δ-subunit containing GABA_A IPSCs are driven by both synaptic and diffusional GABA in mouse dentate granule neurons

Abstract

GABA_A receptors mediate transmission throughout the central nervous system and typically contain a δ subunit (δ receptors) or a γ2 subunit (γ2 receptors). δ IPSCs decay slower than γ2 IPSCs, but the reasons are unclear. Transmitter diffusion, rebinding, or slow deactivation kinetics of channels are candidates. We used gene editing to confer picrotoxin resistance on δ receptors in mice, then pharmacologically isolated δ receptors in mouse dentate granule cells to explore IPSCs. γ2 and δ components of IPSCs were modulated similarly by presynaptic manipulations and manipulations of transmitter lifetime, suggesting that GABA release recruits δ receptors proportionally to γ2 receptors. δ IPSCs showed more sensitivity to altered transmitter release and to a rapidly dissociating antagonist, suggesting an additional spillover contribution. Reducing GABA diffusion with 5% dextran increased the peak amplitude and decreased the decay of evoked δ IPSCs but had no effect on δ or dual-component (mainly γ2-driven) spontaneous IPSCs, suggesting that GABA actions can be local for both receptor types. Rapid application of varied [GABA] onto nucleated patches from dentate granule cells demonstrated a deactivation rate of δ receptors similar to that of δ spontaneous IPSCs, consistent with the idea that deactivation and local GABA actions drive δ spontaneous IPSCs. Overall, our results indicate that δ IPSCs are activated by both synaptic and diffusional GABA. Our results are consistent with a functional relationship between δ and γ2 GABA_A receptors akin to that of slow NMDA and fast AMPA EPSCs at glutamate synapses.

Introduction

GABA_ARs are the major inhibitory neurotransmitter receptors and play an essential role in maintaining the excitatory-inhibitory balance in the brain (Sieghart, 2006; Haider et al., 2006; Mann & Paulsen, 2007; Mann & Mody, 2010). Failure to maintain the balance has been implicated in the pathogenesis of some neurological and neuropsychiatric diseases, including epilepsy, autism, and depression (Eichler, 2008; Fee et al., 2017; Selten et al., 2018). Two distinct populations of GABA_ARs are those containing γ2 and δ subunits, respectively (Jacob et al., 2008; Hörtnagl et al., 2013), referred to here as γ2 receptors and δ receptors. Together these make up the majority of all native GABA_A receptors. Because of their perisynaptic or extrasynaptic localization (Wei et al., 2003), δ receptors are thought to mainly participate in tonic inhibition and a slow, diffusional component of IPSCs (Wei et al., 2003; Brickley & Mody, 2012; Herd et al., 2013; Ye et al., 2013). γ2 receptors are mainly involved in phasic inhibition (Wei et al., 2003) through local, discrete actions on receptors clustered at synapses (Nusser et al., 1998; Sun et al., 2004; Martenson et al., 2017).

The prevailing picture of δ diffusional IPSCs comes mainly from receptor localization studies and genetic deletions (Wei et al., 2003), which do not permit functional comparisons of δ receptor behavior and γ2 receptor behavior in the same cells under the same conditions. We revisited the prevailing view with a chemogenetic approach, which allows pharmacological isolation of δ receptors to investigate their synaptic contribution (Sun et al., 2018). Pharmacologic separation has been critical to our understanding of other dual-receptor signaling, including NMDAR/AMPAR synapses and glycine/GABA synapses (Verdoorn et al., 1990; Lester et al., 1990; Hestrin et al., 1990; Bowery & Smart, 2006; Moore & Trussell, 2017).

If the perisynaptic/extrasynaptic location of δ receptors functionally influences IPSCs compared with γ2 IPSCs, δ IPSCs should be more sensitive than γ2 IPSCs to manipulations of presynaptic output and to manipulations of transmitter lifetime (Wei et al., 2003; Brickley & Mody, 2012; Ye et al., 2013). In contrast, we find limited evidence for a stronger diffusional component of δ IPSCs. Rather, δ and γ2 IPSCs are modulated largely in parallel, and show comparable sensitivity to manipulations of transmitter lifetime, although δ IPSCs exhibit modestly more diffusional influence. We provide evidence that the deactivation kinetics of δ receptors, rather than diffusion, is rate limiting for spontaneous IPSCs (sIPSCs) in dentate granule neurons. This suggests that δ receptors experience transient, synaptic-like pulses of GABA despite their extrasynaptic localization. Overall, our results suggest that δ receptors are recruited by synaptic GABA largely in parallel with γ2 receptors and mediate slow inhibition mostly as a result of their slow deactivation kinetics.

Methods

Ethical Approvals

All animal procedures were performed according to NIH guidelines and approved by the Washington University Institutional Animal Care and Use Committee, protocol 20180208 and under ethical standards outlined previously (Grundy, 2015). Animal pain and suffering was alleviated with 5% isoflurane administration through inhalation in an anesthetic jar under an approved fume hood. Deep anesthesia was monitored by foot and tail pinch, and animals were then decapitated for tissue harvest. Animals were derived in house as described below and were reared under the care of the Washington University School of Medicine Division of Comparative Medicine. Animals had ad libitum access to food and water throughout. Mice were euthanized at the end of studies according to NIH guidelines for minimizing pain. Other aspects of animal studies are described in the sections below.

Animals

δ* and γ2* knock-in (KI) mice were generated using CRISPR/Cas9-based genome editing, as described previously (Sun et al., 2018). Briefly, Cas9 mRNA, sgRNA, and ssODN constructs were injected into the cytoplasm and pronucleus of fertilized eggs. Injected eggs were cultured at 37°C under 5% CO₂ overnight, after which 20–25 two-cell stage embryos were transferred into oviducts of pseudopregnant females. The mutant allele was confirmed by PCR based genotyping from tail DNA. Mosaic founders were bred with WT mice, followed by screening and identification of F1 mice with the desired mutations. We established and maintained colonies with heterozygote crosses. Littermate WT or age/sex-matched WT animals from the colony served as controls (Sun et al., 2018). Mice were maintained on a mixed C57BL/6CBA background.

Slice preparation

Hippocampal slices were prepared from postnatal day 22 (P22) to P52 GABA_AR δ* KI or WT littermates of both sexes. The age range was in accord with previous studies of rodent dentate neurons (Wei et al., 2003; Stell et al., 2003; Glykys et al., 2008; Tao et al., 2013). The ages avoided developmental changes in IPSC kinetics, which stabilize after P21 (Hollrigel & Soltesz, 1997). We used mice of both sexes as in our previous work (Sun et al., 2018). Numbers of males and females were approximately balanced in each experiment.

