MITF-Driven melanoma plasticity as a core mechanism of therapy resistance: integrating microenvironmental signaling, mechanotransduction, and metabolic reprogramming

ABSTRACT

Melanoma remains one of the most aggressive cancers, and although immune checkpoint blockade and MAPK-targeted therapies have transformed clinical management, durable responses occur in only a subset of patients. Converging evidence identifies microphthalmia-associated transcription factor (MITF) – dependent phenotype switching as a central, non-genetic mechanism enabling melanoma cells to escape therapy. Dynamic fluctuations in MITF activity permit transitions between differentiated, proliferative states and invasive, drug-resistant phenotypes. This review synthesizes emerging insights into the tumor microenvironmental, mechanical, and metabolic cues that regulate MITF states. These include cytokine-driven inflammatory signaling, hypoxia, cancer-associated fibroblasts, extracellular matrix remodeling, integrin – YAP/TAZ – mediated mechanotransduction, and metabolic reprogramming involving glycolysis – OXPHOS switching, lipid-regulated MITF control, and nutrient-stress responses. By integrating these pathways, MITF-dependent plasticity shapes melanoma adaptation and persistence under therapeutic pressure. Understanding this interconnected network provides a foundation for developing strategies to target phenotype switching and overcome treatment resistance.

1. Introduction

Melanoma is a highly aggressive skin cancer and remains a major global health challenge, with steadily rising incidence and mortality. According to GLOBOCAN 647 000 new cases and 58 645 deaths were reported worldwide in 2022 [1]. The introduction of ipilimumab in 2011, the first immune checkpoint inhibitor (ICI), marked a turning point in melanoma treatment, leading to a significant increase in overall survival [2]. Subsequent advances in targeted therapies, particularly v-Raf murine sarcoma viral oncogene homolog B (BRAF) and mitogen-activated extracellular kinase (MEK) inhibitors, have further expanded therapeutic options for advanced-stage disease. Nevertheless, durable responses remain limited because the majority of initially responding patients eventually develop resistance [3]. This resistance is increasingly attributed to non-genetic mechanisms, one of which is phenotypic plasticity [4]. Phenotypic plasticity refers to the capacity of melanoma cells to alter their phenotype in response to environmental or genetic changes. Cells exhibiting such plasticity may subsequently undergo phenotype switching, which describes dynamic transitions between distinct transcriptional and functional states. This process shares conceptual similarities with epithelial-mesenchymal transition (EMT), although phenotype switching in melanoma does not imply a directional or binary transition between two fixed states. Instead, melanoma cells can occupy multiple intermediate phenotypic programs and transition reversibly between them, conceptually resembling partial EMT states and the reverse process, mesenchymal-epithelial transition (MET), described in epithelial cancers [5,6]. This adaptive behavior is largely orchestrated through shifts in the expression and activity of lineage-specific regulators, most notably the microphthalmia-associated transcription factor (MITF), originally identified as a key regulator of melanocyte development [7]. Variations in MITF expression and activity underpin melanoma cell-state regulation, linking MITF to proliferation, differentiation and invasive capacity [8–12].

The foundation for understanding melanoma heterogeneity was established by the MITF rheostat model, which proposed that distinct melanoma phenotypes are associated with different levels of MITF activity, including proliferative and invasive states [12]. Advances in transcriptomics and in vivo modeling soon expanded this view, demonstrating that melanoma occupies a broader spectrum of cell states [13–17]. A recent refinement introduced a revised MITF rheostat model defining six melanoma phenotypic states arranged along a continuous gradient of MITF activity, which are hyper-differentiated, melanocytic, intermediate, starved, neural crest stem cell (NCSC)-like, and undifferentiated [18]. These states form a continuous spectrum, with melanoma cells capable of transitioning between them dynamically and reversibly (Figure 1). In general, MITF^High cells are characterized by proliferation, melanocytic differentiation, and relative therapeutic sensitivity, whereas MITF^Low cells exhibit slow cycling, invasiveness, and resistance to both BRAF/MEK inhibitors and immunotherapies. Importantly, resistance also occurs in subsets of highly differentiated MITF^High cells, indicating that therapy resistance in melanoma cannot be attributed exclusively to MITF^Low states [18]. These phenotypic states coexist within melanoma and interact with one another, collectively sustaining tumor adaptation,survival, and progression. MITF activity is further regulated by post-translational modifications including phosphorylation, ubiquitination, SUMOylation and acetylation [19,20]. These modifications may help explain how similar levels of MITF activity can drive distinct phenotypic programs, as BRAF-regulated acetylation shifts MITF DNA-binding from differentiation- toward proliferation-associated target genes [20,21]. In vivo studies using a temperature-sensitive zebrafish mitfa mutant also demonstrate the functional importance of MITF levels, showing that reduced MITF cooperates with BRAF^V600E to drive melanoma development, whereas its inactivation induces tumor regression [22].

Figure 1.

Schematic representation of melanoma phenotypic states along a conceptual MITF activity spectrum.

Beyond cell-intrinsic regulation of MITF, a wide array of extrinsic factors originating from the tumor microenvironment (TME), such as hypoxia, nutrient restriction, inflammatory cytokines and biomechanical signals, further regulate phenotypic switching in melanoma by modulating MITF activity, enabling melanoma cells to adapt dynamically to environmental stressors [23]. Although these pathways are discussed separately for clarity, they should be viewed as interconnected and mutually reinforcing, with their convergence on MITF representing a central organizing principle of this review.

This review aims to synthesize current evidence on how microenvironmental, mechanical, and metabolic cues regulate MITF-dependent melanoma plasticity and therapy resistance. Where applicable, we distinguish between mechanisms supported by robust clinical or multiple independent preclinical studies and those based on limited or preliminary evidence, highlighting where findings await clinical validation. Literature was identified through searches of PubMed and Google Scholar using combinations of keywords including “MITF,” “melanoma plasticity,” “phenotype switching,” “tumor microenvironment,” “inflammatory signaling,” “hypoxia,” “mechanotransduction,” “metabolic reprogramming,” “therapy resistance,” and “targeted therapy.” Priority was given to original research articles, while relevant review articles were included when they provided mechanistic context or comprehensive overviews of a given topic.

From lowest to highest MITF levels, the model comprises six interconvertible phenotypic states: undifferentiated, neural crest stem cell-like, starved, intermediate, melanocytic and hyper-differentiated. High-MITF states are characterized by enhanced proliferative capacity, elevated expression of melanocytic lineage markers, and increased sensitivity to targeted therapies and immune checkpoint blockade, reflecting a more differentiated and therapy-responsive phenotype. In contrast, low-MITF states are associated with a quiescent, slow-cycling behavior, heightened invasive potential, metabolic reprogramming, and robust resistance to both MAPK pathway inhibitors and immunotherapies, underscoring their contribution to tumor heterogeneity and therapeutic failure [18].

