Biological and Pharmaceutical Bulletin
Online ISSN : 1347-5215
Print ISSN : 0918-6158
ISSN-L : 0918-6158
Role of Connexins in Chronic Pain and Their Potential as Therapeutic Targets for Next-Generation Analgesics
Norimitsu MoriokaYoki NakamuraFang Fang ZhangKazue Hisaoka-NakashimaYoshihiro Nakata
ジャーナル フリー HTML

2019 年 42 巻 6 号 p. 857-866


Chronic pain, including inflammatory, neuropathic pain, is a serious clinical issue. There are increasing numbers of patients with chronic pain due to the growing number of elderly and it is estimated that about 25% of the global population will develop chronic pain. Chronic pain patients are refractory to medications used to treat acute pain such as opioids and non-steroidal anti-inflammatory drugs. Furthermore, the complexity and diversity of chronic pain mechanisms hinder the development of new analgesics. Thus, a better understanding of the mechanism of chronic pain is needed, which would facilitate the development of novel analgesics based on novel mechanisms. With this goal, connexins (Cxs) could be targeted for the development of new analgesics. Connexins are proteins with 20 subtypes, and function as channels, gap junctions between cells, and hemichannels that sample the extracellular space and release molecules such as neurotransmitters. Furthermore, Cxs could have functions independent of channel activity. Recent studies have shown that Cxs could be crucial in the induction and maintenance of chronic pain, and modulation of the activity or the expression of Cxs ameliorates nociceptive hypersensitivity in multiple chronic pain models. This review will cite novel findings on the role of of Cxs in the nociceptive transduction pathway under the chronic pain state and antinociceptive effects of various molecules modulating activity or expression of Cxs. Also, the potential of Cx modulation as a therapeutic strategy for intractable chronic pain will be discussed.


The perception of pain is a warning signal and serves to protect the organism from injurious or life-threatening harm. With intense thermal, mechanical and chemical stimulation of peripheral tissues, primary afferent nociceptive neurons (Aδ and C-type) convey nociceptive information to the spinal cord dorsal horn. The primary afferent nociceptive neuron is composed of the dorsal root ganglia, which contain the primary afferent neuron cell body, and surrounding cells, Schwann cells and satellite glial cells, which also have roles in the transduction of nociceptive information. Within the spinal dorsal horn, primary afferent neuron central terminals synapse with spinal dorsal horn neurons that project to several brain regions such as the thalamus and the brainstem. Neurons from these areas project to the somatosensory cortex and limbic areas, which lead to the localization of acute pain and perception of pain as unpleasant.1) In addition, spinal interneurons and terminals from brainstem neurons are found in the spinal dorsal horn, which can act either to dampen or increase pain signaling.

By contrast, chronic pain, including inflammatory pain and neuropathic pain, is not a state in which acute pain is simply prolonged. Chronic pain is a serious clinical issue, with an incidence of about 25% worldwide.2) For example, in the United States, more than 100 million people suffer from chronic pain, and more than $600 billion a year is spent on treatment.3) There are increasing numbers of patients with chronic pain, which is associated with the growing population of elderly people.4,5) Common characteristics of chronic pain include allodynia and hyperalgesia. Allodynia is the abnormal perception of a non-nociceptive stimulus as painful. Hyperalgesia is observed as a decreased threshold to nociceptive stimuli, resulting in exacerbated pain. Allodynia and hyperalgesia are usually refractory to treatment with commonly used analgesics such as opioid and non-steroidal anti-inflammatory drugs.

A number of studies have demonstrated neuroplastic changes within the pain pathway, from primary afferent sensory neurons to spinal cord horn neurons and supraspinal neurons, which contribute to the induction and maintenance of the chronic pain state.68) In addition to neurons, glial cells, including microglia and astrocytes, which are crucial in the formation and maintenance of neuronal networks in the central nervous system (CNS), also show signs of plasticity, in the form of altered functioning and distribution.8,9) Activation of glial cells induces excessive production and release of pronociceptive molecules, including cytokines, chemokines, eicosanoids, neurotrophins and reactive enzymes, which evoke nociceptive responding or decrease the threshold to respond to peripheral non-noxious and noxious stimuli.1013)

The interchange of information between cells, including between neurons and between neurons and glial cells, is crucial in normal physiological functioning of the peripheral and central nervous systems, and dysfunction of cellular communication could underlie chronic pain. One of main mechanisms involving signal transmission between neurons is neurotransmitter release from one neuron and binding to their respective receptors on another neuron. Another mechanism of intercellular communication involves gap junction channels. Gap junctions are transmembrane channels formed by direct connection of six proteins called connexins (Cxs), which are expressed on adjacent cells.14) The gap junction passes molecules with a molecular mass of up to 1 kDa, including ions, cyclic AMP, inositol 1, 4, 5 triphosphate (IP3), ATP and small peptides.1517) Recent studies have demonstrated associations between changes in expression or function of Cxs and the induction and maintenance of chronic pain in animal models (Table 1). Pharmacological and genetic approaches, using various animal models of chronic pain, further support the notion that Cxs have a crucial role in the induction and maintenance of chronic pain (Table 1). A number of Cx subtypes have been identified and their role in the chronic pain state will be examined in the current review.