In accordance with protocols approved by the Washington University IACUC, mice were anesthetized deeply with 5% isoflurane and decapitated. The brain was removed and glued onto a Leica VT1200 specimen holder. Sagittal (300 μm) slices were cut in ice-cold, modified artificial cerebrospinal fluid (ACSF) (in mM: 87 NaCl, 75 sucrose, 25 glucose, 25 NaHCO₃, 2.5 KCl, 1.25 NaH₂PO₄, equilibrated with 95% oxygen - 5% CO₂ plus 0.5 CaCl₂, 3 MgCl₂; 320 mosmol). Slices were then incubated at 32–34°C for 30 min in choline-based ACSF (in mM: 92 choline chloride, 25 glucose, 30 NaHCO₃, 2.5 KCl, 1.2 NaH₂PO₄, 20 HEPES, 2 thiourea, 5 Na ascorbate, 3 Na pyruvate, 2 CaCl₂ and 1 MgCl₂, equilibrated with 95% oxygen-5% CO₂; 300 mosmol), and subsequently stored at room temperature in regular ACSF (in mM: 125 NaCl, 25 glucose, 25 NaHCO₃, 2.5 KCl, 1.25 NaH₂PO₄, equilibrated with 95% oxygen-5% CO₂ plus 2.6 CaCl₂, 1.2 MgCl₂; 310 mosmol), allowing for at least one hour recovery prior to experiments. Except for noted exceptions, drugs were obtained from Sigma (St. Louis, MO).

Whole-cell patch-clamp recording in slices

Slices were transferred to a recording chamber and continuously perfused with oxygenated, regular ACSF at 2 ml min⁻¹. Experiments were performed at 30–32° C. To isolate GABA_A responses, ionotropic glutamate receptors were blocked with 10 μM NBQX (Tocris, Bristol, United Kingdom) and 50 μM D-APV (Tocris) for all experiments unless otherwise stated. To isolate AMPAR responses, GABA_A receptors were blocked with 100 μM gabazine. In some experiments evoked γ2 IPSCs were digitally isolated from δ* slices by applying 100 μM PTX and digitally subtracting the remaining δ IPSC from the baseline IPSC. For experiments altering [Ca²⁺]_o, ACSF containing 1.3 mM CaCl₂ and 2.5 mM MgCl₂ was used. For experiments assessing the impact of TPMPA on IPSC parameters, 200 μM TPMPA (Tocris, Bristol, United Kingdom) was bath-applied. For experiments assessing the impact of dextran (5%) on IPSC parameters, 1 μM CGP-55845 (Tocris) was also included in extracellular solutions to inhibit GABA_BRs.100 μM picrotoxin (PTX; Tocris, Bristol, United Kingdom) was bath-applied to isolate δ* KI-mediated GABA_AR currents, and in some experiments 100 μM gabazine was applied at the end to verify pure GABA_AR currents.

Whole-cell patch-clamp recordings from neuronal somas in the granule-cell layer of the dentate gyrus were performed using differential contrast interference microscopy under infrared illumination. Dentate granule cells were identified on an upright Nikon Eclipse E600FN microscope using a QImaging camera controlled with QCapture (QImaging, Surrey, Canada). Recordings were made with borosilicate patch pipettes (World Precision Instruments, Sarasota, FL; Sutter Instruments, Novato, CA), having open tip resistance of 3–7 MΩ. After a whole-cell configuration was established, cells were allowed to fill with the intracellular solution for ~5 min before recording. Recordings were obtained using a MultiClamp 700B amplifier (Molecular Devices; Sunnyvale, CA), Digidata1550 16-bit A/D converter, and pClamp 10.4 software (Molecular Devices). Pipette capacitance was adjusted using MultiClamp 700B Commander software. Somatic access resistance values were between 10–25 MΩ and were compensated for current amplitudes >2 nA (Figure 1), and cells with unstable access resistance (> 20% change) were excluded from analysis.

Figure 1. Both δ and γ2 IPSC decays are limited by transporters.

A. Representative traces of IPSCs evoked at low vs high stimulus intensity from δ* KI slices before (black) and after (red) PTX application in NO-711. B. The percent contribution of PTX-resistant δ receptors to IPSC peak amplitude (top graph) or charge transfer (bottom graph) at low vs high-level stimulation in NO-711. C-D. Decay weighted time constant of δ and γ2-mediated IPSCs evoked at low vs high-level stimulation, before and after NO-711 application. E-F. Latency to peak of δ and γ2-mediated IPSCs evoked at low vs high-level stimulation, before and after NO-711 application. (N = 7 cells from 2 δ* KI animals for no NO-711 group, and 8 cells from 3 δ* KI animals for + NO-711 group. Comparisons were made by one-way ANOVA, followed by Bonferroni’s multiple-comparisons test. In this and subsequent figures, numbers associated with bars give sample numbers for clarity.

Measurement of phasic GABA_AR currents

Cells were voltage-clamped at −70 mV, with the intracellular pipette solution containing (in mM): 130 CsCl, 10 HEPES, 5 EGTA, 2 MgATP, 0.5 Na₂GTP, and 4 QX-314; pH was adjusted with CsOH to pH 7.3; 290 mosomol. For assessing evoked GABA_AR IPSCs, a glass monopolar stimulating electrode was placed in the molecular layer, ~50 μm away from the recorded cell. Stimulation was applied at 0.05 Hz, with 0.2 ms pulse width. For measuring GABA_AR sIPSCs or spontaneous AMPAR EPSCs (sEPSCs), data were acquired in gap-free mode. All sIPSCs and sEPSCs were acquired at 5 kHz, filtered at 2 kHz using an 8-pole Bessel filter. For quantification of δ contribution to IPSCs, we calculated the effect of PTX on peak evoked IPSC amplitude in δ* KI slices.

Measurement of phasic NMDAR currents

Cells were voltage-clamped at −30 mV, with the intracellular pipette solution containing (in mM): 120 cesium methanesulfonate, 20 HEPES, 10 EGTA, 2 MgATP, 0.3 Na₂GTP, and 5 QX-314 (pH adjusted with CsOH to pH 7.25; 290 mosM). Evoked NMDAR EPSCs were measured in the presence of 10 μM NBQX (Tocris, Bristol, United Kingdom) and 100 μM PTX (Tocris, Bristol, United Kingdom).

Nucleated outside-out patches

To isolate nucleated patches, after a somatic whole cell recording from dentate granule neurons in slices was established, negative pressure (70–200 mbar) was applied and the patch pipette was withdrawn slowly. A small negative pressure (10–30 mbar) was maintained during the recording. The resulting nucleated patch was verified visually as a sphere at the pipette tip. Nucleated patches were held at −70 mV throughout the experiment.