2. The tumor microenvironment as a regulator of MITF states

Melanoma constitutes a dynamic and heterogeneous network in which malignant cells coexist and actively communicate with a TME. The TME consists of diverse cellular components, including cancer-associated fibroblasts (CAFs), adipocytes, keratinocytes, and multiple immune cell populations, such as tumor-infiltrating lymphocytes (TILs) and tumor-associated macrophages (TAMs), as well as non-cellular elements forming extracellular matrix (ECM). Bidirectional interactions between melanoma cells and the TME occur through direct cell – cell and cell – matrix contacts, but also through mediators including cytokines, chemokines and growth factors, affecting tumor proliferation, migration and invasion [24–26]. Increasing evidence indicates that the TME plays a central role in regulating phenotypic plasticity and contributing to the acquisition of drug resistance through a variety of biochemical, mechanical and inflammatory signals [27].

2.1. Inflammatory pathways regulating MITF-Dependent phenotype switching

Inflammatory signaling within the TME has emerged as a major driver of phenotype switching in melanoma, acting on the intrinsic plasticity of melanoma cells to promote dedifferentiation and immune escape. Cytokines and other mediators produced in immune-cell rich melanomas diminish T-cell expansion and function, thereby enabling tumor cells to evade elimination [28].

Tumor necrosis factor alpha (TNF-α) is a key mediator of this transition, directly inducing dedifferentiation in both mouse and human melanoma cells [29,30]. TNF-α activates transcription factor c-Jun in tumor cells, which suppresses MITF and diminishes the transcription of melanocytic differentiation genes. MITF and c-Jun form a regulatory loop in which inflammatory factors lower MITF, and in turn melanoma cells become more permissive to c-Jun induction. Elevated c-Jun then amplifies cytokine-driven signaling. Through this feedback mechanism, a temporary inflammatory trigger may be sustained into a persistent MITF^Low/c-Jun^High state, which is associated with heightened cytokine sensitivity, increased myeloid-cell recruitment and a more invasive phenotype, as demonstrated in preclinical models [31]. Melanoma cells in a MITF^Low/nuclear factor kappa B (NF-κB)^High state display marked resistance to mitogen-activated protein kinase (MAPK) pathway inhibition, whereas cells that maintain high MITF expression remain sensitive to therapy. Prolonged MAPK blockade can also shift melanoma cells toward this resistant state, similar to the one triggered by TNF-α exposure [32]. Notably, macrophage-derived TNF-α can maintain MITF expression in melanoma cells and promote a survival program that protects them from MEK/BRAF-inhibitor-induced apoptosis. In this therapeutic context, a MITF^High state supports resistance [33]. The apparently contradictory effects of TNF-α on MITF may be explained by the kinetics of the inflammatory response. Early TNF-α signaling may transiently modulate MITF expression, whereas prolonged inflammatory stress induces phosphorylation of eukaryotic initiation factor-α (eIF2α) and activation of the integrated stress response, leading to translational reprogramming and suppression of MITF expression [34]. Thus, both MITF^Low and MITF^High melanoma states may contribute to therapy resistance, albeit through distinct mechanisms. Dedifferentiated MITF^Low states are frequently associated with invasiveness and immune evasion, whereas MITF^High states may support tumor survival and persistence under therapeutic pressure [31,33].

Similarly, interferon gamma (IFN-γ), produced by tumor-infiltrating T-cells and myeloid cells, shifts melanoma cells toward a dedifferentiated MITF^Low state. IFN-γ suppresses MITF in vitro by interfering with cAMP response element-binding protein (CREB) activation at the MITF promoter [35]. As a result, a global suppression of protein translation in melanoma cells is induced, triggering an amino-acid deprivation stress response that activates activating transcription factor 4 (ATF4). This leads to establishment of a characteristic ATF4^High/MITF^Low transcriptional state, which has been associated experimentally with increased melanoma susceptibility to cytotoxic T lymphocytes (CTL) [36]. Concurrently, IFN-γ induces immunomodulatory changes such as CD271 and programmed death-ligand 1 (PD-L1) upregulation, reducing antigen expression and limiting CTL activation [37]. Persistent IFN-γ signaling may stabilize this state through chromatin remodeling. This reversible shift from a MITF^High program to a MITF^Low phenotype was also associated with improved outcomes in patients receiving anti-PD-1 therapy, based on paired pre- and post-treatment biopsies [38].

Transforming growth factor β (TGF-β) displays a well-established dual role in cancer, acting as a tumor suppressor in early disease but functioning as a potent promoter of invasion, EMT, metastasis and immune evasion in advanced cancers [39]. Transcriptomic analysis of patient biopsies and cell line experiments indicate that TGF-β can drive melanoma toward a MITF^Low/AXL receptor tyrosine kinase (AXL)^High dedifferentiated phenotype characterized by loss of melanocytic antigens and reduced major histocompatibility complex (MHC) class I expression, thereby diminishing tumor immunogenicity. This state constitutes a common route to both innate and acquired resistance to programmed death receptor 1 (PD-1) inhibitors [40]. This phenotype may also predict broad resistance to targeted therapies, particularly in BRAF- and neuroblastoma RAS viral oncogene homolog (NRAS)-mutant melanoma [41]. Moreover, TGF-β signaling under SMAD7 (mothers against decapentaplegic homolog 7) loss in melanoma models enables a MITF^High/AXL^High state that is simultaneously highly proliferative and invasive, suggesting that AXL-mediated metastatic potential can emerge independently of MITF downregulation [42].

Moreover, interleukin-1 beta (IL-1β) can suppress MITF and melanocytic antigen expression by inducing microRNA-155, driving a dedifferentiated phenotype that impairs CTL recognition. In a subset of melanomas, autocrine IL-1 production maintains a stable MITF^Low inflammatory phenotype, reinforcing an immune-evasive state, though evidence remains primarily preclinical [43,44].

Collectively, these findings suggest that inflammatory signaling can converge on MITF suppression and promote melanoma dedifferentiation An overview of inflammatory pathways regulating MITF-dependent melanoma plasticity is provided in Table 1.

Table 1.

Inflammatory pathways regulating MITF-dependent melanoma plasticity.

Inflammatory factor	Source in TME	Key molecular mediators	Resulting melanoma phenotype	Therapeutic implications	Reference
TNF-α	TAMs, immune cells	c-Jun activation, NF-κB signaling	MITFLow/c-JunHigh; MITFHigh survival state	Resistance to MAPKi: macrophage-derived TNF-α may protect MITFHigh cells from BRAFi/MEKi-induced apoptosis	[31,33]
IFN-γ	TILs, myeloid cells	CREB inhibition, ATF4 induction, chromatin remodeling	ATF4High/MITFLow	PD-L1 and CD271 upregulation, modulation of CTL susceptibility	[35–38]
TGF-β	CAFs, immune cells, tumor cells	SMAD signaling, AXL upregulation	MITFLow/AXLHigh; MITFHigh/AXLHigh upon SMAD loss	Resistance to PD-1 blockade and targeted therapies	[40–42]
IL-1β	tumor cells (autocrine), immune cells	microRNA-155 induction	stable MITFLow	Impaired CTL recognition, sustained immune evasion	[43,44]

2.2. Hypoxia-induced phenotype switching through MITF regulation

Hypoxia is a central feature of the melanoma TME, driving increased invasiveness, angiogenesis and therapy resistance, largely through hypoxia-inducible factor 1 alpha (HIF-1α) activity that supports metabolic adaptation, survival and metastatic progression [45,46]. Elevated expression of hypoxia-related genes is linked to poorer prognosis in melanoma and to a more immunosuppressed tumor microenvironment [47]. Evidence indicates that hypoxia-elicited exosomal HIF-1α promotes drug resistance in melanoma by modulating solute carrier family 7 member 11 (SLC7A11) ubiquitination, as demonstrated in preclinical models [48]. Moreover, in vitro studies show that hypoxia enhances the immunosuppressive activity of CAFs by increasing their secretion of inhibitory factors and further inhibiting T cell cytotoxicity [49].