Table 1. Studies about Involvement of Cxs in Chronic Pain Models
SubtypeExpression changeTissueModelSpeciesDuration of changeReagent usedReferences
Cx43Protein ↑Spinal cordCCIMouse (C57BL6)10 dPeptide5Tonkin RS et al., 2018
Spinal cordCCIMouse (ICR)10, 21 dCarbenoxolone, Gap26, Gap27Chen G et al., 2014
Spinal cordCCIRat (SD)3, 10 dn.e. (carbenoxolone)Wu XF et al., 2011
Spinal cordSNLRat (SD)22 dCORM-2 (carbon monooxide-releasing molecule)Wang H & Sun X, 2017
Spinal cordCIPN (oxaliplatin&pacitaxel)Mouse (C57BL6)13 dn.e. (peptide5)Tonkin RS et al., 2018
Spinal cordCIPN (bortezomin)Rat (SD)30 dCarbenoxoloneRobinson CR & Dougherty PM, 2015
Spinal cordCIPN (oxaliplatin)Rat (SD)7, 14 dCarbenoxoloneYoon SY et al., 2013
Spinal cordBCPMouse (C57BL6)7, 14, 21 dCarbenoxoloneYang H et al., 2018
Spinal cordCarrageenanRat (SD)5, 10 dCarbenoxoloneChoi HS et al., 2017
Spinal cordSpinal cord injuryMouse (ICR)7, 10, 14, 28 dCarbenoxoloneChoi SR et al., 2016
Spinal cordMorphine toleranceRat (SD)3, 5, 7 dGap26Shen N et al., 2014
Spinal cordFormalinRat (SD)90 minNot usedQin M et al., 2006
TG (SGC)Inferior alveolar nerve transectionRat (SD)8 dGap27Kaji K et al., 2016
TG (SGC)CCIRat (SD)10 dsiRNAOhara PT et al., 2008
DRGCCIRat (Wistar)12 dNot usedNeumann E et al., 2015
Sciatic nerveCCIRat (Wistar)12 dNot usedNeumann E et al., 2015
Protein ↓Spinal cordPSNL, Cx43 siRNAMouse (ddy)14 dNot usedMorioka N et al., 2018
Spinal cordPSNLMouse (ddy)14 dLycopeneZhang FF et al., 2016
Spinal cordPSNLMouse (ddy)7, 14, 21 dOverexpression by AV, TNF blocker (etanercept)Morioka N et al., 2015
Spinal cordSNLRat (SD)14 dNot usedXu Q et al., 2014
Phosphorylation ↑Spinal cordBCPRat (Wistar)6, 12, 18 dGap26Hang LH et al., 2016
Spinal cordCIPN (bortezomin)Rat (SD)30 dCarbenoxoloneRobinson CR & Dougherty PM, 2015
Spinal cordCCIRat (SD)3, 10, 20 dn.e. (carbenoxolone)Wu XF et al., 2011
Cx36Protein ↑ACCCCIRat (SD)7 dMefloquine, carbenoxolone, siRNAChen ZY et al., 2016
Protein ↓Spinal cordPSNLMouse (ddy)3, 7, 14, 21 dNot usedNakamura Y et al., 2015
mRNA ↓DRG (neuron and SGC)SNI (sural)Rat (SD)86 dNot usedPérez Armendariz EM et al., 2018
Protein n.e.Spinal cordCarrageenanRat (SD)3 h, 3, 5, 10 dNot usedChoi HS et al., 2017
Spinal cordSpinal cord injuryMouse (ICR)7 dNot usedChoi SR et al., 2016
Spinal cordSNLRat (SD)14 dNot usedXu Q et al., 2014
Spinal cordCIPN (oxaliplatin)Rat (SD)7, 14 dNot usedYoon SY et al., 2013
Cx32Protein ↑Spinal cordFormalinRat (SD)90 minNot usedQin M et al., 2006
Protein n.e.Spinal cordCarrageenanRat (SD)3 h, 3, 5, 10 dNot usedChoi HS et al., 2017
Spinal cordSpinal cord injuryMouse (ICR)7 dNot usedChoi SR et al., 2016
Spinal cordCIPN (oxaliplatin)Rat (SD)7, 14 dNot usedYoon SY et al., 2013
Cx37mRNA ↑Sciatic nerveSciatic nerve crushRat (SD)7, 14 dNot usedLin SH et al., 2002

