A Convergence Model of Bioelectric, Gap Junctional, and Hippo-YAP Signalling in Oral Cancer Stem Cell Maintenance

The bioelectric and GJIC pillars imply upstream reprogramming strategies distinct from direct YAP targeting. Pharmacologic hyperpolarisation using Kv-channel openers (retigabine, minoxidil) is predicted, via the Mukherjee Vmem–MST1 mechanism, to activate LATS1/2 → YAP phosphorylation and reduce CSC frequency in OSCC models—an experiment not yet performed. For GJIC restoration, danegaptide and rotigaptide (connexin-stabilising peptidomimetics developed for cardiac indications) would be predicted to restore Ca2+ wave propagation, re-engage CaMKII–MST1 activation in response to cell density cues, and reduce nuclear YAP independently of direct TEAD inhibition [51]. Neither class has been tested in OSCC CSC models; both are tractable with available tools. miRNA-based strategies targeting YAP post-transcriptionally (miR-27a-3p, miR-381-3p, circHIPK3) have shown preclinical activity but face delivery challenges [38]. miR-27a-3p, in particular, was originally identified by Zeng et al. as a direct post-transcriptional suppressor of YAP1 in OSCC cell lines, with restoration of miR-27-3p reducing the YAP1–OCT4 interaction, downregulating SOX2 and EMT transcription factors and impairing invasive capacity, providing OSCC-specific primary evidence for the miRNA–YAP axis as a tractable therapeutic node within the convergence framework [74]. Because several convergence-directed modalities, notably nucleic-acid agents such as YAP-targeting miRNAs and locally acting connexin or ion channel modulators, face bioavailability and tumour-selectivity constraints, nanoparticle- and nanomaterial-based delivery systems represent a relevant avenue for improving their stability, targeting, and clinical translatability [75].

If the three pillars are organised as a convergence architecture instead of as parallel inputs (Section 8.1), single-system targeting is predicted to be insufficient for durable CSC attractor displacement as the remaining stabilising inputs would continue to reinforce nuclear YAP retention. This prediction is therefore conditional on the convergence architecture itself, which remains to be tested (Section 9.1). Combination strategies that simultaneously address multiple inputs—TEAD + mTOR, TEAD + CXCR1/2, Kv-channel opener + TEAD, or danegaptide + TEAD—are each mechanistically motivated and directly testable in OSCC organoids with CSC frequency and post-withdrawal recurrence as primary endpoints; a finding of non-additive, synergistic benefit from such combinations would itself provide functional support for the convergence architecture, whereas merely additive effects would be equally consistent with parallel regulation. The field cancerization dimension of OSCC pathogenesis further argues for the chemopreventive application of these strategies: the OSF stage, where bioelectric dysregulation, GJIC fragmentation, and Piezo1-mediated YAP activation are each mechanistically plausible but not yet locked into a stable CSC attractor basin, represents a clinically identifiable intervention window with no current pharmacologic equivalent. Bioelectric and GJIC biomarkers that predict malignant conversion in OSF patients would constitute a clinically meaningful output of the mechanistic framework developed here (Figure 3).

Although the convergence framework is mechanistic in construction, its most immediate translational value lies in three clinically actionable directions. First, because the convergence operates substantially at the post-translational level, its readouts are protein-localisation and channel-function assays already routine in diagnostic pathology. Nuclear versus cytoplasmic YAP/TAZ, membrane versus cytoplasmic Cx43 [52], and SOX2 positivity together index the activity of the proposed convergence node and could be combined into a single immunohistochemical panel. Applied to oral potentially malignant disorders, and to OSF, in particular, such a panel offers a candidate biomarker of malignant-conversion risk, distinguishing lesions that are bioelectrically and junctionally “primed” from those that are not, addressing an unmet need, since current OSF surveillance relies on dysplasia grading, a weak predictor of progression [76,77,78,79]. Consistent with such a primed field, SOX2 has been found upregulated in the close resection margins of OSCC relative to healthy mucosa, where its expression correlates with both tumour size and lymph-node compromise [45]. Direct human evidence is that a stemness-associated programme is already active in marginal tissue, supporting the biologically altered field that this prevention argument presupposes.

