Small conductance calcium activated potassium (SK) channel dependent and independent effects of riluzole on neuropathic pain-related amygdala activity and behaviors in rats

Abstract

Background and purpose

Chronic neuropathic pain is an important healthcare issue with significant emotional components. The amygdala is a brain region involved in pain and emotional-affective states and disorders. The central amygdala output nucleus (CeA) contains small-conductance calcium-activated potassium (SK) channels that can control neuronal activity. A clinically available therapeutic, riluzole can activate SK channels and may have antinociceptive effects through a supraspinal action. We tested the hypothesis that riluzole inhibits neuropathic pain behaviors by inhibiting pain-related changes in CeA neurons, in part at least through SK channel activation.

Experimental approach

Brain slice physiology and behavioral assays were done in adult Sprague Dawley rats. Audible and ultrasonic vocalizations and von Frey thresholds were measured in sham and neuropathic rats 4 weeks after left L5 spinal nerve ligation (SNL model). Whole cell patch-clamp recordings of regular firing CeA neurons in brain slices were used to measure synaptic transmission and neuronal excitability.

Key results

In brain slices, riluzole increased the SK channel-mediated afterhyperpolarization and synaptic inhibition, but inhibited neuronal excitability through an SK channel independent action. SNL rats had increased vocalizations and decreased withdrawal thresholds compared to sham rats, and intra-CeA administration of riluzole inhibited vocalizations and depression-like behaviors but did not affect withdrawal thresholds. Systemic riluzole administration also inhibited these changes, demonstrating the clinical utility of this strategy. SK channel blockade in the CeA attenuated the inhibitory effects of systemic riluzole on vocalizations, confirming SK channel involvement in these effects.

Conclusions and implications

The results suggest that riluzole has beneficial effects on neuropathic pain behaviors through SK channel dependent and independent mechanisms in the amygdala.

Introduction

Chronic pain is a significant healthcare issue affecting millions of people worldwide, but treatment remains a challenge (Gaskin and Richard, 2012). Therapeutic options are limited and characterized by variable efficacy and severe side effects, and so there is a need for new and improved strategies. Riluzole is a clinically available drug used for the treatment of amyotrophic lateral sclerosis (ALS). Evidence from preclinical studies suggest that riluzole can have antinociceptive effects in inflammatory pain models such as the formalin (Blackburn-Munro et al., 2002; Coderre et al., 2007; Munro et al., 2007) and carrageenan tests (Abarca et al., 2000), as well as in neuropathic pain models such as chronic constriction injury (Sung et al., 2003), spinal root avulsion injury (Chew et al., 2014), cervical spondylotic myelopathy (Moon et al., 2014), and spinal cord compression injury (Hama and Sagen, 2011). Importantly, in the latter study intracerebroventricular but not intrathecal riluzole injection had antinociceptive effects, suggesting a site of action in the brain. A recent study by our group found evidence for an action of riluzole in the amygdala in an arthritis pain model (Thompson et al., 2015), but this remains to be explored in a chronic pain model.

Mechanisms of antinociceptive effects of riluzole action are not clear. Riluzole can inhibit voltage-gated sodium channels and glutamate transmission, and activate small-conductance calcium-activated potassium (SK) channels (Bellingham, 2011; Cao et al., 2002). SK channels are calcium-sensitive, voltage-insensitive potassium channels that can inhibit neuronal activity by mediating the medium afterhyperpolarization (mAHP) (Adelman et al., 2012), shunting NMDA receptor mediated excitatory transmission (Faber, 2010; Faber et al., 2005), and inducing an inhibitory postsynaptic response (Finch and Augustine, 1998; Fiorillo and Williams, 1998; Morikawa et al., 2000; Morikawa et al., 2003; Nakamura et al., 1999; Takechi et al., 1998). SK channels are widely expressed in the nervous system and have previously been reported to influence pain-related responses in the periphery and spinal cord (Bahia et al., 2005; Hipolito et al., 2015; Pagadala et al., 2013), but their role in pain modulation in the brain remains to be determined.

The amygdala is a limbic brain structure that plays a key role in the emotional affective aspects of pain (Neugebauer, 2015). The lateral-basolateral (LA-BLA) complex of the amygdala receives polymodal sensory information through cortical and thalamic inputs and evaluates and processes biologically significant affective information, which is then relayed to the central nucleus (CeA) of the amygdala. The lateral nucleus of the amygdala is also thought to be the site of convergence for neutral stimuli and nociceptive stimuli involved in fear conditioning (Phelps and LeDoux, 2005). The lateral-capsular division of the CeA (CeLC) integrates information from the LA-BLA with purely nociceptive sensory information from the spino-parabrachio-amygdaloid tract (PB) and serves as an output nucleus by sending projections to various brainstem regions to generate amygdala mediated pain responses and modulate pain behaviors. Excitatory transmission is increased throughout this neurocircuitry in the pain state and results in enhanced drive and output of regular firing CeLC neurons (Ji et al., 2010; Neugebauer et al., 2003). Amygdala activity and output correlate positively with pain and affective behaviors (Carrasquillo and Gereau, 2007; Neugebauer, 2015).

SK channels regulate excitability of CeLC neurons (Lopez de Armentia and Sah, 2004), but not LA-BLA neurons (Faber and Sah, 2007), through the medium but not slow AHP, and the mAHP is abolished in CeLC neurons after application of SK channel blockers (Lopez de Armentia and Sah, 2004). Therefore, inhibiting pain-related amygdala hyperexcitability through SK channel activation could represent a therapeutic strategy for chronic pain. Here we test the hypothesis that riluzole has beneficial effects on neuropathic pain behaviors in an SK channel dependent manner by inhibiting pain-related electrophysiological changes in CeLC neurons.

Methods

Animals

Adult male Sprague-Dawley rats (200–350 g) with unrestricted access to food and water were housed in a temperature-controlled room on a 12-h day/night cycle. This strain and species was used because our previous studies established and validated this preclinical model and approach (Han et al., 2005; Ji et al., 2017; Neugebauer, 2015; Thompson et al., 2015). Experimental procedures were approved by the Texas Tech University Health Sciences Center Institutional Animal Care and Use Committee (IACUC) and are in compliance with the ARRIVE guidelines (Kilkenny et al., 2010; McGrath and Lilley, 2015). Rats were randomly assigned to their respective group for all experiments. Experiments were done in a blinded fashion as much as possible and different investigators performed the tests.

Spinal nerve ligation pain model

The spinal nerve ligation (SNL) model of chronic pain (Kim and Chung, 1992) was used to induce a peripheral neuropathy in the left hindpaw as described in a recent publication from our group (Ji et al., 2017; Kim and Chung, 1992). Isoflurane was used for induction (3%) and maintenance (2%) of anesthesia throughout the procedure. Using sterile procedures, the L5/L6 level paraspinal muscles and underlying L6 transverse process were removed. The L5 spinal nerve was removed from adjacent structures and tightly ligated with 6-0 silk thread. The muscles were sutured closed and the skin was clipped together. Bacitracin was applied, and animals were monitored afterwards for signs of distress. Sham operated animals were used as controls, and received the same surgical procedure but without the L5 spinal nerve ligation. Naïve rats were also used for some experiments. No difference in response was observed between naïve rats and sham rats, and so data were pooled in experiments labeled “control”.