“Duration of change” represents days when expression change in each Cx is observed. “Reagents used” represents the reagents that are used to modulate function or expression of Cxs in the literature. n.e.; no effect, ↑; upregulation, ↓; downregulation, CCI; chronic constriction injury, SNL; spinal nerve ligation, CIPN; chemotherapy-induced peripheral neuropathy, BCP; bone cancer pain, PSNL; partial sciatic nerve ligation, SNI; spared nerve injury, ACC; anterior cingulate cortex, DRG; dorsal root ganglia, TG; trigeminal ganglia, SGC; satellite glial cell.


Connexins have four transmembrane domains, two extracellular loops, one cytoplasmic loop and intracellular N- and C-terminal regions.18,19) Connexins oligomerize into a hexametric channel, termed a connexon, and two connexons link neighboring cells to form a gap junction.20) Homotypic and heterotypic docking of gap junction channels form cell-to-cell connections, and the opening of a gap junction is regulated by depolarization, extracellular alkalization, metabolic inhibition, mechanical stimulation or low concentration of extracellular Ca2+.21) In addition, Cxs act as hemichannels, between the cytoplasm and the extracellular region, forming a membrane permeable channel for small molecules.22) The human and mouse genome contains at least 21 and 20 connexin genes, respectively, which are developmentally regulated and expressed in a tissue-specific manner.23) Furthermore, at least 11 connexin genes have been identified specifically in the mammalian nervous system.24) Each Cx is named based on the molecular weight (in kDa) of the proteins predicted from their cDNA sequences.23)

The expression pattern of Cx subtypes is different between cell types. In the CNS, Cx36 and Cx32 are mainly expressed in neurons and oligodendorocytes, respectively.25,26) Cx26, Cx30 and Cx43 have all been identified in astrocytes, but a number of studies have demonstrated that Cx43 is crucial for astrocytic functioning.2729) Low levels of Cx43 are also observed in resting microglia, but treatment of primary cultured microglia with both tumor necrosis factor (TNF) and interferon-gamma (IFN-γ) increases cellular communication via formation of gap junctions and Cx43 expression.30) However, little is known of the other Cxs that could be present in microglia. The function and role of microglial Cxs in pain perception is still controversial.

Sensory neurons are surrounded by satellite cells, which is a type of glial cell. In the trigeminal ganglion, satellite glial cells express Cx4331,32) and satellite glial cells in dorsal root ganglion express both Cx43 and Cx36.33,34) Some Cx subtypes expressed in either neurons or glial cells, as either gap junctions or hemichannels, are likely to be involved in pain perception, as blocking these channels or preventing their expression through gene knockout prevents the development of hyperalgesia and allodynia in animal models of chronic pain. At the same time, changes in expression and function of Cxs have been observed in animal models of chronic pain.

2.1. Cx43 and Chronic Pain

Among Cxs, Cx43 appears to the most ubiquitous and it has been implicated in preclinical animal models as having a key role in the initiation and maintenance of chronic pain. As mentioned earlier, Cx43 is expressed in CNS astrocytes and satellite glial cells found in the peripheral nervous system (PNS).32,34,35) Spinal cord astrocytic Cx43 expression, in particular, has been shown to play a key role in chronic pain. Deletion of both Cx43 and Cx30 in mice prevented the emergence of heat hyperalgesia and mechanical allodynia following a spinal cord injury.36) However, deletion of Cx30 alone did not prevent the emergence of heat hyperalgesia and mechanical allodynia.36) While Cx43 appears to have a crucial role in the expression of chronic pain, the exact mechanism by which this occurs, including its expression pattern, has yet to be fully elaborated.

2.1.1. Upregulation of Cx43 and Chronic Pain

Increased spinal cord astrocytic Cx43 expression appears to underlie persistent nociceptive hypersensitivity as suggested in various animal models of chronic pain. For example, astrocytic Cx43 protein expression is increased in the ipsilateral spinal dorsal horn following a chronic constriction injury (CCI) of the sciatic nerve, a spinal nerve ligation (SNL), hind paw carrageenan-induced inflammation and unilateral hind limb bone cancer. Cx43 expression appears to be upregulated within the spinal cord grey matter following a spinal cord injury, oxaliplatin and bortezomib-induced peripheral neuropathy.3743) Nociceptive hypersensitivity observed in these models is ameliorated by intrathecal treatment with the gap junction blocker carbenoxolone (CBX).3943) Intrathecal treatment of CBX does not affect pain perception in healthy animals; thus, it appears that channel functioning in the chronic pain state appears to be different from that in the healthy state.

