SalB and CBX modulated astrocytic Cx43 expression, hemichannel permeability, and gap junction communication after OGD/R injury
In this study, we measured subcellular compartment-specific Cx43 protein levels in astrocytes. SalB and CBX similarly reversed OGD/R injury-induced internalization of plasma membrane Cx43 but did not significantly change total cellular Cx43 levels. These results are consistent with our previous observation that ischemic injuries induced cytoplasmic internalization of Cx43 in rat astrocytes but did not change total Cx43 levels [
77]. However, some studies also reported increased Cx43 expression post-ischemic [
78‐
80]. Nakase and colleagues were first to investigate Cx43 expression levels in the human brain, and they found increased Cx43 levels after long-term ischemia [
79]. Another study by Xie et al. revealed that inhibiting Cx43 upregulation significantly increased pyramidal neuron survival and alleviated cognitive impairments after middle cerebral artery occlusion [
12]. We speculate that the discrepancies may arise from differences between in vivo and in vitro models, as elevated Cx43 expression in vivo may result from astrogliosis, whereas in vitro studies usually focus on Cx43 protein levels per cell. In 2010, Orellana and colleagues demonstrated that hypoxia/reoxygenation causes a transient increase in astroglial surface Cx43 protein levels [
80], which has been assumed that surface Cx43 is in the form of hemichannels [
81]. Hence, the authors indicated that the increased hemichannel-related Cx43 proteins could account for the increased Cx43 hemichannel activity, which were also in accordance with works by Retama and colleagues in 2006 [
82]. Here, we applied a commercial plasma membrane protein isolation kit from Invent Biotechnologies, as guided in the protocols; plasma membrane subtraction was separated from a mixture of nuclei, cytosol, and organelles by subsequent differential centrifugation and density centrifugation. The methods for plasma membrane extraction employed here showed no selectivity for gap junctional or hemichannel-related Cx43 proteins; in other words, the methods used in our research allowed collection of both hemichannel- and gap junction channel-related Cx43 proteins. In the current study, we found that, in OGD/R groups, plasma membrane Cx43 protein levels were apparently downregulated, while cytoplasma Cx43 protein levels were upregulated, compared with that in normal cultured astrocytes (Fig.
1). It has been proposed that under control situations, only about 15% of the total astrocytic Cx43 proteins expressed was in hemichannels [
82]; thus, we speculated that the decreased plasma membrane Cx43 proteins levels were mainly those Cx43 proteins composing gap junctional channels, which were internalized into cytoplasma for further degradation after OGD/R stimulation.
In addition to Cx43 protein levels, studies have also investigated astrocytic Cx43 hemichannel activity and GJIC coupling. Here, we found that subjecting astrocytes to OGD/R injury prominently increased their ethidium uptake ability and supernatant ATP concentrations but decreased astrocytic dye coupling degree. SalB or CBX treatment both achieved significant attenuation of the effects on ethidium uptake and ATP release. Furthermore, SalB treatment enhanced astrocytic cellular dye transfer, while CBX application showed inhibited effects (Fig.
2).
Hemichannels release relevant quantities of signaling molecules (e.g., ATP, glutamate, NAD
+ and PGE2) to the extracellular milieu [
83]. In vitro ischemia-like conditions enhance hemichannel activity in astrocytes and many other cell types [
7]. Studies have provided strong evidence that deleterious hemichannels open after cerebral ischemia [
84,
85]. In the current study, we performed dye uptake by astrocytes with EtBr incubation and bioluminescence for determination of eATP concentration, both were indicators for hemichannel activity, and found that increased astrocytic hemichannel opened under OGD/R injury, in accordance with those previous studies. However, it should be noticed that both connexin and pannexin expressed on astrocytes contribute to hemichannels [
7]. Here, we applied CBX, blocker for both Cx and pannexin channels [
86], which showed inhibition for astrocytic hemichannel activity. Further study specially target Panx1 with
10Panx1 may distinguish and investigate the possible contribution of Px1 hemichannels and Cx43 hemichannels [
7]. In particular, Orellana et al. found no significantly increased dye uptake in Cx43-deficient astrocytes after hypoxia. In addition, Cx43 mimetic peptide prevented hypoxia induced dye uptake by hemichannel in astrocytes, but not by pannexin hemichannel blockers [
80]. Iwabuchi and Kawahara have proposed a complex negative feedback loop for pannexin hemichannels, whereby released ATP acts via P2X7 receptors to induce Pan1 hemichannel closure [
87].
