Long non-coding RNAs, or lncRNAs, are RNA transcripts longer than two hundred nucleotides that are not translated into proteins. Once dismissed as transcriptional noise, they are now recognized as key regulators of gene expression, chromatin structure, and cellular signaling. In cancer biology, lncRNAs can behave as oncogenes or tumor suppressors depending on the networks they engage. GAS5, short for growth arrest-specific transcript 5, earned its name because it accumulates when cells halt growth, and earlier work linked it to starvation responses and glucocorticoid receptor signaling. Its role in hepatocellular carcinoma, however, has been described inconsistently across studies, with some reports suggesting tumor-suppressive behavior and others hinting at context-dependent oncogenic activity. Resolving this ambiguity was one of the motivations for the new investigation.

The researchers began by mining multiple public gene-expression datasets to assess GAS5 levels in liver tumors. Across these independent cohorts, GAS5 was markedly reduced in hepatocellular carcinoma tissues compared with healthy liver tissue, a consistent depletion that positioned the molecule as a candidate tumor suppressor rather than a driver of malignancy. The team then examined a panel of hepatocellular carcinoma cell lines, including SNU-387, PLC/PRF/5, SNU-182, Huh7, SNU-449, and HepG2, comparing them with a non-malignant liver cell line called THLE-2. The pattern held: GAS5 expression was depressed in the cancerous lines, and subcellular fractionation experiments revealed that the RNA was distributed between the nucleus and the cytoplasm, a detail that matters because a molecule’s location often hints at its mechanism of action.

To test what GAS5 actually does, the researchers forced its overexpression in hepatocellular carcinoma cells and measured a battery of malignant behaviors. The results were striking. Cells flooded with GAS5 lost viability, formed fewer and smaller colonies, and became less capable of migrating and invading through matrices, the two properties that make tumors lethal by enabling metastasis. Flow-cytometric analysis showed that the cells accumulated in the G0/G1 phase of the cell cycle, the resting checkpoint before DNA replication, indicating that GAS5 blocks the transition into active proliferation. At the same time, the treated cells showed elevated rates of apoptosis, the controlled self-destruction program that cancer cells typically evade. Together, these findings portray GAS5 as a multi-pronged brake on the malignant phenotype.

The mechanistic heart of the study lies in the competitive endogenous RNA hypothesis, a model in which lncRNAs act as molecular sponges. MicroRNAs are short RNA molecules, roughly twenty-two nucleotides long, that silence genes by binding complementary sequences in messenger RNAs and blocking their translation or triggering their degradation. A lncRNA that presents matching binding sites can sequester a microRNA, thereby freeing the microRNA’s true targets from repression. Using bioinformatic prediction algorithms, the team identified miR-18a-5p, a microRNA with well-documented oncogenic activity in liver and other cancers, as a putative partner for GAS5. The prediction was then put to the experimental test.

Luciferase reporter assays provided the decisive evidence. The researchers cloned the predicted GAS5 binding sequence downstream of a luciferase gene, so that if miR-18a-5p bound the sequence, the light-producing enzyme would be dimmed. When miR-18a-5p was added, luciferase activity dropped; when the binding site was mutated, the suppression vanished, confirming a direct and sequence-specific interaction. The same strategy validated the second link in the chain: miR-18a-5p binds the 3-prime untranslated region of the messenger RNA encoding CADM2, a cell-adhesion molecule of the immunoglobulin superfamily that has previously been implicated as a tumor suppressor in glioma and in hepatocellular carcinoma, where low expression after surgical resection predicts high recurrence risk.

With both interactions confirmed, the researchers assembled the full axis and tested it through rescue experiments, a rigorous design in which the proposed mechanism is deliberately overridden to see whether the phenotype follows. When they simultaneously overexpressed miR-18a-5p in GAS5-enriched cells, the tumor-suppressive effects of GAS5 were reversed: viability, colony formation, migration, and invasion rebounded. Silencing CADM2 produced the same cancellation, demonstrating that CADM2 is the functional effector downstream of the GAS5-miR-18a-5p pair. In other words, GAS5 restrains liver cancer not by acting alone but by protecting CADM2 from destruction by miR-18a-5p, restoring the expression of a protein that helps hold cell growth and movement in check.

To confirm that the mechanism operates in living organisms rather than only in culture dishes, the team established xenograft mouse models, implanting hepatocellular carcinoma cells into animals and comparing tumors that expressed high levels of GAS5 with control tumors. The GAS5-overexpressing tumors grew significantly more slowly, showing impaired tumorigenesis in vivo. Molecular analysis of the excised tumors revealed shifts in markers associated with apoptosis and with epithelial-mesenchymal transition, the developmental program that cancer cells hijack to detach, invade, and spread. These in vivo results, obtained under protocols approved by an ethics committee and adhering to ARRIVE guidelines for animal research, lend physiological weight to the cell-culture findings and suggest that the GAS5-miR-18a-5p-CADM2 axis is not an artifact of the laboratory environment.

The study carries broader significance for two reasons. First, it helps settle a controversy. By integrating dataset analysis, mechanistic validation, and animal modeling, the work comes down firmly on the side of GAS5 as a tumor suppressor in hepatocellular carcinoma, consistent with reports that its loss promotes liver cancer proliferation and drug resistance through other targets such as PTEN. Second, it adds a new branch to the growing map of miR-18a-5p signaling in the liver, a microRNA that has already been linked to targets including CPEB3 and to the miR-17-92 oncogenic cluster. Mapping these overlapping circuits matters because tumor cells frequently compensate when a single node is blocked; combination strategies that target multiple points in a network tend to be more durable.

Translating these findings into the clinic remains a distant prospect, and the authors are careful to frame GAS5 and its downstream network as potential therapeutic targets rather than ready-made drugs. Delivering long non-coding RNAs to tumor cells, or designing molecules that mimic their sponge function against oncogenic microRNAs, are active but unsolved challenges in RNA therapeutics. Even so, the study offers a concrete molecular signature: the trio of GAS5, miR-18a-5p, and CADM2 could serve as biomarkers for stratifying patients or monitoring disease, and the axis itself defines a vulnerability that future drug-development efforts might exploit. For a cancer that continues to claim hundreds of thousands of lives each year worldwide, every newly mapped regulatory circuit represents another possible point of attack, and this one has now been charted with unusual completeness, from patient datasets to petri dishes to living tumors.

Subject of Research: The role of the lncRNA GAS5 and the miR-18a-5p-CADM2 axis in suppressing hepatocellular carcinoma

Article Title: LncRNA GAS5 Restrains Hepatocellular Carcinoma by Targeting the miR-18a-5p-CADM2 Axis

Article References: Ke, S., Wang, C., Lin, S., Hong, L., & Wu, G. (2026). LncRNA GAS5 Restrains Hepatocellular Carcinoma by Targeting the miR-18a-5p-CADM2 Axis. Biochemical Genetics. https://doi.org/10.1007/s10528-026-11451-w

Image Credits: AI Generated

DOI: 10.1007/s10528-026-11451-w

Keywords: hepatocellular carcinoma, GAS5, long non-coding RNA, miR-18a-5p, CADM2, competing endogenous RNA, tumor suppressor, liver cancer, apoptosis, epithelial-mesenchymal transition, xenograft model, RNA therapeutics