Epigenetic crosstalk between stem cells and tumors: mechanisms and emerging perspectives

Abstract

Stem cells possess self-renewal and multipotent differentiation capabilities, exhibiting broad applications in regenerative medicine and tissue homeostasis maintenance. Their fate regulation relies heavily on precise epigenetic mechanisms. Cancer stem cells (CSCs), as key drivers of tumor heterogeneity, recurrence, and drug resistance, share extensive epigenetic features with normal stem cells, forming a complex and dynamic regulatory network. Mechanisms including DNA methylation, histone modification, chromatin remodeling, and ncRNAs collectively sustain stem cell pluripotency and tumor stemness, while aberrant epigenetic alterations serve as core drivers of tumor initiation and progression. In recent years, with the advent of single-cell omics and CRISPR-dCas9 epigenetic editing technologies, epigenetic “crosstalk” between stem cells and tumor cells has been progressively uncovered, especially the multidimensional epigenetic reprogramming induced by the tumor microenvironment (TME) that promotes CSC traits and drug resistance. This review systematically summarizes the epigenetic regulatory mechanisms of stem cells, epigenetic abnormalities in tumors, their interactions, and translational potential in therapeutic strategies, focusing on frontier topics such as reversible epigenetic plasticity, metabolic-epigenetic interplay, and liquid biopsy epigenetic biomarkers. Looking forward, artificial intelligence (AI) and big data analysis are expected to deepen the understanding of epigenetic heterogeneity, driving integrative innovations in precision medicine and regenerative interventions. Comprehensive understanding of the epigenetic crosstalk between stem cells and tumors will provide solid theoretical support and technical pathways for CSC-targeted therapies, epigenetic drug development, and stem cell fate manipulation.

Introduction

Stem cells, owing to their unique capacities for self-renewal and multilineage differentiation, have demonstrated broad clinical translational potential in regenerative medicine and tissue repair. Promising advances have already been made in treating hematological disorders, neurodegenerative diseases, and organ injuries. However, the success of clinical applications remains highly dependent on the precise regulation of stem cell differentiation fate. Meanwhile, cancer remains a leading global health burden, characterized by high heterogeneity, recurrence, and metastatic potential - features that remain inadequately explained by conventional research frameworks. The CSC theory posits that within tumors exists a subpopulation of cells with stem-like properties, capable of sustaining tumor heterogeneity, progression, and therapeutic resistance through asymmetric division. This suggests that CSCs may share critical epigenetic regulatory programs with normal stem cells.

In recent years, epigenetic mechanisms have emerged as a pivotal link bridging stem cell biology and tumor development. Processes such as DNA methylation, histone modifications, and non-coding RNA regulation reshape gene expression landscapes without altering the underlying DNA sequence. In stem cells, for example, TET family enzymes mediate the oxidation of 5-methylcytosine (5 mC) to facilitate gene activation [1,2]; histone acetylation mediated by p300 plays essential roles in maintaining pluripotency or inducing lineage-specific differentiation [3]; and ncRNAs such as miR-34a contribute to fate determination [4]. In tumors, however, dysregulated epigenetic control acts as a major driving force. Aberrant promoter hypermethylation can silence tumor suppressor genes, histone deacetylation may enhance invasiveness, and disruption of non-coding RNA networks can promote therapeutic resistance. Moreover, key regulators such as EZH2 exert complex and often bidirectional roles in both stem cells and cancer cells [5,6], underscoring a profound epigenetic overlap between these two cellular contexts.

Emerging technologies such as single-cell omics and CRISPR-dCas9-mediated epigenetic editing have provided powerful tools for dissecting stem cell differentiation trajectories, tumor heterogeneity, and the fine-tuned regulatory mechanisms underlying both processes [7]. Nevertheless, how environmental factors influence stem cell homeostasis and tumor evolution through epigenetic reprogramming remains largely unexplored and requires systematic investigation. Future research should focus on integrating clinical specimens, animal models, and precise intervention tools to elucidate the translational potential of epigenetic mechanisms - both in enhancing the safety and efficacy of stem cell-based therapies and in overcoming therapeutic resistance in cancer.

A comprehensive understanding of the epigenetic interconnections between stem cells and tumors holds great promise for unveiling novel mechanisms of disease pathogenesis and advancing breakthroughs in regenerative medicine and precision oncology. With the continued maturation of multi-omics integration and epigenetic intervention technologies, this interdisciplinary field is poised to drive biomedicine toward a new era of individualized and precision-based healthcare.

Epigenetic regulation of stem cells

Epigenetic features of embryonic stem cells (ESCs)

ESCs, derived from early-stage embryos, exhibit unlimited proliferative potential and pluripotency, enabling differentiation into diverse cell types of all three germ layers. These properties confer fundamental significance to ESCs in the fields of developmental biology and regenerative medicine. The maintenance of pluripotency in ESCs is governed by a highly coordinated molecular regulatory network. The core transcription factors Oct4, Sox2, and Nanog form an interconnected regulatory circuit that sustains self-renewal while repressing differentiation, with Oct4 and Sox2 functioning as indispensable components [8,9].

At the epigenetic level, pluripotency-associated genomic regions generally retain a hypomethylated state accompanied by activating histone modifications, such as H3K4me3, which collectively facilitate robust transcription of pluripotency genes [10]. In addition, several signaling pathways are critically involved in preserving pluripotency. The Wnt/β-catenin pathway enhances stemness through the activation of downstream targets [10]; the Notch pathway regulates cell fate determination and prevents premature differentiation [11,12], and the TGF-β pathway contributes to the stabilization of the pluripotent state.

Long non-coding RNAs (lncRNAs) further participate in this regulatory network and demonstrate dual functional roles. While certain lncRNAs positively support pluripotency, aberrant expression can disrupt homeostasis. For example, overexpression of linc-NSC has been shown to induce apoptosis in ESCs and reduce their tumorigenic capacity [13]. Taken together, pluripotency in ESCs is maintained through the integrated action of transcription factors, epigenetic modifications, and multiple signaling pathways. These mechanisms not only ensure the stability of ESC identity but also provide molecular insights into tumorigenesis and reveal potential therapeutic targets.

The pluripotency and self-renewal capacity of ESCs are governed by multilayered epigenetic mechanisms that orchestrate chromatin remodeling and transcriptional network coupling. In the H1-hESC model, high levels of DNA methylation cooperate with core pluripotency transcription factors such as Oct4, Sox2, and Nanog to maintain an open chromatin configuration and precisely control the temporal expression of key genes [14]. This central regulatory axis operates in concert with the histone modification system, forming a functionally complementary framework: H3K4me1, catalyzed by KMT2B, enhances enhancer-promoter interactions and facilitates transcriptional activation. Meanwhile, the bivalent domains marked by H3K4me3 and H3K27me3 at promoters of developmental genes establish a reversible balance between gene silencing and activation, endowing ESCs with both plasticity and responsiveness [15].

