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

Cancer stem cell (CSC) persistence drives recurrence and therapy resistance in oral squamous cell carcinoma (OSCC), but what keeps cells locked in this stem-like state is poorly understood. In this narrative review, we propose that CSC state is sustained not by any single pathway but by joint dysregulation of three interacting cell-biological systems: membrane potential (Vmem), communication between neighbouring cells through gap junctional intercellular communication (GJIC), and the Hippo–YAP pathway. We argue that these systems act together on one common point—the YAP protein, retained in the nucleus—which switches on a SOX2-centred stemness gene programme and stabilises a self-reinforcing CSC state. Drawing on evidence from cancer genomics, developmental bioelectricity, connexin biology, and OSCC-specific studies, we reconstruct how membrane depolarisation, loss of gap junction coupling, FAT1 mutation, and Hippo pathway inactivation could converge on persistent nuclear YAP, and how betel quid—the principal risk factor across South and Southeast Asia—may engage all three systems at once. Because the model holds that each input reinforces the others, it predicts that targeting several together should displace CSC state more durably than targeting any one alone. We set out the testable predictions this framework generates.

Keywords: bioelectric signalling, cancer stem cells, FAT1, gap junctional intercellular communication, Hippo–YAP signalling, oral squamous cell carcinoma, oral submucous fibrosis

Oral squamous cell carcinoma (OSCC) remains a major cause of cancer-related morbidity worldwide, with recurrence and therapy resistance as persistent clinical challenges [1,2]. Cancer stem cells (CSCs)—a subpopulation endowed with self-renewal, tumorigenic potential, and therapy resistance—are central mediators of these failures [3,4]. The classical hierarchical CSC model, first demonstrated in acute myeloid leukaemia and extended to head and neck squamous cell carcinoma (HNSCC) [5], posited a genetically fixed stem apex generating non-tumorigenic progeny. This model cannot account for the phenotypic plasticity now documented across multiple systems: non-CSC populations regenerate CSC fractions after depletion, CSC states interconvert, and hybrid epithelial–mesenchymal configurations arise stochastically [6,7]. In OSCC specifically, such plasticity correlates with poor prognosis [8,9]. These observations are better accommodated by a systems biology framework in which CSC identity is not a lineage-fixed trait but a dynamic transcriptional state—one of several stable configurations, or attractors, within a high-dimensional gene regulatory landscape [10,11]. Malignant transformation, in this view, involves not merely the accumulation of somatic mutations but a disruption of the regulatory logic governing cell state transitions, locking cells into stem-like attractor basins that are self-reinforcing and difficult to exit.

The Hippo signalling pathway, a conserved regulator of epithelial homeostasis and cell fate, has emerged as a potent driver of CSC maintenance in OSCC. Genomic studies and genome-wide CRISPR fitness screens have identified YAP1 and TAZ, the transcriptional effectors of Hippo, as fitness genes in OSCC cell lines [12,13], and YAP activation has been mechanistically linked to SOX2-driven stemness programmes [14,15]. Yet Hippo–YAP does not operate in isolation: recent evidence positions it downstream of mechanosensory inputs, including transmembrane voltage gradients, with membrane depolarisation favouring YAP nuclear retention through a mechano-electro-osmotic mechanism [16]. Bioelectric signalling—mediated through membrane potential (Vmem), ion channel dynamics, and gap junctional intercellular communication (GJIC)—has emerged as a higher-order regulator of cell fate in both developmental and cancer biology [17,18]. Epithelial tissues maintain coherent bioelectric states through GJIC, through which individual cells integrate into collective mechanosensing fields; disruption of this electrical coordination has been associated with loss of differentiation control and cancer initiation [19]. These mechanisms are particularly relevant in betel quid-associated OSCC, which predominates across South and Southeast Asia. Betel quid induces oral submucous fibrosis (OSF) characterised by extracellular matrix (ECM) stiffening that activates Piezo1-mediated YAP nuclear translocation [20]; arecoline, its principal alkaloid, drives ROS generation, fibrosis, and EMT while directly upregulating YAP/TAZ in oral tissue compartments [21,22,23], positioning Hippo–YAP dysregulation as a functional consequence of betel quid carcinogenesis across multiple cell compartments.

