Epigenetic collapse establishes transcriptional bottlenecks in malignant rhabdoid tumor Authors: Tatsuya Masuda, Takayoshi Watanabe, Toshinori Ozaki, Yasuhiko Kamikubo Published: 2026 ABSTRACT AbstractMalignant rhabdoid tumor (MRT) is an aggressive pediatric cancer defined by biallelic loss of SMARCB1 and characterized by an unusually simple mutational landscape. Despite the absence of recurrent activating oncogenic mutations, MRT exhibits highly malignant behavior, creating a clinical paradox. Loss of SMARCB1 disrupts SWI/SNF-mediated chromatin organization, leading to enhancer redistribution and reorganization of transcriptional control. MRT therefore provides a defined model to examine how structural disruption of chromatin regulation sustains tumor maintenance in the absence of mutation-driven signaling activation. We propose that this epigenetic collapse reduces regulatory diversity and concentrates control of essential survival and cell-cycle programs within limited transcriptional nodes. We define this structural state as transcriptional bottleneck formation, in which tumor viability depends on concentrated transcriptional control rather than recurrent genetic alteration. Bottleneck nodes are characterized by enriched chromatin occupancy at survival and cell-cycle genes, limited compensatory regulation, and functional sensitivity to perturbation. In MRT, members of the RUNX family illustrate a representative concentrated transcriptional node linking chromatin reorganization to regulation of BIRC5 and related survival genes. This disease-focused perspective supports an architecture-centered therapeutic strategy aimed at disrupting concentrated transcriptional control. Identification of biomarkers reflecting this architectural state may facilitate selection of bottleneck-directed interventions in MRT.Graphical abstractKeywords: Malignant rhabdoid tumor, SMARCB1, Epigenetic collapse, Transcriptional bottleneck formation, Chromatin remodeling, Enhancer architecture Introduction Malignant rhabdoid tumor (MRT) is an aggressive pediatric cancer that primarily affects infants and young children. Despite intensive multimodal therapy, outcomes remain poor. MRT is defined by biallelic loss of SMARCB1, a core subunit of the SWI/SNF chromatin-remodeling complex [1–3]. Beyond SMARCB1 inactivation, recurrent driver mutations are rare [4, 5]. This creates a clinical and biological paradox: MRT behaves as a highly malignant tumor despite having one of the simplest mutational landscapes among human cancers [4, 5]. The limited impact of mutation-directed therapeutic strategies indicates that tumor maintenance in MRT is not primarily driven by mutation-driven signaling activation [4, 5]. Loss of SMARCB1 disrupts the structural integrity of the SWI/SNF complex and alters chromatin accessibility across promoters and distal regulatory elements [2–4, 6]. Experimental models show that SMARCB1 restoration can re-establish enhancer activity and lineage-associated transcription, indicating that its absence reshapes the regulatory landscape rather than activating a single oncogenic pathway [6, 7]. In MRT, this remodeling includes redistribution of enhancer activity and changes in super-enhancer organization [6–9]. These alterations involve widespread redistribution of thousands of enhancers and substantial changes in chromatin accessibility, reshaping gene expression at a genome-wide scale. Recent studies further highlight that alterations in three-dimensional genome organization and enhancer–promoter interactions are central features of transcriptional dysregulation in cancer [10]. Importantly, SMARCB1 functions as a canonical tumor suppressor gene, and its biallelic inactivation represents the defining genetic event in MRT. However, the biological consequences of SMARCB1 loss are conceptually distinct from those of classical tumor suppressors such as TP53, RB1, or BRCA1/2. Whereas loss of these tumor suppressors primarily disrupts signaling pathways, genomic integrity, or cell-cycle checkpoints, SMARCB1 loss directly impairs chromatin remodeling and globally alters enhancer architecture [2–4, 6, 7]. This leads to a reduction in regulatory diversity and redistribution of transcriptional control across the genome. We therefore propose that the “biological paradox” of MRT does not arise from tumor suppressor loss per se, but from the unique consequence of chromatin architectural disruption, which concentrates transcriptional control into limited regulatory nodes. This structural reorganization provides the basis for transcriptional bottleneck formation. In this Review, we use the term epigenetic collapse to describe the SWI/SNF-deficient chromatin state in which enhancer and promoter coordination is broadly disrupted. We propose that epigenetic collapse reduces regulatory diversity, leading to concentration of control over essential survival and cell-cycle genes within limited transcriptional nodes. We define this structural state as transcriptional bottleneck formation, in which tumor viability depends on concentrated transcriptional control rather than signaling activation. This concept aligns with broader descriptions of transcription-centered dependencies in cancer [11, 12]. This framework is conceptually related to the notion of “transcriptional addiction” in cancer [11], but differs in that it emphasizes the structural origin of such dependencies arising from chromatin architectural disruption rather than primarily describing their functional manifestation. Importantly, transcriptional bottleneck formation is not intended to redefine transcriptional dependency itself, but rather to describe the structural basis from which such dependencies emerge. In this framework, dependency represents a functional consequence, whereas bottleneck formation reflects the underlying architectural constraint imposed by epigenetic collapse. Transcriptional bottleneck formation does not imply activation of a classical mutation-driven oncogene. Rather, it reflects architectural reorganization of transcriptional control following chromatin remodeling dysfunction [2–4]. Within this reorganized landscape, certain transcription factors can assume a disproportionately central role in maintaining essential programs. In MRT, members of the RUNX family have been implicated in sustaining proliferative and survival-associated transcriptional outputs [13–16]. Functional perturbation of RUNX activity disrupts expression of genes linked to cell-cycle progression and resistance to apoptosis in MRT models, illustrating how a transcriptional regulator can operate as a bottleneck node in the absence of oncogenic mutation [14–16]. MRT also