Kv1.1 channels mediate network excitability and feed-forward inhibition in local amygdala circuits

Kv1.1 containing potassium channels play crucial roles towards dampening neuronal excitability. Mice lacking Kv1.1 subunits (Kcna1−/−) display recurrent spontaneous seizures and often exhibit sudden unexpected death. Seizures in Kcna1−/− mice resemble those in well-characterized models of temporal lobe epilepsy known to involve limbic brain regions and spontaneous seizures result in enhanced cFos expression and neuronal death in the amygdala. Yet, the functional alterations leading to amygdala hyperexcitability have not been identified. In this study, we used Kcna1−/− mice to examine the contributions of Kv1.1 subunits to excitability in neuronal subtypes from basolateral (BLA) and central lateral (CeL) amygdala known to exhibit distinct firing patterns. We also analyzed synaptic transmission properties in an amygdala local circuit predicted to be involved in epilepsy-related comorbidities. Our data implicate Kv1.1 subunits in controlling spontaneous excitatory synaptic activity in BLA pyramidal neurons. In the CeL, Kv1.1 loss enhances intrinsic excitability and impairs inhibitory synaptic transmission, notably resulting in dysfunction of feed-forward inhibition, a critical mechanism for controlling spike timing. Overall, we find inhibitory control of CeL interneurons is reduced in Kcna1−/− mice suggesting that basal inhibitory network functioning is less able to prevent recurrent hyperexcitation related to seizures.

www.nature.com/scientificreports/ neuronal cell loss in the CA1/CA3 region, degeneration of hilar interneurons, and sprouting of mossy fibers 19 . In vitro hippocampal slice recordings from Kcna1 −/− mice showing pathological high-frequency oscillations suggesting that network abnormalities contribute to seizure generation 22 . Given the role of the hippocampus in cognitive function e.g., learning and memory, deficits in hippocampal function are proposed to contribute to cognitive impairments associated with epilepsy 23 . Studies have shown that the amygdala is a critical component for the onset and propagation of temporal lobe seizures 24,25 . Moreover, animal models of TLE show extensive neuropathology signs in the amygdala circuitry resembling those in human TLE 24 , including extensive loss of GABAergic neurons 26 . Additional evidence for a central role of the amygdala in the generation and propagation of seizure activity comes from kindling models of epilepsy in which the amygdala exhibits extreme susceptibility to electrically induced seizures 27 . Of the various amygdalar nuclei, the basolateral amygdala (BLA) both exhibits K v 1.1 immunoreactivity 28 and has the highest propensity to generate seizures 29 . At the molecular and cellular levels, Kcna1 −/− mice experiencing status epilepticus display extensive neuronal cell loss and gliosis in the BLA region 19 and spontaneous seizures resulting in significantly enhanced cFos expression in BLA neurons 20 . Kcna1 −/− mice also possess an enlarged amygdala correlated with seizure activity 30,31 . Overall however, disturbances in non-hippocampal temporal lobe network regions, and their functional contributions to epileptogenesis and behavioral impairments associated with seizures in Kcna1 −/− mice have remained relatively unexplored.
The central amygdala provides the main output of the amygdala complex, projecting to the hypothalamus and brainstem regions involved in the expression of emotional and autonomic responses 32,33 . Within the central amygdala, the central lateral subdivision (CeL) has received increasing interest due to its widespread function towards mediating fear responses 34 , anxiety 35,36 and nociception 37,38 . Dysfunction of the amygdala has been linked to epilepsy-related mood disorders, contributing to the high comorbid occurrence of anxiety and depression in patients with epilepsy 39 . Moreover, it has been suggested that the amygdala is also involved in respiratory abnormalities associated with epilepsy 40 . Notably, Kcna1 −/− mice demonstrate respiratory failures including central apnea during seizures that can result in the phenomenon known as sudden unexpected death in epilepsy (SUDEP) [16][17][18] . A recent hypothesis proposes that spreading depolarization and subsequent brainstem dysfunction triggers respiratory failure in Kcna1 −/− mice 16 . Further, studies from both SUDEP patients [41][42][43] and mice 44 suggest that the amygdala, specifically the central amygdala, has a significant effect on seizure-induced apneas.