In this narrative review, we propose that CSC persistence in OSCC is stabilised by the convergent dysregulation of three interconnected biophysical regulatory systems: tissue-level bioelectric signalling (Vmem), gap junctional intercellular communication (GJIC), and the Hippo–YAP pathway. We argue that these systems do not act independently but converge on a single molecular output—persistent nuclear YAP—which activates the TAZ–TEAD4–SOX2 gene regulatory network and locks oral epithelial progenitor cells into a self-reinforcing CSC attractor state. Drawing on evidence from genomics, developmental bioelectricity, connexin biology, and OSCC-specific studies, we reconstruct the mechanistic links between Vmem depolarisation, Cx43-mediated GJIC fragmentation, FAT1 loss-of-function, and Hippo kinase inactivation, and examine how betel quid exposure simultaneously engages all three dysregulatory axes. The framework generates specific, falsifiable experimental predictions and positions YAP–TEAD-directed combination therapy—targeting the convergence node while co-disrupting upstream biophysical inputs—as a rational strategy for durable CSC state reprogramming rather than cytotoxic elimination (Figure 1).

This narrative review was developed to construct and evaluate a mechanistic hypothesis rather than to provide an exhaustive systematic survey, and the literature was therefore identified through a purposive, theme-structured search as below.

Block A (disease): (“oral squamous cell carcinoma” OR OSCC OR “head and neck squamous cell carcinoma” OR HNSCC OR “oral submucous fibrosis” OR OSF).

Block B (Hippo–YAP): (“Hippo pathway” OR YAP OR YAP1 OR TAZ OR WWTR1 OR TEAD OR LATS OR FAT1).

Block C (bioelectric): (“membrane potential” OR Vmem OR bioelectric OR “ion channel” OR Piezo1 or mechanotransduction”).

Block E (CSC): (“cancer stem cell” OR stemness OR SOX2 OR “tumour-initiating cell”) searched individually and in pairwise or three-way combinations (e.g., A AND B AND C) in PubMed/MEDLINE and Google Scholar, English-language, to March 2026. As the central aim was to assess whether these traditionally separate fields intersect, particular attention was given to pairwise and three-way combinations of these concept blocks (for example, Hippo–YAP + bioelectricity, and Hippo–YAP + gap junctional communication). The scarcity of the literature jointly addressing all three pillars in the oral cancer context is itself a finding that motivates this review. Foundational and mechanistic studies from outside the oral cancer field, in particular, bioelectric signalling and Hippo regulation, were included as they established principles relevant to the proposed framework. The reference list was further developed by backward and forward citation tracking from key mechanistic papers. Sources were selected based on their relevance to the convergence hypothesis and mechanistic informativeness rather than by predefined systematic inclusion criteria, and the review is accordingly interpretive rather than exhaustive. To make explicit the balance between OSCC/HNSCC-specific evidence and principles from extrapolated fields, all cited sources are categorised by domain in Supplementary Table S1.

The canonical hierarchical model of cancer stem cells (CSCs), in which a genetically fixed apex subpopulation generates non-tumorigenic progeny, cannot account for a growing body of experimental observations: non-CSC populations reliably regenerate CSC fractions after depletion, phenotypically distinct CSC states interconvert, and hybrid epithelial–mesenchymal configurations fit neither pole of the differentiation axis [9,24,25]. In OSCC, single-cell transcriptomic profiling confirms that canonical CSC markers—CD44, ALDH1A1, and CD133—fluctuate across transcriptional continua rather than defining immutable subpopulations [26]. Vipparthi et al. demonstrated that hybrid epithelial–mesenchymal states are the predominant stem-like configuration in oral cancer and correlate with poor clinical outcomes [8]. These observations call for a framework that accommodates reversibility, stochastic state transitions, and phenotypic heterogeneity within genetically identical populations.