reflects a broader feature of pediatric cancers, which often exhibit low mutational burdens yet marked epigenetic dysregulation [4, 5, 17]. Unlike many adult malignancies that accumulate multiple signaling mutations over time, pediatric tumors frequently arise within developmental transcriptional programs that remain highly sensitive to chromatin remodeling [5, 17]. MRT exhibits biological heterogeneity. Molecular profiling studies have identified subgroups with distinct transcriptional features and developmental signatures, including ATRT-MYC, ATRT-SHH, and ATRT-TYR, which differ in developmental origin, transcriptional programs, and epigenetic features [17–19]. In this context, disruption of a single chromatin regulator such as SMARCB1 can have disproportionate effects on lineage specification and transcriptional coordination [2–4, 6]. Together, these observations support a model in which epigenetic collapse, rather than mutation-driven signaling activation, establishes transcriptional bottlenecks in MRT (Fig. 1). Epigenetic collapse establishes transcriptional bottlenecks in malignant rhabdoid tumor. A Genetic simplicity. Malignant rhabdoid tumor (MRT) is defined by biallelic loss of SMARCB1 and contains few additional recurrent driver mutations. Despite this simple mutational landscape, MRT exhibits highly malignant clinical behavior, illustrating a central biological paradox. B Epigenetic collapse Loss of SMARCB1 disrupts the SWI/SNF chromatin remodeling complex and destabilizes enhancer organization across the genome. Regulatory architecture becomes disordered, resulting in enhancer redistribution and reduced regulatory diversity. C Transcriptional bottleneck formation. As regulatory diversity declines, control of essential survival and cell-cycle genes becomes concentrated within limited transcriptional nodes. We define this structural state as transcriptional bottleneck formation, whereby tumor viability depends on concentrated transcriptional control rather than recurrent genetic alteration Current clinical and molecular landscape of MRT Clinical presentation and disease behavior MRT is a highly aggressive pediatric malignancy that occurs predominantly in infants and very young children. It most commonly arises in the kidney and central nervous system. Central nervous system MRT is referred to as atypical teratoid/rhabdoid tumor (ATRT) [17–20]. Regardless of anatomical site, these tumors share biallelic loss of SMARCB1 and exhibit similar biological behavior [17, 18]. Clinical outcomes remain poor despite intensive therapy, and long-term survival is limited, particularly in patients diagnosed before one year of age [18, 19]. Relapse is common, and metastatic disease may be present at diagnosis [18, 19]. In infants, radiation therapy is often delayed or avoided because of the risk of neurodevelopmental toxicity, which further restricts treatment options [18, 19]. Large cooperative studies have improved risk stratification and supportive care, yet improvements in survival have been modest [18, 19]. The continued gap between aggressive treatment and durable control highlights the need for strategies that address the biological basis of MRT [18, 19]. Standard multimodal therapy Current treatment combines surgery, intensive chemotherapy, and radiation therapy when feasible [18, 19]. Gross total resection improves local control but is not always achievable, particularly in central nervous system disease [18, 19]. Chemotherapy regimens include combinations of alkylating agents, platinum compounds, anthracyclines, and high-dose methotrexate [18, 19]. High-dose chemotherapy followed by autologous stem cell rescue has been explored in selected patients [18, 19]. Although some patients achieve remission, relapse remains frequent and treatment-related toxicity is substantial [18, 19]. Radiation therapy improves local control but presents major challenges in infants [18, 19]. Early cranial irradiation is associated with long-term neurocognitive and endocrine effects [18, 19]. Delaying or modifying radiation may reduce toxicity but can compromise disease control [18, 19]. These limitations illustrate the limits of intensifying conventional cytotoxic therapy in very young patients [18, 19]. Molecularly informed therapeutic approaches Because MRT is defined by loss of SMARCB1, therapeutic efforts have focused on chromatin biology [2–4]. One approach targets enhancer of zeste homolog 2 (EZH2), the catalytic subunit of Polycomb repressive complex 2 [21]. Preclinical studies show that EZH2 inhibition suppresses growth of SMARCB1-deficient tumor cells [21]. Early-phase trials have reported activity in some patients, but responses are variable and often incomplete [22]. Other epigenetic regulators have also been evaluated, including alternative SWI/SNF-related regulatory dependencies and synthetic lethal interactions [23–25]. Histone deacetylase inhibitors, bromodomain and extraterminal domain (BET) inhibitors, and cyclin-dependent kinase (CDK) inhibitors demonstrate preclinical activity in MRT models [25]. However, durable clinical benefit has not yet been consistently achieved [18, 19]. These findings suggest that targeting a single chromatin-modifying enzyme may not fully reverse the transcriptional state established by SMARCB1 loss [2–4, 6, 7]. Immunotherapy has also been explored. MRT typically has a low mutational burden and few neoantigens, which may limit responsiveness to immune checkpoint blockade [18, 19]. The tumor microenvironment remains under investigation, but immunotherapy has not yet changed overall clinical outcomes [18, 19]. Representative clinical and molecularly informed therapeutic strategies currently explored in MRT are summarized in Table 1. Table 1Current clinical and molecularly informed therapeutic strategies in MRTTherapeutic categoryRepresentative agents / approachesMechanistic rationale in MRTSupporting evidence (references)Clinical statusConventional multimodal therapySurgery, intensive chemotherapy, radiation therapyCytotoxic control of rapidly proliferating tumor cellsClinical cooperative studies and treatment protocols [18, 19]Standard of care; limited long-term survivalEZH2 inhibition (PRC2 targeting)Tazemetostat; EZH1/2 dual inhibitorsCounteracts altered chromatin repression following SMARCB1 lossPreclinical and early-phase clinical studies in SMARCB1-deficient tumors [21, 22, 26]Early clinical evaluation; variable responsesSWI/SNF-related synthetic lethal targetingBRD9 inhibition; ncBAF targetingExploits altered SWI/SNF complex composition in SMARCB1-deficient chromatin statesFunctional studies in SMARCB1-deficient cancers [23, 24, 27]Preclinical developmentEpigenetic