In recent years, increasing evidence has highlighted the critical role of RNA modifications in regulating stem cell fate. Epigenetic crosstalk between N6-methyladenosine (m⁶A) and 5mC not only modulates transposable element activity but also plays a pivotal role in cell fate decisions, thereby expanding the conceptual boundaries of epigenetic regulation [16]. The plasticity of chromatin architecture is also fundamental to the maintenance of stemness. For example, Dppa3 induces a 2-cell embryo-like chromatin state, which is essential for zygotic genome activation [17]. At the same time, the DNA methylation machinery exhibits spatiotemporally specific regulatory patterns: DNMT1 controls the timing of de novo methylation by DNMT3 through both catalytic activity-dependent and independent pathways, ensuring the precise inheritance of epigenetic information [18].

The non-coding RNA regulatory network is also deeply embedded within the chromatin regulation system. For instance, Lnc530 modulates local chromatin states by forming a complex with DDX5 and TDP-43 proteins [19]. At the level of differentiation regulation, epigenetic factors exhibit marked functional heterogeneity: miR-146a directs the differentiation trajectory of vascular smooth muscle cells by targeting KLF4 [20], whereas TET-mediated active DNA demethylation influences mitotic fidelity in ESCs via regulation of KHDC3 expression, a process closely linked to tumorigenesis [21]. Notably, the enzymatic activity of TET proteins is finely modulated by the Idax/Rinf signaling axis, ensuring the precise execution of differentiation programs [22]. Dysregulation of such epigenetic regulatory networks is particularly prominent in cancer. For example, BRD9 not only sustains self-renewal in stem cells but also participates in the epigenetic reprogramming associated with pancreatic and breast cancer pathogenesis, underscoring a deep homology between stem cell epigenetic regulation and oncogenic mechanisms [23]. Figure 1 illustrates the key epigenetic features of ESCs, including low DNA methylation levels, enrichment of active histone marks, and distinct chromatin configurations, which together establish the foundational framework for maintaining pluripotency and enabling lineage-specific differentiation.

Figure 1.

Illustrates the key epigenetic features of ESCs.

Epigenetic regulation of adult stem cells

The epigenetic regulatory network precisely governs fate decisions in adult stem cells through mechanisms including DNA methylation, histone modifications (such as H3K27me3), and non-coding RNAs, thereby maintaining a dynamic balance between their quiescent state and differentiation potential. Environmental stimuli or pathological factors can reshape stem cell behavior by altering chromatin accessibility. Mechanistic studies have shown that Wedelolactone promotes chondrogenesis in mesenchymal stem cells (MSCs) by activating the transcription factor FOXO1, which in turn suppresses the activity of EZH2, the catalytic subunit of polycomb repressive complex 2 (PRC2), thereby lifting the epigenetic silencing of chondrogenesis-related genes [24]. In contrast, CTR9 deletion disrupts the bone morphogenetic protein 2 (BMP-2) signaling pathway and impairs the functional integrity of MSCs [25]. In a cardiac injury repair model, inhibition of EZH2 markedly attenuates the cardioprotective effects of MSC-derived exosomes (MSC-EXOs) and accelerates myocardial fibrosis. Interestingly, high mobility group protein A2 (HMGA2) can counteract this pathological effect. Further mechanistic exploration revealed that MSC-EXOs mitigate post-infarction fibrosis by inhibiting EZH2 activity, a process involving EZH2-mediated repression of HMGA2 expression and downstream damage to the phosphoinositide 3-kinase/protein kinase B (PI3K/AKT) signaling pathway [26].

Aberrant epigenetic regulation can lead to impaired stem cell differentiation and contribute to the onset and progression of various diseases. For instance, the long non-coding RNA CIR inhibits the expression of ATOH8, thereby blocking the chondrogenic differentiation of human umbilical cord-derived MSCs (hUC-MSCs). Interestingly, ATOH8 itself positively regulates chondrogenesis via the EZH2/H3K27me3 axis, suggesting a feedback mechanism involving both transcriptional and epigenetic layers [27]. Similarly, in a pathological pregnancy model, exosomes derived from adipose-derived MSCs (AD-MSCs) mediate EZH2-dependent inactivation of the mammalian target of rapamycin (mTOR) signaling pathway, which plays a pivotal role in inducing protective autophagy in hypoxic trophoblasts [28].

The above studies have systematically demonstrated the central role of the EZH2/H3K27me3 axis in regulating stem cell differentiation and revealed how its dysregulation disrupts key signaling pathways such as BMP-SMAD, PI3K/AKT, and mTOR, thereby affecting tissue regeneration and repair (e.g., in the heart and cartilage) or mediating pathological processes (e.g., fibrosis, trophoblast dysfunction). These findings not only deepen our understanding of the molecular mechanisms underlying complex diseases such as tumorigenesis but also provide a critical theoretical foundation for developing precise therapeutic strategies based on exosome delivery systems or epigenetic inhibitors. Adult stem cells play essential roles in maintaining tissue homeostasis and driving regeneration, and their fate decisions are likewise governed by complex epigenetic mechanisms. Compared to ESCs, adult stem cells are generally maintained in a relatively stable yet reversible quiescent state. NcRNAs also contribute to the regulation of adult stem cell activation, proliferation, and differentiation. Table 1 summarizes the major epigenetic features of various adult stem cell types and highlights their critical roles in functional maintenance and fate determination, providing valuable insights into their underlying regulatory networks.

Table 1.

