Slow GABA Transient and Receptor Desensitization Shape Synaptic Responses Evoked by Hippocampal Neurogliaform Cells

Prolonged duration of GABA after vesicular release at NGFC synapses. A, Top traces, Representative traces of NGFC-IPSCs showing the effect of 300 μm TPMPA (each trace is the average of 5 sweeps). Bottom traces, Closer examination by scaling the attenuated trace to control amplitude reveals that TPMPA prolongs rise time of the IPSC. B, C, Representative traces showing superimposed control and IPSCs during the application of 300 nm bicuculline or 1 μm SR95531 (each trace is the average of 5 sweeps). Scaling of the attenuated trace reveals that neither bicuculline nor SR95331 significantly altered the 20–80% rise time of the unitary IPSCs. D, Quantification of the actions of TPMPA, bicuculline, or SR95531 on 20–80% rise time and amplitude of NGFC-IPSCs. Statistical significance is noted (***p < 0.005; *p < 0.05). Number of observations were 15 for TPMPA, 5 for bicuculline (BIC), and 6 for SR95331. CTRL, Control. Error bars indicate SEM.

Kinetics of NGFC-IPSCs at low release probability and in the presence of a GAT-1 blocker. A, B, Representative traces of NGFC-IPSCs in control or in the presence of 10 μm CdCl₂ showing the effect of application of 25 μm SKF89976A (each trace is an average of 5 sweeps). C, Quantification of the effects of cadmium or SKF89976A on 20–80% rise time, amplitude, and decay time constant of the IPSCs. Note that SKF89976A is applied either alone (left red bars for each graph) or in the presence of Cd²⁺ (right red bars for each graph). Any statistical difference is highlighted with asterisks (*p < 0.05, **p < 0.01, ***p < 0.005; n = 5 for SKF89976A, n = 8 for Cd²⁺, n = 4 for Cd²⁺ plus SKF89976A). Error bars indicate SEM. D, Representative examples of fast sIPSCs and slow sIPSCs recorded in NGFCs in control conditions and after application of 25 μm SKF89976A. E, Quantification of the effect of SKF89976A on sIPSCs. The left graph shows the analysis of 10–90% rise time (in milliseconds) versus decay time constant (in milliseconds) of 500 sIPSCs recorded from NGFCs (n = 5). Note the appearance of two populations: a fast rising, fast decaying events (fast sIPSCs), and slowly rising (>3 ms), slowly decaying (>40 ms) events (slow sIPSCs). The two distributions were highly significantly different (p < 0.001, Kolmogorov–Smirnov test here and onward). The application of SKF89976A enhanced the rise time and the decay time constant of the slow sIPSCs (p < 0.001) but not of the fast sIPSCs (p > 0.1). In addition, SKF89976A enhanced the amplitude of the slow sIPSCs (p < 0.001) but not of the fast sIPSCs (p > 0.1). This latter action is shown in the cumulative plots (right) of the amplitude of fast sIPSCs (number of events, 485) and slow sIPSCs (number of events, 15) in control or during SKF89976A as indicated.

Effects of L-655708, zolpidem, or diazepam on hippocampal NGFC-IPSCs. Representative traces of NGFC-IPSCs in control and after application of 25 nm L-655708 (A), 100 nm zolpidem (B), or 1 μm diazepam (C). Each trace is the average of five individual sweeps. D, Quantification of the drug effects on 20–80% rise time, amplitude, and decay time constant of NGFC-IPSCs. Statistical significance is noted where appropriate (*p < 0.05, **p < 0.01, ***p < 0.005; n = 6 for L-655708, n = 17 for zolpidem, n = 4 for diazepam). Error bars indicate SEM.

