Serotonergic control of GABAergic inhibition in the lateral amygdala

Abstract

Much evidence implicates the serotonergic regulation of the amygdala in anxiety. Thus the present study was undertaken to characterize the influence of serotonin (5-HT) on principal neurons (PNs) of the rat lateral amygdala (LA), using whole cell recordings in vitro. Because inhibition is a major determinant of PN activity, we focused on the control of GABAergic transmission by 5-HT. IPSCs were elicited by local electrical stimulation of LA in the presence of glutamate receptor antagonists. We found that 5-HT reduces GABA_A inhibitory postsynaptic currents (IPSCs) via presynaptic 5-HT_1B receptors. While the presynaptic inhibition of GABA release also attenuated GABA_B currents, this effect was less pronounced than for GABA_A currents because 5-HT also induced a competing postsynaptic enhancement of GABA_B currents. That is, GABA_B currents elicited by pressure application of GABA or baclofen were enhanced by 5-HT. In addition, we obtained evidence suggesting that 5-HT differentially regulates distinct subsets of GABAergic synapses. Indeed, GABA_A IPSCs were comprised of two components: a relatively 5-HT-insensitive IPSC that had a fast time course and a 5-HT-sensitive component that had a slower time course. Because the relative contribution of these two components varied depending on whether neurons were recorded at proximity versus at a distance from the stimulating electrodes, we speculate that distinct subtypes of local-circuit cells contribute the two contingents of GABAergic synapses. Overall, our results indicate that 5-HT is a potent regulator of synaptic inhibition in LA.

NEW & NOTEWORTHY We report that 5-HT, acting via presynaptic 5-HT_1B receptors, attenuates GABA_A IPSCs by reducing GABA release in the lateral amygdala (LA). In parallel, 5-HT enhances GABA_B currents postsynaptically, such that GABA_B inhibitory postsynaptic currents (IPSCs) are relatively preserved from the presynaptic inhibition of GABA release. We also found that the time course of 5-HT-sensitive and -insensitive GABA_A IPSCs differ. Together, these results indicate that 5-HT is a potent regulator of synaptic inhibition in LA.

INTRODUCTION

Converging lines of evidence implicate a dysregulation of the amygdala, possibly mediated by serotonin (5-HT), in pathological anxiety. First, the amygdala plays a critical role in the genesis of defensive behaviors (LeDoux 2000) and it is hyperactive in anxiety disorders (Bremner 2005; Shin et al. 2006). Second, serotonin (5-HT) uptake inhibitors figure among the recommended pharmacological treatments for anxiety disorders (reviewed in Bandelow et al. 2017) and the amygdala receives serotonergic inputs from the dorsal raphe (Ma et al. 1991; Sadikot and Parent 1990). Third, dorsal raphe neurons fire at higher rates during emotional arousal (Cohen et al. 2015; Grahn et al. 1999; Schweimer and Ungless 2010), causing a rise in amygdala levels of 5-HT (Kawahara et al. 1993; Kirby et al. 1995; Yokoyama et al. 2005).

To shed light on how 5-HT regulates emotional behaviors, several in vitro studies have examined the effects of 5-HT in the basolateral complex of the amygdala (BLA). Unexpectedly, contrasting results were obtained in the lateral (LA) and basolateral (BL) nuclei. In principal neurons (PNs) of LA but not BL, 5-HT was found to exert various excitatory effects, including a slow depolarization, the enhancement of depolarization-induced slow afterdepolarizations, and the inhibition of Ca²⁺-dependent K⁺ currents (Faber and Sah 2002; Yamamoto et al. 2012, 2014). In BL by contrast, most studies reported that 5-HT increases the frequency of spontaneous IPSCs in PNs, an effect dependent on the depolarization of interneurons via 5-HT_2A receptors (Bocchio et al. 2015; Jiang et al. 2009; Rainnie 1999).

So far, the regulation of inhibition by 5-HT has not been examined in LA. Yet, this question is of particular significance because LA is the main input station of the amygdala for sensory inputs and inhibition is a major determinant of PN activity in LA (reviewed in Paré et al. 2003). Indeed, in vivo intracellular recordings of PNs have revealed that their activity is dominated by inhibitory synaptic potentials of high amplitude and rapid onset (Lang and Paré 1997). As a result, PNs rarely spike, generally at rates <0.1 Hz (Paré and Gaudreau 1996). Consistent with these results, synaptic boutons immunoreactive for glutamic acid decarboxylase (GAD) or GABA are concentrated in the perisomatic region of PNs, although they are also present in their dendritic tree (Carlsen 1988; McDonald et al. 2002). Since deafferenting the BLA only causes slight decreases in local GAD levels (Le Gal LaSalle et al. 1978), it is likely that these inhibitory inputs mostly arise from GABAergic cells intrinsic to LA. In support of this, LA contains a similar variety of GABAergic interneurons subtypes (Mascagni and McDonald 2003; reviewed in Spampanato et al. 2011) as in cortex (Freund and Buzsáki 1996; Markram et al. 2004). For example, these include a numerically dominant parvalbumin group and somatostatin-expressing neurons, which, respectively, target the somatic region and dendrites of PNs (Muller et al. 2006, 2007a; Smith et al. 1998).

Thus, given the importance of inhibition in the functioning of LA, the present study aimed to characterize how 5-HT regulates GABA_A and GABA_B inhibition, identify the receptor subtype mediating these effects, and examine the possibility that these receptors are differentially expressed by different subsets of inhibitory inputs.

MATERIALS AND METHODS

Experiments were performed in accordance with the guiding principles of the Physiological Society of Japan and were approved by the Animal Care Committee of Kanazawa Medical University.

Slice preparations.