GABA and glutamate iontophoresis

Iontophoresis was performed using an Axoclamp-2B amplifier (Molecular Devices). The iontophoretic pipette, with 50–100 MΩ tip resistance, was filled with 1 M GABA solution (pH was adjusted to ~5 with HCl) or 150 mM L-glutamate (pH was adjusted to ~7.4 with NaOH). Before penetrating slice tissues, the iontophoretic pipette capacitance was compensated using a 10 nA, 10 ms test pulse. The iontophoretic pipette was then brought close to the nucleated patch (< 1 μm) without touching it. To prevent GABA or glutamate from leaking out of the pipette, a retaining current (−10 to −40 nA for GABA; 10–40 nA for glutamate) was applied until no change of the holding current in the nucleated patch recording was observed as the iontophoretic pipette approached the targeted nucleated patch. To evoke iontophoretic GABA IPSCs, 1 ms current pulses ranging from 50 to 550 nA were applied to generate an input/output (I/O) curve. To evoke iontophoretic glutamate EPSCs, 1 ms current pulses ranging from −50 to −550 nA were applied to generate a range of receptor responses from low to high agonist concentration.

Theta-tube GABA application

Fast application (~1 ms exchange times) of GABA (100 μM) solution was applied to a nucleated patch, achieved by rapidly switching between two solution streams from a theta-pipette attached to a piezo-electric actuator. Junction currents obtained following patch blow-out are shown above traces to indicate drug delivery and removal speed.

Statistics

For evoked or iontophoretic IPSC, or currents evoked by fast GABA application, mean values of peak/weighted decay time constant were calculated from an average waveform obtained from 10–20 sweeps at baseline and again following stabilization of drug effects. sIPSCs were detected using a template-matching algorithm in Clampfit; and templates were created by averaging >20 events from a representative cell. Decay times were assigned based on single or bi-exponential least-squares fits as appropriate. For bi-exponential fits, we collapsed the two components into an overall weighted decay time constant to facilitate averaging across cells. The weighted time constant was calculated as ΣAi*τi, where Ai is the fractional amplitude and τi is the time constant of the exponential component.

Cells served as the statistical basis of N for all experiments; animal number is reported in figure legends. Student’s independent, two-tailed t-test was used to compare means of two groups. For comparison of >2 groups, one-way ANOVA followed by Bonferroni corrected, post-hoc analyses were performed. For detection of effects within cells, a paired t-test or one-way repeated measures ANOVA was performed. Statistical analysis was performed using GraphPad Prism 7 or 8 (GraphPad Software, San Diego, California). For clarity, specific tests are described in the Results and Figure Legends. In figures significance is displayed at the level of p ≤ *0.05, **0.01, and ***0.001. Summary data are presented as mean ± standard deviation (SD). Raw data are available at the Figshare data repository (https://figshare.com/articles/delta_IPSC_Dataset_xlsx/11586696).

Results

δ and γ2 IPSC components are modulated in parallel by stimulus strength and by transport inhibition

Because of the perisynaptic or extrasynaptic localization of δ receptors (Wei et al., 2003), physical spillover must occur for δ IPSCs. However, whether diffusion is more important for governing the amplitude and time course of δ IPSCs compared with γ2 IPSCs remains unclear. For instance, the high affinity of δ receptors may compensate for the extrasynaptic localization of δ receptors, leading to little functional impact of diffusion. Our previous study found that in the presence of intact GABA uptake, stimulus intensity recruited γ2 and δ receptors mostly in parallel (Sun et al., 2018). If spillover is functionally more important for δ IPSCs than for γ2 IPSCs, inhibiting GABA transport while manipulating the number of activated presynaptic fibers with increased stimulus intensity should increase the extrasynaptic GABA concentration (Isaacson et al., 1993; Rossi & Hamann, 1998) and result in higher δ receptor contribution to IPSCs. In contrast, we found that increased stimulus intensity with uptake blockade by 10 μM NO-711, the GABA transporter inhibitor, did not alter the relative contribution of δ to IPSC peak amplitude (Figure 1A, B). δ contribution to IPSC peak amplitude was 31 ± 10 % when evoked at low stimulus intensity and was 26 ± 11 % when evoked at high stimulus intensity (n = 8, p = 0.41). Similarly, δ contribution to IPSC charge did not differ between low vs high stimulus intensity (50 ± 13 % vs 50 ± 19 %, n =8, p =0.99). The amplitudes analyzed represented 32 ± 9% (low stimulus intensity) and 111 ± 16 % (high stimulus intensity) of the amplitude evoked by maximum stimulus intensity. This suggests that δ receptors are recruited proportionally to synaptic γ2 receptors, consistent with our previous findings (Sun et al., 2018).

Surprisingly, applying NO-711 not only prolonged the decay of δ IPSCs, but also prolonged γ2 IPSCs at both low and high stimulus intensity, showing that the IPSC decays of both receptor classes are limited by transporters (Figure 1C, D). NO-711 increased δ decay from 71.5 ± 5.9 ms (n = 7) to 190.4 ± 45.2 ms (n = 8) at low stimulus intensity (p = 0.02) and from 98.3 ± 10.5 ms (n = 7) to 450 ± 131.2 ms (n = 8) at high stimulus intensity (p < 10⁻⁴; Figure 1C). Figure 1D shows that NO-711 increased γ2 decay at low intensity from 36.8 ± 9.9 ms (n = 7) to 78.1 ± 37.3 ms (n = 8, p = 0.03) and from 33.1 ± 11.5 ms (n = 7) to 142 ± 34 ms (n = 8) at high intensity (p < 10⁻⁴). The decay with increased stimulus intensity in the presence but not absence of NO-711 was increased for both δ and γ2 IPSCs (n = 8, p < 10⁻⁴, n = 8, p < 3 × 10⁻⁴). On the other hand, δ IPSCs exhibited a substantially longer latency to peak than γ2 IPSCs, and NO-711 prolonged the rise time of δ IPSCs from 34.4 ± 5.6 ms (n = 7) to 58.4 ± 11.3 ms (n = 8, p < 10⁻⁴) at low stimulus intensity, and from 32.1 ± 2.6 ms (n = 7) to 49.9 ± 7.9 ms (n = 8, p = 6 × 10⁻⁴) at high stimulus intensity (Figure 1E). NO-711 did not prolong the rise time of γ2 IPSCs evoked at low stimulus intensity (from 9.2 ± 3.5 ms, n =7, to 8.9 ± 2.2 ms, n = 8, p > 0.99) or high stimulus intensity (from 5.7 ± 2.1 ms, n = 7, to 7.7 ± 2.8 ms, n = 8, p = 0.63; Figure 1F), consistent with more GABA spillover for δ IPSCs due to their extrasynaptic localization (Figure 1E,F). These results suggest that over a wide range of activated synapses, δ and γ2 IPSCs are affected mostly in parallel by increased synaptic recruitment and by transport inhibition, with the δ component exhibiting evidence for more GABA spillover, reflected mainly in the rise of IPSCs.