regulator targetingHistone deacetylase inhibitors; BET inhibitors; CDK inhibitorsModulates transcriptional programs altered by chromatin remodeling dysfunctionPreclinical MRT studies and related chromatin-driven tumor models [25]InvestigationalTranscription factor interference (RUNX1 as representative example)RUNX1-targeted DNA-binding inhibitors (e.g., PI-polyamides)Disrupts concentrated transcriptional control sustaining survival and cell-cycle programsGenetic suppression and pharmacologic inhibition in MRT models [15, 16, 28–30]Preclinical proof-of-conceptImmunotherapeutic approachesImmune checkpoint blockadeExplores immune activation despite low mutational burdenEarly clinical exploration in pediatric solid tumors including MRT [18, 19]Limited evidence of durable benefitTherapeutic categories are summarized to reflect representative clinical and preclinical approaches evaluated in MRT. Listed strategies do not imply equivalent clinical efficacy, and most molecularly informed approaches remain under active investigatioThis table summarizes conventional and molecularly informed therapeutic approaches evaluated in MRT. Strategies targeting chromatin regulators reflect attempts to modulate transcriptional states established by SMARCB1 loss. Transcription factor interference is included as an emerging architecture-centered approach aimed at disrupting concentrated transcriptional control sustaining survival and cell-cycle programs. Clinical efficacy remains variable, and most targeted strategies are in early-phase or preclinical development Current clinical and molecularly informed therapeutic strategies in MRT Therapeutic categories are summarized to reflect representative clinical and preclinical approaches evaluated in MRT. Listed strategies do not imply equivalent clinical efficacy, and most molecularly informed approaches remain under active investigatio This table summarizes conventional and molecularly informed therapeutic approaches evaluated in MRT. Strategies targeting chromatin regulators reflect attempts to modulate transcriptional states established by SMARCB1 loss. Transcription factor interference is included as an emerging architecture-centered approach aimed at disrupting concentrated transcriptional control sustaining survival and cell-cycle programs. Clinical efficacy remains variable, and most targeted strategies are in early-phase or preclinical development Limits of mutation-centered and pathway-centered models The limited impact of targeted therapies in MRT highlights a broader issue [18, 19]. Many modern cancer treatments are built on a mutation-driven model in which recurrent activating mutations define druggable signaling pathways [12]. In such tumors, inhibition of a dominant oncogenic kinase can produce substantial clinical responses [12]. MRT does not fit this model. Beyond SMARCB1 loss, recurrent activating mutations are rare [4, 5], and no single dominant oncogenic kinase can be targeted to shut down tumor growth. Instead, MRT is characterized by widespread transcriptional dysregulation resulting from chromatin remodeling failure [2–4, 6, 7]. Pathway-centered strategies may therefore address downstream signaling without correcting the structural reorganization of gene control [2–4]. If essential survival and cell-cycle programs are maintained through reorganized transcriptional architecture, targeting individual signaling nodes is unlikely to be sufficient [2–4, 6, 7]. Why a transcription-centered model is needed in MRT Conventional cytotoxic therapy remains toxic and only partly effective, and mutation-centered approaches have not produced durable breakthroughs in MRT [18, 19]. The defining event in this disease is loss of SMARCB1, which disrupts chromatin organization and reshapes transcriptional control at a global level [2–4, 6, 7]. Because MRT lacks recurrent activating kinase mutations, tumor survival cannot be explained by a single dominant signaling pathway [4, 5]. Instead, tumor persistence appears to depend on reorganized transcriptional control [2–4, 6, 7]. If tumor viability is sustained by concentrated transcriptional control rather than discrete oncogenic mutations, a model centered on transcriptional architecture may better explain disease persistence and guide the search for more effective therapeutic targets [11, 12]. In the following sections, we examine how SMARCB1 loss leads to epigenetic collapse and how this process may establish transcriptional bottlenecks that sustain MRT viability. SMARCB1 loss and chromatin reorganization in MRT Architecture of the SWI/SNF complex The SWI/SNF complex regulates nucleosome positioning and controls access of transcription factors to DNA [2–4]. In mammalian cells, SWI/SNF exists in at least two major forms: the BRG1/BRM-associated factor (BAF) complex and the polybromo-associated BAF (PBAF) complex [2, 3]. These complexes share several core subunits but differ in specific accessory components that confer distinct regulatory functions [2, 3]. SMARCB1 is a core subunit present in canonical BAF and PBAF complexes [2, 3]. It contributes to complex stability and chromatin targeting. Structural and biochemical studies indicate that SMARCB1 supports interactions between SWI/SNF and nucleosomal DNA, thereby coordinating chromatin remodeling at promoters and enhancers [2–4]. Loss of SMARCB1 alters SWI/SNF composition and changes genome-wide chromatin occupancy [6, 31–33]. Although other SWI/SNF subunits remain expressed, the absence of SMARCB1 modifies how the complex engages regulatory elements [6, 31–33]. In MRT, this disruption represents the initiating event that reshapes transcriptional control [1, 31]. Enhancer remodeling after SMARCB1 loss Genome-wide analyses in MRT models demonstrate widespread changes in chromatin accessibility following SMARCB1 loss [6, 7]. Assay for transposase-accessible chromatin using sequencing (ATAC-seq) reveals redistribution of accessible chromatin regions, particularly at distal enhancers [6, 7]. Restoration of SMARCB1 partially reverses these enhancer changes, supporting a direct role for SMARCB1 deficiency in enhancer remodeling [6]. Loss of SMARCB1 is associated with both gain and loss of enhancer activity, indicating reorganization rather than uniform suppression [6, 7]. Super-enhancers, which cluster regulatory elements that control genes essential for cell identity and survival, are also reshaped in MRT [7–9]. Altered super-enhancer composition can shift transcriptional priorities and reinforce tumor-specific gene expression programs [7–9]. These observations indicate that SMARCB1 loss reconfigures enhancer architecture and reshapes chromatin accessibility at