Wistar rats (21–30 days old, 29 males and 11 females) were decapitated under deep isoflurane anesthesia. The brains were dissected out and then sectioned using a vibrating microtome (Pro7; Dosaka, Kyoto, Japan) at a thickness of 300 µm in an ice-cold cutting solution composed (in mM) of 103 N-methyl-d-glucamine, 2.5 KCl, 10 MgSO₄, 30 NaHCO₃, 1.2 NaH₂PO₄, 0.5 CaCl₂, 25 glucose, 20 HEPES, 2 thiourea, 3 Na-pyruvate, and 12 N-acetyl-l-cysteine. Slices were kept at 30°C for 5 min in the above solution and then incubated at room temperature for more than 1 h in a normal artificial cerebrospinal fluid (ACSF) composed (in mM) of 124 NaCl, 2.5 KCl, 2.0 CaCl₂, 1.2 MgSO₄, 1.3 NaH₂PO₄, 26 NaHCO₃, and 20 glucose (~310 mOsm; ~7.2 pH). All solutions were oxygenated with a mixture of 95% O₂-5% CO₂. Recordings were made from slices submerged in normal ACSF at 30°C.

Electrophysiological recordings.

Slices were placed in a recording chamber on the stage of an upright microscope (BX51WI; Olympus, Tokyo, Japan) with a ×60 water-immersion objective (LUMPlanFI/IR, Olympus). Whole cell recordings were made from the soma of visually identified pyramidal-like neurons located in LA using glass pipettes (resistance 5–8 MΩ) pulled from borosilicate glass capillaries with a Narishige pipette puller (PP-830, Tokyo, Japan). Glass pipette electrodes were filled with a solution containing (in mM) 130 K-gluconate, 10 KCl, 2 MgCl₂, 2 Na₂-ATP, 0.4 Na₂-GTP, 0.2 EGTA, 10 HEPES, and 5 K₂-phosphocreatine with pH adjusted to 7.2–7.3 with KOH.

For miniature IPSC (mIPSC) recordings, pipettes were filled with a cesium-based solution containing (in mM) 50 Cs-methanesulfonate, 85 CsCl, 2 MgCl₂, 2 Na₂-ATP, 0.4 Na₂-GTP, 0.2 EGTA, 10 HEPES, and 5 QX-314 with pH adjusted to 7.2–7.3 with CsOH. Osmolarity was adjusted to 295–300 mOsm. The liquid junction potential for the potassium-based solution was ~10 mV and for the cesium-based solution was ~1 mV. However, the membrane potential values reported below were not corrected for the junction potential. We only considered neurons with a stable resting membrane potential negative to −55 mV. Series resistance was monitored. If it changed by >20%, the recording was discarded. We recorded only one neuron per slice and used two to five slices per animal.

To evoke synaptic responses, a bipolar tungsten stimulation electrode was placed in the center of LA. During the recordings, electrical stimuli (100 µs, 0.06 mA) were applied through this electrode every 15 s. Whole cell recordings were obtained 50–350 µm from the tips of the stimulating electrodes. To isolate inhibitory responses, AMPA/kainate and NMDA receptor antagonists were added to the perfusate [10 µM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and 50 µM dl-2-amino-5-phosphonopentanoic acid (dl-APV)]. To isolate GABA_A IPSCs, the GABA_B antagonist CGP55845 (1 µM) was also added to the perfusate. Voltage-clamp recordings of electrically evoked inhibitory postsynaptic currents (IPSCs) were performed at a holding potential of −50 mV. GABA_A miniature IPSCs (mIPSCs) were recorded in the presence of tetrodotoxin (1 µM) at a holding potential of −80 mV for 1 min. A Picospritzer II (Parker Hannifin, Hollis, NH) was used for pressure application of drugs in the vicinity of recorded neurons. Recordings were obtained with a Multiclamp 700A amplifier (Molecular Devices, Sunnyvale, CA) and digitized at 20 kHz with a Digidata-1550 interface (Molecular Devices) controlled by pClamp-10.3 (Molecular Devices). When mIPSCs were recorded, the data were filtered at 1 kHz.

Drugs.

Depending on the purpose of the experiments, one or more of the following drugs were added to the perfusate: baclofen (20 µM; Tocris, Bristol, UK), CGP55845 (1 µM; Sigma, St. Louis, MO), CNQX (10 µM; Wako, Osaka, Japan), CP93129 (5 µM; Tocris), dl-APV (50 µM; Sigma), DOI (10 µM; Tocris), GABA (50 µM; Wako), picrotoxin (100 µM; Wako), QX-314 (5 mM; Tocris), serotonin (5-HT; 5 µM; Wako), sumatriptan (5 µM; Tocris), tetrodotoxin (TTX; 1 µM; Latoxan, Portes lès Valence, France), and WAY100635 (10 µM; Tocris).

Data analysis.

Data are expressed as means ± SE. Paired or unpaired t tests as well as one-way ANOVA with post hoc Dunnett tests were used for assessing statistical significance using a threshold P < 0.05. For statistical testing and computing linear regressions, we used SPSS (IBM, Chicago, IL) or MATLAB (Mathworks, Natick, MA). mIPSCs were automatically detected offline using the software Mini Analysis (Synaptosoft, Decatur, GA) followed by manual corrections.

RESULTS

5-HT suppresses inhibitory inputs to PNs via 5-HT_1B receptors.

Whole cell recordings of PNs (n = 141) were obtained under visual guidance with infrared and differential interference contrast microscopy. Their electrophysiological characteristics were consistent with previous reports (Faber et al. 2001; Washburn and Moises 1992). Synaptic responses were elicited by electrical stimuli (100 µs, 0.06 mA), delivered every 15 s through bipolar electrodes positioned in the center of LA. To isolate the inhibitory components of these responses, CNQX (10 µM) and dl-APV (50 µM) were added to the perfusate. In five cells, we confirmed that the residual response was abolished by a combination of picrotoxin (100 µM) and CGP55845 (1 µM).

Pharmacologically isolated inhibitory postsynaptic currents (IPSCs) were studied in voltage-clamp mode, from a holding potential of −50 mV (Fig. 1). Addition of 5-HT (5 µM) to the perfusate reduced IPSC amplitudes by 41.6 ± 6.0% (Fig. 1, A and C; n = 10). Although this effect of 5-HT outlasted the period of drug application by several minutes (Fig. 1B), IPSC amplitudes gradually recovered (control, 119.8 ± 11.0 pA; 5-HT, 73.6 ± 12.5 pA; washout, 101.6 ± 11.4 pA; one-way repeated ANOVA; df = 2; F = 3.93; P = 0.03; post hoc Dunnett test, control vs. 5-HT, P = 0.017; control vs. washout, P = 0.445; n = 10 from 5 rats). Consistent with our previous report (Yamamoto et al. 2014), 5-HT also induced an inward current (15.7 ± 6.4 pA; n = 10 from 5 rats).