Hypoxia can also promote melanoma plasticity by suppressing MITF and facilitating transitions toward more dedifferentiated, invasive cell states. Hypoxia suppresses MITF through HIF-1α-dependent induction of transcriptional repressors, including basic helix-loop-helix family members e40 [50] and b2 [51] (BHLHE40 and BHLHB2), which bind and inhibit the MITF promoter, linking hypoxia-driven MITF suppression with increased phenotypic plasticity and metastatic progression [50]. This mechanism is consistent with immunohistochemical data demonstrating an inverse correlation between MITF and the hypoxia marker glucose transporter 1 (GLUT-1) in human tumors [52]. In line with these findings, hypoxia induces a switch from a receptor tyrosine kinase-like orphan receptor (ROR1)-positive to an ROR2-positive phenotype, accompanied by increased HIF-1α, reduced MITF and elevated wingless-type family member 5A (WNT5A) expression, changes associated with mesenchymal-like, invasive melanoma states [53]. Consistently, studies have identified fibronectin 1 (FN1)^High/MITF^Low melanoma cells predominantly within hypoxic, necrotic regions, exhibiting EMT-like and stem-associated features [54]. HIF-1α has also been reported to contribute to MITF repression under normoxic conditions [55], and pharmacologic inhibition of HIF-1α dimerization has been shown to lower MITF, disrupt metabolic pathways supporting melanoma survival and induce cell death, both demonstrated in vitro [56].

Furthermore, MITF directly regulates succinate dehydrogenase subunit B (SDHB) and thus influences the tricarboxylic acid (TCA) cycle. Under hypoxic conditions, reduced MITF levels decrease SDHB expression and elevate succinate, contributing to a sustained hypoxic response [57]. MITF also regulates HIF-1α expression and participates in the control of HIF-1α SUMOylation, a modification associated with increased proliferation, migration and invasion, indicating the reciprocal MITF-HIF axis [58]. These interactions are schematically summarized in Figure 2.

Figure 2.

Hypoxia-MITF feedback loops driving melanoma phenotypic plasticity.

Hypoxia within the TME enhances the immunosuppressive activity of CAFs, leading to reduced T-cell cytotoxicity and HIF-1α stabilization [49]. HIF-1α suppresses MITF via induction of the transcriptional repressors BHLHE40 and BHLHB2, promoting dedifferentiated melanoma states [50,51]. Reduced MITF downregulates SDHB, resulting in succinate accumulation and altered TCA flux that sustain hypoxic state, while MITF-dependent modulation of HIF-1α SUMOlyation reinforces a positive feedback loop [57,58].

2.3. Cancer-associated fibroblasts as modulators of melanoma phenotype

Importantly, nonmalignant stromal cells also contribute to the inflammatory niche, with CAFs representing the predominant and highly heterogeneous stromal population. They shape immune interactions and support melanoma progression, metastasis and therapy resistance through the secretion of factors such as TGF-β, interleukin-6 (IL-6), fibroblast growth factor 2 (FGF2) and podoplanin (PDPN) [59,60]. Beyond paracrine signaling, they remodel the ECM by depositing components including collagen and fibronectin, enhancing melanoma cell adhesion, motility and cytoskeletal dynamics, changes that have been linked to invasive phenotype and emerging resistance to MAPK-directed therapies [61].

CAFs secrete high levels of IL-6 and interleukin-8 (IL-8), cytokines that enhance melanoma motility and contribute to resistance to immunotherapy [62]. Moreover, CAFs can suppress cytotoxic T-cell function through soluble mediators, including elevated arginase activity and C-X-C motif chemokine ligand 12 (CXCL12), which impair CTL signaling [63]. Importantly, it has been reported that melanoma regions with high CAF abundance show a negative correlation with the MITF^High program and a positive correlation with the AXL^High state, indicating a transcriptomic correlation between CAF-rich niches and reduced MITF expression, though a direct causal relationship remains to be established [64]. Additionally, melanoma phenotype switching has also been described as transdifferentiation toward CAF-like or endothelial-like states [18,65]. It has been proposed that MITF^Low/AXL^High melanoma cells can acquire CAF-like traits, including the ability to autonomously deposit and remodel a fibrillar ECM enriched in collagen and fibronectin, a process that reinforces resistance to BRAF and/or MEK inhibitors [66].

Moreover, in preclinical models, CAFs promote resistance to BRAF/MAPK inhibition through fibronectin-driven β1-integrin/focal adhesion kinase (FAK)/proto-oncogene tyrosine-protein kinase Src (SRC)-dependent extracellular signal-regulated kinase (ERK) reactivation [67] and, independently, through collagen-mediated discoidin domain-containing receptor 1 and 2 (DDR1/DDR2) clustering with activation of the pro-survival NF-κB-inducing kinase (NIK)/I-kappa-B kinase alpha (IKKα)/NF-κB2 pathway [68].

Although MITF^Low melanoma states exhibit CAF-like features, direct evidence that CAFs modulate MITF expression or control MITF-dependent phenotype switching is currently lacking. Determining whether CAF-derived signals influence MITF dynamics represents an important direction for future research, given their established roles in metastasis, immune evasion and drug resistance.

3. Mechanotransduction and ECM remodeling in MITF plasticity

ECM consists of networks of extracellular proteins, proteoglycans and glycoproteins. Beyond providing a structural framework, it delivers biochemical and biomechanical signals that regulate cell growth, survival, migration and differentiation, as well as vascular development and immune responses. Increasing evidence indicates that tumor-associated ECM signaling represents a critical determinant of malignancy [69,70]. In melanoma, progressive ECM remodeling is marked by loss of basement membrane components (collagen IV, laminin) and accumulation of interstitial collagens, tenascin and fibronectin, forming a dense matrix that promotes invasion and metastases [71]. Signals derived from the extracellular matrix, including stiffness, integrin involvement and yes-associated protein (YAP)/transcriptional coactivator with PDZ-binding motif (TAZ)-mediated mechanotransduction, have been implicated in the regulation of MITF expression and consequently influence melanoma malignancy [72–74].