a global scale (Fig. 2) [6, 7]. Chromatin reorganization following SMARCB1 loss in MRT. A Normal state. In MRT cells prior to SMARCB1 loss, the SWI/SNF complex containing SMARCB1 supports balanced enhancer activity and maintains chromatin accessibility across multiple regulatory elements. PRC2-mediated repression is limited to defined chromatin domains, preserving high regulatory diversity. B SMARCB1 loss and epigenetic collapse. Loss of SMARCB1 disrupts SWI/SNF complex integrity and destabilizes enhancer organization genome-wide. Genome-wide alterations in chromatin accessibility across regulatory elements are accompanied by expanded PRC2-associated repressive domains, resulting in reduced regulatory diversity. C Regulatory concentration and transcriptional bottleneck formation. As regulatory diversity declines, control of essential survival and cell-cycle genes becomes concentrated within limited regulatory domains, including super-enhancer clusters. This structural concentration establishes a transcriptional bottleneck, in which tumor viability depends on focused transcriptional control rather than mutation-driven signaling activation Interaction with polycomb repression SWI/SNF and Polycomb complexes exert opposing effects on chromatin structure [2, 3]. Polycomb repressive complex 2 includes EZH2 as its catalytic subunit and catalyzes trimethylation of histone H3 lysine 27, a mark associated with gene repression [21, 26]. In the absence of SMARCB1, PRC2 activity is often increased at specific genomic regions [21]. This shift promotes silencing of developmental regulators and contributes to transcriptional imbalance. Experimental inhibition of EZH2 partially restores expression of selected genes in SMARCB1-deficient cells, highlighting functional interplay between SWI/SNF and Polycomb pathways [21, 22, 26]. However, enhanced PRC2 activity alone does not fully explain the transcriptional landscape of MRT. Broader chromatin reorganization caused by SMARCB1 loss involves both reduced SWI/SNF-mediated activation and altered repression dynamics [6, 7]. This imbalance contributes to widespread transcriptional instability [6, 7]. Developmental transcriptional instability MRT develops during early childhood, a period characterized by active developmental gene regulation [5, 17]. Genes controlling lineage specification and differentiation remain under dynamic chromatin control during this stage of life [5, 17]. Loss of SMARCB1 during this developmental window may therefore have amplified effects. Disruption of enhancer coordination can destabilize lineage programs and impair normal differentiation [6, 7, 17]. Instead of progressing along defined developmental trajectories, tumor cells may adopt persistent aberrant transcriptional states [6, 7]. This instability creates conditions in which survival and cell-cycle programs are maintained through reorganized transcriptional control. In pediatric tumors such as MRT, chromatin remodeling failure intersects with developmental timing, contributing to sustained transcriptional dysregulation [5, 6, 17]. Structural basis of transcriptional bottleneck formation Loss of regulatory diversity In normal cells, essential gene expression programs are maintained through coordinated activity of multiple enhancers and transcription factors [8, 9]. This layered regulatory organization permits flexibility and partial compensation when individual components are perturbed. Super-enhancer architecture and distributed transcription factor occupancy support coordinated regulation of cell identity programs [8, 9]. In MRT, epigenetic collapse disrupts enhancer coordination and redistributes chromatin accessibility [6, 7]. As enhancer landscapes are reshaped, parallel regulatory inputs that normally support essential genes may be reduced or destabilized. This reduction in regulatory diversity limits the capacity of cells to sustain core programs through alternative regulatory pathways [6, 7]. When multiple enhancers or transcription factors no longer contribute equally, transcriptional control becomes less distributed and increasingly dependent on a smaller number of regulators. This shift from distributed regulation to concentrated transcriptional control represents a structural consequence of chromatin remodeling dysfunction. Concentration of transcriptional control As regulatory diversity decreases, control over essential survival and cell-cycle genes becomes concentrated within limited transcriptional nodes [6, 7]. Such concentration does not necessarily reflect genetic amplification or signaling activation. Instead, it emerges from reorganization of enhancer architecture and redistribution of transcription factor occupancy [6–9]. Genes required for tumor maintenance may cluster within shared regulatory domains, including altered enhancer and super-enhancer domains [7–9]. In this context, disruption of a dominant transcriptional regulator can exert disproportionately large effects because compensatory regulatory inputs are reduced. This structural concentration of transcriptional control distinguishes transcriptional bottleneck formation from classical oncogene activation. Whereas mutation-driven oncogenes derive importance from recurrent genetic alteration, bottleneck regulators assume importance through architectural centrality within a reorganized regulatory landscape [11, 12]. Defining bottleneck nodes We define a transcriptional bottleneck node as a regulator that satisfies three criteria in the context of epigenetic collapse. First, it exhibits enriched chromatin occupancy at genes essential for survival and cell-cycle progression. Chromatin mapping approaches, including chromatin immunoprecipitation and accessibility profiling, can identify such occupancy patterns [6, 7]. Second, compensatory regulatory alternatives are limited. In contrast to distributed regulatory systems described in normal developmental contexts [8, 9], epigenetic collapse reduces parallel enhancer inputs, increasing reliance on the remaining dominant regulator [6, 7]. Third, functional perturbation of the regulator is expected to result in growth arrest or apoptosis in relevant preclinical models. This operational definition aligns conceptually with prior descriptions of transcription-centered dependencies in cancer [11, 12], while emphasizing the architectural consequences of chromatin remodeling dysfunction. Together, these features distinguish bottleneck nodes from classical oncogenes. Their importance arises from structural concentration of transcriptional control rather than mutation-driven signaling activation. Importantly, the concept of transcriptional bottleneck formation is intended