δ receptors are more sensitive to alteration of vesicle release probability than γ2 receptors

Experiments in Figure 1 address spillover by manipulating the number of activated presynaptic axons without altering synaptic release probability. Spillover of neurotransmitter may also be affected by altered vesicle release through manipulation of vesicle availability or vesicle release probability (Mennerick & Zorumski, 1995; Lozovaya et al., 1999; Overstreet & Westbrook, 2003). We manipulated vesicle availability with paired-pulse stimulation (300 ms interval) to evoke IPSCs (Figure 2A). GABA release is known to show presynaptic depression in response to paired stimulation (Davies et al., 1990; Davies & Collingridge, 1993). The peak amplitude of IPSCs showed stronger paired-pulse depression for δ receptors (53 ± 7 %) than for γ2 receptors (33 ± 20 %; n = 7, p = 0.04, Figure 2B). On the other hand, this small difference did not translate into a change in the percentage of δ component contributing to the conditioning (22 ± 10 %) versus test IPSC (16 ± 7 %, n = 7, p = 0.06, Figure 2C). The decay time constant was reduced from 72 ± 10.5 ms to 64 ± 6.5 ms (n = 7, p = 7 × 10⁻³) for the δ component and from 33 ± 13.7 ms to 19.5 ± 6 ms (n = 7, p = 4 × 10⁻³) for the γ2 component (Figure 2D), while the level of depression of total charge transfer did not differ between δ (58 ± 6.6 %) and γ IPSCs (59 ± 9.5 %; n = 7, p = 0.77, Figure 2E). The faster decay of test (second) IPSCs for both γ2 and δ receptors suggests that GABA diffusion influences the decay of both receptor classes when transmitter output is high. The stronger depression of peak δ IPSCs following paired-pulse stimulation, although small, could result from reduced GABA accumulation or from postsynaptic factors, such as slow recovery from desensitization (Bright et al., 2011). These latter factors are addressed by studies of postsynaptic properties below.

Figure 2. δ receptors are more sensitive to paired-pulse depression than γ2 receptors.

A. Representative traces of paired-pulse stimulation of IPSCs evoked in δ* KI slices. Black: IPSCs evoked before PTX application. Red: IPSCs evoked after PTX application, mediated by δ receptors. Grey: Subtracted IPSCs (Black – Red), mediated by γ2 receptors. B. Percentage of peak reduction of δ or γ2-mediated IPSCs with paired-pulse stimulation (N = 7 cells from 3 animals, paired t test, p = 0.04). C. δ contribution to total peak amplitude of the first vs second IPSCs during paired stimulation (N = 7 cells from 3 δ* KI animals, paired t test, p = 0.06). D. Left panel: weighed time constant of decay for the first vs second δ IPSCs following paired-pulse stimulation. Right panel: decay weighed time constant of the first vs second γ2 IPSCs following paired pulse stimulation. (N = 7 cells from 3 δ*KI animals, paired t test, p = 7 × 10⁻³ and 4 × 10⁻³, respectively). E. Percentage of charge reduction of δ or γ2-mediated IPSCs with paired-pulse stimulation (N = 7 cells from 3 animals, paired t test, P = 0.77).

Manipulation of vesicle release probability by lowering extracellular [Ca²⁺] ([Ca²⁺]_o) from 2.6 mM to 1.3 mM (Augustine, 2001; Markwardt et al., 2009; Thanawala & Regehr, 2013) showed a similar reduction of charge transfer for δ IPSCs (40 ± 22 %) compared with dual IPSCs (32 ± 25 %, n = 5, p = 0.81) or γ2 IPSCs (29.8 ± 28 %, n = 5, p = 0.68; Figure 3A, B), but the manipulation more strongly depressed peak δ IPSCs (39.7 ± 16.8 %) compared with dual IPSCs (20.3 ± 15.4 %, n = 5, p = 0.01) or γ2 IPSCs (17.1 ± 18 %, n = 5, p = 9 × 10⁻³; Figure 3C). Consistent with these findings, δ contribution to IPSC peak amplitude was reduced from 15.3 ± 6.8 %, in high [Ca²⁺]_o, to 12 ± 6.4 %, in low [Ca²⁺]_o (Figure 3D; n = 5, p = 3 × 10⁻³). Dual-component IPSC decays were reduced from 61.1 ± 12.2 ms to 48.9 ± 6.2 ms by lowering Ca²⁺ (Figure 3E; n = 8, p = 0.01). However, this effect appeared accounted for by marginal effects on both γ2 and δ decay components. δ IPSC decays were reduced from 60.7 ± 6.4 ms to 59.7 ± 5.7 ms by lowering Ca²⁺ (Figure 3E; n = 5, p = 0.78). γ2 IPSC decays were reduced from 51.8 ± 17.7 ms to 40 ± 12.5 ms by lowering Ca²⁺ (Figure 3E; n = 5, p = 0.08). Thus, spillover does not clearly affect the decay of δ IPSCs more than γ2 IPSCs. The additional reduction of peak δ IPSCs (Figure 3C, D) could nevertheless indicate a larger spillover component for δ IPSCs. The parallel behavior of the dual-component IPSCs and γ2 component (Figure 3B, C, E) gave us confidence that we could use the dual-component as a proxy for γ2 IPSCs in subsequent experiments. In these experiments, pharmacological manipulations were more complex, making it impractical to assess PTX sensitivity in each experimental condition.

Figure 3. δ contribution to peak IPSCs exhibits higher sensitivity to reduced vesicle release.

A. Representative traces of evoked IPSCs from δ* KI slices before (black) and after (red) PTX application, in high (2.6 mM) vs low (1.3 mM ) [Ca²⁺]_o. Grey: Subtracted IPSCs (Black – Red), mediated by presumed γ2 receptors. B-C. Total charge and peak reduction of δ, γ2, or dual-component (dual) IPSCs in reduced [Ca²⁺]_o. Comparisons were made by one-way repeated measures ANOVA, followed by Bonferroni’s multiple-comparisons test. p =0.01, p = 9 × 10⁻³ (N = 5 cells from 3 δ* KI animals). D. δ contribution to total peak amplitude of the dual IPSCs in high vs low [Ca²⁺]_o (N = 5 cells from 3 δ* KI animals, paired t test, p = 3 × 10⁻³). E. Decay weighted time constant of δ, γ2, or dual IPSCs in high vs low [Ca²⁺]_o. (N = 8 cells from 4 δ* KI animals for dual IPSC, and 5 cells from 3 δ* KI animals for δ and γ IPSC, paired t test, p = 0.01).