3.1. ECM stiffness and integrin-mediated control of MITF expression

The composition and stiffness of the matrix are key factors determining melanoma cell behavior. Increased cross-linking and alignment of collagen fibers enhance integrin signaling, activating downstream mechanosensitive pathways that affect gene expression and cell fate. Tumor-associated matrix remodeling has been linked to tumor growth, survival and metastatic progression in a context-dependent manner, as distinct ECM compositions can promote different melanoma phenotypic states [73,75,76]. Notably, experimental modeling indicates that melanoma invasion displays a non-linear relationship with ECM stiffness: both very loose and highly rigid matrices restrict cell motility, whereas intermediate stiffness is most permissive [77]. In melanoma, ECM stiffness has been shown experimentally to affect MITF expression via the YAP/paired box gene 3 (PAX3)/MITF signaling axis. Increased collagen density and rigidity induce a differentiated state by promoting YAP translocation to the nucleus and changing MITF expression. Conversely, inhibition of YAP activity or disruption of integrin binding reduces MITF expression, indicating that ECM-derived mechanical stimuli can be translated into transcriptional control of melanocytic differentiation. These findings highlight that integrin signaling is not only structural but also an important determinant of melanoma phenotype [73]. Integrin expression also correlates with metastatic risk and has been evaluated as a prognostic marker in primary melanoma [78]. Moreover, integrins have been extensively explored as therapeutic targets in melanoma [79], although clinical translation has proven challenging, with trials failing to demonstrate significant efficacy [80].

3.2. YAP/TAZ signaling as a mechanosensitive regulator of MITF plasticity

Cells are constantly exposed to various mechanical factors such as stress, changes in stiffness, pressure and tissue deformation, which they sense and integrate through mechanotransduction mechanisms. The transcription factors YAP and TAZ interpret a wide range of mechanical signals, from shear stresses to cell shape and ECM stiffness, and translate them into cell-specific transcriptional programs. Their nuclear localization is promoted by cytoskeletal tension, increased matrix stiffness and actin reorganization [81,82]. In melanoma, YAP/TAZ activation correlates with reduced MITF expression and acquisition of an invasive state [74]. In preclinical models, YAP/TAZ signaling has also been linked to resistance to BRAF inhibitors, while their partner transcription factors of the TEAD (transcriptional enhanced associate domain) family are key regulators of the MITF^Low phenotype. Mechanical stimuli integrated through YAP/TAZ may facilitate reversible transitions between MITF^High and MITF^Low cell states [83]. It is worth noting that, in uveal melanoma, aberrant activation of YAP/TAZ signaling represents a particularly prominent oncogenic driver, fueled by activating mutations in GNAQ and GNA11 (G protein subunit alpha q and alpha 11) that promote YAP nuclear accumulation independently of canonical Hippo pathway inhibition [84,85].

Different ECM components can shape MITF activity by altering MAPK signaling. In vitro studies indicate that collagen I weakens ERK signaling and increases nuclear MITF, supporting a more differentiated state, while fibronectin strengthens FAK-ERK signaling, lowers nuclear MITF and favors dedifferentiation. Blocking MEK/ERK restores nuclear MITF on fibronectin, suggesting that ECM cues regulate MITF largely through ERK-dependent control of its localization. ECM cues may additionally shape MITF programs through WNT/β-catenin pathway, as fibronectin exposure upregulates mediators such as FRAT regulator of WNT signaling pathway 2 (FRAT2), stabilizing β-catenin and providing an auxiliary route through which ECM composition influences MITF-dependent transcription [86].

3.3. ECM remodeling and positive feedback loops in melanoma plasticity

Melanoma cells not only respond to ECM-derived signals but also actively remodel their microenvironment to maintain a specific phenotypic state. MITF has been reported to act as a transcriptional repressor of ECM, focal adhesion and EMT-related genes, indicating that matrix remodeling may represent both a driver and a consequence of low-MITF states [87]. In experimental systems, MITF^Low invasive cells secrete ECM components such as type I collagen, fibronectin and lysyl oxidase (LOX), increasing matrix stiffness and enhancing mechanotransduction. This forms a positive feedback loop, in which ECM remodeling promotes YAP/TAZ activation and further suppresses MITF [88]. Furthermore, exposure to BRAF inhibitors induces biomechanical remodeling of the ECM, including increased collagen I and fibronectin deposition by melanoma cells and CAFs. The resulting ECM stiffening has been proposed to promote melanoma survival via FAK/SRC signaling leading to resistance to MAPK-targeted treatment [61]. Consistent with this model, integrin-SRC signaling has been identified as a key mediator of BRAFi resistance in vivo, with SRC activation linking ECM remodeling to YAP/TAZ-mediated MITF suppression [67]. These reciprocal interactions between MITF, ECM remodeling and mechanotransduction are schematically summarized in Figure 3

Figure 3.

ECM remodeling-driven feedback mechanisms stabilizing MITF plasticity.

Stiff and fibronectin-rich extracellular matrix activates integrin-FAK/SRC signaling, promoting nuclear YAP/TAZ-TEAD activity and parallel MEK/ERK signaling. These pathways converge on suppression of nuclear MITF, while MITF^Low melanoma cells further remodel the ECM through secretion of type I collagen, fibronectin and lysyl oxidase, establishing a positive feedback loop that stabilizes the invasive phenotype and contributes to therapy resistance [72,83,86,88].

4. Metabolic reprogramming as a driver and consequence of MITF dynamics

Tumor cells remodel their metabolism to sustain proliferation, survival and metastatic progression, and this metabolic reprogramming is recognized as a fundamental hallmark of cancer driven by oncogenic signaling [89]. The metabolic state of melanoma is further shaped by systemic factors. Specific dietary patterns and nutrient-derived metabolites can modulate melanoma metabolism and alter antitumor immunity, influencing the efficacy of immune checkpoint inhibitors through effects on CD8⁺ T-cell activity, nutrient-stress pathways and the gut microbiome [90]. Microbial determinants such as reduced Faecalibacterium prausnitzii correlate with adverse lipid and bile acid metabolism and poorer recurrence-free survival [91]. MITF functions as a central metabolic sensor in melanoma. High MITF supports proliferation by regulating the TCA cycle, fatty-acid desaturation and mitochondrial metabolism, whereas low MITF activates starvation-like programs through succinate-HIF signaling that promote dedifferentiation, invasion and therapy resistance [92,93]

4.1. Metabolic switching between glycolysis and OXPHOS

Under normoxic conditions, oxidative phosphorylation (OXPHOS) is the most efficient source of adenosine triphosphate (ATP), yet many cancers, including melanoma, preferentially use aerobic glycolysis. This metabolic pattern, known as the Warburg effect, supports rapid growth by supplying intermediates required for nucleotide, amino-acid and lipid synthesis. In metastatic and therapy-resistant melanoma, however, OXPHOS is upregulated. MITF contributes to this metabolic switch through transcriptional control of peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC1α) [94]. MITF directly induces PGC1α expression, thereby increasing OXPHOS, and PGC1α can in turn promote MITF transcription, suggesting a positive feedback loop [95,96]. The MITF^High phenotype often correlates with high PGC1α expression and an OXPHOS dominance, whereas in MITF^Low state are frequently associated with HIF1α-mediated glycolysis and enhanced glutamine usage [97]. Although PGC1α supports progression of malignant cells and their survival under oxidative stress conditions, its low expression has been associated with increased invasiveness and metastasis [98]. Evidence suggests that dedifferentiation and loss of MITF may render melanoma cells PGC1α-independent [96]. Resistant BRAF-mutant melanoma cells shift toward elevated OXPHOS and glutamine dependence, creating a mitochondrial vulnerability [99]. Studies further indicate that oncogenic BRAF suppresses MITF-PGC1α axis, driving melanoma cells toward glycolysis, while BRAF inhibition reverses this shift and leads to response to oxidative stress, as demonstrated in vivo [100]. BRAF inhibition also induces nicotinamide phosphoribosyltransferase (NAMPT) upregulation, lowering nicotinamide adenine dinucleotide (NAD+) levels, promoting invasive and drug-resistant states, as shown in preclinical models [101]. These findings show that melanoma relies on a highly adaptable metabolic landscape, where transitions between glycolytic, oxidative and hybrid states, enable cells to cope with environmental stress and evade therapy [102].

