Introduction

Major depression disorder (MDD), one of the most severe mental disorders troubling over 350 million people worldwide, is also the main reason for disability [1]. Besides, MDD imposes a considerable economic cost and brings a significant burden to society [2]. The current consensus view is that cellular and molecular abnormalities caused by genetic and environmental interactions contribute a lot to depression [3]. It is traditionally believed that depression is related to the monoaminergic neurotransmitter system in the brain [3]. However, the pathological phenomena of depression are complicated, and the monoaminergic hypothesis has many limitations, which cannot fully explain these phenomena. Thorough cognition of depression pathophysiology and pathogenesis is still lacking. Thus, it is essential to consider more pharmacological mechanisms of MDD to explore new therapeutic targets. The GJC is currently a hot research content, and it may also be a potential target for antidepressant therapy.

Gap junction dysfunction in the prefrontal cortex (PFC) can induce depressive-like behaviors [4]. GJCs are formed by connexins [5]. Connexin is synthesized by the endoplasmic reticulum (ER) and transported to the membrane surface [6]. The six penetrating gap junction proteins constitute homomeric or heteromeric connexons [7]. A pair of connexons on the corresponding surfaces of adjacent cell membranes form a GJC which allows small molecules whose molecular weight is <1 kDa and diameter is <1.5 nm. These small molecules include ions, metabolic molecules, and second messengers to pass through mediating information exchange between cells [8]. Uncoupled connexon acts as a hemichannel (HC) to promote the chemical connection between the intracellular and extracellular spaces [7]. GJCs act as a critical role in regulating nerve cell growth, differentiation, and physiological functions by participating in the metabolic coupling of material exchange between cells, the electrical coupling of electrical signal transmission, and the transmission of information between cells [9]. Under physiological conditions, HCs on the cell membrane keep closed. Only under special circumstances can the HCs be activated, including stress and acute injury, causing some molecules to enter and exit the cell through the channel [10]. A short-term opening is beneficial to increase the adaptability of cells, but long-term activation may cause damage to the cell, which further induces many diseases, e.g., depression [11]. There are two families of gap junction proteins in mammals: connexins (21 members in humans) [12] and pannexins (three members) [13]. However, several pieces of evidence indicates that HCs formed by pannexins cannot be assembled into GJCs [14]. Thus, this article only discusses connexin. Connexin is widely expressed in all tissues except differentiated skeletal muscle, circulating erythrocytes, and mature sperm cells [15]. More than half of the connexin is expressed in the nervous system and 1/3 in the central nervous system (CNS), mainly in the glial. Cx43 is the main subtype of connexin of astrocytes [7, 16], and other subtypes including Cx30 [17], Cx26, Cx45, Cx40, and Cx46 [18] are also slightly expressed. Astrocyte dysfunction is an important pathological feature and pathogenesis of depression, of which Cx43 dysfunction plays an important role [4]. Inactivation of the Cx43 gene in astrocytes increased the acute antidepressant effect of fluoxetine [19]. Notably, Cx43 is widely distributed [20]. The abnormal function of Cx43 in other tissues and cells such as liver tissue [21] and microglia [22] also contributes directly or indirectly to the induction of depression-like behaviors, which will be elaborated in this article. In short words, the relationship between Cx43 dysfunction and depression-like behaviors is far more complicated than expected. Therefore, in the present review, the authors summarize the latest views on the role of Cx43 in depression-like behaviors.

Cx43 abnormalities and dysfunction in depression

At the earliest, postmortem studies found that Cx43 was downregulated in the locus coeruleus [23], frontal cortex [24], mediodorsal thalamic nucleus [25], and caudate nucleus [26] in patients with MDD compared to healthy individuals. It suggests that Cx43 expression in the above areas may act a vital role in the pathophysiology of depression. Later, studies on Cx43 expression and function changes were carried out (Table 1). Cx43 gene levels are reduced in the orbitofrontal cortex [26] and neocortex [25] in patients with depression. However, there was no difference in the DNA methylation of the Cx43 gene in the PFC between patients with clinically well-defined depressed patients and healthy people. It suggests that the contribution of Cx43 to depression may be derived from downstream protein levels and functions. This section will discuss Cx43 abnormalities of several related brain areas in the depression model in vivo and in vitro.

Table 1 Abnormalities of Cx43 in experimental models.
Full size table

Prefrontal cortex (PFC)

Rats subjected to chronic unpredictable stress (CUS) showed a significant decrease in Cx43 protein [4] and mRNA expression [27], accompanied by GJC dysfunction in the PFC [4]. Loss of Cx43 delayed the growth rate of astrocytes [28]. The gap between the two neighboring astrocytes was around 1.5-fold wider than the control group [4]. CUS suppressed gap junction permeability and decreased gap junction density [4]. In another study, the Cx43 content of the orbitofrontal cortex of rats subjected to CUS decreased, and the myelin basic protein area fraction was positively correlated with the density of Cx43-positive puncta in the orbitofrontal cortex, suggesting that the change of Cx43 may be related to the myelin morphology disorder in depression [29]. Furthermore, rats subjected to CUS have higher levels of endogenous corticosterone (CORT) [30]. Rats treated with chronic CORT administration consistently showed decreased Cx43 protein and GJC dysfunction in the PFC [31]. The observations suggested that the increase of endogenous CORT caused by CUS might cause reduced Cx43 expression and dysfunction of Cx43 [32]. Rats exposed to chronic restraint stress also showed decreased Cx43 protein and GJC dysfunction in the PFC [33]. Another study found that Cx43 expression decreases in internal prefrontal cortex (mPFC) astrocytes of chronic social defeated stress (CSDS) mice was associated with neuronal activity decreases [34]. Both lipopolysaccharide (LPS) and CORT have been proposed as an inducer of depressive-like context [35]. In rat cultured astrocytes, chronic administration of exogenous CORT reduced Cx43 expression via enhancing the degradation and suppressing the synthesis of Cx43 [36, 37]. The passways of degradation affected include the ubiquitin-proteasomal and autophagy-lysosomal pathways of Cx43 [38]. CORT damaged GJC function by reducing the distribution of Cx43 and enhancing the phosphorylation of Cx43 at Ser368 [37]. LPS activated Cx43 HCs in primary cultured cortical astrocytes of the mouse but had no effects on Cx43 protein [39].

