Long-term cortical synaptic changes contribute to chronic pain and emotional disorders

Increasing evidence consistently indicates that cortical mechanisms play important roles in chronic pain and its emotional disorders. Central synapses, especially excitatory synapses, are undergoing long-term memory-like plastic changes after peripheral injury. These changes not only occur at the single synaptic level, but also take place at cortical and subcortical circuits. Consequently, neuronal responses to peripheral sensory stimuli, or even to sensory inputs triggered by normal physiological signals such as touch and movement, are significantly potentiated or increased. Such prolonged cortical excitation likely contributes to chronic pain and its related emotional changes. In this short review article, I will summarize recent progress using animal models and explore possible different mechanisms that may contribute to chronic pain in the brain.


Introduction
Since the discovery of cortical plasticity in adults, much progress has been towards the understanding of chronic pain mechanisms in the brain. Both brain imaging and modern neuroscience studies have elicited certain brain regions in pain perception, unpleasantness, and fearful memory. Interrelationships between pain and emotional responses such as anxiety, fear and depression have been reported and investigated. The brain not only interprets sensory inputs from the lower spinal cord level to affective responses or pain perception, but also enhances, amplifies or exaggerates such sensory experience. Perhaps the useful pain protection and memory functions also causes the suffering of patients. This would explain why many previous drugs developed for blocking pain transmission from periphery to brain fail to effectively control or inhibit chronic pain. The abuse of analgesics, such as opioids, in recent years clearly shows the need for novel pain medicine. It is insufficient to block pain transmission alone to reduce pain, it is equally or more important to inhibit or reverse pain plasticity in order to treat chronic pain.

Multiple cortical areas involved in pain conditions
Animal electrophysiological studies and human brain imaging studies have revealed several key cortical regions involved in pain including acute and chronic pain [1][2][3][4]. They include the anterior cingulate cortex (ACC), insular cortex (IC), primary somatosensory cortex (S1), secondary somatosensory cortex (S2) and prefrontal cortex (PFC). In particular for the ACC and IC, hyperexcitation or hyperactivation has been observed in different chronic pain conditions. This observation in humans has also been confirmed by animal studies that neurons in these cortical areas respond to peripheral noxious stimuli or injury. For example, in adult rats, peripheral amputation caused long-term potentiation (LTP) in the ACC in vivo [5]. There are increasing studies from both animal and human studies that demonstrate the importance of ACC/IC as well as PFC and S1/S2 in chronic pain and related emotional disorders in different models of chronic pain [6][7][8][9][10][11][12][13].

Activation of activity-dependent immediate early genes after injuries
One strong piece of evidence supporting the potential plasticity changes in cortical regions comes from immunostaining studies of activity-dependent early genes. These immediate early genes such as c-fos and Egr1 provide a useful tool to map the cortical areas that contribute to the induction and expression of chronic pain conditions in animals. Different peripheral injuries, such as inflammation, nerve injury, https://doi.org/10.1016/j.neulet.2018.11.048 chronic visceral pain, and formalin activates a set of immediate early genes in the ACC, IC and other related areas [14]. Calcium-stimulated adenylyl cyclase subtype 1 (AC1) and subtype 8 (AC8) contribute to the activation of injury induced immediate early genes [14]. Furthermore, Wei and Zhuo [15] reported that the extracellular signal-regulated kinase (Erk) activity is activated in the ACC by mechanical allodynia at two weeks after the injury. These observations strongly indicate that ACC and IC neurons are activated after peripheral injuries, and longterm plastic changes can further sensitize neurons and synapses in the same areas even after many days or weeks after the injury. This observation also raises another possibility that cortical synapses may undergo secondary plastic changes when animals experience allodynia after injury.