Fig. 1.

5-HT-induced suppression of inhibitory postsynaptic currents (IPSCs) in principal lateral amygdala (LA) neurons. A: representative IPSC traces induced by local electrical stimulation of LA in the presence of CNQX (10 µM) and dl-APV (50 µM). Bath application of 5-HT (5 µM) reduced IPSC amplitudes. B: averaged time course of normalized IPSC amplitudes before and after 5-HT application from 10 neurons. C: modulation of IPSC amplitudes by 5-HT in 10 neurons. 5-HT significantly reduced IPSC amplitudes (control vs. 5-HT: from 119.8 ± 11.0 to 73.6 ± 12.5 pA; one-way repeated ANOVA; df = 2; F = 3.93; P = 0.03; post hoc Dunnett test, control vs. 5-HT, P = 0.017; control vs. washed out: from 119.8 ± 11.0 to 101.6 ± 11.4 pA; post hoc Dunnett test, control vs. washout, P = 0.445; n = 10). *P < 0.05.

To identify the 5-HT receptor subtype responsible for the reduction of GABA_A IPSCs, we tested the effects of 5-HT agonists that differ in their receptor selectivity, in the presence of the GABA_B antagonist CGP55845 (1 µM). Addition of DOI (10 µM) to the perfusate, a broad 5-HT₂ receptor agonist, had no effect on GABA_A IPSCs (Fig. 2A; paired t test; df = 8; t = −0.88; P = 0.405; n = 9 from 4 rats), while sumatriptan (5 µM), a 5-HT₁ receptor agonist, attenuated GABA_A IPSCs by 37.2 ± 3.6% (Fig. 2B; paired t test; df = 10; t = 11.03; P < 0.001; n = 11 from 5 rats). CP93129 (5 µM), a selective 5-HT_1B receptor agonist, reduced GABA_A IPSCs by 35.5 ± 5.6%, (Fig. 2C; paired t test; df = 7; t = 6.80; P < 0.001; n = 8 from 3 rats), nearly as much as sumatriptan. To determine whether 5-HT_1A receptors are involved in the 5-HT induced suppression of IPSCs, we also tested the effect of WAY100635 (10 µM), a 5-HT_1A receptor antagonist. Even in the presence of WAY100365, 5-HT reduced the amplitude of IPSCs by 39.2 ± 2.7% (Fig. 2D; paired t test; df = 5; t = 9.15; P < 0.001; n = 6 from 3 rats). Together, these results indicate that 5-HT_1B receptors mediate the inhibition of GABA_A IPSCs by 5-HT.

Fig. 2.

5-HT_1B receptor subtype mediates the GABA_A inhibitory postsynaptic current (IPSC) reduction by 5-HT. The GABA_A IPSCs were isolated with CNQX (10 µM), dl-APV (50 µM), and CGP55845 (1 µM). A–D: experiments testing the effects of the broad 5-HT₂ receptor antagonist DOI (10 µM; A), the 5-HT₁ receptor agonist sumatriptan (5 µM; B), the 5-HT_1B receptor agonist CP93129 (5 µM; C), and 5-HT in the presence of the 5-HT_1A receptor antagonist WAY100635 (D; 10 µM). For each type of experiment, the left shows representative IPSC traces (black: control; red: drug), the middle shows averaged time course of normalized IPSC amplitudes, and the right shows the result obtained in all tested cells (A, n = 9; B, n = 11; C, n = 8; D, n = 6). The broad 5-HT₂ receptor agonist DOI (10 µM) failed to alter IPSC amplitudes (A; from 138.0 ± 13.1 to 140.7 ± 12.9 pA; paired t test; df = 8; t = −0.88; P = 0.405, n = 9). The 5-HT₁ receptor agonist sumatriptan reduced the IPSC amplitudes (B; from 115.6 ± 9.9 to 73.1 ± 8.4 pA; paired t test; df = 10; t = 11.03; P < 0.001; n = 11). The 5-HT_1B receptor agonist CP93129 reduced the IPSC amplitudes to the same extent as sumatriptan (C; from 109.8 ± 13.1 to 74.0 ± 14.3 pA; paired t test; df = 7; t = 6.80; P < 0.001; n = 8). 5-HT reduced the IPSC amplitude in the presence of Way100635 (D; from 119.6 ± 7.6 to 73.2 ± 5.7 pA; paired t test; df = 5; t = 9.15; P < 0.001; n = 6). *P < 0.05.

The reduction of GABA_A IPSCs by 5-HT depends on a presynaptic mechanism.

To test whether the reduction of GABA_A IPSCs by 5-HT depends on pre- or postsynaptic mechanisms, we used two approaches. In the first, we examined the effects of 5-HT on the paired-pulse ratio. In the second, we tested whether 5-HT also reduced IPSCs elicited by exogenous application of GABA. Both tests were carried out in the presence of the GABA_B antagonist CGP55845 (1 µM).

For the first test, we elicited IPSCs by applying two identical electrical stimuli in rapid succession (50-ms interstimulus interval). Addition of 5-HT (5 µM) to the perfusate attenuated the GABA_A IPSCs evoked by the first stimulus by 40.2 ± 3.4% (paired t test; df = 10; t = 8.69; P < 0.001; n = 11 from 4 rats; Fig. 3, A and B), similar to the effect observed in our first experiment. In contrast, IPSCs elicited by the second stimulus were more resistant to 5-HT (reduction of 33.9 ± 3.5%; paired t test; df = 10; t = −2.49; P = 0.032; n = 11 from 4 rats) such that the amount of paired-pulse depression was reduced (Fig. 3, C and D; IPSC2/IPSC1; from 0.84 ± 0.02 to 0.93 ± 0.02; paired t test; df = 10; t = −2.58; P = 0.027; n = 11 from 4 rats).

Fig. 3.