Epigenetic regulatory mechanisms of adult stem cells

Stem Cell Type	Epigenetic Mechanism	Key Regulatory Factors/Enzymes	Target Genes or Pathways	Functional Impact	References
Hepatic Stellate Cells (HSCs)	Regulation via MIR4435-2HG/miR-506-3p axis;	CXCL1, TGF-β1;	TGF-β signaling pathway;	Exacerbates the malignant behavior of HCC cells;	[214,215]
	Regulation by miR-500a-3p transported through HCC-derived exosomes	TGF-β1, IL-10, PD-L1	SOCS2/JAK3/STAT5A/STAT5B signaling pathway	Promotes HCC growth, migration, and regulation of the immune microenvironment
Breast CSCs (BCSCs)	LINC00115 mediates SETDB1-driven methylation of PLK3;	SETDB1, PLK3, HIF1α, ALKBH5;	LINC00115-SETDB1-PLK3-HIF1α-ALKBH5 positive feedback loop;	Induces the BCSC phenotype, enhancing chemoresistance and metastasis;	[108,216]
	lncROPM regulates the mRNA stability of PLA2G16	lncROPM and PLA2G16	PLA2G16 and PI3K/AKT signaling pathways	Maintains BCSC stemness, promoting chemoresistance and tumorigenesis
Colorectal CSCs (CRCSC)	Epigenetic regulation of miR-8063;	miR-8063 and its target hnRNPAB;	The hnRNPAB gene and the Wnt/β-catenin signaling pathway regulate microtubule stability via STMN1	Regulates the self-renewal capacity of colorectal CSCs;	[217,218]
	miR-210-3p-mediated targeting of STMN1	miR-210-3p and STMN1		Promotes the migratory and deformability potential of CRCSC cells
Prostate CSCs (PCSCs)	lncRNA MBNL1-AS1 competitively binds to miR-221-3p;	lncRNA MBNL1-AS1 and miR-221-3p;	Targets CDKN1B to inhibit the Wnt signaling pathway;	Regulates PCSC stemness and tumor proliferation and invasion;	[219,220]
	miR-34a targets and regulates cancer-associated molecules	miR-34a and folate receptor alpha (FOLR1)	Targets molecules such as folate receptor alpha (FOLR1)	Inhibits prostate cancer cell growth and stemness
Pancreatic CSCs (PCSCs)	lncRNA NORAD competitively binds to miR-202-5p;	lncRNA NORAD, miR-202-5p, ANP32E;	ANP32E gene;	Promotes PCSC proliferation and self-renewal;	[221,222]
	miR-630 targets and regulates PRKC	miR-630	PRKCI-Hedgehog signaling axis	Inhibits PCSC self-renewal and tumorigenicity
GSCs	lncRNA INHEG regulates rRNA 2’-O-methylation;	lncRNA INHEG, SUMO2 E3 ligase TAF15, and NOP58;	rRNA 2’-O-methylation pathway mediated by SUMOylation of NOP58;	Promotes GSC self-renewal and tumorigenicity	[223,224]
	EZH2-mediated enhancement of ZNF596/STAT3 signaling	LINC00115 and EZH2	ZEB1 gene and the ZNF596/EZH2/STAT3 pathway
Esophageal CSCs (ECSCs)	SNHG16 functions as a miR-802 sponge to regulate gene expression;	miR-802, PTCH1;	Hedgehog signaling pathway and PTCH1 gene;	Promotes esophageal cancer cell proliferation and self-renewal;	[225,226]
	SNHG12 acts as a sponge for miR-6835-3p and regulates mRNA stability	miR-6835-3p, BMI1, IGF2BP2, β-catenin	BMI1 and CTNNB1 (β-catenin)-related pathways	Enhances cell proliferation, migration, EMT transition, and maintenance of stemness
Bladder CSCs (BCSCs) (CSCs)	DNMT3A mediates hypermethylation of the miR-129-2 promoter;	DNMT3A, SNHG1;	SNHG1/DNMT3A/miR-129-2-5p/Rac1 signaling pathway;	Induces stem-like behavior and cellular invasion;	[227,228]
	LUCAT1 enhances mRNA stability in an m6A-dependent manner	LUCAT1, IGF2BP2, HMGA1	LUCAT1/IGF2BP2/HMGA1 axis	Enhances stem cell phenotype, chemoresistance, and malignant tumor progression
Thyroid CSCs (TCSCs)	CDKN2B-AS1 regulates by sponging miR-122-5p;	miR-122-5p and P4HA1;	miR-122-5p/P4HA1 axis;	Promotes thyroid cancer cell growth, migration, and invasion;	[229,230]
	DOCK9-AS2 regulates gene expression by sponging miR-1972	SP1 and β-catenin (CTNNB1)	Wnt/β-catenin pathway	Promotes tumor proliferation, migration, invasion, EMT, and stemness
Endometrial CSCs (CSCs)	miR-326 targets and regulates gene expression;	GPR91, STAT3, VEGF;	GPR91/STAT3/VEGF signaling pathway;	Inhibits HuECSC proliferation, invasion, angiogenesis, and tumorigenicity;	[231,232]
	miR-137 is upregulated and suppresses PITX2 expression through targeted inhibition	miR-137, PITX2	miR-137/PITX2/MyoD regulatory axis	Suppresses the myogenic differentiation capacity of EN-MSCs

Epigenetic reprogramming of induced pluripotent stem cells

Induced pluripotent stem cells (iPSCs) achieve somatic cell fate reversal through the systematic remodeling of the epigenetic landscape. The core mechanisms involve dynamic regulation of DNA methylation, reconfiguration of histone modifications, and decondensation of higher-order chromatin structures. This process relies on the targeted erasure of genomic imprints by DNA demethylases (e.g., the TET family), chromatin opening mediated by histone acetyltransferases (e.g., p300), and the overcoming of epigenetic barriers such as X chromosome reactivation, ultimately reconstructing an embryonic stem cell-like gene regulatory network. However, residual epigenetic memory - such as donor cell-specific DNA methylation heterogeneity - may impair the differentiation potential of iPSCs and increase their tumorigenic risk. Recent single-cell multi-omics studies have shed light on the molecular characteristics of such residual signatures: in iPSCs with homozygous deletions of different exons in DNMT3A, near-complete loss of DNA methylation is observed, yet the cells retain the capacity for mesenchymal and hematopoietic differentiation. Notably, exon 23-deleted iPSC-derived hematopoietic progenitor cells (iHPCs) partially recapitulate the methylation features of DNMT3A-mutant acute myeloid leukemia (AML) [29]. This phenomenon suggests that functional redundancy among epigenetic enzymes may buffer the consequences of global methylation loss.

Further studies have revealed that histone modification regulation plays a pivotal role in cell fate determination. For instance, HDAC2 modulates the expression of Endophilin-B1 in neuronal subtypes, serving not only as a marker of differentiated neurons but also directly influencing neuronal and mitochondrial metabolic pathways [30].

Moreover, epigenetic reprogramming can be precisely modulated through signaling pathways. Interferon-γ (IFNγ) stimulation enhances the activity of TET enzymes via activation of the STAT3 signaling pathway, thereby promoting X chromosome reactivation - a process directly associated with TET-mediated DNA demethylation [31]. Current research is increasingly focused on integrating CRISPR-based screening technologies with epigenetic editing tools - such as dCas9-TET1 fusion proteins - to explore novel strategies for eliminating epigenetic noise, with the ultimate goal of enabling precise control over iPSC reprogramming trajectories and enhancing their clinical safety.

Similar to ESCs, iPSCs exhibit various “bivalent” epigenetic states during early reprogramming. The interplay between ncRNAs and epigenetic regulators plays a pivotal role in shaping their phenotypic characteristics. Table 2 summarizes the major molecular events and regulatory mechanisms involved in the epigenetic reprogramming of iPSCs, providing critical insights into the principles of reprogramming and its potential links to epigenetic abnormalities in cancer.

Table 2.

