Nuclear translocation of Cx43 promotes to CRC progression and associates with β-catenin accumulation.

Introduction
Colorectal cancer (CRC) is one of the most common malignancies of the digestive system and has the third highest incidence and mortality rate globally.1 Metastasis is a major cause of high mortality in patients with CRC.2 In some patients, distant metastases to the liver, lungs, peritoneum, and other organs are common.3 Metastatic CRC is associated with a poor prognosis, and existing treatments can appropriately prolong survival; however, a cure remains uncommon.4 Therefore, a deeper understanding of the mechanisms underlying CRC metastasis is crucial for its effective management.
The tumour microenvironment (TME) refers to the environment of the tumours or cancer stem cells, which includes surrounding immune cells, blood vessels, extracellular matrix (ECM), fibroblasts, lymphocytes, and signalling molecules.5 Fibroblasts are the main components of the TME, and they primarily release growth factors for regenerative repair.6 Previous research has indicated that the TME can promote the development of various tumour stages and plays a key role in the pathogenesis of cancer.7 Normal fibroblasts (NFs) inhibit malignant tumour metastasis.8 Cancer-associated fibroblasts (CAFs) are important components of TME. Activated CAFs can promote tumour growth, angiogenesis, invasion, and metastasis, as well as ECM remodelling and chemoresistance.9,10 Similarly, CAFs can be induced by C-X-C motif chemokine ligand 12 (CXCL12) and TGF-β and play a role in angiogenesis, invasion, and metastasis in CRC.11 Accordingly, CAFs can profoundly affect intercellular communication and invasive abilities of cancer cells.
The Wnt signalling pathway is involved in the progression of CRC. Depending on whether β-catenin is activated, the Wnt signalling pathway can be classified as a canonical and non-canonical pathways.12 During CRC progression, the activated Wnt signalling pathway promotes proliferation and metastasis. The core protein molecule β-catenin mediates signalling and is involved in the activation of downstream target gene transcription through the formation of transcription complexes in the nucleus. A recent report demonstrated a close association between the Wnt signalling pathway and connexin. Wnt signalling can activate the intracellular mobilisation of Connexin43 (Cx43). Wnt5a was reported to reduce membrane and cytoplasmic localisation, whereas it increases nuclear localisation of Cx43.13 Although previous reports have illustrated the Wnt signalling pathway, several studies have highlighted the need for further exploration.
Cx43 is a classical protein belonging to the gap-junction family encoded by GJA1, which mainly functions as a gap junction. However, growing evidence reflects the additional physiological role of Cx43. This process appears to be dysregulated during CRC progression.14 Abnormal Cx43 expression may lead to cancer metastasis and a poor prognosis.15 However, Cx43 exhibits distinct promoting or inhibiting effects on tumour progression. Some reports indicate that Cx43 inhibits tumour progression,16 while others present the opposite view.17,18 Recently, an increasing number of studies have suggested a procarcinogenic role for Cx43, which may be associated with its phosphorylation.19,20 The non-gap junction function of Cx43 is associated with CRC progression; however, the underlying molecular mechanisms remain unknown.
In this study, we systematically identified the role of Cx43 in facilitating CRC progression. Cx43 exhibits nuclear translocation and carries β-catenin into the nucleus. Furthermore, Cx43 expression was elevated during the interaction between CRC cells and fibroblasts, which may be mediated by TGF-β stimulation. Therefore, nuclear translocation of Cx43 regulates the promotion of CRC progression via phosphorylation.

Methods

Cell culture
SW620 (RRID: CVCL_0547), SW480 (RRID: CVCL_0546), and RKO (RRID: CVCL_0504) cells were purchased from the ATCC. Human embryonic lung fibroblasts (HELF) were purchased from Kaiji (Nanjing, China). Human umbilical vein endothelial cells (HUVEC) (RRID: CVCL_2959) and human aortic vascular smooth muscle cells (HAVSMC) were purchased from Procell Life Science&Technology (Wuhan, China). The three main cell lines used in this study (SW620, SW480, and RKO) have been authenticated using short tandem repeat (STR) profiling within the last 3 y. STR Profiling was performed according to the ISO 9001:2008 and ISO/IEC 17025:2005 quality standards (Genetic Testing Biotechnology, China). All the experiments were conducted using mycoplasma-free cells. SW620 and SW480 cells were cultured in RPMI 1640 (Hyclone) supplemented with 10% foetal bovine serum (Hyclone) and 1% penicillin and streptomycin (Biosharp). RKO cells,HELFs, HUVECs, and HAVSMCs were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Hyclone) supplemented with 10% foetal bovine serum (Hyclone) and 1% penicillin and streptomycin (Biosharp). Cells were maintained at 37 °C in a humidified atmosphere containing 5% CO2.

Primary fibroblasts culture
To obtain normal fibroblasts (NFs) from noncancerous colorectal tissues, we isolated and cultured stromal fibroblasts from normal colorectal tissues of patients with CRC, and cancer-associated fibroblasts (CAFs) were isolated from primary CRC tissues of the same patient. Primary fibroblasts were isolated with permission from all patients. NFs were obtained from the surgical margins (5 cm away from the tumour margins), and CAFs were obtained from the centre of the tumour tissue. The excised tissues were stored in sterile PBS buffer containing 400 U/mL penicillin/streptomycin on ice. After washing five times, the tissues were cut into small pieces of approximately 1mm3 using sterile scalpels and scissors and then adhered and cultured on the bottom of tissue culture flasks (added 2 mL DMEM supplemented with 15% foetal bovine serum and 100 U/mL penicillin/streptomycin). After 2 weeks of culture, fibroblasts were passaged and purified using the differential adhesion method. Primary fibroblasts used in this study were retained in the third to tenth passage. Fibroblasts in the microenvironment must be distinguished from endothelial cells and smooth muscle cells. CD31 is a marker for endothelial cells and caldesmon is a marker for smooth muscle cells. α-SMA is a fibroblast marker

Double layer co-culture of tumour cells and fibroblasts
To assess the impact of fibroblasts on CRC cells under co-culture conditions, we used a Transwell co-culture system. Fibroblasts were seeded in the lower chamber of 6-well plates, while CRC cells were seeded onto a 0.4μm transwell upper chamber (Corning). The fibroblasts used in this study were NFs, CAFs, and HELFs. After adherence, replaced 1 mL of fresh culture medium was added to the upper chamber and 2 mL of fresh culture medium was added to the lower chamber. The cells were collected after the desired time point was reached. Replaced 1 mL of fresh culture medium in both the upper and lower chambers to ensure that the cells had sufficient nutrition. If the culture time was not indicated, all cells were cultured for 2 d.