Here, we examined whether electrophysiologically distinct neuronal cell types from the CeL and BLA nuclei of Kcna1 −/− mice exhibit alterations in intrinsic excitability and synaptic activity. Since the BLA appears to be a key site for seizure initiation in TLE, we first examined the functional characteristics of BLA pyramidal neurons and found that the lack of K v 1.1 subunits alters spontaneous synaptic activity in BLA pyramidal neurons with no changes in firing responses evoked by somatic stimulation. We also provide evidence for the loss of K v 1.1 subunits towards enhancing intrinsic excitability and impairing feed-forward inhibition into CeL GABAergic neurons. Together, we predict that these changes in synaptic and circuit properties contribute to neuronal hyperexcitability in Kcna1 −/− mice and to promote seizure activity.

Results
Loss of K v 1.1 subunits in BLA pyramidal neurons enhances excitatory synaptic activity but does not affect inhibitory synaptic activity or intrinsic excitability. Seizure related neuronal cell loss, gliosis and enhanced Fos immunostaining have been observed in the BLA of Kcna1 −/− mice 19,20 , however experimental evidence showing functional alterations associated to BLA in this animal model remain lacking. Employing whole-cell current clamp recordings of BLA pyramidal neurons from Kcna1 +/+ and Kcna1 −/− mice, we initially examined whether the loss of K v 1.1 subunits in BLA pyramidal neurons affects their intrinsic excitability and synaptic activity (Figs. 1 and 2). In response to incremental steps of current injection, BLA pyramidal neurons from both Kcna1 +/+ and Kcna1 −/− mice exhibited similar action potential firing profiles with similar frequencies as shown in frequency-current (f-I) plots (Fig. 1b). We did not observe any difference between Kcna1 +/+ and Kcna1 −/− mice in resting membrane potential (Fig. 1c), rheobase-the minimal current required to evoke action potential firing (Fig. 1d) or action potential threshold (Fig. 1e). Thus, loss of K v 1.1 subunits does not appear to modify the intrinsic excitability of BLA pyramidal neurons. These data are consistent with previous results showing that kindling or repeated seizures in the amygdala do not alter membrane properties and intrinsic excitability of BLA pyramidal neurons 45,46 .
Over the last decades, there has been growing evidence that K v 1.1 channel activity regulates neurotransmitter release, providing fine-tuning mechanisms to modulate synaptic strength 47,48 . Although studies have shown that presynaptic K v 1.1-containing channels are localized in the BLA neurons 49 , their functional contributions to synaptic activity have not been reported.
We examined spontaneous excitatory (sEPSCs) and inhibitory (sIPSCs) postsynaptic currents in BLA pyramidal neurons in Kcna1 +/+ and Kcna1 −/− mice (Fig. 2). Figure 2a shows an example of sEPSCs recorded in BLA neurons of Kcna1 +/+ and Kcna1 −/− mice. The mean frequency of sEPSCs was significantly increased in Kcna1 −/− mice (18.8 ± 0.4 Hz, n = 7 cells from 5 mice) compared with Kcna1 +/+ (12.8 ± 0.4 Hz, n = 9 cells from 5 mice, p = 0.0004; Fig. 2c). In contrast, the mean amplitude of sEPSCs showed no difference between Kcna1 +/+ and Kcna1 −/− BLA pyramidal neurons (Kcna1 +/+ = 14.7 ± 1.1 pA, n = 9 cells from 5 mice; Kcna1 −/− = 15.2 ± 1.1 pA, n = 7 cells from 5 mice p = 0.64; Fig. 2d). These data suggest that the increase in sEPSCs frequency observed in Kcna1 −/− BLA neurons is likely a result of an increased glutamate release probability at nerve terminals, consistent with previous reports of synaptic activity of BLA neurons from kindled rats 46 . We next recorded sIPSCs (Fig. 2b) and found no significant changes in either sIPSCs frequency or amplitude between Kcna1 +/+ and Kcna1 −/− mice (Fig. 2e,f). The mean sIPSCs values were 8.2 ± 0.9 Hz and 49.7 ± 6.03 pA for Kcna1 +/+ mice (n = 5 cells from 4 mice), and 8.03 ± 0.7 Hz and 58.1 ± 4.9 pA for Kcna1 −/− mice (n = 5 cells from 4 mice), respectively. To assess the K v 1.1-containing channels regulate firing properties of late-spiking GABAergic interneurons in the central lateral amygdala. Loss of GABAergic interneurons and alterations in inhibitory activity has been demonstrated in the amygdala of epileptic animal models, suggesting that GABAergic neurons may play key roles in the generation and spread of seizures 26,51 . Within the amygdala complex, the CeL consists almost exclusively of GABAergic interneurons that shape amygdala output and function as a relay station between the amygdala complex and other downstream targets [52][53][54] . K v 1.1-containing channels play crucial roles in regulating near-threshold properties and repetitive firing of fast-spiking neocortical 6,55 and deep cerebellar nuclear GABAergic neurons 56 thus here we studied their contributions towards regulating excitability of CeL GABAergic interneurons.
