The clinical starting point was a group of patients whose stomachs suffer as a consequence of their livers. The team enrolled 111 patients with hepatitis B-related liver cirrhosis or hepatic failure at the Third Affiliated Hospital of Sun Yat-sen University between January 2022 and August 2024, alongside 118 control subjects. Using the gastrointestinal symptom rating scale, a validated 15-item questionnaire, the researchers found that patients with liver disease scored significantly higher overall than controls, 51.49 versus 42.40, with heartburn, acid reflux, nausea, bloating and diarrhea all markedly more frequent. Endoscopy told a matching story: the diseased mucosa showed more erosion and bleeding, and statistical analysis confirmed that mucosal lesions correlated with higher symptom scores, while normal mucosa tracked with lower scores.

Why would a congested stomach become a hypoxic one? The explanation is vascular. In portal hypertension and hepatic failure, restricted blood flow through the portal vein causes gastric venous congestion, and the oxygen supply to the epithelium falls behind demand. Using Hypoxyprobe-1 staining, the team demonstrated that hypoxia was indeed present in the gastric epithelium of these patients. In parallel, they established a mouse model of portal hypertension by partially ligating the portal vein, and again the hypoxia signal rose in step with a quantified gastric injury index. Crucially, the transcription factor HIF-1α, the master regulator of the cellular response to low oxygen, accumulated in the nuclei of epithelial cells in both human and mouse tissue, poised to reprogram gene expression.

That reprogramming showed up first as a metabolic shift. RNA sequencing of clinical gastric samples revealed enrichment of glycolysis and anaerobic metabolism pathways, and untargeted metabolomics of epithelial cells from the mouse model confirmed the biochemical signature: lactic acid rose while isocitric acid and citric acid, intermediates of the mitochondrial citric acid cycle, fell. This is the classic Warburg-like pattern in which cells, deprived of the oxygen needed for oxidative phosphorylation, burn glucose to lactate instead. In the short term this is a survival strategy, but the study suggests that under sustained hypoxia it becomes pathological, feeding a vicious cycle of metabolic disorder, reactive oxygen species generation and cell death.

The central molecular discovery concerns FUNDC1, an outer mitochondrial membrane protein previously known to help build mitochondria-associated ER membranes, or MAMs. These contact sites are far more than structural curiosities: they regulate calcium transfer from the ER to mitochondria, lipid exchange and mitochondrial dynamics. The researchers found that HIF-1α binds directly to a hypoxia-responsive element in the FUNDC1 promoter, driving its transcription, and that a second, post-transcriptional layer reinforces this. The m6A methyltransferase METTL3 chemically modified FUNDC1 messenger RNA, and the RNA-binding protein IGF2BP3, the most upregulated m6A reader in cirrhotic gastric tissue, recognized and stabilized that modified transcript. In mice carrying a mutant METTL3, FUNDC1 upregulation was blunted, confirming the epitranscriptomic mechanism.

With FUNDC1 elevated, the ER and mitochondria embraced more tightly. Transmission electron microscopy showed expanded contact zones in epithelial cells from portal hypertensive mice and from cirrhotic patients, and a proximity ligation assay for the IP3R2–VDAC1 tethering pair quantified the same increase. Coimmunoprecipitation confirmed that FUNDC1, IP3R2, VDAC1 and GRP75 formed complexes under hypoxic conditions. Truncation experiments pinpointed amino acids 7 to 48 of FUNDC1 as the domain essential for MAMs formation, a detail the authors highlight as a blueprint for designing inhibitors that could block the tethering interface specifically.

The consequences of this hyper-tethering were metabolic and oxidative. Seahorse extracellular flux analysis showed that epithelial cells under hypoxia had reduced maximal respiration, increased proton leak, elevated non-mitochondrial oxygen consumption and heightened basal and compensatory glycolysis, all of which were reversed when FUNDC1 was knocked down. Levels of electron transport chain subunits NDUFA9, SDHA, cytochrome b and ATP5A fell, along with ATP production. Meanwhile, mitochondrial superoxide, lipid peroxidation products such as 4-HNE and malondialdehyde, and oxidized mitochondrial DNA all climbed, while antioxidant defenses, superoxide dismutase activity and the glutathione redox ratio, collapsed. Calcium handling was also disturbed, with both cytosolic and mitochondrial calcium rising and membrane potential falling.

The final executioner in the cascade was NINJ1, a transmembrane protein that has recently emerged as the mediator of plasma membrane rupture during lytic cell death. In its inactive state NINJ1 is scattered across the membrane; when activated it oligomerizes into polymers that lyse the membrane and spill cellular contents. The team found NINJ1 oligomerization increased in gastric tissue from cirrhotic and hepatic failure patients, alongside elevated cleaved caspase-3 and cleaved caspase-1, markers of apoptosis and pyroptosis. Silencing NINJ1 in mice with portal hypertension reduced lactate dehydrogenase release, a readout of membrane permeabilization, restored epithelial viability and lessened mucosal injury. Notably, NINJ1 blockade did not change caspase activation, indicating it acts at the terminal step of membrane rupture rather than upstream of cell death signaling.

Two intervention experiments tied the whole axis together and pointed toward the clinic. Mito-TEMPO, a mitochondria-targeted antioxidant, suppressed NINJ1 oligomerization, reduced LDH release and protected the gastric mucosa in the mouse model, placing mitochondrial reactive oxygen species upstream of NINJ1 activation. The HIF-1α inhibitor PX-478 suppressed FUNDC1 expression and attenuated injury. Confocal imaging also showed that fragmented mitochondrial DNA released under oxidative stress colocalized with NINJ1, hinting at a direct or indirect role for escaping mtDNA in activating the rupture machinery, though the authors acknowledge the precise mechanism remains unresolved.

The authors are candid about limitations. The partial portal vein ligation model mechanically constricts the portal vein without reproducing hepatic parenchymal injury, systemic inflammation or the other comorbidities of clinical cirrhosis, and only knockdown rather than tissue-specific knockout approaches were used. Yet the reversibility of the pathway is encouraging: because MAMs remodeling is dynamic rather than covalent, targeted disruption of excessive ER-mitochondria tethering, inhibition of METTL3 or IGF2BP3 with emerging small molecules, or blockade of NINJ1 oligomerization with monoclonal antibodies could, in principle, interrupt the cycle before irreversible tissue damage sets in. If those strategies survive preclinical and clinical testing, the humble contact points between two organelles may become the address at which doctors finally treat the stomach’s silent suffocation.

Subject of Research: The FUNDC1–MAMs–NINJ1 signaling axis in hypoxia-induced gastric mucosal injury

Article Title: Mitochondria-associated endoplasmic reticulum membranes facilitate gastric epithelial injury in hypoxia-induced gastric mucosal lesions

Article References: Tan, S., Luo, J., Ou, X., Guo, X., Ke, B., & Tan, S. (2026). Mitochondria-associated endoplasmic reticulum membranes facilitate gastric epithelial injury in hypoxia-induced gastric mucosal lesions. Journal of Advanced Research. https://doi.org/10.1016/j.jare.2026.10.007

Image Credits: AI Generated

DOI: 10.1016/j.jare.2026.10.007

Keywords: hypoxia, gastric mucosal injury, mitochondria-associated ER membranes, FUNDC1, HIF-1α, m6A methylation, METTL3, IGF2BP3, NINJ1, portal hypertension, mitochondrial oxidative stress, plasma membrane rupture