Presynaptic origin of the suppression of GABA_A inhibitory postsynaptic currents (IPSCs) by 5-HT. The GABA_A IPSCs were isolated with CNQX (10 µM), dl-APV (50 µM), and CGP55845 (1 µM). A: representative traces of paired-pulse IPSCs in the absence (black) and presence (red) of 5-HT. B: amplitude of the first IPSC induced by paired-pulse stimulation in control conditions and in the presence of 5-HT (from 170.0 ± 11.3 to 101.0 ± 8.7 pA; paired t test; df = 10; t = 8.69; P < 0.001; n = 11). The extent of suppression was much the same as for the single pulse IPSCs. C: scaled traces for comparing the paired-pulse ratio. D: paired-pulse-ratio in 11 tested neurons, in the absence (black) or presence (red) of 5-HT. The paired-pulse ratio was significantly increased by 5-HT (IPSC2/IPSC1; from 0.84 ± 0.02 to 0.93 ± 0.02; paired t test; df = 10; t = −2.58; P = 0.027; n = 11). E: representative traces of GABA_A current induced by a direct GABA (50 µM) puff to the recorded neuron. F: amplitude of GABA_A current induced by direct GABA puff applications in 6 tested neurons (from 114.2 ± 14.0 to 111.7 ± 14.03 pA, paired t test; df = 5; t = 0.85; P = 0.431, n = 6). *P < 0.05.

In prior experiments with such paired-stimulus paradigms, it was observed that depending on the synapse type, the second stimulus elicited PSCs of higher or lower amplitude (paired-pulse facilitation or depression, respectively) and that manipulations that increased release probability decreased the amount of facilitation (or enhanced the depression) and conversely (Manabe et al. 1993). As a result, transmitter release probability is thought to be inversely proportional to the amount of facilitation (or proportional to the amount of depression).

Thus the above results suggest that the reduction of IPSCs by 5-HT depends on a presynaptic mechanism, namely a reduction of GABA release. If this interpretation is correct, one would expect GABA_A IPSCs evoked by exogenous GABA to be insensitive to 5-HT. To test this prediction, we performed local pressure application of GABA in the vicinity of recorded cells (see materials and methods). As predicted, addition of 5-HT (5 µM) to the perfusate did not alter the IPSCs elicited by exogenous GABA (Fig. 3, E and F; from 114.2 ± 14.0 to 111.7 ± 14.03 pA; paired t test; df = 5; t = 0.85; P = 0.431; n = 6 from 4 rats), again supporting the conclusion that 5-HT presynaptically inhibits IPSCs by reducing GABA release.

To test this idea further, we examined the effects of 5-HT on pharmacologically isolated miniature GABA_A IPSCs (mIPSCs) recorded from a holding potential of −80 mV and in the presence of tetrodotoxin (1 µM) and CGP55845 (1 µM). In this experiment, we used a cesium-based internal solution to improve space clamp. Application of 5-HT (5 µM) decreased the frequency of mIPSCs by 18.1 ± 4.2% (Fig. 4, A and C1; paired t test; df = 7; t = 3.01; P = 0.020; n = 8 from 5 rats), leaving their amplitude and decay time unchanged (Fig. 4, C2, C3, and E; amplitude, 3.1 ± 3.0%; paired t test; df = 7; t = 0.98; P = 0.362; decay, −7.0 ± 3.1%; paired t test; df = 7; t = −1.89; P = 0.100; n = 8 from 5 rats). Moreover, the 5-HT_1B receptor agonist CP93129 (5 µM) yielded qualitatively identical results. That is, CP93129 decreased mIPSC frequency by 19.7 ± 5.5% (Fig. 4, B and D1; paired t test; df = 7; t = 3.32; P = 0.013; n = 8 from 4 rats) but did not alter their amplitude or decay time (Fig. 4, D2, D3, and F; amplitude, 0.3 ± 3.3%; paired t test; df = 7; t = 0.02; P = 0.982; decay, −2.3 ± 1.3%; paired t test; df = 7; t = −1.53; P = 0.169; n = 8 from 4 rats). We also examined the effect of 5-HT and CP93129 on the distribution of mIPSC decay time, but no change was noted (Fig. 4, E and F).

Fig. 4.

Effect of 5-HT on GABA_A miniature inhibitory postsynaptic currents (mIPSCs). Recordings were performed in the presence of TTX (1 µM), CNQX (10 µM), dl-APV (50 µM), and CGP55845 (1 µM). A and B: representative traces of mIPSCs before (top) and after (bottom) 5-HT (5 µM; A) or CP93129 (5 µM; B) application. V_H, holding potential. C1–3: 5-HT reduced the frequency of mIPSC (C1; from 2.3 ± 0.3 Hz to 1.9 ± 0.3 Hz; paired t test; df = 7; t = 3.01; P = 0.020; n = 8) but did not alter their amplitude (C2; from 44.2 ± 2.0 pA to 42.7 ± 2.1 pA; paired t test; df = 7; t = 0.98; P = 0.362; n = 8) and decay time of mIPSC (C3; from 7.4 ± 0.3 ms to 7.8 ± 0.2 ms; paired t test; df = 7; t = −1.89; P = 0.100; n = 8). D1–3: CP93129 reduced the frequency of mIPSC (D1; from 2.9 ± 0.4 Hz to 2.3 ± 0.4 Hz; paired t test; df = 7; t = 3.32; P = 0.013; n = 8) but did not alter their amplitude (D2; from 47.4 ± 2.9 to 47.4 ± 3.6 pA; paired t test; df = 7; t = 0.02; P = 0.982; n = 8) and decay time of mIPSC (D3; from 7.9 ± 0.4 to 8.0 ± 0.4 ms; paired t test; df = 7; t = −1.53; P = 0.169; n = 8). E and F: cumulative probability distribution of the decay time of mIPSCs in control conditions (black) and after application (red) of 5-HT (E) or CP93129 (F). *P < 0.05.

These results further support the conclusion that 5-HT decreases GABA release probability by acting on presynaptic 5-HT_1B receptors.

GABA_B IPSCs are also regulated by a postsynaptic action of 5-HT.

