Review maps epigenetic control of PI3K/AKT signaling in liver cancer

A 2026 review identifies a reciprocal molecular network in hepatocellular carcinoma, or HCC, in which DNA methylation, non-coding RNAs and histone modifications sustain activation of the PI3K/AKT pathway, while activated PI3K/AKT signaling can also reshape the tumor epigenome. The evidence for epigenetic control of the pathway is more extensive than the evidence for the reverse direction, which remains incomplete. No treatment combining PI3K/AKT and epigenetic agents has been approved for HCC, and most relevant drugs remain in preclinical development or early-phase clinical testing.

PI3K/AKT signaling is frequently and persistently activated in HCC, an aggressive malignancy of the digestive system that is often diagnosed at an intermediate or advanced stage. The pathway regulates cell survival, growth, proliferation, apoptosis and metabolic homeostasis, and its principal components include PI3K, AKT, PTEN and numerous upstream and downstream regulators. When extracellular growth factors bind receptor tyrosine kinases, class I PI3K converts phosphatidylinositol 4,5-bisphosphate into phosphatidylinositol 3,4,5-trisphosphate, or PIP3. PIP3 recruits AKT and PDK1 to the cell membrane, where AKT is activated by phosphorylation at Thr308 by PDK1 and at Ser473 by mTORC2. PTEN normally restrains the signal by dephosphorylating PIP3, while persistent pathway activation can support tumor-cell proliferation, invasion and metastasis through downstream effectors.

The review separates canonical epigenetic regulation from RNA-mediated control. Epigenetic mechanisms alter gene activity without changing the underlying DNA sequence and include DNA methylation, histone modifications and ATP-dependent chromatin remodeling. DNA methyltransferases, including DNMT1, DNMT3A and DNMT3B, establish or maintain methylation patterns, whereas TET enzymes support active demethylation. Histone acetylation, methylation, phosphorylation, ubiquitination and lactylation are controlled by opposing enzyme systems. Non-coding RNAs provide another regulatory layer: microRNAs, long non-coding RNAs and circular RNAs can influence messenger-RNA stability, translation, alternative splicing, protein ubiquitination and protein degradation. Long non-coding RNAs may also recruit chromatin-modifying complexes. These processes occur within a disease environment shaped by chronic injury, inflammation, fibrosis, cirrhosis, metabolic dysregulation, hypoxia and immune-stromal remodeling.

DNA methylation provides several experimental routes to sustained PI3K/AKT activation. ZNF191 can bind the DNMT1 promoter and increase DNMT1 expression in HCC models, altering genome-wide methylation and activating the pathway. Six2-driven DNMT1 activity hypermethylates the E-cadherin promoter, reducing E-cadherin expression while promoting epithelial-to-mesenchymal transition, PI3K/AKT/mTOR activation and resistance to 5-fluorouracil. TERT and Sp1 can cooperate to increase DNMT3B, producing promoter hypermethylation and silencing of PTEN and RASSF1A. Other methylation events silence NNAT or PDE7B, whose loss weakens restraint on pathway kinases. In the opposite direction, hypomethylation of the FBXO32 promoter increases FBXO32, which promotes ubiquitin–proteasome degradation of the AKT inhibitor PHLPP2. TET1-mediated demethylation has also been associated with PI3K/AKT/mTOR activation, increased proliferation and reduced apoptosis in HCC cell models.

Non-coding RNAs can sustain the pathway by altering RNA metabolism or the stability of signaling proteins. The oncogenic lncRNA IGFRIL recruits PTBP1, accelerating degradation of IGFBP3 messenger RNA and increasing activity of the IGF1R/AKT/mTOR axis. SP1-induced USP27X-AS1 recruits the deubiquitinase USP7, reducing AKT ubiquitination and degradation and thereby increasing AKT stability. Hypoxia can activate additional feedback loops. HIF-1α induces KDM4A-AS1, which sequesters miR-411-5p, increases KPNA2 expression and promotes AKT signaling; elevated KPNA2 and AKT activity can in turn increase HIF-1α. A parallel loop involving MAPKAPK5-AS1, miR-154-5p and PLAGL2 links hypoxia to EGFR/PI3K/AKT activation. Other examples include RP11-295G20.2 binding PTEN and promoting its lysosomal degradation, and circMRPS35 sequestering miR-148a so that increased STX3 drives PTEN degradation.

Histone modifications provide further links between chromatin state and pathway activity. Reduced PCK1 expression lowers S-adenosylmethionine synthesis and the repressive H3K9me3 mark at the S100A11 promoter, derepressing S100A11, which binds and activates AKT1. Overexpressed EZH2 deposits H3K27me3 at the miR-622 promoter, silences that microRNA and releases the CXCR4 axis, leading to PI3K/AKT and RAF/MEK/ERK activation. In HCC cells, MBD3 recruits the NuRD complex to deacetylate the TFPI2 promoter and silence a gene that normally restrains PI3K/AKT signaling. Metabolic reprogramming adds another connection: PYCR1-driven glycolysis raises intracellular lactate, promotes H3K18 lactylation at the IRS1 promoter and increases IRS1 expression. Lactate-associated H3K14 lactylation can increase NEDD4, which promotes PTEN degradation and has been linked to chemoresistance in experimental models.

