Amygdala Plasticity and Pain

The amygdala is a limbic brain region that plays a key role in emotional processing, neuropsychiatric disorders, and the emotional-affective dimension of pain. Preclinical and clinical studies have identified amygdala hyperactivity as well as impairment of cortical control mechanisms in pain states. Hyperactivity of basolateral amygdala (BLA) neurons generates enhanced feedforward inhibition and deactivation of the medial prefrontal cortex (mPFC), resulting in pain-related cognitive deficits. The mPFC sends excitatory projections to GABAergic neurons in the intercalated cell mass (ITC) in the amygdala, which project to the laterocapsular division of the central nucleus of the amygdala (CeLC; output nucleus) and serve gating functions for amygdala output. Impairment of these cortical control mechanisms allows the development of amygdala pain plasticity. Mechanisms of abnormal amygdala activity in pain with particular focus on loss of cortical control mechanisms as well as new strategies to correct pain-related amygdala dysfunction will be discussed in the present review.


The Amygdala and Pain
e amygdala is an almond-shaped limbic structure located in the medial temporal lobe and is well known for its role in conveying emotional signi cance to a sensory stimulus, emotional and a ective states, and related behavioral adaptations in response to changes in the internal and external bodily environment [1][2][3][4]. e amygdala has also emerged as an important site in the brain for the emotional-a ective dimension of pain and pain modulation [5][6][7][8][9][10][11][12].
A pain-related function was rst suggested by the discovery of a dedicated nociceptive pathway from the spinal cord through the external lateral parabrachial (PB) nucleus to the central nucleus of the amygdala [13,14]. Reevaluation of an historical example of reduced pain sensitivity also suggests amygdala involvement in pain processing. Patient H.M. was a man that underwent bilateral resection of the temporal lobe including the uncus, amygdala, anterior hippocampus, and parahippocampal gyrus to correct severe and intractable epilepsy [15][16][17]. After the surgery, H.M. did not perceive even the highest thermal stimulus intensity as painful when control groups did. It is now thought that this de cit was likely due to amygdala resection [16,17], illustrating the importance of the amygdala in pain processing in the brain. Importantly, this de cit in pain perception occurred despite an intact nociceptive system and was not accompanied by the tissue injury characteristic of pain insensitivity disorders, indicating that protective pain functions were intact.
e clinical relevance of these ndings has been corroborated by human neuroimaging studies that demonstrate amygdala activation in response to experimental noxious stimuli, including mechanical compression, thermal stimulation, and capsaicin application [10], as well as increased amygdala activity in migraineurs compared to healthy controls when presented with negative but not positive or neutral emotional stimuli [45]. In addition, functional connectivity between the left amygdala and the PFC, cingulate cortex, and basal ganglia is di erent in patients with complex regional pain syndrome (CRPS) [46], and corticolimbic reverberating loops have been implicated in the prediction of and transitioning to chronic pain [11]. Patients with irritable bowel syndrome (IBS) had higher positive resting-state functional connectivity between the amygdala and the insula, pre-and postcentral gyri, and supplementary motor area compared to healthy controls, and this increased connectivity positively correlated to pain intensity [47]. A separate study demonstrated that IBS patients that did not have visceral hypersensitivity had decreased positive resting-state functional connectivity of the amygdala within the default mode network compared to healthy controls as well as IBS patients with visceral hypersensitivity [48]. In female twin pairs with and without chronic pelvic pain, connectivity between the right PAG and the right amygdala, connectivity between the left PAG and the right and left basolateral amygdala, and connectivity of the right basolateral amygdala to the medial orbital frontal cortex, anterior cingulate cortex (ACC), right insula, left thalamus, and hypothalamus differed between the twin with pelvic pain compared to the healthy twin before and after bladder distension by an oral water bolus [49].