Collectively, the above experiments indicate that 5-HT causes a reduction in the amplitude of GABA_A IPSCs by activating presynaptic 5-HT_1B receptors that reduce GABA release. Thus, in principle, 5-HT should also reduce GABA_B IPSCs and to a similar extent as for GABA_A IPSCs. To test this prediction, we examined the effect of 5-HT on pharmacologically isolated GABA_B IPSCs; that is, IPSCs we recorded in the presence of CNQX (10 µM), dl-APV (50 µM), and picrotoxin (100 µM). However, when elicited by single electrical stimuli, the residual (GABA_B) IPSCs were unstable and of low amplitude (Fig. 5A, inset). Repetitive electrical stimuli (5 stimuli at 25 Hz) were required to elicit robust GABA_B currents. As expected, 5-HT attenuated these synaptically evoked GABA_B IPSCs but by only 31.7 ± 1.1% (Fig. 5, A and B; paired t test; df = 7; t = 10.61; P < 0.001; n = 8 from 3 rats), a smaller reduction than seen with GABA_A IPSCs (41.6 ± 6.0%) elicited by single electrical stimuli. When we compared the 5-HT-induced reduction of GABA_A IPSCs elicited by repetitive stimuli, as in the GABA_B experiments, the difference was even larger: the GABA_A IPSC was reduced by 45.9 ± 0.1% whereas the GABA_B IPSC was reduced by 31.7 ± 1.1% (t test; df = 10; t = 2.47; P = 0.033).

Fig. 5.

The enhancement of the GABA_B current by 5-HT is mediated postsynaptically. Recordings in A–D were performed in the presence of picrotoxin (100 µM). A–C: experiments testing the effects of 5-HT (red, 5 µM; control, black) on electrically evoked (5 stimuli at 25 Hz) GABA_B inhibitory postsynaptic currents (IPSCs; A) and on IPSCs evoked by pressure application of GABA (50 µM; B) or baclofen (20 µM; C). GABA_A and glutamate receptor antagonists were present throughout (see materials and methods). For each type of experiments, the left shows traces from a representative cell whereas the right shows the results obtained in all tested cells (A, n = 8; B, n = 5; C, n = 5). While 5-HT significantly reduced the amplitude of electrically evoked GABA_B IPSCs (A: from 39.1 ± 6.7 to 27.1 ± 7.5 pA; paired t test; df = 7; t = 10.61; P < 0.001, n = 8), it enhanced GABA_B responses to exogenous GABA (B; from 71.9 ± 23.0 pA to 104.0 ± 25.8 pA; paired t test; df = 4; t = −3.82; P = 0.019; n = 5) and baclofen (C; control, 33.3 ± 1.3 pA; baclofen, 46.3 ± 3.3 pA; washout, 38.0 ± 2.2 pA; one-way repeated ANOVA; df = 2; F = 7.703; P = 0.03; post hoc Dunnett test, control vs. baclofen, P = 0.004; control vs. washout, P = 0.302; n = 5). A, inset: representative examples of GABA_B currents evoked by single versus repetitive electrical stimuli. *P < 0.05.

This discrepancy suggested that the regulation of GABA_B IPSCs by 5-HT might be more complex than for GABA_A currents. To examine this possibility, in the presence of picrotoxin (100 µM), we performed local pressure applications of GABA in the vicinity of recorded cells, allowing us to examine the effects of 5-HT on GABA_B currents, independently of any presynaptic effects. Addition of 5-HT (5 µM) to the perfusate enhanced postsynaptic GABA_B currents by 57.1 ± 21.5% (Fig. 5, C and D; paired t test; df = 4; t = −3.82; P = 0.019; n = 5 from 3 rats).

To obtain an independent corroboration of this finding, we also examined the effect of 5-HT on GABA_B currents elicited by pressure application of baclofen. In control conditions, a brief pulse of baclofen (0.5 s) induced a long-lasting (>30 s) GABA_B current (Fig. 5E; control). Addition of 5-HT to the perfusate enhanced the GABA_B current by 39.4 ± 9.5%, and the current returned to control values after washout (Fig. 5, E and F; control, 33.3 ± 1.3 pA; baclofen, 46.3 ± 3.3 pA; washout, 38.0 ± 2.2 pA; one-way repeated ANOVA; df = 2; F = 7.703; P = 0.03; post hoc Dunnett test, control vs. baclofen, P = 0.004; control vs. washout, P = 0.302; n = 5 from 2 rats). Together, these results indicate that 5-HT enhances GABA_B currents postsynaptically, thereby partially counteracting the reduction caused by the presynaptic inhibition of GABA release.

Negative correlation between GABA_A IPSC amplitudes and their attenuation by 5-HT.

The above experiments indicate that GABA_B currents are more resistant to the attenuating effects of 5-HT than GABA_A IPSCs. However, in contrast with GABA_A IPSCs, repetitive stimuli were required to elicit robust GABA_B currents (Fig. 6A, inset), raising the possibility that 5-HT effects on GABA_A currents might differ depending on IPSC amplitudes. These considerations led us to revisit our data set on the attenuation of electrically evoked GABA_A IPSCs by 5-HT (n = 20 from 9 rats) and CP93129 (n = 8; recorded in Fig. 2). Specifically, we examined the relationship between the initial amplitude of GABA_A IPSCs and the magnitude of their attenuation by 5-HT and CP93129 in the presence of the GABA_B antagonist CGP55845 (1 µM). Indeed, even though we used a fixed stimulation intensity, IPSC amplitudes varied widely between cells (72.5 to 216.2 pA; n = 28 from 13 rats). This effect was strongly related to the distance between the stimulating electrodes and recorded neurons: the lower the distance, the higher the IPSC amplitude (Pearson r = –0.60; P = 0.004; n = 20 from 4 rats).

Fig. 6.