Epigenetic reprogramming mechanisms of iPSCs

Reprogramming Stage	Epigenetic Mechanism	Key Regulatory Factors/Enzymes	Target Sites or Regions	Biological Effects	Experimental Model or Cell Type	References
Early Stage	DNA demethylation	TET1/2/3; miR-29a/b	Genome-wide CpG islands;	Early transcriptional reprogramming;	Mouse pulmonary epithelial premalignant cells;	[233,234]
			Promoter region	Silencing of tumor suppressor genes	TET2-knockout RWPE-1 cells
Early Stage	Histone modification	H3K4 methyltransferase, BRD4	Transgene promoter	Pro-metastatic phenotype	Primary PDAC cells; PLC	[235,236]
Intermediate Stage	Non-coding RNA	LncRNA; BMP7	BMP7 promoter	BMP7 expression	Oct4/JDP2-induced iPS-like cells;	[237,238]
					Human ESCC cell line
Intermediate Stage	Non-coding RNA	miRNA; miR-155	Pluripotency gene promoter;	Phenotypic reversibility;	Genome-wide miRNA microarray;	[239,240]
			Ship1 mRNA 3’ untranslated region (3’UTR)	Inhibition of terminal differentiation	Antigen-specific CD8+ T cells
Intermediate Stage	Chromatin remodeling	Core driver factors;	MYC promoter;	Epigenetic heterogeneity;	Patient-derived organoids (PDOs);	[241,242]
		BRG1	FASN promoter	Proliferation dependence	TNBC cell lines TNBC
Maturation Stage	Gene expression activation	SOX2; Ezh2	Super-enhancer-associated oncogenes;	Tumor dependency;	Mouse xenograft model;	[243,244]
			Neuroendocrine gene promoters	Heterogeneous tumor	Patient-derived xenograft (PDX) model
Maturation Stage	Transcription factor regulation	GLI2-NANOG positive feedback loop;	Metastasis-specific enhancer regions;	Distant organ colonization;	Clonosphere-derived tumor model;	[245,246]
		NANOG-MYC positive feedback loop	MYC target gene promoters	Therapeutic resistance (castration/chemotherapy)	CRPC xenograft tumor (LAPC9)
					CRPC
Late Stage	Histone modification	OCT4	OCT4 promoter region	Promotes cellular phenotype transformation	MCF7 breast cancer cell line	[247,248]
Late Stage	Inhibits the Wnt/β-catenin signaling pathway	DNMTs, TET	Wnt/β-catenin signaling pathway	Enhances antitumor and antimetastatic effects	Breast CSCs; LUAD mouse model (LUAD: Lung Adenocarcinoma)	[233,249]
Early Stage	Histone modification	EZH2, miR-490-3p	Host gene of miR-490-3p;	Affects the expression of its target genes;	GBM cell lines (GBM Multiforme cell lines);	[250,251]
			Gene regulatory region of GC B cells (GC B cells: Germinal Center B cells)	Inhibits proliferation	FL cell lines (Follicular Lymphoma cell lines)

Epigenetic aberrations in cancer

Aberrant DNA methylation in tumorigenesis

Aberrant DNA methylation profoundly influences tumor initiation and progression through dual-layered epigenetic mechanisms. On a global scale, hypomethylation can induce chromosomal instability (CIN) and aberrant activation of repetitive genomic elements, thereby significantly enhancing the genomic mutational potential of cancer cells. Concurrently, hypermethylation of CpG islands in the promoter regions of specific tumor suppressor genes (e.g., p16, BRCA1) and DNA damage repair genes leads to transcriptional silencing, collectively forming a core oncogenic mechanism that drives cancer progression. Epidemiological studies indicate that approximately 70% of tumor-associated genes exhibit aberrant methylation. This epigenetic alteration is implicated in the maintenance of CSC properties and clonal evolution, and has emerged as a critical therapeutic target in the field of epigenetic therapy. In solid tumors, global hypomethylation is significantly associated with tumor aggressiveness and clinical prognosis. In pancreatic cancer, clinical studies have demonstrated that DNA hypomethylation can serve as a biomarker for predicting occult metastasis risk [32]. In hepatocellular carcinoma (HCC), gastric cancer, and colorectal cancer, hypomethylation within intragenic regions of the SMAD3 gene suggests dysregulation of the TGF-β signaling pathway [33]. Additionally, hypomethylation of the S100P gene has been identified as a driver of malignant transformation in the serrated adenoma-carcinoma sequence [34]. Notably, in ovarian cancer, hypermethylation of the BRCA1 promoter detectable in patient serum has demonstrated promising potential as a non-invasive diagnostic marker [35].

At the molecular level, aberrant methylation is primarily mediated by the DNA methyltransferase (DNMT) family. In breast cancer, the cooperative action of DNMT1 and DNMT3a induces hypermethylation of the RGMA gene, which subsequently activates the FAK/Src/PI3K/AKT signaling pathway [36]. In endometrial carcinoma, the long non-coding RNA FAM83H-AS1 has been shown to epigenetically regulate ferroptosis by recruiting DNMT1 to enhance methylation of the CDO1 gene [37]. Cancer type-specific methylation landscapes further underscore the complexity of this regulatory network. In gallbladder cancer, hypermethylation of the DLC1 gene directly contributes to oncogenic transformation [38]; in gastric cancer, hypomethylation of LIN28B is independently associated with poor prognosis [39]. In HCC, PDE7B promoter methylation modulates epithelial-mesenchymal transition (EMT) via the PI3K/AKT pathway [40], while in colorectal cancer, hypomethylation of SLCO4A1-AS1 activates the Hsp90/Cdk2/c-Myc axis to promote tumor progression [41]. In glioblastoma (GBM), FOSL2 hypomethylation has been found to drive M2 polarization of tumor-associated macrophages [42]. In hematological malignancies, BCL7A gene silencing via promoter methylation has been implicated in the regulation of leukemic cell proliferation [43], and in colorectal cancer, CXCL14 hypermethylation reveals a novel epigenetic mechanism for TME modulation [43].

Epigenetic therapeutic strategies are making breakthroughs in clinical translation. DNMT inhibitors have shown promising effects in restoring target gene expression and enhancing the sensitivity of tumors to immunotherapy. For example, decitabine combined with PAS1-30nt-RNA effectively suppresses breast cancer metastasis by reactivating silenced genes [44]. Additionally, the regulation of DNMT3A by the SYNCRIP protein has identified a novel therapeutic target for reversing aberrant p16 gene methylation [45]. These advances strongly support the notion that DNA methylation abnormalities are not only hallmark epigenetic features of cancer, but also highly actionable targets in the era of precision medicine. Their clinical potential warrants further in-depth exploration and development.

Aberrant histone modifications in cancer

Aberrant histone modifications play a central role in tumor initiation and progression by altering chromatin accessibility and gene expression patterns. Dysregulated modifications - such as acetylation, methylation, and ubiquitination - can disrupt the dynamic balance of chromatin structure, leading to oncogene activation or tumor suppressor gene silencing. For example, hypoxia promotes sustained EMT in pancreatic ductal adenocarcinoma (PDAC) via the histone methylation-MAPL axis, which is targetable by multidrug therapy, offering a potential strategy to overcome chemoresistance [46]. Further studies have shown that hyperactivation of histone deacetylases (HDACs) induces chromatin condensation and suppresses tumor suppressor gene expression. Meanwhile, aberrant overexpression of the histone methyltransferase EZH2 promotes cancer cell proliferation by catalyzing H3K27me3-mediated gene silencing [47]. In addition, mutations in histone-modifying enzymes - such as MLL gene rearrangements - can lead to abnormal methylation patterns, driving the progression of hematologic malignancies such as leukemia [48]. These epigenetic disruptions not only act as independent oncogenic drivers but also interact with other mechanisms, including DNA methylation and non-coding RNAs, collectively shaping the epigenomic landscape of cancer.