Electrophysiology

Brain slice preparation

Brain slices containing the CeA (1.6–3.3 mm posterior to bregma) were obtained from SNL and sham rats. Brains were removed quickly as described before (Ji et al., 2013; Ji et al., 2015; Ren et al., 2013) and immersed in ice-cold oxygenated sucrose-based physiological cutting solution, containing the following (in mM): 87 NaCl, 75 sucrose, 25 glucose, 5 KCl, 21 MgCl₂, 0.5 CaCl₂, and 1.25 NaH₂PO₄ (Kasanetz et al., 2013). Coronal brain slices (400 μm) containing the CeLC were obtained using a Vibratome (Series 1000 Plus) and incubated for at least 1 h before recordings in oxygenated ACSF (35°C) containing (in mM): 117 NaCl, 4.7 KCl, 1.2 NaH₂PO₄, 2.5 CaCl₂, 1.2 MgCl₂, 25 NaHCO₃, and 11 glucose. For electrophysiological experiments, one brain slice was transferred to the recording chamber and bathed in ACSF (35±1°C) using gravity driven superfusion at a rate of ~2 ml/min. Only one neuron was recorded in each slice, and new slices were used for each series of drug applications.

Patch-clamp recordings

Whole-cell patch-clamp recordings were made from visually identified CeLC neurons in the right hemisphere using infrared DIC-IR videomicroscopy as described previously (Ji et al., 2017; Kiritoshi et al., 2016; Kiritoshi and Neugebauer, 2015). Borosilicate glass recording electrodes (3–5 MΩ) were filled with a potassium gluconate or potassium methyl sulfate based internal solution for mAHP recordings and a potassium gluconate based internal solution for action potential and synaptic recordings. No difference was found between mAHP responses using potassium gluconate and potassium methyl sulfate based internal solutions, and so data were combined for analysis. The potassium gluconate based internal solution contained the following (in mM): 122 K-gluconate, 5 NaCl, 0.3 CaCl₂, 2 MgCl₂, 1 EGTA, 10 HEPES, 5 Na₂-ATP, and 0.4 Na₃-GTP; KOH was used to adjust pH to 7.2–7.3 and sucrose was used to adjust osmolarity to 280 mOsm/kg. The potassium methyl sulfate based internal solution contained the following (in mM): 135 KMeSO₄, 8 NaCl, 10 HEPES, 2 Na₂ATP, 0.3 Na₃GTP; KOH was used to adjust pH to 7.2–7.3 and sucrose was used to adjust osmolarity to 280 mOsm/kg. The system for data acquisition and analysis consisted of a dual four-pole Bessel filter (Warner Instruments, Harvard Apparatus), low-noise Digidata 1440A interface (Axon Instruments, Molecular Devices), Axoclamp-2A amplifier (Axon Instruments), and pClamp10.2 software (Axon Instruments).

Neuronal excitability was investigated in current clamp. Action potentials were evoked from a holding potential of −60 mV using 0.5 s depolarizing current steps of increasing amplitude. Number of action potentials formed at a given magnitude of injected current was used to assess action potential firing frequency-current (F–I) relationship.

AHP amplitude was measured in current clamp from a holding potential of −60 mV. The AHP was evoked using a large 1 s depolarizing current injection. The mAHP amplitude was determined by finding the antipeak amplitude in the mAHP portion of the AHP (within 100 ms of termination of current injection). If the membrane potential did not return to threshold potential within this period, the mAHP was considered 0 mV.

Synaptic transmission was assessed in current clamp from a holding potential of −60 mV by focal electrical stimulation (150 μs square wave pulses; S88 stimulator; Grass Technologies) of the parabrachial (PB) input into the CeA as described before (Fu and Neugebauer, 2008; Neugebauer et al., 2003; Ren et al., 2013). For stimulation of the PB input, a concentric bipolar stimulating electrode (David Kopf Instruments) was positioned dorsomedial to the CeA and outside of the caudate-putamen. The peak of the excitatory (EPSP) and inhibitory (IPSP) responses were used to assess changes in synaptic transmission. If after the EPSP the membrane potential did not drop below the prestimulation holding potential, the IPSP was considered 0 mV. The current-clamp approach was used to allow the assessment of synaptic integration of excitatory and inhibitory inputs.

Behavioral assays

Vocalizations

Vocalizations were measured as described previously (see (Cragg et al., 2016; Han et al., 2005; Kiritoshi et al., 2016; Medina et al., 2014; Neugebauer et al., 2007; Thompson et al., 2015). Rats were briefly anesthetized with isoflurane (2–4%) and lightly restrained in a custom designed recording chamber (US Patent 7,213,538) placed at a fixed distance from the detectors. Rats completely recovered from anesthesia before testing. Audible vocalizations (20 Hz–16 kHz; nocifensive response) were recorded by condenser microphone connected to a preamplifier, and ultrasonic vocalizations (25±4 kHz; averse affective response) were recorded by a bat detector connected to a filter and amplifier (UltraVox 4-channel system; Noldus Information Technology). Vocalizations were evoked using brief (15 s) innocuous (100 g/6mm²) and noxious (500 g/6mm²) stimuli applied to the left hindpaw using forceps calibrated with a force transducer whose output was displayed in grams on an LED screen as described before (Neugebauer et al., 2007). This output was used to ensure a consistent force was applied to the hindpaw during the innocuous or noxious stimulations. Vocalizations were recorded for 1 min starting at the onset of the stimulus and analyzed using Ultravox 2.0 software (Noldus Information Technology). Total durations of vocalizations over the 1 min period were used to assess supraspinally organized pain behaviors. Vocalizations were measured before and after systemic application of drug or vehicle, and before, during and after stereotaxic drug or vehicle administration.

Von Frey test

Von Frey spinal withdrawal thresholds were determined using the up-down method (Chaplan et al., 1994; Dixon, 1980). Rats were isolated in bottomless plastic chambers on top of a wire mesh grid. Von Frey monofilaments (Touch Test®, Stoelting Co., Wood Dale, IL) of varying thickness, corresponding to different bending forces (0.4, 0.6, 1, 2, 4, 6, 8, 15 g), were used for mechanical stimulation of the plantar surface of the left hindpaw. If a withdrawal in response to the stimulus occurred, the next lowest monofilament was used for the subsequent trial. Otherwise, the next highest was used. The 2 g monofilament was used for the first trial, and six trials beginning after the first positive withdrawal response were used to calculate the withdrawal threshold to assess spinally organized pain behaviors. Mechanical thresholds were measured before and after systemic application of drug or vehicle, and before, during and after stereotaxic drug or vehicle administration.

Forced swim test (FST)

The forced swim test was used to assess depression-related behaviors (Slattery and Cryan, 2012). Rats were acclimated to the laboratory for at least 1 h before experimentation. On Day 1, rats were placed in a plastic cylinder (20 cm × 50 cm) partially filled with water (30 cm; 23–25°C) for a 15 min pre-test. On Day 2, rats underwent cannula implantation surgery for stereotaxic application of riluzole into the amygdala (see “Microdialysis for stereotaxic drug application into the amygdala”). On Day 3, rats were randomly assigned to receive riluzole or ACSF (vehicle control) in order to prevent confounding effects from the order of drug administration. Rats were briefly anesthetized and the microdialysis probe was inserted into the cannula. Rats completely recovered from anesthesia before testing. After drug or vehicle administration, rats were placed in the partially water-filled cylinder and movement was recorded for 5 min by a camera (Basler AG, Ahrensburg, Germany) and analyzed by EthoVisionXT software (Noldus Information Technology). On Day 4, rats received the opposite treatment as on Day 3 (rats receiving riluzole on Day 3 received ACSF on Day 4 and vice versa) and were again placed in the partially filled cylinder and recorded for 5 min. The cylinder was cleaned carefully after each test. Previous studies indicated that antidepressant-like effects on the FST persist over multiple tests for at least 14 days without intervening pretests (Mezadri et al., 2011). Total duration of time spent immobile over the 5 min trial was used to assess depression-like behaviors.