as a structural model of regulatory concentration and does not imply exclusivity of a single factor. Representative bottleneck nodes in MRT RUNX as a candidate bottleneck node Within the reorganized transcriptional landscape of MRT established by epigenetic collapse, members of the RUNX family meet the proposed criteria of transcriptional bottleneck nodes [11, 14, 34]. First, chromatin occupancy analyses demonstrate enrichment of RUNX1 at regulatory regions controlling genes essential for survival and cell-cycle progression in MRT models [15, 16]. These observations satisfy the criterion of concentrated binding at essential transcriptional outputs. Second, enhancer redistribution following SMARCB1 loss reduces regulatory diversity, limiting compensatory transcriptional alternatives [6, 7]. In this narrowed regulatory environment, RUNX activity assumes a disproportionately central position, consistent with architectural concentration of transcriptional control [11, 14]. Third, functional perturbation of RUNX activity produces marked impairment of tumor viability. Genetic suppression or pharmacologic interference, including RUNX-targeted DNA-binding inhibitors, reduces expression of survival and cell-cycle genes and induces growth arrest or apoptosis in preclinical MRT models [15, 16, 29]. Importantly, RUNX genes are not recurrently mutated or amplified in MRT [4, 5]. Their functional importance reflects architectural centrality within a reorganized chromatin landscape rather than mutation-driven signaling activation [11, 14]. Among RUNX family members, RUNX1 has been most consistently implicated in MRT models [15, 16]. These findings support its designation as a representative transcriptional bottleneck node within SMARCB1-deficient tumors (Fig. 3). RUNX1 sustains survival and cell-cycle programs in MRT. A RUNX1-associated transcriptional output. In MRT, RUNX1 contributes to transcriptional regulation of genes involved in survival and cell-cycle progression. BIRC5 is shown as a representative downstream target associated with these programs. Sustained RUNX1 activity supports maintenance of survival and cell-cycle transcriptional output and tumor maintenance. B Disruption of RUNX1-associated transcriptional output Reduction of RUNX1 activity decreases BIRC5 expression and reduces transcription of survival and cell-cycle genes. Impairment of RUNX1-associated transcriptional output disrupts these programs and compromises tumor maintenance RUNX-dependent transcriptional control Within this reorganized transcriptional landscape, RUNX family transcription factors have emerged as candidate regulators of survival-associated programs in MRT [13, 34]. RUNX1 has been implicated in maintaining expression of genes required for proliferation and resistance to apoptosis [14–16]. One well-characterized downstream target of RUNX activity in MRT is BIRC5 (also known as survivin), a regulator of mitotic progression and apoptosis resistance [16]. Suppression of RUNX activity reduces BIRC5 transcription and disrupts survival and cell-cycle programs in MRT models [15, 16]. This link provides a mechanistic connection between a bottleneck node and essential transcriptional outputs. In addition to BIRC5, RUNX-regulated genes include representative mediators of cell-cycle progression and mitotic control, including CCNB1, CDK1, and AURKB [13, 14, 34]. These coordinated outputs illustrate how a transcriptional bottleneck node sustains tumor viability through concentrated regulatory control rather than mutation-driven signaling activation [11, 14]. Heterogeneity within MRT MRT exhibits biological heterogeneity. Molecular profiling studies have identified subgroups with distinct transcriptional features and developmental signatures [17–19]. The contribution of individual transcriptional nodes may therefore vary across molecular subtypes. The developmental stage at tumor initiation may also influence regulatory architecture [5, 17]. Because MRT arises during early childhood, variation in developmental context can affect enhancer landscapes and patterns of transcriptional control. Defining transcriptional bottleneck nodes within each molecular subgroup will be important for precision targeting. Although RUNX activity represents a candidate bottleneck node in many MRT models, systematic analysis across patient-derived datasets remains necessary [15, 16]. Representative transcriptional bottleneck nodes and their associated survival and cell-cycle programs in MRT are summarized in Table 2. Table 2Representative transcriptional bottleneck nodes and associated survival and cell-cycle programs in MRTBottleneck node (representative)Representative downstream programs (survival → cell-cycle)Supporting evidence (references)Functional consequence in preclinical models RUNX1 BIRC5 and other survival and cell-cycle programsGenetic suppression and pharmacologic inhibition studies in MRT models [15, 16]Impaired tumor viability with growth arrest and apoptosisBRD9 (ncBAF subunit)Enhancer-linked survival and cell-cycle transcriptional programsFunctional dependency and pharmacologic inhibition in SMARCB1-deficient cancers [23, 24]Impaired tumor viability and proliferationPRC2/EZH2Survival and cell-cycle programs influenced through altered chromatin repression stateEZH2/EZH1/2 inhibition studies in INI1/SMARCB1-loss tumors including MRT [21, 22, 26]Growth suppression with variable and often incomplete responsesNodes are listed as representative examples of concentrated transcriptional control in MRT. Classification is based on evidence of chromatin occupancy, reduced compensatory regulation in the context of epigenetic collapse, and functional sensitivity to perturbation in preclinical modelsImportantly, these nodes are not mutually exclusive and may exhibit context-dependent relevance across molecular subgroups. Functional redundancy or compensatory interactions, particularly among transcription factor families such as RUNX, may influence dependency and therapeutic responseThis table summarizes representative transcriptional bottleneck nodes identified in MRT and their associated survival and cell-cycle transcriptional programs. Nodes are selected based on evidence of concentrated chromatin occupancy, limited compensatory regulatory alternatives in the context of epigenetic collapse, and functional sensitivity to genetic or pharmacologic perturbation in preclinical modelsRUNX1 is shown as a representative example supported by direct functional evidence in MRT models. Additional regulators, including BRD9 and PRC2/EZH2, are included as candidate nodes associated with SMARCB1-deficient chromatin states, although their relative