Connexin43 functions as a non-canonical phenotypic stability factor in promoting hybrid Epithelial/Mesenchymal phenotype in glioblastoma cells

Abstract

Glioblastoma (GBM) is a highly malignant and aggressive brain tumor with patients typically experiencing a median survival of 15-18 months after diagnosis. Gap junction protein Connexin43 (Cx43) plays a crucial role in GBM by having both tumor-suppressing and tumor-promoting roles. Here, we identify a critical tumor-promoting role of Cx43 in GBM by functioning as a non-canonical phenotypic stability factor and driving partial EMT, which enhances the acquisition of stemness properties in the cells. Using high-grade mouse astrocytoma cell lines, we found that CT2A cells had higher Cx43 gap junction assembly as compared to KR158 cells. The increased Cx43 assembly in CT2A cells activates the NF-κB signaling pathway, promoting a hybrid E/M phenotype and thereby enhancing self-renewal properties. CT2A cells also exhibited collective cell migration, a characteristic feature of hybrid E/M phenotype, stress resistance, and proliferative properties. To verify Cx43's role in NF-κB pathway activation, DBT and DBT-Erp-29 cells (with higher Cx43 expression) were studied, showing increased NF-κB activation in DBT-Erp-29 cells. Interestingly, KR158 cells formed longer tunneling nanotubes to expedite alternative cellular communication due to reduced gap junctional intercellular communication (GJIC). These results offer valuable insights into targeting Cx43-mediated signaling pathways due to the potential tumor-promoting role of Cx43.

Keywords: Cancer stem cells; Connexin43; Glioblastoma; Hybrid E/M phenotype; Partial EMT.

Fig. 1

Connexin 43 expression in CT2A and KR158 cells. (A-H) CT2A and KR158 cells were immunostained with anti-Cx43 antibody, and Alexa Fluor 568-tagged Phalloidin and counterstained with DAPI. (I-J) The total protein upregulation of Connexin 43 expression in CT2A cells as compared to KR158 cells was further confirmed by Western Blotting. CT2A and KR158 cells were subjected to Triton X solubilization assay, then immunostained with anti-Cx43 antibody and counterstained with DAPI to check the surface expression of Cx43 in the Glioma cells. CT2A cells express more surface expression of Cx43, as shown in (K-Q). Data expressed as the mean ± SD of three independent experiments. The significant differences are indicated by *P < 0.05, * *P < 0.01, * **P < 0.001, * ** *P < 0.0001. The scale bar used represents 50 µm.

Fig. 2

CT2A cells have higher functional GJIC and hemichannel formation. The LY dye was scrape-loaded in CT2A (A) and KR158(C) cells. (B-D) denote the bright-field images of the respective images. cells. It is observed that LY dye uptake and the distance traveled are greater in CT2A cells as compared to KR158 cells. The distance traveled by the dye was quantified in ImageJ software, and absolute values were plotted in a violin plot (E). An EtBr dye uptake assay was performed to determine hemichannel activity, where CT2A cells demonstrated more hemichannel activity than KR158 cells(F-G). The absolute value for intensity was measured in ImageJ software and was plotted in a violin plot (H). The scale bar for the LY dye transfer assay images is 100 µm, and that for all the EtBr uptake assay images is 50 µm. Data are expressed as mean ± SEM of three individual replicates. The significant differences are indicated by *P < 0.05, * *P < 0.01, * **P < 0.001, * ** *P < 0.0001.

Fig. 3

CT2A cells are more proliferative and stress-resistant than KR158 cells. CT2A(A) and KR158(B) cells were seeded in 35 mm dishes with around 500–800 cells per dish and incubated for approximately 7 days. It is then noted that CT2A cells have a higher propensity to form colonies as compared to KR158 cells, as also shown in the quantification graph(C). The glioma cells were subjected to serum starvation stress for 48 h, and then cell viability was measured using MTT assay(D) to check the apoptosis; cells were stained with Propidium Iodide. CT2A (E) demonstrates less staining than KR158 (F) cells. ImageJ software measured The mean fluorescence intensity and plotted it in a Bar graph (G). The significant differences are indicated by *P < 0.05, * *P < 0.01, * **P < 0.001, * ** *P < 0.0001. The scale Bar used for (E-F) is 50 µm.

Fig. 4

Increased Cx43 expression correlates with collective cell migration, a feature of the hybrid E/M phenotype. A scratch assay was performed to compare the rate of cell migration between CT2A(A-D) and KR158(E-H) cells. It is observed that CT2A cells demonstrate Collective cell migration specifically noted in (C-D), which is a characteristic feature of Partial EMT. Images were captured at 0, 12, 24 and 36 h. (I) Results are expressed as the percentage of wound area covered by cells, considering the area measured at 0 h as 0 %. Immunoblotting experiments used anti-occludin, anti-vimentin, and anti-DJ-1 antibodies (J). CT2A cells express higher Vimentin and DJ-1, whereas expression levels of Occludin are comparable to those of KR158 cells(L-K). CT2A cells demonstrate co-expression of Occludin and Vimentin when immunostained with anti-occludin and anti-vimentin antibodies. The images were captured using STED microscope at 60X magnification (N—O). The significant differences are indicated by *P < 0.05, * *P < 0.01, * **P < 0.001, * ** *P < 0.0001. The scale bar used for (A-H) is 100 µm and (N—O) is 10 µm.