Evidence that activated PI3K/AKT remodels the epigenome is comparatively limited. One reported mechanism involves AKT increasing DNMT3B transcription while raising HuR expression, extending DNMT3B messenger-RNA half-life and increasing its translation. The resulting rise in DNMT3B disrupts genome-wide methylation in HCC models. AKT can also phosphorylate the acetyltransferase p300 at Ser1834, increasing its activity and H3K9 acetylation at the HMGA2 promoter. Related experimental work links AKT-mediated p300 activity to H3K9 acetylation at the DKK1 locus, while MEK/ERK-dependent PKM2 activity contributes to H3T11 phosphorylation at the same locus. These histone changes increase DKK1 expression and the invasive or metastatic potential of the cells. Another study reports that EGF-associated PI3K/AKT activation raises HDAC2 expression, which increases cell-cycle proteins, accelerates the G2/M transition and promotes proliferation.

Clinical experience with targeted agents that might interrupt this network remains limited. In a small advanced-HCC cohort within a phase II basket trial, the pan-class I PI3K inhibitor buparlisib had poor tolerability, including elevated transaminases, anxiety and depression. Dual PI3K/mTOR inhibitor dactolisib produced severe diarrhea and liver-function abnormalities in solid-tumor studies without dedicated advanced-HCC cohorts, and development was suspended after limited antitumor activity and poor long-term tolerability. Samotolisib produced transient disease control in a subset of patients, but hyperglycemia and elevated transaminases limited prolonged treatment. AKT inhibitors have reached early-phase solid-tumor testing, but no phase II HCC data were available for MK-2206, and capivasertib had not entered a dedicated HCC trial. These findings establish clinical evaluation of the drug classes, not proof of efficacy in HCC.

Epigenetic agents face their own constraints. Decitabine and azacitidine inhibit DNA methyltransferases, but dose-limiting myelosuppression and possible liver injury complicate their use, particularly in patients with hepatic dysfunction. In a phase I study of vorinostat plus sorafenib, thrombocytopenia and liver injury were frequent dose-limiting toxicities, and no recommended phase II dose was established. Resminostat plus sorafenib was evaluated in the phase I/II SHELTER trial, where mild gastrointestinal adverse events predominated; cross-trial comparisons suggest better tolerability for the resminostat combination, but do not establish comparative efficacy. The RNA-targeting agent TKM-080301 produced modest antitumor activity in 43 patients with Child-Pugh A HCC, with grade 4 thrombocytopenia as the dose-limiting toxicity. LNA-i-miR-221 was administered to only one patient with HCC, precluding disease-specific conclusions.

Preclinical combination strategies have provided mechanistic support but not clinical validation. In cell and animal models, the HDAC inhibitor vorinostat reduced AKT phosphorylation at Ser473 and enhanced cytotoxicity from targeted agents, providing a rationale for combining HDAC and AKT inhibitors. Dual PI3K/HDAC compounds CUDC-907 and BEBT-908 showed greater preclinical antitumor activity than the corresponding single-agent combinations, while DNMT inhibitors can restore PTEN expression and lower AKT activation. The review identifies biomarker selection, treatment sequence, dose adjustment, liver reserve and tumor heterogeneity as central obstacles. Multi-omics markers that combine pathway activity, epigenetic state, genomic background and liver function may eventually guide patient selection, but direct comparative studies across HCC etiologies remain limited. Dedicated HCC trials, longitudinal sampling and validation of nanocarrier-based co-delivery are needed before these approaches can define a clinical standard.

Subject of Research: Bidirectional regulation between PI3K/AKT signaling and epigenetic controls in hepatocellular carcinoma

Article Title: Crosstalk between PI3K/AKT signaling and epigenetic regulation in hepatocellular carcinoma: molecular mechanisms and therapeutic opportunities

Article References: Zhang, Z., Liu, J., Lei, G., & Hu, X. (2026). Crosstalk between PI3K/AKT signaling and epigenetic regulation in hepatocellular carcinoma: molecular mechanisms and therapeutic opportunities. Frontiers in Cell and Developmental Biology, 14, Article 1975140. https://doi.org/10.3389/fcell.2026.1975140

Image Credits: AI Generated

DOI: 10.3389/fcell.2026.1975140

Keywords: hepatocellular carcinoma, PI3K/AKT signaling, epigenetic regulation, DNA methylation, histone modifications, non-coding RNAs, targeted therapy, drug combinations, tumor microenvironment, liver cancer, Crosstalk, between