The gene regulatory network (GRN) attractor framework, developed by Huang, Ernberg, and Kauffman, provides this framework [10]. In dynamical systems terms, a cell’s phenotype is the product of GRN dynamics that channel cells toward a limited set of stable configurations—attractors—corresponding to distinct, reproducible cell phenotypes. The collection of attractors and the barriers between them constitute the gene regulatory landscape, a formal instantiation of Waddington’s epigenetic landscape metaphor [27]. Cancer arises, in this view, not only through somatic mutation but through deformation of this landscape: mutations lower energy barriers, rendering aberrant attractor states accessible, many of which resemble embryonic or progenitor configurations [6,28]. The critical implication is that the same genome, differently regulated, can generate both normal and malignant phenotypes—establishing the theoretical basis for state-reprogramming rather than purely cytotoxic therapeutic strategies. Intratumoral heterogeneity, long attributed to clonal genetic evolution, is now understood to have a substantial nongenetic component driven by GRN dynamics and stochastic state transitions [29].

The attractor model makes a specific, testable prediction: subpopulations perturbed by sorting or drug treatment should reconstitute the original phenotypic distribution as the system returns to its attractor basin. Li et al. confirmed this bidirectionally in a clonal breast cancer line—both CSC-enriched and non-CSC-enriched fractions regenerated the full phenotypic spectrum within days to weeks—establishing the hallmark behaviour of a dynamical system organised around stable attractors [30]. In OSCC, formal attractor mapping has not yet been performed, but indirect evidence is consistent with this organisation: the CancerSEA atlas reveals discrete co-occurring transcriptional modules (stemness, EMT, proliferation, stress) across OSCC-relevant data [26], and hybrid EMT states, which may represent intermediate basins, exhibit the highest plasticity and worst clinical outcomes [8].

The depth and accessibility of the CSC attractor basin are determined by the architecture of self-reinforcing GRN circuits. In OSCC, the TAZ–TEAD4–SOX2 positive feedback loop—in which nuclear YAP/TAZ transcriptionally activate SOX2, and SOX2 in turn suppresses Hippo-mediated YAP/TAZ phosphorylation by repressing upstream activators NF2 and WWC1—exemplifies this architecture [14,15]. Co-expression of OCT4 and NANOG with CSC markers in OSCC suggests that embryonic pluripotency GRN modules are reactivated as part of the malignant attractor programme [31], consistent with Huang and Kauffman’s prediction that cancer attractors resemble progenitor states encoded within conserved regulatory circuitry. Taken together, the OSCC CSC state is best understood as a dynamically stabilised GRN attractor maintained by self-reinforcing transcriptional circuits within a deformed regulatory landscape. The central question—what biophysical forces shape this landscape in OSCC—is the subject of the sections that follow.

The Hippo pathway operates as a kinase cascade: MST1/2 phosphorylate and activate LATS1/2, which phosphorylate YAP and TAZ, sequestering them in the cytoplasm via 14-3-3 binding and targeting them for proteasomal degradation [32]. When the pathway is inactive, unphosphorylated YAP and TAZ translocate to the nucleus and bind TEAD family transcription factors to drive proliferation, survival, EMT, and stem cell maintenance. Upstream regulation is rich: apicobasal polarity complexes, adherens junctions, and F-actin cytoskeletal tension all modulate LATS1/2 activity, positioning Hippo as an integrator of mechanical, polarity, and intercellular adhesion cues [33]. This mechanosensory function extends to transmembrane voltage gradients, as discussed in Section 4.5, a finding central to the three-pillar convergence hypothesis of this review.

The case for Hippo–YAP as a driver of OSCC biology begins at the genomic level. Comprehensive characterisation of HNSCC identified recurrent focal amplification at chromosome 11q encompassing YAP1, placing the primary Hippo effector within one of the most recurrently amplified regions in this cancer type [34]. Pan-cancer TCGA analysis confirmed that 14% of HNSCCs harbour combined YAP1 or TAZ amplifications (mutually exclusive pattern), and elevated YAP/TAZ transcriptional target gene activity (22-gene downstream signature) was independently associated with decreased overall survival [35]. FAT1, encoding a protocadherin that scaffolds the Hippo signalling complex, is among the most frequently mutated genes in HNSCC (29.8%), and its loss-of-function activates YAP1 through failure of MST1 membrane recruitment [36]. Atypical, non-canonical Hippo alterations independent of upstream kinase suppression have additionally been characterised as emerging hallmarks of head and neck cancer biology [12].

