EC₈₀ (Figure 4b). Similarly, at human α₄β₃δ receptors, GABA displayed an EC₅₀ value of 0.36 μM (Figure 4d), and gabazine displayed an IC₅₀ of 0.44 μM (Figure 4e). The constitutive activity at both 37°C and 22°C was negligible and we did not observe any current induced by gabazine at either receptor subtype (Figure 4c, f). We also tested if spontaneous currents were observed in α₄β₁δ expressing HEK‐293 cells recorded in regular whole‐cell patch clamp electrophysiology at 34°C using intracellular solutions as used for condition b (n = 10) and d (n = 12) in slice experiments (Figure 4g, h). GABA (100 μM) was applied for 15 s followed by 100 μM picrotoxin for 15 s. In each of these conditions, the noise at baseline was not significantly different from the noise in picrotoxin (for 50‐μM EGTA intracellular solution, BL noise was 12.7 ± 5.4 vs. 12.8 ± 3.8 pA in PTX, for 10‐mM EGTA intracellular solution, BL noise was 13.8 ± 4.9 vs. 13.1 ± 4.8 pA in PTX).

FIGURE 5.

Gabazine (GBZ) does not antagonize the effect of Thio‐THIP. Recording conditions used here were b: 34°C and 50‐μM EGTA in the intracellular solution (ICS) and d: 34°C and 10‐mM EGTA in the intracellular solution (ICS). (a) Thio‐THIP, GBZ and picrotoxin (PTX) on DGGCs in recording condition b in adult rats (n = 8). (b) Same experiment as (a), recording condition d. Note that 10‐μM GBZ neither blocks nor increases the TCD already induced by Thio‐THIP (right panel, n = 13). (c,d) Thio‐THIP and GBZ recorded in condition d in adult β₁ ^+/+ and β₁ ^−/− mice (n = 18 and 22, respectively). (e) The noise (SD) induced by Thio‐THIP and GBZ in β₁ ^+/+ and β₁ ^−/− mice. The noise in Thio‐THIP + GBZ is significantly lower in β₁ ^−/− than β₁ ^+/+ mice. (f,g) The average mIPSC peak is similarly reduced by Thio‐THIP in β₁ ^+/+ and β₁ ^−/− mice. (h) Left: Representative example of PCR genotyping of samples from β₁ ^+/+, β₁ ^+/− and β₁ ^−/− mice. Right: Representative preliminary Western blot of GABA_A β₁‐subunit in cortical brain homogenate from β₁ ^+/+ and β₁ ^−/− mice showing absence of β₁‐subunit immunoreactivity in β₁ ^−/− mice (β₁ ^−/−, n = 4, β₁ ^+/+, n = 5)

FIGURE 4.

Gabazine (GBZ) antagonizes the effect of GABA but does not induce a current alone in constitutively silent α₄β₁δ and α₄β₃δ receptors expressed in HEK‐293 Flp‐In cells. (a,d) GABA (■) activates human α₄β₁δ (a) and α₄β₃δ (d) receptors with a similar EC₅₀s. (b,e) GBZ (▼) concentration‐dependently blocks the response to an EC₈₀ concentration of GABA with IC₅₀ = 0.34 μM at α₄β₁δ receptors (b) and 0.44 μM at α₄β₃δ receptors (e). (c,f) GBZ does not induce a current alone in any receptor type. All datapoints in graphs (a–f) are means of triplicate measurements ± SD. FU, fluorescence unit. Symbols below both figures are the same in the three vertical panels. (g) Trace from a HEK‐293 cell expressing α₄β₁δ receptors and recorded from in patch clamp mode using an intracellular concentration of 50‐μM EGTA, showing effect of GABA (100 μM) and PTX. (h) The noise values for baseline and PTX period in two different intracellular solutions (containing either 50‐μM or 10‐mM EGTA) which was not significantly different (Student's t‐test), indicating no constitutive current

