The GIST-T1 cell line was derived from a metastatic pleural tumor originating from a G-GIST in a 47-year-old Japanese woman with a heterozygous mutation in KIT exon 11 (p.V560_Y578del). This cell line was procured from Cosmo Bio (Cosmo Bio Co., Ltd., Tokyo, Japan) and cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) supplemented with 10% fetal bovine serum (FBS) (Biowest, Nuaillé, France), 100 U/mL penicillin G, and 100 μg/mL streptomycin (Invitrogen; Thermo Fisher Scientific, Waltham, MA, USA) at 37 °C in a 5% CO₂ atmosphere. Human umbilical vein endothelial cells (HUVECs), sourced from the endothelium of veins in the human umbilical cord, were obtained from Takara Bio (Takara Bio, Inc., Shiga, Japan) and maintained in Endothelial Cell Growth Medium 2 (Takara Bio, Inc.).

Female NOD.Cg-PrkdcscidIl2rgtm1Sug/ShiJic (NOG) mice, aged 7 weeks and sourced from Jackson Laboratory Japan, Inc., Japan, with an average body weight of 17–20 g, were utilized and allocated into two distinct groups. The mice were housed in cages with three or four mice per cage, provided with unrestricted access to food and water, and maintained under a 12-hour light and 12-hour dark (12:12 LD) cycle.

Plasmid construction and transfection were performed as previously described [32]. Full-length Cx43 cDNA was synthesized via reverse transcriptase polymerase chain reaction (RT-PCR) using the forward and reverse primers specified in Supplementary Table S1, Ampli Taq Gold (Thermo Fisher Scientific), and mRNA extracted from a SI-GIST exhibiting high Cx43 mRNA expression. The amplified DNA fragments were subjected to electrophoresis, collected, digested with restriction enzymes, and subcloned into the BamHI and XhoI sites of the pcDNA3.1/Zeo (+) mammalian expression vector with the CMV promoter (Thermo Fisher Scientific). To confirm the authenticity of the product, the vector containing the full-length Cx43 cDNA was sequenced using ABI BigDye Terminator version 3.1 (Applied Biosystems; Thermo Fisher Scientific) and ABI Prism 3100Avant Genetic Analyzer (Applied Biosystems; Thermo Fisher Scientific). A total of 1 × 10⁶ GIST-T1 cells were suspended in 2 µg of the vector with full-length Cx43 cDNA in 100 µL of Cell Line Nucleofector kit V solution (Lonza, Basel, Switzerland) and electroporated using an Amaxa Nucleofector II machine (program T20) (Lonza) in accordance with the manufacturer’s protocol. Stable GIST-T1 cells with elevated Cx43 expression (GIST-T1-Cx43 cells) were selected by incrementally increasing the concentration of Zeocin (Thermo Fisher Scientific) to 250 μg/mL, and a single monoclonal clone was isolated using the limiting dilution method. Cloning was performed twice. Western blot analysis was performed as described previously [32, 33]. Equivalent amounts of protein lysates from the two GIST cell lines were applied to Bolt 4%–12%, Bis-Tris Plus WedgeWell^TM Gels, electrophoresed, and transferred to an iBlot 2 PVDF transfer membrane (Invitrogen, CA, USA). Polyclonal rabbit anti-Connexin 43 antibody (Sigma-Aldrich; Merck KGaA, C6219, dilution 1:2000), polyclonal rabbit anti-human CD117, c-kit antibody (DAKO Cytomation, Glostrup, Denmark, A4502, dilution 1:500), and monoclonal mouse anti-beta actin antibody (Abcam, Cambridge, UK, ab8226, dilution 1:1000) were used as primary antibodies, and HRP-conjugated goat anti-rabbit IgG antibody (DAKO Cytomation, dilution 1:2000) and HRP-conjugated goat anti-mouse IgG antibody (DAKO Cytomation, dilution 1:1000) were used as secondary antibodies. The membranes were incubated with primary and secondary antibodies using the iBind^TM Solution Kit (Invitrogen). Protein bands were enhanced using Amersham ECL Prime Western blotting Detection Reagent (GE Healthcare Life Science, Buckinghamshire, UK) and visualized using a ChemiDoc Imaging System (Bio-Rad Laboratories, CA, USA).

Thirteen female NOG mice, aged 7 weeks, were allocated to two groups. Anesthesia was administered using isoflurane at a concentration of 4%–5% for induction and 1.5%–3% for maintenance. The peritoneum was excised to a length of approximately 15 mm to expose the spleen. Two suspensions of GIST cell lines in DMEM supplemented with 10% FBS (2 × 10⁶ cells/100 µL) were meticulously injected into the spleen using a 29-G needle. After confirming the absence of bleeding at the injection site, the spleen was repositioned within the abdominal cavity, and the peritoneum and skin were sutured with stainless-steel wound clips. To ensure stable cell transplantation, the spleen was retained in the abdominal cavity without dissection. Four weeks after the intrasplenic injection of tumor cells, the mice livers were carefully excised, and the number of liver metastases was analyzed histologically (Figure 1A).