In colorectal cancer (CRC), methylation of H3R117 promotes the citrullination of histone H3 (H3cit), which in turn increases methylation of the IGFBP1 promoter. This process enhances the enrichment of H3K9me2, heterochromatin protein 1 (HP1), and DNMT1 at the promoter region. IGFBP1 is believed to function as a tumor suppressor gene, while MARylation of H3R117 may facilitate CRC progression [49]. Mechanistically, HBO1 acetylates H3K14, H4K8, and H4K12, thereby upregulating CTNNB1 expression and activating the Wnt/β-catenin signaling pathway. A novel small-molecule HBO1 inhibitor, WM-3835, has been shown to effectively suppress the progression of B-cell acute lymphoblastic leukemia (B-ALL) [50]. Similarly, in gastric cancer, hesperetin modulates the activity of CBP, affecting the stability of the DOT1L protein, which leads to a significant reduction in H3K79 methylation and the expression of metastasis-related genes [51].

Key nodes within the epigenetic regulatory network have further revealed potential therapeutic targets in cancer. ASH2L, through its interaction with COMPASS components and specific genomic loci, promotes invasion and migration in triple-negative breast cancer (TNBC) cells, likely via H3K4 methylation at inflammation- and immune response-related genes [52]. Elevated nuclear levels of SREBF1 and histone H2A lysine 130 acetylation (H2A-K130ac) are directly associated with advanced prostate cancer. Reversal of these epigenetic changes can sensitize castration-resistant prostate cancer (CRPC) to abiraterone treatment [53]. In addition, HMGCL-mediated metabolic regulation of histone acetylation activates the FOXM1/β-catenin pathway in GBM, and targeting HMGCL with JIB-04 has been shown to effectively inhibit tumor growth [54].

The synergistic interaction between ncRNAs and histone modifications also warrants significant attention. The transcription factor NONO directly interacts with nuclear PKM2, guiding PKM2-mediated phosphorylation of histone H3 at threonine 11 (H3T11ph) to promote metastasis in TNBC. H3T11ph, in cooperation with TIP60-mediated H3K27 acetylation (H3K27ac), jointly activates SERPINE1 expression, thereby enhancing tumor invasiveness [55]. In ovarian cancer, high expression of PRMT1 facilitates BRD4 methylation, which promotes its phosphorylation. Silencing PRMT1 significantly inhibits tumor progression, highlighting its potential as a therapeutic target [56].

Epigenetic-targeted therapeutic strategies have made breakthrough advances. HDACi alleviate the suppression of PRELP mRNA expression in bladder cancer cells by promoting histone acetylation, and several HDAC is have already entered clinical application [57]. Belinostat effectively kills KRAS-mutant lung cancer cells by modulating mitochondrial metabolism, while Panobinostat exerts anticancer effects via the Nrf2 signaling pathway [58]. Moreover, the combination of HDACi with ferroptosis inducers has been shown to enhance the ferroptotic sensitivity of colorectal cancer cells [59]. In prostate cancer, reversing the SREBF1/H2A-K130ac axis can resensitize CRPC to abiraterone [53], and NIT2, by inhibiting HBO1-mediated H3K14 acetylation and oxidative phosphorylation gene expression, is positively correlated with favorable prognosis in gastric cancer [60].

In summary, aberrant histone modifications reshape the epigenomic landscape of tumors through multilayered regulatory networks, providing a solid theoretical foundation for the development of targeted therapies, such as HDACi, EZH2is, and PRMT1 antagonists. Future research should focus on elucidating the crosstalk mechanisms among different histone modifications to enable precise epigenetic intervention strategies.

Dysregulation of non-coding RNA in cancer

Dysregulated non-coding RNAs (ncRNAs) have become a central component of epigenetic disruption in cancer, playing critical roles in tumor initiation, progression, and therapeutic resistance through multilayered regulatory networks involving epigenetic modifications, post-transcriptional regulation, and protein interactions. As key regulatory hubs of gene expression, different classes of ncRNAs exhibit distinct molecular mechanisms: MicroRNAs (miRNAs) can simultaneously regulate tumor suppressor genes and oncogenes via bidirectional control, influencing tumor behavior in a context-dependent manner; Long non-coding RNAs (lncRNAs) function both by recruiting chromatin-modifying complexes (such as PRC2) to mediate histone methylation, and by acting as competing endogenous RNAs (ceRNAs) to construct intricate epigenetic regulatory networks; Circular RNAs (circRNAs) typically participate in the fine-tuning of oncogenic signaling pathways by functioning as miRNA sponges, thereby modulating downstream gene expression.

Mechanistic studies have revealed the specific roles of ncRNAs in tumor regulation. At the level of microRNA-mediated control, miR-204-3p significantly suppresses pancreatic cancer metastasis by targeting the MGAT1 gene [61], while miR-1269b modulates the malignant behavior of gastric cancer cells by regulating METTL3 expression [62]. The lncRNA MALAT1 promotes breast cancer progression via the miR-561-3p/TOP2A signaling axis [63], whereas miR-488 [64] and miR-432 [65] exert tumor-suppressive effects in breast cancer by targeting FSCN1 and AXL, respectively. In colorectal cancer, miR-330 inhibits tumor cell proliferation by downregulating TYMS expression [66], and miR-195-5p affects metabolic reprogramming in TNBC through the regulation of PSAT1 [67].

The regulatory network of lncRNAs exhibits multidimensional characteristics in tumor biology. SNHG1 promotes breast cancer growth and angiogenesis by regulating STAT6 phosphorylation [68], and also facilitates oral squamous cell carcinoma progression via the RUNX2/GDF10 signaling axis [69]. SNHG14 drives gastric cancer metastasis by activating the EMT program [70], while MYO16-AS1 enhances invasive capacity of bladder cancer cells, contributing to disease progression [71]. In contrast, NBR2 suppresses breast cancer cell proliferation by inhibiting the autophagy pathway [72].

In the field of circRNA regulation, circ_0007379 promotes the maturation of miR-320a, thereby negatively regulating RUNX1 expression [73]. Exosome-derived circPLEKHM1 drives tumor metastasis by upregulating OSMR protein translation [74], while circNEIL3 suppresses colorectal cancer metastasis by binding to and promoting the degradation of the YBX1 protein [75]. These studies highlight how ncRNAs construct intricate interactive regulatory networks, forming complex hierarchical layers of control throughout the entire course of tumor development and therapeutic resistance.

In summary, ncRNA-mediated epigenetic regulatory networks offer a novel perspective for understanding the biological behavior of tumors. Their multilayered and multipathway regulatory characteristics not only deepen our understanding of the mechanisms underlying tumorigenesis but also open new avenues for the development of ncRNA-based precision diagnostic and therapeutic strategies.