Because the effects of gabazine and Thio‐THIP were both dependent on the presence of the δ‐subunit and both ineffective at producing a significant tonic current density when the constitutive gating was present (condition b, Figure 2 and Table 1), we hypothesized that their effects were mediated by the same receptors. We tested in conditions b and d if gabazine (1 and 10 μM) could affect the tonic current density induced by Thio‐THIP (100 μM). In condition b, Thio‐THIP induced a tonic current density of −0.3 ± 0.3 which was not changed by 10‐μM gabazine (Figure 5a). In condition d, the tonic current density induced by Thio‐THIP (−1.2 ± 0.2, n = 13, Figure 5b) also was not significantly changed by 1‐μM gabazine or 10‐μM gabazine.

Because the effects of gabazine (at 10 μM, maximum effect, Figure 5b) and Thio‐THIP were not additive, this suggested that their effects could be mediated through the same (saturable) mechanism. Thio‐THIP displays a selectivity for the α₄β_1/3δ receptor subtype, sparing other hippocampal δ‐subunit partnerships. To assess if a β‐subunit could be involved in this response, we repeated the Thio‐THIP–gabazine interaction study in mice ablated of the β₁‐subunit (Figure 5h) and made recordings in condition d (Figure 5c–g). In the β₁ ^+/+ mice (n = 18), Thio‐THIP induced a tonic current density of −0.9 ± 0.3 compared to −0.7 ± 0.4 in β₁ ^−/− mice. However, the ensuing effect of gabazine (10 μM) produced different effects on the tonic current density between genotypes. In β₁ ^+/+ mice, gabazine added after Thio‐THIP produced a tonic current density of −0.9 ± 0.4 (Thio‐THIP vs. Thio‐THIP + gabazine, non‐significant, Student's paired t‐test, n = 18) but significantly reduced the tonic current density in β₁ ^−/− mice to −0.4 ± 0.4 (Thio‐THIP vs. Thio‐THIP + gabazine, n = 22). The noise (SD) during baseline and during Thio‐THIP application was independent on genotype (Figure 5e), whereas gabazine significantly reduced noise in β₁ ^−/− to a level below that of the β₁ ^+/+ (gabazine + Thio‐THIP in β₁ ^+/+ produced a noise of 1.37 ± 0.26 pA but 1.18 ± 0.29 pA in β₁ ^−/− mice (n = 18 β₁ ^+/+ and 22 β₁ ^−/−). An additional qualitative observation was that the onset of the noise by Thio‐THIP, or the shutting of it by gabazine, was fast (see Figure 5a, b, g), reaching a plateau within tens of seconds after onset, whereas the tonic current density by either Thio‐THIP or gabazine continued to develop for typically 2 min after onset. These data obtained in β₁ ^−/− mice infer that the gabazine response involves, at least in part, β₁‐containing receptor subtypes—in addition to δ‐subunit‐containing receptors, possibly in a common subtype, for example, α₄β₁δ.

4. DISCUSSION

In the present study, we show that the magnitude of the constitutive and the agonist‐induced current passed by α₄β_1/3δ GABA_A receptors in adult rodent dentate gyrus granule cells (DGGCs) is determined by temperature, free intracellular Ca²⁺ levels and PKC activity. By comparing effects of the orthosteric agonists THIP and Thio‐THIP, the orthosteric antagonists bicuculline and gabazine, and the channel blocker picrotoxin in four different recording conditions in adult rats, δ^+/+, δ^−/−, β₁ ^+/+ and β₁ ^−/− mice, we demonstrate the existence of an inverse relationship of the constitutive versus the agonist‐induced current in DGGCs. Further, we show that only under the conditions of a shut constitutive channel does gabazine induce a picrotoxin‐sensitive current. This current is completely dependent on the expression and incorporation of the δ‐ and to some degree also the β₁‐subunit in functional GABA receptors, elucidated with the first in vitro phenotypical description of a β₁ ^−/− mouse strain. However, this effect of gabazine could not be replicated in human α₄β_1/3δ receptors expressed in HEK‐293 cells, suggesting a component missing in HEK cells, compared to DGGCs. Our data suggest that in vivo efficacy of δ‐subunit preferring agonists (i.e. THIP) in DGGCs is diminished in conditions of increased PKC activity.