Dissected liver tissues from NOG mice with intrasplenic transplantation of the two GIST cell lines were fixed in 10% neutral-buffered formalin. The tissues were cross-sectioned at 2 mm intervals, dehydrated, and embedded in paraffin. Three-micrometer-thick sections were cut and used for conventional hematoxylin and eosin (H&E) staining. The number of liver metastases in all H&E-stained specimens was evaluated histologically. Three or more tumor cells were counted as one metastatic focus in the liver.

Two GIST cell lines were seeded into six wells per cell line at a density of 1 × 10³ cells/well in DMEM supplemented with 10% FBS in 96‐well plates (Corning, Inc., NY, USA). Following a 7-day incubation period, 10 μL of Cell Counting Kit-8 (Dōjindo Laboratories, Kumamoto, Japan) was added to each well, and the plates were incubated at 37 °C for an additional 2 h, according to the manufacturer’s protocol. Cell viability was assessed using a microplate reader (2030 ARVO X4, PerkinElmer Life and Analytical Sciences, Shelton, USA). The experiments were conducted in triplicate.

As previously described with certain enhancements [32], a migration assay was conducted using Matrigel-free Falcon cell culture inserts (Corning, Inc.), while an invasion assay was performed using 24-well BD BioCoat Matrigel Invasion Chambers (BD Biosciences, NJ, USA), according to the manufacturer’s protocol. Given the stability of the results, two GIST cell lines were suspended at a concentration of 2 × 10⁶ cells/mL in DMEM supplemented with 0.4% FBS and introduced into the upper chambers of the inserts, followed by incubation for 48 h. The cells that migrated, invaded, and adhered to the bottom surface of the membranes were permeabilized in 100% methanol, fixed in 10% neutral buffered formalin, and stained with Giemsa stain for 1 h at room temperature. The total number of migrated and invaded cells on the membrane was quantified using a Hybrid Cell Count BZ-X710 system (Keyence, Corp., Osaka, Japan). The experiments were conducted in triplicate.

To investigate the interactions between tumor cells and HUVECs, a static adhesion assay was conducted as previously described, with some modifications [32]. A migration assay was performed using Matrigel-free Falcon cell culture inserts (Corning, Inc.), and an invasion assay was performed using 24-well BD BioCoat Matrigel Invasion Chambers (BD Biosciences, NJ, USA) according to the manufacturer’s protocol [32]. HUVECs were cultured at a density of 2.5 × 10⁵ cells/well in 96-well plates overnight. Two GIST cell lines were labeled with 2 μg/mL Calcein-AM (Dōjindo Laboratories) at 37 °C for 30 min, washed thrice with phosphate-buffered saline (PBS), and suspended at a density of 2 × 10⁶ cells/mL in DMEM supplemented with 0.4% FBS to ensure stable results. The two GIST cell lines were seeded at a density of 5 × 10⁵ cells/well on confluent HUVEC monolayers. Following a 2-h co-culture period, non-adherent tumor cells were removed along with the medium, and tumor cells adhered to HUVECs were washed three times with PBS along with HUVECs. Fluorescence intensity was measured using a fluorescence microplate reader (2030 ARVO X4, PerkinElmer Life and Analytical Sciences) at excitation and emission wavelengths of 485 and 535 nm. The experiments were conducted in triplicate.

To evaluate the transmigration of tumor cells through the endothelial monolayer, a transendothelial migration assay was performed as previously described, with some modifications [32]. HUVECs were pre-seeded at a density of 2 × 10⁵ cells per well onto 24-well Transwell Inserts (Corning, Inc.) and cultured until a monolayer was established. Subsequently, 5 × 10⁴ gastrointestinal stromal tumor (GIST) cells labeled with 2 μg/mL CalceinAM were introduced into the upper chamber of the insert. Following a 48-h co-culture period, non-migrated cells on the upper side of the membrane were removed using a cotton swab. Transmigrated cells on the lower side of the membrane were fixed with 10% neutral-buffered formalin. Transmigrated cells were visualized and quantified in five random fields at ×100 magnification using a fluorescence microscope equipped with a Hybrid Cell Count BZ-X710 system (Keyence, Corp.). The experiments were conducted in triplicate.