δ receptors, but not γ2 receptors, experience modestly reduced synaptic [GABA] following reduced release probability

Rapidly dissociating competitive antagonists more strongly depress PSCs at low release probability when transmitter diffusion contributes to the PSCs. This is because low release probability results in a lower effective transmitter concentration and more effective antagonist competition. By contrast, canonical uniquantal local interaction of transmitter with postsynaptic receptors resists altered antagonist sensitivity since each active synapse experiences a similar agonist concentration, regardless of overall output (Foster et al., 2005). Thus, to test further whether δ receptors experience lower [GABA] during reduced release probability, we applied 200 μM TPMPA, a low affinity, rapidly dissociating GABA_AR antagonist (Markwardt et al., 2009) (Figure 4A). The results revealed no difference of peak IPSC antagonism in high [Ca²⁺]_o (65 ± 12.3 %) vs in low [Ca²⁺]_o (69 ± 10.2 %, n = 6, p = 0.39) in the absence of PTX (Figure 4B, dual—mainly γ2 mediated). TPMPA also exhibited similar effects on dual IPSC charge antagonism in high [Ca²⁺]_o (78.1 ± 8.2 %) vs in low [Ca²⁺]_o (81.6 ± 5.6 %; n = 6, p = 0.21). IPSCs mediated by δ receptors, isolated by PTX, exhibited modestly stronger TPMPA antagonism when lowering [Ca²⁺]_o (from 72 ± 18.2 % to 82 ± 7.5 % IPSC peak antagonism; n = 10, p = 0.04; from 75.7 ± 15.8 % to 90 ± 10.7 % IPSC charge antagonism; n = 10, p = 1 × 10⁻³; Figure 4C). This pattern of results suggests that δ receptors, but not γ2 receptors, experience lower synaptic [GABA] during reduced transmitter output.

Figure 4. δ receptors experience marginally lower [GABA] with reduced release probability.

A. Top: Representative traces of evoked dual-component (dual) IPSCs, in the absence of PTX, from δ* KI slices in high (2.6 mM; black) vs low (1.3 mM; red) [Ca²⁺]_o, before and after TPMPA application. Bottom: Representative traces of evoked δ IPSCs, in the presence of PTX, from δ* KI slices in high (2.6 mM; black) vs low (1.3 mM; red) [Ca²⁺]_o, before and after TPMPA application. B. Peak and charge reduction of dual IPSCs mediated by TPMPA in high vs low [Ca²⁺]_o. (N = 6 cells from 2 δ* KI animals; paired t test) C. Peak and total charge reduction of δ IPSCs mediated by TPMPA in high vs low [Ca²⁺]_o. (N = 10 cells from 5 δ* KI animals; paired t test).

Hindering diffusion has higher impact on the decay of δ IPSCs

Our results so far suggest that diffusion and/or transmitter rebinding modestly influence the peak amplitudes of δ IPSCs more than that of γ2 IPSCs. If GABA escapes significant distances to activate δ receptors, hindering diffusion should alter the time course and peak amplitude of δ IPSCs. Application of high molecular weight dextran has been used previously to probe the role of diffusion (Min et al., 1998; Savtchenko & Rusakov, 2005; Szabadics et al., 2007; Markwardt et al., 2009). Bath-applied dextran (5%) increased peak amplitude of evoked δ (from −62 ± 38.8 pA to −75.7 ± 44.4 pA, n = 9, p = 0.04) and dual-component (γ2-dominated) IPSCs (from −64.7 ± 42.9 pA to −177 ± 93 pA, n = 8, p = 2 × 10⁻³), suggesting neither δ nor γ2 receptors are saturated by GABA at peak (Figure 5A,C). The result also suggests that both evoked δ and γ2 IPSCs are driven by synaptic and diffusional GABA. On the other hand, dextran consistently but modestly reduced the decay time constant of δ IPSCs (from 66 ± 9.6 ms to 59 ± 11.8 ms; n = 9, p = 0.03), without changing the decay of dual IPSCs (from 30.8 ± 6.8 ms to 29.5 ± 4 ms, n = 8, p = 0.43), suggesting that evoked δ IPSCs exhibited higher sensitivity to manipulation of GABA diffusion than γ2 IPSCs (Figure 5A, B). No dextran effect on IPSC rise time (latency to peak) was evident. For dual IPSCs, latency to peak was 9.7 ± 3.2 ms before dextran and 8.2 ± 2.2 ms in dextran (n = 8, p = 0.13). For δ IPSCs, latency to peak was 19 ± 14 ms before dextran and 21.2 ± 13.2 ms in dextran (n = 9, p = 0.32). These results suggest that the onset of evoked IPSCs is not dictated strongly by diffusion (Figure 5A, D). As a comparator, NMDAR EPSC peak amplitudes were increased by dextran from −115.9 ± 78.3 pA to −186.6 ± 106.4 pA (n = 8, p = 4 × 10⁻³), with an effect size on peak EPSC intermediate between that of δ and dual-component IPSCs and with no consistent effect on either EPSC decay (from 29.8 ± 6.8 ms to 32.4 ± 11.5 ms, n = 8, p = 0.33) or rise time (from 9.5 ± 2.7 ms to 11.5 ± 3.5 ms, n = 8, p = 0.16; Figure 5). These effects on NMDAR EPSCs are similar to previously reported results in CA1 (Savtchenko & Rusakov, 2005).

Figure 5. Manipulation of GABA diffusion has similar impacts on evoked dual-component or δ IPSCs.

A. Representative traces of evoked dual-component (dual) or δ IPSCs before (black) and after (red) dextran application. Dextran effects on evoked NMDA EPSCs is also shown as a positive control. B. Decay weighted time constant of IPSCs and EPSCs before and after dextran application. C. Peak amplitude of IPSCs and EPSCs before and after dextran application. D. Latency to peak of IPSCs and EPSCs before and after dextran application. (N = 8 cells from 2 δ* KI animals for dual-component IPSCs; N = 9 cells from 5 δ* KI animals for δ IPSCs; N = 8 cells from 2 WT animals for NMDA EPSC; paired t test).

