The gene in question is GNG4, which encodes the gamma-4 subunit of heterotrimeric G proteins, the signaling molecules best known for relaying messages from G protein-coupled receptors at the cell surface. Rather than studying GNG4 in isolation, a research team led by Xiangchou Yang, Yi Xu, and Chenwei Pan, working across institutions including Wenzhou Medical University and Shanghai Jiao Tong University School of Medicine, took an unusually comprehensive approach. They combined single-cell RNA sequencing, which profiles gene expression in thousands of individual cells, with spatial transcriptomics, which maps where those genes are active within intact tissue, and multiplex immunofluorescence, which visualizes multiple proteins simultaneously under the microscope. This triangulation allowed them to move beyond correlation and ask where, precisely, in the three-dimensional architecture of a liver tumor the immune-exclusion program operates.

The first finding to emerge was epidemiological in character. GNG4 turned out to be enriched in malignant hepatocytes, the cancerous liver cells themselves, and high expression of the gene was strongly associated with poor patient prognosis and elevated scores on a single-cell-based risk stratification model the team had constructed. In other words, patients whose tumors lit up for GNG4 fared measurably worse than those whose tumors did not. More striking still, when the researchers examined tumors from patients who had received neoadjuvant immunotherapy, treatment given before surgery to prime the immune system, they found that GNG4 expression was spatially concentrated in the regions of the tumor that had failed to respond. The gene was not merely a passive marker of aggressive disease; its geography within the tumor mapped directly onto therapeutic failure.

To test whether GNG4 was causally involved rather than a bystander, the researchers used short hairpin RNA, or shRNA, a technique that silences a specific gene by triggering its degradation, to knock down GNG4 in hepatocellular carcinoma cells. The effects were immediate and broad. Deprived of GNG4, the cancer cells proliferated more slowly, migrated less efficiently, and lost much of their invasive capacity in laboratory cultures. When the team implanted these GNG4-deficient cells into animal models, tumor growth was significantly suppressed. What makes this result particularly interesting is that the suppression was not purely a cell-autonomous phenomenon. The tumors lacking GNG4 were not just weaker; they were also more visible to the immune system.

That second, immunological mechanism is where the study becomes genuinely novel. In the animal models, GNG4-deficient tumors showed markedly increased infiltration by CD8-positive T cells, the cytotoxic lymphocytes responsible for directly killing infected and malignant cells. Not only were more of these cells present, but they were functionally hotter: the infiltrating T cells expressed elevated levels of interferon-gamma, the signature cytokine of activated cellular immunity, and granzyme B, the protease enzyme that T cells use to punch lethal holes in target cells. The authors describe this as a dual mechanism, in which removing GNG4 simultaneously restrains the tumor cells’ own growth and hands the surrounding microenvironment back to the immune system. Both arms of the effect converged on tumor restriction.

The mechanistic thread connecting the tumor cell to the T cell turned out to run through an unexpected molecule called MID1, short for Midline-1, a protein whose expression state in neighboring CD8-positive T cells changed in lockstep with GNG4 expression in adjacent malignant cells. Computational models of cellular communication, which infer signaling interactions between cell types from their co-localization and receptor-ligand expression patterns, together with spatial transcriptomics, revealed that regions of GNG4-high tumor tissue were consistently accompanied by an altered MID1 expression state in the T cells hovering at the tumor’s edge. The implication is that malignant cells broadcasting GNG4-associated signals somehow reach across the microenvironment and reprogram the identity of the immune cells nearest to them.

To establish that this relationship was functional rather than merely correlational, the researchers ran an in vitro rescue experiment using primary murine CD8-positive T cells, freshly isolated rather than immortalized. When T cells were exposed to conditioned medium, the nutrient broth in which malignant cells had been growing, from GNG4-silenced cancer cells, the T cells recovered function. Critically, that functional restoration was tied to the recovery of Mid1 expression within the T cells themselves. This is the kind of experimental loop that elevates a spatial observation into a mechanistic claim: perturb the tumor cell, watch the T cell change, and then show that the change in the T cell depends on a specific molecular marker reverting to its healthy state.

The final layer of validation came from human tissue. Using multiplex immunofluorescence on human hepatocellular carcinoma samples, the team stained for GNG4 in malignant cells and for the combination of CD8 and MID1 in infiltrating T cells, then quantified their spatial relationships across the tissue sections. The result confirmed the animal and single-cell findings: GNG4 expression in malignant cells was inversely correlated with the local density of CD8-positive MID1-positive T cells. Where the gene was loud, the killer T cells were scarce; where the gene was quiet, they clustered. In human disease, as in the models, the GNG4 axis appears to define an immune-excluded architecture, a tumor state in which immune cells are present at the periphery but barred from penetrating the malignant core.

The translational implications are considerable. Immune exclusion is one of the recognized reasons checkpoint inhibitors fail, because a drug that releases the brakes on T cells cannot help cells that never arrive at the tumor in the first place. If GNG4-high tumors represent a distinct, identifiable subset of hepatocellular carcinoma characterized by this exclusion phenotype, then GNG4 itself, or the signaling pathways it feeds into, becomes a candidate target for combination therapies designed to tear down the exclusion barrier before or alongside checkpoint blockade. The gene’s association with prognosis also suggests it could serve as a biomarker, helping clinicians identify up front which patients are unlikely to benefit from immunotherapy alone and might need a different strategy.

As with any study that moves from computational atlases to mouse models to human tissue, important questions remain before these findings reach the clinic. The precise signaling route by which GNG4 in a malignant hepatocyte influences MID1 in a neighboring T cell has not yet been fully mapped, and whether pharmacological inhibition of GNG4 in established tumors can reverse immune exclusion in patients is a question only clinical trials can answer. What the study delivers now is a clearly articulated hypothesis with unusually strong spatial evidence behind it: liver cancer is not just a disease of malignant cells, but of the conversations those cells conduct with their surroundings, and GNG4 appears to be one of the words in that vocabulary that tells the immune system to stay outside.

Subject of Research: The role of GNG4 in tumor immune exclusion and CD8+ T cell infiltration in hepatocellular carcinoma

Article Title: GNG4 associated with tumor immune exclusion and compromised CD8+ T cell infiltration via microenvironmental crosstalk in hepatocellular carcinoma

Article References: Yang, X., Zhou, Y., Jin, X., Ying, Z., Chen, Q., Chen, J., Ahmed, A.-A. W. H., Shan, Y., Xu, Y., & Pan, C. (2026). GNG4 associated with tumor immune exclusion and compromised CD8+ T cell infiltration via microenvironmental crosstalk in hepatocellular carcinoma. Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-08967-5

Image Credits: AI Generated

DOI: 10.1186/s12967-026-08967-5

Keywords: hepatocellular carcinoma, GNG4, CD8+ T cells, tumor microenvironment, immune exclusion, MID1, single-cell RNA sequencing, spatial transcriptomics, multiplex immunofluorescence, immunotherapy resistance, liver cancer, tumor immunology