contribution to concentrated transcriptional control may vary across molecular subgroupsFunctional consequences listed reflect preclinical observations and do not imply uniform clinical efficacy Representative transcriptional bottleneck nodes and associated survival and cell-cycle programs in MRT Nodes are listed as representative examples of concentrated transcriptional control in MRT. Classification is based on evidence of chromatin occupancy, reduced compensatory regulation in the context of epigenetic collapse, and functional sensitivity to perturbation in preclinical models Importantly, these nodes are not mutually exclusive and may exhibit context-dependent relevance across molecular subgroups. Functional redundancy or compensatory interactions, particularly among transcription factor families such as RUNX, may influence dependency and therapeutic response This table summarizes representative transcriptional bottleneck nodes identified in MRT and their associated survival and cell-cycle transcriptional programs. Nodes are selected based on evidence of concentrated chromatin occupancy, limited compensatory regulatory alternatives in the context of epigenetic collapse, and functional sensitivity to genetic or pharmacologic perturbation in preclinical models RUNX1 is shown as a representative example supported by direct functional evidence in MRT models. Additional regulators, including BRD9 and PRC2/EZH2, are included as candidate nodes associated with SMARCB1-deficient chromatin states, although their relative contribution to concentrated transcriptional control may vary across molecular subgroups Functional consequences listed reflect preclinical observations and do not imply uniform clinical efficacy Therapeutic implications of targeting transcriptional bottlenecks Conceptual shift from mutation targeting Most modern cancer therapies are designed to inhibit recurrent activating mutations in dominant signaling pathways [12]. In MRT, this strategy has had limited success because tumor maintenance is not driven by mutation-driven kinase activation but by chromatin remodeling failure following SMARCB1 loss [2–4, 6]. Targeting transcriptional bottlenecks represents a conceptual shift from mutation-centered therapy to architecture-centered intervention [11, 14]. Rather than inhibiting a genetically altered enzyme, this approach seeks to disrupt concentrated transcriptional control sustaining essential survival and cell-cycle programs within a reorganized regulatory landscape [11, 14]. This strategy does not replace mutation-based approaches but addresses a distinct biological dependency created by epigenetic collapse in MRT [2–4, 6, 7]. Direct transcriptional interference One strategy is to interfere directly with DNA binding of bottleneck transcription factors. Molecules that block access to specific regulatory sequences can suppress activation of essential survival and cell-cycle genes [29]. In addition to direct DNA-binding interference, alternative therapeutic strategies include disruption of protein–protein interactions, targeted protein degradation approaches, and modulation of transcriptional regulators through chemical-induced proximity. These approaches may provide complementary means to perturb transcriptional control in MRT. However, it should be noted that most transcription factor–targeting strategies remain at a preclinical stage, with limited clinical validation to date. In MRT models, inhibition of RUNX activity reduces tumor growth and induces apoptosis, illustrating that interference at the level of transcription factor–DNA interaction can impair tumor viability [13–16]. These findings provide proof-of-concept that concentrated transcriptional control can be pharmacologically disrupted in chromatin-remodeling–deficient tumors [15, 16, 29]. Importantly, such approaches aim to interrupt structural concentration of transcriptional control rather than inhibit mutation-driven enzymatic activation [11, 12]. Targeting downstream essential programs An alternative strategy focuses on downstream programs controlled by a transcriptional bottleneck node. If survival and cell-cycle genes represent convergent outputs of concentrated regulatory control, their inhibition may produce coordinated antitumor effects [11, 14]. Because regulatory diversity is reduced following epigenetic collapse [6, 7], disruption of core survival programs may exert broader effects than inhibition of individual upstream signaling intermediates. This principle differs from pathway-centered approaches that target isolated signaling components without addressing the underlying concentration of transcriptional control [11, 12]. Evidence from chromatin-driven cancers suggests that targeting individual signaling pathways is often insufficient to reverse transcriptional states established by chromatin remodeling defects. In such contexts, gene expression programs are maintained through coordinated regulatory architectures rather than isolated signaling nodes. Enhancer-centered regulatory strategies have been explored in chromatin-driven cancers such as Ewing sarcoma, where disruption of disease-associated enhancer elements has been shown to suppress tumor growth in preclinical models [35, 36]. These findings support the concept that targeting regulatory architecture can produce coordinated effects on downstream transcriptional programs. Although analogous enhancer- or chromatin-regulatory strategies remain largely preclinical in MRT and related SMARCB1-deficient tumor models [23–25], these observations provide a conceptual framework for exploring enhancer-based interventions in chromatin-remodeling–deficient tumors. In MRT, this downstream-oriented strategy may be particularly relevant when direct targeting of transcription factors remains technically challenging. Biomarker development Effective implementation of bottleneck-directed strategies requires biomarkers of concentrated transcriptional control in MRT. Transcriptional bottleneck formation reflects architectural reorganization of chromatin and enhancer landscapes rather than recurrent genetic alteration [11, 14]. Biomarker development should therefore focus on structural features of transcriptional control. Chromatin accessibility and enhancer architecture Genome-wide chromatin accessibility profiling demonstrates that SMARCB1 loss reshapes enhancer landscapes and alters super-enhancer organization in MRT [6, 7]. In this context, candidate biomarkers include: Reduced regulatory diversity. Reduced regulatory diversity. Enrichment of chromatin occupancy at concentrated transcriptional nodes. Enrichment of chromatin occupancy at concentrated transcriptional