Human tissue material
From January 2008 to February 2016, 226 primary colorectal adenocarcinoma and paraneoplastic adenoma specimens were collected from the Ningbo Yinzhou No.2 Hospital, and the distal margins (normal intestinal mucosa) of the 226 specimens were selected as controls. These samples were obtained from 226 patients, who underwent surgical resection without prior exposure to chemotherapy or radiation therapy. Samples were collected from the tissue bank of our laboratory for paraffin sectioning and immunohistochemistry. The samples selected had a central district, intraepithelial neoplasia around cancer, and normal-appearing colonic mucosa of the incisal edge for the matched samples, confirmed by two experienced colon pathologists.

Judging Criteria in IHC
Determination of paracancerous adenoma tissue: Heterogeneous hyperplastic epithelial cells often exist between colorectal carcinoma and normal intestinal mucosa, characterised histologically by large and deeply stained nuclei, varying degrees of nuclear sarcoidosis, multilayers, and lack of polarisation. Depending on the degree of complexity of the glandular structure and the degree of nuclear heterogeneity, heteroplasia may be classified as low- or high-grade, and in our experiments, we classified it as intestinal adenoma tissues.
IHC scoring: Given the uneven distribution of markers in the tumour, the three most positive regions for each marker were counted and the average value was taken as the final score. The scoring criteria for colour intensity were as follows: 0 points for no staining, 1 point for pale yellow colour, 2 points for yellow or deep yellow colour, and 3 points for yellow-brown colour. The range of positive cells was scored as follows: 0 points for no positive cells, 1 point for 1%–25%, 2 points for 26%–50%, 3 points for 51%–75%, and 4 points for more than 75%. The final score is the product of these two scores. A score of less than 3 was negative and a score of more than 3 was positive. A score of 4 or 5 was considered as “ + ”, 6 to 8 as “ + + ”, and more than 9 as “ + + + .” Each sample was confirmed by two experienced pathologists.

Immunoblotting
Proteins were separated by SDS-PAGE, transferred onto a nitrocellulose membrane (BioRad), and probed with primary antibodies followed by horseradish peroxidase (HRP)-linked secondary antibodies. The primary antibodies used for western blotting were as follows: Connexin 43 (Cell Signalling Technology, 1:1000), GAPDH (Hangzhou Lianke Biology Co., Ltd, 1:10000), β-actin (Hangzhou Lianke Biology Co., Ltd, 1:10000), Lamin B1 (Abcam, 1:10000), β-catenin (Cell Signalling Technology, 1:1000), E-cadherin (Cell Signalling Technology, 1:1000), N-cadherin (Cell Signalling Technology, 1:1000), vimentin (Cell Signalling Technology, 1:1000), fibronectin (Cell Signalling Technology, 1:1000), phospho-β-catenin (S552) (Cell Signalling Technology, 1:1000), Wnt5a (Cell Signalling Technology, 1:1000), Dishevelled Segment Polarity Protein 3 (DVL3) (Cell Signalling Technology, 1:1000), TCF4 (Cell Signalling Technology, 1:1000), α-SMA (Sigma, 1:100), Flag (Sigma, 1:1000), and Phospho-(Ser/Thr) Phe (Abcam, 1:1000). The secondary antibodies used for western blotting were horseradish peroxidase (HRP)-linked secondary antibodies (Cell Signalling Technology, 1:5000). We performed intensity analysis with ImageJ. The relative intensity of each target protein was normalised to the intensity of the loading control. The normalised values for the experimental groups were expressed relative to those of the control group, setting target protein of the control group value to 1.

Immunofluorescence
Cells were placed in 6-well plates with glass coverslips and cultured according to double-layer co-culture. Cells were fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X−100, and blocked with 10% BSA/PBS. Samples were counterstained with 4’, 6-Diamidino−2-phenylindole dihydrochloride (DAPI) before being mounted onto slides for visualisation. The primary antibodies used for immunostaining were caldesmon (Cell Signalling Technology, 1:1000), CD31 (Cell Signalling Technology, 1:1000), α-SMA (Sigma, 1:100), connexin 43 (Cell Signalling Technology, 1:1000) (Proteintech, 1:1000), and β-catenin (Cell Signalling Technology, 1:1000). The secondary antibodies used for immunostaining were goat anti-rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488 (Invitrogen, 1:500), and goat anti-rabbit IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 546 (Invitrogen, 1:300).

Immunohistochemistry
Sections (4 μm thick) were processed for immunohistochemistry (IHC). Histological specimens were fixed with 10% neutral formalin, carefully sampled, embedded in conventional paraffin, sliced, and stained using the EnVision method. The primary antibodies used were Cx43 (Cell Signalling Technology, 1:100), Ki67 (Fuzhou Maxim), and β-catenin (Fuzhou Maxim), while the secondary antibodies and DAB staining were purchased from Fuzhou Maxim Company. Known positive sections were used as positive controls and PBS was used as a negative control instead of the primary antibody. The experimental procedures were performed in accordance with the manufacturer’s instructions.

Immunoprecipitation
The cells were lysed in RIPA Lysis Buffer (Beyotime). Immunoprecipitation was performed using Protein G Agarose (Beyotime), according to the manufacturer's instructions. Added 0.2 to 2 μg of primary antibody to each lysate and incubated overnight at 4 °C with constant rotation. The next day, 20 μL of protein G Sepharose beads were added to each lysate for 1 to 3 h at 4 °C with constant rotation. Unbound proteins were removed by five washes using lysis buffer or PBS, and proteins were eluted from the beads by the addition of 20–40μL of sample buffer. Immunoblotting (IB) assays were used to analyse the samples.

