Wnt Signaling in HCC The Wnt/beta-catenin pathway is one of the most frequently activated signaling cascades in hepatocellular carcinoma. Upon Wnt ligand binding to Frizzled receptors, cytosolic beta-catenin escapes GSK3beta-mediated phosphorylation and degradation, translocates to the nucleus, and activates target gene transcription through TCF/LEF transcription factors. CTNNB1 mutations - the second most common mutation in HCC after TP53 - constitutively stabilize beta-catenin and hyperactivate this pathway.
The Problem: Multiple Wnt-Related Classes Without Unified Characterization Several independent groups had identified Wnt-related HCC molecular subclasses but characterized them in different ways: some focused on CTNNB1 mutations and liver-specific Wnt targets, others on TGFbeta-activated Wnt through classical targets. Whether these represent the same or distinct biological entities, and how their molecular mechanisms differ, was not systematically addressed.
This Study: 642 HCCs, Two Classes, Sorafenib Testing This study analyzed Wnt-pathway gene expression in 642 HCC samples, validated findings in six independent datasets, correlated with immunohistochemistry (beta-catenin and glutamine synthetase), and used a CTNNB1-mutation gene signature. It then tested whether sorafenib - the only approved HCC drug - modulates Wnt signaling in cell lines and xenograft models representing each Wnt class.
Central Finding: CTNNB1-Class vs. Wnt-TGFbeta-Class Wnt pathway activation was detected in 315 of 642 HCCs (49.1%), further divided into CTNNB1-class (21.5%, characterized by liver-specific Wnt targets and nuclear beta-catenin) and Wnt-TGFbeta-class (27.6%, characterized by classical Wnt targets and membranous beta-catenin only). Sorafenib modulated beta-catenin/Wnt signaling in experimental models of both classes.
Dataset Scale and Architecture The study used a training set of 91 HCV-related HCC samples and an independent validation set of 144 mixed-etiology HCCs, supplemented by 407 HCC samples from five additional published datasets. Gene expression profiling of 210 Wnt-pathway mRNAs and 49 Wnt-related miRNAs was performed. CTNNB1-mutation data were available for 91 samples, and immunohistochemistry was performed on 108 samples.
NTP-Based Class Assignment Wnt class membership was assigned using the Nearest Template Prediction (NTP) method, which classifies individual samples against pre-defined class templates without requiring batch normalization. CTNNB1-class and Wnt-TGFbeta-class-specific Wnt gene signatures (the WntGenes-signatures) were derived from differentially expressed Wnt pathway genes in each class and then validated in six independent datasets.
CTNNB1-Mutation Signature A gene expression signature predicting CTNNB1 mutation status was generated from samples with and without CTNNB1 mutations (analogous to published TP53-mutation signatures in breast cancer), with 91% prediction accuracy (specificity 87%, sensitivity 100%) in an independent validation dataset. This allowed mutation status estimation even in cohorts where direct sequencing data were unavailable.
Cell Line and Xenograft Models HepG2 cells (CTNNB1-class signature-enriched, truncated beta-catenin), SNU398 (CTNNB1 Ser37C point mutation), SNU387 (Wnt-TGFbeta-class-enriched, wild-type beta-catenin), and Huh7 (wildtype, LiCl-activated Wnt) were used for in vitro sorafenib studies. Sorafenib effects were quantified via TCF/LEF luciferase reporter, Western blot, immunofluorescence, and RT-PCR for liver-related Wnt targets (GLUL, LGR5, TBX3). The HepG2 xenograft model received 30 mg/kg/day sorafenib by gavage.
CTNNB1-Class: Liver-Specific Wnt Targets and Nuclear Beta-Catenin CTNNB1-class HCCs (21.5% of all cases) show significant upregulation of 7 of 9 liver-related Wnt target genes: GLUL (glutamine synthetase), LGR5, TBX3, MERTK, EPHB2, SPARCL1, and REG3A. At the protein level, 80% of CTNNB1-class samples show nuclear/cytoplasmic beta-catenin staining vs. only 16.6% of non-Wnt HCCs, and 74.3% show cytoplasmic glutamine synthetase staining vs. 21.3% in non-Wnt HCCs (both p<0.001). The CTNNB1-mutation signature is enriched in this class (91% accuracy), and 19 of 31 Wnt-related miRNAs are dysregulated.