To explore the impact of Cx43 overexpression on the downstream signaling pathways of KIT in GIST cells, Western blot analysis was performed using the same method used to confirm the stable cell line. After assessing the protein concentrations in the lysates from the two GIST cell lines, equivalent amounts of protein from each sample were electrophoresed. Mouse anti-phospho-KIT (pTyr936) antibody (Affinity BioReagents, Inc., Golden, USA, dilution 1:1000), polyclonal rabbit anti-human CD117, c-kit antibody (DAKO Cytomation, A4502, dilution 1:500), phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) antibody (Cell Signaling Technology, Inc., Beverly, MA, USA, 9101, dilution 1:1000), p44/42 MAPK (Erk1/2) antibody (Cell Signaling Technology, Inc., 9102, dilution 1:1000), phospho-Akt (Ser473) antibody (Cell Signaling Technology, Inc., 9271, dilution 1:1000), Akt antibody (Cell Signaling Technology, Inc., 9272, dilution 1:1000), and monoclonal mouse anti-beta actin antibody (Abcam, ab8226, dilution 1:1000) were used as primary antibodies, and HRP-conjugated goat anti-rabbit IgG antibody (DAKO Cytomation, dilution 1:2000) and HRP-conjugated goat anti-mouse IgG antibody (DAKO Cytomation, dilution 1:1000) were used as secondary antibodies.

To compare the number of liver metastases in the two groups of mice transplanted with GIST cell lines in the in vivo experiment, the mean and standard deviation (SD) for each group were calculated. The significance of the differences (p < 0.05) between groups was assessed using a two-tailed t-test with Welch’s correction. Grubbs’ test (α = 0.1) was used to identify and statistically exclude a single outlier in each group after confirming data normality. The significance of the in vitro experiments was evaluated using an unpaired Student’s t-test. All statistical analyses were performed using GraphPad Prism 10.6.0 (GraphPad Software, Inc., Boston, MA, USA).

As previously reported, Cx43 protein expression was absent in G-GISTs [20]. Western blot analysis indicated that Cx43 protein expression in GIST-T1-Cx43 cells was significantly higher than that in the original GIST-T1 cells (Figure 2). These cells were established as stable GIST-T1 cells expressing elevated levels of Cx43, which was achieved by transfecting full-length Cx43 cDNA into the GIST-T1 cells.

To investigate the impact of Cx43 on the hepatic metastatic potential of GIST cells in vivo, an intrasplenic transplantation experiment was performed in mice. Four weeks after the injection of tumor cells into the spleens of 13 mice, the number of liver metastases was assessed using H&E-stained specimens. The number of hepatic metastatic foci in mice transplanted with GIST-T1-Cx43 cells was significantly 2.53 times greater than that in mice transplanted with GIST-T1 cells (p = 0.010) (Figures 1B,C). A single outlier in each group was statistically identified and excluded following confirmation of data normality (Supplementary Figures S1, S2).

To assess the influence of Cx43 on GIST cell proliferation, we conducted a comparative analysis of the proliferative capacities of the two GIST cell lines. Following the seeding of 1 × 10³ cells from each cell line and a 7-day incubation period, GIST-T1-Cx43 cells exhibited significantly reduced growth compared to GIST-T1 cells (p < 0.001) (Figure 3). Additionally, we evaluated the impact of Cx43 on the migratory and invasive behaviors of GIST cells using Matrigel-free and Matrigel-coated transwell membranes, respectively. The GIST-T1-Cx43 cells demonstrated a 3.43-fold increase in migration through the transwell membrane compared to GIST-T1 cells (p < 0.001) (Figure 4A), and a 1.70-fold increase in invasion through the Matrigel-coated transwell membrane (p = 0.036) (Figure 4B).

To assess the influence of Cx43 on the adhesion of GIST cells to HUVECs, a static adhesion assay was performed. The results indicated that CalceinAM-labeled GIST-T1-Cx43 cells exhibited significantly higher adhesion to HUVECs, with a 1.67-fold increase compared to CalceinAM-labeled GIST-T1 cells (p < 0.001) (Figure 5). Furthermore, to evaluate the impact of Cx43 on the transmigration of GIST cells through HUVECs, a transendothelial migration assay was performed using three distinct cell lines. The findings revealed that CalceinAM-labeled GIST-T1-Cx43 cells migrated through HUVECs 1.56 times more than CalceinAM-labeled GIST-T1 cells (p < 0.001) (Figure 6).

KIT expression, its phosphorylation, and the phosphorylation of downstream molecules of the KIT signaling pathways, including PI3K-AKT and RAS-RAF-MAPK, were examined. MAPK phosphorylation was activated, whereas AKT phosphorylation was suppressed in GIST-T1-Cx43 cells compared to that in the original GIST-T1 cells (Figure 7). The RAS-RAF-MAPK pathway was significantly activated in GIST-T1-Cx43 cells.