Fluorescence recovery after photobleaching
Fluorescence recovery after photobleaching (FRAP) experiments were performed using confocal microscopy. The cells were maintained at 37 °C in an air-stream stage incubator. Cells were stained with 10 μg/mL 5(6)-CFDA diluted in PBS and incubated in the dark for 15 min. Following incubation, the cells were washed three times with PBS to remove excess dye. Bleaching was performed with a circular spot using the 488 nm and 514 nm lines from a 40mW argon laser operating at 75% laser power. A single iteration was used for the bleach pulse, which lasted 0.8 to 40 ms depending on the bleach spot size. Fluorescence recovery was monitored at low laser intensity (0.2% of a 40mW laser) at 0.8 to 40 ms intervals, depending on the experiment.

Measurement of cell migration and invasion
Transwell chambers were used to measure the migration and invasion abilities of the CRC cells. Briefly, 1 × 105 cells in serum-free RPMI 1640 or DMEM were added to the upper chamber of the Transwell plates. The upper side of the chamber membrane was coated with Matrigel (for invasion) or without Matrigel (for migration). Complete RPMI 1640 or DMEM supplemented with 10% foetal bovine serum was added to the lower chamber of the Transwell plates. After culturing for a suitable period of time, the cells that migrated or invaded the bottom side of the membrane were fixed with paraformaldehyde and stained with crystal violet (Shanghai Biyuntian Biotechnology Co., Ltd) for 5 to 10 min. Cells were counted and photographed under a microscope. Each measurement was performed in triplicate and the experiments were repeated three times.

Wound healing assay
Cells transfected with various constructs were grown in six-well plates until they reached confluence. Wound injury was induced with the tip of a sterile micropipette, and detached cells were removed by washing with PBS. The cells were then incubated with serum-free medium and allowed to migrate for 48 h. Photographs were taken at 0 h, 24 h and 48 h. Each measurement was performed in triplicate and the experiments were repeated three times.

Detection of intracellular calcium content
SW620 cells were added to Chromogenic Reagent and Calcium Assay Buffer and mixed gently. The reaction was incubated for 5–10 min at room temperature in the dark. The absorbance of the cells was measured at a wavelength of 575 nm.

Scrape loading dye transfer assay
SW480 and SW620 cells and HELFs were cultured to density saturation and soaked in PBS with added dyeing agents (Lucifer yellow). The cell surface was lightly scratched using a scalpel and incubated for 5 min at room temperature. The cells were then washed with dye and imaged.

Subcutaneous tumour formation experiment in mice
Female nude mice (4-weeks-old) were purchased from the Shanghai Slake Company. The mice were placed in controlled environment (12 h light/dark cycle; 20−26℃; 50% humidity). For the tumorigenesis assay, the mice were randomly divided into three groups using a random number table, each group having five mice. The three groups were HELFs, SW620 cells and SW620 cells along with HELFs (9:1). The mice were subcutaneously inoculated with a tumour cell suspension and the total number of cells in all the groups was 1 × 106. Tumour growth was monitored every 2−3 d by measuring the tumour dimensions with a caliper. The mice were sacrificed when the size of the tumours was between 10 and 15 mm, photographed, and the weights of the tumours were recorded. The mice were euthanized with an overdose of pentobarbital sodium. Treatment of mice were not blinded as they were performed by one single investigator; however, results were analysed in blinded way. All animal experiments were performed in accordance with the protocol approved by the Institutional Animal Care and Use Committee of Zhejiang University.

Stable transfection
For stable overexpression of Cx43, the full-length or mutated CDS fragment of Cx43 was cloned from human cDNA by PCR and linked to pCDNA3.1 (Addgene Cambridge) with a Flag-tag at the N-terminus. For shRNA of Cx43, shRNA was designed and linked to pLKO.1 (Addgene Cambridge). The in vitro transfection reagent LipoD293TM was purchased from SignaGen Laboratories (Rockville, MD, USA). Cells were plated one day before transfection into 6-well plates. Cells were transfected with pCDNA3.1 or pLKO.1 were screened using G418 or puromycin (4 μg/mL). The sequences of the oligonucleotides used are listed in Supplemental Table 3.

Nuclear and cytoplasm fractionation
Nuclear and cytoplasmic fractions were obtained using a cytoplasmic and nuclear protein extraction kit (KeyGEN Biotech), according to the manufacturer’s instructions. Briefly, special reagents were added to the cells and centrifuged at 3000 rpm, resulting in a final upper layer of cytoplasmic proteins and a lower layer of nuclear proteins. Subsequently, other reagents were added and centrifuged at 12000 rpm. The supernatants were stored as nuclear extracts.

RNA extraction and real-time fluorescence quantitative PCR
RNA was extracted from cells in good condition using TRIzol (Thermo Fisher Scientific) according to standard TRIzol protocols. A Reverse Transcription kit (Takara) was used to reverse-transcribe 1000 ng of RNA to cDNA. RT-PCR experiments were performed according to the instructions of the Takara Fluorescence Quantitative Kit.

Dual-luciferase reporter activity detection
Cells were plated into 24-well plates until they adhered to the wall, and the density reached 80%. TopFlash, FopFlash, and pRL-TK (Beyotime Biotech) were used to detect activation of the Wnt signalling pathway. The plasmid in serum-free DMEM (TOP/FOP: pRL-TK = 50:1 in the plasmid) was mixed with an appropriate amount of Lipofectamine 2000 in serum-free DMEM. After 5 min of stewing, 100 μL of the transfection mixture was added to each well. The activity of the fluorescent plum reporter gene was detected 48 h after transfection. Cells in each well plate were lysed with 100 μL of lysis solution PLB after washing with PBS. The cell lysate was centrifuged, de-precipitated after shaking, and incubated for 15 min. Activity was detected using the Chemiluminescence Instrument Berthold Luminor LB 9507 (Berthold Company). Detection was performed using the Dual Luciferase Reporter Assay System (Promega, Madison, WI, USA). Each measurement was performed in triplicate and the experiments were repeated three times.