Fig. 5

CX43 may promote a phenotypically stable E/M state in glioblastoma by modulating key EMT/MET regulators. (A,B) Scatter plots comparing mRNA expression (log₂ TPM) levels for two gene pairs: GJA1 versus SNAI1 (A) and GJA1 versus GRHL3 (B). In panel A, the Spearman correlation coefficient is –0.19 (p = 0.0131) and the Pearson correlation is –0.23 (p = 3.39×10⁻³), with the linear regression described by y = –0.25x + 4.68 (R² = 0.05). In panel B, Pearson correlation analysis reveals a positive relationship (r = 0.25, p = 9.79×10⁻⁴). Data points are annotated by mutation status, distinguishing cases where neither gene is mutated, only one gene is mutated, or mutation profiling was not available. (C, D) Scatter plots for GJA1 versus NFATC1 (C) and GJA1 versus CD44 (D). For the GJA1–NFATC1 comparison, the Spearman correlation is 0.24 (p = 2.03×10⁻³) and the Pearson correlation is 0.23 (p = 2.82×10⁻³), with the regression line y = 0.14x + 1.17 (R² = 0.05). For GJA1 versus CD44, a stronger association is observed with a Spearman correlation of 0.35 (p ≈ 1.04×10⁻⁵) and a y = 0.29x + 4.94 regression equation. Similar to panels A and B, sample points are grouped by their mutation status (e.g., both genes mutated, a single mutation, or not profiled). (E) Heatmap displaying mRNA expression z-scores (relative to diploid samples, based on U133 microarray data) for a panel of genes implicated in epithelial-mesenchymal transition, MET and cell adhesion, including GJA1, TWIST1, TWIST2, SNAI1, SNAI2, GRHL1, GRHL2, GRHL3, OVOL1, OVOL2, OVOL3, NFATC1, CD44, NFκB1, and RELA. We took results with p < 0.05 as a significant result.

Fig. 6

Partial EMT is not activated by the canonical Wnt-β-catenin signaling pathway. Immunoblotting experiments with anti-GSK-3β and anti-β-catenin antibodies revealed that the Canonical Wnt/β-catenin signaling pathway is not activated in Glioma cells(A-C). The lung epithelial carcinoma (A549) cells were used as a control. The significant differences are indicated by *P < 0.05, * *P < 0.01, * **P < 0.001, * ** *P < 0.0001.

Fig. 7

The NF-κB pathway drives the retention of the hybrid E/M phenotype, followed by the acquisition of stem cell marker expression in GBM cells. To check if CT2A cells demonstrate more stemness properties, the CT2A (A) and KR158 (B) cells were immunostained with an anti-CD44 antibody, which is a prevalent stem cell marker. CT2A cells show higher expression levels of CD44 than KR158 cells. Further quantitative analysis was performed using Immunoblotting(C) followed by graphical representation(D).Immunoblotting experiments with anti-P-65 and anti-IKK2 antibodies demonstrated that the NF-κB pathway is activated more in CT2A as compared to KR158 cells (C, E, and F), which drives the Partial EMT in Glioma cells. The significant differences are indicated by *P < 0.05, * *P < 0.01, * **P < 0.001, * ** *P < 0.0001. The scale Bar used for (A-B) is 25 µm.

Fig. 8

Cx43 trafficking to the cell surface regulated by Erp-29 leads to increased activation of the NF-κB pathway. (A) CT2A cells demonstrate a higher expression of Erp-29 cells than KR158 cells, verified through Western blot analysis tagged with anti-Erp-29 antibody. To check the activation of the NF-κB pathway in cells with increased Cx43 trafficking, immunoblotting was performed in DBT and DBT-Erp-29 cells with anti-P-65 and anti-IKK2 antibodies (B, D, and F). It was observed that DBT-Erp-29 expresses more P-65 and IKK2 as compared to DBT cells, owing to their difference in the formation of gap junction channels at the cell surface. The significant differences are indicated by *P < 0.05, * *P < 0.01, * **P < 0.001, * ** *P < 0.0001.

Fig. 9

Alternative Intercellular communication is facilitated by tunneling nanotube formation. CT2A (A-D) and KR158 (E-H) cells were immunostained with anti-Cx43 antibody, Alexa Fluor 568-tagged Phalloidin, and counterstained with DAPI and imaged with an Olympus IX81 microscope. In our study, we observed that reduced GJIC leads to the formation of longer tunneling nanotubules to facilitate intercellular communication (A-H). Thus, KR158 cells form larger nanotubes than CT2A cells due to reduced GJIC formation. (I-K) To further elucidate this idea, STED microscopic images were captured at 60X magnification, and it was observed that KR158 cells form larger tunneling nanotubes as compared to CT2A cells. The significant differences are indicated by *P < 0.05, * *P < 0.01, * **P < 0.001, * ** *P < 0.0001. The scale Bar used for (A-H) is 50 µm, and for (I and J) is 10 µm.