Mechanism of antrocytic connexin-43 autophagy degradation in oxygen-glucose deprivation and its effect on neuroinflammation

Background : Connexin 43 (Cx43) are the most widely distributed gap junction proteins in the nervous system. Cx43 enables cell-to-cell communication and plays an important role in ion transport, substrate exchange and delivery of information , which have been implicated in cerebral ischemia injury. Our previous work revealed the relationships between Cx43 and glia-mediated neuroinflammation through the release of ATP in oxygen-glucose deprivation (OGD), which means degradation of Cx43 may improve neuroinflammatory damage during OGD injury . However, the roles of Cx43 degradation and neuroinflammation caused by OGD remain unclear. Methods: We used primary cultured astrocytes treated with OGD as an in vitro model of cerebral ischemia injury and we used middle cerebral artery occlusion (MCAO) model as an in vivo model of cerebral ischemia. HeLa cells were used in overexpression experiments. Cx43 protein levels were determined by western blotting. The interaction between Cx43 and related autophagy receptors was determined by co-immunoprecipitation and immunofluorescence. The gene knockdown (KD) of ATG5, OPTN, NDP52, PINK1 and Cx43 was applied by siRNA transfection. Related cytokines were detected by cytometric bead assay. Results: We found that Cx43 protein levels increased after ischemia in gene KD of ATG5, OPTN, NDP52 and PINK1 primary astrocytes. The interaction of Cx43 with OPTN, NDP52 and PINK1 was increased after cerebral ischemia injury in vitro and vivo. While the interaction was weakened after point mutation of Cx43 at Ser368, Tyr265 and Tyr247. Meanwhile, IL-10 upregulated during OGD after KD of ATG5, OPTN, NDP52 and PINK1 in astrocytes , while TNF downregulated during OGD after KD of ATG5, OPTN, NDP52 and PINK1 in astrocytes. Conclusions: Our results suggest that degradation of Cx43 is caused by selective autophagy during ischemia injury and the autophagy degradation of Cx43 plays important roles in neuroinflammation mediated by OGD injury. Treatment targeting Cx43 degradation pathway can improve neuroinflammation responses induced by OGD injury , which provide novel therapeutic strategies and crosstalk between autophagy and neuroinflammation. of Cx43. In this study, identify the


Western blotting
Cell lysates were retrieved using RIPA lysis buffer (Abcam, ab156034) supplemented with Protease Inhibitor Cocktail (Thermo Scientific, A32955) and PhosSTOP (Roche,4906845001). Then cells were sonicated on ice and centrifuged at 4°C at 14000 g for 20 min, followed by a BCA assay (Thermo Fisher Scientific, 23225) for protein concentrations. Next All cell lysates were boiled with 4× LDS sample buffer (Invitrogen, NP0007), 30-40 μg of total proteins were run out on a 4-12% Sure PAGE Bis-Tris gel (Genscript, China, M00654) and transferred to PVDF membranes (Thermo Scientific,88585). Membranes were probed with the indicated primary antibodies overnight at 4°C, followed by the appropriate HRP-conjugated secondary antibodies for 2h at room temperature. The blots were imaged on the ChemiDoc developer system (Bio-Rad). All band detection was in the linear range. Related Information was provided in in Supplementary Figure 2.

Co-immunoprecipitation (Co-IP)
The cell lysates were extracted from the treated cells using cell lysis buffer (Cell Signaling Technology, 9803S) supplemented with Protease Inhibitor Cocktail (Thermo Scientific, A32955) and PhosSTOP (Roche, 4906845001), followed by a BCA assay (Thermo Fisher Scientific, 23225) for protein concentrations at 1mg/ml and incubated with indicated primary antibodies overnight at 4°C with a constant shaking speed. Then, the complexes were mixed with Protein G Agarose (Roche, 11243233001) and shaken for 3 h at 4°C to capture the antigen-antibody mixture.
The beads were then washed five times with cell lysis buffer and boiled in SDS loading buffer. The eluted proteins were analyzed by western blotting.