Several factors influence MITF-PGC1α axis. ATP-citrate lyase (ACLY), enhances acetyltransferase p300 activity and promotes histone acetylation at the MITF locus, increasing its expression, leading to increased OXPHOS and resistance to MAPK inhibitors. Accordingly, ACLY inhibition restores sensitivity to MAPK pathway inhibitors (MAPKi) in preclinical models [103]. Oxygen availability further modulates this axis. Hypoxia suppresses MITF and promotes HIF-1α-driven glycolysis via activation of pyruvate dehydrogenase kinase 1 (PDK1), which inhibits pyruvate dehydrogenase (PDH), whereas normoxia can stabilize PGC1α protein even in MITF^Low cells, supporting a shift toward OXPHOS [104]. HIF-1α increases GLUT-1 expression, lactate dehydrogenase A (LDHA)-mediated lactate production and reduces mitochondrial TCA flux, promoting glutamine as the primary carbon source [105,106].

In vitro studies indicate that S100A4 protein, also known as metastasin (Mts1) or fibroblast-specific protein 1 (FSP1), suppresses this axis by downregulating MITF, which in turn shifts melanoma toward a more glycolytic, invasive state [107].

As a consequence of glycolysis, the increased lactate production results in more acidic TME, and low extracellular pH drives phenotypic switching. Sustained acidosis represses MITF, promoting a MITF^Low/AXL^High phenotype, associated with therapy resistance in vitro [108]. The dynamic metabolic plasticity associated with MITF state transitions is schematically summarized in Figure 4.

Figure 4.

Metabolic feedback loops linking MITF dynamics with melanoma plasticity.

Melanoma cells dynamically transition between MITF^High and MITF^Low states in response to microenvironmental stress and therapy pressure. The dashed arrow indicates the reversibility of this transition. Arrows (↑/↓) indicate increased or decreased activity of metabolic processes and metabolites, respectively. Protein nodes are shown without directional labels, as their regulation is conveyed by the connecting arrows in the figure. Red blunt-ended lines indicate inhibition. In MITF^High state, MITF cooperates with PGC1α (MITF-PGC1α axis) to promote OXPHOS, leading to oxidative stress tolerance. ACLY-derived acetyl-CoA epigenetically reinforces this axis [96,103]. In contrast, MITF^Low cells adopt a glycolytic stress-adapted program under hypoxic conditions (↓O2), characterized by HIF-1α activation, increased lactate production and acidification (↓pH). HIF-1α promotes glycolysis through LDHA induction, while inhibiting mitochondrial metabolism via PDK1-mediated PDH repression. MITF suppression leads to succinate accumulation and altered TCA cycle flux, stabilizing HIF-1α. TAMs-derived Mts1 additionally suppresses MITF [57,97,104,106].

4.2. Lipid metabolism as a regulator of MITF-dependent plasticity

Lipid metabolism is profoundly rewired in melanoma to sustain tumor growth and progression. Melanoma cells display increased de novo lipogenesis driven by enhanced acetyl coenzyme A (acetyl-CoA) production through ACLY and upregulation of key enzymes such as acetyl-CoA carboxylase (ACC1) and fatty acids synthase (FASN), associated with poor prognosis. Under metabolic stress, BRAF-mutant melanomas additionally acquire acetate dependence, while altered fatty-acid utilization, cholesterol biosynthesis and lipid signaling further promote metastasis and therapy resistance [109,110]. Lipid metabolism also appears to influence melanoma phenotype switching through its impact on MITF expression. Key lipid metabolic pathways regulating MITF expression are summarized in Table 2.

Table 2.

Lipid metabolic pathways regulating MITF-dependent melanoma plasticity.

Lipid pathway	Key enzymes or receptors	MITF-centered regulatory relationship	References
Fatty acid desaturation	SCD	MITF induces SCD expression; SCD prevents SFA accumulation and ER stress, sustaining MITF activity	[111,112]
SFA accumulation/ ER stress	SCD	ER stress activates eIF2α-ATF4/NF-κB signaling, leading to MITF suppression	[111,112]
Bioactive phospholipids	LPA, S1P	ERK-dependent MITF phosphorylation and degradation	[113,114]
Sphingolipid metabolism	A-SMase,	Sustained ERK/AKT signaling under A-SMase inhibition	[115]
Ceramide signaling	C2-ceramide ASAH1	Sustained ERK/AKT activation; αvβ5/FAK activation under ASAH1 inhibition; reciprocal MITF-ASAH1 regulation	[116–118]
Cholesterol uptake	SR-BI	Reciprocal MITF-SR-BI regulation likely through AKT signaling;	[119]
Environmental lipid uptake	FATP1 WNT5A AXL SRC	MITFHigh drives FATP1-mediated lipid uptake; MITFLow/AXLHigh induces adipocyte lipolysis via WNT5A and AXL-SRC-driven invasion	[120,121]
Eicosanoid metabolism	COX-2	COX-2 inhibition suppresses MITF and enhances BRAFi efficacy through ERK and PI3K/AKT modulation	[123,124]

MITF regulates fatty acid desaturation in melanoma through direct control of stearoyl-CoA desaturase (SCD), which catalyzes the conversion of long-chain saturated fatty acids (SFAs) into monounsaturated fatty acids (MUFAs). High MITF drives SCD expression and prevents SFA accumulation, thereby preventing endoplasmic reticulum (ER) stress and apoptosis. In contrast, reduced SCD activity induces ER stress, eIF2α phosphorylation and ATF4/NF-κB signaling, leading to MITF suppression and stabilization of the MITF^Low phenotype. This establishes a positive feedback loop in which altered fatty acid composition actively reinforces the dedifferentiated state [111,112].

Bioactive lipids such as lysophosphatidic acid (LPA) and sphingosine-1-phosphate (S1P) similarly repress MITF activity by activating ERK-dependent MITF phosphorylation and degradation [113,114], while reduced acid sphingomyelinase (A-SMase) activity amplifies this axis by sustaining ERK activation. Although A-SMase levels inversely correlate with metastasis, this likely reflects the requirement for transient MITF suppression during invasion followed by MITF re-expression during metastatic outgrowth [115]. Moreover, C2-ceramide also suppresses MITF by inducing sustained ERK and AKT activation [116]. Low acid ceramidase (ASAH1) downregulates MITF via integrin alpha-v beta-5 (αvβ5)-FAK signaling, while MITF itself activates ASAH1, forming a feedback loop that links sphingolipid metabolism to melanoma differentiation state and plasticity [117,118].