Wnt-TGFbeta-Class: Classical Wnt Targets Without Nuclear Beta-Catenin Wnt-TGFbeta-class HCCs (27.6% of all cases) are characterized by upregulation of classical Wnt target genes: MMP7, PLAU (urokinase-type plasminogen activator), RUNX2, CDC2, and CCND3. These tumors also show upregulation of Wnt activating genes FZD7, FZD6, TCF4, FRAT2, and ARRB2, which can activate Wnt signaling even without CTNNB1 mutations. Beta-catenin staining remains membranous rather than nuclear in this class, and only 4 Wnt-related miRNAs are dysregulated.
Only Four Genes Overlap Between Classes - In Opposite Directions Of the 210 Wnt-pathway genes analyzed, only SALL1, PRKCD, PLAU, and MAP1B were significantly dysregulated in both classes - and in opposite directions for most of them. This near-complete non-overlap establishes the two classes as biologically distinct entities rather than a spectrum of the same pathway aberration.
Consistent Across Six Independent Datasets The CTNNB1-WntGenes-signature was significantly associated with CTNNB1-class in 4 of 5 external datasets, and the Wnt-TGFbeta-WntGenes-signature was validated in 5 of 5 external datasets. Combined analysis of 642 HCCs confirmed 315 cases with Wnt activation (49.1%): 21.5% CTNNB1-class and 27.6% Wnt-TGFbeta-class - remarkably consistent across training, validation, and five public datasets.
Sorafenib Reduces TCF/LEF Reporter Activity Across All Wnt Models Sorafenib decreased TCF/LEF luciferase reporter activity in all four cell line models tested: HepG2 (CTNNB1-class, truncated beta-catenin), SNU398 (CTNNB1 Ser37C mutation), SNU387 (Wnt-TGFbeta-class, wild-type beta-catenin), and LiCl-stimulated Huh7 (GSK3beta-inhibited wild-type beta-catenin). Baseline TCF/LEF activity ranged over 1,000-fold across these lines (HepG2>SNU398>Huh7>SNU387), but sorafenib reduced signaling in all contexts.
Beta-Catenin Protein Levels Decrease and Nuclear Translocation Is Blocked Western blot analysis showed decreased beta-catenin protein levels in HepG2 and SNU398 cells following sorafenib treatment. Immunofluorescence in LiCl-treated Huh7 cells demonstrated that sorafenib prevents beta-catenin nuclear translocation, confirming that the drug acts upstream of or at the level of the beta-catenin nuclear/cytoplasmic shuttle.
Liver-Related Wnt Targets Downregulated by Sorafenib RT-PCR analysis in HepG2 cells showed significant decreases in GLUL, LGR5, and TBX3 mRNA levels after sorafenib treatment - three of the key liver-related Wnt target genes that define the CTNNB1-class. This confirmed that sorafenib reduces signaling not only at the TCF reporter level but also suppresses the specific target gene program characteristic of CTNNB1-class HCC.
In Vivo Xenograft: Reduced Tumor Volume and Extended Survival In the HepG2 subcutaneous xenograft model, sorafenib treatment (30 mg/kg/day) significantly reduced tumor volume and extended median survival from 22 days in controls to 34 days in treated animals (n=7 per group). Tumors from sorafenib-treated animals showed significant downregulation of GLUL and LGR5 compared to controls, validating the in vitro Wnt suppression in a living tumor context.
Half of HCC Patients Have Targetable Wnt Activation The finding that 49.1% of HCC cases show Wnt pathway activation - consistent across training, validation, and five independent public cohorts - establishes Wnt signaling as the most common recurrent pathway aberration in HCC. This prevalence makes Wnt-targeted therapy potentially relevant to a major fraction of the HCC patient population.
The Two Classes Have Different Clinical Correlates CTNNB1-class HCCs tend to have larger tumor diameter (>3cm) and are associated with CTNNB1 mutations and well-differentiated tumor biology. Wnt-TGFbeta-class tumors are associated with vascular invasion, satellitosis, and greater early recurrence risk after surgery. These different clinical profiles suggest distinct surveillance strategies and different adjuvant therapy needs for the two Wnt-activated patient groups.
Sorafenib as a Wnt Modulator: Implications for CTNNB1-Class Patients Current sorafenib trials are not stratified by Wnt pathway status. The finding that sorafenib suppresses Wnt signaling specifically in CTNNB1-class-like experimental models suggests that CTNNB1-class patients (identifiable by CTNNB1 mutation sequencing, glutamine synthetase IHC, or gene expression profiling) may be preferential responders to sorafenib. This hypothesis warrants prospective testing in biomarker-stratified clinical trials.