High-amplitude inhibitory postsynaptic currents (IPSCs) are less sensitive to 5-HT than low-amplitude IPSCs. Recordings were performed in the presence of CNQX (10 µM), dl-APV (50 µM), and CGP55845 (1 µM). A: representative traces showing IPSCs in control conditions (black) and in the presence of 5-HT agonists (5- HT; 5 µM, n = 20; CP93129; 5 µM; n = 8; red) in cells with high- (top) and low-amplitude (bottom) IPSCs. The low-amplitude IPSC was suppressed to a greater extent by the 5-HT agonists than the high-amplitude IPSC. B: IPSC suppression ratio (y-axis) as a function of control IPSC amplitude (x-axis). C: differing time course of low (blue)- and high-amplitude (red) IPSCs. Low amplitude IPSC was scaled up to match the peak amplitude of a high-amplitude IPSC, highlighting the slower decay time of low-amplitude IPSCs. D: IPSC half-width (y-axis) as a function of IPSC amplitude (x-axis). E: for high (left)- and low-amplitude (right) IPSCs, 3 superimposed traces are shown: control IPSC (black), IPSC in the presence of 5-HT agonist (red), and 5-HT-sensitive component (blue). The 5-HT-sensitive IPSC was estimated by subtracting the 5-HT-depressed IPSC from the initial IPSC. F: for high (left)- and low-amplitude (right) IPSCs, we show 3 superimposed traces scaled so that their peak amplitude would match: control IPSC (black), IPSC in the presence of the 5-HT agonist (red) and 5-HT-sensitive component (blue). G: averaged IPSCs recorded in the experiments testing the effects of 5-HT (left) or CP93129 (right), irrespective of control IPSC amplitudes. In both cases, we show 3 superimposed traces scaled so that their peak amplitude would match: control IPSC (black), IPSC in the presence of 5-HT agonist (red), and the difference between these 2 traces.

Unexpectedly, the impact of 5-HT and CP93129 varied depending on control IPSC amplitudes. As shown in the representative examples of Fig. 6A, control IPSCs of high amplitude were less sensitive to 5-HT than IPSCs of low amplitude. In fact, there was a significant negative correlation between initial IPSC amplitudes and their sensitivity to 5-HT and CP93129 (Fig. 6B; Pearson r = –0.47; P = 0.012; n = 28 from 13 rats).

The contrasting sensitivity of high versus low-amplitude GABA_A IPSCs to 5-HT and CP93129 suggests that they are mediated by different complements of GABAergic synapses, with a larger proportion of the synapses generating low-amplitude IPSCs expressing 5-HT_1B receptors than in high-amplitude IPSCs and a greater fraction of those mediating high-amplitude IPSCs devoid of 5-HT_1B receptors than in low-amplitude IPSCs. To further characterize the two sets of inhibitory synapses, we first examined the time course of the pharmacologically isolated GABA_A IPSCs and determined if it correlated with their amplitude. As shown in Fig. 6D, IPSC durations (decay time) correlated negatively with their amplitude (Pearson r = –0.45; P = 0.017; n = 28 from 13 rats), whereas their rise time (20–80%) did not (Pearson r = 0.12; P = 0.528; n = 28 from 13 rats).

Furthermore, 5-HT and CP93129 accelerated the decaying phase of the IPSCs (from 18.9 ± 1.5 to 14.6 ± 1.4 ms; paired t test; df = 27; t = 6.51; P < 0.001; n = 28 from 13 rats). This is evident in Fig. 6, E and F, which illustrates, for the 14 highest (left) and 14 lowest (right) amplitude IPSCs, the control IPSC, the IPSC in the presence of 5-HT or CP93129, and the difference between the two (i.e., the 5-HT-sensitive component). As shown in Fig. 6, in both subsets of cells, the 5-HT-sensitive component had a markedly slower time course than the 5-HT-resistant component (the decay time was ~12 ms longer; 26.5 ± 2.1 vs. 14.6 ± 0.9 ms; paired t test; df = 27; t = −5.82; P < 0.001; n = 28 from 13 rats). Importantly, while the amplitude of the 5-HT-sensitive current (Table 1) did not differ between the two subsets of cells (131.8%; high, 51.5 ± 5.2; low, 39.1 ± 3.7 pA; t test, df = 26; t = 1.96; P = 0.06), that of the 5-HT-resistant component was more than twice higher (219.2%) in cells with high (120.0 ± 9.3 pA) than low (54.7 ± 4.4 pA) IPSC amplitudes (t test; df = 26; t = 6.34; P < 0.001). We also analyzed the effect of 5-HT (5 µM) and CP93129 (5 µM) on the decay time of IPSCs recorded separately. With 5-HT, the decay time of GABA_A IPSCs became faster (Fig. 6G, left; from 18.4 ± 1.4 to 14.2 ± 1.3 ms; paired t test; df = 19; t = 4.86; P < 0.001; n = 20 from 9 rats). With CP93129, also the decay time of GABA_A IPSCs became faster (Fig. 6G, right; from 20.2 ± 1.0 to 15.7 ± 0.7 ms; paired t test; df = 7; t = 5.17; P < 0.002; n = 8 from 4 rats). This result indicates the serotonergic modulation of the IPSCs time course is also dependent on 5-HT_1B receptors.

Table 1.

Properties of low- and high-amplitude IPSCs in control conditions in the presence of drugs as well as the difference between them

	Amplitude, pA	Rise Time, ms	Decay Time, ms	Half Width, ms
Control IPSCs
High	167.2 ± 7.5*	1.91 ± 0.15	17.3 ± 1.5	14.3 ± 1.0
Low	92.3 ± 3.8*	2.21 ± 0.13	20.6 ± 1.2	16.2 ± 0.9
Drug-resistant component
High	120.0 ± 9.3*	1.76 ± 0.16	13.7 ± 1.5	12.5 ± 0.9
Low	54.7 ± 4.4*	2.05 ± 0.18	15.5 ± 1.2	14.0 ± 0.8
Drug-sensitive component
High	51.5 ± 5.2	2.64 ± 0.35	24.9 ± 3.7	19.7 ± 2.9
Low	39.1 ± 3.6	2.32 ± 0.22	28.1 ± 1.9	20.1 ± 1.6

Values are means ± SE for control conditions in the presence of drugs (5-HT or CP93129; drug-resistant component) and the difference between them (drug-sensitive component); n = 14. In cells with low- and high-amplitude inhibitory postsynaptic currents (IPSCs), the 5-HT-sensitive component had a slower decay time than the 5-HT-resistant component (P < 0.001; see detailed statistics in main text). Whereas the amplitude of the 5-HT-sensitive current did not differ between the 2 subsets of cells, that of the 5-HT-resistant component was significantly higher in cells with high than low IPSC amplitudes (P < 0.001). CP93129 also shortened the decay time of GABA_A IPSCs (P < 0.002).