Epigenetic interactions between stem cells and cancer

Epigenetic regulation of CSCS

Stem cells and somatic cells exhibit substantial convergence in their epigenetic regulatory mechanisms, encompassing DNA methylation, histone modifications, and non-coding RNA-mediated control. In stem cells, global hypomethylation serves to preserve pluripotency and self-renewal, whereas differentiated cells rely on selective gene silencing through DNA methylation to ensure phenotypic stability. CSC-associated surface markers are similarly subject to methylation-dependent regulation; for example, methylation of Lgr5 has been correlated with favorable prognosis in CRC, highlighting the pivotal role of DNA methylation in both cell fate determination and tumor progression, albeit not as the sole regulatory mechanism [76,77]. Histone modifications further contribute to the plasticity of epigenetic states. The coexistence of the “bivalent” marks H3K4me3 and H3K27me3, enriched at promoters of ESCs and developmental regulators, establishes a poised configuration that facilitates transcriptional activation upon differentiation cues. Although the functional implications of UTX-mediated H3K27 demethylation remain under debate, accumulating evidence indicates its essential contribution to the regulation of these state transitions [78,79]. Moreover, non-coding RNAs, particularly miRNAs and lncRNAs, exert multilayered regulatory effects by modulating translational repression, chromatin remodeling, and transcriptional control, thereby maintaining cellular homeostasis and directing lineage commitment. A notable example is HOTAIR, which activates core stemness factors through the NF-κB pathway, thereby sustaining ovarian CSC self-renewal and driving tumor progression. Therapeutic approaches that integrate epigenetic modulators with strategies targeting HOTAIR have been proposed as promising avenues to mitigate recurrence risk [6].

The maintenance of CSCs and their role in driving tumor heterogeneity and treatment resistance are fundamentally governed by a dynamic regulatory network formed through the interplay between epigenetic mechanisms and non-coding RNA signaling. At the epigenetic level, DNA methylation contributes to oncogenesis by silencing tumor suppressor genes (such as CDKN2A and APC) and activating stemness-associated pathways, notably Wnt/β-catenin, thereby establishing a persistent pro-oncogenic epigenetic memory. Of particular interest is the role of PRMT9 in regulating leukemia stem cell survival and immune evasion, suggesting its potential as a therapeutic target [80-82]. Histone-modifying enzymes, including EZH2, HDACs, and KDM1A/LSD1, dynamically modulate chromatin accessibility to fine-tune the expression thresholds of key stemness genes such as SOX2, OCT4, and LGR5. Representative examples include BEX1, which maintains liver CSC properties via the Wnt/β-catenin pathway, and SLC25A22, which reprograms succinate metabolism to activate the same pathway and induce 5-fluorouracil resistance in colorectal cancer [81,83-87]. Moreover, the cooperative interaction between H3K4 methyltransferases (KMT2B/KMT2D) and DNA methylation has opened new avenues for understanding breast CSC dormancy mechanisms. Interventions targeting UHRF1 or disrupting the DNA methylation-progesterone signaling axis have also demonstrated clinical potential in attenuating CSC phenotypes in both HCC and leukemia [82,85,88].

NcRNAs play a pivotal role in reinforcing epigenetic regulatory effects by establishing multilayered control networks. At the axis level, HotairM1 promotes CSC self-renewal via the HOXA1-Nanog signaling axis, while m6A modification mediated by METTL14/IGF2BP2 regulates malignant transformation of lung cancer stem-like cells through lncRNA AC026356.1 [89,90]. In breast cancer, the lncROPM-PLA2G16 axis has been shown to maintain stem cell pluripotency, while lnc408 supports the stem-like phenotype via the CBY1-β-catenin axis; HOTAIR significantly enhances breast CSC activity by activating the NF-κB pathway [91-93]. In prostate cancer, lncRNA NORAD accelerates CSC proliferation via the miR-202-5p/ANP32E axis. In gastric cancer, WT1-AS suppresses stemness through downregulation of WT1 or hsa-miR-15a-5p via the ONECUT2/β-catenin axis [94-96].

In esophageal squamous cell carcinoma (ESCC), the UPF1/LINC00963/miR-508-5p/SOX2 axis, along with SOX2OT, forms an integrated regulatory network mediating chemoresistance and tumorigenesis. In oral squamous cell carcinoma, miR-146a maintains CSC survival via the CD24-AKT-β-catenin axis, while LSD1 promotes stemness through the KPNA2/PI3K/AKT pathway [87,97,98]. In HCC, TINCR maintains stem-like properties through autophagy regulation, while the DIO3OS-NONO-ZEB1 axis exhibits tumor-suppressive activity. Moreover, m6A modification of MIR4435-2HG/NOP58 drives maintenance of the CSC phenotype [99-101]. In glioma, lncRNA INHEG supports CSC characteristics by linking rRNA methylation to protein translation, and GAS5 suppresses GBM proliferation through epigenetic regulation [102,103]. In ovarian cancer, combining HOTAIR inhibition with epigenetic drugs can reverse platinum resistance, while in breast cancer, LINC00115 has been implicated in chemoresistance via SETDB1-mediated epigenetic regulation [91,104,105].

Epigenetic reprogramming endows CSCs with unique properties of drug resistance and dormancy, enabling them to evade therapy through reversible shifts in epigenetic states under microenvironmental stress. This forms the biological foundation for tumor recurrence and metastasis. Systematic dissection and precise targeting of these regulatory networks hold significant translational potential for the development of innovative combination therapies aimed at overcoming drug resistance, eradicating CSCs, and preventing tumor relapse.

Epigenetic mechanisms underlying the transformation of stem cells into tumor cells

In recent years, advances in stem cell engineering and epigenetic research have highlighted the pivotal role of stem cells in tumor initiation, progression, and therapeutic intervention. MSCs and iPSCs, owing to their tumor-tropic properties, have been utilized as vehicles for targeted drug delivery. By carrying anti-tumor agents or gene-editing tools, they enable precise targeting of the tumor microenvironment, thereby enhancing therapeutic efficacy while minimizing adverse effects. Evidence indicates that MSCs can promote fatty acid oxidation (FAO), which in turn enhances GC cell stemness and chemoresistance; conversely, the combination of FAO inhibitors with chemotherapy has been proposed as a promising strategy to overcome this resistance. Furthermore, the MSC-associated long non-coding RNA HCP5 has been shown to aggravate chemoresistance through specific signaling axes [106,107].

Beyond drug delivery, stem cells contribute to immunotherapy by secreting cytokines or serving as carriers for tumor vaccines to potentiate immune responses. They also participate in epigenetic regulation and gene repair delivery strategies designed to reverse drug resistance. Notably, the LINC00115/SETDB1 axis and the lncROPM-PLA2G16 signaling pathway have been identified as key mediators of chemoresistance in breast CSCs [108,109]. Collectively, these findings underscore that stem cells not only exert essential regulatory functions in tumorigenesis but also provide novel strategies for drug delivery, immunotherapeutic enhancement, and the reversal of therapeutic resistance.