The concentration and duration of GABA that drives NGFC-IPSC. A, Experimentally recorded NGFC-IPSCs (dots, normalized to the peak of the control current in each recording) were fit to simulations (solid lines) using the kinetic model shown in supplemental Figure 6 (available at www.jneurosci.org as supplemental material). For each condition, the rate constants of the model were fixed (see supplemental Fig. 6, available at www.jneurosci.org as supplemental material), and only the amplitude, rise, and decay time constants of the synaptic GABA concentration transient (stippled lines, peak normalized for display) were allowed to vary as free parameters. Where GABA_A receptor antagonists or zolpidem were applied, the model was forced to find the single GABA transient that simultaneously minimized error for both control (black) and drug (gray) conditions. In all cases, NGFC-IPSCs were best fit when the GABA transient was low and long lasting. For comparison, sIPSCs recorded in GCs of the dentate gyrus were also fit, and required a high but brief GABA transient. B, A plot of the best-fitting peak GABA concentration versus its decay time constant reveals a clear separation between the concentration profiles underlying fast (GC) and slow NGFC-IPSC, and shows that the latter is driven by GABA exposures of <100 μm that last for tens to hundreds of milliseconds. The dashed line is the boundary that best separates the parameters for NGFCs and GCs, obtained using Fisher's linear discriminant analysis. C, Qualitatively different receptor behaviors are induced by brief and high versus long and low GABA exposure. Simulations using the average GABA concentration transients (top, normalized) for GCs (stippled) and NGFCs (solid) predict that long and low GABA exposure results in lower peak open probability (middle) because of greatly enhanced occupancy of slow desensitized states (bottom). Note the much longer time axis in the bottom plot.

In vivo firing of NGFC during theta oscillations elicits midterm synaptic depression of NGFC-IPSCs. A, Responses of presynaptic (left) and postsynaptic (right) NGFCs recorded in voltage clamp in vitro to the injection into the presynaptic cell of NGFC firing activity recorded in vivo. In the experiment shown, the presynaptic NGFC displayed autaptic IPSCs and also evoked IPSCs in a postsynaptic NGFC. Left top trace, The vertical bars represent a depolarizing voltage pulses (100 stimuli, 1 ms each) used to evoke action currents that elicit autaptic IPSCs (left) and IPSCs in a postsynaptic NGFC (right). The bottom traces show some responses at the onset of the train at expanded current and time scales. B, Plot of the normalized amplitude of the IPSC for each stimulus in the train (autaptic and synaptic IPSCs were pooled; n = 9). The data were best fitted with exponential function with double decay time constants of 0.3 and 3.9 s. C, Summary graph showing normalized recovery from depression at different time points after the train of stimuli (the data were best fitted with exponential function with a single decay time constants of 10.4 min). Error bars indicate SEM.

NGFC-IPSCs also depress after low-frequency stimulation. A, Stimulation protocol used to study synaptic depression and recovery. Each vertical bar represents a 0.3 ms depolarizing voltage pulse used to evoke an action current that elicits an autaptic and/or synaptic slow IPSCs. B, Left, Single traces of autaptic NGFC-IPSCs evoked at varying intervals (shown next to each sweep) from the onset of the train. Right, Plot of the normalized amplitude of the IPSC, for the experiment shown, for each stimulus in the train. The data were fitted with a double exponential with a τ₁ of 21.9 s and τ₂ of 239.3 s. C, Left, Single traces of NGFC-IPSCs elicited at varying intervals (shown next to each sweep) after a 0.1 Hz train of stimuli. Right, Summary graph showing percentage recovery from depression at different time points after the train of stimuli for the experiment illustrated (middle graph) and for all data (right graph; n = 26). In the pooled data plot, the recovery from depression had a midpoint of 143.1 s. Error bars indicate SEM.

Slow receptor desensitization leads to the depression of low-frequency evoked NGFC-IPSCs. A, A train of 100 depolarizing pulses applied at 0.1 Hz produced a marked depression in the IPSC (n = 12). Increasing the ambient GABA concentration by application of 25 μm SKF89976A caused a marked downward shift in the IPSC amplitude versus time relationship (p < 0.005; n = 6; F test). In contrast, the fast-off GABA_A receptor antagonist TPMPA (300 μm) elicited a significant upward shift in this relation (p < 0.005; n = 10; F test). B, In the presence of 1–2 μm GABA, a similar protocol of 0.1 Hz stimulation induced a significantly faster synaptic depression than in controls (p < 0.05; n = 4; F test). C, Graph showing the recovery from desensitization expressed as change in paired-pulse ratio with increasing interstimulus interval (range, 0.3–100 s). The relationship can be described by a single exponential curve that reaches a plateau after ∼40 s (n = 10). Error bars indicate SEM.