Functional evidence comes from a genome-wide CRISPR-Cas9 fitness screen of OSCC cell lines in which YAP1 and TAZ emerged as essential genes: genetic ablation induced apoptosis and substantially reduced viable cell populations, demonstrating a fitness addiction to sustained YAP/TAZ activity [13]. Transcriptional profiling of dependency classes revealed distinct programmes: YAP1-dependent tumours enriched for cell cycle signatures, WWTR1 (TAZ)-dependent tumours enriched for immune and interferon-response signatures. YAP1 was independently identified as a candidate oncogene on the chromosome 11q22 amplicon [37], and its oncogenic credentials are further supported by a systematic review of 13 OSCC studies finding pro-oncogenic direction in 12/13 (92%), with unfavourable direction in all studies examining EMT, migration, invasion, and in vivo tumour growth [38].

YAP functions not merely as a proliferative switch but as a reprogramming factor capable of inducing stem-like transcriptional states in committed epithelial cells. Omori et al. demonstrated using an inducible mouse model that endogenous YAP1 hyperactivation (through Hippo pathway inactivation) in oral epithelium is sufficient to drive the onset and progression of OSCC, establishing YAP1 as a potent oncogenic driver rather than a passive amplification bystander [39]. The most mechanistically resolved evidence for YAP as a CSC attractor-inducing factor comes from Faraji et al., who used spatiotemporally controlled oncogene activation with multiomics in KRT14+ oral epithelial progenitors [40]. Constitutively active YAP induced invasive oral carcinoma with high penetrance, and single-cell RNA sequencing identified convergence into a discrete tumour-initiating (TI) cell cluster absent from normal epithelia, enriched for YAP, mTORC1, E2F, MYC, partial EMT, and squamous differentiation signatures. TI cell programmes were conserved in human TCGA-HNSC tumours where above-median YAP and mTORC1 activity independently associated with worse overall and disease-free survival. YAP directly activated NRG1 and AXL (EGFR/HER3 axis driving mTOR signalling) and PDPN (pEMT marker), and mTOR inhibition substantially reduced carcinoma formation in vivo, identifying a YAP→mTOR vulnerability with direct therapeutic implications.

At the histologic and clinical level, Amano et al. performed the first comprehensive IHC characterisation of major Hippo pathway proteins across OED, CIS, and OSCC (n = 109): MST1, LATS1, and LATS2 were consistently low across all stages, while YAP1 nuclear expression increased with tumour grade and infiltrative invasion pattern, correlating with EMT markers (E-cadherin loss, vimentin, Slug) and with PRMT1/5 co-expression as an independent predictor of relapse-free survival [41]. YAP and TAZ are essential for basal and squamous cell carcinoma initiation [42], and nuclear YAP consistently correlates with aggressive tumour behaviour, lymph node metastasis, and reduced survival across the OSCC clinical literature. This body of evidence collectively positions nuclear YAP as the convergence output of upstream biophysical dysregulation and as the functional engine of the OSCC CSC attractor state.

The mechanistic bridge between Hippo–YAP pathway activity and the CSC gene regulatory network is most clearly established through the TAZ–TEAD4–SOX2 transcriptional axis. Li et al. demonstrated that in head and neck squamous cell carcinoma, TAZ directly binds to TEAD4 and the complex occupies the transcriptional promoter of SOX2, driving its expression and thereby activating the broader pluripotency-associated GRN that sustains CSC self-renewal [14]. SOX2, a master transcriptional factor of embryonic and adult stem cell identity, is a critical node in the OSCC CSC GRN: its overexpression is sufficient to promote tumour initiation/sphere formation in squamous carcinoma models [43], and its expression correlates with poor prognosis in OSCC patients [44]. The TAZ–TEAD4–SOX2 circuit therefore represents a direct transcriptional mechanism by which Hippo pathway inactivation, signaled by loss of upstream kinase activity or by mechanical/bioelectric cues that prevent LATS1/2-mediated phosphorylation, translates into activation of the stemness GRN.

The feedback architecture of this circuit is equally important. SOX2 has been shown to antagonise the Hippo pathway through direct transcriptional repression of two upstream Hippo activators, NF2 (Merlin) and WWC1 (Kibra), thereby relieving upstream restraint on YAP and reinforcing its nuclear retention [15]. It should be noted that this mechanism was established in osteosarcoma and glioblastoma cell lines; direct evidence in oral epithelial or HNSCC models has not been reported. This creates a double-positive feedback loop—TAZ activates SOX2, and SOX2 in turn stabilises TAZ/YAP nuclear activity—that constitutes precisely the kind of self-reinforcing transcriptional circuit that, in the attractor framework described in Section 3, deepens the CSC attractor basin and raises the energy barrier to exit.