Rotarod test

Motor function was assessed using the rotarod performance test described previously (Monville et al., 2006). Rats were trained on the RotaRod setup (Panlab, Harvard Apparatus) over 2 consecutive days with 4 trials each day. For each training trial, rats were placed on the rod rotating at a constant speed of 12 rpm for 5 min with a 20 min rest between trials (modified from (Urbach et al., 2014)). Rats that did not consistently remain on the beam for the full 5 min trial by the end of the training period were excluded from the study. For the actual test, rats were placed on the rod rotating at 4 rpm that smoothly accelerated to 40 rpm over 5 mins. The latency to falling from the rod was used to assess motor function. Experiments using systemic or stereotaxic drug administration were conducted at least 24 hours apart for each drug or dose. To prevent confounding effects, the order of drug administration was randomly assigned to each rat.

Drug application in behavioral assays

For systemic drug application, riluzole or vehicle (2-hydroxypropyl-β-cyclodexterin, HBC; 30%) was injected intraperitoneally. Riluzole or potassium channel blockers (see “Materials”) or vehicle (ACSF) were also administered stereotaxically into the amygdala using reverse microdialysis.

Microdialysis for stereotaxic drug application into the amygdala

Reverse microdialysis for drug application into specific brain regions is a well-established technique in our laboratory, and experiments were conducted as described previously (for recent publications, see (Cragg et al., 2016; Kiritoshi et al., 2016; Thompson et al., 2015)). This approach was used instead of drug injection because it does not result in a volume effect and allows for drug concentration to be maintained throughout drug administration (Stiller et al., 2003). Rats were anesthetized with isoflurane (2–4%) and placed in a stereotaxic frame (David Kopf Instruments). A guide cannula (CMA/Microdialysis, Solna, Sweden) was stereotaxically inserted to the dorsal margin of the right central nucleus of the amygdala (CeA; 2 mm caudal to bregma, 4 mm lateral to midline, and 6.5 mm deep) after a small craniotomy on the right side. The cannula was fixed to the skull with dental acrylic (Plastic One, Roanoke, VA, USA). Bacitracin was applied to the exposed tissue. For offsite control experiments, cannulas were implanted lateral to the CeA into the basolateral amygdala (BLA) or medial to the CeA into the internal capsule. On the test day, a microdialysis probe (CMA/Microdialysis 11, Solna, Sweden) extending 1 mm beyond the cannula was inserted into the right amygdala and connected to an infusion pump (Harvard Apparatus, Holliston, MA, USA) using polyethylene tubing. Drugs or artificial cerebrospinal fluid (ACSF, vehicle control; containing in mM: 117 NaCl, 4.7 KCl, 1.2 NaH₂PO₄, 2.5 CaCl₂, 1.2 MgCl₂, 25 NaHCO₃, and 11 glucose) were applied for 15 min at a rate of 5 μl/min before testing to establish tissue equilibrium. Microdialysis probe tip locations were verified histologically after experimentation (see “Histological verification of injection sites”).

Histological verification of microdialysis probe locations

Microdialysis probe positions were verified after experimentation histologically. Rats were euthanized by decapitation using a small guillotine (Harvard Apparatus Decapitator). Brains were rapidly removed and submerged in 4% paraformaldehyde and kept at 4°C overnight. Brains were then transferred to 30% sucrose in 0.1 M phosphate buffer and kept at 4°C until sectioning. Sections were made at 30 μm using a cryostat (Vibratome UltraPro 5000) and mounted on gel-coated glass slides. Sections were stained with hematoxylin and eosin (H&E) and coverslipped. Microdialysis probe tip locations were identified from the slides and plotted on diagrams adapted from (Paxinos and Watson, 1998).

Experimental protocol

Behaviors

Vocalizations and mechanical sensitivity (see “Behaviors”) were assessed before and weekly after SNL to confirm development and assess stability of neuropathy. Vocalizations were assessed before mechanical sensitivity with at least 5 min intervals between tests. All other experiments, including testing for drug effects, occurred approximately 4 weeks after surgery. The time course of behavioral changes in this model is well established (see for example Ji et al., 2017). Effects of systemic (i.p.) riluzole were determined 1 h postinjection in SNL and sham rats, because preliminary data indicated a maximum effect of 8 mg·kg⁻¹ riluzole at this time point, when withdrawal thresholds were tested every 30 min for 3 hours. Effects of stereotaxic drug application into the amygdala on vocalizations, mechanical thresholds, depression-like behaviors, and motor function (see “Behaviors”), were measured 15 min after starting drug application by microdialysis using a probe inserted into a surgically implanted guide cannula. To investigate site and mechanism of action of systemic riluzole, pain behaviors were measured 1 h after i.p. injection of riluzole and 15 min after concomitant application of potassium channel blockers (or ACSF) into the amygdala by microdialysis (microdialysis beginning 45 min post-injection of riluzole). Unless otherwise specified, separate cohorts of rats were used for each experiment.

Electrophysiology experiments

Brain slices were obtained from SNL or sham rats 4 weeks after surgery. Spinal withdrawal thresholds were measured in SNL and sham rats at least 45 min before brain slices were obtained to confirm presence of neuropathic hypersensitivity in SNL rats. This is a well-established protocol in our laboratory (see (Ji et al., 2017; Kiritoshi et al., 2016; Kiritoshi and Neugebauer, 2015)) and is unlikely to result in changes in neuronal responses as in vivo electrophysiology experiments using even noxious stimuli result in a return to baseline activity (see (Ji et al., 2015; Ji and Neugebauer, 2009; Kim et al., 2017)).

Data and statistical analysis

Statistical significance was accepted at the level P < 0.05. All averaged values are presented as means ± SEM. GraphPad Prism 3.0 software (Graph-Pad Software, San Diego, CA) was used for analysis. One-way and two-way ANOVA, repeated measures where appropriate, with Bonferroni post hoc tests, and paired or unpaired t-tests were used as indicated in the text. Group sizes were determined by power analysis using pilot data to estimate the effect size to obtain statistical significance at an alpha of 0.05 for a power of 95%. The data and statistical analyses comply with the recommendations on experimental design and analysis in pharmacology (Curtis et al., 2015).

Materials

The following drugs were used for behavioral experiments: riluzole (2-amino-6-(trifluoromethoxy)benzothiazole; Sigma-Aldrich) dissolved in 2-hydroxypropyl-β-cyclodextrin (HBC, Sigma-Aldrich; 30 %; HBC also served as the vehicle control for systemic drug injection). Total volume was increased to 1 ml with 0.9 % isotonic saline for i.p. injections (containing <10 % HBC). For stereotaxic drug administration by microdialysis, riluzole was diluted in ACSF from a stock solution dissolved in HBC (30 %). For electrophysiology experiments, a stock solution of riluzole dissolved in DMSO was created and diluted in ACSF to the final concentration (containing <0.1 % DMSO). For stereotaxic drug administration and electrophysiology experiments, apamin (SK channel blocker; Tocris Bioscience, R&D Systems, Minneapolis, MN) and charybdotoxin (BK channel blocker; Alomone Labs, Jerusalem, Israel) were diluted in ACSF from stock solutions dissolved in ACSF, which served as a vehicle control. For electrophysiology experiments, DL-AP5 (NMDA receptor antagonist, Tocris Bioscience) was diluted in ACSF from a stock solution dissolved in water.