In addition to the ipsilateral spinal dorsal horn, a unilateral CCI increases Cx43 protein expression within the sciatic nerve and in the dorsal root ganglion.34) Ohara et al. have shown that a CCI of the infraorbital nerve increased Cx43 protein expression in the trigeminal ganglion, and reducing Cx43 expression with RNA interference prevented nociceptive hypersensitivity.32) These findings indicate that nerve injury increases Cx43 along the pain pathway, in both the CNS and the PNS.

A number of molecules induce expression of Cx43. For example, the pronociceptive cytokine TNF has been identified as upregulating Cx43 expression in mice with a sciatic nerve CCI.44) Blockade of sigma-1 receptors led to reduced Cx43 expression in lumber spinal dorsal horn astrocytes following a spinal cord injury, indicating that spinal cord injury evokes sigma-1 receptor activity, which then leads to increased spinal Cx43 expression.39) However, the ligand that activates sigma-1 receptors in response to spinal cord injury has yet to be identified. Intrathecal treatment with the N-methyl-D-aspartate (NMDA) receptor antagonist MK801 prevented increased Cx43 expression in the spinal dorsal horn in a mouse model of bone cancer pain, indicating that the excitatory neurotransmitter glutamate is involved in increasing Cx43 expression.43) The findings thus far suggest that injury-induced release of cytokines and neurotransmitters upregulates Cx43 expression.

Recent studies have shown that post-transcriptional regulation is involved in mRNA stability and translation of Cxs. Micro-RNAs (miRs), which destabilize mRNAs or interfere with mRNA translation, could be involved in the regulation of Cx43 expression.45) It has been shown that miR-1, which targets Cx43 mRNA, was downregulated following a sciatic nerve CCI, which was associated with an increase in Cx43 expression.34) However, the exact mechanism wherein miR-1 is downregulated in this model has yet to be identified. Other studies have suggested the importance of miRs in the induction of chronic pain.46,47) Thus, it is possible that miR-mediating regulation of Cx43 expression could be a key cellular mechanisms underlying nociceptive hypersensitivity.

There are a number of possible mechanisms that could link increased Cx43 to nociceptive hypersensitivity. Chen et al. demonstrated that sciatic nerve CCI induces the release of chemokine CXCL1 via newly formed Cx43 protein-based hemichannels and enhanced hemichannel activity in spinal astrocytes.44) CXCL1, in turn, promotes both peripheral and central sensitization through its receptor, CXCR2.48) Chronic treatment of morphine in rats leads to a morphine tolerance, which is behaviorally characterized by hyperalgesia, and increased spinal astrocytic Cx43 expression.49) Intrathecal injection of Cx43-gap junction specific blocker gap26 in morphine tolerant rats prevented not only spinal astrocytic Cx43 upregulation but also the expression and phosphorylation of NMDA receptors and the reduction of glutamate transporter GLT-1 expression, suggesting that morphine tolerance increases Cx43 and the enhancement of glutamatergic neurotransmission in the spinal dorsal horn, thereby leading to behavioral hyperalgesia.49) The findings further suggest a role of Cx43 in mediating morphine tolerance-induced hyperalgesia.

Another possible mechanistic link between Cx43 and nociceptive hypersensitivity is via nucleotide-binding oligomerization domain-like receptor protein 3 (NLRP3) inflammasome complex, a key mediator of neuroinflammation, and functions in the production of pronociceptive molecules interleukin (IL)-1β and IL-18.50) The NLRP3 inflamasome complex was activated following a CCI in mice.37) Although a direct link between increased astrocytic Cx43 and activation of NLRP3 inflammasome was not demonstrated, they speculated that nerve injury evoked increased Cx43 hemichannel activity, resulting in increased ATP release, which in turn, activated NLRP3 inflammasome.37) Other yet-to-be-identified substances that fit though the hemichannel could also activate other enzymes and lead to nociceptive hypersensitivity.