Immunofluorescence
Slide-cultured cells were treated as indicated in the figure legends. After treatment, the cells were fixed with 4% paraformaldehyde at room temperature for 20 min and washed with PBS for 5 min. Then, the cells were permeabilized with 0.1% Triton X-100 and blocked using 3% goat serum in PBS for 40 min. Next, the cells were incubated with primary antibodies (as indicated in the figure legends) diluted in 3% goat serum overnight at 4°C, and then washed with PBS and incubated with antimouse Alexa-Fluor-488-conjugated secondary antibodies and anti-rabbit Dylight-633-conjugated secondary antibodies for 1 h at room temperature. The cells were then washed thrice for 5 min each with 1% Triton X-100 in PBS. During the final wash step, they were incubated with DAPI (Solarbio, C0056) in PBS for 5 min. We used the Leica TCS SP5 confocal microscope and LSM 510 Zeiss confocal microscope for observing the immunofluorescence results.

Statistics
Experiments were not randomized. All statistical data were calculated and graphed using GraphPad Prism6. All data are presented as means ± SD. Statistical differences were detected using a two-tailed Student's t-test. A p-values less than 0.05 was considered statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001.

Results
The degradation of Cx43 is caused by autophagy after OGD Firstly, we found that the Cx43 protein levels decreased after OGD (Fig. 1a). We then wanted to confirm that this reduction was caused by autophagy. We first treated astrocytes with baf A1, an autophagy inhibitor, for 3 hours before and after OGD. Cx43 and P-Cx43 was found to have accumulated (Fig. 1b). Cx43 and P-Cx43 protein levels increased in Atg5 KD astrocytes after OGD for 6 hours (Fig. 1a). We also investigated the immunofluorescence of Cx43 and LC3B in primary astrocytes, which was found that overlay of two proteins increased after OGD (Fig. 1e). These results prove that the degradation of Cx43 is caused by autophagy after OGD.
OPTN and NDP52 are the autophagy receptors required for the autophagy of Cx43 in OGD To ascertain whether OPTN and NDP52 were associated with the autophagy of Cx43, we chose siRNAs for their KDs. As shown in Fig. 2a, Cx43 and P-Cx43 protein levels were increased in OPTN KD astrocytes and NDP52 KD astrocytes after 6 hours OGD.
To evaluate the relationship between Cx43 and OPTN and Cx43 and NDP52 at the spatial level, we used immunofluorescence analysis. The overlay of Cx43 and OPTN ( Fig. 2c) and Cx43 and NDP52 (Fig. 2c) were found to increase in the primary astrocytes after 3 hours of OGD. Verifying the interaction between proteins requires Co-IP analysis. As shown in Fig. 5a, the protein levels of OPTN and NDP52 that were combined with Cx43 increased after OA treatment, which means that the interactions between Cx43 and OPTN, as well as Cx43 and NDP52, had increased.
Also, we overexpressed RFP-OPTN and RFP-NDP52 in the HeLa cells. The interactions between Cx43 and RFP-OPTN, as well as Cx43 and RFP-NDP52, had increased after OGD (Fig. 3a), meanwhile, overlay of Cx43 and RFP also increased (Fig. 3b).
Collectively, these findings suggest that OPTN and NDP52 are the autophagy receptors for the autophagy of Cx43 in OGD.
PINK1 is the molecule involved upstream of the autophagy of Cx43 before the connections with NDP52 and OPTN in OGD As shown in Fig. 4a, Cx43 protein levels were increased in PINK1 KD astrocytes after OGD, demonstrating that PINK1 was related to Cx43 degradation. We used Co-IP to detect binding between Cx43 and PINK1. We found that the PINK1 content bound to Cx43 had increased after OA treatment (Fig. 5a). Besides, overlay of Cx43 and PINK1 in primary astrocytes was found to had increased after OGD (Fig. 4c). These findings suggest PINK1 is associated with the autophagy of Cx43.
Next, we find that protein levels of OPTN and NDP52 that were combined with Cx43 disappeared after PINK1 KD in OGD (Fig. 5a). Meanwhile, the overlay between Cx43, RFP-OPTN and PINK1-Myc increased after OGD, the same as the overlay between Cx43, RFP-NDP52 and PINK1-Myc (Fig. 5b). All the above results prove that PINK1 is the molecule involved upstream of the autophagy of Cx43 before the connections with NDP52and OPTN under OGD.
The autophagy of Cx43 is related to ubiquitination in OGD, which is composed of pUb(S65) Protein degradation requires ubiquitination, as does Cx43. The ubiquitin kinase PINK1 is closely related to the autophagy of Cx43. In this study, identify the ubiquitination used in the autophagy degradation of Cx43, we overexpressed HA-Ubquitin in Hela cells, which were then treated with OGD. We found that amounts of ubiquitin bound to Cx43 were increased (Fig. 6a), As previously stated, the ubiquitin phosphorylation site was located at Ser65, which was phosphorylated by PINK1. Therefore, we next investigated the role of pUb (S65) in the autophagy of Cx43 in OGD. Co-IP showed that, after OGD, pUb(S65) modified Cx43 increased in HA-Ubquitin overexpressed Hela cells (Fig. 6a). Modification by pUb(S65) in the total protein content had increased in all groups after OGD ( Fig. 6c). Altogether, these findings suggested that the ubiquitination of Cx43 consisted of pUb(S65) induced Cx43 autophagy.