4.1. Constitutive activity of GABA_A receptors

The four different recording conditions (a: 24°C and low EGTA intracellular solution, b: 34°C and low EGTA intracellular solution, c: 34°C low EGTA including PKC inhibitor intracellular solution, and d: 34°C and high EGTA intracellular solution) were chosen because of previous findings that implicate temperature and kinase activity as regulators of δ‐subunit‐containing GABA_A receptors (Bright et al., 2011; Bright & Smart, 2013; Houston, Lee, Hosie, Moss, & Smart, 2007; Wei et al., 2003). The results obtained from recording condition b (Table 1) compared to condition c or d suggest that the endogenous tonic current density is mediated by high intracellular kinase activity most likely mediating specific phosphorylation of receptor subunits (discussed below). Recordings at 24°C gave similar results as kinase inhibition at 34°C, which we suggest is a consequence of low kinase activity at 24°C, because PKC in this case does not readily translocate to the plasma membrane in mammalian cells (Chemin et al., 2007). Since the responses to the orthosteric agonists Thio‐THIP and THIP were increased in conditions a, c and d relative to b, (Figure 2 and Table 1, see below), the same conditions for which the tonic current density was absent (a, c and d, Figure 1; Table 1), two conclusions are made. First, the absence of an endogenous tonic current density in conditions a, c and d is not due to internalization of δ‐subunit‐containing receptors and second, the tonic current density in condition b must be the result of constitutively active receptors that can be shut by PKC inhibition, bicuculline and picrotoxin (Figures 1 and 2; Table 1) but not by gabazine (see below). The constitutively active receptors recorded in condition b were also detected in δ^+/+ but not in δ^−/− mice, identifying the δ‐subunit as necessary and sufficient for the measured constitutive current.

4.2. Response to THIP and Thio‐THIP in different recording conditions

Results obtained with THIP and Thio‐THIP were qualitatively similar. However, the ratio between tonic current density values obtained in conditions c and b (i.e. tonic current density(c)/tonic current density(b) ~ 2 for THIP but ~4 for Thio‐THIP, calculated from the tonic current density values in Table 1) suggests that PKC inhibition or chelation of intracellular Ca²⁺ affect receptors that are targets for Thio‐THIP more selectively. We shall therefore focus this discussion on α₄β_1/3δ receptors due to the selectivity of Thio‐THIP for these (Hoestgaard‐Jensen et al., 2014). Since the α₄β_1/3δ receptors, but not α₄β₂δ receptors, exhibit constitutive gating in oocytes and upon PKA‐activation in HEK cells (Hoestgaard‐Jensen et al., 2014; Jensen et al., 2013; Tang et al., 2010), these subtypes appear plausible candidates or the constitutive gating in DGGCs. Studies examining the co‐existence of a PKA‐activated constitutive current and a GABA‐activated current in α₄β₃δ GABA_A receptors in HEK cells demonstrated that the constitutive current sets the baseline floor, under which sufficiently low concentrations of agonist are ineffective in increasing channel open probability (Tang et al., 2010). Our data in DGGCs are in support of this, since the effect of Thio‐THIP on the tonic current density is minimal in condition b (high constitutive current), but ~4 times more effective in conditions a, c and d (no constitutive current, Figures 1a and 2a).