Hippocampus

CUS reduced the expression of Cx43 and the ratio of Cx43/glial fibrillary acidic protein (GFAP) in rat hippocampal CA1 area and impaired the function of GJCs in the rats. However, there are no differences in GFAP expression. It suggests that CUS treatment influenced the expression of Cx43 in astrocytes [40]. In addition, the expression of Cx43 was significantly decreased in the hippocampus of CSDS mice and was strongly associated with decreases in neuronal activity [34]. Acute restraint-induced activation of Cx43 HCs in astrocytes of the mouse was further enhanced by chronic restraint stress [11]. Interestingly, the enhancement of neuronal HC activities caused by chronic restraint stress was inhibited by Cx43 HC blockers, e.g., Gap26, Gap27, and Cx43E2. It indicated that Cx43 HCs was involved the enhancement of neuronal HC activities. However, neurons were confirmed to express Panx1 HCs [41] and Cx36 HCs [42], but not Cx43 HCs. The author speculated that Cx43 HC activities of astrocytes were a pre-requisite condition of enhancing neuronal HC activities induced by chronic stress [11]. It suggests that the enhancement of Cx43 HC opening in astrocytes caused by acute stress is an adaptation to stress in the short term. In contrast, the enhancement of neuronal HC activities induced by long-term stress can damage the crucial functions of brain physiology [43] involved in the mechanism of depression-like behavior. However, the level of Cx43 protein did not change under both acute and chronic resistant stress [11]. It can be speculated that dysfunction of Cx43 in the depression model is more important than changes of Cx43 protein levels. Consistently, CORT exposure increased the level of phosphorylated Cx43 at Ser368 in mice hippocampus [44], but not affect total Cx43 expression [37, 44].

In summary, reduced Cx43 protein levels and dysfunction, including GJC dysfunction and HC activation, were shown in the PFC and hippocampus in various models of depression.

Cx43 affects behaviors in depression

Prefrontal cortex (PFC)

Mitterauer et al. considered that decrease of connexins was related to the pathogenesis of depression firstly [45]. After that, various drugs that could regulate the function of GJCs and HCs were used to study the role of connexins in depression-like behavior [46, 47]. Cx43 overexpression in the mPFC increased neuronal activity and improved depressive-like behaviors of CSDS mice, while Cx43 suppression in normal mice reduced neuronal activity and induced depressive-like behaviors [34]. Futhermore, infusion of the non-selective GJC inhibitor carbenoxolone (CBX) into the PFC of healthy rats induced anhedonia in the sucrose preference test [4]. Cx43 specific GJC inhibitor Gap26 and Gap27 also displayed similar effects reversed by fluoxetine [4]. Reduced Cx43 protein levels and GJC dysfunction caused by CBX infusion in the PFC could be reversed by classic antidepressant fluoxetine [33]. Notably, CBX also inhibits HCs [48]. Gap26 and Gap27 specifically block Cx43 GJCs and HCs [49, 50]. It suggested that inhibiting the expression and function of Cx43 in the PFC contributed to the pathological mechanism of depression-like behaviors.

Hippocampus

Overexpression of Cx43 in the hippocampus astrocytes increased neuronal activity and inhibited depressive-like behaviors of CSDS mice, while suppression of Cx43 in normal mice was sufficient to reduce neuronal activity and induced depressive-like behaviors [34]. Bilateral infusion of gap junction blocker CBX into hippocampal CA1 area of healthy rats induces depression-like behavior [40]. This may be caused by GJC dysfunction in astrocytes. It was found that the Cx43 content in astrocytes increased [40], which may be a compensatory increase. Bilateral infusion of CBX into the ventral hippocampus (vHIP) decreased anxiety-like behavior in the elevated plus maze test and the open field test (OFT) [51]. It may be due to the electrical signaling, which generates synchronized activities between vHIP and mPFC, drives anxiety-like behaviors [51], and there is a unidirectional ipsilateral nerve projection from the CA1 area of the hippocampus to mPFC in rat brain [52]. In fact, unilateral injection of CBX into vHIP combined with contralateral injection into mPFC produced similar anxiolytic effects [51]. However, no change in anxiety behaviors were observed with CBX in single unilateral vHIP [51]. Moreover, the dorsal hippocampus (dHIP) is more related to the memory function of the hippocampus [53]. Consistently, infusion of CBX into bilateral dHIP did not induce changes in anxiety-like behavior [51]. CBX is a non-selective blocker that can block the gap junctions of both neurons and astrocytes [54]. It influenced electrical signaling due to disturbing the theta rhythm in the vHIP and mPFC [51]. Although Cx43 is not expressed on neurons, studies have found that inhibiting Cx43 in astrocytes reduced the activities of neurons [34], and Cx43 GJCs could regulate the synaptic plasticity of neurons. However, mice with constitutive deficiency of Cx43 in hippocampal astrocytes showed less despair behaviors in the TST and more exploratory behaviors in the OFT [44]. These animals lack Cx43 throughout development, leaving open the possibility of long-term compensatory mechanisms [51]. Furthermore, Cx43 levels decreased in the hypothalamus of these mice, so the incompletely consistent function of Cx43 in different brain regions may be an important factor [44].