RNA-seq
PolyA RNA was extracted, sequenced, and analysed using RiboBio. Detailed information is provided in Supplementary Table 4. The sequencing coverage and quality statistics for each sample are summarised in Supplementary Table S3. The cells were set up into three replicate samples and plated into a 6-well plate. A paired-end reading was performed. The platform used was Illumina HiSeq X Ten. Differential gene expression analysis was performed using Cuffdiff. Benjamini–Hochberg false discovery rate method was used to verify multiple hypothesis testing. Differentially expressed genes were chosen as candidates for further analysis, with P < 0.05 and fold change >1.5 or <0.67. The clean Data were compared with the Homo sapiens genome. Gene set enrichment analysis was performed using GSEA. Enriched pathways within the network were identified using ClueGO.

Bioinformatics analyses
Five-year survival and prognosis data of tumour patients were obtained from the The Cancer Genome Atlas (TCGA) and GEO databases, including age, sex, target protein expression, and recurrence status of tumour patients. Data analysis was performed using statistical methods.

Statistical analyses
Statistical analyses were performed using GraphPad Prism6.0 or SPSS19.0. The χ2 test and Fisher’s exact test were used to compare the rates among all groups. Stata15 was used to plot the forest. Spearman’s test was used to analyse the correlation between Cx43, Ki67, and β-catenin. The Breslow test was used for survival analysis. Discrepancy analysis data between groups were analysed using t-tests, one-way analysis of variance (ANOVA), or two-way ANOVA. Statistical significance was set at *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Results

Fibroblasts can influence the epithelial-mesenchymal transition (EMT) process in CRC cells through the Wnt signalling pathway
NFs and CAFs were successfully isolated from primary cultures. The morphologies of the NFs and CAFs were analysed (Figure 1A). Immunofluorescence (IF) assays of vascular markers (CD31 and Caldesmon) and a mesenchymal marker (α-smooth muscle actin (α-SMA)) were performed in NFs and CAFs (Figure 1B-1D). Immunoblotting (IB) assays for E-cadherin (E-cad) and vimentin (Vim) demonstrated that both NFs and CAFs were mesenchymal cells (Figure 1E). These results indicated that fibroblasts, rather than epithelial cells, were sorted in the established cell culture system. Dishevelled Segment Polarity Protein 3 (DVL3) and phospho-β-catenin (S552) are important regulators of Wnt/β-catenin signalling.21 Double-layer co-culture was performed, and IB assays for the detection of Wnt signalling pathway activation by NFs were performed in SW620 cells. The results showed that NFs inhibited the expression of S552 and DVL3 after 2 d and 4 d of co-culture (Figure 1F). Correspondingly, CAFs promoted the expression of S552 and DVL3 after 2 d of co-culture (Figure 1G). These results illustrate the different effects of NFs and CAFs on the Wnt signalling pathway in SW620 cells. Total β-catenin expression was downregulated in cells by co-cultured with NFs (Figure 1H), in contrast to CAFs co-culture, which did not alter its levels (Data not shown). These results indicated that the stability of β-catenin is regulated by specific components within the native TME. The expression of Wnt5a was significantly upregulated in SW620 cells at 2 to 4 d of co-culture, suggesting its potential association with the non-classical Wnt signalling pathway (Figure 1H). Wnt5a inhibits the Wnt/β-catenin signalling pathway by activating the Wnt/Ca2+ pathway.22 Therefore, we detected Ca2+ concentration through IF assays and found that NFs increased Ca2+ activation in SW620 cells, indicating activation of the Wnt/Ca2+ pathway (Figure 1I). Luciferase reporter assays revealed that CAFs promoted, whereas NFs suppressed, Wnt pathway activity in CRC cells (Figure 1J). The Wnt signalling pathway is closely related to EMT in the early stages of metastasis and may directly or indirectly affect EMT. Therefore, we examined EMT-related markers and found that CAFs downregulated E-cadherin and upregulated vimentin and fibronectin in 4 d, and therefore promoted EMT (Figure 1K). In contrast to CAFs, NFs upregulated E-cadherin, downregulated N-cadherin (N-cad), and inhibited EMT (Figure 1L and 1M). The cell migratory capacity assay revealed that CAFs promoted SW620 migration, whereas NFs inhibited it (Figure 1N and 1O). Overall, these results suggest that fibroblasts have a significant effect on CRC.

CRC cells activate fibroblasts to conjointly facilitate the development of tumours
In the TME, cells interact with and influence each other through the secretion of various factors.23 SW620 and SW480 cells were derived from the same parent and presented the same genetic background; however, SW620 is a highly metastatic cell line, whereas the SW480 cell line exhibits a low metastatic potential.24 To mimic the tumour microenvironment after metastasis, lung fibroblast HELFs were transplanted with CRC cells using chambers. α-SMA is a marker of fibroblast activation, while TGF-β is capable of stimulating fibroblast activation.25 IB assays revealed that both SW620 and SW480 cells were able to increase α-SMA expression levels (Figure 2A and 2B). IF assays revealed that CRC cells were able to activate HELFs, inducing effects similar to those of TGF-β (Figure 2C). Elevation of α-SMA expression indicates HELFs activation in CRC cells. Cell scratch assays showed that SW620 and SW480 cells promoted the mobility of HELFs; however, this effect was not as significant as that of TGF-β stimulation (Figure 2D). Previously, we found that fibroblasts activate the Wnt signalling pathway in CRC cells. We tested the effect of CRC cells on the Wnt signalling pathway in fibroblasts. CRC cells showed increased expression of DVL3 and TCF4 in HELFs, suggesting activation of the Wnt pathway (Figure 2E and 2F). IF assays demonstrated that both SW620 and SW480 cells stimulated β-catenin expression in HELFs (Figure 2G). These results suggest that HELFs are activated through interactions with CRC cells, which leads to a greater migratory capacity and activation of the Wnt signalling pathway. Moreover, HELFs and CAFs activated the Wnt signalling pathway in SW620 cells, which involved elevation of β-catenin, DVL3, and S552 (Figure 2H). This interaction was associated with EMT and elevated expression of N-cad, Vim, and E-cad (Figure 2I). Cell-derived xenografts were established in nude mice to study the synergistic effect of SW620 cells and HELFs (Figure 2J). The tumour volume and weight indicated that SW620 cells and HELFs activated each other, contributing to the development of CRC cells (Figure 2K and 2L). Collectively, these findings suggest that CRC cells are capable of activating fibroblasts.