Postsynaptic phosphorylation and membrane receptor translocation
Electrophysiological recordings from individual cortical pyramidal neurons offer direct evidence of the involvement of the ACC and IC in chronic pain conditions. AMPA receptor-dependent EPSCs, recorded from pyramidal neurons of layers II/III and V are found to be significantly enhanced after injury in different animal models of pain, including chronic inflammation and neuropathic pain [6,8]. Biochemical and pharmacological studies found that these changes are associated with an alteration in the rectification of AMPA receptor-mediated synaptic transmission and the development of sensitivity to an inhibitor of CP-AMPA receptors [16,17]. Genetic studies using genetic deletion of different subtypes of AMPA receptors show that deletion of GluA1, but not GluA2, significantly reduces peripheral injury-triggered c-fos activation in the ACC [6]. Electron microscopy studies demonstrate that nerve injury increases the postsynaptic expression of AMPA receptor GluA1 in pyramidal neurons in the layer V of the ACC, including the corticospinal projecting neurons [17]. Biochemical studies further confirm the involvement of AMPA receptor GluA1 by showing the increase of membrane-bound GluA1-containing AMPA receptors in the ACC after peripheral nerve injury [16]. Enhancement of AMPA receptor mediated responses in the ACC has also been reported in animal models of visceral pain [18,19].
In the IC, Qiu et al. [20] found that peripheral nerve injury triggered the enhancement of AMPA receptor-mediated excitatory synaptic transmission. The synaptic GluA1 subunit of AMPA receptor, but not the GluA2/3 subunit, was increased after nerve ligation. Genetic knockin mice lacking phosphorylation of the Ser845 site, but not that of the Ser831 site, blocked the enhancement of the synaptic GluA1 subunit, indicating that GluA1 phosphorylation at the Ser845 site by protein kinase A (PKA) was critical for this upregulation after nerve injury. Furthermore, A-kinase anchoring protein 79/150 (AKAP79/150) and PKA were translocated to the synapses after nerve injury.
NMDA receptors that are important for triggering LTP in the ACC, also undergo upregulation after peripheral injuries. Peripheral nerve injury leads to increased GluN2B-containing NMDA receptor -mediated responses. The expression of GluN2B subunits in the ACC is upregulated after peripheral inflammation [21,22]. Similar upregulation of NMDA GluN2B has been reported in the IC [23]. Qiu et al. [23] reported that there is a long-term increase in the amount of synaptic NMDA receptors, but not that of extrasynaptic NMDA receptors. Furthermore, the increase in the amount of NMDA receptors required phosphorylation of the NMDA receptor subunit GluN2B at Tyr (1472) by a pathway involving adenylyl cyclase subtype 1 (AC1), PKA and Src family kinases. Microinjection of NMDA receptor or GluN2B-specific antagonists into the IC reduced behavioral responses to normally non-noxious stimuli in the mouse model of neuropathic pain, indicating that enhanced NMDAR function in the IC contributes to chronic pain.

Protein synthesis and protein degradation
It has been reported that cAMP-dependent signaling pathways play important roles in protein dependent memory, including activation of CREB signaling pathways. Although direct investigation of new protein synthesis in the ACC or IC have not been reported, there is evidence that indicates that protein synthesis is likely taking place in pain-related areas. Late-LTP in the ACC is inhibited or reduced by anisomycin, a protein synthesis inhibitor [8]. Recently, Ko et al. [24] reported that activation of the ACC induced by peripheral nerve injury increases the turnover of specific synaptic proteins in a persistent manner. Neural cell adhesion molecule 1 (NCAM1) is one of the molecules involved. NCAM1 contributes to spine reorganization and contributes to the behavioral sensitization. Thus, it is likely that neuronal activity is capable of affecting both the rate of new protein synthesis as well as protein degradation; such biphasic modulation offers great flexibility for neuronal activities to alter brain structure in a rapid and fine manner. Future studies are needed to map long-term changes in signaling proteins in these areas.

Presynaptic release of glutamate
In addition to long-term postsynaptic changes in receptors, and signaling proteins, the release of glutamate is found to be enhanced in animal models of chronic pain [6]. In both animal models of neuropathic pain and chronic inflammatory pain, the ratio of paired-pulse facilitation, an indicator of presynaptic enhancement of excitatory synaptic transmission, is reduced [6,16,25]. Spontaneous release of glutamate, as detected by measuring the frequency of AMPA receptormediated miniature EPSCs (mEPSCs), are also increased. Unlike postsynaptic changes, more studies are needed to reveal presynaptic basic mechanisms for such enhancement. Koga et al. [26][27][28] reported a presynaptic form of LTP (pre-LTP) in the ACC, providing a useful in vitro model for examining presynaptic mechanism for the potentiation. By using genetic approaches, Koga et al. [26] found that ACC pre-LTP was abolished in Fmr1 KO mice. Furthermore, SCRAPPER, an E3 ubiquitin ligase expressed in presynaptic terminals, is also found to be required for pre-LTP in the ACC [28]. In addition, SCRAPPER also affects the spontaneous release of glutamate in the ACC. Similar pre-LTP has been reported in the IC, and AC1-cAMP-signaling pathway is required for IC pre-LTP [29].