Generally, the evidence we have collected suggests that Cx43 dysfunction of PFC and hippocampus may play a role in the onset of depression-like behaviors. Improving GJC dysfunction and inhibiting the activity of HCs may be new directions for improving depression-like behaviors. Moreover, it is more important to consider the function of Cx43 (HCs and GJCs) associated with depression but not Cx43 individual kinetics. Also, mice with the conditional knockout of Cx43, as achieved by crossing Cx43fl/fl mice with GFAP-cre mice, may not be the best tool for studying Cx43 function in a particular brain region. Injecting pseudotyped lentivirus containing the Cre-recombinase locally to drive the inactivation of Cx43 can be a better strategy. Furthermore, since gap junction blockers using currently can block both GJCs and HCs, there is no definite research and evidence for the isolated role of GJCs or HCs in the development of depression-like behaviors, and more relevant studies are needed in the future. Roles of Cx43 in other areas in the brain related to depression need to explore as well further because Cx43 in different brain regions have different functions [55]. Also, the Cx43 gap junction has a regulatory effect on the activities and function of neurons [34]. The neural circuit of vHIP-mPFC has been shown to play a role in depression-like behaviors. Therefore, further research on the effect of Cx43 on neural circuits is also a potential new direction.

Effects of antidepressants on Cx43

The antidepressants currently in clinical use mainly include tricyclic and tetracyclic antidepressants (TCAs), selective 5-HT reuptake inhibitors (SSRIs), 5-HT, and NE reuptake inhibitors (SNRIs) [56]. Rapid antidepressants, e.g., ketamine, are also a hot spot of current research. Many studies reported that the treatment with these antidepressants caused alterations in the expression of Cx43 in astrocytes.

Monoamine antidepressants

Ten monoamine antidepressants from four therapeutic classes were tested (Table 2). The results showed that 24 and 48 h treatment induced an increase in the Cx43 expression at the mRNA and protein level. The effect of fluoxetine on Cx43 was pronounced, which has been verified in several models in vivo and in vitro [4, 57, 58]. Thus, it can be inferred that the expression of Cx43 in the brain of patients with MDD decreases, and antidepressant treatment is beneficial to the upregulation of Cx43 [59]. Jeanson et al. systematically tested the gap junction communication and HC activities in primary cultured astrocytes with seven antidepressants from four categories (TCA, SSRI, NRI, SNRI) [39]. It is known that these cells only express Cx43 [60]. The results showed that these antidepressant drugs had different effects on both Cx43 GJC and HC functions of astrocyte, although the level of Cx43 did not change significantly (Table 2) [39]. These reports have contradictions and inconsistencies. Treatment with fluoxetine in vivo [61] and in vitro showed an increase in Cx43 expression level [4, 57, 58]. Given treatment with amitriptyline to primary cultured rat astrocytes, it showed an increase in Cx43 expression level [62] and inhibition of GJC function. However, both fluoxetine and amitriptyline did not change the expression level of Cx43, and only treatment with amitriptyline showed an inhibitory effect on GJCs [39]. Moreover, in the preliminary research of our laboratory, fluoxetine and duloxetine were administered to rats chronically in the control group and CUS model group. Through dye tracer experiment and electron microscopy analysis, fluoxetine and duloxetine had reversal effects on the GJC dysfunction of astrocytes in the PFC caused by CUS, but no impact on the GJC function without CUS in rats [4]. However, fluoxetine had an inhibitory effect on the GJC function of mouse frontal astrocytes cultured in vitro, while duloxetine had no noticeable impact [39].

Table 2 Summary of the effects of antidepressants for classes: TCA, SSRI, SNRI, NRI, and NMDAR antagonist.
Full size table

As a whole result, the effect of antidepressants on Cx43 function is far more complicated than the current literature reports. That opposed effects are observed within the same therapeutic class. There may be several reasons for these contradictions: First, the models used in these studies are different. The preliminary research in our laboratory worked on rats in vivo [4], whereas Jeanson et al. used cultured mouse astrocytes [39]. Furthermore, other cell culture models, e.g., human astrocytoma [58] and rat astrocytes [62], were also used. Secondly, the dosage and period of treatment differed. Cell culture models from different species and cell types may have different sensitivity and tolerance to antidepressants. Moreover, in the experiments that were operated in vivo models, animals are given chronic treatment indeed involved more integrated and complex mechanisms.

In addition, the effects of antidepressants on the Cx43 HCs of astrocytes have been less studied. LPS is considered an inducer of a depression-like context [35, 63]. It can activate Cx43 HCs in astrocytes by releasing pro-inflammatory factors tumor necrosis factor-α (TNF-α) and interleukin-1β (IL-1β), and astrocytes stimulated by LPS can be used to study the mechanism of antidepressant drugs [64, 65]. Seven monoamine antidepressants tested (fluoxetine, amitriptyline, paroxetine, imipramine, reboxetine, duloxetine, and venlafaxine) all had an inhibitory effect on LPS-induced HC activities in astrocytes [39]. At the same time, fluoxetine [66, 67], amitriptyline [66, 68], paroxetine [69], and imipramine [70] have all been found to inhibit the production of TNF-α and IL-1β. In contrast, venlafaxine has been found to increase the level of TNF-α [67]. Meanwhile, the inhibitory effect of venlafaxine on LPS-induced HC activities was the lowest [67]. Therefore, it can be preliminarily speculated that antidepressants may inhibit the activation of Cx43 HCs to decrease the activity of astrocytes by reducing the generation of TNF-α and/or IL-1β levels [39]. Besides, Cx43 HCs in astrocytes have also been found to mediate glutamate release [48, 71]. Combined with the hypothesis that glutamate cycle disorders cause depression [72], the antidepressant inhibits actions on HC activities can also support the current hypothesis [39]. However, more studies are needed to confirm and further explore the effects of antidepressants on the function of Cx43, Cx43 GJCs, and HCs.