The preceding results suggest that diffusion participates in evoked GABA transmission onto both δ and γ2 receptors. We next tested whether hindering diffusion affects sIPSCs, where transmitter output is reduced to single-axon events. Dextran failed to affect peak, decay or rise time of sIPSCs mediated by δ or by γ2 receptors (Figure 6A–D). For dual sIPSCs, the decay was 8.7 ± 1.6 ms before dextran and 8.2 ± 1.7 ms in dextran (n = 10, p = 0.5). For δ sIPSCs, the decay was 17.3 ± 12.1 ms before dextran and 22.6 ± 20.1 ms in dextran (n = 9, p = 0.2). For dual sIPSCs, peak amplitude was −31.5 ± 25 pA before dextran and −50.9 ± 45.6 pA in dextran (n = 10, p = 0.14). For δ sIPSCs, peak amplitude was −7 ± 2.6 pA before dextran and −7 ± 2.5 pA in dextran (n = 9, p = 0.96). For dual sIPSCs, the rise time was 1.9 ± 0.6 ms before dextran and 1.6 ± 0.5 ms in dextran (n = 10, p = 0.18). For δ sIPSCs, rise time was 4.8 ± 1.2 ms before dextran and 6.1 ± 2.7 ms in dextran (n = 9, p = 0.14). These results suggest that peak δ sIPSCs, similar to γ2 sIPSCs, are driven by local, near saturating GABA effects (Figure 6B–D), suggesting a limited impact of GABA diffusion on sIPSCs mediated by δ receptors. Interestingly, the decay of δ sIPSCs was still slower than the decay of dual-component sIPSCs (17.3 ± 12.1 ms for δ sIPSCs, n = 9, vs 8.7 ± 1.6 ms for dual-component sIPSCs, n = 10, p = 0.04, Student’s independent t test).

Figure 6. Manipulation of GABA diffusion has similar impacts on dual component or δ sIPSCs.

A. Representative traces of dual-component (dual) or δ sIPSCs before (black) and after (red) dextran application. B. Decay weighted time constant of sIPSCs before and after dextran application. C. Peak amplitude of sIPSCs before and after dextran application. D. Rise time of sIPSCs before and after dextran application. (N = 10 cells from 4 δ* KI animals for dual-component sIPSC; N = 9 cells from 5 δ* KI animals for δ sIPSC; paired t test).

Explore intrinsic channel deactivation kinetics of δ vs γ2 receptors

Figure 6 shows that slowing diffusion does not affect the decay of δ sIPSCs. This is in contrast to some bona fide diffusion mediated PSCs (Lu et al., 2017) and raises the possibility that δ sIPSC decay time course is limited by slow deactivation kinetics of receptors rather than diffusion or transmitter rebinding. To explore this possibility, we employed rapid GABA application to nucleated outside-out patches from dentate granule neurons. Using a theta barrel to rapidly apply and remove 100 μM GABA to δ* patches in the presence of PTX, we found that decay/deactivation kinetics of δ* receptors (24.6 ± 5.5 ms, n = 5) was comparable to δ sIPSCs (17.3 ± 12.1 ms, n = 9, p = 0.23), implying that slow deactivation is a major driver of δ sIPSCs. However, kinetics of dual-component patch currents (24.2 ± 6.8 ms, n = 6) were slower than dual-component sIPSCs (8.7 ± 1.6 ms, n = 10, p < 10⁻⁴; Figure 7A,C; Figure 6B).

Figure 7. δ receptors exhibit channel deactivation comparable to δ sIPSCs.

A. Representative traces of GABA-induced dual-component or δ currents evoked by theta-barrel applications onto nucleated patches from WT (no PTX) and δ* KI (in 100 μM PTX) dentate granule neurons, respectively (N = 6 patches from 2 WT animals; N = 5 patches from 3 δ* KI animals). B. Iontophoretic application of varied GABA concentrations onto nucleated patches from the indicated genotypes and conditions. Insets show the effect of PTX. C. The summary of deactivation kinetics across GABA concentrations, obtained with varied iontophoretic currents (closed circles), showed similar values to theta-barrel applications (purple, open diamonds shown at saturating [GABA]). For WT dual-component current deactivation, comparisons of tau values below 30% of maximal peak amplitude (N = 9 patches from 2 WT animals) vs. those at maximal peak amplitude (N = 9 patches from 2 WT animals) showed no difference (19.7 ± 8.5 ms vs. 22.8 ± 7.3 ms, p = 0.35, Student’s paired t test). For δ current deactivation, comparisons of tau values below 30% of max peak amplitude (22.2 ± 11.9 ms, N = 4 patches from 3 δ* KI animals) to those at 100% of max peak amplitude, either within patch (36.3 ± 14.7 ms, N = 4 patches from 3 δ* KI animals; p = 0.13, Student’s paired t test) or to the group mean (30.3 ± 11 ms, N = 9 patches from 3 δ* KI animals; p = 0.26, Student’s independent t test) also failed to reveal evidence for an effect of GABA concentration on deactivation kinetics. Nucleated patches from δ* KI mice in the absence of PTX showed indistinguishable dual-component GABAR deactivation rate constants (open circles) compared to patches from WT mice (closed circles). Pooled data from both theta-barrel application and iontophoresis showed a slower deactivation rate (averaged across GABA concentrations) for δ receptors compared to γ2-dominated dual-component receptors (26.4 ± 7 ms vs 21.5 ± 6 ms, respectively; p = 0.04, Student’s independent t test). D-F. To examine whether fast GABA removal following iontophoretic application may rate limit decay kinetics, we performed control experiments with patches excised from dentate granule cells expressing γ2* KI receptors (D), and with glutamate iontophoresis (E). The faster deactivation of γ2* was detected (D,F) and the deactivation rate of AMPARs is indistinguishable from the decay of sEPSCs (blue inset in E, blue open triangles in F, p = 0.3, Student’s independent t test ). N = 9 patches from 2 WT animals, 6 patches from 2 γ2* KI animals, 4 patches for iontophoretic AMPA currents and 4 dentate granule cells for sEPSC recording; AMPA iontophoresis and sEPSC data were collected from the same two animals (1 WT and 1 δ* KI animal). Statistical comparisons indicated in panel F: Bonferroni’s multiple comparisons test following one-way ANOVA.