Amygdala Pain Neurocircuitry
e amygdala receives multiple lines of input ( Figure 1) relevant for pain processing, and multiple nuclei in the amygdala are involved in its pain processing functions. ese include the lateral-basolateral complex (LA/BLA), the central nucleus (CeA), and the intercalated cell mass (ITC); see Figure 2 and [7][8][9]. e LA/BLA is predominantly composed of pyramidal glutamatergic projection neurons that receive polymodal sensory, including nociceptive, inputs from the midline and posterior nuclei of the thalamus, insular cortex, and sensory association cortices, as well as inputs from the ACC and medial prefrontal cortex (mPFC) [1,[7][8][9]12]. rough associative processing, the LA/BLA attaches emotional-a ective content to the sensory inputs and transmits that highly processed information to the amygdala output region in the CeA for further processing as part of amygdala fear and anxiety circuitry [3,50,59,60]. is LA/BLA-CeA projection is now known to generate and modulate pain-related behaviors [7]. e BLA also projects to di erent cortical areas, including the infra-and prelimbic mPFC, ACC, and perirhinal and insular cortices [51,[61][62][63][64][65][66][67].
e CeA serves as the major output nucleus for amygdaladriven pain-related functions. e laterocapsular region of the CeA (CeLC, including both the lateral and capsular divisions) receives purely nociceptive information via the spinoparabrachio-amygdaloid tract [72] and possibly via direct projections from the spinal cord [73,74]. e parabrachial input is characterized by its peptidergic nature and serves as the exclusive source of calcitonin gene-related peptide (CGRP) for the amygdala [75].

Pain Research and Management
Interposed between LA/BLA and CeLC is a group of GABAergic interneurons in the intercalated cell mass (ITC). ITC cells receive excitatory inputs from the infralimbic mPFC and LA/BLA and are activated during extinction of negative emotional responses [2,3,50,[78][79][80][81][82]. erefore, ITC cells serve a gating function for amygdala output from the CeA through feedforward inhibition that involves activation of neuropeptide S (NPS) receptors on ITC cells [39,83,84].
CeA output arises from medial CeA (CeM) and CRF-CeLC projection neurons. ese projections target other limbic, hypothalamic, and brainstem regions including the periaqueductal gray (PAG) [3,60,[85][86][87]. Projections to hypothalamus and brainstem regions such as the nucleus of the solitary tract (NTS) contribute to amygdala-driven autonomic responses. Projections to the PAG generate or modulate vocalizations and startle responses. CeA projections to other limbic structures such as the bed nucleus of the stria terminalis (BNST) and to monoaminergic regions such as locus coeruleus are thought to drive amygdala in uences over anxiety and depression [3,88,89]. CeLC neurons can also a ect output from the CeM through intra-amygdala connections involving GABAergic protein kinase C delta (PKCδ) positive and other interneurons in the lateral CeA [50,[90][91][92], but details of this intra-amygdalar circuitry remain to be determined.
CeA neurons expressing PKCδ (PKCδ + ) are inhibited by exposure to the fear-conditioned stimulus and have been termed "OFF cells,", whereas PKCδ-negative neurons are excited when presented with the conditioned stimulus, and reciprocal inhibitory connections exist between these cell types [1,90,91]. Inhibition of PKCδ + cells is believed to contribute to emotional responses such as freezing through disinhibition of CeM output neurons [50]. Importantly, PKCδ + neurons are distinct from CRF-containing CeLC projection neurons characterized by their nonaccommodating spike ring pattern [87,91]. And so, it is reasonable to assume that it is the population of PKCδ-negative CRF-CeLC neurons that have been studied for their nociceptive PB input and participation in pain processing [7]. CRF-CeLC neurons are uniquely positioned to provide and regulate amygdala output through their projections to extra-amygdalar targets and through local inhibition of PKCδ + cells to disinhibit CeM output neurons [1,3,60,[85][86][87] (Figure 2).