The effect of a high concentration of EGTA or inhibitor of PKC in the intracellular solution on the endogenous tonic current density and response to agonists suggests an already high level of endogenous kinase activity and receptor phosphorylation at 34°C in the low concentration of EGTA in the intracellular solution (see also Brandon et al., 2000; Bright et al., 2011; Bright & Smart, 2013). It is therefore interesting to note that the effect of Thio‐THIP on the synaptic response was independent on the recording condition (Figure 2b), suggesting that Thio‐THIP‐responding receptors close to the synapse are not similarly sensitive to kinase activity as extrasynaptic receptors. GABA_A receptors interact with, among others, gephyrin, GABA receptor associated protein, calreticulin (gC1q‐R) and AKAP79/150 through a ~30 AA long domain in the intracellular TM3‐TM4 loop (Brandon et al., 2003; Essrich, Lorez, Benson, Fritschy, & Luscher, 1998; Schaerer, Kannenberg, Hunziker, Baumann, & Sigel, 2001; Wang, Bedford, Brandon, Moss, & Olsen, 1999). Receptor anchoring at the synapse requires gephyrin interaction and the binding domain for gephyrin has been mapped in the β‐subunits and shown to overlap the S383/S409 (CaMK2 site) and the S408/409/410 (PKA and PKC site) (Kowalczyk et al., 2013; Maric et al., 2017; Maric, Mukherjee, Tretter, Moss, & Schindelin, 2011). Trafficking of GABA_A and glycine receptors is sensitive to calcineurin and PKC activity due in part to phosphorylation of gephyrin (Bannai et al., 2009; Specht et al., 2011; Tyagarajan et al., 2011; Zacchi, Antonelli, & Cherubini, 2014). However, phosphorylation of receptor subunits is likely to also play a role in anchoring to gephyrin, as phosphorylation of the S409 residue in a GABA_A receptor subunit β₁ fragment also reduces the affinity to gephyrin fragments (Maric, personal communication).

4.3. Gabazine and the constitutive current in rodent dentate gyrus granule cells (DGGCs)

The two orthosteric GABA_A receptor antagonists, gabazine and bicuculline, exert different effects against spontaneously active GABA_A receptors:‐ bicuculline shuts spontaneously active receptors (Birnir et al., 2000) and is termed an inverse agonist whereas gabazine is ineffective towards these (McCartney et al., 2007; Wlodarczyk et al., 2013) and is termed a neutral antagonist. The constitutive current in recording condition b can be closed by bicuculline and possibly displaced by gabazine for receptors to resume constitutive gating (see McCartney et al., 2007; Wlodarczyk et al., 2013 and Figure 3b). However, since gabazine is capable of introducing this current by itself, but only under conditions of shut constitutive channels, we conclude that the tonic current density is the result of an action by gabazine at a receptor associated with a shut channel and not bicuculline displacement. Further, since 20 μM of bicuculline did not in the least affect the magnitude of current induced by gabazine (Figure 3b), this would argue against that bicuculline compete with gabazine to reverse this effect. Similar to the effect of Thio‐THIP in recording condition b described above, the effect of gabazine was hidden in the already active constitutive current. This effect of gabazine in DGGCs was picrotoxin‐ and DS2‐sensitive, completely δ‐subunit dependent, TPMPA‐insensitive, partially β₁‐dependent and displayed an EC₅₀ of 2.1 μM. We have summarized the essence of this relation between constitutive current and agonist induced current in Figure 6. Since the mechanism of PKC mediated GABA_A receptor desensitization both generate spontaneous current, but also limit agonist mediated current, it is not clear how the net excitability is affected. However, in the condition of desensitized extrasynaptic receptors, the effect of GABA from inhibition of uptake or release from neurogliaform interneurons is likely diminished. However, we were unable to demonstrate that α₄β_1/3δ receptors expressed in HEK‐293 cells display a gabazine‐induced current when measured by the FMP assay (Figure 4a–f). Nor were we able to see spontaneous activity in recombinant α₄β₁δ receptors when recorded from in conventional patch clamp mode in similar conditions to condition b (Figure 4g, h). It is possible that intracellular components acting through the TM3‐TM4 loop at β‐subunits mediating the effects of gabazine in DGGCs are missing in the HEK‐293 cells, as argued by the relative insensitivity of GABA_A receptors to CaMK2 modulation in HEK‐293 cells (Houston & Smart, 2006). Since gabazine had no effect on the tonic current density induced by Thio‐THIP in β₁ ^+/+ mice, but reduced the tonic current density by Thio‐THIP in β₁ ^−/− mice (Figure 5c, d), we conclude that, in addition to the δ‐subunits, the β₁‐subunit‐containing receptors are also involved in the response by gabazine, but not exclusively so. We hypothesize that phosphorylation of the receptor induces a state of the receptor with a high degree of spontaneous activity. Gabazine can, in the absence of phosphorylation, induce a similar state of high spontaneous activity. We believe that some of the effects of gabazine alone on DGGCs have been observed by others before us, but not subjected to methodological analysis. This may in part be because the connection to the constitutive gating was not assessed. However, specific methodologies used in the field, such as recordings at different temperatures, adding GABA to the ACSF, including or excluding TTX and using intracellular solutions of different Ca‐chelating capacity here are confounders which can be obviated to determine the effect of gabazine on α₄β_1/3δ in DGGCs. Furthermore, we suggest that the relationship between PKC and desensitized receptors implies that the use PKC activators or inhibitors in the presence of GABA and desensitized GABA_A receptors can lead to ambiguous interpretation of data.