Ketamine

In addition to mainstream monoamine antidepressants, other antidepressants, e.g., N-methyl-D-aspartic acid receptor (NMDAR) antagonist ketamine, also received attention. There are also a small number of studies on the effects of ketamine on Cx43 (Table 2). Acute administration of ketamine (20 µM, 30 min) had a significant inhibitory effect on Cx43 HCs of mouse cortical astrocytes, whereas ketamine (300 µM, 30 min) had an inhibitory effect on the GJCs [73]. It is generally believed that the plasma concentrations of ketamine exerting antidepressant effects in humans and rats are 10 and 20 µM, separately [74, 75]. Therefore, the acute treatment with ketamine at a therapeutic-relevant concentration had an inhibitory effect on the activity of the Cx43 HCs but not GJCs [32]. The mechanism may be inhibiting the release of inflammatory factors TNF-α and IL-1β [73]. The mechanism of high concentration ketamine to inhibit the function of GJCs may be directly acting on NMDARs and gamma-aminobutyric acid receptors (GABARs) on astrocytes [76, 77].

Candidate pathological mechanism of depression-like behaviors associated with Cx43

Cx43 and gliotransmission

Synaptic plasticity is the specific structural and functional change of synapses caused by the continuous activity of neurons, and it is closely related to the pathophysiological process of a variety of neuropsychiatric diseases [78]. Astrocytes are the most abundant and functional glial cells in the CNS [79]. They can affect neurons’ synaptic strength and function by expressing receptors and regulating neurotransmitters, e.g., adenosine triphosphate (ATP), glutamate, γ-aminobutyric acid [80, 81]. Its dysfunction can cause synaptic abnormalities, which are related to various mood disorders, e.g., depression [82]. Therefore, the concept of “tripartite synapse” was first proposed over 20 years ago to describe the intimate relationship between neurons and glutamatergic synaptic astrocytes [83]. In addition, microglia in resting states can also interact with astrocytes and neurons, so the hypothesis of “quad-partite synapse” has been proposed [84]. Due to the significant contribution of astrocyte dysfunction to depression [85], this section only discusses the contribution of Cx43 expression and function in the “tripartite synapse” to gliotransmission and synaptic plasticity.

Cx43 and glutamate-glutamine cycle

Presynaptic neurons release glutamate through vesicles [86]. Glutamate in the synaptic cleft binds to the glutamate receptors on postsynaptic neurons. Then, glutamate is quickly cleared up from the synaptic cleft, and more than 90% [87] is absorbed by astrocytes via the astrocyte-specific glutamate transporter GLT-1 [88], which is homologous to the excitatory amino acid transporter (EAAT2) in the brain of human [89]. In astrocytes, glutamate is transformed to glutamine by glial-specific glutamine synthetase. Then glutamine shuttles back to the presynaptic neurons and is converted into glutamate by neuron-specific phosphoric acid-activated glutaminase to supplement and maintain the glutamate storage of presynaptic neurons [87] (Fig. 1).

Fig. 1: Cx43 and glutamate-glutamine cycle under physiological conditions.
figure 1

Under physiological conditions, the function of Cx43 GJCs is normal, and most of the HCs remain closed. Presynaptic neurons release glutamate through vesicles. Glutamate in the synaptic cleft binds to the glutamate receptors on postsynaptic neurons, including NMDAR and AMPAR. Then, glutamate is quickly removed from the synaptic cleft, and more than 90% is taken up by astrocytes via the astrocyte-specific glutamate transporter GLT-1. In astrocytes, glutamate is converted to glutamine by glial-specific glutamine synthetase. A part of glutamine shuttles back to the presynaptic neurons. It is converted into glutamate by neuron-specific phosphoric acid-activated glutaminase to supplement and maintain the glutamate storage of the presynaptic neurons. Also, a part of glutamine enters GABAergic interneurons to synthesize GABA, binding to GABAR on presynaptic neurons and inhibiting glutamate release. Through the above two aspects of regulation, the concentration of glutamate in the synaptic cleft keeps low. AMPAR a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor, GABA γ-aminobutyric acid, GJC gap junction channel, GLT-1 glutamate transporter 1, HC hemichannels, NMDAR N-methyl-D-aspartate-receptor.

Full size image

In the PFC and hippocampus of rats exposed to CUS, the function of Cx43 GJCs was impaired [4] (Fig. 2). Moreover, after blocking the Cx43 GJCs in astrocytes, GLT-1 expression decreased [90]. Recent studies show that pharmacological inhibition of central astrocytic glutamate uptake with the GLT-1 inhibitor dihydrokainic acid (DHK) can induce anhedonia-/depressive-like behaviors. Brain region-specific inhibition of GLT-1 is sufficient to induce depression-like behaviors. The possible mechanism is that glutamate intake through GLT-1 is reduced, resulting in a decrease of the glutamate pool and excess glutamate in the synaptic cleft [91, 92]. Glutamate binds to over-activated NMDARs outside the synapse (particularly in GluN2B-containing NMDARs), increasing cell death and neuron loss [91, 93]. Also, dysfunction of GLT-1 damages glutamate release and uptake which further inhibits a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) and NMDAR action to affect the cortex, resulting in reduced synapse density and diameter and dendritic length and arborization [93] (Fig. 2). In addition, stress can also activate Cx43 HCs of astrocytes, and the outflow of a large amount of glutamate through HCs leads to a further increase of glutamate in the synaptic cleft [94]. Interestingly, postmortem studies found that glutamate decreased in the frontal limbic of patients with MDD [95]. Consistently, neuroimaging studies using proton magnetic resonance spectroscopy revealed reduced levels of glutamate in the brain areas, including PFC of patients with MDD [96]. However, ketamine might produce rapid antidepressant-like effects, at least in part by transiently increasing glutamate cycling in the PFC. It has been demonstrated in rodent and human studies [97, 98]. Therefore, we suspect that the effects of GJC and HC dysfunction on the glutamate cycle may be time-dependent. In the short term, glutamate release from activated HCs can be beneficial to adaptation to the stress [11]. However, continuous activation of HCs and dysfunction of GJCs induce the accumulation of glutamate in the synaptic cleft. The functional degradation of astrocytes caused by prolonged stress may cause the decreased glutamate levels in depressed individuals [85]. Thus, inhibiting the activated HC may be helpful to delay the progression of depression-like behavior, while activating HCs to release glutamate may produce a rapid antidepressant effect in individuals with severe depression.