Pain-Related Changes in Amygdala Neurocircuitry
Importantly, changes in the amygdala neurocircuitry have been detected in di erent preclinical models of pain and linked mechanistically to pain behaviors, indicating that these maladaptive neuroplastic changes are a brain mechanism of pain. In systems (whole animal) electrophysiology studies of amygdala neurons in anesthetized rats, background activity and responses to mechanical compression of peripheral tissues increased in an arthritis pain model in the CeLC [36,37,[93][94][95][96][97] and BLA [71], as well as in the CeLC in a neuropathic pain model [42,44]. ese activity changes are not simply a re ection of changes along the pain pathways to the amygdala but arise from synaptic plasticity within the amygdala network. Brain slice physiology studies showed enhanced excitatory transmission at the PB-CeLC and BLA-CeLC synapses [33-35, 38, 40], decreased ITC-mediated synaptic inhibition of CeLC neurons [39,40], and increased neuronal excitability in the CeLC [33,34,38], as well as enhanced excitatory transmission at the LA-BLA synapse and increased excitability of BLA neurons [71] in an arthritis pain model. Enhanced transmission at the PB-CeLC synapse and neuronal excitability [25] and increased levels of neurochemical activity markers (c-Fos and extracellular signalregulated kinase, ERK) [22,23,28,98] and CRF [24,26] were found in the CeA in visceral pain models. Enhanced PB-CeLC and BLA-CeLC neurotransmission and excitability [43,99] and increased expression of CRF and glucocorticoid receptors [100,101] in the CeA were also observed in neuropathic pain models. It should be noted that not all of these neuroplastic changes have been described in all pain models. It is possible that amygdala neuroplasticity could  di er based on time course (acute versus chronic pain) or etiology (in ammatory versus neuropathic) of pain. e full scope of pain-related amygdala plasticity remains an area of active investigation.
e clinical relevance of these neuroplastic changes in preclinical pain models is supported by neuroimaging studies, indicating that amygdala activity is increased in subjects with previously diagnosed pain conditions, including osteoarthritis, IBS, and bromyalgia compared to matched controls [10].
Interestingly, this amygdala pain-related plasticity exhibits hemispheric lateralization. Under normal conditions, nociceptive inputs into the CeLC are received by and activate both the right and left amygdala [30,36]. No apparent di erence has been detected in background activity or responses to innocuous and noxious mechanical test stimuli between neurons in the right and left amygdala (CeLC), although neurons in the left amygdala appear to have more restricted receptive elds in peripheral tissues [36]. Exogenous activators also can increase neuronal activity in the right and left CeLC [36] or induce hypersensitivity [102]. In in ammatory [36] and neuropathic [42] pain conditions, neurons in the right but not left amygdala exhibit a sustained increase in background and evoked activity irrespective of the side of injury, and the receptive eld size of neurons in the right but not left amygdala is increased in an arthritis pain model [36]. In humans, di erences in connectivity between the right and left amygdala have been observed in IBS patients compared to healthy controls [47], IBS patients compared to those without visceral hypersensitivity [48], and twin pairs with one twin with and one without chronic pelvic pain before and after bladder distension [49].