The phosphorylation of Cx43 was involved in its autophagy
Changes in Cx43 ubiquitination revealed that some modifications to Cx43 must be occurring after OGD. Therefore, we chose phospho-Cx43(Ser368) as our target.
PKC and Src kinases are the start for Cx43 autophagy As mentioned above, Cx43 is phosphorylated by PKC and Src kinases at Ser368, Tyr265 and Tyr247 in OGD. We mutated the phosphorylation site of Cx43 to study the role of kinases. There is no doubt that NDP52, OPTN, PINK1, pUb(S65) bound to GFP-Cx43 had increased after OA treatment, but these connections decreased in the overexpressed GFP-Cx43(368A) Hela cells or overexpressed GFP-Cx43(247A + 265A) Hela cells (Fig. 7a). These findings suggest that PKC and Src kinases are the start of Interestingly, TNF showed no significant difference in KD of ATG5, OPTN, NDP52 and PINK1 astrocytes after OGD, and IL-10 increased in these groups after OGD. IFN-γ still increased and IL-6 decreased more (Fig. 8). These data suggest that blocking Cx43 autophagy degradation pathway can reduced pro-inflammatory cytokine concentrations (TNF and IL-6) and increased anti-inflammatory cytokine concentrations (IL-10) in OGD.

Discussion
Cx43 plays a vital role in maintaining the homeostasis of the nervous system.
Previous studies have confirmed that Cx43 is dysfunctional under hypoxic conditions in the brain. Cx43 increases the release of excitatory glutamate, exacerbating the damage after ischemia [26,27] . Therefore, attention has been paid towards finding methods that can inhibit the function of Cx43, such as the addition of the Cx43 inhibitor CBX [28] or the Cx43-specific inhibitory peptide Gap19 [29,30] . However, only a small number of studies focus on changes in the amounts of Cx43 rather than channel permeability. Interestingly, some studies have reported that the degradation of Cx43 is achieved by selective autophagy under normal conditions. A related autophagy receptor that has been identified is p62 [8] . Therefore, in this study, our hypothesis was that the brain has a targeting mechanism to clear Cx43 under ischemia conditions, which is selective autophagy.
Some previous research has shown that Cx43 increased after treatment with proteasomal inhibitors [31] , while other studies have described the genetic silencing of autophagy-associated genes resulted in accumulation of Cx43 and reduced colocalization of LC3B with Cx43 [32] . Simultaneously, decrease in Cx43 caused by starvation was sensitive to lysosomal inhibitors [32,33] . All above results predict both proteasomal pathway and autophagy can cause degradation of Cx43, but that the latter is more likely. Similarly, we demonstrated that the degradation of Cx43 under hypoxic conditions was achieved by autophagy. All these results predict that the degradation of Cx43 mainly depends on autophagy.
Since Cx43 degradation occurs via its autophagy under hypoxic conditions, the relevant autophagy receptors must play an important role. Previous literature has reported that OPTN and NDP52 are a pair of mitochondrial autophagy-related receptors [10] , while mitochondria contain Cx43 [11] , which is sensitive to ischemia. Therefore, the autophagy degradation of Cx43 may take a similar pathway.
Unexpectedly, we show that NDP52 and OPTN are related to the autophagy receptors of Cx43 degradation. NDP52 and OPTN have UBDs as autophagy receptors [34] , which means the ubiquitination of Cx43 can happen before its connection with NDP52 and OPTN. The relationship between Cx43 degradation and ubiquitination has been reported in many studies. For example, E3 ubiquitin ligase Nedd4 interacts with the carboxyl terminus of Cx43 via its WW domains and ubiquitinates Cx43 [19,35] . Several E3 ubiquitin ligases are involved in Cx43 ubiquitination, including Wwp1 [36] , Smurf2 [37] , and TRIM21 [38] . Meanwhile, AMSHmediated deubiquitination of Cx43 regulates Cx43 degradation [39] . These studies showed that Cx43 ubiquitination was closely related to its degradation, and it was necessary to identify the ubiquitin kinase or ubiquitin ligase that involved. Recently, some studies have shown that the ubiquitin kinase PINK1 phosphorylates ubiquitin at Ser65 to activate mitophagy [23,40] , meaning that pUb(S65) may be involved in the degradation of Cx43. However, we are not clear about the relationship between Cx43 ubiquitination and its autophagy degradation in cerebral ischemia. We have shown that PINK1 is the ubiquitin kinase involved in this process and pUb(S65) modifies Cx43 in OGD injuries. In brief, Cx43 was phosphorylated in response to ischemia. Meanwhile, ubiquitin was phosphorylated by PINK1, and p-Cx43 was then modified with pS65-UB. Thus, this study confirmed that a modified form of ubiquitin, pS65-UB, modified Cx43.