A similar regulatory circuit has been suggested for scavenger receptor class B type 1 (SR-BI), a high-density lipoprotein (HDL)-cholesterol uptake receptor, which may activate MITF via AKT signaling, while MITF in turn induces SR-BI expression, thereby linking cholesterol availability to MITF activity [119].

Beyond cell-intrinsic pathways, environmental lipids supplied by stromal cells regulate melanoma behavior in a phenotype-dependent manner. MITF-dependent states display distinct mechanisms of lipid acquisition, with MITF^High cells relying primarily on fatty acid uptake through transporters such as fatty acid transporter protein 1 (FATP1) [120], whereas invasive MITF^Low/AXL^High cells can induce adipocyte lipolysis via WNT5A and utilize released fatty acids to support invasion through an AXL-SRC signaling axis [121]. Stromal cells, including aged dermal fibroblasts, can further promote resistance to BRAFi/MEKi therapy through fatty acid transporter protein 2 (FATP2)-mediated lipid import, highlighting the role of microenvironment-derived lipids in melanoma metabolic plasticity [122].

Furthermore, eicosanoid pathways also contribute to metabolic control of cell phenotype through MITF regulation. Cyclooxygenase-2 (COX-2) inhibition by resveratrol suppresses MITF via ERK and phosphoinositide 3-kinase (PI3K)/AKT pathways [123], while non-steroidal anti-inflammatory drugs (NSAIDs) such as diclofenac and lumiracoxib enhance BRAF inhibitors (BRAFi) efficacy, block metabolic rewiring toward OXPHOS, and delay resistance by preventing MITF upregulation in vitro [124].

4.3. Nutrient starvation and integrated stress responses driving MITF plasticity

Nutrient limitation is a common stress in tumors and forces cancer cells to rebalance metabolic supply and biosynthetic demand. This adaptation has been proposed to rely on translational reprogramming driven by eIF2α phosphorylation, which reduces anabolic activity while promoting invasive behavior as a survival strategy [125]. Positron emission tomography (PET) imaging indicates that nutrient utilization within the tumor microenvironment is intrinsically programmed, with cancer cells preferentially relying on glutamine, while glutamine metabolism limits glucose uptake even when glucose is not scarce [126].

Microenvironmental stresses such as glutamine deprivation activate the integrated stress response (ISR) via eukaryotic initiation factor 2B (eIF2B) inhibition, leading to translational reprogramming, ATF4 induction, and suppression of MITF at both translational and transcriptional levels, promoting a MITF^Low phenotype. Importantly, mechanistic studies indicate that translational reprogramming rather than MITF loss alone drives invasion. Although ATF4 can suppress MITF even in nutrient-rich conditions, generating an ATF4^High/MITF^Low state, this alone is insufficient to induce invasiveness. However, under glutamine starvation translation reprogramming has been shown to promote oxidative-stress resistance, anoikis resistance, and immunotherapy-resistance, thereby enabling efficient metastatic survival [34].

Restriction of glucose also activates ATF4, leading to MITF suppression. As a consequence, loss of MITF impairs cell-cycle progression and shifts cells toward a slow-cycling, MITF^Low/AXL^High invasive phenotype. Glucose deprivation also promotes pro-angiogenic and pro-invasive programs, including vascular endothelial growth factor (VEGF) and IL-8 expression. Mechanistically, VEGF induction is mediated in part through direct transcriptional activation by ATF4, linking metabolic stress to metastatic potential [127,128].

Moreover, MITF directly regulates a melanoma-specific subset of lysosomal and autophagy genes by binding coordinated lysosomal expression and regulation (CLEAR) elements, thereby supporting starvation-induced autophagy and tumor cell survival independently of transcription factor EB and E3 (TFEB and TFE3), classical transcriptional regulators of lysosomal biogenesis and autophagy belonging to the microphthalmia/transcription factor E (MiT/TFE) family. Accordingly, MITF depletion attenuates the autophagic response to nutrient starvation, whereas its overexpression increases autophagosome formation, indicating that autophagy serves as an adaptive survival mechanism for melanoma cells under nutrient-limiting condition in cellular models [129]. Under MITF^Low conditions driven by metabolic stress, a hierarchical transition from MITF to TFEB and ultimately TFE3 is engaged. Nuclear translocation of TFE3 can drive a distinct transcriptional program associated with mesenchymal behavior, metastasis, and metabolic adaptation including OXPHOS upregulation, acting antagonistically to MITF [130,131]. Moreover, the three MiT/TFE family members differentially regulate immune cell infiltration [131].

4.4. Integrative signaling nodes controlling MITF-dependent plasticity

The signaling pathways described in preceding sections converge on various intracellular regulators that control MITF activity in a hierarchical and context-dependent manner. Among these, ERK acts as a major post-translational regulator of MITF and links diverse signals to rapid changes in MITF activity [132]. These signals include fibronectin-driven integrin-FAK signaling [86], bioactive lipids [113,114,116], and eicosanoid pathways [123]. Notably, ERK signaling is context-dependent: transient ERK activation can enhance MITF transcriptional activity, whereas sustained activation promotes MITF degradation [132]. AKT represents another regulatory node integrating metabolic and microenvironmental cues. Lipid-derived mediators activate PI3K/AKT signaling [116,119], while hypoxia promotes functional synergy between AKT and HIF-1α that supports melanoma progression [133]. Upon MAPK pathway suppression, PI3K/AKT can emerge as a compensatory survival pathway that can suppress MITF-dependent differentiation [134]. YAP/TAZ signaling intersects with these pathways through mechanosensory stimuli from the ECM. ECM components influence both ERK signaling and YAP/TAZ activity, linking mechanotransduction to regulation of MITF-dependent cell states [86]. The ISR provides an additional regulatory pathway centered on ATF4. Nutrient deprivation [127,128], inflammatory cytokine signaling [34,36], HIF-1α-dependent signaling [56], oncogenic BRAF activity [135], and ER stress induced by lipid imbalance [111] can each activate ATF4, making it a convergence point for diverse signals that sustain MITF suppression. Together, these pathways form an interconnected regulatory network controlling MITF dynamics.

5. Therapeutic strategies targeting melanoma plasticity

A major challenge in melanoma therapy is the emergence of treatment-resistant states that arise under MAPK inhibition and immunotherapy. Resistant phenotypes may involve either dedifferentiated MITF^Low populations or metabolically adapted MITF^High states, and are shaped by tumor metabolism and by microenvironmental stimuli such as inflammation, hypoxia, and ECM remodeling. Consequently, effective therapeutic strategies will likely need to target the pathways governing phenotype switching rather than focusing solely on oncogenic signaling. However, despite promising mechanistic rationale and encouraging preclinical results, most strategies targeting melanoma plasticity remain at early developmental stages Current therapeutic strategies targeting melanoma plasticity are summarized in Table 3.

Table 3.

Therapeutic strategies targeting MITF-dependent melanoma plasticity.