Cx43 expression is upregulated by interactions of CRC cells and fibroblasts
Cx43 is normally expressed on cell membranes and plays a role in intercellular junction communication. During CRC progression, the localisation of Cx43 in the cell membrane is inhibited, and its communication function is restricted.15 Previous reports have shown that CRC cells and fibroblasts can activate each other.26-28 We further explored the correlation with Cx43 expression. IB and IF assays revealed that both HELFs and CAFs increased Cx43 expression in SW620 and SW480 cells, confirming our hypothesis (Figure 3A-3D). Interestingly, co-culture with CRC cells also increased Cx43 expression in HELFs, and the IB and IF assays revealed consistent results (Figure 3E and 3F). TGF-β stimulation also increased Cx43 levels in HELFs, suggesting a correlation between elevated Cx43 levels and fibroblast activation (Figure 3G). However, the interaction between CRC cells and fibroblasts did not affect the gap junctional intercellular communication (GJIC) (Fig. S1). Although gap junction formation is an important function of Cx43, this finding suggests that interaction-induced changes in Cx43 may have physiological significance. These results suggest that the interaction between tumour cells and fibroblasts increases Cx43 expression in both cell types.

Nuclear Cx43 was associated with CRC progression
Cx43 is the most frequently expressed connexin in human tissue. The abnormal expression profile of Cx43 is upregulated or downregulated during the development of many cancers, showing two distinct effects: the promotion and suppression of cancer.29,30 While verifying Cx43 expression in CRC, IHC analysis of a cohort of 226 paired normal tissue, intestinal adenoma, and CRC samples revealed that Cx43 was predominantly expressed in the cell membrane of normal tissue samples, whereas its expression was considerably increased in the cytoplasm of intestinal adenoma. Surprisingly, Cx43 showed marked nuclear expression in the intestinal adenomas (Figure 3H). The expression patterns of Cx43 in different samples comprising the tumour centre were as follows: 64.2% (145/226) in the nucleus, 23% (52/226) in the cytoplasm, and 0.8% (2/226) in the cell membrane. The positive expression rates in the intestinal adenoma area were 42.0% (95/226) in the nucleus, 65.9% (149/226) in the cytoplasm, and 1.3% (3/226) in the cell membrane. In the normal intestinal mucosa, Cx43 is fully expressed in the cell membrane and cytoplasm; however, no expression was detected in the nuclear region. The positive expression rate of nuclear Cx43 in the tumour centre area was significantly higher than that in the intestinal adenoma area (χ2 = 22.209, P < 0.05), and the positive expression rate in the intestinal adenoma area was significantly higher than that in the normal intestinal mucosa (χ2 = 74.527, P < 0.05). The positive expression rate of cytoplasmic Cx43 was lower in the tumour centre area than in the intestinal adenoma area (χ2 = 25.675, P < 0.05). To identify the tumorigenesis stage associated with the crucial role of Cx43, we compared Cx43 expression levels at the tumour invasion site and the central region of the tumour in patient tissue sections and found that Cx43 expression was significantly higher in cells at the tumour invasion site (Figure 3I). This finding implies a potential involvement of Cx43 in the initiation of metastasis. Ki67 has been widely considered a tumour marker in pathological studies and has prognostic and diagnostic values for various cancers. The expression of Ki67 is correlated with tumour stage in the context of assessing survival and cancer progression. Cx43 expression in the nuclear region of tumour cells in the tumour centre was weakly correlated with Ki67 expression (r = 0.171, P < 0.05) (Figure 3J and Table S1). Analysis of 418 samples in TCGA dataset showed that high expression of Cx43 was associated with poor overall survival (Figure 3K). Furthermore, a forest plot based on the TCGA and GEO databases demonstrated the promoting effect of Cx43 in most databases; however, Cx43 showed a suppressive effect in a few databases (Fig. S2). Analysis of Cx43 expression and disease-free survival showed similar results (Fig. S3). Furthermore, we analysed the paired samples that were collected. The Breslow test was used to compare the differential expression in the nucleus in overall survival, and the results showed that overall survival with negative expression in the nuclear region was significantly better than that with positive expression in the nuclear region (Breslow = 11.533, P < 0.05, Figure 3L). Negative expression in the cytoplasm was significantly higher than positive expression in the cytoplasm (Breslow = 15.615, P < 0.05; Figure 3M). Correlations between Cx43 expression in the nuclear regions and clinico-pathological parameters were assessed, and the results revealed differential Cx43 expression in the nuclear region among sexes (χ2 = 3.924, P < 0.05), infiltration (χ2 = 6.992, P < 0.05), lymph node metastasis (χ2 = 12.680, P < 0.05), and vessel carcinoma emboli (χ2 = 28.432, P < 0.05) (Table S2). Specifically, nuclear Cx43 expression was significantly associated with the infiltration of the deep muscle layer, lymph node metastasis, and vessel carcinoma emboli. Our analyses linked the nuclear expression of Cx43 to clinicopathological parameters. The nuclear expression of Cx43 did not vary significantly with age (χ2 = 1.397, P > 0.05) or histologic grade (χ2 = 3.822, P > 0.05). Previously, we detected the association of interactions between CRC cells and fibroblasts with EMT and performed mRNA correlation analysis between GJA1 and CDH1, and CDH2 and Vim using the TCGA database and confirmed the correlation between Cx43 and the EMT process (Figure 3N). Although some studies have reported Cx43-mediated tumour suppression, we considered it an oncogene in CRC, which confirmed that nuclear expression of Cx43 was associated with poor prognosis.