Many studies have demonstrated that the phosphorylation of proteins affects their downstream autophagy. The phosphorylation of mAbg9 was shown to affect the molecular mechanism underlying its autophagic transport [41] . Another study showed that PTEN phosphorylation promotes the nuclear translocation and autophagy of PTEN in response to treatment with DNA-damaging agents in cancer cells [42] . Phosphorylation is also related to the degradation of Cx43 [17] . Similarly, the phosphorylation of Cx43 may be associated with its autophagy. We chose the common Ser368, Tyr265 and Tyr247 sites of Cx43 according to previous studies [43,44] . After repeating the above experiments, it was found that the phosphorylation of Cx43 is closely related to its autophagy. When giving the above site a point mutation, we observed that the connection between cx43 and OPTN, NDP52 and PINK1 is weakened, simultaneously, the number of pS65-UB modified Cx43 also reduced. We show that phosphorylation of Cx43 is the starting point of its autophagy degradation. IL-10, as an anti-inflammatory cytokine, has been shown to play an antiinflammatory protective role in cerebral ischemia [45,46] , it enhances resistance to cerebral ischemia and reduces ischemic damage. Different from IL-10, TNF-ɑ is a pro-inflammatory cytokine, as same as IL-6. A study showed that enhanced autophagy by rapamycin can upregulate IL-6 expression in mice astrocytes [47] .
More interestingly, some links between Cx43 autophagy degradation and neuroinflammation were found in our work. Blocking Cx43 autophagy degradation pathway can reduced pro-inflammatory cytokine concentrations (TNF and IL-6) and increased anti-inflammatory cytokine concentrations (IL-10) in OGD. The above results reveal the relationship between autophagy degradation of Cx43 and astrocyte-mediated neuroinflammation. However, how autophagy degradation of Cx43 affects the production of TNF-ɑ, IL-6 and IL-10 in cerebral ischemia remains unclear.

Study limitations
This study has some limitations. Firstly, the phosphorylation site of Cx43 is not just the above three sites during OGD. Roles of MAPK and PKC in Cx43 autophagy degradation during OGD remains unknown. Furthermore, Cx43 autophagy degradation pathway in OGD is similar to mitophagy. The relationships between Cx43 and mitophagy need deeper research. Second, according to our results, Cx43 may be an anti-inflammation molecule. How Cx43 affects cytokines expression remains unclear. Vivo studies may provide more evidences.

Conclusions
In conclusion, we showed that under OGD, Cx43 is phosphorylated by PKC and Src kinases. Then phosphorylated Cx43 combines with pUb(S65) phosphorylated by PINK1. These modified Cx43 recruits the autophagy receptors OPTN and NDP52 for its autophagy-based degradation. Preventing the degradation of Cx43 autophagy is beneficial to the transformation of astrocytes to anti-inflammatory in cerebral ischemia. This study offers new ideas and intervention methods for Cx43-targeted therapy, bringing more crosstalk between autophagy and neuroinflammation.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests. Figure 1 The degradation of Cx43 is caused by autophagy during OGD in primary astrocytes. a. Atg The interaction between Cx43 and RFP-NDP52 increased in HeLa cells during OGD, as well   The autophagy of Cx43 depends on ubiquitination during OGD, which is composed of pUb(     The interaction between Cx43 and RFP-NDP52 increased in HeLa cells during OGD, as well as The autophagy of Cx43 depends on ubiquitination during OGD, which is composed of pUb(S65