Beyond direct activation of the stemness gene programme through TAZ–TEAD4–SOX2, nuclear YAP additionally drives EMT-associated transcriptional reprogramming in oral epithelium. Xu et al. demonstrated that ECM stiffness-induced Piezo1 activation in oral epithelial cells produces YAP nuclear translocation that upregulates vimentin and downregulates E-cadherin, the canonical EMT transcriptional signature [20]. EMT, particularly in hybrid epithelial–mesenchymal configurations, is closely associated with CSC self-renewal, plasticity, and tumour-initiating capacity in OSCC and oral epithelial cancer, with single-cell transcriptomic studies in OSCC documenting hybrid epithelial–mesenchymal states enriched for CSC markers [8]. The YAP convergence node therefore drives at least two complementary transcriptional outputs that together produce the full CSC phenotype: the TAZ–TEAD4–SOX2 axis activates stemness genes directly, while the YAP–EMT axis produces the phenotypic plasticity characteristics of CSC states. SOX2 functions as a shared transcriptional integrator of both axes, since it has additionally been implicated in driving mesenchymal gene expression while self-amplifying through NF2/WWC1 repression. Although this dual-axis architecture has not been formally tested in OSCC, the convergent activation of stemness and EMT programmes through a shared YAP–SOX2 transcriptional core provides a parsimonious mechanism for the depth and self-reinforcing nature of the OSCC CSC attractor state, and predicts that experimental disruption of either axis alone, without simultaneous disruption of the other, would produce only transient CSC displacement.

Direct support for both arms of this dual SOX2 role has since been reported in OSCC. In CAL27 and SCC15 tongue squamous cell carcinoma cells, transient SOX2 knockdown reduced the mesenchymal markers vimentin and SNAIL, with a modest increase in E-cadherin, while simultaneously lowering 3D sphere-forming capacity and cell migration and attenuating AKT signalling [45]. These data substantiate, within oral tissue, both stemness-maintaining and EMT/invasion-promoting functions attributed to SOX2 in the convergence model and link the SOX2 node to the PI3K/AKT/mTOR axis discussed in Section 7. The SOX2→NF2/WWC1 repression that closes the self-reinforcing loop, however, remains established in osteosarcoma and glioblastoma; confirming this specific repression in oral cells is identified as a priority in Section 9.1.

Li et al. further showed that TAZ promotes EMT and CSC maintenance in oral cancer through regulation of EMT-associated genes including vimentin and fibronectin, connecting the Hippo–SOX2 axis to the broader phenotypic plasticity programme [44]. The co-amplification of YAP1 and SOX2 at chromosomes 11q and 3q, respectively in HNSCC, both within the most recurrently gained chromosomal regions [34], is consistent with a model in which genomic co-amplification reinforces the TAZ–SOX2 positive feedback loop at the gene dosage level [34].

A critical and recently published mechanistic insight positions Hippo–YAP not only as a downstream effector of mechanical inputs but as a sensor of transmembrane voltage states, directly connecting the bioelectricity pillar of this review’s hypothesis to the Hippo–YAP pillar at a molecular level. Mukherjee et al. demonstrated in a landmark Cell 2026 study that membrane potential mediates the cellular response to mechanical pressure through a mechano-electro-osmotic mechanism with direct consequences for YAP nuclear localisation [16].

In their model, a reduction in tissue biomass density, as occurs during wound healing when cells are mechanically stretched, triggers membrane depolarisation, promotes YAP nuclear translocation (via FAT1-mediated disassembly of the MST1 signalling complex), stimulating cell growth and cycle progression to restore tissue homeostasis. Conversely, high tissue density imposes mechanical compression and membrane hyperpolarisation, which activates MST1, leading to LATS1/2-mediated YAP phosphorylation and cytoplasmic sequestration, thereby suppressing proliferation. This bidirectional, causal connection between membrane potential and Hippo pathway activity established Vmem as an upstream regulatory input to the Hippo kinase cascade; not merely a correlate of cell state but an active determinant of YAP nuclear availability. Importantly, these experiments were conducted in kidney (MDCK) and mammary epithelial cell lines (MCF10A, EpH4-Ev, primary HMECs); direct demonstration of this mechanism in oral epithelial or OSCC systems has not yet been reported.