Gap junction channels have evident physiological significance in morphogenesis, development, and tissue synchronization, but two opposing hypotheses exist in regard to their role in cell death. The transfer of caspase-derived apoptotic peptides through gap junction channels supports a “bystander” hypothesis, as studies showing that non-selective gap junction blockers, such as octanol [
88] and CBX [
89,
90], provide protection in models of brain ischemia. In contrast, a “good Samaritan” role is supported by studies showing that Cx43 gene knockout is associated with larger stroke lesions, amplified apoptosis, and inflammation [
91,
92]. Furthermore, post-injury gap junction channel inhibition correlates with glutamate cytotoxicity and neuronal injury aggravation [
93,
94]. Our findings seem to support the “good Samaritan” hypothesis, but there may be a balance between the “bystander” and “good Samaritan” hypotheses. The known discrepancies probably arise from the use of non-selective gap junction blockers that also inhibit hemichannels.
In conclusion, we observed opened hemichannels, weakened GJIC, and Cx43 internalization in astrocytes after OGD/R injury. Both CBX and SalB inhibited Cx43 redistribution. CBX suppressed the opening of hemichannels and gap junctions; SalB enhanced cell communication while reducing hemichannel openings.
Effects of ACM from SalB- and CBX-treated astrocytes on microglial activation after OGD/R injury
Inflammatory responses contribute to secondary neuronal damage, which substantially affects acute ischemic injuries. After ischemia, newly activated microglia produce both detrimental and neuroprotective mediators, with the balance between them determining the injured neurons’ fates. Activated microglia can exhibit either the classic M1 pattern, in which they secrete pro-inflammatory cytokines and exacerbate neuronal injuries, or the alternative M2 pattern, in which they promote reparative anti-inflammatory responses [
27]. Several receptors expressed on microglia recognize specific ligands, including heat shock proteins, ATP, and nucleic acids [
95,
96]. Ischemia-induced neuronal death results in ATP release and microglial activation via P2 receptors. This corresponds with significant post-ischemic elevation of microglial P2X4 and P2X7 receptor expression [
97,
98]. Although many factors mediate the migration of activated microglia to the injured region, ATP is among the most important mediators [
99]. Extracellular ATP induces endogenous ATP release from microglia, which attracts distant microglia to the injury site [
123].
ATP release through astrocytic hemichannels establishes an ATP gradient that is a critical trigger for microglial responses. In 2005, Davalos et al. showed that local ATP injections mimicked traumatic brain injuries and induced microglial activation, which was inhibited by connexin channel blockers [
100]. This indicates that extracellular ATP released from damaged tissues and surrounding astrocytes mediates a rapid microglial injury response. Furthermore, Huang et al. showed that Cx43 knockout mice exhibited diminished areas of post-traumatic ATP release, suppression of astrogliosis and microgliosis, and less tissue loss following spinal cord injuries [
101]. Similarly, another study showed that partial deletion of astrocytic Cx43 expression similarly reduced pro-inflammatory cytokine levels after systemic lipopolysaccharide injections [
26]. Moreover, partial Cx43 deletion inhibits microglial activation in mice, and hemichannel modulators such as Cx43 mimetic peptide [
24] and Cx43 antisense oligodeoxynucleotide [
102] effectively inhibit post-spinal cord injury inflammation. These results suggest that connexin hemichannels act as a “switch” for the inflammasome signaling cascade by contributing extracellular ATP both during and after an injury. Here, we found that SalB attenuated OGD/R injury-induced microglial activation, including the morphology changes, M1/M2 polarization, and release of pro-inflammatory or anti-inflammatory cytokines. Furthermore, when applied to microglia, OGD/R + SalB-ACM and OGD/R + CBX-ACM induced weaker microglial inflammatory reactions than OGD/R-ACM did, which is consistent with the results of the previous studies (Figs.
4 and
5).