Calcium-stimulated AC1 is critical for both presynaptic and postsynaptic changes
Experiments using integrative approaches demonstrate that AC1 activity is critical for both postsynaptic and presynaptic changes in the ACC and/or IC [7,8,29]. AC1 is a key signaling enzyme for the production of cAMP in neurons [7,30,31]. Unlike in hippocampal neurons where other protein kinases such as CaMKII play important roles in LTP, ACC and IC neurons require AC1 activity to undergo both pre-LTP and post-LTP [6,29]. In AC1 knockout mice, activation of several immediate early genes by injuries are significantly reduced or abolished [14]. Potentiation of AMPA receptor mediated responses and postsynaptic trafficking of GluA1 receptors are also blocked in AC1 knockout mice [16,20]. These results indicate that AC1 is critical for injury-triggered postsynaptic changes in the ACC and IC. Similar findings are found with presynaptic changes. The changes of PPF and spontaneous release of glutamate are blocked in AC1 knockout mice [16]. In parallel, behavioral allodynic and hyperalgesic responses are reduced or inhibited in different animal models of chronic pain in AC1 knockout mice [31]. The role of AC1 is further confirmed by the use of AC1 inhibitor NB001 [31]. In the IC, genetic deletion of AC1 prevented the translocation of AKAP79/150 and PKA, as well as the upregulation of synaptic GluA1-containing AMPA receptors [20].

Neuromodulation
Recent studies have demonstrated that nerve injury also caused enhanced neuronal excitability in the dendrites of ACC deep layer pyramidal neurons by using dendritic whole-cell patch-clamp recording method [12]. The dysfunction of HCN channels are involved. Furthermore, activation of the serotonin receptor subtype 7 (5-HT 7R ) alleviated the injury. At the behavioral level, application of LP-211, a new 5-HT 7R agonist, increased the mechanical withdrawal threshold in neuropathic animals [32], These results indicate serotonin mediated modulation in the ACC neurons play important roles in enhanced neuronal excitability of dendrites during neuropathic pain. In addition to directly contribute to pain process in the ACC, it has been reported that activation of ACC deep pyramidal cells can facilitate spinal nociceptive excitatory transmission [33]. Future studies are clearly needed to investigate molecular mechanisms for such long-term changes, and possible link to spinal sensitization in chronic pain conditions.

Disinhibition
In addition to enhanced excitatory transmission and neuronal excitability, it has also been reported that reduced ACC GABA level [34] or the loss of connection between excitatory and inhibitory neurons in ACC deep layer [11] in animal models of chronic pain. Using multiple whole-cell recordings from neurons in layer 5 of the ACC of adult mice, Blom et al. [11] observed a striking loss of connections between excitatory and inhibitory neurons in both directions after nerve injury. Such loss in connection may also contribute to cortical excitation observed in ACC neurons during chronic pain.

Glial mechanism
Glial related mechanisms have been nicely investigated at the spinal cord dorsal horn [35]. Activated microglia may alter synaptic transmission and plasticity by releasing microglia-derived signaling molecules such as BDNF. In the cortical regions such as ACC and IC, the possible role of astrocytes and microglia activation in animal models of chronic pain remain to be investigated [36,37]. In the ACC and IC, few microglia are activated after peripheral nerve injury [38]. By contrast, it has been reported by others that spinal cord microglia cells are activated in the same animals. Furthermore, in the ACC, both post-LTP and pre-LTP are insensitive to minocycline, an inhibitor of microglial activation, suggesting that the activation of microglia is not involved in long-term synaptic plasticity [39]. A transitory microglial reaction has been reported in the infralimbic cortex of rats but not mice at 7 days after nerve injury. No changes have been detected in other regions of the cortex such as ACC and IC [37]. For astrocytes, there are reports of activation of astrocytes in the ACC after peripheral inflammation [35]. Future studies are clearly needed to examine the time-dependent contribution of astrocyte activation to cortical neuronal excitation.