2.1.2. Downregulation of Cx43 and Chronic Pain

While an association between upregulation of Cx43 in chronic pain models and nociceptive hypersensitivity has been extensively documented, decreased Cx43 expression also appears to lead to nociceptive hypersensitivity. Knockdown of Cx43 expression with small interfering RNA (siRNA) in the trigeminal ganglion of naïve animals leads to facial mechanical hypersensitivity.32) In the spinal dorsal horn, Cx43 is localized in astrocytes, but not microglia and neurons.35) Downregulation of spinal dorsal horn astrocytic Cx43 has been reported in rats following a SNL51) and in mice following a partial sciatic nerve ligation (PSNL).35) Furthermore, restoring astrocytic Cx43 expression to normal or basal levels with intrathecal injection of a recombinant adenovirus vector ameliorated PSNL-induced mechanical hypersensitivity.35) Intrathecal injection of naïve mice with Cx43 siRNA induced mechanical hypersensitivity, which is line with findings that observed increased nociception with decreased spinal Cx43 expression following an injury.35) Interestingly, in line with findings from the CCI, spinal cord injury, oxaliplatin treatment pain models, over-expression of Cx43 by an adenovirus vector in naïve animals induced significant mechanical allodynia.35) Based on these findings, it appears that the absolute change of Cx43 expression in spinal astrocytes from basal levels is crucial for the induction of the chronic pain state and that the direction of change appears to be etiologically dependent. Therefore, the exact nature and consequence of the change in Cx43 expression, whether it is increased or decreased, will need to be evaluated in each pain model.

There are a number of mechanisms that reduce Cx43 expression in spinal astrocytes, which, in turn, leads to persistent nociceptive hypersensitivity. In several models of chronic pain, TNF has been shown to be increased in spinal dorsal horn, which is mainly produced by activated spinal microglia.52) Partial sciatic nerve ligation-induced Cx43 downregulation in spinal dorsal horn is prevented by intrathecal treatment with TNF inhibitor etanercept.35) In addition, intrathecal treatment of naïve animals with TNF decreases Cx43 expression.35) The findings suggest that injury-evoked expression of TNF could be crucial in the downregulation of Cx43 in spinal dorsal horn. Furthermore, incubation of cultured spinal astrocytes with proinflammatory cytokines such as TNF and IFN-γ reduces expression of Cx43 through the activation of ubiquitin-proteasome system, which is a key system for protein degradation.53,54) Previous studies have shown that TNF and IL-1β released from activated microglia attenuates Cx43 expression in brain astrocytes.55,56) Furthermore, proinflammatory stimuli lead to reduced Cx43 expression and gap junction communication in astrocytes.57,58) However, it is not yet known whether other cytokines, aside from TNF, or proinflammatory stimuli downregulates Cx43 expression in spinal astrocytes, which in turn contribute to the development and maintenance of nociceptive hypersensitivity in chronic pain models.

2.1.3. Phosphorylation of Cx43 and Chronic Pain

Phosphorylation of Cx43 is a crucial step in the regulation of function and assembly of gap junctions, and various protein kinases, such as protein kinase C and Src tyrosine kinase, may be involved in the phosphorylation of Cx43.59) Cx43 phosphorylation has been demonstrated in chronic pain models. After sciatic nerve CCI, increased phosphorylation at Ser368 of spinal astrocytic Cx43 was observed during the maintenance phase (10–20 d after CCI), and this response was induced by the downregulation of KATP channels.60) Wu et al. also suggested that increased Cx43 phosphorylation could be involved in interrupting gap junction function.60) Furthermore, increased phosphorylation at Ser368 of spinal astrocytic Cx43 has been reported in the bone cancer pain model. Although it was also reported that phosphorylated Cx43 is involved in mechanical allodynia through the production of chemokine CXCL12, the mechanism mediating phosphorylated Cx43-induced CXCL12 production was not elaborated.61) Phosphorylated Cx43 was increased in the spinal dorsal horn of bortezomib-treated rats, which show significant mechanical hypersensitivity.41) In this study, three different bands detected by Western blotting using a Cx43 antibody corresponded to non-phosphorylated (P0) and phosphorylated (P1, P2) Cx43, and the expression of phosphorylated (P1) Cx43 was significantly increased in the spinal dorsal horn of bortezomib-treated rats. A previous study indicated that the P1 phosphorylation state of Cx43 corresponded to increased trafficking of Cx43 from the cytoplasm to the membrane, which, in turn, led to increased Cx43-gap junction activity.62) While increased trafficking of Cx43 to the cell membrane may be involved in nociceptive hypersensitivity observed in the bortezomib-induced neuropathic pain model, membrane trafficking of astrocytic Cx43 and a link between phosphorylated Cx43 and nociceptive hypersensitivity have yet to be uncovered.