To our knowledge, no study has yet demonstrated the Vmem → MST1 → Hippo mechanism directly in oral epithelial or HNSCC cells. The evidence in oral tissue is at present indirect, comprising the Piezo1-mediated mechanotransduction route reported by Xu et al. in oral submucous fibrosis and the established genomic and functional roles of FAT1 and Hippo effectors in OSCC [13]. The bioelectric-to-Hippo connection in oral tissue is therefore proposed here as an extrapolation requiring direct experimental confirmation, not an established fact.

A second feature of the Mukherjee mechanism, underemphasised but critical for the OSCC context, is that bioelectric homeostasis in epithelium operates at two distinct scales. At the single-cell scale, Vmem responds to mechanical pressure through the mechano-electro-osmotic coupling described above. At the tissue level, Mukherjee et al. observed that this Vmem-dependent growth control is calibrated against the achievement of tight barrier function, with growth proceeding until barrier integrity establishes the homeostasis endpoint signal that constrains further proliferation [16]. Tight junctions, through their role in establishing transepithelial potential difference (TEPD), generate the tissue-level bioelectric reference that single-cell Vmem signalling drives forward. Saw et al. independently demonstrated that TEPD itself governs epithelial homeostasis through electromechanical coupling [46]. This two-scale architecture has direct implications for the OSF-to-OSCC continuum: OSF is characterised by epithelial atrophy, loss of normal architecture, and disruption of barrier integrity, all of which would compromise the tissue-level TEPD that normally anchors the homeostatic feedback signal. In OSF tissue, oral epithelial cells may therefore experience not only individual Vmem dysregulation but also the loss of tissue-level reference that tells them when proliferation should cease—a homeostatic system without a reachable endpoint.

The Mukherjee mechanism additionally couple bioelectric and mechanotransductive signalling at the molecular level. Mukherjee et al. note that membrane potential is well-positioned to regulate any signalling pathway originating at the plasma membrane, and they specifically identify calcium signalling as an “ideal effector” of Vmem-mediated mechanical transduction, citing the strong voltage-dependence of calcium ion flux. Critically, they observe that flux through key calcium channels including Piezo is voltage-dependent, with Piezo channels themselves shown to be voltage-gated [16]. This intrinsic voltage-sensitivity of Piezo, demonstrated experimentally by Moroni et al., means that Vmem state co-regulates Piezo channel calcium influx in response to a given mechanical stimulus [47]. The Piezo1-mediated mechanotransduction route demonstrated by Xu et al. [20] in OSF therefore does not operate independently of cellular Vmem state. In OSF/OSCC, where both Vmem dysregulation and ECM stiffening occur simultaneously, the bioelectric and mechanotransductive systems are not parallel insults but coupled signalling architectures whose dysregulation compounds at the channel level.

The implications for OSCC carcinogenesis are direct. Ion channel dysregulation, a well-documented feature of cancer cells, alters membrane potential in ways that may render cells unresponsive to the mechanical cues that normally activate the Hippo pathway and suppress YAP [19]. A depolarised membrane state, whether arising from ion channel mutation, altered expression, or disrupted gap junctional coupling, would—by extension of the Mukherjee biomass-density model, though this specific extrapolation has not yet been directly demonstrated in cancer—be expected to favour YAP nuclear retention independently of tissue density cues. In the context of the OSF-to-OSCC continuum, the Piezo1-mediated mechanosensory pathway identified by Xu et al. [20] provides a complementary upstream route—increased ECM stiffness activates Piezo1, driving calcium influx and YAP nuclear translocation, priming the oral epithelium toward a pro-malignant configuration. Together, the bioelectric and mechanotransductive pathways, coupled at Piezo1 voltage-gating and converging at intracellular calcium, suggest that the oral epithelium in betel quid-associated OSCC is subject to convergent biophysical inputs that both favour sustained YAP nuclear activity and CSC attractor stabilisation.