Effects of inhibiting plastic changes in behavioral models of pain
Direct evidence for the contribution of ACC enhanced synaptic transmission comes from pharmacological and behavioral studies. Microinjections of different inhibitors or antagonists into the ACC that prevented or reduced ACC LTPs produced analgesic and/or anxiolytic effects in animal models of chronic pain. For inhibiting excitatory Fig. 1. Different basic mechanisms that may contribute to cortical excitation after injury. After injuries, including peripheral or central insults, neuronal synapses and circuits within the ACC, IC as well as other related cortical and subcortical regions are likely undergoing long-term plastic changes. Consequently, central responses to subsequent sensory stimuli as well as possible spontaneous pain are dramatically enhanced. Based on recent studies and reports, there are at least five different mechanisms for such changes: (a) Postsynaptic potentiation of excitatory synapses (Post-LTP): In rodent models of chronic pain, ACC synapses undergo long-term presynaptic and postsynaptic changes.; at postsynaptic sites, the expression of AMPA receptors is increased. These presynaptic and postsynaptic changes that occur in chronic pain models share some common mechanisms with presynaptic LTP and postsynaptic LTP that can be induced by experimental protocols in vitro. In addition to postsynaptic increases in AMPA receptor expression, postsynaptic NMDA receptors, particularly those containing the GluN2B subunit, are also increased in models of chronic pain. AC1 is essential for both the presynaptic enhancement of glutamate release, and the postsynaptic potentiation of AMPARs and NMDARs; (b) Presynaptic potentiation of excitatory synapses (Pre-LTP): At presynaptic sites, glutamate release is increased; (c) Recruitment of 'silent' excitatory synapses: Silent synapses have been reported in different regions of the brain. Due to limitation of recording methods, few studies have been reported in mature cortical synapses. However, it remains one potential possibility that some of the cortical synapses may be 'silent'. This silence can be due to presynaptic ineffective release of glutamate upon stimulation, or absence of a postsynaptic receptor that responds to the release of glutamate. Recent multiple channel electrode studies in both ACC and IC have revealed increases of active channels after LTP induction [8]. More importantly, such phenomena disappear after injury; providing a very likely mechanism for cortical spread in disease state. It may help to explain phantom pain or phantom sensation as well; (d) Dis-inhibition or loss of inhibition of inhibitory transmission: Although this possibility has not been directly investigated, it is possible that inhibitory transmission may be affected after injury. Such inhibitory mechanisms may affect excitatory synapses by acting on presynaptic terminals and/or postsynaptic sites. Loss of such inhibition may indirectly enhance the release of glutamate or potentiate postsynaptic excitatory responses of excitatory pyramidal cells; (e) Glial mechanism: Although no obvious activation of microglia has been reported in the ACC/IC, it is possible that some microglia or astrocytes derived signaling molecules may be released after injuries. These molecules, such as BDNF, may then affect excitatory synaptic transmission and plasticity in cortical synapses. In addition, they may affect the uptake of glutamate, and its action in synapses.
AMPA mediated transmission, microinjection of CNQX into the ACC affects behavioral allodynia in animal models of neuropathic pain [6]. Since electrophysiological recordings show that potentiation of postsynaptic responses was reversed by Ca 2+ permeable AMPA receptor antagonist NASPM in the ACC, behavioral studies show that microinjection of NASPM into the ACC inhibited behavioral sensitization caused by nerve injury [17]. These observations strongly indicate that inhibiting potentiated AMPA receptor mediated responses in the ACC produces analgesic effects. Furthermore, injection of the PKMζ inhibitor ZIP, that reduced the expression of postsynaptic LTP in the ACC, also produced analgesic effects [40]. For NMDA receptors, microinjection of AP5, or inhibiting NR2B receptors in the ACC produced analgesic effects in animal models of chronic pain [6,41]. HCN channels are important for the maintenance of pre-LTP in the ACC [26]. Injection of the HCN channel inhibitor ZD7288 produced inhibitory effects on injury induced anxiety [26]. Similarly, injecting NMDA receptor or GluN2Bspecific antagonists into the IC reduced behavioral responses to normally non-noxious stimuli in the mouse models of neuropathic pain. Together, these findings consistently indicate that enhanced excitatory synaptic responses in the ACC/IC contribute to behavioral sensory responses as well as anxiety-like behaviors in animal models of chronic pain.

Conclusion and future directions
In summary, there are many different mechanisms that may contribute to cortical potentiation that underlies cortical mechanisms for chronic pain (see Fig. 1). Both neuronal and glial mechanisms may contribute to increased synaptic responses in animal model of chronic pain. For neuronal mechanisms, pre-LTP and post-LTP can contribute to enhanced synaptic responses. In addition, increased postsynaptic expression of NMDA receptors also contributes to such enhancement. Postsynaptic trafficking of AMPA receptors may be attributed to the recruitment of silent responses in the ACC and IC. Potential changes in cortical inhibitory transmission may also play important roles in chronic pain and anxiety. Future studies of plasticity of inhibitory transmission and how inhibitory transmission interacts with excitatory transmission and plasticity are clearly needed. For glial mechanisms, although evidence is lacking, signaling molecules derived from glial cells may contribute to enhanced glutamate responses indirectly.
Recent studies indicate that cortical changes may be time-dependent [42]. At early time points after injuries, rapid synaptic changes may lay the foundation of triggering prolonged synaptic changes. Furthermore, it is possible that some of these changes may cause additional homo-and hetero-synaptic changes in cortical circuits. For example, synaptic tagging that has been reported in the hippocampus may also play important roles in pain-related cortical synapses. Unlike classic learning events, prolonged changes after injuries may cause ongoing synaptic plasticity through the connections in the central nervous system. Enhanced synaptic potentiation allows normal physiological inputs from the body system to induce pathological changes among adjacent or wired systems. Some 'silent' connections or synapses may become active in such conditions. Different forms of synaptic plasticity, including LTPs, long-term depression (LTD) and synaptic tagging will provide excellent synaptic models to examine basic mechanisms for cortical changes in chronic pain. In addition, to confirm or investigate different pathways into the cortex using optogenetic approaches, molecular mechanisms for cortical plasticity are greatly needed for future development of new medicines to treat patients with chronic pain and/ or emotional disorders.