Second, the framework defines a molecularly selected population for convergence-directed therapy. FAT1 loss-of-function (~30% of HNSCC) and YAP1/WWTR1 amplification (11q/3q) are recurrent, routinely assayable events that the model predicts should sensitise tumours to YAP–TEAD-directed therapy; FAT1-altered HNSCC has already been nominated as an enriched-response population for TEAD inhibitors [36]. Coupled with the protein-level panel above, these markers could support biomarker-stratified enrolment in early-phase trials of clinical-stage TEAD inhibitors (IK-930, VT3989) and their combinations. This stratification cuts in both directions. Because the upstream bioelectric and gap junctional interventions considered here act on Hippo through the FAT1-dependent Vmem → MST1 → LATS1/2 route, they are predicted to be effective principally in FAT1-intact tumours; in FAT1-altered tumours, where that route is uncoupled, direct YAP–TEAD inhibition is the more logical intervention. The model therefore predicts a FAT1-status-dependent division of therapeutic strategy (upstream biophysical modulation for FAT1-intact disease and direct nuclear-YAP/TEAD targeting for FAT1-altered disease)—a concrete, assayable prediction since FAT1 status is a routinely obtainable genomic readout.

Third, and most distinctively, the framework identifies the OSF stage as a clinically detectable chemoprevention window in a population of approximately 600 million betel quid users for whom approved targeted antifibrotic or molecularly guided preventive strategies are still lacking [80,81,82,83]. Because several of the implicated upstream agents are already approved or in clinical development for non-oncological indications, for example, Kv-channel openers (retigabine, minoxidil), connexin-stabilising peptidomimetics (danegaptide, rotigaptide), and Piezo1 inhibitors, the convergence logic is unusually amenable to repurposing at the prevention stage, before the CSC attractor is established. A computational companion to this review (manuscript in preparation) formalises the convergence architecture as a dynamical (Boolean and reduced-ODE) model and uses it to rank these single-agent and combination strategies by predicted durability; in that analysis, a non-additive, synergistic response to specific combinations emerges as the signature that would distinguish a true convergence architecture from parallel regulation, and prioritises these combinations for the experimental agenda in Section 9.1.

The convergence framework developed here generates five prioritised experimental questions, each tractable with existing tools and OSCC-relevant model systems. First, formal attractor landscape mapping in OSCC—using single-cell RNA sequencing, GRN reconstruction (SCENIC, Arboreto), and phenotypic reconstitution experiments—is needed to confirm that the CSC state constitutes a discrete transcriptional attractor basin. The Faraji et al. scRNA-seq datasets (GEO: GSE276778–GSE276783) provide a genetically defined oral epithelial CSC entry system for cross-species attractor alignment. Second, electrophysiological characterisation of resting Vmem in OSCC CSC versus non-CSC subpopulations, combined with pharmacologic hyperpolarisation using Kv-channel openers, would directly test whether Vmem manipulation alters YAP nuclear localisation and CSC frequency in an OSCC-specific system. Third, connexin isoform profiling and functional GJIC measurement (FRAP assay) across OSCC CSC and non-CSC subpopulations, followed by pharmacologic GJIC restoration using danegaptide, would test whether GJIC re-engagement reduces CSC frequency through Hippo kinase reactivation without requiring direct YAP inhibition. Fourth, characterisation of Piezo1–YAP axis activation in OSF-derived primary epithelial cells and patient organoids on stiffness-graded matrices would test whether Piezo1 inhibition (Dooku1) prevents CSC state enrichment at the OSF stage, a potential chemoprevention window.

Fifth, systematic preclinical combination testing—pairing TEAD inhibition with mTOR inhibition, CXCR1/2 blockade, Kv-channel openers, or GJIC restoration—in OSCC cell lines and patient-derived organoids, with CSC frequency and post-withdrawal recurrence as primary endpoints, would test the core prediction of the convergence model: that multi-system perturbation produces the non-additive, durable CSC attractor displacement that single-pathway targeting cannot achieve. Collectively, these experiments would transform the framework from synthesis to validated mechanistic model and define the evidence base for a biomarker-selected OSCC trial of TEAD-directed combination therapy.

Cancer stem cells in OSCC are best understood not as a fixed, genetically distinct subpopulation but as a dynamic transcriptional attractor state, a stable configuration of the gene regulatory network that is self-reinforcing, resistant to displacement and continuously regenerated from the tumour populations. The persistence of this attractor state, and the failure of conventional CSC-targeting strategies to durably eliminate it, reflects the stability of the regulatory landscape that sustains it rather than the intrinsic resilience of any particular cell.