In EpCAM-positive hepatic progenitor stem cells (HpSCs), aberrant epigenetic regulatory networks drive tumorigenesis through multilayered synergistic interactions. The DNA methylation landscape exhibits a paradoxical pattern characterized by global hypomethylation alongside localized hypermethylation, resulting in oncogene activation and transcriptional silencing of key tumor suppressor genes such as p16 and BRCA1. This bidirectional epigenetic dysregulation disrupts the threshold of cell fate determination and acts as a crucial molecular switch for the malignant transformation of HpSCs into CSCs, contributing to the development of hepatoblastoma (HB) and HCC [110].

Aberrant regulation at the level of histone modifications also constitutes a core mechanism in malignant transformation. The histone demethylase KDM5B remodels chromatin architecture by specifically removing H3K4me3 marks, while functional inactivation of its negative regulator, hsa-miR-448, further lifts the repression of the KDM5B-MALAT1 signaling axis, forming a positive feedback loop that sustains tumor cell stemness. Notably, this mechanism has been experimentally validated in a TNBC metastasis model [111]. Non-coding RNA-mediated epigenetic reprogramming also plays a pivotal role. The long non-coding RNA DUXAP10, by recruiting PRC2, induces local CpG island hypermethylation and reshapes chromatin accessibility landscapes during cadmium-induced carcinogenesis [112]. In HCC, m6A methylation serves as an upstream driver of lncRNA MIR4435-2HG expression, which - through recognition by the YTHDF protein family - initiates a NOP58-dependent stemness maintenance program [101].

The epigenetic hub protein SATB2 exhibits a unique dual regulatory function: by integrating DNMT activity with histone acetylation, it simultaneously silences differentiation-related gene expression and activates the Wnt/β-catenin signaling pathway. This dual mode of regulation directly drives the malignant transformation of prostate epithelial progenitor cells (PrECs) into a CSC-like phenotype [113]. In ESCC, miR-191-3p regulates the CXCR4/PI3K/AKT signaling axis by targeting and suppressing RGS1 expression. Its stemness-regulating effect shows a strong correlation with HDAC activity [114]. The histone demethylase KDM1A establishes a cross-regulatory network between the histone code and non-histone modifications by coordinately removing H3K4me1/2 marks and stabilizing HIF-1α protein via deubiquitination. This leads to transcriptional repression of the Wnt pathway antagonist APC2 and aberrant activation of DKK1, a mechanism identified as a central molecular event in maintaining thyroid cancer stemness [86].

Of particular interest is the TET1/FOXO4 transcriptional regulatory axis, which modulates β-catenin nuclear translocation via 5-hydroxymethylcytosine (5hmC) modification. During gastric cancer metastasis, this axis forms a positive feedback loop with the canonical Wnt signaling pathway, revealing the dual functional nature of TET family enzymes as key executors of epigenetic reprogramming [115]. Collectively, these studies outline a molecular landscape of epigenetic memory remodeling, wherein imbalanced DNA methylation, disrupted histone modifications, and non-coding RNA regulatory networks drive a cascade amplification effect that facilitates the phenotypic transformation of stem cells into tumor-initiating cells. This provides a novel epigenetic perspective for understanding tumor initiation and progression. The transformation of stem cells into cancer cells is typically accompanied by aberrant DNA methylation, dysregulated histone modifications, reorganization of higher-order chromatin structure, and perturbation of non-coding RNA networks - all of which collectively promote the aberrant activation of stemness-associated genes and silencing of tumor suppressor genes. Figure 2 provides a systematic overview of the key epigenetic events, molecular pathways, and their interactions involved in the transformation of stem cells into tumor cells, offering a visual framework for deciphering the mechanisms of stem cell malignant conversion and identifying potential intervention strategies.

Figure 2.

A comprehensive schematic of critical epigenetic events.

Epigenetic influence of the TME on stem cells and cancer cells

The epigenetic interplay between stem cells and tumors is reflected not only in their shared gene regulatory mechanisms but also in the establishment of complex multilayered networks mediated by DNA methylation, histone modifications, and non-coding RNAs. These networks play critical roles in sustaining the characteristics of CSCs and in driving therapeutic resistance [116]. Although stem cells exhibit considerable potential in drug delivery and immune modulation, their genomic instability and epigenetic plasticity remain major challenges to clinical application. Emerging epigenetic editing technologies, such as CRISPR/Cas9, offer promising opportunities to target key regulatory modifications, optimize stem cell-based therapies, and enhance immune responses [117,118].

In addition, stem cells actively remodel the tumor epigenetic landscape within the tumor microenvironment (TME) by modulating immune cells, fibroblasts, and cytokine networks. For example, exosomal H19 derived from cancer-associated fibroblasts (CAFs) has been shown to promote stemness and drug resistance in CRC [119]. These findings underscore the necessity for future research to systematically dissect the stem cell-tumor epigenetic regulatory networks, while integrating epigenetic editing, immune modulation, and stem cell engineering to accelerate the clinical translation of precision medicine and personalized anticancer strategies.

The TME plays a critical role in driving malignant tumor progression through multidimensional epigenetic regulatory networks. Under hypoxic conditions, a central component of the TME, the expression of DNA methyltransferase 3B (DNMT3B) is upregulated, leading to hypermethylation of the promoter region of microRNA-485-3p (miR-485-3p). This epigenetic alteration not only directly sustains CSC properties and chemoresistance in PDAC [120], but also reveals a key regulatory mechanism by which hypoxia suppresses TET dioxygenase activity, resulting in a global increase in DNA methylation. Further mechanistic analysis has shown that the hypoxia-activated spliced variant of XBP1 (XBP1s), independent of the hypoxia-inducible factor 1α (HIF1α) signaling pathway, forms a transcriptional regulatory complex with histone deacetylase 2 (HDAC2) and histone methyltransferase EZH2. This complex remodels the chromatin architecture associated with ΔNp63α, providing an epigenetic foundation for breast cancer metastasis [121].

This regulatory network exhibits dual epigenetic effects at the level of histone modifications. Hypoxia-induced histone H3 lysine 9 lactylation (H3K9la) enhances the invasive capacity of ESCC by activating the expression of laminin subunit gamma 2 (LAMC2) [122]. Meanwhile, E2F transcription factor 1 (E2F1) mediates the aberrant alternative splicing of serine/arginine-rich splicing factor 7 (SRSF7), providing novel insight into how epigenetic plasticity influences breast cancer progression through alternative splicing regulation [123].

The cytokines secreted by stromal fibroblasts, such as transforming growth factor-β (TGF-β) and interleukin-6 (IL-6), constitute another crucial regulatory axis. Pan-cancer analysis of the long non-coding RNA MIR210HG has shown a significant positive correlation between its expression levels and both tumor hypoxia and immune cell infiltration [124], highlighting the synergistic role of microenvironmental factors in driving epigenetic imbalance. On the level of metabolic regulation, lactate dehydrogenase (LDH) activity between tumor and stromal cells leads to the accumulation of L-2-hydroxyglutarate (L-2HG), which perpetuates a malignant cycle by maintaining CSC traits and promoting an immunosuppressive microenvironment [125]. Meanwhile, ALL1 fusion gene-associated protein 9 (AF9) bridges HIF1α and acetylated c-Myc to form a transcriptional complex, thereby revealing the central role of epigenetic regulators in HCC progression from a metabolic reprogramming perspective [126].