To address whether the slower deactivation of dual-component current was the result of technical issues with theta application, we employed the alternative method of iontophoresis. Iontophoresis also allowed us to explore a range of agonist concentrations, since deactivation of some GABA_AR subunit combinations is dependent on GABA concentration (Banks & Pearce, 2000), and we do not know the relevant GABA concentration at peri-synaptic/extrasynaptic δ receptors following synaptic release. Varied iontophoretic currents induced increasing current amplitudes. Current decay showed no dependence on agonist concentration in dual-component or PTX-isolated δ* receptors (Figure 7B, C). WT dual-component current deactivation below 30% of maximal peak amplitude was 19.7 ± 8.5 ms, comparable to the deactivation rate at 100% of maximal peak amplitude (22.8 ± 7.3 ms, n = 9, p = 0.35). δ current deactivation below 30% of maximal peak amplitude was 22.2 ± 11.9 ms (n = 4), similar to the deactivation rate at 100% of maximal peak amplitude when comparing within patch (36.3 ± 14.7 ms, n = 4, p = 0.13) or between patches (30.3 ± 11 ms, n = 9, p = 0.26). These results suggest that deactivation does not depend on GABA concentration for either receptor population. In δ* patches in the absence of PTX, deactivation rates of dual-component GABA current matched those of WT patches (Figure 7C, open circles vs closed circles). Hence the data from both groups were pooled for further analysis. Deactivation rates of dual-component GABA iontophoretic current matched the kinetics observed with theta-barrel application (Figure 7C, purple open diamonds). In pooled data sets, average deactivation rate across GABA concentrations was slower for δ receptors than for γ2-dominated dual-component current (26.4 ± 7 ms, n = 13, vs 21.5 ± 6 ms, n = 20, respectively; p = 0.04, Student’s independent t test), consistent with the idea that slower kinetics of δ receptors is a major contributor to slower δ sIPSCs.

The lack of decay dependence on GABA concentration seems to exclude slow agonist removal as an important contributor to patch kinetics. However, we were still puzzled by the slower dual-component patch decays compared with dual-component sIPSCs (Figure 7C vs. Figure 6). To examine the possibility that fast GABA removal from patches may be rate limiting, we performed control experiments with patches excised from dentate granule neurons expressing γ2* KI receptors, which display a faster baseline sIPSC decay in the absence of PTX (Sun et al., 2018). We detected faster deactivation of dual-component GABA currents evoked from γ2* KI slices (10.7 ± 4.2 ms, n = 6) than from WT γ2 slices (23 ± 7 ms, n = 9, p = 3 × 10⁻⁴; Figure 7D,F), consistent with the idea that patch responses were not limited by agonist removal time, although these responses were also slower than sIPSCs from the same genotype (Sun et al., 2018). Finally, we performed fast glutamate application using glutamate iontophoresis onto patches to evoke AMPAR currents. The evoked AMPAR currents exhibit deactivation rates at 3.9 ± 1.1 ms (n = 4), comparable to decay of sEPSCs (5.2 ± 1.9 ms; n = 4, p = 0.3, Student’s independent t test), further suggesting that agonist removal does not rate limit decay kinetics (Figure 7E–F). The pooled AMPAR current deactivation (4.5 ± 1.6 ms, n = 8) was significantly faster than the dual component GABA current deactivation from WT γ2 slices (23 ± 7 ms, n = 9, p < 10⁻⁴). Reasons for the mismatch between dual-component patch deactivation vs. dual-component sIPSCs are addressed in the Discussion.

We also used patches to explore a role for postsynaptic factors in paired-pulse depression of δ vs. γ2 IPSCs in Figure 2. Neither dual-component currents nor δ receptor currents exhibited depression when probed with maximum iontophoretic current with an interval of 300 ms. Dual-component currents showed 0.8 ± 1.8% depression (n = 3) and δ receptors showed 9 ± 9.1% potentiation (n = 5). Therefore, it is unlikely that differences in desensitization to brief pulses of GABA can explain the stronger paired-pulse synaptic depression observed in Figure 2.

Discussion

Our work explored the role of diffusion in generating δ IPSCs in dentate granule neurons using mice that allow pharmacological isolation of δ receptors. The work was motivated by the slow time course of δ IPSCs relative to γ2 IPSCs (Wei et al., 2003; Sun et al., 2018); prevailing understanding suggests a diffusional component of δ transmission because of the extrasynaptic location of δ subunit-containing GABA_A receptors (Wei et al., 2003). In contrast to a strong diffusional component to transmission via δ receptors, we found that δ and γ2 receptors are mostly modulated in parallel by manipulations that affect the concentration of GABA at extrasynaptic receptors. For the first time we surveyed the deactivation kinetics of δ receptors in a native cell type and find a good match with δ sIPSC kinetics. This suggests that at least a fraction of δ receptors receive a brief, synaptic-like pulse of GABA following vesicular release that mimics rapid application to a patch. Overall, our results suggest that GABA functions as a synaptic transmitter at both δ receptors and γ2 receptors, with some limited additional diffusional contribution to δ IPSCs compared with γ2 IPSCs. Apparently, the high affinity of δ receptors for GABA helps compensate for their extrasynaptic localization.

Several caveats and limitations warrant discussion. Strictly speaking, the component of GABA IPSCs that we isolated in the present work by PTX-subtraction was a non-δ receptor component. However, our previous work has shown that γ2-containing receptors are the main mediators of IPSC current in dentate granule neurons (Sun et al., 2018). Therefore, we feel confident in assigning the PTX-sensitive current as γ2-mediated. We also acknowledge that our study is premised on the idea that receptors containing a mutated δ subunit behave similarly to WT δ receptors. Our previous study performed comprehensive baseline IPSC comparisons between WT and δ* slices and failed to find differences. Nevertheless, it remains a formal possibility that the subunit mutation modestly alters trafficking or biophysics of functional receptors.

The peri/extrasynaptic localization of δ receptors has been demonstrated morphologically (Wei et al., 2003). Therefore, some escape of GABA from the synaptic cleft must be required to activate δ receptors. On the other hand, our physiological assays demonstrate that diffusion is of limited importance for the activation of the receptors unless there is sufficient synchronous co-release from many synapses. δ receptors possess > 4-fold higher affinity for GABA than typical γ2 receptors (Brown et al., 2002; Wohlfarth et al., 2002; You & Dunn, 2007; Karim et al., 2013; Lagrange et al., 2018). Therefore, the extrasynaptic location of δ receptors may not have particularly strong physiological implications without synchronous GABA release from multiple sites, which gives rise to slower IPSCs for both δ and γ2 IPSCs. Indeed the largely parallel modulation of the two receptor classes by presynaptic manipulations may be among the more surprising results of our work.