Glutamine Synthetase and Beta-Catenin IHC as Clinical Biomarkers The study demonstrates that glutamine synthetase (GS) cytoplasmic staining and nuclear beta-catenin IHC are strongly enriched in CTNNB1-class HCC (74.3% and 80% positivity, respectively) and represent simple, widely available assays that could serve as clinical biomarkers to identify CTNNB1-class patients for biomarker-stratified treatment trials without requiring gene expression profiling.
CTNNB1-Class: Mutation-Driven with Liver-Specific Transcription Factor Program The CTNNB1-class is primarily driven by CTNNB1-activating mutations (exon 3 mutations removing the Ser/Thr domain required for GSK3beta-mediated phosphorylation) and CDH1 (E-cadherin) downregulation. The resulting constitutively active beta-catenin activates a liver-specific target gene program orchestrated by class-specific transcription factors (SALL1, TLE1, TCF7, TCF7L1) rather than the ubiquitous TCF4 program seen in other cancers.
Wnt-TGFbeta-Class: Receptor-Level Activation Through Frizzled Upregulation The Wnt-TGFbeta-class shows no CTNNB1 mutations but instead upregulates Wnt receptors and co-activators FZD7, FZD6, TCF4, FRAT2, and ARRB2, which can activate canonical Wnt signaling in the absence of CTNNB1 mutations. TGFbeta pathway co-activation in this class provides a second cooperating oncogenic signal. Beta-catenin remains membranous rather than nuclear, possibly because TCF4-driven Wnt activation in this context uses a nuclear effector mechanism that does not require cytoplasmic beta-catenin accumulation.
miRNA Differences Reinforce Distinct Regulatory Programs The 19 Wnt-related miRNAs dysregulated in CTNNB1-class vs. only 4 in Wnt-TGFbeta-class reflects fundamentally different post-transcriptional regulatory environments. The close alignment between CTNNB1-class and the previously defined miRNA class A in HCC (from a parallel study) integrates the two approaches, suggesting that CTNNB1-driven Wnt activation produces a distinct miRNA regulatory network that the TGFbeta-activated Wnt pathway does not.
Sorafenib's Wnt Mechanism: Crosstalk or Direct Target? Sorafenib is primarily a multi-kinase inhibitor of VEGF receptor, PDGFR, Raf, and ERK/AKT. Its effects on Wnt signaling likely occur through indirect crosstalk: Ras/Raf/MAPK signaling can activate GSK3beta-independent beta-catenin stabilization in cancer cells, and sorafenib's inhibition of these pathways could secondarily reduce beta-catenin levels. The finding that sorafenib suppresses Wnt in wild-type and mutant beta-catenin contexts argues for a mechanism acting upstream of beta-catenin stabilization.
Prospective Biomarker-Stratified Trial of Sorafenib The in vitro and xenograft evidence that sorafenib suppresses Wnt signaling in CTNNB1-class models is hypothesis-generating, not clinical evidence. A randomized trial stratifying HCC patients by Wnt subtype (using CTNNB1 mutation testing or glutamine synthetase/nuclear beta-catenin IHC) and assessing sorafenib benefit in each stratum would directly test whether CTNNB1-class patients derive greater survival benefit from sorafenib.
Direct Wnt Inhibitors for the Wnt-TGFbeta-Class Since Wnt-TGFbeta-class HCC activates Wnt at the receptor/ligand level (through FZD7, FZD6, Wnt ligands), porcupine inhibitors (which block Wnt ligand secretion) or anti-FZD antibodies could theoretically specifically target this class. Testing these agents in Wnt-TGFbeta-class cell line and patient-derived xenograft models would determine their class-specific efficacy.
Why Does Wnt-TGFbeta-Class Lack Nuclear Beta-Catenin? The paradox that Wnt-TGFbeta-class tumors have transcriptional evidence of Wnt activation (upregulated classical targets, altered Wnt pathway genes) but lack the nuclear beta-catenin and glutamine synthetase staining expected of Wnt-active tumors is mechanistically unexplained. Identifying the nuclear effector mechanism - whether through a TCF4-dependent, beta-catenin-independent mode or through negative feedback loops - is essential for understanding this class and designing targeted interventions.
Combining CTNNB1 and Wnt Classification with Other HCC Molecular Subclasses The HCC molecular landscape includes other major subclasses (S1/Wnt-TGFbeta, S2/stem cell/EpCAM, S3/cell cycle/proliferative, G1-G6 Boyault classes). Mapping the CTNNB1 and Wnt-TGFbeta classes onto these parallel classification frameworks and onto clinical parameters (BCLC stage, etiology, CTNNB1 mutation, AFP level) in prospectively collected cohorts would build an integrated molecular-clinical framework suitable for trial design.