Mechanisms of hemispheric lateralization of amygdala plasticity and function are not yet clear. Protein kinase A (PKA) activation has been implicated in the development of central sensitization [103] and synaptic plasticity [33][34][35]104] in the CeLC in an arthritis pain model, and application of a PKA inhibitor into the right but not left amygdala resulted in a reduction in neuronal activity in this pain model [36], indicating that PKA is endogenously activated in the right but not left amygdala in the pain state. ERK is also known to play an important role in amygdala plasticity and amygdala-driven behaviors in in ammatory pain models [30,104], and activation of the metabotropic glutamate receptor 5 (mGluR5) has been shown to increase amygdala ERK activation and ERK-dependent pain behaviors [102,105]. Application of a blocker of ERK activation [30,31] and pharmacological blockade or conditional deletion of mGluR5 [102] prevented formalin-induced hypersensitivity when administered into the right but not left amygdala. Chemical (pituitary adenylate cyclase-activating polypeptide) or optogenetic activation of the right but not left CeA increased nociceptive visceromotor responses in a murine urinary bladder distension model, and optogenetic silencing of the left but not right CeA increased bladder distensioninduced visceromotor responses [27]. In a visceral pain model cyclophosphamide-induced cystitis, optogenetic activation of the left but not right CeA inhibited abdominal mechanosensitivity, whereas activation of the right CeA further increased visceromotor responses in this model [27] ese ndings may suggest that while neuroplasticity in the right CeA drives pain-related behaviors, the left CeA may be linked to antinociception. Evidence from neuroimaging studies in humans also suggests right hemispheric lateralization of amygdala responses to acute experimental pain stimuli [10], whereas in clinical pain conditions, right hemispheric deactivation [106] and left hemispheric signal increases [10] have been reported. is could be due to a compensatory increase in inhibitory transmission in the left amygdala that is not present in the right amygdala in the chronic pain state. Painrelated hemispheric lateralization may re ect a general principle of lateralized emotional processing. Right hemispheric amygdala activation has been found in response to masked fearful faces [107] and was exaggerated in veterans with posttraumatic stress disorder (PTSD) [108]. Enhanced activity of the right, but not the left, amygdala in men was related to encoding and long-term memory of lms judged as arousing negative emotions compared to neutral lms [109,110]. Mechanisms and signi cance of pain-related amygdala lateralization remain to be determined.

Pain-Related Amygdala-Centered Corticolimbic Interactions
Information processing in the amygdala can be regulated by inhibitory gating mechanisms centered on ITC cells and their activation by cortical control systems (Figure 2). e mPFC in uences amygdala function through feedforward inhibition of CeLC neurons via excitatory projections to ITC cells as an important mechanism of cognitive modulation of emotions such as a fear [2,3,50,78,79,81,82,88,111]. Evidence suggests that mPFC-driven feedforward inhibition of CeLC output neurons is impaired in pain [7,112]. Electrical stimulation of the external capsule, including infralimbic mPFC inputs into the amygdala, resulted in a non-N-methyl-D-aspartate (non-NMDA) receptor-mediated monosynaptic excitatory synaptic response (EPSC) in dorsomedial ITC cells and non-NMDA receptor-driven synaptic inhibition (IPSC) of CeLC neurons [39]. In brain slices from arthritic rats, the monosynaptic EPSC in ITC cells and the glutamate-driven IPSP in CeLC neurons were reduced [39], suggesting painrelated impairment of mPFC-driven feedforward inhibition of amygdala output. Decreased infralimbic mPFC activity has been implicated in extinction de cits [113][114][115][116]. Accumulating evidence points to mPFC deactivation in pain, which could explain impaired control of amygdala processing [7,112]. Functional and structural abnormalities in the mPFC have been detected in human pain patients [117,118] and in preclinical pain models [71,[119][120][121]. As a consequence, activity of output neurons in the infralimbic and prelimbic mPFC is decreased in acute [71,122,123] and chronic pain models [120,124,125]. Decreased glutamatergic drive of pyramidal cells [120] and abnormally enhanced glutamatergic activation of parvalbumin-expressing GABAergic interneurons [71,125,126] have been implicated in the mPFC deactivation in pain.

Pain Research and Management
Hyperactivity in the BLA plays an important role in pain-related mPFC deactivation [7,112]. e BLA sends glutamatergic projections to the pre-and infralimbic mPFC [61,66,67]. Importantly, while some of these BLA axon terminals make direct contact with pyramidal cells, the majority of synapses on neighboring parvalbumin and somatostatin-positive interneurons form GABAergic connections with pyramidal cells, targeting mainly the somatic and proximal axonal regions [62,127].