Previous studies correlating drug effects in brain slices with drug effects using microdialysis in the intact rat showed that a 100-fold greater drug concentration in the microdialysis fiber than the desired target concentration in the tissue was needed due to the concentration gradient across the microdialysis membrane and diffusion into the brain tissue (Fu and Neugebauer, 2008; Ji et al., 2013; Ji et al., 2010). The concentrations for drug application by microdialysis (riluzole (Cao et al., 2002); apamin (Faber and Sah, 2007); ChTx (Sun et al., 2015)) and for drug application in brain slices (AP5 (Bird et al., 2005)) were chosen based on previously published electrophysiological studies.

Results

Riluzole increases mAHP and inhibits neuronal excitability in CeLC neurons

We first determined riluzole effects on the SK channel mediated mAHP, neuronal excitability (F–I relationship), and synaptic responses of regular firing neurons in the CeLC using whole-cell patch-clamp electrophysiology in brain slices from control and SNL rats. In brain slices from sham rats (Fig. 1A) riluzole (10 μM) significantly increased mAHP amplitude in 9 of 12 CeLC (n = 12 neurons; P < 0.05; F₄₁ = 24.40, repeated measures ANOVA with Bonferroni post hoc tests). In brain slices from SNL rats (Fig. 1B), an mAHP was detected in 5 CeLC (“mAHP present”, Fig. 1B, left) whereas 6 CeLC had no mAHP (Fig. 1B, right). Riluzole increased the mAHP amplitude in CeLC neurons with mAHP significantly (P < 0.01, F₁₇ = 31.55, repeated measures ANOVA with Bonferroni post hoc tests) but had no effect in neurons without mAHP (P > 0.05, F₁₇ = 1.47, repeated measures ANOVA). The riluzole effect in sham and SNL rats was confirmed to be SK channel mediated by subsequent blockade by apamin (100 nM) application.

Figure 1. Increase of mAHP and inhibition of neuronal excitability by riluzole in regular firing CeLC neurons in brain slices from sham and SNL rats.

Whole cell patch-clamp recordings. (A – B) Effects of riluzole (10 μM) on regular firing CeLC neurons in brain slices from sham rats (A; n = 7 neurons) and on CeLC neurons with mAHP (B; n = 5 neurons) and without mAHP (B; n = 6 neurons) in slices from SNL rats. The mAHP was eliminated by subsequent coapplication of an SK channel blocker (apamin, 100 nM). Traces show representative examples. Scale bars, 5 mV, 100 ms. Bar histograms show means ± SEM. ## P < 0.01 compared to riluzole, n.s. (non-significant); one-way ANOVA with Bonferroni post-tests. (C) Excitability (F–I relationship) of neurons from in brain slices from SNL rats (n = 9 neurons) is increased compared to sham controls (n = 8 neurons). Symbols show means ± SEM. *, ** P < 0.05, 0.01, n.s. (non-significant); two-way ANOVA with Bonferroni posttests. (D) Concentration-dependent inhibitory effects of riluzole on action potentials evoked with a depolarizing current of 250 pA in slices from sham rats (n = 4 neurons, 0.1 μM; n = 5 neurons, 1 μM; n = 5 neurons, 3 μM; n = 10 neurons, 10 μM; n = 5 neurons, 25 μM) and SNL rats (n = 4 neurons, 0.1 μM; n = 5 neurons, 1 μM; n = 5 neurons, 3 μM; n = 12 neurons, 10 μM; n = 3 neurons, 25 μM). Symbols show means ± SEM. (E – L) Riluzole (10 μM) inhibits action potential F–I relationship in regular firing CeLC neurons in brain slices from sham (E; n = 5 neurons) and SNL (I; n = 5 neurons) rats. Spikes were evoked by stepwise depolarizing current injections of increasing magnitude. Subsequent coapplication of apamin (100 nM) had no effect in either group. Peak (F, J) and maximum rise slope (G, K) of the first action potential formed by a 250 pA depolarizing current injection were inhibited by riluzole application in slices from SNL rats (J–L; n = 11 neurons) but not sham rats (F–H; n = 9 neurons). The maximum decay slope (H, L) was not affected by riluzole application in either group. Traces show representative examples. Scale bars: 50 mV, 100 ms, 250 pA for full current step traces; 50 mV, 5 ms for individual action potential traces. Symbols show means ± SEM. *, **, *** P < 0.05, 0.01, 0.001, n.s. (non-significant); two-way ANOVA with Bonferroni posttests (E, I); paired t-tests (F – H, J – L).

SK channels can modulate neuronal excitability of CeA but not LA-BLA neurons through the mAHP (see Introduction). Consistent with previous studies in various pain models (Bird et al., 2005; Ikeda et al., 2007; Neugebauer et al., 2003), excitability of CeLC neurons was significantly increased in brain slices from SNL rats compared to sham rats (Fig. 1C; n = 8 neurons, sham; n = 9 neurons, SNL; F_1,105 = 26.44, main effect of pain; two way ANOVA). Riluzole dose dependently inhibited the F–I relationship in CeLC neurons in sham and SNL brain slices (Fig. 1D; Sham: n = 4–10 neurons; SNL: n = 3–12 neurons). Bath application of riluzole (10 μM) significantly inhibited the CeLC frequency-current (F–I) relationship of action potential firing in neurons in brain slices from sham rats (Fig. 1E; n = 5 neurons; F_2,56 = 89.60, main effect of drug; repeated measures two way ANOVA) and SNL rats (Fig. 1I; n = 5 neurons; F_2,56 = 251.8, main effect of drug; repeated measures two way ANOVA). However, the inhibitory effect of riluzole was unaffected by subsequent addition of apamin (100 nM; Fig. 1E and 1I). Since we did not find any difference in F–I relationships and their modulation by riluzole in CeLC neurons with and without mAHP, the data were pooled for the analysis of excitability.

Analysis of the kinetics and properties of the first action potential formed by a 250 pA depolarizing current showed that riluzole had no significant effect on the action potential peak (Fig. 1F), maximum rise slope (Fig. 1G), or maximum decay slope (Fig. 1H; n = 9 neurons; P > 0.05; t-value = 1.74, F; t-value = 1.45, G; t-value = 1.93, H; paired t-tests) in CeLC neurons from sham rats but significantly reduced the peak (Fig. 1J) and maximum rise slope (Fig. 1K; n = 11 neurons; t-value = 3.18, J; t-value = 2.92, K; paired t-tests) without affecting the minimum decay slope (Fig. 1L; n = 11 neurons; P > 0.05; t-value = 1.93; paired t-test) in CeLC neurons from SNL rats. This suggests that riluzole has a selective inhibitory effect on the transient sodium current of the action potential in the SNL model but not under normal conditions.

These data suggest that while riluzole can increase the SK channel mediated mAHP in CeLC neurons, it inhibits neuronal excitability through a mechanism independent of SK channel activation.