Whole cell current-clamp recordings of CeL interneurons was used to compare firing properties between Kcna1 +/+ and Kcna1 −/− mice. In response to depolarizing current injections, CeL neurons were shown to display two major firing patterns: late spiking (LS) and early spiking (ES), typical of that previously reported 52,54,57,58 . In control Kcna1 +/+ mice, LS neurons exhibited a substantial delay before the onset of the first AP (~ 0.8 to 1 s; see   To explore mechanisms underlying the observed changes in synaptic strength we next measured the pairedpulse ratio (PPR) as an index of short-term facilitation, a known form of plasticity at mITC-CeL synapses. Figure 4d show representative traces of eIPSCs evoked by pair-pulse stimulation at an interval of 100 ms from Kcna1 +/+ (black traces) and Kcna1 −/− mice (red traces). In response to the paired-pulse stimuli, facilitation of synaptic events was observed in both Kcna1 +/+ and Kcna1 −/− mice. Notably, the PPR obtained by two successive GABAergic eIPSCs was increased in Kcna1 −/− compared with Kcna1 +/+ mice (Fig. 4e). Changes in PPR reflect alterations in release probability from the presynaptic terminals 65,66 thus the enhanced PPR suggests that the release probability from GABAergic presynaptic terminals is decreased at mITC-CeL synapses in Kcna1 −/− mice. Together, the responses evoked by local stimulation of mITCs demonstrate that inhibitory synaptic transmission onto CeL neurons is significantly impaired in Kcna1 −/− mice.

Feed-forward inhibition mediated by mITCs neurons onto CeL is impaired in Kcna1 −/− mice.
As noted, the mITCs mediate feed-forward inhibition (FFI) onto neurons of the CeL, playing an important role in sculping amygdala network dynamics 53,62,63 . To test whether loss of K v 1.1 subunits results in altered disynaptic FFI of CeL neurons, we applied electrical stimulation in the lateral amygdala while recording synaptic responses in CeL neurons from Kcna1 +/+ and Kcna1 −/− mice (Fig. 5a). At a holding potential of -20 mV, electrical stimulation of the lateral amygdala elicited a biphasic response in CeL neurons (Fig. 5b) that is composed of a monosynaptic excitatory (eEPSC) component and a disynaptic inhibitory (eIPSC) response. The inhibitory component was blocked by bath perfusion of the non-NMDA receptor antagonist CNQX (20 µM) (Fig. 5b, blue trace), indicating that the eIPSC component is driven by excitatory synapses onto inhibitory mITCs neurons projecting onto CeL neurons 62 . In Kcna1 −/− mice, bath application of the GABA A receptor antagonist picrotoxin (100 µM) www.nature.com/scientificreports/ also completely blocked the inhibitory component (Fig. 5b, right, violet trace) confirming that disynaptic transmission is mediated by mITCs GABAergic neurons. Interestingly, the eIPSC amplitude was significantly reduced in Kcna1 −/− mice (16.2 ± 3.0 pA, n = 5 cells from 4 mice) compared with Kcna1 +/+ mice (38.8 ± 6.6 pA, n = 8 cells from 5 mice; p = 0.004, Fig. 5c). No changes on the eEPSC component in CeL neurons were observed (Fig. 5d). The eEPSC amplitude values were − 25.2 ± 3.3 pA for Kcna1 +/+ mice (n = 5 cells from 4 mice), and − 25.8 ± 3.6 pA for Kcna1 −/− mice (n = 8 cells from 5 mice), respectively. It is known that feedforward inhibitory circuits permit a narrow window between excitatory and inhibitory inputs so that disynaptic inhibition occurs after a short delay from the onset of monosynaptic eEPSCs; this synaptic delay restricts the time window for temporal summation of excitatory inputs 67 . Given the importance of FFI inhibition in regulating the timing of neuronal responses, we measured the timing of each component of FFI in CeL neurons in response to stimulation of LA by voltage clamping the membrane at their respective reversal potentials (− 70 mV for eEPSC and 0 mV for eIPSC; Fig. 5e).