In particular, it should be noticed that neither SalB nor CBX was a specific Cx43 hemichannel or gap junctional blocker [
49,
54,
103]. Hence, we further applied specific Cx43 hemichannel blocker-Gap 19 to explore the role of astrocytic Cx43 hemichannel during OGD/R injury. Results indicated that Gap 19 application significantly blocked OGD/R-induced Cx43 hemichannel opening and ATP release. Furthermore, OGD/R-ACM promoted microglial activation and HT-22 neuronal apoptosis, while after incubation with apyrase for 30 min, OGD/R + apyrase-ACM attenuated microglial activation and HT-22 neuronal injury. Also, OGD/R-Gap19-ACM obviously attenuated microglial activation, while addition of ATP in OGD/R-Gap19-ACM enhanced microglial activation and HT-22 neuronal damage. Another gap junctional inhibitor Gap26 showed similar results to those of Gap19 (Figs.
8,
9,
10, and
11).
We conclude that OGD/R injuries induce astrocytic Cx43 hemichannel opening and thus cause substantial ATP release, which plays an important role in microglial activation and HT-22 neuronal survival during OGD/R injury process. SalB and CBX may exert their protective effects by reducing ATP release; further study using Cx43 mimetic peptide Gap19 established the critical role of astrocytic Cx43 hemichannel and the secondary released ATP during OGD/R injury-induced neuroinflammatory responses.
Effects of MCM on astrocytic hemichannels and gap junctions after OGD/R injury
Previous studies showed that incubating astrocytes with pro-inflammatory cytokines or a high proportion of microglia caused reduced Cx43 expression and dye coupling accompanied with extensive microglial activation. Adding the anti-inflammatory mediator transforming growth factor β1 reverses the microglial activation and restores functional coupling [
28,
30]. We cannot exclude the possibility that cytokines directly affect astrocytic properties like Cx43 expression, especially given the evidence that the pro-inflammatory cytokine IL-1β directly affects astrocytic gap junctions [
104,
105]. Similarly, it has been reported that amyloid β (Aβ) induces microglial activation and thereby influences astrocytic gap junctions [
106] and that CB treatment prevented Aβ-induced astrocytic hemichannel activation [
107].
In our study, treating astrocytes with OGD/R-MCM induced a prominent increase in ethidium uptake but reduced cell coupling, but using OGD/R + SalB-MCM reversed these effects (Fig.
6). The mechanism remains unclear, though the functional interference may involve phosphorylation, since Cx43 function is quite sensitive to various kinases and phosphatases, including MAPK. For instance, brain slice studies have shown that ischemia, which upregulates the expression of cytokines such as IL-1β and TNF-α, induces Cx43 dephosphorylation [
108]. Furthermore, cytokines affect other astrocytic properties. For example, the pro-inflammatory cytokine TNF-α activates PKC, which causes depolarization of astrocytes [
105]. Further research is necessary to clarify the mechanisms.
In conclusion, our findings indicate that activated microglia and their pro-inflammatory cytokine secretions differentially regulate astrocytic gap junctions and hemichannel activity, which may in turn aggravate ATP release from opened hemichannels and thus form a vicious circle after OGD/R injury.
Effects of SalB and CBX on Src, PKC, and PKB and the corresponding Cx43 regulatory sites in astrocytes after OGD/R injury
Phosphorylation of Cx43’s C-terminal domain regulates GJIC. This domain is phosphorylated at over a dozen residues [
37‐
40]. Many kinases phosphorylate Cx43, and the predominant effect is a decrease in GJIC [
41]. In the ischemic penumbra, significant changes happen in the states of numerous signaling pathways involving those protein kinases, including MAPK family members, PKB and PKC kinases [
43‐
47]. In our study, we assessed protein expression in OGD/R-injured astrocytes and found that Ser368-phosphorylated Cx43 levels were decreased in the plasma membrane but increased in the cytoplasm. Furthermore, PKC, which phosphorylates the Cx43’s Ser368 site, was significantly upregulated and activated in the plasma membrane. Our results were similar to those of a previous immunohistochemistry study that showed ischemia-induced dephosphorylation of astrocytic Cx43 [
109]. However, it remains unclear how both Cx43 dephosphorylation and PKC activation occur during OGD/R injury, as under normal conditions, Ser368-phosphorylated Cx43 levels remain whereas no PKC activation is observed. OGD/R injury may induce some other unknown factors. Research has showed that uncoupling of Cx43-based GJIC was more a cause than a consequence of Cx43 dephosphorylation because post-hypoxic decreases in astrocytic coupling occurred before Cx43 dephosphorylation [
41]. In cultured astrocytes exposed to hypoxia, Cx43 dephosphorylation occurs in conjunction with reduced GJIC [
110,
111]. In our study, we found that SalB inhibited PKC activation and upregulated Ser368-phosphorylation of Cx43, which may be related to enhanced astrocytic coupling. However, CBX inhibited PKC activation and reduced Ser368-phosphorylation of Cx43, which indicates that Cx43 or Cx43-related GJIC may also exert regulatory effects on PKC activity.