Phenotypes of Cx43 in CRC cell lines
Next, we confirmed the function of Cx43 in CRC cells in vitro. The expression of Cx43 in different CRC cell lines was analysed using immunofluorescence (IF) assays (Fig. S4). Overexpression and silencing of Cx43 decreased and increased the invasiveness of SW480 and SW620 cells (Figure 4A-4F), respectively. SW620 and SW480 cells carried an APC mutation that resulted in constitutive nuclear accumulation of β-catenin under basal conditions. To circumvent this limitation, we utilised RKO cells that retained wild-type APC. Overexpression of Cx43 in RKO cells increased cell viability (Figure 4G and 4H). To further confirm this association, Cx43 was silenced in Cx43-overexpressing SW620 cells and an increase in cellular invasiveness was detected (Figure 4I and 4J). Changes in cellular invasiveness indicated that Cx43 overexpression inhibits invasion in cells harbouring APC mutation-driven β-catenin accumulation, whereas it promotes invasion in cells lacking β-catenin accumulation.

Upregulated Cx43 expression was mediated by TGF-β and induced diverse nuclear translocation profiles
To determine the mechanisms underlying the CAF-mediated upregulation of Cx43 expression, we analysed gene expression differences between NFs and CAFs in a GEO dataset (GSE121946). Significant upregulation of TGF-β was detected in the CAFs with distant macrometastases (Figure 5A). We established optimal conditions to explore the role of TGF-β in Cx43 expression (Fig. S5A and S5B). The duration of TGF-β action was 12 h, and the concentration was 5 ng/mL, which resulted in the highest expression of Cx43. We examined other cytokines, such as EGF and FGF, We examined other cytokines, such as EGF and FGF, which are abundantly expressed by CAFs.31 Among them, TGF-β exerted the most pronounced effect on Cx43 regulation, markedly elevated Cx43 expression (Fig. S5C). Based on the data shown in Figure 3, we separately detected changes in Cx43 levels in the nucleus Cx43 and cytoplasm. In SW480 cells, TGF-β stimulation led to the upregulation of Cx43 and β-catenin expression in the nucleus, whereas their expression in the cytoplasm was not affected (Figure 5B). However, in TGF-β-stimulated SW620 cells, the Cx43 and β-catenin levels did not change (Figure 5C). TGF-β stimulation enhanced cell invasiveness but silenced Cx43 attenuated the pro-invasive effect of TGF-β on SW480 cells (Figure 5D and 5E). Luciferase reporter gene assays confirmed that TGF-β activated the Wnt signalling pathway in CRC cells (Figure 5F). We overexpressed Cx43 in SW620, SW480, and RKO cells (Figure 5G-5I). Cx43 was expressed in the cytoplasm of all cells, revealing its significant expression in the nuclei of SW480 and RKO cells. Cx43 was rarely expressed in the nuclei of SW620 cells. TGF-β stimulation upregulated Cx43 expression in the nuclei of SW480 cells, but not in SW620 cells (Figure 5G and 5H). Furthermore, we used a TGF-β inhibitor (A83−01) for a rescue experiment and found that A83−01 downregulated the expression of nuclear Cx43, which confirmed the TGF-β associated upregulation of Cx43 in the nucleus (Figure 5J). The above results indicate that TGF-β enhanced Cx43 expression mainly by promoting nuclear translocation, which caused differential expression profiles of Cx43 in different CRC cell lines.

Nuclear translocation of Cx43 was regulated by PRKCA-mediated phosphorylation sites
Recently, an increasing number of studies have reported a relationship between Cx43 and cancer.32 Cx43 plays different roles at different stages of CRC progression.15 To investigate whether the nuclear translocation of Cx43 influences β-catenin accumulation and Wnt pathway activation, we performed RNA sequencing on RKO cells, which lack constitutive β-catenin accumulation due to their wild-type APC background (Figure 6A), which revealed that Cx43 was associated with the Wnt signalling pathway (Figure 6B). A total of 222 and 818 genes were upregulated and downregulated, respectively, in the overexpression (OE) group compared to those in the empty vector (EV) group. We verified these gene expression patterns based on significant fold-change values by qPCR (Figure 6C and 6D). The genes that interacted with Cx43 were investigated using the BioGrid (Set B) website and intersected with the differentially expressed genes obtained using RNA sequencing (RNA-seq) (Set A) to obtain four alternative genes, PRKCalpha, STS, NT5E, and S100A1 (Figure 6E). In the present study, protein kinase C alpha (PRKCA) was downregulated in RKO cells and upregulated in SW620 cells (Figure 6C and 6D). Using cNLS Mapper, we demonstrated that Cx43 does not have a classical nuclear localisation sequence (Fig. S6). Considering that PRKCA is a protein kinase, we hypothesised that the phosphorylation of PRKCA affects the nuclear translocation of Cx43. Analysis of TCGA and GEO databases revealed that low expression of PRKCalpha leads to poor prognosis (Figure 6F and 6G). IP assays performed on Cx43-overexpressing SW480 cells demonstrated the presence of a serine phosphorylation site on Cx43 (Figure 6H). GO6983, a pan-PKC inhibitor, decreased the invasiveness of SW480 cells in a concentration-dependent manner (Figure 6I). The amino acid residues at site 368 of Cx43 may be modified through PRKCA-mediated phosphorylation.33 Ser368 phosphorylation has become a key marker for assessing Cx43 function.34 Phosphorylation at the Ser368 site can slow cardiac conduction,35 suppress arrhythmic events36 and reduce gap-junction intercellular communication.37 To investigate the mechanism of action of PRKCA on Cx43, three mutants, Cx43-S368E, Cx43-S368A, and Cx43-S368Δ, were constructed, which mimicked sustained activation and sustained inhibition of phosphorylation, respectively (Figure 6J). IB assays revealed that the phosphorylation state of site 368 influences Cx43 translocation into the nucleus. Compared with S368E-associated sustained phosphorylation activation, S368A-associated sustained inhibition and S368Δ-associated deletion of site 368 resulted in a reduction in Cx43 translocation into the nucleus (Figure 6K). Moreover, the migratory ability and invasiveness of SW480 cells were enhanced by sustained phosphorylation. The invasiveness of SW480 cells decreased with loss of phosphorylation at site 368 (Figure 6L). Thus, PRKCA-mediated phosphorylation of Cx43 promotes its entry into the nucleus and contributes to malignant progression of CRC. Nuclear Cx43 contributes more to CRC progression than cytoplasmic Cx43. These results suggest that Cx43 has different abilities to translocate into the nucleus in different CRC cell lines, which is regulated by PRKCA-mediated phosphorylation at site 368, affecting cell invasion ability.