In both, Kcna1 +/+ (black trace) and Kcna1 −/− (red trace) mice the eIPSC component occurred with a delay following the onset of eEPSC component. However, the mean delay between the eEPSC and eIPSC was found to be significantly longer in Kcna1 −/− mice (4.86 ± 0.6, n = 5 cells from 4 mice) compared with Kcna1 +/+ mice (2.58 ± 0.3 ms, n = 6 cells from 5 mice, p = 0.026; Fig. 5f). These data suggest K v 1.1-containing channels play an important role in regulating the spike integration window in CeL GABAergic interneurons by setting a limit for the spike generation provoked by the temporal summation of excitatory synaptic activity.

Discussion
To our knowledge, the current study is the first to provide evidence of key roles of K v 1.1-containing potassium channels in regulating network excitability of the amygdala circuitry. Using Kcna1 null mice (Kcna1 −/− ), a wellknown mouse model of temporal lobe epilepsy, here we show that deletion of K v 1.1 subunits alters intrinsic excitability and synaptic activity in neurons from two amygdalar subdivisions known to be involved in epilepsy and epilepsy-related comorbidities. At the cellular level, we found no significant differences in either passive membrane properties or action potential parameters of BLA pyramidal neurons between wild-type and Kcna1 −/− mice (Fig. 1). Unaltered excitability has been previously reported in hippocampal CA3 and layer V neocortical pyramidal neurons from Kcna1 −/− mice 21,68 . Although immunocytochemical studies have shown K v 1.1 subunits to be localized in BLA pyramidal neurons 49 it is likely that the lack of changes in somatic firing responses observed here could reflect the low level of K v 1.1 channel expression in the somatic compartment. This notion is consistent with a pharmacological study of lateral amygdala pyramidal neurons in which application of dendrotoxin-K, a selective blocker of K v 1.1-containing channels, did not modify intrinsic firing properties 69 .
Presynaptic K v 1.1-containing channels localized in BLA neurons play a critical role in mediating the inhibitory effect of µ-opioids on neurotransmitter release from GABAergic inputs to the BLA 49 . Here, we tested whether the complete loss of K v 1.1 subunits affected the spontaneous synaptic activity of BLA pyramidal neurons. Recordings of spontaneous postsynaptic currents provide information concerning the overall synaptic drive within an intact network, which in turn has functional implications for the overall output of a given neuron 66 . It is well established that alterations in the glutamatergic and GABAergic systems contribute to hyperexcitability in many animal models of epilepsy and other neurodevelopmental disorders including autism and schizophrenia 70 . Examining www.nature.com/scientificreports/ whether there exists any differences in excitatory and inhibitory spontaneous synaptic activities in the BLA of Kcna1 −/− mice we found that the frequency of sEPSCs was increased in BLA neurons in Kcna1 −/− mice (Fig. 2c). This result is in agreement with studies in animal models of epilepsy showing that alterations in BLA pyramidal neurons induced by repeated seizures or kindling are confined to the presynaptic glutamatergic terminals 45,46 . In contrast to sEPSCs, analysis of sIPSCs shows no differences in the frequency or amplitude of inhibitory GABAergic currents between Kcna1 −/− and Kcna1 +/+ mice (Fig. 2e,f). We hypothesize that the noted shift in the E-I balance due to the increase in sEPSC frequency may contribute to the epileptiform activity of the Kcna1 −/− mice. Loss of K v 1.1 function is known to affect the near-threshold excitability and repetitive firing properties of fast-spiking neocortical 6,55 and auditory GABAergic interneurons 71 . Within the amygdala complex, the lateral subdivision of central amygdala (CeL) is under strong local inhibitory control and it has been reported that the CeL mediates autonomic and behavioral responses associated with comorbidities