Ser373-phosphorylated Cx43 is also associated with dramatically increased gap junction size and gap junctional communication [
114]. Akt induces Ser373 phosphorylation of Cx43 [
101], and inhibiting Akt causes gap junction losses [
39]. Here, in the OGD/R group astrocytes, we found increased levels of Ser373-phosphorylated Cx43 in both the plasma membrane and cytoplasm and reduced cytoplasmic levels of PKB’s Thr308-phosphorylated activated form. Solan and Lampe explored post-injury gap junctional upregulation and turnover in a model of wound healing and found that under conditions of injury or growth factor treatment, the first step is characterized by increased gap junction size and gap junctional communication and Akt activation [
40]. Besides, Ser373-phosphorylation of Cx43 occurs within 5–15 min after injury [
113]. They suggested that this initial step served to deplete the plasma membrane of non-junctional Cx43 by rapidly incorporating it into gap junctions and efficiently internalizing it. In our study, we detected these proteins after a 48-h reperfusion period, which may account for the opposite effects on PKB activity and Ser373-phosphorylated Cx43 levels, because after 48 h, most of the astrocytic Cx43 would already have been internalized. Further studies are necessary to investigate these changes in the early periods after OGD/R injuries. An interesting direction for such studies was indicated by Bejarano et al. [
114], in which they showed that connexins modulate autophagosome biogenesis and observed internalization of connexin-autophagy protein complexes. Furthermore, novel electron microscopy techniques have also been used to show localization of phosphorylated Cx43 in mouse ovarian follicles [
10]. Furthermore, we found that SalB reduced plasma membrane levels of Ser373-phosphorylated Cx43 but increased cytoplasmic Thr308-phosphorylated PKB levels, whereas CBX exerted no such effects. Besides its function in phosphorylating Cx43, PKB is involved in myriad cellular processes including cell survival, metabolism, and protein synthesis [
116]. SalB-induced Thr308- phosphorylation of PKB may also provide protection.
Src has long been known to downregulate gap junction communication and cause gap junction disassembly by phosphorylating Cx43 [
117,
118]. Src directly phosphorylates Cx43’s Tyr247 and Tyr265 residues [
119,
120]. In this study, we showed that OGD/R injury significantly activated Src, as indicated by the upregulation of cytoplasmic and plasma membrane levels of Tyr416-phosphorylated Src. Furthermore, the OGD/R group also exhibited increased plasma membrane levels of Tyr265-phosphorylated Cx43. This is consistent with previous studies [
41,
42]. Interestingly, in a wound healing model in which Akt phosphorylated Cx43 within 5–15 min of the injury, Src exerted its function within 30 min and continued doing so for 24 h or longer. This was accompanied by rapid downregulation of gap junctional communication and gap junctional internalization, which is critical to later steps in effective wound healing [
121]. Similar phenomena are also observed in ischemic pathologies. For example, Li et al. found that chemical ischemia/hypoxia induced marked astrocytic Cx43 dephosphorylation, and the “dephosphorylated” form of connexin-43 was immunoprecipitated by a phosphotyrosine antibody [
41], suggesting tyrosine phosphorylation of connexin-43 by Src. Furthermore, inhibiting Cx43 dephosphorylation blocked Src-Cx43 interactions. Naitoh et al. showed that in isolated rat hearts, PKCε was co-immunoprecipitated with Cx43 in the non-ischemic myocardium and that the levels of both increased after the onset of ischemia [
42]. Cx43-Src complexes were detected 35 min after ischemia but not under the baseline condition or at 10 min after ischemia. We therefore conjecture that after the 48-h reperfusion period in our study, Src had been activated, Cx43’s Tyr265 site had been phosphorylated, and large-scale Cx43 internalization was underway.