2.1.4. Channel-Independent Function of Cx43 and Chronic Pain

Increased nociceptive hypersensitivity associated with increased Cx43 expression is reversed with gap junction blocker treatment, suggesting that enhanced gap junction communication or hemichannel activity could underlie nociceptive hypersensitivity. However, increased nociceptive hypersensitivity observed with decreased Cx43 expression could be independent of decreased channel expression and channel functioning. In fact, previous studies have demonstrated that Cx43 likely has roles beyond that as a channel.6365) Cx43 has long cytosolic C-terminus which appears to mediate multiple domains-specific interactions with other proteins.66) Cx43 moderates cross-communication between regulatory proteins located in cell membranes and cytoskeletal proteins.65,67) Furthermore, knockdown of CNS astrocytic Cx43 expression induces changes in the expression of various genes.68) Downregulation of spinal astrocytic Cx43 in mice with RNA interference was accompanied by decreased GLT-1 expression, which in turn reduced synaptic clearance of glutamate and increased glutamatergic neurotransmission.35) Furthermore, expression of pronociceptive molecules IL-6 and cyclooxygenase-2 (COX-2) were also upregulated in spinal astrocytes following downregulation of Cx43 expression.63) Thus, suppression of Cx43 expression could lead to expression of genes that may enhance nociception. Spinal astrocytes treated with the gap junction inhibitor CBX downregulated Cx43 expression in addition to inhibiting gap junction activity. By contrast, treatment of spinal astrocytes with selective Cx43-gap junction blocker Gap27, which has no effect on Cx43 expression, did not change expression of astrocytic GLT-1, IL-6 and COX-2.63) Furthermore, phosphorylation at Ser9 of glycogen synthase kinase-3β (GSK-3β) was decreased, which indicates increased enzyme activity, following downregulation of Cx43, and this enzyme is crucial in expression change of GLT-1, IL-6 and COX-2 in spinal astrocytes.63) Akt1 is a kinase upstream of GSK-3β, and decreased phosphorylation of Akt1, which indicates decreased enzymatic activity, was observed following downregulation of astrocytic Cx43.63) Currently, the connection between decreased Cx43 expression and decreased Akt1 phosphorylation in spinal astrocytes is not yet clear. As mentioned above, Cx43 interacts with various intracellular signaling molecules. Therefore, it is possible that Cx43 could be involved in the regulation of other intracellular signaling molecule and expression of genes related to pain modulation.

2.2. Cx26 or Cx30 and Chronic Pain

Although it has been reported that astrocytes express Cx26, Cx30 and Cx43,29) there are regional differences in the expression of Cx26 and there is no clear consensus on whether they are involved in the formation of gap junctions in astrocytes.69) Currently, there are no reports that shown an involvement of Cx26 in chronic pain.

Mice that do not express Cx30 display mechanical allodynia and heat hyperalgesia following a spinal cord injury, similar to their wild-type litter mates.36) Furthermore, the expression of Cx30 is not changed in spinal dorsal horn following PSNL.35) Thus, it is likely that astrocytic Cx30 might be less important than Cx43 in mediating chronic pain.

2.3. Cx36 and Chronic Pain

Cx36 is the main neuronal connexin and has been detected in mature rat, mouse and human neurons.26,70) Cx36 is constitutively expressed in various brain regions, including spinal cord, cortex, hippocampus, and cerebellum.71) Cx36 has been shown to regulate inhibitory interneurons in the cerebellum, and it is possible that Cx36 could have a role in the modulation of neuronal activity.72) Dysfunction of Cx36 has been implicated in neurological disorders such as cerebral ischemia, traumatic brain injury, inflammation and epilepsy.70,7375)

Decreased expression of Cx36 in spinal dorsal horn following PSNL paralleled the time course of nociceptive hypersensitivity.76) Furthermore, as mentioned earlier, knockdown of Cx36 with intrathecal injection of Cx36 siRNA induced mechanical hypersensitivity through the enhancement of glutamatergic excitatory neurotransmission in spinal dorsal horn.76) In addition, Cx36 was shown to be expressed in inhibitory glycinergic, but not γ-aminobutyric acid (GABA)ergic, spinal interneurons.76) The findings suggest a link between Cx36 expressed on glycinergic spinal interneurons and enhancement of spinal glutamatergic excitatory neurotransmission, in that reducing Cx36 expression may lead to downregulation of glycinergic interneurons, which in turn, increases excitatory synaptic transmission and nociceptive hypersensitivity. Moreover, others have also demonstrated that mRNA expression of Cx36 is decreased in both primary afferent neurons and satellite glial cells in dorsal root ganglion following a sural nerve injury.33) Thus, it appears that both CNS and PNS Cx36 expression is involved in nociceptive hypersensitivity.