of epilepsy such as depression, anxiety and pain via projections to the brain stem 35,36,72 . As such, alterations in the intrinsic excitability of CeL neurons can affect the functional output of the amygdala circuitry. A previous report on CeL neurons showed that α-dendrotoxin (DTX)-sensitive K v 1-containing channels of the K v 1 family (K v 1.1-1.6) regulate spike latency of LS cells 58 , although effects of K v 1.1-containing channels on CeL interneurons was not explicitly tested. In the present study, we demonstrate that the lack of K v 1.1 subunits has a major influence on the excitability of LS cells of the CeL by reducing the characteristic delay to first AP (Fig. 3b), decreasing the rheobase (Fig. 3c), and increasing the number of APs in response to current injections (Fig. 3e). These data suggest that K v 1-containing channels contribute to attenuating depolarizing voltage transients caused by excitatory synaptic inputs in LS cells, allowing for a delay in the response towards a sustained barrage of excitatory synaptic inputs [73][74][75] . Altogether, an increase in action potential firing in response to somatic stimulation in late firing CeL neurons, presumably corresponding to a subset of GABAergic neurons known to exert inhibitory control of brainstem projecting neurons of the central medial amygdala (CeM) 76,77 , would result in disinhibition of CeM output. Overall, it is likely that the enhanced excitability and AP firing of CeL LS interneurons contribute to the hyperexcitability and seizure occurrence in Kcna1 −/− mice.
The synaptic plasticity of the CeL is implicated in a variety of behavioral phenotypes and neurological disorders, including in models of conditioned fear and pain 78,79 . CeL interneurons, receive inputs from amygdalar and extra amygdalar sites and send GABAergic outputs to various downstream targets including the hypothalamus and brainstem. Within the amygdala complex, the CeL receives excitatory inputs from the lateral amygdala and feedforward inhibitory inputs from the GABAergic medial intercalated cells (mITCs) and send outputs to the CeM and brainstem regions such as parabrachial nucleus and nucleus tractus solitarius. Having shown that K v 1.1-containing channels play a critical role in regulating the intrinsic excitability of CeL neurons, we also explored the potential impact of K v 1.1 subunits on synaptic plasticity of CeL neurons. We first studied the effect of K v 1.1 loss on the direct source of inhibitory input onto CeL neurons (Fig. 4). Recording from CeL neurons and stimulating monosynaptic inputs from local GABAergic interneurons from the mITCs (Fig. 4a), we found that in Kcna1 −/− mice monosynaptic IPSCs evoked at the mITC-CeL synapse had lower amplitudes than those recorded from Kcna1 +/+ mice (Fig. 4c). This suggests that basal neurotransmission at mITC-CeL synapses are impaired in Kcna1 −/− mice. To explore the mechanisms underlying this synaptic impairment, we investigated the possible involvement of presynaptic mechanisms by measuring the pair-pulse ratio (PPR) of eIPSCs, a parameter affected by changes in release probability from presynaptic terminals 65 . We found that the PPR at mITC-CeL synapses was significantly increased in Kcna1 −/− mice compared to Kcna1 +/+ mice, suggesting that the release probability from the GABAergic terminals is decreased in the mITC-CeL pathway in Kcna1 −/− mice (Fig. 4e). Enhanced short-term facilitation of mITC-evoked IPSCs indicates modification in the frequency filtering properties in the mITC-CeL pathway via rapid changes in synaptic strength 80 , allowing the sensitization towards high frequency or coincident activation of inhibitory mITC inputs onto CeL. Therefore, Kcna1 subunits seems to be critical to maintain a precise control of temporally organized activity in the CeL. Together, these data show that K v 1.1-containing channels are crucial contributors towards the synaptic strength and short-term plasticity of CeL neurons.