Targeted pathway	Therapeutic agents	Biological and therapeutic effects	Therapeutic context	Developmental stage	References
Directed phenotype redifferentiation	Methotrexate + TMECG	Induction of a MITFHigh state, apoptosis via thymidine, depletion and DNA damage.	Effect independent of BRAF, MEK and p53 status	Preclinical (in vivo)	[136]
YAP-TAZ-TEAD	Verteporfin + anti-PD-1	Inhibition of YAP1 activity, reduced tumor size, promotion of pro-inflammatory TME.	anti-PD-1- resistance	Preclinical (in vivo)	[138]
mTORC1/2 signaling/ OXPHOS	AZD8055, AZD2014 + selumetinib	Suppression of OXPHOS, restored sensitivity to MEKi.	MEKi-resistance	Preclinical (in vivo)	[141]
Mitochondrial complex I/ OXPHOS	Metformin + vemurafenib	Antiproliferative effects in BRAFV600E melanoma cell lines. No clinical efficacy data available from the phase I/II study.	BRAFV600E melanoma, MAPKi- resistance	Preclinical; Phase I/II (NCT01638676)	[143]
Mitochondrial complex I/ OXPHOS	Metformin + pembrolizumab	No significant impact of metformin on pembrolizumab efficacy in phase III study. No clinical efficacy data available from the phase I study	Advanced melanoma	Phase III (NCT02362594); Phase I (NCT03311308)	[145]
Mitochondrial complex I/ OXPHOS	Metformin + anti-PD-1 mAb	No clinical efficacy data available.	Advanced melanoma and other tumors	Phase II (NCT04114136)
Mitochondrial complex I/ OXPHOS	Phenformin + dabrafenib + trametinib	Synergistic antitumor activity, delayed resistance, tumor regression in preclinical study. No clinical efficacy data available from the phase Ib study.	MAPKi-resistance	Preclinical; Phase Ib (NCT03026517)	[146,147]
Mitochondrial complex I/ OXPHOS	IACS-010759	Marked tumor regression as monotherapy, lack of synergy with MAPKi.	MAPKi- resistance	Preclinical (in vivo)	[148]
Mitoribosome/ ISR	Tigecycline, doxycycline + MAPKi	Sensitization to MAPKi, suppression of drug-tolerant cell population, complete rensponse in one patient	MAPKi-resistance	Preclinical (in vivo); Single case report	[149]
Glutamine metabolism	Telaglenastat	Suppression of glutamine-driven OXPHOS, enhanced T-cell cytotoxicity; limited clinical efficacy.	BRAFi-resistance	Preclinical; Phase I/II (NCT02771626)	[151,152]
Fatty acid β-oxidation	Thioridazine	Inhibition of peroxisomal FAO, suppression of drug-tolerant persister cells	MAPKi-resistance	Preclinical (in vivo)	[154]
Fatty acid β-oxidation	Ranolazine	Delayed BRAFi resistance; reduction of NGFRHigh persisters; improved antitumor immunity.	BRAFi-resistance	Preclinical (in vivo)	[155]
Cholesterol metabolism	ACAT2 or SOAT inhibition + vemurafenib	Tumor growth suppression; induction of ferroptotic cell death.	BRAFi-resistance	Preclinical (in vitro)	[157]
Lipid remodeling	Coronatine	Increased pro-apoptotic ceramides, reduced pro-metastatic phosphatidylinositols, selective melanoma suppression.	Lipidomic profiling in melanoma cells	Preclinical (in vitro)	[158]
RXR signaling	RXR agonists (bexarotene), RXR antagonists (HX531)	Modulation of NCSC-like MRD states, unintended expansion of AXLHigh persisters	MAPKi-resistance	Preclinical (in vivo)	[15]
Ferroptosis	Erastin; GPX4 inhibitors (RSL3, ML162, ML210)	Selective killing of dedifferentiated MITFLow melanoma cells; reduced resistant outgrowth	BRAFi-resistance	Preclinical (in vitro)	[14]
Autophagy	PARP inhibitors + MAPKi	Induction of lysosomal autophagic cell death, restored MAPKi sensitivity.	MAPKi-resistance	Preclinical (in vivo)	[161]

Directed phenotype switching has been explored using methotrexate (MTX) to force melanoma cells into a MITF^High, differentiated, tyrosinase-positive state. This differentiation makes them vulnerable to the tyrosinase-activated antifolate prodrug TMECG, a potent dihydrofolate reductase (DHFR) inhibitor. In combination, MTX and TMECG deplete thymidine, induce DNA damage and apoptosis, achieving melanoma-selective effects in vitro and in mouse models, irrespective of BRAF, MEK or tumor protein p53 status [136]. Independently, low-dose MTX limits distant metastasis in vivo [137]. To date, however, neither of these preclinical strategies has been evaluated in clinical trials.

5.1. Targeting mechanotransduction pathways

Another potential target is the YAP/TAZ-TEAD axis, a central effector of mechanotransduction known to suppress MITF and promote invasive states in melanoma. YAP1 overexpression has been identified as a biomarker of anti-PD-1 resistance, and pharmacologic YAP1 inhibition with verteporfin enhanced anti-PD-1 efficacy in vivo by promoting a more inflamed, T-cell-rich TME [138]. The first-in-human TEAD inhibitor IK-930 was evaluated in a phase I trial (NCT05228015) in advanced solid tumors, based on strong preclinical synergy with MEK and epidermal growth factor receptor (EGFR) inhibition in xenograft models of colorectal cancer and non-small cell lung cancer; however, the trial was terminated prior to completion [139]. Clinical proof-of-concept for TEAD inhibition is further provided by the palmitoylation inhibitor VT3989 (NCT04665206), currently tested in a phase I/II trial in patients with refractory solid tumors, predominantly malignant pleural mesothelioma. VT3989 demonstrated a favorable safety profile and meaningful antitumor activity, including objective responses and prolonged progression-free survival in a subset of patients [140]. Together, these data indicate that pharmacologic targeting of YAP/TAZ-TEAD signaling is feasible, though clinical evidence in melanoma is currently lacking and extrapolation from other tumor types should be interpreted with caution.

5.2. Targeting metabolic reprogramming

Metabolic reprogramming constitutes an additional component of melanoma plasticity, enabling dynamic adaptation to environmental and therapeutic stress, and therefore represents another target for future treatment strategies. Importantly, therapeutic targeting of melanoma metabolism must be interpreted in the context of the pronounced metabolic heterogeneity described above. Melanoma cells dynamically switch between glycolytic and oxidative states depending on genetic background, microenvironmental conditions and therapeutic pressure, meaning that interventions targeting OXPHOS, FAO or glutamine metabolism are unlikely to be universally effective and may require biomarker-guided patient selection. Resistance to MEK inhibitors can shift melanoma toward a high OXPHOS state, and inhibiting the mammalian target of rapamycin complex 1/2 (mTORC1/2) signaling may resensitize such cells. Dual mTORC1/2 inhibitors (AZD8055 or AZD2014) combined with selumetinib resulted in markedly superior tumor suppression in mice than either drug alone. Mechanistically, mTORC1/2 blockade reduces PGC1α expression, thereby suppressing OXPHOS and restoring sensitivity [141]. Consistently, PGC1α inhibition reduces melanoma tumorigenicity and suppresses stem-like populations in vitro [142].