This review has proposed that three interconnected regulatory systems shape this landscape in OSCC. The Hippo–YAP pathway—genomically amplified, functionally essential, and mechanosensorily responsive—is the transcriptional stabiliser of the CSC attractor basin, sustaining nuclear YAP, activating the TAZ–TEAD4–SOX2 positive feedback circuit, and driving the mTORC1 and pEMT programmes that define the tumour-initiating cell state. Bioelectric signalling—mediated through membrane potential gradients and ion channels dynamics—acts as an upstream regulator of Hippo pathway activity, with Vmem depolarisation favouring YAP nuclear retention through the mechano-electro-osmotic mechanism recently established by [16] GJIC—the tissue-level integrator of individual bioelectric states—maintains the collective error-correction capacity that normally prevents cells from sustaining aberrant attractor states; its fragmentation, through connexin downregulation and post-translational channel inactivation driven by oncogenic Src/EGFR signalling, removes this protective mechanism and allows local epithelial clusters to enter the CSC basin independently of tissue-level cues.

In the context of betel quid-associated OSCC, which predominates across South and Southeast Asia—and for which no molecularly targeted prevention strategy currently exists—these three systems converge with particular biological plausibility. Betel quid exposure induces OSF, with ECM stiffening that activates Piezo1-mediated YAP nuclear translocation: arecoline generates ROS that alter ion channel function and shift Vmem toward depolarisation; fibrotic remodeling constrains gap junction plaque formation and fragments GJIC. Each of these is a biophysical consequence of betel quid exposure that, according to the framework developed here, contributes to deepening the CSC attractor basin before frank malignancy is clinically detectable. The OSF stage therefore represents both a mechanistic entry point into the attractor landscape and a clinical window for chemopreventive intervention, one that the experimental priorities outlined in Section 9.1 are designed to exploit.

The direct experimental evidence linking all three systems simultaneously in OSCC is absent, and this absence is the central finding of this review. It is not a weakness but a frontier. The mechanistic framework is internally coherent, grounded in primary evidence from multiple converging fields, and generates specific, falsifiable predictions that are tractable with existing experimental tools and publicly available genomic resources. Critically, because the convergence mechanism operates substantially at the post-translational level, YAP nuclear localisation, ion-channel gating, and connexin channel assembly, its predictions are most appropriately tested by assays of protein localisation and channel function rather than transcript abundance alone. A first transcript-level evaluation of the convergence predictions in public OSCC single-cell datasets is underway [84]; consistent with the post-translational nature of the proposed mechanism, it indicates that the convergence is not resolvable at the level of transcript co-regulation and reinforces the need for the protein- and functional-level assays prioritised in Section 9.1. Demonstrating that Vmem dysregulation alters CSC frequency in OSCC, that GJIC restoration re-engages Hippo pathway suppression, and that multi-system combination targeting achieves more durable CSC attractor displacement than single-pathway inhibition, these experiments would transform the framework from synthesis into a validated mechanistic model.

The therapeutic implication is correspondingly clear. If the CSC state in OSCC is stabilised by a convergent regulatory architecture that cannot be permanently disrupted by targeting any single component, then durable therapeutic reprogramming requires simultaneous or sequential perturbation of the landscape itself, deforming the attractor basin rather than depleting its inhabitants. The convergence node that makes this tractable is nuclear YAP: a protein whose activity is regulated by all three upstream systems, whose inhibition by TEAD-blocking small molecules is now entering clinical translation, and whose downstream transcriptional program—including mTOR activation via NRG1/AXL, pEMT programming via PDPN, and SOX2-mediated self-renewal—is now characterised at single-cell resolution in the oral epithelial progenitor cell that gives rise to OSCC. Targeting this node, informed by the bioelectric and gap junctional context that determines its upstream regulation, represents a rational and mechanistically grounded strategy for therapeutic reprogramming of OSCC that has not yet been clinically pursued. This review provides the conceptual foundation for that pursuit.

Declaration of Generative AI and AI-assisted technologies in the writing process. During the preparation of this manuscript, the authors used Claude (Anthropic) for language editing, narrative structuring, citation cross-checking, iterative refinement of mechanistic arguments, and generation of figures. The authors conceived the convergence framework, conducted the literature analysis, made all intellectual and scientific decisions, and reviewed and edited all output. The authors take full responsibility for the content of the publication.

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/ijms27156649/s1.

S.K.A.—conceptualization, literature search/filtering, manuscript writing/formatting, reviewed and verified figures construction; W.C.N.—manuscript review and revision; F.H.—manuscript review and revision; F.F.L.—manuscript review and revision, reviewed and verified figures construction, Y.F.C.—conceptualization, literature search/filtering, manuscript writing, figures construction. All authors have read and agreed to the published version of the manuscript.

No new data were created or analyzed in this study. Data sharing is not applicable to this article.

No new data were created or analyzed in this study. Data sharing is not applicable to this article.