CeL neurons establish reciprocal connections whose functional role has not yet been clarified 54 , and it is difficult to exactly infer how Kcna1-containing channels regulate CeL overall output activity. The CeL receives glutamatergic afferents from the nociceptive parabrachial pontine nucleus of the brainstem 81 and from the LA, where thalamic and cortical multimodal sensory inputs converge, conveying information associated to fear learning and emotional processing 82 . We speculate that different local inhibitory networks regulate distinct behavioral outcomes depending upon context and that dysregulation of amygdala circuits in Kcna1 −/− mice might result in responses exacerbated or inappropriate for that context.
Feedforward inhibition (FFI) is an important mechanism previously characterized in hippocampal 67 and cerebellar 83 neurons, where it is shown to play a key role in controlling spike timing. In thalamic networks 84 FFI restrains the propagation of epileptiform waves 85,86 . Dysfunction of FFI is known to cause abnormal circuit dynamics that underlie seizures; reduced FFI in fast-spiking interneurons has been implicated in the fast spreading of epileptic seizures in a mouse model of Dravet syndrome 87 . Here we showed the involvement of K v 1.1-containing channels in FFI of CeL neurons in an animal model of TLE. Our results are consistent with previous reports that excitatory projections from the lateral amygdala not only project to the CeL directly but also target inhibitory interneurons in the mITC which in turn inhibit CeL neurons via a FFI mechanism 53,62,63 . The lack of K v 1.1 subunits specifically decreased the amplitude of the inhibitory component of disynaptic responses in CeL neurons after LA stimulation (Fig. 5c). It is interesting to note that there were no changes in the amplitude of the eEPSC component (Fig. 5d), indicating that reduction of FFI onto CeL neurons must be located downstream of the excitatory input. FFI mediated by mITCs is shown to underlie fear extinction 88 thus any reduction could induce impairment of fear extinction and/or anxiety in Kcna1 −/− mice. www.nature.com/scientificreports/ It is well established that FFI sets a limit for overexcitation by temporally restricting the time window for temporal synaptic integration during which action potentials can be generated 67 . Our study suggests that K v 1.1-containing channels contribute to the regulation of the spike integration window in CeL interneurons. The spike integration window in Kcna1 −/− mice was wider (long delay between the eEPSC and eIPSC component) compared to Kcna1 +/+ mice (Fig. 5e,f), which could potentially allow multiple scattered inputs to trigger action potentials in CeL neurons. Fundamentally, altering the spike integration window will alter how information is processed as a function of frequency. Together, our results suggest that K v 1.1-containing channels basally regulate disynaptic feed-forward inhibition in LA-CeL pathway. Impairment of feedforward inhibitory control of CeL outputs could affect the multisensory integration of cortical and thalamic inputs converging onto the amygdala, ultimately modifying emotional behaviors and nociceptive responses in epilepsy-related conditions. Such impairment may also compromise the function of the central autonomic network involved in cardiorespiratory regulation increasing the risk of SUDEP in Kcna1 −/− mice.
GABAergic neurotransmission has long been considered as a mechanism to restrain excessive excitation in neuronal networks and thus prevent the occurrence of seizures 89 . However, in experiments using human tissue from temporal lobe epilepsy patients it was shown that synchronous activity of pyramidal cells mediates transition from inter-ictal discharges to focal seizure events while interneuron firing leads to inter-ictal epileptiform discharges relying on both GABAergic and glutamatergic activity 90 . Furthermore, interneuron hyperexcitability contributes to seizure onset by promoting extracellular K + accumulation, allowing the recruitment of surrounding areas and increasing the propensity to sustain synchronous seizure activity 91,92 . Our data showing increased firing activity in Kcna1 −/− CeL interneurons is consistent with network hyperactivity which may promote seizures through loss of spike timing control in the local circuit.
In conclusion, our results provide the first evidence concerning critical roles of K v 1.1-containing channels in synaptic integration and shaping the properties of a feedforward inhibitory circuit within the CeL. While Kcna1 −/− mice exhibit spontaneous seizures, it remains to be determined how these K v 1.1-related mechanisms directly or indirectly affect seizure initiation and propagation within the amygdala and translate into complex phenotypes characterized by comorbid behaviors and respiratory dysfunction.