nodes. Super-enhancer clustering linked to survival and cell-cycle programs [8, 9]. Super-enhancer clustering linked to survival and cell-cycle programs [8, 9]. ATAC-seq–derived accessibility patterns and enhancer mapping may serve as structural indicators of concentrated transcriptional control by capturing the architectural consequences of epigenetic collapse [6, 7]. Transcriptional signatures of concentrated control RNA expression profiling may reveal coordinated activation of survival and cell-cycle programs associated with specific concentrated transcriptional nodes. Enrichment of RUNX-associated transcriptional output, including BIRC5 and related genes, may indicate dependency on a representative bottleneck node [15, 16]. Integration of transcriptional data with chromatin occupancy mapping will be necessary to distinguish dominant nodes from transcription factors with distributed regulatory roles [6, 7]. Functional sensitivity as a biomarker dimension Because transcriptional bottleneck formation implies limited compensatory regulation, functional sensitivity to perturbation represents a defining biological feature [11, 14]. In preclinical MRT models, suppression of RUNX activity impairs survival and cell-cycle programs and reduces tumor viability [13–16]. Although direct functional testing is not routine in clinical practice, ex vivo assays using patient-derived models may help identify tumors with strong dependency on specific concentrated transcriptional nodes. Such approaches align with a precision strategy based on architectural state rather than mutational status [11, 12]. Variation across molecular subgroups MRT comprises molecular subgroups with distinct enhancer landscapes and developmental programs [17–19]. The identity of dominant concentrated transcriptional nodes may therefore differ among these groups. Biomarker strategies should focus on detecting concentrated transcriptional control within each tumor rather than assuming a single universal regulator. Clinical implementation considerations In pediatric oncology, feasibility and safety are central considerations. Chromatin accessibility assays and transcriptomic profiling increasingly rely on low-input sequencing platforms compatible with limited biopsy material. Standardized analytical workflows will be essential for reproducible identification of concentrated transcriptional nodes. Biomarker development in MRT does not aim to identify a recurrent oncogenic mutation. Instead, it seeks to define tumors in which survival and cell-cycle programs are maintained through concentrated transcriptional control established by epigenetic collapse [2–4, 6, 7]. Such identification may support application of architecture-centered therapeutic strategies targeting representative bottleneck nodes [11, 14]. Conceptual implications and future directions MRT provides a well-defined example of a chromatin-driven cancer in which epigenetic collapse establishes transcriptional bottlenecks despite a simple mutational landscape [4, 5]. This model explains how extensive transcriptional dysregulation can coexist with few recurrent activating mutations [4, 5]. The perspective outlined here shifts attention from mutation-driven signaling activation to structural concentration of transcriptional control [11, 14]. In MRT, loss of SMARCB1 reshapes chromatin organization and reduces regulatory diversity [2–4, 6, 7]. As control over essential survival and cell-cycle programs becomes concentrated within limited transcriptional nodes, tumor viability depends on architectural organization rather than mutation-driven signaling activation [11, 14]. Although the identity of specific bottleneck nodes may vary among SMARCB1-deficient subgroups, the underlying principle of concentrated transcriptional control remains consistent [17–19]. Defining these nodes in a context-dependent manner will be necessary for biologically grounded therapeutic development. Future studies should prioritize systematic integration of chromatin accessibility profiling, enhancer mapping, and functional perturbation analyses in patient-derived models [6, 7, 15, 16]. Such approaches will clarify which regulators operate as dominant transcriptional nodes in distinct MRT subtypes. Because MRT arises during early childhood, translation of architecture-centered strategies requires careful attention to developmental context and safety [5, 17]. Therapeutic approaches that selectively disrupt tumor-specific transcriptional bottlenecks while preserving normal developmental gene regulation represent a critical objective. In this disease context, epigenetic collapse represents not merely a consequence of chromatin remodeling failure but a structural event that reorganizes transcriptional control [2–4, 6, 7]. Recognizing and targeting transcriptional bottleneck formation provides a mechanistically grounded path toward improving outcomes in chromatin-driven pediatric cancers [11, 14]. Discussion This Review proposes that epigenetic collapse following SMARCB1 loss restructures chromatin organization and reduces regulatory diversity in MRT [2–4, 6, 7]. Rather than reflecting activation of a dominant oncogenic mutation, this structural disruption concentrates control of essential survival and cell-cycle programs within limited transcriptional nodes, establishing transcriptional bottleneck formation [11, 14]. MRT combines SMARCB1 inactivation with a simple mutational landscape yet remains highly aggressive, illustrating how tumor maintenance can arise from architectural reorganization of transcription [4, 5]. Importantly, this framework does not redefine tumor suppressor loss as a unique mechanism in MRT, but rather highlights that SMARCB1 functions as a canonical tumor suppressor whose loss leads to a distinct form of chromatin architectural disruption. This distinction clarifies that the defining feature of MRT lies not in tumor suppressor loss itself, but in how such loss reshapes transcriptional control through reduction of regulatory diversity and concentration of transcriptional output [2–4, 6, 7]. This distinction further clarifies that transcriptional bottleneck formation is conceptually distinct from transcriptional dependency or synthetic lethality, as it describes the structural basis that gives rise to such functional requirements. This framework also differs from lineage-specific dependency identified through large-scale perturbation studies, such as Project DRIVE [37]. While lineage-specific dependency reflects context-dependent functional requirements, transcriptional bottleneck formation provides a structural explanation for why such dependencies arise under conditions of reduced regulatory