Riluzole increases synaptic inhibition at the PB-CeLC synapse

Riluzole effects on synaptic transmission were assessed by recording excitatory (EPSP) and inhibitory (IPSP) postsynaptic potentials evoked by electrical stimulation of presumed parabrachial (PB) input into the CeA as in our previous studies (Fu and Neugebauer, 2008; Neugebauer et al., 2003; Ren et al., 2013). Individual examples of current-clamp recordings of EPSPs/IPSPs in a CeLC neuron in a brain from a sham rat and another neuron from an SNL rat are shown in Fig. 2A and 2B, respectively. Riluzole (10 μM) with and without apamin (100 nM) had no significant effect on the amplitude of the evoked EPSPs in brain slices from sham rats (Fig. 2C; n = 7 neurons; F_2,84 = 0.8459, main effect of drug, repeated measures two way ANOVA) and in slices from SNL rats (Fig. 2D; n = 6 neurons; F_2,70 = 5.254, main effect of drug; repeated measures two way ANOVA). There was a non-significant effect of riluzole on the IPSPs in brain slices from sham rats, and subsequent addition of apamin (100 nM) reduced the IPSP significantly compared to riluzole alone (Fig. 2E; n = 7 neurons; F_2,84 = 9.039, main effect of drug; repeated measures two way ANOVA with Bonferroni posttests). In brain slices from SNL rats, riluzole (10 μM) significantly increased the evoked IPSPs (Fig. 2F; n = 6 neurons; F_2,70 = 3.920, main effect of drug; repeated measures two way ANOVA with Bonferroni posttests), which was blocked by the addition of apamin (100 nM). These data suggest that riluzole increases synaptic inhibition in an SK channel dependent manner.

Figure 2. Increased synaptic inhibition of CeLC neurons by riluzole in the SNL model.

Current-clamp recordings of excitatory (EPSP) and inhibitory (IPSP) synaptic responses of regular firing CeLC neurons to PB input in brain slices from sham rats (left column) and from SNL rat s(right column). (A, B) Individual examples showing lack of effect of riluzole (10 μM) on EPSPs but facilitation of IPSPs. Scale bars, 5 mV, 100 ms. (C–F) Input output functions of synaptic transmission. Riluzole (10 μM) with and without an SK channel blocker (apamin, 100 nM) had no effect on EPSPs in brain slices from sham rats (C, n = 7 neurons) and from SNL rats (D, n = 6 neurons). Riluzole (10 μM) had a non-significant effect on IPSPs in brain slices from sham rats (E, n = 7 neurons) and significantly increased IPSP amplitude in brain slices from SNL rats (F, n = 6 neurons). Coapplication of apamin (100 nM) reversed the effects of riluzole significantly. Symbols show means ± SEM. * P < 0.05 compared to predrug, # P < 0.05 compared to riluzole; two-way ANOVA with Bonferroni posttests.

Intra-amygdala riluzole inhibits pain and depression-like behaviors but not motor function in SNL rats

Next, we sought to correlate these inhibitory electrophysiological effects to their behavioral consequences. We first characterized pain-related behavioral changes in the SNL model of neuropathic pain. SNL induction resulted in a significant and stable increase in the durations of evoked audible (Fig. 3A) and ultrasonic (Fig. 3B) vocalizations and a significant reduction in von Frey withdrawal thresholds (Fig. 3C) compared to sham controls (n = 6 sham rats; n = 11 SNL rats; F_1,75 = 26.55, audible, F_1,75 = 38.71, ultrasonic, F_1,135 = 6392, thresholds; P < 0.01-0.001, main effect of pain, repeated measures two way ANOVA with Bonferroni posttests), indicating development of nocifensive and averse affective pain behaviors and mechanosensitivity.

Figure 3. Inhibition of pain-related behaviors by intra-amygdala riluzole.

(A – F) Audible (A) and ultrasonic (B) vocalizations evoked by brief noxious stimuli and von Frey withdrawal thresholds (C) were measured before and repeatedly after left L5 spinal nerve ligation or sham surgery. Vocalizations were significantly increased and withdrawal thresholds were significantly reduced in rats after SNL (n = 11 rats) compared to sham (n = 6 rats) surgery. Stereotaxic administration of riluzole (1 mM, concentration in the microdialysis probe, 15 min) into the CeA inhibited these changes by significantly reducing evoked audible (D) and ultrasonic (E) vocalizations (n = 6 rats) but had no effect on von Frey withdrawal thresholds (F; n = 5 rats) in SNL rats compared to predrug values. Symbols and bar histograms show means ± SEM. *,**,*** P < 0.05, 0.01, 0.001; n.s. (non-significant); repeated measures two-way ANOVA with Bonferroni post-tests (A - C) or paired t-tests (D – F). (G) Immobility on the FST was increased in SNL rats (n = 25 rats) compared to control rats (n = 18 rats). (H, I) Stereotaxic application of riluzole into the CeA significantly decreased time spent immobile in the FST compared to ACSF application in SNL rats (I; n = 4) but not sham rats (H; n = 5). The same rats were tested with riluzole and ACSF. (J) The riluzole effect was blocked by coapplication of apamin (1 μM, concentration in the microdialysis probe, 15 min) and riluzole into the CeA (n = 6 rats). The same rats were tested with ACSF and apamin on subsequent days. (G–J) Bar histograms show means ± SEM. *,** P < 0.05, 0.01, n.s. (non-significant); unpaired t-test (G) and paired t-tests (H – J). (K – N) Latency to falling from the rotarod was not different between sham rats (n = 8) and SNL rats (n = 6; K). Systemic administration of riluzole significantly and dose-dependently (2 - 8 mg/kg riluzole) decreased the latency until falling from the rotarod in sham (L; n = 6 rats) and SNL (M; n = 6 rats) rats compared to vehicle injection in the same rat. Stereotaxic application of riluzole (1 mM, concentration in the microdialysis probe, 15 min) into the right CeA had no effect on the latency to falling from the rotarod compared to ACSF application in the same rat (n = 11; N). Bar histograms and symbols show means ± SEM. *, **, *** P < 0.05, 0.01, 0.001, n.s. (non-significant); unpaired t-test (K), repeated measures one-way ANOVA (L, M), paired t-test (N).

To determine behavioral effects of riluzole application into the amygdala, riluzole was administered stereotaxically (1 mM, concentration in the microdialysis probe, 15 min) into the right CeA of SNL rats. Intra-CeA riluzole significantly inhibited durations of evoked audible (Fig. 3D) and ultrasonic (Fig. 3E) vocalizations (n = 6 rats; P < 0.05; t-value = 2.96, D; t-value = 2.58, E; paired t-tests) but had no effect on von Frey withdrawal thresholds (Fig. 3F; n = 5 rats; P > 0.05, t-value = 0.93; paired t-test) compared to predrug values. Microdialysis probe position was verified histologically after experimentation (vocalizations, Fig. 4A; von Frey test, Fig. 4B). These data suggest that intra-amygdala riluzole inhibits pain-related nocifensive and averse affective behaviors in the SNL model.

Figure 4. Location of microdialysis probes for stereotaxic drug application into the amygdala.

Locations of microdialysis probes for stereotaxic administration of riluzole into the CeA are shown for vocalization experiments (A), the von Frey test (B), the forced swim test (FST) (C), and the rotarod (D). For vocalization experiments, locations of microdialysis probes are shown for stereotaxic administration of apamin into the CeA or offsite (E) and of charybdotoxin (ChTx) into the CeA (F). Locations of microdialysis probes are shown for stereotaxic administration of apamin into the CeA for the von Frey test (G). Diagrams show coronal brain slices. Numbers indicate distance from the bregma.