Fig. 2: Cx43 and glutamate-glutamine cycle under stress conditions.
figure 2

(1) Stress causes dysfunction of Cx43 GJCs and activation of Cx43 HCs. (2) Dysfunction of GJCs reduces the expression of GLT-1. A possible intermediate target is GFAP. The activated HCs release large amounts of glutamate into the synaptic cleft. (3) The clearance of glutamate through GLT-1 decreases, and glutamine synthesis in astrocytes decreases. (4) Glutamine shuttled back to the presynaptic neurons decreases, and glutamate storage of presynaptic neurons decreases. (5) Glutamine entering into GABAergic interneurons decreases, and the synthesis of GABA reduces. The inhibitory effect of GABAergic neurons on the release of glutamate from presynaptic neurons is weakened. (6) Reduced glutamate clearance by GLT-1, increased release of glutamate by HCs, and dysfunction of GABAergic system together leads to the accumulation of glutamate in the synaptic cleft. Accumulated glutamate in the synaptic cleft leads to excessive and continuous activation of NMDAR and AMPAR, resulting in a decrease of BDNF release, thereby contributing to the onset of depression. AMPAR a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor, BDNF brain-derived neurotrophic factor, GABA γ-aminobutyric acid, GFAP glial fibrillary acidic protein, GJC gap junction channel, GLT-1 glutamate transporter 1, HC hemichannels, NMDAR N-methyl-D-aspartate-receptor.

Full size image

Furthermore, glutamate-glutamine cycling is considered the primary source of glutamine in the brain [86]. The reduction of GLT-1 expression caused by GJC dysfunction also reduces glutamine synthesis. Glutamine is the precursor for synthesizing the inhibitory neurotransmitter gamma-aminobutyric acid (GABA) [99]. Therefore, GJC dysfunction also leads to GABAergic system dysfunction. Gamma-aminobutyric acid (GABA) is responsible for fine-tuning and controlling excitatory transmission [100, 101], which is also a potential factor in the pathophysiology of MDD [102] (Fig. 2).

However, the specific mechanism by which Cx43 GJC dysfunction in astrocytes leads to reducing GLT-1 expression is still unclear, and a possible intermediate target is GFAP. The decrease in GFAP expression following stress exposure led to a strong glutamate transporter GLT-1 on the hippocampus and cortical astrocytes [85], which may be through a cAMP-protein kinase A-dependent pathway [103]. GFAP seems to act as a scaffolding protein to involve the expression of GLT-1 on the membrane surface [85]. Furthermore, in previous studies of our laboratory, rats were given a single bilateral infusion of CBX in the hippocampal CA1 area and sacrificed 2 days later. Interestingly, the expression of GFAP in the hippocampus increased [40]. A cellular self-protection mechanism may induce a compensatory increase in GFAP expression, and long-term gap junction dysfunction may decrease GFAP expression.

Cx43 and ATP

ATP is generally considered the primary energy currency of cells [85]. It widely mediates astrocyte-neuronal signal communication [104] and participates in the regulation of synaptic plasticity [105] (Fig. 3). Glutamatergic signaling triggered Ca2+ influx into neurons through AMPAR or NMDAR, resulting in a localized decrease in the extracellular Ca2+ concentration [106]. Cx43 HCs in astrocytes open in response to low extracellular Ca2+ conditions and mediate the efflux of ATP. ATP binds to astrocytic P2Y1R and P2Y2R, and then activates the G-protein coupled with P2YRs [107]. Activated G-protein further activates phospholipase C (PLC), leading to the release of intracellular inositol triphosphate (IP3) [108]. Then, IP3 binds to IP3 receptors on the ER membrane, and subsequent calcium releases from the ES calcium stores into the cytoplasm [85]. The localized increase in cytosolic Ca2+ is named a calcium wave [109]. The intracellular Ca2+ signal in astrocytes triggers the vesicular or non-vesicular release of gliotransmitters, e.g., glutamate, GABA, ATP, into the synaptic cleft [110, 111]. These gliotransmitters, especially ATP and glutamate, bind to synaptic receptors, providing a feedback signal on synaptic transmission and neuronal activities [85, 112].