is synaptic arrangement was shown to account for amygdala-driven mPFC deactivation by glutamate-driven feedforward inhibition in an arthritis pain model [71,126]. Feedforward inhibition involves activation of GABAergic interneurons mediated by non-NMDA receptors and mGluR1 but not mGluR5 [122,128]. It should be noted that mGluR5 in the mPFC is expressed mostly on postsynaptic elements [129] to exert excitatory e ects on pyramidal cells [130][131][132]. In contrast, GABAergic inputs to mGluR5 expressing mPFC pyramidal cells are regulated by cannabinoid CB1 receptors under normal conditions [131,133]. In brain slices from arthritic rats, IPSCs evoked by electrical or optogenetic activation of BLA axon terminals in pre-and infralimbic mPFC pyramidal cells were increased; IPSCs could be blocked with non-NMDA glutamate receptor and GABA A receptor antagonists [71,126]. Systems electrophysiology studies in anesthetized rats showed that pain-related decreases in background and evoked activity of prelimbic mPFC pyramidal-like neurons were reversed by a GABA A receptor antagonist and attenuated by an mGluR1 but not mGluR5 antagonist [122]. e decrease in mPFC pyramidal cell activity was causally linked to increased BLA neuronal activity in the arthritis pain model because restoring BLA activity with a CRF1 antagonist increased background and evoked activity of prelimbic mPFC neurons [71].

Pharmacological Strategies Targeting Amygdala Pain Neurocircuitry
Interventions that increase amygdala output, even in the absence of acute injury, elicit pain behaviors [27,30,102,134,135], whereas those that decrease amygdala activity generally inhibit pain behaviors (see [7] for review). erefore, controlling amygdala activity is a desirable therapeutic strategy for chronic pain. Interventions that were found to have some bene cial e ect in preclinical studies include non-NMDA and NMDA receptor antagonists, mGluR1 and mGluR5 antagonists, agonists for group II mGluR2/3 and group III mGluR, including mGluR8, antagonists for CGRP1 and CRF1 receptors, neuropeptide S activating NPS receptors, and inhibitors of ERK and PKA (reviewed in [7]). Here, we will discuss strategies to control amygdala activity by restoring cortical control as well as interventions targeting the amygdala that have emerged from recent studies.

Strategies Targeting Pain-Related Corticoamygdala
Dysfunction. ere is good evidence to suggest that mPFC deactivation in pain results in loss of amygdala control (see the Pain-Related Amygdala-Centered Corticolimbic Interactions section). A CRF1 receptor antagonist (NBI27914) inhibited the pain-related increase in synaptic excitation and background and evoked activity of BLA neurons in arthritic rats and increased the background and evoked activity of mPFC neurons that was decreased in the pain model [71].
is intervention also inhibited increased mechanosensitivity (spinal withdrawal re exes), averse a ective responses (audible and ultrasonic vocalizations), and anxietylike behaviors (measured in the elevated plus maze) and restored normal decision-making on a rodent gambling task in arthritic rats [71].
Another strategy to restore mPFC output used a group II mGluR antagonist (LY341495) to increase synaptically evoked spiking of mPFC pyramidal cells in brain slices from normal and arthritic rats [136]. E ects of a group II agonist (LY379268) showed that these receptors act on glutamatergic synapses from BLA to inhibit direct excitatory transmission and feedforward inhibition onto pyramidal cells, but their net e ect is decreased pyramidal cell output, possibly because the e ect on EPSCs preceded that on IPSCs. Facilitatory e ects of the antagonist suggest that the system may be tonically active to control pyramidal output.
Activation of mGluR5 was tested because of its location on mPFC pyramidal cells (see the Pain-Related Amygdala-Centered Corticolimbic Interactions section). A positive allosteric modulator (PAM) of mGluR5 (VU0360172) increased synaptically evoked spiking in mPFC pyramidal cells using electrical and optogenetic stimulation of BLA inputs [126,131]. is facilitatory e ect on mPFC output involved inhibition of synaptic inhibition by engaging endocannabinoid signaling because CB1 antagonists (AM251 and AM281) and an intracellular inhibitor of diacylglycerol lipase DAGL (tetrahydrolipstatin, THL) blocked the e ect of VU0360172 [126,131]. While this strategy worked under normal conditions, the facilitatory e ect of VU0360172 was lost in the arthritis pain model due to a breakdown of mGluR5-driven endocannabinoid signaling in the mPFC resulting in a lack of 2-arachidonoylglycerol (2-AG) [126].