Recently, Pan and co-workers showed that SalB directly inhibited Src activity [
57]. We found that SalB increased astrocytic plasma membrane levels of Src’s Tyr527-phosphorylated deactivated form but did not significantly decrease plasma membrane levels of Tyr416-phosphorylated Src, which may be due to incomplete dephosphorylation [
122]. However, SalB did decrease cytoplasmic levels of Tyr416-phosphorylated Src. As for Tyr265-phosphorylated Cx43, SalB decreased plasma membrane levels but increased cytoplasmic levels. These results indicate that SalB inhibited Src and reduced Tyr265-phosphorylated Cx43 levels. Combined with our observations that SalB decreased Ser373-phosphorylated Cx43 levels and increased Ser368-phosphorylated Cx43 levels in the plasma membrane, we conclude that SalB-induced Src inhibition may promote Ser368-phosphorylation of Cx43, which is associated with Cx43-related GJIC under normal conditions.
CBX is a semisynthetic derivative of glycyrrhetinic acid [
124]. It has been demonstrated that CBX produced inhibition of the both hemichannel and gap junctional intercellular communication [
125,
126]. In the current study, 10 μM of CBX was selected based on MTT-viability tests for astrocytes implanted for OGD/R injury, as shown in Additional file
1: Figure S1B. Further, WB analysis for various phosphorylated Cx43 proteins and related protein kinases showed that CBX treatment induced obviously downregulation of p-Cx43(Ser368), accompanied by decreased p-PKC(Ser729) protein levels in plasma membrane, while showing no significantly regulation for p-Cx43(Tyr265) and p-Cx43(Ser373). Besides, CBX treatment inhibited plasma membrane’s Src kinases activity, with markedly decreased p-Src(Tyr416) protein levels. Here, several issues need to be mentioned. First, although it has been widely used, the potential inhibitory mechanism of CBX is still not completely understood until now. Verselis and Srinivas indicated that it may be that these reagents work through protein internalization or turnover, or perhaps an indirect mechanism involving binding to cytoplasmic intermediate molecules [
127,
128]. Thus, the results in our research prompt further investigation for CBX’s potential action targets. Second, studies have found that CBX with different concentrations of 10, 20, and 100 μM (10 min) caused a dose-dependent attenuation of dye coupling by − 71 ± 4%, − 85 ± 4%, and − 92 ± 2%, respectively, as assessed by the scrape-loading technique [
129]. Another study showed that, compared with untreated cells, 50 μmol/L CBX reduced the strength of homocellular coupling between astrocytes by about 70% using fluorescence recovery after photobleaching (FRAP) method [
90]. Study also showed that EtBr uptake is reduced by Cx knockdown (50.4% inhibition) and CBX treatment (40.8%; 100 μM) [
70]. In our research, dye uptake is reduced by about 80% after CBX treatment, compared with the OGD/R groups, while for dye coupling, no obvious changes observed in those two groups, for OGD/R in the current research also induced attenuation of gap junction communication among astrocytes. Conclusively, the inhibitory effects of CBX on astrocytic hemichannels and GJ were concentration dependent, while different methods for evaluating the channels’ permeabilities probably account for inconsistency of results obtained. In summary, further studies still need to focus on the concentration related-regulation for Cx43 hemichannels and gap junctional permeability. Also, the methods used for evaluating hemichannel activity or gap junction conduction need to be taken into consideration. Otherwise, the mechanism of Panx1 channel inhibition of carbenoxolone has been elucidated and it seems that a mutation in the first extracellular loop reverses its action polarity [
130].
Besides, we assumed that Cx43 may be one of the most important targets of SalB, for we found accidentally that, following CBX pre-treatment of astrocytes for 30 min, SalB exerted more deleterious effect than protection for astrocytes suffering OGD/R injury; in addition, Gap19 or Gap26 exerted similarly results. On the other hand, Gap 19, Gap26, or CBX treatment following SalB pre-treatment showed similar protection with the effect of single-drug application for astrocytes exerted to OGD/R injury (Additional file
1: Figure S1C). One possible explanation for this phenomenon may be derived from the mechanism of drug resistance—cell adhesion-mediated drug resistance (CAM-DR). Fulda et al. reported that glioblastoma multiform (GBM) cells are able to employ CAM-DR by forming spheres via cell–cell interactions. Intriguingly, when inhibiting cell–cell interactions by inhibition of gap junctions through chemical inhibition with carbenoxolone or connexin-mimicking Gap27, GBM cells were sensitized to drug-induced apoptosis. Nevertheless, further research to explore the real truth is needed.