Fig. 3: Under stress conditions, ATP mediates the formation of ICWs through Cx43 GJCs and HCs.
figure 3

(1) Glutamatergic signaling triggers Ca2+ influx into neurons through AMPAR or NMDAR, resulting in a localized decrease in the extracellular Ca2+ concentration. (2) Cx43 HCs in astrocytes open in response to low extracellular Ca2+ conditions and mediate the efflux of ATP. (3) ATP binds to P2YRs on astrocytes and then activates G protein and PLC, promoting the release of intracellular IP3. (4) The combination of IP3 and IP3R on the ER membrane causes calcium release from ER, forming calcium waves. (5) Calcium waves regulate the release of neurotransmitters from presynaptic neurons and further promote the release of ATP from HCs. (6) ATP is degraded to ADP. ADP activates interneuronal P2Y1 receptors, stimulating depolarization and firing, thereby enhancing inhibitory transmission. (7) ICWs transmit through Cx43 GJCs and HCs. Both Ca2+ and IP3 can enter the cytoplasm of adjacent astrocytes through GJCs and then induce the calcium waves in adjacent astrocytes. ATP released from HCs activates P2YRs on the membrane of neighboring cells and then induces the release of IP3 and subsequent calcium release from the ER calcium stores. (8) A large-scale calcium wave is formed in the astrocyte network to regulate neuronal activities under stress conditions. ADP adenosine diphosphate, AMPAR a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor, ATP adenosine triphosphate, ER endoplasmic reticulum, GJC gap junction channel, HC hemichannels, ICW intercellular calcium waves, IP3 inositol triphosphate, NMDAR N-methyl-D-aspartate-receptor, PLC phospholipase C.

Full size image

Furthermore, astrocytes can transmit the calcium signals to neighboring non-stimulated astrocytes, forming intercellular Ca2+ waves (ICWs) [113]. There are two possible pathways by which ICWs can be transmitted (Fig. 3). One is mediated by Cx43 GJCs [114]. Both Ca2+ [109] and IP3 [115] can enter the cytoplasm of adjacent astrocytes through GJCs and induce the calcium waves in adjacent astrocytes. The other pathway is through ATP in the synaptic cleft. ATP activates P2YRs on the membrane of neighboring cells and then induces the release of IP3 and subsequent calcium release from the ES calcium stores [109]. These two pathways are not mutually exclusive but work together to promote the transmission of ICWs [109]. However, based on current research about ICWs in astrocytes, it seems to act crucial roles under pathological conditions but not physiological conditions [109, 116]. Spontaneous ICWs are sparsely seen in the mature CNS, even after physiological stimulation [109, 117]. This may be related to the low activities of the HCs under physiological conditions [118]. The low concentration of ATP is not enough to support the long-distance propagation of ICWs. In addition, the P2X7 receptor may also be involved in the propagation of ICWs [119]. Several GJC/HC blockers, e.g., heptanol, octanol, CBX, and mefloquine, prevented ATP-dependent spread of ICWs under low divalent cation solutions by affecting P2X7R [119]. But the specific mechanism needs to be further explored.

Moreover, ATP is degraded to ADP, activating P2YR1 on the interneuron [107, 120]. The interneuron is stimulated to depolarize and increase firing [107]. This pattern of neuron-glia signaling may be a negative feedback mechanism that increases inhibitory transmission while glutamatergic activities are excessive [107]. Also, blocking ATP release from astrocytes can induce a decrease in the number of hippocampal nerve spines in mice, leading to neurological disorders and depression-like behaviors [121]. Exogenous ATP or endogenous activation of astrocytes to promote ATP release can quickly reverse depression-like behaviors within a week [122]. This may be related to the ATP release of astrocytes, which can enhance inhibitory transmission relying on local excitation and serve as a brake on network excitatory output [107].

Astrocytes rapidly take up glutamate in the synaptic cleft to provide primary metabolic substrates glutamine to neurons and control synaptic activities by releasing ATP and generating calcium waves [109]. Cx43 HCs mediate ATP efflux [123]. Thus, Cx43 may be a promising antidepressant target worth developing.

Cx43 and metabolic disorders

As we all know, depression is a disease of environmental and genetic factors [3]. Its pathogenesis is very complicated, and it is not limited to brain cells and molecular abnormalities. Therefore, considering other related factors in the body is essential for understanding depression. This section will explain the relationship among Cx43, metabolic disorders, and depression from the hypothalamus-pituitary-adrenal axis (HPA axis) dysfunction, insulin resistance, and thyroid hormone dysfunction (Fig. 4).

Fig. 4: Cx43 dysfunction contributes to depression through peripheral mechanisms.
figure 4

(1) ERS in hepatocytes spreads through Cx43 GJCs and promotes insulin resistance. (2) Insulin resistance causes Cx43 GJC dysfunction by increasing the level of H2O2 in vascular smooth muscle cells. (3) TSH promotes TH production by increasing the synthesis of Cx43 and inducing the opening of Cx43 GJCs in thyrocytes. (4) Excessive CORT produced by adrenocortical cells causes Cx43 GJC dysfunction of astrocytes in the brain. (5) Under conditions of stress, activated microglia releases inflammatory factors, promoting Cx43 HC opening and Cx43 GJC dysfuntion of astrocytes. (6) Opened HCs further contribute to the activation and spread of the inflammasome pathway. (7) Insulin resistance can promote neuroinflammation and affect the brain function. CORT corticosterone, ERS endoplasmic reticulum stress, GJCs gap junction channels, HC hemichannel, TH thyroid hormone, TSH thyroid stimulating hormone.