Because of the strong association between chronic pain and emotional-affective disorders like depression (Bushnell et al., 2013) and evidence for depression-like behaviors in the SNL model (Ji et al., 2017), we assessed riluzole effects on depression-like behaviors in SNL rats using the forced swim test (see Methods). SNL rats had significantly greater immobility on the FST compared to control rats (Fig. 3G; n = 18 rats, control; n = 25 rats, SNL; P < 0.01, t-value = 2.82, unpaired t-test), indicating development of pain-related depression-like behaviors. We used stereotaxic administration of riluzole (1 mM, concentration in the microdialysis probe, 15 min) into the amygdala (CeA) to test its effect on depression-like behaviors on the FST. To avoid confounding effects from the order of riluzole or ACSF administration, rats were randomly assigned to receive riluzole or ACSF on the first day of FST testing. Three sham rats received ACSF on Day 3 and riluzole on Day 4, and two received riluzole on Day 3 and ACSF on Day 4. Two SNL rats received ACSF on Day 3 and riluzole on Day 4, and two received riluzole on Day 3 and ACSF on Day 4. Intra-amygdala riluzole application significantly decreased the duration of time spent immobile on the FST in SNL rats (Fig. 3I; n = 4; P < 0.05, t-value = 3.77, paired t-test) but not sham rats (Fig. 3H; n = 5; P > 0.05, t-value = 0.46, paired t-test) compared to intra-CeA ACSF (vehicle control) application in the same rat, suggesting inhibition of pain-related depression-like behaviors. The inhibitory effect of riluzole was blocked by stereotaxic coapplication of apamin (1 μM, concentration in the microdialysis probe, 15 min) with riluzole into the CeA (Fig. 3J; n = 6 rats; P > 0.05, t-value = 1.55, paired t-test) compared to ACSF application in the same rat, suggesting SK channel involvement in the antidepressant effects of stereotaxic riluzole. There was no effect of apamin alone on immobility in a cohort of 7 rats, in which this was tested, compared to ACSF application in the same rat (immobility = 143.8 s, ACSF; immobility = 141.6 s, apamin; P > 0.05, t-value = 0.40, paired t-test). The data suggest that the lack of riluzole effects in the presence of apamin was due blockade of SK channel mediated riluzole actions by apamin rather than an independent action of apamin. Microdialysis probe tip position was verified histologically after experimentation (Fig. 4C).

Next we tested if stereotaxic drug administration of riluzole (1 mM, concentration in the microdialysis probe, 15 min) into the amygdala (CeA) had any effects on motor function using the rotarod test (see Methods). Intra-CeA riluzole had no effect on latency to fall compared to ACSF (vehicle control) in the same rat (Fig. 3N; n = 11; P > 0.05, t-value = 0.26, paired t-test). Microdialysis probe position was verified histologically after experimentation (Fig. 4D). Importantly, no motor deficit was found in the rotarod test in SNL rats compared to sham controls (Fig. 3K). Latency to fall was not significantly different (P > 0.05, t-value = 0.50, unpaired t-test) between SNL rats (n = 6) and sham rats (n = 6). Therefore, there should be no confounding motor impairment in the SNL pain model affecting FST performance or drug effects. These data suggest that effects of riluzole on vocalizations were not due to a sedating or anesthetic effect, and provide more support to antidepressant-like effects of riluzole on the FST instead of confounding motor changes.

Systemic riluzole inhibits vocalizations and increases spinal reflex thresholds

Next, we determined the effects of systemic riluzole injection on pain behaviors in order to assess the clinical utility of this therapeutic strategy. In sham operated rats, systemic injection of riluzole (8 mg·kg⁻¹, i.p.; n = 5 rats) had no effect on the duration of evoked audible (Fig. 5A) or ultrasonic (Fig. 5B) vocalizations or on spinal withdrawal thresholds (Fig. 5C) compared to those after vehicle (2-hydroxypropyl-β-cyclodexterin, HBC; 30%, i.p.) injection in the same rat (P > 0.05, paired t-tests). In SNL rats, systemic injection (i.p.) of riluzole significantly and dose-dependently (2 mg·kg⁻¹, n = 9 rats; 4 mg·kg⁻¹, n = 9 rats; 8 mg·kg⁻¹, n = 9 rats) inhibited durations of evoked audible (Fig. 5A; apparent EC₅₀=2.0 mg·kg⁻¹) and ultrasonic (Fig. 5B; apparent EC₅₀=1.8 mg·kg⁻¹) vocalizations and increased von Frey withdrawal thresholds (Fig. 5C; apparent EC₅₀=2.8 mg·kg⁻¹) compared to those after vehicle injections (HBC, 30%, i.p.; n = 9 rats) significantly (F₃₅ = 10.46, audible, F₃₅ = 6.20, ultrasonic, F₄₀ = 4.681, thresholds, P < 0.05-0.001, one way ANOVA with Bonferroni posttests). The data show inhibitory effects of riluzole on nocifensive and averse affective pain behaviors and mechanosensitivity.

Figure 5. Contribution of SK channels in the CeA to the inhibitory effects of systemic riluzole in SNL rats.

(A – C) Systemic (i.p.) riluzole injection significantly and dose dependently attenuated evoked audible (A) and ultrasonic (B) vocalizations (n = 9 rats for each dose) and dose dependently increased withdrawal thresholds (C) in SNL rats (n = 9 rats [2, and 4 mg·kg ⁻¹]; n = 14 rats [8 mg·kg ⁻¹]) measured 1 h post-injection compared to vehicle (n = 9 rats) injection. However, in sham rats (n = 5 rats) systemic riluzole (8 mg·kg ⁻¹) injection had no effect on vocalizations or withdrawal thresholds compared to vehicle injection in the same rat. Symbols show means ± SEM. *, **, *** P < 0.05, 0.01, 0.001, n.s. (non-significant); one-way ANOVA with Bonferroni post-tests (riluzole effects in SNL rats), or paired t-test (riluzole effects in sham rats). (D – G) Systemic riluzole (8 mg·kg ⁻¹) with stereotaxic coapplication of ACSF (15 min; vehicle control; n = 9 rats) into the CeA resulted in significant inhibition of audible and ultrasonic vocalizations (D). However, systemic riluzole with intra-CeA SK channel blockade (apamin; 1 μM, concentration in the microdialysis probe, 15 min; n = 6 rats) eliminated this effect (E). Stereotaxic coapplication of a BK channel blocker (F; charybdotoxin, ChTx, 1 μM, concentration in the microdialysis probe, 15 min; n = 7 rats) into the CeA or stereotaxic coapplication of apamin outside of the CeA (G; BLA, n = 4 rats; IC, n = 2 rats; combined n = 6 rats) did not block the inhibitory effects of systemic riluzole on audible and ultrasonic vocalizations. Bar histograms show means ± SEM. ** P < 0.01; n.s. (non-significant); paired t-tests. (H, I) Systemic riluzole with stereotaxic application of ACSF (H; n = 6 rats) and apamin (I; 1 μM, concentration in the microdialysis probe, 15 min; n = 7 rats) into the CeA significantly increased von Frey withdrawal thresholds compared to predrug values. Bar histograms show means ± SEM. *, *** P < 0.05, 0.001; paired t-tests.

We also determined systemic riluzole effects on motor function using the rotarod test. Systemic injection (i.p.) of riluzole decreased the latency to falling from the rotarod in sham rats (Fig. 3L; n = 6; apparent IC₅₀ = 3.9 mg/kg) and SNL rats (Fig. 3M; n = 6; apparent IC₅₀ = 3.9 mg/kg) dose-dependently and significantly (F₂₃ = 6.08, sham; F₂₃ = 4.41, SNL; P < 0.05-0.001, repeated measures one way ANOVA with Bonferroni posttests) compared to vehicle administration in the same rat, suggesting inhibition of motor function. However, the lack of motor effects with stereotaxic drug administration suggest that this is an off target action and does not result from increasing SK channel activity in the amygdala.