e facilitatory e ect of mGluR5 activation on mPFC output was restored with inhibitors of the postsynaptic 2-AG hydrolyzing enzyme ABHD6 (intracellular WWL70) and the monoacylglycerol lipase MGL (JZL184) to increase availability of 2-AG in the postsynaptic cell or with a GABA A receptor blocker (intracellular picrotoxin) [126]. Coapplication of a CB1 receptor agonist (ACEA) with the mGluR5 PAM also increased synaptically evoked spiking of mPFC pyramidal cell neurons in brain slices from arthritic rats by decreasing abnormally enhanced feedforward inhibition from the BLA through depolarization-induced suppression of synaptic inhibition [126]. Systems electrophysiology studies in anesthetized rats with arthritis showed that coadministration of VU0360172 and ACEA into the mPFC increased background and evoked activity of pyramidal-like cells in the mPFC and inhibited the pain-related increase of background and evoked activity in amygdala (CeLC) neurons [123].
is combination strategy also inhibited increased mechanosensitivity (spinal withdrawal re exes) and audible and ultrasonic vocalizations and mitigated cognitive de cits in the reward-based decision-making in a rodent Pain Research and Management gambling task in the arthritis pain model [126]. e data further con rm the inverse link between mPFC and amygdala activity and that restoring mPFC output with a combination strategy of mGluR5-CB1 activation can engage cortical control of abnormally enhanced amygdala output to inhibit pain behaviors.
Neuropeptide S (NPS) binds to the G q /G s -coupled NPS receptor (NPSR), which is expressed in several brain regions including the dorsomedial ITC cell cluster in the amygdala, and produces anxiolytic e ects [84,[137][138][139][140][141]. NPS increased mPFC-driven feedforward inhibition of CeLC neurons by activating ITC cell drive and output in brain slices from arthritic rats through a PKA-dependent mechanism [39]. Intra-ITC as well as nasal application of NPS resulted in decreased background and evoked activity of CeLC neurons in anesthetized rats with arthritis pain, and this e ect was blocked by stereotaxic administration of an NPSR antagonist ([D-Cys( t Bu) 5 ]NPS or SHA68) into the ITC area [142]. Intra-ITC or nasal application of NPS also inhibited painrelated increases in audible and ultrasonic vocalizations as well as anxiety-like behaviors on the elevated plus maze but had no e ect on mechanosensitivity; the inhibitory e ects were blocked by stereotaxic administration of [D-Cys( t Bu) 5 ] NPS or SHA68 [39,142].
ese studies provide strong evidence for the concept that engaging mPFC control of amygdala processing may be a useful therapeutic strategy for pain management. is concept is supported by studies in humans that have implicated corticolimbic loops, including mPFC-amygdala interactions, in the prediction of and transitioning to chronic pain [11,143].

erapeutic Strategies Targeting Pain-Related Amygdala
Hyperactivity. Pharmacological interventions targeting glutamate receptors and neuropeptide systems in the amygdala have been reviewed recently [7]. Here, additional strategies involving the serotonergic system and potassium channels will be discussed.
Serotonergic descending pathways are involved in endogenous antinociceptive signaling from the brain to the spinal cord [52,144,145]. However, serotonin (5-HT) actions can be excitatory or inhibitory depending on the speci c receptor subtype and its associated neurotransmitter [52]. It is therefore not surprising that selective serotonin reuptake inhibitors (SSRIs) have shown inconsistent e cacy for neuropathic pain treatment [146][147][148]. One of the at least 14 5-HT receptors, the G q/11 -coupled 5-HT 2C receptor, has been implicated in adverse and inconsistent e ects of SSRIs for neuropathic pain [149,150] and, speci cally in the BLA, in the generation of anxiogenic behaviors [151][152][153].