Mitochondrial complex I inhibitors have also been explored as a metabolic co-treatment to MAPK inhibition. Metformin combined with vemurafenib demonstrated synergistic antiproliferative effects in a subset of BRAF^V600E melanomas, though responses were heterogenous [143], and synergy with binimetinib has been confirmed in 2D and 3D models, including vemurafenib-resistant cells [144]. A phase I/II clinical trial (NCT01638676) is currently evaluating metformin in combination with vemurafenib, and no efficacy data are yet available. Additionally, clinical evaluation of metformin in combination with immune checkpoint inhibitors is ongoing (NCT03311308; NCT04114136); however, available data from completed phase III study have not yet demonstrated a clear improvement in anti-PD-1 efficacy [145]. Phenformin, another mitochondrial complex I inhibitor, has shown strong preclinical activity in melanoma, where its combination with BRAF inhibition produced synergistic antitumor effects, delayed resistance, and induced tumor regression in mouse models [146]. A phase Ib study combining phenformin with dabrafenib and trametinib reported manageable toxicity and objective responses in over half of treated patients, though as a dose-finding study it was not powered to assess efficacy [147]. Consistent with these findings, the OXPHOS complex I inhibitor IACS-010759 (OPi) produced marked tumor regression in vivo, however this effect has not been observed in combination with MAPKi [148]. Furthermore, ISR activation upregulates mitochondrial protein synthesis, conferring vulnerability to mitoribosome-targeting tetracycline antibiotics, including tigecycline and doxycycline, that can sensitize BRAF-mutant melanoma to MAPK inhibition and overcome both targeted therapy and immunotherapy resistance in preclinical models, though clinical validation remains limited to a single case report [149].

Importantly, the rise in OXPHOS activity in BRAFi-resistant melanoma cells relies predominantly on glutamine [99,150], suggesting that glutaminase inhibition could counteract this metabolic shift. The glutaminase inhibitor telaglenastat exhibited immunomodulatory and antitumor effects in preclinical melanoma models, enhancing T-cell cytotoxicity and improving responses to adoptive T-cell therapy and immune checkpoint blockade [151]. However, in a phase I/II clinical trial with nivolumab (NCT02771626) the combination showed acceptable safety but only modest clinical efficacy in melanoma [152].

Lipid metabolism also shapes melanoma immunogenicity and therapeutic response. MAPKi-treated melanoma cells increase their reliance on fatty acid β-oxidation (FAO) [153], and inhibiting this pathway, including blockade of peroxisomal FAO with thioridazine, suppresses the emergence of drug-tolerant persister cells and enhances MAPKi efficacy in vivo [154]. Similarly, ranolazine delays BRAFi resistance, reduces nerve growth factor receptor (NGFR)^High persisters and improves antitumor immunity [155]. FAO additionally contributes to melanoma progression, as its inhibition increases autophagosome formation and suppresses migration and invasion [156]. Acquired resistance to BRAF inhibition has also been linked to altered lipid metabolism, with resistant cells displaying a remodeled lipid profile and increased dependence on extracellular lipids. Targeting cholesterol-processing enzymes such as acetyl-CoA acetyltransferase 2 (ACAT2) or sterol O-acyltransferase (SOAT) synergised with vemurafenib, reduced proliferation and promoted ferroptotic cell death in preclinical models [157]. In parallel, lipidomic profiling demonstrated that coronatine increases pro-apoptotic ceramides and reduce phosphatidylinositol species linked to metastasis, selectively suppressing melanoma proliferation while sparing normal melanocytes in vitro [158].

5.3. Other potential targets

Beyond the major phenotypic and metabolic axes, several additional vulnerabilities have been identified within drug-tolerant melanoma populations. Retinoid X receptor gamma (RXRG) regulates the NCSC-like drug-tolerant state within minimal residual disease (MRD). RXR activation with bexarotene promotes MAPKi-induced entry into a dormant NCSC state, whereas RXR antagonism (HX531) suppresses NCSC emergence but enriches AXL^High invasive persisters [15]. This suggests that RXR modulation may need to be combined with AXL-directed therapies, such as AXL-107-MMAE [159] to effectively target heterogeneous perister populations [18].

Dedifferentiated MITF^Low melanoma cells have been experimentally shown to be highly susceptible to ferroptosis. They showed marked sensitivity to erastin and glutathione peroxidase 4 (GPX4) inhibitors (RSL3, ML162, ML210), and BRAFi-induced dedifferentiation further enhances this vulnerability. Combining vemurafenib and ferroptosis inducers reduces the outgrowth of resistant populations in vitro [14]. This ferroptosis vulnerability may be contextually modulated by oleic acid, which is abundant in lymph [160], suppresses ferroptosis, and is preferentially acquired by MITF^Low/AXL^High cells to drive AXL-dependent invasion [121].

Additionally, poly(ADP-ribose) polymerase (PARP) inhibitors have been shown to reverse MAPKi-induced phenotype switching. They triggered lysosomal autophagic cell death, increased mitochondrial lipid metabolism and antigen presentation, and shifted melanoma cells from an invasive to a proliferative, MAPKi-sensitive phenotype. Combined PARP and MAPK inhibition resulted in synergistic tumor regression in patient-derived models [161].

Although multiple therapeutic strategies targeting melanoma plasticity have shown encouraging activity in experimental models, translation into durable clinical benefit remains limited, highlighting the need for more rigorous clinical validation and better identification of responsive patient subgroups.

6. Future perspectives and conclusions

Melanoma adapts to therapeutic and environmental pressures through a complex interplay of metabolic shifts, inflammatory signaling and mechanotransduction, which collectively reshape MITF activity and drive reversible transitions between differentiated and invasive states. These regulatory layers are not independent but converge on shared signaling nodes such as ERK, AKT, ATF4, and YAP/TAZ, which function as central integrators of MITF-dependent phenotype switching.

Importantly, the relative contribution of individual pathways to phenotype switching is likely context-dependent and varies across tumor genotypes, microenvironmental conditions, and treatment settings. Current evidence suggests that multiple regulatory inputs converge on MITF activity, but a universal hierarchy of signaling pathways governing melanoma plasticity has not yet been established. A deeper understanding of how these regulatory layers integrate to stabilize resistant phenotypes will be essential for developing combination strategies that limit cellular plasticity and prevent the persistence of therapy-tolerant melanoma cells. Durable disease control will likely require approaches that simultaneously reshape metabolic, epigenetic and microenvironmental programs governing phenotype switching and drug tolerance. Integrated metabolic-epigenetic strategies, including interventions that modulate lipid-driven epigenetic dysregulation [112], together with therapies targeting hypoxia [162], acidosis [163], immune-mediated inflammatory factors or mechanotransduction may help prevent the emergence of therapy-resistant states.

Stromal-directed approaches are also under preclinical investigation., Selective tubulin-binding agents were shown to normalize tumor vasculature, reduce hypoxia and improve responses to adoptive T-cell therapy and checkpoint blockade [164]. Complementary experimental work has demonstrated that engineered nanoclusters may selectively reduce viability of both epithelial- and mesenchymal-like melanoma cells through intrinsic apoptosis [165]. Despite promising preclinical activity across multiple plasticity-targeting strategies, whether any of these approaches will translate into durable clinical benefit remains to be established

Funding Statement

This research was supported by Statutory Subsidy Funds of the Department of Molecular and Cellular Biology, Wroclaw Medical University, grant no. SUBZ.D260.26.013.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

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