Upregulated expression of both Cx43 and β-catenin in the nucleus reflected their close relation
In this study, we confirmed that CAFs increase the expression level of Cx43 and activate the Wnt signalling pathway. Next, we investigated whether Cx43 and β-catenin are functionally correlated. IF assays for Cx43 and β-catenin revealed two upregulated expression levels in the nucleus (Figure 7A and 7B). CAFs stimulation upregulated Cx43 expression in the nucleus and colocalization in the nucleus was detected. Consistently, overexpression of Cx43 increased β-catenin levels in SW480 cells (Figure 7C and 7E). The increase in SW620 cells was not significant (Figure 7D and 7F). Co-immunoprecipitation assays confirmed the interaction between Cx43 and β-catenin in the SW480 cells (Figure 7G and 7H). IHC indicated that β-catenin expression differed between the cytoplasm and nucleus as the CRC malignancy progressed (Figure 7I). This cellular localisation trend was similar to that observed for Cx43. The expression of Cx43 and β-catenin in the nucleus correlated (r = 0.208, P < 0.05) (Figure 7J and Table S1). This reflects a correlation between the expression of the two proteins, and patients with high Cx43 expression tended to exhibit high β-catenin expression. We investigated the association of Cx43 with β-catenin based on TCGA database, and mRNA correlation analysis confirmed a weak correlation between Cx43 and β-catenin (Figure 7K). Altogether, β-catenin and Cx43 could upregulate their expression in the nucleus upon stimulation with CAFs.
Overall, our study elucidated a mechanism for β-catenin activation via the nuclear translocation of Cx43, which may be a complementary process in the Wnt/β-catenin signalling pathway (Figure 8). The abnormal expression or nuclear translocation of cell membrane proteins significantly affects cancer and other diseases. Further research can establish the nuclear translocation of Cx43 as a novel clinical indicator for diagnosis, treatment, and prognostic evaluation.

Discussion

Subcellular localisation may explain the functional paradox of Cx43
Report has demonstrated that Cx43 inhibits metastasis, and the data from that study are consistent with our own finding that Cx43 overexpression inhibits invasion in cells harbouring APC mutation-driven β-catenin accumulation (Figure 4). In contrast, our work reveals that when Cx43 translocates to the nucleus, it promotes metastasis by facilitating the nuclear entry of β-catenin. Therefore, it promotes invasion in cells lacking β-catenin accumulation (Figure 4). The functional duality of Cx43 indicates that its biological effects are inherently determined by subcellular compartmentalisation, where nuclear localisation promotes tumour invasion, while membrane localisation suppresses it. Nuclear Cx43 has been reported to be associated with survival in the colon, breast and lung.38-40 We propose a localisation-dependent functional model: cytoplasmic Cx43 inhibits the invasion capability, whereas nuclear Cx43 promotes it. In SW480 and SW620 cells, APC mutations cause β-catenin accumulation; thus, Cx43 overexpression-induced β-catenin nuclear translocation cannot further activate the Wnt signalling pathway. Consequently, the suppression of invasion observed upon Cx43 overexpression likely results primarily from the tumour-suppressive role of membrane-localised Cx43. In RKO cells, wild-type APC prevents β-catenin accumulation. Therefore, the increased invasion capacity induced by Cx43 overexpression is mainly attributable to Cx43 nuclear translocation. Critically, the spatial redistribution trajectory of Cx43 from the plasma membrane to the nucleus signifies its functional shift from gap junction activity to oncogenic functions upon nuclear translocation. In summary, this study establishes aberrant Cx43 nuclear translocation as a novel mechanistic driver of metastatic initiation, operating primarily in cells with normal β-catenin contexts rather than in those with pre-existing β-catenin dysregulation. Cx43 possesses strong potential as a therapeutic target for CRC; however, its dual functionality necessitates precise patient stratification prior to clinical translation.

Cx43 may specifically link to CRC processes
Nuclear Cx43 possesses oncogenic potential distinct from its classical gap-junction functions and may serve as a novel CRC biomarker.35 At the plasma membrane, Cx43 exerts its tumour-suppressive effects through functional gap-junction channels. Ser368 phosphorylation induces closure of Cx43 gap-junction channels.41 Phosphorylation of Cx43 at Ser368 not only attenuates this membrane based tumour-suppressive effects but also facilitates the nuclear translocation of Cx43, thereby endowing it with oncogenic functions within the nucleus. Consequently, inhibition of Ser368 phosphorylation represents a promising therapeutic target for CRC. The Ser368 site is located in the C-terminal. This could be achieved using developed Cx43-targeting agents, such as peptide mimetics (αCT1, CT10), small-molecule inhibitors (Carbenoxolone), or nanocarriers, which may specifically impede Cx43 nuclear translocation.42-44 Cx43 may influence through modification of cellular communication in the TME. Hemichannels open under hypoxic or inflammatory conditions, releasing deleterious signals such as ATP and glutamate that activate CAFs, thereby promoting immunosuppression.35 CAFs express high levels of Cx43 in CRC and establish intercellular gap-junctions that supply metabolic support to tumour cells.45 In metastatic lesions, CAFs frequently exhibit high Cx43 expression. Aberrant expression of this connexin correlates significantly with patient prognosis, and connexins play a pivotal role in the TME.46 Therefore, therapeutic strategies directed against Cx43 necessitate precise patient stratification. For CRC patients exhibiting high nuclear Cx43, inhibiting Cx43 nuclear translocation or disruption the Cx43/β-catenin interaction could attenuate metastatic potential.