One challenge remains in reconciling the behavior of δ mIPSCs, which our previous work revealed to decay faster (6–15 ms) than δ sIPSCs and faster than the deactivation times observed in the present study (Sun et al., 2018). Because IPSCs cannot decay faster than the deactivation kinetics of the channels involved, it is possible that the previous δ mIPSC estimates were not resolved accurately due to the limited signal:noise ratio of the small events. Although our rapid application studies revealed no evidence for dependence of deactivation on GABA concentration (Banks & Pearce, 2000), we also cannot exclude the possibility that δ mIPSC deactivation kinetics represent a lower GABA concentration, with faster deactivation, than we were able to resolve in nucleated patches. Finally, δ sIPSC decays could be driven mainly by multivesicular release, involving more spillover or receptor reactivation than univesicular δ mIPSCs (Overstreet & Westbrook, 2003). The lack of dextran effect on δ sIPSC decay causes us to favor the idea that δ sIPSCs are mostly driven by local, synaptic GABA.

Results in Figure 1 reveal a similar contribution of GABA spillover or rebinding for both γ2 and δ IPSCs (evident as the sensitivity of IPSCs to NO-711), suggesting that extrasynaptic γ2 receptors are recruited proportionally with δ receptors as the number of activated presynaptic fibers is increased. Dentate granule cells exhibit some extrasynaptic GABA_A receptors (Glykys et al., 2008; Engin et al., 2015). It is therefore possible that these extrasynaptic γ2 receptors or others are recruited following increased synaptic stimulation. Although the decay of both γ2 and δ IPSCs exhibit similar sensitivity to NO-711 (Fig. 1C–D), the rise time (latency to peak) of δ IPSCs was more strongly influenced by NO-711 (Fig. 1E–F). This pattern suggests a relatively stronger diffusional role in δ IPSCs. Other results suggesting a stronger role for diffusion at δ receptors include the sensitivity to paired stimulation, sensitivity to lowered extracellular Ca²⁺, sensitivity to TPMPA, and sensitivity to dextran. Each of these effects exhibited a rather modest effect size compared with γ2 IPSCs, leading to our conclusion of a marginal difference in the contribution of diffusion to δ IPSCs compared with γ2 IPSCs.

The slower decay of γ2-driven IPSCs under conditions of elevated output, such as evoked IPSCs, may be caused by both GABA diffusion and slower channel kinetics of extrasynaptic γ2 (Banks & Pearce, 2000). The decay time constant of evoked γ2 IPSCs was highly sensitive to paired-pulse depression (Figure 2). This may suggest a stronger spillover or rebinding effect on the initial IPSC, where transmitter output is higher. Slow GABA clearance is also suggested by γ2 IPSC decay sensitivity to NO-711 (Figure 1D). If spillover onto extrasynaptic γ2 receptors is involved, different biophysical properties of these receptors could also participate in slow IPSC decays (Banks & Pearce, 2000). We found that γ2-mediated, dual-component currents from nucleated patches showed deactivation kinetics in line with evoked γ2 IPSCs (Figure 5B) but slower than synaptic γ2-driven sIPSCs (Figure 7, Figure 6B). These results suggest that bona fide synaptic γ2 receptors have faster kinetics than those found outside synapses, which dominate responses of receptors in patches. Similar results have been found in CA1 neurons (Banks & Pearce, 2000). One possibility is that auxiliary subunits or other synaptic proteins responsible for kinetics (Han et al., 2019) are absent from receptors in nucleated patches. Alternatively, patch excision may directly alter γ2 receptor properties. These possibilities could apply to δ receptor deactivation as well, but the best available evidence when combining outcomes of dextran, TPMPA, and nucleated patch experiments suggests a surprisingly limited role of GABA diffusion in δ sIPSC kinetics.

Past work using dextran to probe diffusional contributions has led to different results, some of which qualitatively match our observations, and others that do not. For instance, dextran slows a glutamate EPSC in unipolar brush cells (Lu et al., 2017) but speeds GABA IPSCs generated by neurogliaform cells in hippocampus (Szabadics et al., 2007). The latter is similar to our results. The specific effects of slowing diffusion likely depend on the concentration of transmitter involved and the specific spatial configuration of receptors relative to release site.

Our results revealed that sIPSCs are saturated (evident as no effect of dextran on the peak, Figure 6C), whereas evoked IPSCs are not saturated (slowed diffusion augmented peak, Figure 5C). A possible explanation is that intersynaptic crosstalk is actually increased by slowing diffusion (Rusakov & Kullmann, 1998; Kullmann et al., 1999). Evoked GABA release by activating many synapses simultaneously, may recruit receptors that are perisynaptic or extrasynaptic but that normally experience low GABA levels. When GABA diffusion is slowed by dextran, these perisynaptic or extrasynaptic receptors may experience higher local GABA concentration, depending on their distance from release sources (Rusakov & Kullmann, 1998). On the other hand, sIPSCs may involve receptors that are located directly opposite release sites and experience locally saturating GABA. Therefore, slowing GABA diffusion will not alter the sIPSC peak amplitude (Rusakov & Kullmann, 1998).

Our results provide the first estimates of deactivation kinetics for δ receptors in native neurons. Previous results have suggested that δ receptors may exhibit fast kinetics and weak desensitization (Haas & Macdonald, 1999), slow kinetics and strong desensitization (Bright et al., 2011), or fast kinetics and weak desensitization (Lagrange et al., 2018) perhaps depending on the α subunit partner. Our results suggest intermediate deactivation kinetics with no desensitization evident by 300 ms following a saturating pulse of GABA. Deactivation seemed mildly dependent on GABA concentration, possibly suggesting a role for mono-liganded receptors in generating δ current.

The time course of IPSCs has a major impact on excitation (Rossi & Hamann, 1998; Sun et al., 2018). Here we have investigated the reasons for the slow time course of δ IPSCs. We find that slow deactivation kinetics play a major role in δ IPSCs in dentate granule neurons, with a role for diffusion in the time course of both δ and γ2 IPSCs when synchrony of release is augmented. Our results suggest a functional relationship between δ and γ2 receptor-mediated IPSCs in dentate granule neurons akin to that of slow, small synaptic NMDA and fast, large AMPAR receptor-mediated EPSC at many glutamate synapses.

Key points summary.

• Current views suggest γ2 subunit-containing GABA_A receptors mediate phasic IPSCs while extrasynaptic δ subunits mediate diffusional IPSCs and tonic current.

• We have re-examined the roles of the two receptor populations using mice with picrotoxin resistance engineered into receptors containing the δ subunit.

• Using pharmacological separation, we find that in general δ and γ IPSCs are modulated in parallel by manipulations of transmitter output and diffusion, with evidence favoring modestly more diffusional contribution to δ IPSCs.

• Our findings also reveal that spontaneous δ IPSCs are mainly driven by channel deactivation, rather than by diffusion of GABA.

• Understanding the functional contributions of the two receptor classes may help understand the actions of drug therapies with selective effects on one population over the other.