In contrast to decreased Cx36 expression in peripheral nerve and spinal cord in chronic pain models, Cx36 expression is increased in the anterior cingulate cortex (ACC) following CCI. The ACC is believed to be involved in the modulation of pain-related emotionality.77,78) At the cellular level, increased Cx36 expression could enhance synaptic neurotransmission.79) Surprisingly, the expression of Cx36 is not changed in spinal dorsal horn in several models of chronic pain, such as spinal cord injury, spinal nerve ligation, oxaliplatin-induced neuropathic pain and hind paw carrageenan treatment.39,40,42,51)

As there are few studies that have examined the role of Cx36 in chronic pain, the mechanism that regulates Cx36 expression has yet to be elucidated. As Cx36 is mainly expressed in neurons, a change in Cx36 expression would involve not only local neuronal functioning but also neuronal circuits and networks. Therefore, further elaboration of distinct roles and involvement of Cx36 in each region of the pain pathway will be needed.

2.4. Cx32 and Chronic Pain

In the CNS, although Cx32 is expressed in neurons and microglia, oligodendrocytes abundantly express Cx32.80) While it is known that oligodendrocytes are involved in induction of chronic pain,81) the role of oligodendrocytic Cx32 in mediating nociceptive hypersensitivity has yet to be elucidated. In fact, previous studies have shown that protein expression of Cx32 is not changed in spinal dorsal horn following spinal cord injury, oxaliplatin and carrageenan treatment.39,40,42) In other models, there appears to be a role of neural Cx32. Qin et al. showed increased numbers of heterotypic gap junction channels between astrocytic Cx43 and neuronal Cx32, and this response paralleled the induction of heat hyperalgesia following subcutaneous injection of formalin into the rat hindpaw.82) It appears that Cx32 is involved in the induction of certain pain states, but more studies are needed to establish a role of Cx32 in nociceptive transduction.

2.5. Cx37 and Chronic Pain

Cx37 is expressed in the developing rat brain cortex and in spinal motor neurons.83,84) Currently, few studies have demonstrated an involvement of Cx37 in nociceptive hypersensitivity. Cx37 mRNA expression is increased in the sciatic nerve, in parallel with the expression of heat hyperalgesia, but not in the spinal dorsal horn, following a nerve crush.85) The origin of Cx37 mRNA in this study is not entirely clear, as it could have originated from Schwann cells or transported down the axon from dorsal root ganglion neurons–expression levels in dorsal root ganglion neurons were not measured. A role of newly expressed peripheral nerve Cx37 in nociceptive hypersensitivity remains to be elucidated.

2.6. Cx45 and Chronic Pain

Cx45 is highly expressed in brain neurons during embryogenesis and the neonatal period, and the level of expression declines thereafter.84) In adult rats, it is found in abundance only in the thalamus and hippocampal CA3 region.86) By contrast, Cx45 has been identified in superficial spinal dorsal horn excitatory interneurons of mice.87) Furthermore, Cx45 expressing neurons in mice are in close proximity to serotonin (5-HT)-containing terminals from monoaminergic descending neurons originating from the brainstem.87) However, the role of Cx45 in nociceptive transduction is unknown, and further investigation is necessary to determine whether there are changes in function or expression in chronic pain models.


The current review shows that modulation of Cx43 expression or activity could be a significant therapeutic strategy for treating chronic pain. Blocking Cx43-gap junction or hemichannels, in particular, could be crucial in reducing key mechanisms underlying chronic pain, as upregulation of Cx43 along the pain pathway, including the spinal cord, satellite glial cells in the trigeminal ganglion and sciatic nerve, has been well-documented. The non-selective gap junction blocker CBX has demonstrated efficacy in various rodent chronic pain models.3944) In addition, the selective Cx43 blockers, Gap26 and Gap27, which are Cx43 mimetic peptides, also demonstrated antinociceptive efficacy in a CCI and bone cancer pain model.44,61) Intrathecal injection of Peptide5, which is a short peptide mimetic of the extracellular EL1 loop of the Cx43 channel, ameliorated mechanical hypersensitivity in mice with a CCI.37) Suppression of Cx43 expression with RNA interference, in addition to blocking Cx functioning, could be a powerful approach to reducing chronic pain.32)