Riluzole inhibits vocalizations through an action on SK channels in the CeA

To assess SK channel contributions in the CeA to the inhibitory behavioral effects of systemic riluzole, a selective SK channel blocker (apamin; 1 μM, concentration in the microdialysis probe, 15 min) was stereotaxically applied into the right CeA beginning 45 min after systemic injection of riluzole (8 mg·kg⁻¹, i.p.) for vocalization measurements 1 h post-injection in SNL rats. The dose of 8 mg·kg⁻¹ was chosen because it resulted in the most consistent inhibitory effects when systemically administered (see Fig. 5A–C). Systemic administration of riluzole in combination with stereotaxic application of ACSF (15 min) into the right CeA significantly inhibited durations of audible (Fig. 5D) and ultrasonic (Fig. 5E) vocalizations (evoked by brief noxious stimuli, see Methods) compared to predrug values (n = 9 rats; P < 0.01 and 0.001; t-value = 4.48, D; t-value = 5.88, E; paired t-tests). However, systemic administration of riluzole together with stereotaxic application of apamin into the right CeA attenuated those actions, resulting in no significant effect of systemic riluzole on evoked vocalizations compared to predrug values (n = 6 rats; P > 0.05,; t-value = 1.30, D; t-value = 1.86; paired t-tests), suggesting that SK channels in the CeA are involved in systemic riluzole actions. There was no effect of apamin alone on vocalizations in a cohort of 7 rats, in which this was tested, compared to predrug values (audible = 4.60 s, ACSF; audible = 4.28 s, apamin; ultrasonic = 4.39 s, ACSF; ultrasonic = 3.75 s, apamin; P > 0.05, t-value = 0.38, audible; P > 0.05, t-value = 0.74, ultrasonic; paired t-tests). The data suggest that the lack of significant riluzole effects in the presence of apamin was due blockade of SK channel mediated riluzole actions by apamin rather than an independent action of apamin.

To determine if this effect was selective to SK channels in the CeA, inhibitory effects of systemic riluzole on audible (Fig. 5F) and ultrasonic (Fig. 5G) vocalizations were assessed in the presence of a large-conductance calcium-activated potassium (BK) channel blocker (charybdotoxin, ChTx) as well as with apamin offsite application into nearby structures. Significant inhibitory effects of systemic riluzole persisted with stereotaxic blockade of BK channels (ChTx, 1 μM, concentration in the microdialysis probe, 15 min) in the right CeA (n = 7 rats; P < 0.01; t-value = 4.42, F; t-value = 4.66, G; paired t-test) and with stereotaxic coapplication of apamin (1 μM, concentration in the microdialysis probe, 15 min) into nearby sites (n = 4 rats, BLA; n = 2 rats, internal capsule, IC; combined offsite, n = 6 rats; P < 0.01; t-value = 4.60, F; t-value = 5.05, G; paired t-test). Microdialysis probe position was verified histologically after experimentation (apamin, Fig. 4E; ChTx, Fig. 4F).

To assess SK channel contributions to the inhibitory effects of riluzole on mechanosensitivty, von Frey withdrawal thresholds were determined using the same procedures and time points described above for vocalization experiments. Systemic administration of riluzole in combination with stereotaxic application of ACSF (15 min) into the right CeA significantly increased von Frey withdrawal thresholds compared to predrug values (Fig. 5H; n = 6 rats; P < 0.001, t-value = 7.69 paired t-test). This effect persisted after SK channel blockade with apamin in the CeA (Fig. 5I; n = 7 rats; P < 0.05, t-value = 2.90, paired t-test). There was no effect of apamin alone on withdrawal thresholds in a cohort of 7 rats, in which this was tested, compared to predrug values (threshold = 1.04 g, ACSF; threshold = 0.89 g, apamin; P > 0.05, t-value = 0.59, paired t-test). The data suggest that SK channels in the CeA are not involved in the inhibitory effects of systemic riluzole on mechanosensitivity, and also argue against a non-specific action in the amygdala.

Discussion

Riluzole is an FDA-approved drug indicated for the treatment of amyotrophic lateral sclerosis. In this study, we demonstrated that riluzole inhibits pain-related amygdala hyperactivity in a partially SK channel-dependent manner, which corresponded to antinociceptive and antidepressant-like effects at the behavioral level. To the best of our knowledge, this is the first study demonstrating SK channel mediated mechanisms of antinociceptive and antidepressant-like activity of riluzole in the central amygdala in a model of chronic neuropathic pain. This data is significant because it provides preclinical support for targeting SK channels in the CeA for neuropathic pain treatment and describes the mechanisms by which these antinociceptive effects occur.

Riluzole increased the amplitude of the SK channel mediated mAHP in some but not all neurons in the CeA. There was an apparent reduction in mAHP amplitude and absent mAHP in SNL neurons compared to control conditions, and these groups exhibited differential effects of riluzole on the mAHP. SNL neurons with a reduced predrug mAHP had an increased mAHP amplitude after riluzole application, whereas neurons without a predrug mAHP were unaffected. Because SK channels in the central amygdala regulate action potential firing frequency (Lopez de Armentia and Sah, 2004), this deficit could contribute to amygdala pain-related hyperactivity. Contributions of SK channel dysfunction in the amygdala to pain-related plasticity remain to be determined.

Riluzole also inhibited the F–I relationship in CeA neurons, but this was not reversed with SK channel blockade, suggesting that mechanisms other than SK channel activation are involved in riluzole-mediated inhibition of neuronal excitability. Riluzole is a potent inhibitor of sodium currents (Bellingham, 2011). We found that riluzole decreased action potential rise slope and peak voltage in neurons from SNL brain slices, suggesting inhibition of sodium currents. The EC₅₀ of riluzole inhibition of neuronal excitability (1.0 μM in sham and 2.6 μM in SNL neurons) is consistent with an action on the persistent sodium current (EC₅₀ of 2 μM; (Bellingham, 2011)). This action could contribute to SK channel-independent inhibition of neuronal excitability.

Stimulation of the PB-CeLC synapse results in a monosynaptic excitatory and polysynaptic inhibitory response in regular firing CeLC neurons (Lopez de Armentia and Sah, 2004). The EPSP amplitude formed from stimulation of this pathway was unaffected by riluzole application in both sham and SNL conditions, suggesting that increased SK channel-mediated shunting of excitatory transmission at this synapse is not an effect of riluzole. However, riluzole application increased the IPSP amplitude and was reversed by subsequent SK channel blockade. This indicates that riluzole increases synaptic inhibition at this synapse in an SK channel dependent manner. Synaptic responses were determined in current clamp to allow for an integrative analysis of synaptic function. Because excitatory and inhibitory responses were not isolated, modulation of the EPSP could result in a corresponding change in IPSP amplitude. The lack of riluzole effects on EPSP amplitude in either sham or SNL rats argues against such confounding effects and suggests that riluzole specifically enhances inhibitory synaptic transmission. A direct effect of SK channels on an IPSP has previously been described in the midbrain as a consequence of mGluR1 activation and subsequent inositol 1,4,5-triphosphate (IP3)- and cyclic ADP-ribose (cADPR)-induced mobilization of endogenous calcium reserves from intracellular stores (Fiorillo and Williams, 1998; Morikawa et al., 2000; Morikawa et al., 2003). Group I mGluRs in the amygdala circuitry are involved in the arthritis pain-related imbalance of excitatory and inhibitory synaptic transmission, and because they are coupled to G_q/11 proteins and mobilize intracellular calcium stores (Li et al., 2011; Neugebauer, 2015; Neugebauer et al., 2003; Ren and Neugebauer, 2010), enhanced activation of group I mGluR in the pain state could affect SK channel function. Potential links between amygdala mGluR function and synaptic SK channel function remain to be determined.