Full size image

Cx43 and HPA axis dysfunction

The HPA axis is the primary link in coping with stress [124]. Acute or chronic stress can cause changes in the function of the HPA axis [125]. HPA axis dysfunction is one of the recognized biochemical changes in depression [126]. The stress activates the HPA axis and then induces the paraventricular nucleus to release corticotropin-releasing hormone (CRH). CRH triggers the hormone cascade release, eventually triggering the release of glucocorticoid (GC). GC inhibits the activity of the HPA axis through negative feedback [127]. Cortisol in humans and CORT in rodents are the primary GC that acts on central GC receptors and affects the function of the metabolic regulation hormone [127]. Reduced expression of Cx43 in the hippocampus and frontal cortex was found in several stress-related depression animal models, including CUS [4, 29, 128], acute/chronic restraint stress [11], and exogenous CORT models [44]. In vitro, CORT inhibits gap junctional intercellular communication (GJIC) in prefrontal and hippocampal astrocytes [37]. Furthermore, the expression of Cx43 was significantly reduced [129], while phosphorylated Cx43 significantly increased [37]. The possible mechanism of Cx43 reduction induced by CORT includes decreasing Cx43 biosynthesis and membrane distribution, increasing Cx43 degradation, and regulating Cx43 stability [37]. In addition, the phosphorylation of Cx43 at Ser368 might cause the decreased expression of Cx43 and GJC dysfunction [130, 131]. There are relatively few studies on Cx43 HCs. Studies have initially found that CORT activated HCs [11]. Thus, Cx43 may be involved in the pathophysiological changes of depression under the HPA axis hypothesis by weakening GJC functions and enhancing HC functions [32]. More experiments are needed to confirm the effect of CORT on Cx43 HCs and explore possible mechanisms. In addition, the effect of changes in Cx43 expression and functions on endogenous CORT secretion is also needed to explore.

Cx43 and insulin resistance

The results of multiple studies have shown that insulin resistance may be the cause of depression induced by obesity [132]. Insulin administration had antidepressant effects [133]. Alterations of Cx43 are related to insulin resistance. High-dose insulin increased the phosphorylation of Cx43 at Ser368 in vascular smooth muscle cell GJCs and decreased the expression of Cx43, causing GJC dysfunction. High-dose insulin treatment largely enhanced the H2O2 level in cells. Furthermore, pretreatment with catalase increased the expression of Cx43, decreased the phosphorylation level of Cx43 at Ser368, and recovered cellular GJIC function compared with cells only treated with high-dose insulin [134]. The results suggest that high-dose insulin impairs intercellular GJIC through the oxidative stress-activated signaling pathway [134]. It may be a possible mechanism by which excessive insulin secretion in the body leads to insulin resistance [135, 136]. However, the increased expression of Cx43 and cell-cell coupling in hepatocytes might be a factor that promoted the development of insulin resistance. Endoplasmic reticulum stress (ERS) is an essential character of obesity and type 2 diabetes [137]. It is also a crucial factor in the network of stress signals regulating insulin resistance and diabetes [138]. In hepatocytes, ERS increases Cx43 expression and cell-cell coupling. The liver Cx43 knockout mice can avoid the effects of highly nutritious diet-induced ERS and insulin resistance. In obese mice, Cx43-mediated cell-cell coupling promotes the transmission of ERS between cells [21]. It suggests that Cx43 and GJCs provide a pathway for the transmission of chronic stress signals [21], and may play a role in promoting the onset of insulin resistance and depression. As a way of cell communication, Cx43 GJC mediates the necessary communication between cells, however, it also facilitates the spread of harmful substances [139, 140]. Thus, the contribution of Cx43 to insulin resistance is also bidirectional, which may be related to the role of Cx43 in this tissue.

Cx43 and thyroid hormone dysfunction

The thyroid hormone plays a vital role in developing a healthy brain [141] because it relates to neuron precursors’ proliferation, migration, differentiation, and synapse formation [142]. The effect is mediated mainly by glial cells, particularly astrocytes [143]. Clinical studies have found that the levels of thyroxine (T4), free triiodothyronine (FT3), and free thyroxine (FT4) in female patients with depression were significantly lower than those in healthy groups, suggesting that patients with depression might be accompanied by hypothyroidism [144]. Also, the hyperfunction of the hypothalamus-pituitary-thyroid axis (HPT axis) affects brain function [145]. Various studies suggest that the HPT axis is involved in the pathophysiology and prognosis of depression [146]. Notably, some patients with depression have normal thyroid functions [147, 148]. The specific relationship between depression and thyroid hormone dysfunction has been demonstrated in cross-sectional studies, and findings have been inconsistent [149]. The existing literature suggests a link between subtle thyroid dysfunction and depression, but more data are needed [150]. In fact, screening for thyroid dysfunction is part of the routine assessment of depressed patient [144].

Like most tissues, thyrocytes communicate with each other through various types of intercellular junctions, including GJCs. There are two connexins expressed in thyrocytes: Cx43 and Cx32 [151]. Thyroid-stimulating hormone (TSH), produced in the brain’s hypothalamus, promotes thyroid hormone production by increasing the synthesis of these two connexins and inducing the opening of GJCs in thyrocytes [20]. Cx43 GJC-mediated cell-to-cell communication is likely involved in controlling thyrocyte proliferation [151]. In patients with Hashimoto disease, the expression of Cx43 in thyroid epithelial cells decreased [152].

Furthermore, some studies showed that thyroid hormone had a regulatory function on the expression of connexin [153]. Thyroid hormone inhibits the growth and proliferation of Sertoli cells by inducing the expression of Cx43 [20]. Gap junction coupling inhibitors have been shown to significantly reverse the inhibitory effects of thyroid hormones on the growth and proliferation of Sertoli cells [20]. Antidepressant fluoxetine has been reported to cause male sexual dysfunction [154], but the target that induces this side effect is still unknown. Sertoli cells have nutritional and protective effects on developing sperm [155]. Fluoxetine may impair the integrity of the peritubular myoid cell-Sertoli cell and disturb spermatogenesis, which may be one reason why fluoxetine induces male sexual dysfunction [156]. As the critical target that regulates the growth and proliferation of Sertoli cells [157], Cx43 may be a potential target to improve this side effect of fluoxetine. In addition, cellular therapy is an emerging treatment strategy for depression [158]. As ‘nurse cells’ within the testis, Sertoli cells are considered an ideal substitute for neurotransmitter antagonists and inhibitors due to their anti-inflammatory, neuroprotective, and nutritional effects [157]. As the target of regulating the growth and proliferation of Sertoli cells, Cx43 also has great potential value for developing a new type of antidepressant therapy and understanding the cellular and molecular processes that underlie depression.

