The protein in question is connexin 26, or Cx26, encoded by the gene GJB2. In healthy tissue, six connexin subunits assemble into hemichannels that dock with neighboring cells to form gap junctions, pores through which small molecules and ions pass freely. Cx26 was long considered a tumor suppressor, yet accumulating evidence has linked its overexpression to metastasis and poor prognosis in several cancers. Earlier work had noted that Cx26 accumulates in the cytoplasm of colorectal cancer cells rather than sitting on the plasma membrane, but whether this mislocalization was merely a passive hallmark of aggressiveness or an active driver of it remained unresolved.

To answer that question, the team led by researchers at Nanjing University turned to a uniquely informative pair of cell lines: SW480, isolated from a primary colon adenocarcinoma, and SW620, derived a year later from a lymph node metastasis in the same patient. The metastatic SW620 cells expressed far more Cx26 than their primary counterparts, yet their gap junction function was weaker, not stronger. Surface-level Cx26 was nearly absent in SW620 cells. Instead, fractionation and immunofluorescence experiments revealed that the protein had accumulated in an unexpected destination: the mitochondria. When the researchers induced epithelial-mesenchymal transition, or EMT, in SW480 cells with TGF-β, Cx26 levels rose and the protein shifted toward mitochondria in step with the cells’ acquisition of invasive traits.

The pattern held beyond the laboratory dish. In a human colorectal cancer tissue microarray, multiplex immunofluorescence showed that Cx26 overlapped strongly with the mitochondrial marker TOM20 in metastatic lesions of the liver, lung, and ovary, while primary tumors retained substantial membrane-localized Cx26 with little mitochondrial colocalization. A parallel comparison of the mouse melanoma sublines B16F1 and B16F10—which share a genetic background but differ sharply in metastatic capacity—reproduced the same relationship: the highly metastatic B16F10 cells carried less membrane Cx26 and more mitochondrial Cx26. Mitochondrial translocation of Cx26, the authors conclude, is a defining feature of the aggressive, metastatic phenotype.

What steers the protein to mitochondria? Signal peptide predictions pointed to the first forty amino acids of Cx26, but deleting them changed nothing. The answer emerged from a co-immunoprecipitation screen coupled to mass spectrometry, which identified PIMT—protein L-isoaspartyl/D-aspartyl O-methyltransferase, encoded by PCMT1—among 118 candidate interacting proteins. PIMT is a conserved repair enzyme that methylates isoaspartyl residues created by the spontaneous deamidation of asparagine, a chemical decay process that kinks protein backbones and undermines stability. In the metastatic SW620 cells, the interaction between PIMT and Cx26 was far stronger than in SW480 cells, and silencing PIMT pushed Cx26 back toward the cell membrane, restored gap junction communication, and drained the protein out of mitochondria.

Mass spectrometry then pinpointed the chemistry: PIMT methylates Cx26 at asparagine 176, a residue within the extracellular loop that normally helps dock hemichannels between adjacent cells. Mutating that residue to alanine shortened the protein’s half-life, reduced its mitochondrial abundance, and increased its cytoplasmic pool. PIMT overexpression, by contrast, protected wild-type Cx26 from degradation while leaving the N176A mutant untouched. The team showed that Cx26 is degraded by the ubiquitin-proteasome system, and that PIMT suppresses Cx26 ubiquitination through its enzymatic activity. In short, PIMT acts as a chaperone of sorts—repairing Cx26’s conformation, stabilizing it, and licensing its journey to the mitochondrial compartment.

Inside the mitochondria, Cx26 proved functionally potent. Overexpression of Cx26 increased mitochondrial calcium uptake, measured with a genetically encoded calcium sensor, while the N176A mutation blunted that uptake. Electron microscopy and co-immunoprecipitation revealed that Cx26 strengthens contacts between the endoplasmic reticulum and mitochondria by bridging the calcium release channel IP3R with the mitochondrial channel VDAC1. The consequences for tumor behavior were dramatic: Cx26 overexpression accelerated EMT in SW480 cells, enlarged xenograft tumors, and multiplied pulmonary metastatic burden in mouse models, while Cx26 knockdown in SW620 and B16F10 cells did the opposite. Blocking mitochondrial calcium uptake with ruthenium red or the MCU inhibitor MCU-i4 reversed the EMT phenotype, establishing calcium flux as the mechanistic link between mitochondrial Cx26 and metastatic capacity.

The therapeutic twist came from an unbiased screen. Querying the Genomics of Drug Sensitivity in Cancer databases across 23 colorectal cancer cell lines, the researchers found that higher GJB2 expression generally predicted resistance to standard chemotherapeutics, including 5-fluorouracil and oxaliplatin. Bortezomib, a 26S proteasome inhibitor used as first-line therapy for multiple myeloma and mantle cell lymphoma, broke the pattern: cells with high GJB2 expression were more sensitive, showing lower IC50 values. In direct comparisons, bortezomib killed metastatic SW620 cells more effectively than primary SW480 cells, while the conventional colorectal cancer drugs showed no such preference. Cx26 overexpression sensitized SW480 cells to the drug, confirming the connection.

Curiously, the mechanism had little to do with the proteasome. Cellular thermal shift assays, molecular docking, differential scanning fluorimetry, and surface plasmon resonance all converged on an unexpected target: PIMT itself. Bortezomib bound PIMT directly, raised its melting temperature by 5.6 degrees Celsius, and enhanced its methyltransferase activity in vitro. Treatment with the drug strengthened the PIMT-Cx26 interaction, drove more Cx26 into mitochondria, and triggered a catastrophic surge of intracellular calcium. That calcium flood generated reactive oxygen species and activated ERK signaling, producing PARP cleavage and LC3-II accumulation—hallmarks of apoptosis and autophagy. Scavenging reactive oxygen species, blocking mitochondrial calcium uptake, inhibiting ERK with PD0325901, or knocking down Cx26 or PIMT all blunted bortezomib’s killing of SW620 cells, mapping the pathway end to end.

The study reframes Cx26 mislocalization from a passive biomarker into an organelle-specific engine of malignancy, and it challenges the assumption that solid tumors are inherently resistant to proteasome inhibitors. Bortezomib, in this context, would not work as a proteasome poison but as a molecular lever that amplifies the very PIMT-Cx26 axis on which metastatic cells have come to depend, converting their addiction into a death signal. The authors caution that mitochondrial Cx26 localization must be validated as a survival predictor in larger clinical cohorts, and that direct biochemical confirmation of the N176 modification with purified substrates is still needed. Even so, the work offers a concrete strategy for a disease state with few options: identify tumors addicted to mitochondrial Cx26, and turn that dependency against them with a drug already sitting on the pharmacy shelf.

Subject of Research: PIMT-mediated mitochondrial translocation of connexin 26 in metastatic colorectal cancer and its therapeutic targeting by bortezomib

Article Title: Mitochondrial Localization of Cx26 Promotes Colorectal Cancer Malignancy and Exposes a Fatal Vulnerability to Drug Targeting

Article References: Wang, J., Chen, L., Qin, J., Wang, X., Ye, S., An, N., Bai, M., Shen, Y., Wu, X., & Xu, Q. (2026). Mitochondrial Localization of Cx26 Promotes Colorectal Cancer Malignancy and Exposes a Fatal Vulnerability to Drug Targeting. Advanced Science, Article e78117. https://doi.org/10.1002/advs.78117

Image Credits: AI Generated

DOI: 10.1002/advs.78117

Keywords: colorectal cancer, connexin 26, Cx26, PIMT, mitochondria, metastasis, EMT, bortezomib, calcium signaling, gap junctions, drug sensitivity, Advanced Science