The data suggest that the SK channel-dependent behavioral effects of riluzole are best explained by actions on the mAHP and on synaptic inhibition. Riluzole is known to directly activate SK channels (Cao et al., 2002), although we cannot exclude riluzole actions on possible upstream mechanisms including altered SK channel phosphorylation as contributing to these effects. The effects of riluzole on neuronal excitability were not reversed by SK channel blockade and are consistent with an action on sodium currents, and so it is possible that this riluzole effect contributes to SK channel-independent inhibitory behavioral effects. Riluzole has a lower potency for rat SK2 channels (43 μM; (Cao et al., 2002)) than for inhibition of the persistent sodium current (EC₅₀ of 2 μM; (Bellingham, 2011)), and so it is possible that SK channel independent riluzole effects on neuronal excitability could predominate at the dose tested in this study. It is also possible that impaired SK channel function in the pain state could result in a dissociation between mAHP and excitability.

Inhibitory electrophysiological effects correlated with reduced pain and depression-like responses in the SNL pain model. Intra-CeA application of riluzole inhibited vocalizations and depression-like behaviors without affecting spinal withdrawal thresholds, indicating that the amygdala is involved in mediating supraspinally organized riluzole actions. Different brain regions including the amygdala contribute to reflex behaviors through top-down pain modulation, and some manipulations can affect these behaviors by modulating this neurocircuitry (Neugebauer, 2015; Thompson and Neugebauer, 2017). The lack of effect on withdrawal thresholds with intra-CeA riluzole indicates that effects on this neurocircuitry are not a significant drug action, and is consistent with previous studies have indicated that riluzole may have a supraspinal site of action (Abarca et al., 2000; Hama and Sagen, 2011; Palazzo et al., 2002). We demonstrated the clinical relevance of this therapeutic strategy through systemic drug injection. Systemic riluzole administration inhibited evoked vocalizations and increased spinal withdrawal thresholds in SNL but not sham rats, arguing against an anesthetic effect. Because latency to fall on the rotarod test was reduced in both SNL and sham rats, it is unlikely that impaired motor function contributed to withdrawal threshold effects observed only in SNL. Riluzole can block voltage-gated sodium channels in human myotubes (Deflorio et al., 2014), leading to the possibility that motor impairment could result from an off-target peripheral action on muscle tissue. Vocalizations but not spinal withdrawal thresholds returned to near-normal levels in SNL rats after riluzole application, which is consistent with greater riluzole effects on supraspinal areas of the nervous system involved in mediating vocalization responses rather than on neurocircuitry involved in mediating spinal reflex responses. Furthermore, this effect is likely mediated by an extra-amygdala site because it was not present with stereotaxic drug administration. A previous study demonstrated increased spinal withdrawal thresholds with intracerebroventricular riluzole administration but not intrathecal administration (Hama and Sagen, 2011), suggesting that another brain region involved in top-down pain modulation may contribute to riluzole effects on spinal withdrawal thresholds.

We confirmed involvement of SK channels in the CeA in these inhibitory actions of riluzole because SK channel blockade in the CeA attenuated beneficial effects on supraspinally organized behaviors. This was a selective effect because SK channel blockade in neighboring structures as well as blockade of BK channels did not affect systemic riluzole actions. Interestingly, in contrast to our previous study in a model of acute arthritic pain (Thompson et al., 2015), intra-CeA apamin did not fully reverse the inhibitory behavioral effects of systemic riluzole. This could indicate that compared to the acute arthritis pain state, SK channel function is deficient and cannot be pharmacologically corrected in the chronic neuropathic pain state and provides support to a maladaptive change in SK channel regulation contributing to chronic pain-related amygdala plasticity. Because of this, alternative molecular targets of riluzole might have greater contributions to the observed antinociceptive effects in neuropathic pain.

Although we did not observe changes in pain-related behaviors in sham rats, we did see electrophysiological effects including significant inhibition of the F–I relationship. Previous studies have reported elevated amygdala activity in the pain state (Neugebauer, 2015; Thompson and Neugebauer, 2017), and our data are consistent with these changes. Manipulations in the amygdala could require the elevated activity present in the pain state to result in significant behavioral effects; neuronal activity might not be sufficient under normal conditions for further inhibition by riluzole to result in behavioral consequences. It is also possible that the predrug vocalizations and spinal withdrawal thresholds were too near the limits of detection for our tests, resulting in a floor and ceiling effect respectively that reduced the ability to detect behavioral changes associated with riluzole administration.

Throughout this study, we have focused on interventions in the right amygdala because it is contralateral to the side of injury and preclinical data indicate that the amygdala response to pain is hemispherically lateralized to the right side in acute inflammatory (Carrasquillo and Gereau, 2007, 2008; Ji and Neugebauer, 2009) and chronic SNL-induced neuropathic (Goncalves and Dickenson, 2012) pain. However, we cannot exclude the possibility that the left amygdala, or other brain regions or molecular targets, could play a role in mediating riluzole effects.

In summary, riluzole is a clinically available drug that inhibits neuronal activity and increases synaptic inhibition in the central amygdala in a partially SK channel dependent manner, resulting in inhibition of neuropathic pain-related behaviors. However, data from the present study showing loss of SK-channel mediated mAHP in some cells in the neuropathic pain model (Fig. 1B) suggest that SK channel function in the CeA under neuropathic pain conditions may be deficient and could contribute to maladaptive pain-related amygdala plasticity (see Thompson and Neugebauer, 2017). This novel concept remains to be explored. The present study suggests that SK channels in the CeA are accessible to systemically administered drugs and supports investigation into SK channel-targeted therapeutic strategies for chronic pain.

Highlights.

• Riluzole, a clinically available therapeutic, inhibits averse affective neuropathic pain behaviors.

• Behavioral effects of riluzole are blocked by potassium (SK) channel blockade in the amygdala.

• Riluzole increases SK-channel dependent medium afterhyperpolarization (mAHP) of amygdala neurons.

• Riluzole increases synaptic inhibition in amygdala brain slices in an SK channel-dependent manner.

• Inhibitory effects of riluzole on excitability of amygdala neurons are SK-channel independent.

List of abbreviations

ACSF

artificial cerebrospinal fluid

ALS

amyotrophic lateral sclerosis

BK channel

large-conductance calcium-activated potassium channel

BLA

basolateral nucleus of the amygdala

CeA

central nucleus of the amygdala

CeLC

laterocapsular division of the central nucleus of the amygdala

ChTx

charybdotoxin

EPSP

excitatory postsynaptic potential

F–I

frequency-current

FST

forced swim test

H&E

hematoxylin and eosin

HBC

2-hydroxypropyl-β-cyclodextrin

IC

internal capsule

I_NaP

persistent sodium channel current

IPSP

inhibitory postsynaptic potential

LA-BLA

lateral-basolateral complex of the amygdala

mAHP

medium afterhyperpolarization

PB

parabrachial area (external lateral)

SK channel

small-conductance calcium-activated potassium channel

SNL

spinal nerve ligation

TTX

tetrodotoxin