However, previous studies have also shown that downregulation of spinal Cx43 expression leads to nociceptive hypersensitivity, so upregulation, to basal levels, would be called for in some pain states.35,51,63,88) In these cases, there are various compounds that increase Cx43 expression. For example, lycopene, a non-provitaminic carotenoid and found in abundance in tomatoes, upregulated Cx43 expression in various cell lines.89,90) Upregulation of spinal Cx43 expression by lycopene in PSNL mice was accompanied by amelioration of nociceptive hypersensitivity.88) Furthermore, overexpression of Cx43 via an adenovirus vector also ameliorated mechanical hypersensitivity in PSNL mice.35) Antidepressants have been shown to be antinociceptive in certain chronic pain states.91) Treatment of cultured cortical astrocytes with antidepressants, including amitriptyline, clomipramine and fluvoxamine, upregulated Cx43 protein expression.92) Furthermore, chronic systemic administration of antidepressants, of either fluoxetine or duloxetine, increased Cx43 protein expression in the rat prefrontal cortex.93) Although these studies showed upregulation of Cx43 in context with anti-depressive treatment, these findings could extend to understanding of the mechanism of the antinociceptive effect of antidepressants.

Treatment with CBX suppressed nociceptive hypersensitivity through a Cx36-dependent manner.79) The anti-malarial agent quinine is known to specifically block Cx36-gap junction,94) yet quinine has not been preclinically tested in a chronic pain model. Like Cx43, however, opposite directions of Cx36 expression has been observed in various chronic pain models,76,79) it is possible that a Cx36 blocker may be useful for treating some types of chronic pains whereas other chronic pains may benefit from increased Cx36 expression or functioning.

As mentioned previously, the development of drugs targeting Cx-channels is difficult because the direction of the expression change appears to be dependent on etiology of the pathology. Therefore, compounds may need special characteristics, such as partial agonism, that is, acting as an antagonist when Cx is activated or increased, and act as an agonist when Cx is suppressed or decreased. Thus, most important would be to reverse the expression or activity of Cx from the abnormal state back to the normal or basal condition rather than completely suppress expression or increase expression well beyond what is normal. However, the physiological level of each Cx is not entirely clear and whether drugs that have multipurpose functioning can be developed is not yet known.

Connexin subtypes are present throughout the body. In addition to the CNS and PNS, Cx43 is abundantly expressed in cardiac muscle, and dysfunctional Cx43 contributes to arrhythmia.95) Thus, it is possible that Cx43-targeting compounds (blockers or inducers) could induce adverse effects, when they are systemically administered. In addition, as Cx36 is generally expressed in various types of neurons,71) and involved in a wide range of neurological disorders,70,7375) Cx36-targeting compounds could affect normal brain functioning such as memory, cognition and motor function. Therefore, in addition to special pharmacological properties, it may be important to deliver drugs to a specific site of the pain pathway.


Studies in rodent chronic pain model have demonstrated that various types of Cx could have key roles in the mechanism of chronic pain, and because of this, designing drugs that target Cxs could lead to useful therapeutics for pain management.

In addition to spinal dorsal horn, Cx43 expression appears to be downregulated in the prefrontal cortex and hippocampus contralateral to a PSNL in mice (Morioka et al., unpublished data). The prefrontal cortex and hippocampus are crucial brain regions involved in the affective-motivational dimension of chronic pain.78,96,97) Significant changes to the affective as well as the sensory-discriminatory component of pain are observed in chronic pain.98,99) Downregulation of Cx43 in the prefrontal cortex was suggested to underlie disturbances in cell communication, which, in turn, underlie with depressive-like behavior in rats.93) Similarly, reduced expression of Cx43 mRNA has been observed in post-mortem brains from patients with major depressive disorders.100) Thus, treatment targeting Cxs could lead to amelioration of both pain and the depression that accompanies it.

At present, the exact role and involvement of Cx-gap junctions in patients with chronic pain is unknown. Perhaps in vivo neuroimaging, such as positron emission tomogram imaging, could lead to identifying the distribution and function of Cxs in chronic pain patients. While almost all preclinical chronic pain studies on Cx have utilized models developed in rodents and that Cxs are highly conserved across species, the limitation of rodents is that they are a species that is phylogenetically distant from human. Perhaps examining Cxs in other nonhuman animal species, such as nonhuman primates, may help to further clarify the actual role of this target in chronic pain.97,101)

In total, understanding the distinct roles of Cx-gap junctions in regulating nociceptive hypersensitivity could aid in the development of a novel approach to overcoming chronic pain.


We also thank Dr. Aldric T. Hama for his careful editing of the manuscript.

Conflict of Interest

The authors declare no conflict of interest.

© 2019 The Pharmaceutical Society of Japan