In a rat model of neuropathic pain, viral vector-mediated 5-HT 2C receptor knockdown in the BLA inhibited mechanical hypersensitivity, aversive a ective pain behaviors (vocalizations), anxiety-like behaviors, and depression-like behaviors [44]. Pharmacological blockade of 5-HT 2C receptors (SB242084) in the BLA conveyed e cacy to a systemically applied SSRI ( uvoxamine) for inhibition of emotional responses (vocalizations) and anxiety-like pain behaviors but not mechanical hypersensitivity [154]. e bene cial behavioral e ects of 5-HT 2C receptor knockdown in the BLA involved inhibition of irregular and burst ring and evoked activity of CeLC neurons in neuropathic rats [44]. At the synaptic level, 5-HT 2C receptor knockdown in the BLA blocked the increase in excitatory transmission at the BLA-CeLC synapse in brain slices from neuropathic rats but had similar inhibitory e ects on feedforward inhibition under control conditions and in the neuropathic pain model. 5-HT 2C receptor is predominantly expressed in GABAergic neurons, but increased expression in non-GABAergic BLA cells was detected in the neuropathic pain state. e underlying mechanisms of this switch remain to be determined.
Another recent strategy to mitigate pain-related amygdala hyperactivity is activation of small-conductance calcium-activated potassium (SK) channels in the CeA. SK channels are calcium-sensitive, voltage-insensitive potassium channels that are expressed in somatic and dendritic regions of the neuron in a brain region-speci c manner [155][156][157][158]. Somatically expressed SK channels regulate neuronal excitability by mediating the medium afterhyperpolarization (mAHP) to decrease action potential ring rate [155]. In the amygdala, SK channels regulate action potential ring of neurons in the lateral CeA [159] but not LA [160]. SK channels also regulate dendritic excitability to modulate synaptic transmission and plasticity. In the amygdala, activation of synaptic SK channels in the LA acts as a postsynaptic shunt to reduce excitatory synaptic transmission [161], whereas removal of SK channels from the postsynaptic membrane of LA neurons by a PKAdependent mechanism facilitates excitatory transmission and synaptic plasticity [162].
A clinically available compound that can inhibit SK channels is riluzole, an FDA approved drug for the treatment of amyotrophic lateral sclerosis (ALS) that easily crosses the blood-brain barrier [163,164]. It should be noted that other actions of riluzole include inhibition of voltage-gated calcium channels, rapidly inactivating voltage-gated and persistent sodium channels, and glutamate receptor currents [165][166][167]. Systemically applied riluzole had antinociceptive e ects in the formalin test [168][169][170], in the carrageenan model of hindpaw inammation [171], and in neuropathic pain models [172][173][174][175]. Riluzole also produced pain relief in patients with irritable bowel syndrome [176]. e site and mechanism of painrelated riluzole e ects were not identi ed in these studies. Systemic application of riluzole inhibited emotional responses (audible and ultrasonic vocalizations), but not mechanosensitivity (spinal withdrawal re exes), in a rodent model of arthritic pain, and these inhibitory e ects were reversed by stereotaxic (intra-CeA) administration of a blocker of SK channels (apamin) but not of largeconductance calcium-activated potassium BK channels (charybdotoxin) [177].
An interesting observation is that not every intervention targeting the amygdala to inhibit emotional-a ective responses to pain a ects mechanosensitivity. is is true for 6 Pain Research and Management riluzole [177] as well as for NPS [142], an mGluR5 antagonist [178], and an SSRI [154], and may suggest di erential roles of neurochemically distinct intra-amygdala circuits.

Conclusions
e amygdala is a key node in the interaction of emotionala ective factors with sensory and cognitive aspects of pain. e synaptic and cellular analysis of amygdala function and plasticity as the neurobiological basis of certain pain behaviors has provided a model system for the study of brain mechanisms of pain. e better understanding of relevant intra-and extra-amygdalar circuits and their neurochemical and molecular signatures should yield novel targets for therapeutic interventions because amygdala activity is causally linked to pain behaviors, and therefore, controlling abnormally enhanced amygdala activity is a desirable goal for pain management.

Conflicts of Interest
ere are no con icts of interest.