^*P < 0.05, statistically significant differences between high vs. low.

Together, these findings suggest that high- and low-amplitude control IPSCs both contain a fast, relatively 5-HT-insensitive component and a slower, 5-HT-sensitive component. The fast, 5-HT-resistant component accounts for a larger portion of the overall IPSC in cells with high- compared with low-amplitude IPSCs. The slow, 5-HT-sensitive component accounts for a larger portion of the overall IPSC in cells with low- compared with high-amplitude IPSCs. The absolute size of the slow component is about the same for both the large and small control IPSCs. However, because the fast component is much larger in high-amplitude IPSCs, they appear relatively resistant to 5-HT.

DISCUSSION

This study was undertaken to shed light on the regulation of GABAergic inhibition by 5-HT in LA. The interest of this question stems from prior observations indicating that inhibition is a major determinant of PN activity in LA (reviewed in Paré et al. 2003) and converging lines of evidence implicating a dysregulation of the amygdala, in part mediated by 5-HT, in pathological anxiety (Bandelow et al. 2017; Bremner 2005; LeDoux 2000; Lowry et al. 2005; Shin et al. 2006).

Our results indicate that by binding to presynaptic 5-HT_1B receptors and inhibiting GABA release, 5-HT attenuates GABAergic IPSCs. However, this effect was more evident when monitoring GABA_A than GABA_B IPSCs because 5-HT also caused a postsynaptic enhancement of GABA_B currents, which partially counteracted for the presynaptic inhibition of GABA release. Unexpectedly, we also observed that 5-HT differentially regulates distinct subsets of GABAergic synapses. That is, 5-HT did not seem to have much influence on a set of inhibitory inputs that mediates fast GABA_A IPSCs but strongly inhibited another one that generates slower GABA_A responses.

Prior work on the regulation of neuronal excitability by 5-HT in the basolateral complex of the amygdala.

The regulation of neuronal excitability by 5-HT is extremely complex because the density of 5-HT inputs varies between structures (Steinbusch 1981) and the expression of the many 5-HT receptor subtypes is heterogeneous, regionally as well as between cell types (for example, see Andrade 1998; Chameau and van Hooft 2006; Palacios 2016). The basolateral complex of the amygdala is no exception. For instance, a higher density of 5-HT axons is found in the BL than LA (Muller et al. 2007b; Sengupta et al. 2017) and the concentration of 5-HT receptors varies between these two nuclei. For instance, 5-HT_2C receptors are more concentrated in LA than BL (Greenwood et al. 2012), whereas the opposite is true for 5-HT_1A and 5-HT_1B receptors (Saha et al. 2010). There is also heterogeneity at the level of single cells. For instance, in BL at least, 5-HT_2A receptors are expressed by PNs and parvalbumin-positive interneurons but not by cholecystokinin interneurons (McDonald and Mascagni 2007). In contrast, 5-HT_3A receptors are only found in a specific subtype of local-circuit cell that does not express calcium-binding proteins (Mascagni and McDonald 2007; see Bocchio et al. 2016 for an extensive description of the cell-type specific expression of 5-HT receptors).

Expectedly given these variations, electrophysiological studies on the effect of 5-HT in LA and BL have yielded different results. In LA PNs, 5-HT was found to exert a variety of excitatory effects including a depolarization, the inhibition of Ca²⁺-dependent K⁺ currents, and the enhancement of depolarization-induced slow afterdepolarizations (Faber and Sah 2002; Yamamoto et al. 2012, 2014). In BL by contrast, optogenetic excitation of 5-HT terminals inhibited most PNs, an effect mediated by the activation of 5-HT_1A receptors combined with a direct excitation of local-circuit cells (Sengupta et al. 2017). Indeed, the optogenetic activation of 5-HT terminals excited a large proportion of interneurons via 5-HT_2A receptors; only a minority of interneurons were inhibited and via 5-HT_1A receptors (Sengupta et al. 2017). Consistent with this, the pharmacological activation of 5-HT_2A receptors depolarized interneurons in the BLA (Bocchio et al. 2015; Jiang et al. 2009).

In addition to the postsynaptic actions reviewed above, 5-HT also acts presynaptically to regulate transmitter release in the basolateral complex of the amygdala. Indeed, an electron microscopic study found numerous axon terminals immunopositive for 5-HT in close apposition with nonserotonergic axon terminals (Muller et al. 2007b). Consistent with this, several studies reported that 5-HT presynaptically inhibits (Cheng et al. 1998; Rainnie 1999; Yamamoto et al. 2012) transmitter release in the basolateral complex of the amygdala.

5-HT inhibits GABA release via presynaptic 5-HT_1B receptors.

In LA and BL, the influence of 5-HT over glutamate release appears to be always inhibitory (Cheng et al. 1998; Guo et al. 2017; Yamamoto et al. 2012). While several candidate 5-HT receptors were identified for this effect (Cheng et al. 1998; Guo et al. 2017; Yamamoto et al. 2012), the receptors responsible for the modulation of GABA release remained unknown. By contrasting the impact of a variety of 5-HT receptor agonists and antagonists, our study established that 5-HT_1B receptors mediate the inhibitory influence of 5-HT on GABA release in LA.

The 5-HT_1B receptor is a G_i/o-coupled receptor that reduces neural activity, is predominantly found on axon terminals (Boschert et al. 1994) and is similar in structure to the 5-HT_1A receptor, another G_i/o-coupled receptor (Hoyer et al. 2002). In many other brain regions, activation of 5-HT_1B receptors causes a suppression of both excitatory and inhibitory synaptic transmission. These include the suprachiasmatic nucleus (Pickard et al. 1999), basal forebrain (Nishijo and Momiyama 2016), cerebellar nuclei (Saitow et al. 2009), bed nucleus of stria terminalis (Guo and Rainnie 2010), and various cortical regions (Kjaerby et al. 2016; Liao and Lee 2014; Mlinar et al. 2003). In all these cases, the suppression depended on a presynaptic mechanism.

Our findings in LA agree with these earlier observations. Indeed, we also found that the 5-HT_1B receptor-dependent suppression of GABA transmission is mediated presynaptically. That is, even though 5-HT reduced electrically evoked GABA_A IPSCs, it did not alter the inhibitory responses elicited by pressure application of GABA. Moreover, we found that 5-HT reduced the amount of paired-pulse depression normally exhibited by electrically evoked IPSCs and the frequency (but not amplitude) of GABA_A mIPSCs.