GJCs and HCs formed by aggregated connexin are essential for direct communication between cells [159]. Adjacent cells share cytoplasmic contents through GJCs, including ions, second messengers, and small metabolites [160, 161]. Under physiological conditions, hormones can regulate the expression of Cx43 through endocrine, autocrine, and paracrine methods. Under stress conditions, on the one hand, functional changes of Cx43 GJCs and HCs Cx43 can control the dramatic changes in the internal and external environment of cells caused by stress within a certain range. On the other hand, GJC can provide a way for the rapid spread of toxic substances, intracellular redox components, or signal molecules in neighboring cells, contributing to metabolic disorders [162]. Therefore, in a short time, changes of Cx43 may be beneficial to the adaptation of cells and the body to external factors. However, long-term stress will damage the function of Cx43 [4], inducing metabolic disorders. In turn, metabolic disorders affect the secretion of biologically active molecules, e.g., hormones, chemokines, cytokines, and lipids [162]. Changes in the secretion of these substances dynamically alter the communication among cells, neighboring and distant cells [162], thereby affecting the function of connexins [153]. In short, the function of connexin and metabolic disorders affect each other, and metabolic disorders are inseparable from depression. It suggests that improving metabolic disorders by regulating the function of connexin may be a potential target for depression treatment.

Cx43 and neuroinflammation

The critical role of neuroinflammation in the occurrence and development of MDD has been testified by many experimental studies [163]. Neuroinflammation and Cx43 also have a complex relationship (Fig. 4).

On the one hand, neuroinflammation affects the expression and function of Cx43. In patients with MDD, the expression of pro-inflammatory factors, e.g., TNF-α [164] and interleukin- 6 (IL-6) [165], increased, which was related to the severity of MDD. Some antidepressants can inhibit the production of these inflammatory factors [166]. Under conditions of neuroinflammation, microglia are activated and release these inflammatory factors to promote the opening of Cx43 HCs in astrocytes [48]. Opened HCs further contribute to the activation and spread of the inflammasome pathway [167]. Other inflammation and stress factors include fibroblast growth factor-2 (FGF-2) [168], transforming growth factor-β (TGF-β) [168], and arachidonic acid (AA) [169] have also been found to reduce the expression of Cx43 and inhibit the opening of GJCs. Another study found that NO and NO-derived compounds could enhance Cx43 HCs and reduce GJ communication [170, 171]. The Cx43 GJC dysfunction in astrocytes has been an essential condition for the onset of depression. Some studies suggest that the activation of Cx43 HCs may be a factor that induces depression [11]. Therefore, Cx43 may be a critical intermediate factor in inflammation-induced depression; however, the underlying complex mechanism is still unclear.

On the other hand, changes in the expression and function of Cx43 may promote the production of inflammatory responses. A recent study has found an encouraging result that inhibitors of Cx43 HCs (tonabersat) could inhibit inflammasome activation and damage in the CNS [172]. However, research on the mechanism is relatively scarce. Astrocytes cultured in vitro were found to release NAD+ through HCs in the resting state, and then NAD+ was transported from the cytosol to the active site of CD38 [173]. CD38 is a multifunctional enzyme that can enhance Ca2+ levels in cells and Ca2+-dependent functions accordingly by autocrine and paracrine mechanisms [174]. Enhancing Ca2+ levels upmodulates various intercellular signaling pathways, e.g., inflammation pathways [175]. In addition to astrocytes, the function of Cx43 in other cells may also be related to inflammation-induced depression. Depression and Alzheimer’s disease (AD) always co-occur, and depression increases the risk of AD [176]. High Cx43 HC activities of mastocytes in the brain may play an essential role in the occurrence and development of AD [177]. Moreover, mastocytosis, rare activation and aggregation of mast cells in different kinds of tissues, may be involved in the process of inflammation-induced depression [178]. Thus, although the global study is sparse, as in the case of AD, high HC activities in mastocytes probably were early factors in the inflammation pathways linked to depression [177]. However, the cascade of events taking place among activation in mastocytes, involvement of Cx43 GJCs/HCs in astrocytes, and the damage of neurons need further studies [177].

Conclusion and perspectives

In this review, we summarized recent advances in Cx43 related to depression. It mainly includes the abnormalities of Cx43 in depression, the effects of Cx43 changes on depression, and the effects of antidepressants on Cx43. In addition, we emphatically describe the candidate pathological mechanisms of depression-like behaviors associated with Cx43. Cx43 dysfunction contributes to depression by the following aspects: affecting the gliotransmission, e.g., glutamate-glutamine cycle and ATP release; inducing dysfunction of HAP axis, insulin, and thyroid hormone; inducing neuroinflammation. However, these mechanism needs further studies to confirm.

The following outstanding questions need to be studied: (1) The current research on Cx43 changes under depression mainly focuses on astrocytes in the brain, but how does Cx43 change in other tissues and cells? (2) How do Cx43 GJCs and HCs independently participate in the pathogenesis of depression?

In addition, the glutamate cycle, ATP release, neuroinflammation, and metabolic disorders are all targets that can be used for the treatment of depression. Cx43 may be an upstream target that caused these changes, but the sequence of these changes needs further study. The etiology of depression is complex, and further research on Cx43 would contribute to shedding new light on the pathogenesis of depression. Cx43 also can develop into a novel, root cause-based antidepressant treatment target for depression.