FIGURE 6.

A graphical illustration of the relationship between the agonist‐induced and the spontaneous current in dentate gyrus granule cells (DGGCs). Central for the model is the observation that the agonist‐induced (by Thio‐THIP) and the spontaneous current, although mediated by same receptors, are not additive but better described as complimentary. The maximum current gated by a receptor is fixed, irrespective of it being due to spontaneous or ligand‐gating, but the baseline floor is set by the constitutive current which is determined by the endogenous kinase activity towards GABA_A receptors: the larger the constitutive current, the less room for agonist induced current. The physiological relevance of this relation may become apparent when considering diseases involving a much increased PKC activity in distinct neurons (e.g. epilepsy or addiction including alcoholism) against which certain GABAergic compounds are surprisingly ineffective. The model is based on data of the present manuscript and those presented by Tang et al. (2010)

In conclusion, we have described the existence of kinase‐dependent behaviour of δ‐subunit‐containing GABA_A receptors that disguises the efficacy of orthosteric agonists and gabazine. This is likely to concern the effects of orthosteric activation of δ‐subunit‐containing GABA_A receptors in specific pathologies such as epilepsy and addiction including alcoholism (Anstee et al., 2013; Garcia‐Pardo, Roger‐Sanchez, Rodriguez‐Arias, Minarro, & Aguilar, 2016; Kia et al., 2011).

AUTHOR CONTRIBUTION

N.O.D. conceived the study together with P.W., conducted and analysed all slice electrophysiology experiments and was main responsible for writing the manuscript. P.W. conceptualized the studies involving the β₁ knockout mouse strain. C.B.F.‐P. performed and analysed whole‐cell patch‐clamp and FMP experiments on GABA receptors expressed in HEK‐293 cells. U.L. performed Western blots and genotyping of transgenic mice together with P.W. P.S. supplied GABA receptor subtype specific antibodies. J.K. and B.F. synthesized THIP and Thio‐THIP. P.W. co‐wrote the manuscript together with N.O.D. All authors contributed by correcting and proof‐reading the manuscript.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

DECLARATION OF TRANSPARENCY AND SCIENTIFIC RIGOUR

This declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research as stated in the BJP guidelines for Natural Product Research, Design and Analysis, Immunoblotting and Immunochemistry, and Animal Experimentation, and as recommended by funding agencies, publishers and other organizations engaged with supporting research.

Dalby NO, Falk‐Petersen CB, Leurs U, et al. Silencing of spontaneous activity at α4β1/3δ GABA_A receptors in hippocampal granule cells reveals different ligand pharmacology. Br J Pharmacol. 2020;177:3975–3990. 10.1111/bph.15146