Postsynaptic enhancement of GABA_B IPSCs by 5-HT.

We noticed that GABA_A and GABA_B IPSCs were not reduced to the same extent by 5-HT, which seemed inconsistent with the notion that 5-HT acts presynaptically, by inhibiting GABA release. However, control experiments revealed that GABA_B IPSCS were also potentiated by 5-HT through a postsynaptic mechanism. Indeed, when GABA_B currents were induced by directly puffing GABA or baclofen, 5-HT actually enhanced GABA_B currents. Together, these results suggest that GABA_B currents are partially spared from the 5-HT-induced suppression of GABA release by a competing postsynaptic enhancement of GABA_B currents.

While the 5-HT receptors responsible for this effect remain unknown, it should be noted that there are precedents in the literature for this effect (Jacobson et al. 2017; Strahlendorf et al. 1991). Also, it seems unlikely that the direct convergence of 5-HT and GABA acting via GABA_B receptors on the same K⁺ conductance (Andrade et al. 1986) is responsible for the postsynaptic enhancement of GABA_B currents since this mechanism would be expected to reduce GABA_B currents through occlusion.

A clue pointing to the possible significance of this dual regulation of GABA_B currents by 5-HT comes from the contrasting intensity of the afferent drive required to elicit GABA_A compared with GABA_B IPSCs. Whereas a single low-intensity stimulus reliably elicits GABA_A IPSCs, activation of robust GABA_B currents requires repetitive stimulation. Thus the dual regulation of GABA_B IPSCs might preserve the dynamics of inhibitory tone in LA.

5-HT differentially regulates distinct subsets of GABAergic synapses.

Several observations suggest that the electrical stimuli we used to elicit IPSCs recruited different contingent of GABAergic synapses and that 5-HT regulates them differentially. First, depending on the distance between the stimulating electrodes and recorded cells, IPSCs varied in amplitude, with cells closer to the stimulating electrodes exhibiting higher amplitude IPSCs than more distant neurons. Second, 5-HT caused a greater attenuation of low- than high-amplitude IPSCs. Third, for both low- and high amplitude IPSCs, the time course of the 5-HT-sensitive and -resistant IPSCs differed markedly: the 5-HT-insensitive component rose (~2 ms) and decayed faster (~15 ms) than the 5-HT-sensitive component (~3 and 30 ms, respectively).

The differing time course of the 5-HT-resistant and -sensitive components is reminiscent of the GABA_A,fast and GABA_A,slow first described in the hippocampus (Banks et al. 1998; Pearce 1993) and later in a variety of other brain regions (reviewed in Capogna and Pearce 2011). In these various structures, mounting evidence suggests that GABA_A,slow is generated by dendrite-targeting interneurons such as the NPY-expressing neurogliaform (NGF) cells (Price et al. 2005; Szabadics et al. 2007). These cells are unusual because they have an extensive axonal field that often crosses regional boundaries. Various mechanisms have been invoked to explain the prolonged time course of GABA_A,slow (reviewed in Armstrong et al. 2012), including dendritic filtering, asynchronous vesicular release, GABA spillover, and differences in GABA receptor subunit composition. While these factors may contribute to explain the slow time course of GABA_A,slow, the main one appears to be the spatiotemporal profile of GABA concentration at the synapse (Karayannis et al. 2010; Krook-Magnuson and Huntsman 2007; Szabadics et al. 2007). Indeed, many of the synapses formed by NGF cells are appositions rather than conventional synapses with narrow clefts, such that the GABA transient is prolonged (Oláh et al. 2009; Szabadics et al. 2007).

Since LA also contains NGF neurons (McDonald 1984), could they be responsible for the 5-HT-sensitive GABA_A,slow we observed? In support of this possibility, an electron microscopic study in the basolateral complex of the amygdala reported that 5-HT-positive axonal boutons contact NPY-positive somata (Bonn et al. 2013). Moreover, NGF cells have an extensive axonal arbor, contribute frequent en passant axonal varicosities, and form many appositions, as opposed to conventional synaptic contacts, which should prolong GABA transients (Mańko et al. 2012). Consistent with this, it was reported that the inhibitory responses elicited by the activation of NGF cells last longer than those evoked by other interneuronal classes of the basolateral amygdala (Mańko et al. 2012; reviewed in Capogna 2014).

While LA also contains NGF neurons (McDonald 1984), lacking experiments directly testing the impact of 5-HT on the IPSCs generated by NGF cells, we can only speculate as to their involvement in the genesis of the slow 5-HT-sensitive IPSC component. Furthermore, it is currently unclear why the slow IPSC component accounted for a greater proportion of the IPSCs seen in PNs distant from our stimulating electrodes. We presume that it reflects differences in the axonal arbors of the various interneuronal subtypes, with those devoid of 5-HT_1B receptors having narrower axonal fields.

An important challenge for future investigations will be to ascertain whether NGF cells play a critical role in the regulation of LA by 5-HT. This could be achieved by examining whether NPY+ axon terminals express 5-HT_1B receptors and by studying the impact of 5-HT on the IPSCs generated by optogenetically excited NGF cells.

GRANTS

This work was supported partly by KAKENHI Grants 18K07572 and 18KK0468 (to R. Yamamoto), 19K16192 (to T. Furuyama), 19K09918 (to M. Ono), and 17H02223 (to N. Kato), Kanazawa Medical University Grant C2018-1 (to N. Kato), a grant from the Nakatomi Foundation (to R. Yamamoto), and National Institute of Mental Health Grant MH-119854 (to D. Pare).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.Y. conceived and designed research; R.Y. and T.F. performed experiments; R.Y., T.F., T.S., D.P., and N.K. analyzed data; R.Y., T.S., M.O., D.P., and N.K. interpreted results of experiments; R.Y. prepared figures; R.Y., D.P., and N.K. drafted manuscript; R.Y. and D.P. edited and revised manuscript; R.Y., T.F., T.S., M.O., D.P., and N.K. approved final version of manuscript.