diversity. Reduced regulatory diversity shifts transcriptional control from distributed enhancer networks toward concentrated regulatory nodes [8, 9]. In normal cells, essential genes are supported by coordinated enhancer activity and partially redundant transcription factors [8, 9]. In MRT, loss of enhancer coordination narrows regulatory inputs, increasing reliance on remaining dominant regulators [6, 7]. This transition defines transcriptional bottleneck formation and represents a structural consequence of chromatin remodeling dysfunction [2–4]. Within this reorganized landscape, specific transcription factors may assume central roles because of architectural position rather than genetic alteration [11, 14]. In MRT, members of the RUNX family exemplify this principle [13, 34]. RUNX1 has been linked to maintenance of survival and cell-cycle programs, including regulation of BIRC5, and functional perturbation impairs tumor growth in preclinical models [14–16]. Although RUNX represents a strong candidate bottleneck node, broader validation in patient-derived datasets will be necessary to determine context-specific dependencies [17–19]. Potential compensatory mechanisms among RUNX family members may influence the effectiveness of targeting individual RUNX factors. Functional redundancy between RUNX1, RUNX2, and RUNX3 has been suggested in certain contexts, raising the possibility that inhibition of a single member could be partially offset by others. Such compensatory effects may represent a potential mechanism of resistance and should be considered in the development of therapeutic strategies targeting RUNX-dependent transcriptional control. In addition, other chromatin-associated regulators may function as alternative bottleneck nodes in SMARCB1-deficient contexts. Candidates include SWI/SNF-associated factors and epigenetic regulators such as BRD9 and PRC2/EZH2, which have been implicated in maintaining transcriptional programs in these tumors [27]. These observations support a shift in therapeutic strategy. If tumor viability depends on concentrated transcriptional control, disruption of dominant transcriptional nodes may produce coordinated downstream effects [11, 14]. This architecture-centered approach differs fundamentally from mutation-centered therapy by targeting structural concentration of transcriptional control established by SMARCB1 loss [2–4, 6]. This perspective aligns with broader efforts to define transcription-centered dependencies in SWI/SNF-deficient cancers [27]. Transcriptional bottleneck formation is likely context dependent. Distinct SMARCB1-deficient subgroups exhibit differences in enhancer landscapes and developmental programs, suggesting that dominant transcriptional nodes may vary by lineage and developmental timing [17–19]. Systematic integration of chromatin accessibility profiling, enhancer mapping, and functional perturbation studies in patient-derived models will be essential to define actionable bottleneck nodes across subtypes [6, 7, 15, 16]. Clinical translation requires biomarkers capable of identifying tumors characterized by concentrated transcriptional control [11]. Structural indicators, including altered enhancer architecture and chromatin occupancy patterns, together with coordinated transcriptional outputs, may enable stratification of tumors according to architectural dependency [6–9]. In pediatric oncology, safety considerations and developmental context remain central to implementation of architecture-centered therapeutic strategies [5, 17]. Conclusion Epigenetic collapse caused by SMARCB1 loss reshapes chromatin organization and reduces regulatory diversity in MRT [2–4, 6, 7]. This structural reorganization concentrates control of essential survival and cell-cycle programs within limited transcriptional nodes, thereby establishing transcriptional bottleneck formation despite a simple mutational landscape [4, 5, 11, 14]. In MRT, members of the RUNX family provide a representative example of such a bottleneck node [13, 34]. Their functional importance reflects architectural concentration of transcriptional control rather than mutation-driven oncogenic activation [11, 14]. Interference with RUNX activity provides proof-of-concept that disruption of concentrated transcriptional control can impair tumor viability [14–16]. Recognition of transcriptional bottleneck formation clarifies how chromatin remodeling failure sustains aggressive tumor behavior in the absence of recurrent oncogenic mutations [4, 5]. By integrating chromatin biology with transcriptional organization, this disease-focused perspective supports development of architecture-centered therapeutic strategies in MRT and potentially in other chromatin-driven pediatric cancers [11, 27]. Acknowledgements The authors acknowledge the use of generative artificial intelligence (ChatGPT, OpenAI) to assist with English language editing, preparation of schematic figures (Figs. 1, 2 and 3) and tables (Table 1, and 2), and manuscript formatting tasks. All AI-assisted outputs were reviewed, verified, and finalized by the authors, who take full responsibility for the content. We thank Mr. Sohei Urakami for his understanding of this project and for his continuous support in our efforts to overcome intractable cancers. Abbreviations Malignant rhabdoid tumor SWI/SNF Chromatin-remodeling complex SWI/SNF Complex subunit (also known as INI1/BAF47) Polycomb repressive complex 2 Enhancer of zeste homolog 2 Runt-related transcription factor family Baculoviral IAP repeat–containing protein 5 (survivin) BRG1/BRM-associated factor complex Polybromo-associated BAF complex Atypical teratoid/rhabdoid tumor Author contributions T.M. conducted the literature search and drafted the manuscript.T.W. contributed to the literature search and manuscript drafting.T.O. critically evaluated the literature and refined the conceptual model.Y.K. conceived and designed the overall conceptual framework of the Review, supervised the project, and finalized the manuscript.All authors reviewed and approved the final manuscript. Funding This work was supported in part by a Grant-in-Aid for Scientific Research (KAKENHI; grant numbers 22K07168, 24K11817, 25K11143, 25K11216, and 25K19717), by the Joint Research Project of the Institute of Medical Science, The University of Tokyo, and by an Extramural Collaborative Research Grant from the Cancer Research Institute, Kanazawa University. Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate Approval of the research protocol by an Institutional Review Board: Not applicable. Consent for publication Not applicable. Informed consent Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.