Regulatory mechanisms of Cx43 nuclear translocation
This study elucidated two regulators of Cx43 nuclear translocation. First, TGF-β is a key stimulator of synergistic cellular effects within the TME. As a classical activator of CAFs, TGF-β upregulates Cx43 expression.47 Results in Figure 5 indicated that the TGF-β mediated enhancement of CRC invasiveness is, at least in part, dependent on the nuclear localisation of Cx43. Our findings indicated that tumour cell-fibroblast interactions involve mutual activation via secretory factors, although additional research may identify other critical factors. Second, PRKCA potentially resolves conflicting reports on Cx43 subcellular localisation. Despite extensive studies on Cx43, few studies have investigated its nuclear translocation, as exemplified by the predominant focus on Cx43 mutations that correlate with gap junction dysfunction. Previous reports have indicated that phosphorylation regulates Cx43 behaviour, including membrane trafficking and gap junction assembly.48 The C-terminus of Cx43 is modulated by various kinases that contain multiple phosphorylation sites.49 PRKCA phosphorylates Cx43 at S368 and phosphorylation at this site is correlated with nuclear aggregation. In our study, analysis of TCGA and GEO databases revealed that low expression of PRKCA leads to poor prognosis (Figure 6F and 6G). A predictive analysis indicated that PRKCA is associated with the activation of the WNT signalling pathway.50 Interestingly, the predictions from this study are consistent with our experimental findings, suggesting that PRKCA may facilitate the nuclear translocation of Cx43, thereby initiating downstream activation of the Wnt signalling pathway. A copy number variation (CNV) analysis study demonstrated that alterations in PRKCA increase susceptibility to the development of CRC.51 The BRAFV600E mutation is a frequent alteration in CRC and PRKCA has been identified as a contributor to the mechanisms of intrinsic resistance to BRAFV600E targeted therapy in CRC,52 thereby implicating PRKCA as a pro-oncogenic factor in CRC. PRKCA may exert its oncogenic effects through mediation by miRNA.53 Investigating the role of PRKCA in modulating Cx43 nuclear translocation during the promotion of CRC metastasis warrants further exploration.

Potential mechanism of Cx43/β-catenin cotranslocation and Wnt activation
Our data indicated a protein-protein interaction between Cx43 and β-catenin (Figure 7), suggesting that their co-translocation may drive Wnt pathway activation, which is a pivotal event in CRC progression. Consequently, we speculated that the oncogenic function of Cx43 is ultimately mediated by nuclear translocation of β-catenin. Although structural constraints might impede the nuclear translocation of full-length Cx43 (which contains four transmembrane domains), current evidence suggests that Cx43 may lose gap junction function through mutations, epigenetic alterations, or post-transcriptional events.54 Proteolytic fragments (e.g., the 20 kDa C-terminal fragment generated via internal translation initiation) can translocate to the nucleus and regulate gene transcription.55 We hypothesised that these truncated Cx43 isoforms facilitate β-catenin nuclear import and subsequent Wnt pathway activation. However, the core mechanistic questions remain unresolved. Further research should clarify the molecular basis for concomitant Cx43/β-catenin nuclear translocation and determine whether S368 phosphorylation modulates Cx43/β-catenin interaction. Nonetheless, whether nuclear Cx43 can promote tumorigenesis independently of β-catenin remains an open question that warrants further investigation.

Limitations of the study and anticipated further exploration
Despite these mechanistic insights, this study has limitations that merit future investigation. First, the form of Cx43 undergoing nuclear translocation remains ambiguous, and whether full-length protein, C-terminal fragments, or other proteolytic derivatives mediate nuclear effects requires the biochemical isolation of nuclear Cx43 complexes. Second, the precise mechanisms by which nuclear Cx43 governs CRC phenotypes need to be elucidated, potentially involving chromatin immunoprecipitation, to identify direct transcriptional targets. Third, while PRKCA-mediated S368 phosphorylation is implicated in translocation, in vivo validation of this phospho-regulatory axis using phosphorylation-mimetic/defective mutants in metastatic models is essential. Fourth, the functional contribution of the 20 kDa Cx43 fragment to metastasis requires genetic or pharmacological perturbation. Fifth, the determinants of cell type-specific variability in Cx43 nuclear translocation across cancer lineages should be explored. Sixth, the functional interdependence between nuclear Cx43, β-catenin, and Wnt activation requires pathway inhibition studies. Seventh, luciferase reporter gene plasmids are commonly used to assess β-catenin mediated TCF/LEF transcriptional activity in the Wnt signalling pathway. We employed this method in our analysis of cancer cells (as shown in Figure 1J and 5E). However, for the experiments in Figure 2, transfection of fibroblasts proved to be particularly challenging. Consequently, we sought to use changes in TCF4 levels as an indirect indicator of alterations in Wnt pathway transcriptional activity. However, it is important to note that an increase in TCF4 expression only serves as important supporting evidence for the activation of the Wnt pathway. Eighth, in vivo evidence of nuclear Cx43-driven metastasis requires animal models with spatial tracking of translocated Cx43. Finally, physiological relevance could be strengthened by validating key findings in patient-derived organoids to better recapitulate TME complexity.

Conclusion
We identified Cx43 nuclear translocation as a novel mechanism for β-catenin activation. Nuclear translocation of Cx43 promotes CRC progression, and nuclear Cx43 is associated with CRC progression. Moreover, the Wnt signalling pathway was activated in the co-culture of CRC cells and fibroblasts. Thus, Cx43 nuclear translocation has emerged as a potential biomarker for diagnosis, therapeutic development, and prognosis.

Ethics statement
This research complies with all relevant ethical regulations. Animal experiments were approved by the Ethics Committee of Zhejiang University (Ethics Committee number: ZJU20160023). Ethical approval for patients was obtained from approved by the Yinzhou No.2 Hospital of Ningbo China (Ethics Committee number: 2023−013). This study was conducted in compliance with the Declaration of Helsinki and all applicable ethical guidelines. Informed consent is not required for this study, primarily because the risks involved are no greater than minimal, the waiver of informed consent will not adversely affect the rights and welfare of the subjects, and the privacy and identifiable information of the participants are protected.

Supplementary Material

Supplementary Material
All the supplementary figures together