Materials and methods
Animals. Experiments were performed on littermate wild-type (Kcna1 +/+ ) and null (Kcna1 −/− ) mice. For this study a total of 29 male and 30 female mice (P22-P28) were used. Breeding pairs of heterozygous (Kcna1 −/+ ) mice were on a C3HeB/FeJ congenic background and colonies were maintained in the Animal Resources Unit at the University of British Columbia. Mice were given food and water ad libitum and kept on a 12-h light/dark cycle. All the experimental protocols were approved by the University of British Columbia Animal Care Committee (UBC ACC Protocol A19-0233) and were in accordance with the Canadian Council on Animal Care (CCAC) guidelines. The present study complies with the pertinent aspects of ARRIVE guidelines. Acute brain slice preparation. Under isoflurane anesthesia [5% (vol/vol) in oxygen], mice were decapitated, and brains removed and transferred immediately to an ice-cold oxygenated (95% O 2 -5% CO 2 ) sucrose cutting solution containing the following (in mM): 214 Sucrose, 26 NaHCO 3 , 1.25 NaH 2 PO 4 , 11 glucose, 2.5 KCl, 0.5 CaCl 2 , and 6 MgCl 2 (pH 7.4). 300 µm-thick coronal slices of amygdala complex containing basolateral and central amygdala were collected using a vibratome (VT 1200; Leica). Slices were incubated in artificial cerebral spinal fluid (ACSF) containing (in mM):126 NaCl, 26 NaHCO 3 , 2.5 KCl, 1.5 NaH 2 PO 4 , 2 MgCl 2 , 2 CaCl 2 ,10 glucose (pH 7.4) with 95% O 2 -5% CO 2 at 37 °C for 45 min prior to recording.
Whole cell patch clamp recordings. After incubation in ACSF for ~ 40 min, individual slices were transferred to the recording chamber, superfused with ACSF and maintained at 30 °C. BLA and CeL amygdala neurons were visualized with infrared differential interference contrast (IR-DIC; Slicescope 6000 Scientifica, UK) in combination with a X40 water immersion objective. Whole-cell patch-clamp recordings were performed to record intrinsic excitability and synaptic properties. All recordings were collected using a Multiclamp 700B amplifier and signals were digitized and acquired using Digidata 1550 and pClamp 10 and/or 11 software (Molecular devices). The recording chamber was grounded with an Ag/AgCl pellet. Patch pipettes (4-6 MΩ) were pulled from borosilicate glass using a P-1000 micropipette puller (Sutter Instruments).
Intrinsic excitability was recorded in current clamp mode using an internal recording solution containing the following (in mM): 120 K-gluconate, 10 HEPES, 1 MgCl 2 , 1 CaCl 2 , 11 KCl, 11 EGTA, 4 MgATP, 0.5 NaGTP, with pH adjusted to 7.2 using KOH and osmolarity adjusted to 290 mOsm/kgH 2 O using D-mannitol. The liquid junction potential was + 13.3 mV and the data was reported without subtraction. Bridge balance values were monitored during recordings and cell displaying bridge balance values > 30 MΩ were excluded from the analysis. Action potentials were evoked with 1.2 -s long square-pulse current injections from -100 to 200 pA with increments of 10 pA. Data under current clamp conditions were sampled at 50 kHz and low-pass filtered at 10 kHz.
Synaptic activity such as spontaneous and evoked postsynaptic currents were recorded in voltage clamp mode using a cesium based internal solution containing the following (in mM): 140 Cs-methanesulfonate, 5 CsCl, 5 tetraethylammonium-Cl, 1 EGTA, 10 HEPES, 4 MgATP, 0.5 NaGTP, the pH was adjusted to 7.2 with CsOH, and osmolality was adjusted to 290 mOsm/kgH 2 O with D-mannitol. To isolate spontaneous excitatory post synaptic currents (sEPSCs), the GABA A receptor antagonist picrotoxin (PTX, 100 µM) was added to the ACSF. To isolate spontaneous inhibitory postsynaptic currents (sIPSCs), the NMDA receptor antagonist D-2-amino-5-phosphonovalerate (D-APV, 50 µM), and AMPA receptor antagonist cyanquixaline 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX, 20 µM) were added to the ACSF. To evaluate sIPSCs, cells were held at a membrane potential of 0 mV, and for sEPSCs cells were held at a membrane potential of − 70 mV during a 60-s gap-free recording. www.nature.com/scientificreports/ Data acquisition was sampled at 20 kHz and low-pass filtered at 2 kHz. Recordings with a series resistance > 25 MΩ were excluded from analysis. To evoke synaptic responses electric pulses were generated with an S48 stimulator via a stimulus isolation unit (SIU5; Grass Instruments) and applied with a concentric bipolar electrode (CBAPC100; FHC Inc.). Evoked inhibitory postsynaptic currents (eIPSCs) were elicited in CeL amygdala GABAergic interneurons when the stimulating electrode was placed on the medial intercalated cells (mITC). As shown in Fig. 4a (middle panel) anatomical landmarks were used to identify the mITC, located within the intermediate capsule.
eIPSCs were isolated in the presence of CNQX (20 µM), D-AP5 (50 µM) and GABA B blocker CGP52432 (1 µM) in ACSF recording solution. For feed-forward experiments, the stimulating electrode was placed in the lateral amygdala (LA) and evoked synaptic responses were recorded in CeL neurons at a holding potential of − 20 mV. For morphologic identification of BLA and CeL neurons, recorded neurons were labeled with biocytin (0.05% in the internal solution) by applying hyperpolarizing current for 20 min in current-clamp mode. Subsequently, the recording pipette was withdrawn slowly and the slice was immediately fixed into 4% paraformaldehyde (PFA) solution at 4 °C. Biocytin-filled neurons were visualized by incubating the slices in avidin-biotinylated-horseradish peroxidase (ABC). Images of biocytin-filled neurons were obtained using an Olympus BX-53 Widefield Microscope with a 40×/NA 0.8 semi-apochromat objective in order to visualize cell morphology and confirm localization (Fig. 4a, lower panel). z-stack acquisition was performed using the Fiji Image processing software (version 1.53j).
Data analysis and statistics. Electrophysiological recordings were analyzed using Clampfit 11 (Molecular devices); data plotting, figures generation and statistical analysis were performed using Origin 8.6 (Origin Lab). The steady state frequency of action potentials was obtained from the last 500 ms period of the depolarizing current pulses and plotted as a function of normalized current injection (f-I relation). Rheobase was measured as the minimum intensity of 1 s current pulse required for initiation of AP. The spike delay was measured from the start of the current injection to the start of the rising phase of first AP evoked by the rheobase. Detection and analysis of the spontaneous synaptic postsynaptic currents was performed by creating a template in Clampfit 11. For calculation of sEPSCs to sIPSCs ratio, mean values of each measure (frequency and amplitude) for each cell were calculated and expressed as a ratio. The ratios were then pooled within groups and compared the ratio of the mean values. The paired-pulse ratio (PPR) was measured by delivering two pulses with an increasing interstimulus interval, and the PPR was calculated from the amplitude of the synaptic response to the second pulse divided by the first pulse. To measure the timing of feed-forward inhibition (synaptic delay between the eEPSCs and eIPSCs), the eEPSCs and eIPSCs were separated by voltage clamping at their respective reversal potentials (eEPSCs were recorded at − 70 mV and eIPSCs at 0 mV, respectively). We used respective 10% rise time points to determine the eEPSC-eIPSC delay. All data are expressed as mean ± SEM. Statistical comparison was performed with one-way ANOVA using Tukey post hoc test. Differences with P-value < 0.05 were considered statistically significant. The n value represents the number of cells tested. In most experiments, only one neuron was recorded from an individual animal; the sample size indicates the number of animals used for recordings. Synaptic and action potential parameters from age-matched male and female mice were pooled together for statistical comparison.
Pharmacological agents. D-AP5, CNQX and CGP52432 were purchased from Tocris Bioscience. Picrotoxin (PTX) was purchased from Sigma Aldrich. All drugs were prepared as stock solutions (CNQX was dissolved in DMSO and the other drugs in nanopure H 2 O) and stored at − 20 °C; working aliquots were thawed and added to ACSF. Drugs were applied by bath perfusion at a flow rate of 1 ml/min.