These dynamic interactions indicate that the TME orchestrates a multilayered epigenetic reprogramming system -comprising DNA methylation, histone modifications, and non-coding RNA regulation - which not only intensifies molecular heterogeneity but also confers drug resistance and metastatic potential to tumor cells. This theoretical framework provides a critical scientific basis for the development of therapeutic strategies targeting the epigenetic-microenvironmental interaction axis and underscores the need for future interventions to consider the synergistic regulatory network involving metabolism, epigenetics, and the immune microenvironment. Components of the TME - such as inflammatory cytokines, hypoxic conditions, tumor-associated fibroblasts, and immune cells - can modulate the activity of DNMTs and histone-modifying enzymes, thereby influencing chromatin accessibility and non-coding RNA expression. These changes, in turn, promote the maintenance of stemness, drug resistance, and invasive phenotypes. Figure 3 summarizes the key molecular pathways and mechanisms by which the TME regulates the epigenetic states of stem and tumor cells, offering a visual reference for understanding their dynamic interactions.

Figure 3.

Effects of the TME on the epigenetics and metabolism of stem cells and tumor cells.

Epigenetic therapy strategies targeting stem cells and tumors

Application of epigenetic drugs in cancer treatment

DNA methylation inhibitors such as 5-aza-2’-deoxycytidine (5-Aza-CdR, decitabine), 5-azacytidine (5-AzaC, Vidaza), and 5’-fluoro-2’-deoxycytidine (FdCyd) can suppress DNMT1 activity, thereby reactivating the INK4a/ARF and CIP/KIP families, significantly inhibiting the growth of colorectal cancer HCT-116 cells and inducing apoptosis [127]. The DNA demethylating agent 5-azacytidine (5-aza) has been shown to reduce cell proliferation and increase the expression of glial fibrillary acidic protein (GFAP). Additionally, 5-aza can enhance the therapeutic efficacy of the DNA-damaging agent temozolomide (TMZ) in both subcutaneous and orthotopic PDX models of IDH1 R132H-mutant gliomas. As a monotherapy, 5-Aza offers survival benefits, but its combination with TMZ demonstrates optimal efficacy in two distinct IDH1 R132H-mutant glioma models [128]. The combination of nivolumab and 5-aza has shown acceptable safety, increased CD3+ and CD8+ T cell infiltration, and good tolerability in heavily pretreated pediatric patients with relapsed/refractory acute myeloid leukemia (R/R AML) at the RP2D (3 mg/kg/dose) [129]. PTEN-deficient GBMs evade immune surveillance through inhibition of the ERV-MAVS-IFN axis. Combined treatment with an EZH2i and 5-Aza epigenetically reactivates viral mimicry pathways, overcoming this resistance and offering a promising strategy to enhance antitumor immunity and improve prognosis in patients with PTEN-deficient GBM [130]. Epigenetic priming with oral azacitidine (CC-486) prior to R-CHOP chemotherapy shows acceptable safety in previously untreated patients with intermediate- to high-risk diffuse large B-cell lymphoma (DLBCL) or grade 3B/transformed follicular lymphoma. Moreover, CC-486 combined with vincristine and prednisone (R-CHOP) demonstrates initial signs of safety and antitumor activity in this patient group [131]. In human breast cancer cell lines, decitabine induces transcriptome reprogramming mediated by DNA methylation [132]. Vitamin C enhances DNA hydroxymethylation induced by decitabine or azacytidine in colon cancer cells and subsequently reactivates the epigenetically silenced tumor suppressor CDKN1A [133]. Furthermore, 5-aza-2’-deoxycytidine (decitabine) increases the expression of cancer-testis antigens and alters immune checkpoint expression in head and neck squamous cell carcinoma, particularly in CD39-positive CD8 and CD4 T cells [134].

HDAC inhibitors, either as monotherapy or in combination with other agents, can reactivate silenced genes by remodeling tightly condensed chromatin structures. This reactivation subsequently induces differentiation, cell cycle arrest, or apoptosis [135]. The HDAC inhibitor Vorinostat exerts anticancer effects in TNBC cells by preventing nitric oxide-driven histone deacetylation [136]. Small molecules that bind G-quadruplex DNA structures synergize with the HDAC inhibitor SAHA (Vorinostat) to enhance antitumor efficacy in both gemcitabine-sensitive and gemcitabine-resistant pancreatic cancer cells [137]. The combination of Vorinostat and isotretinoin (13-cis-retinoic acid) is safe and tolerable in patients with advanced renal cell carcinoma and has shown response in refractory metastatic cases [138]. Vorinostat also inhibits the expression of small-cell transcription factors and induces cell death in giant congenital melanocytic nevus cells [139]. The combination of 5-Aza-CdR and SAHA can induce growth arrest and apoptosis via the JAK/STAT signaling pathway [140]. HDACi such as valproic acid (VPA) and SAHA enhance both trastuzumab-mediated phagocytosis and trastuzumab-independent cytotoxicity. The immunomodulatory properties of these HDACis support their rational combination with monoclonal antibodies for cancer therapy [141]. The combination of SAHA and cisplatin synergistically induces apoptosis in hepatic-like adenocarcinoma cells that produce alpha-fetoprotein [142]. In lung adenocarcinoma, SAHA enhances antitumor immunity through the HDAC1/JAK1/FGL1 axis, affecting CD8+ T cell-mediated cytotoxicity [143]. The HDAC inhibitor Romidepsin (FK228) inhibits the proliferation of endometrial cancer (EC) cells and induces apoptosis via activation of the p53/p21-caspase/PARP signaling cascade [144]. FK228 also modulates the expression of the immune checkpoint ligand PD-L1 in colorectal cancer and suppresses immune cell function [145]. Furthermore, combination therapy with FK228 and radiotherapy sensitizes fusion-positive rhabdomyosarcoma cells to radiation [146].

Panobinostat (LBH589), another HDAC inhibitor, shows potent anti-myeloma activity both in vitro and in vivo when combined with the β-catenin inhibitor Tegavivint (BC2059) [147]. Panobinostat also inhibits the Wnt/β-catenin signaling pathway by upregulating APCL expression in breast cancer cells [148]. More broadly, pan-HDAC inhibitors can reverse breast cancer metastasis induced by NEDD9 overexpression [149].

In the treatment of CSCS via epigenetic regulation, specific DNA methylation sites play pivotal roles. iDNMT-induced DNMT1 inhibition enhances the expression of miR-27-3p in lung squamous cell carcinoma (LSCC), thereby suppressing activation of the Notch pathway. Furthermore, the combination of iDNMT and radiofrequency ablation (RFA) shows promise as a potential therapeutic strategy for LSCC [150]. SOX17, a tumor-suppressive transcription factor, is downregulated in ESCC due to promoter hypermethylation. Dysregulation of the SOX17/NRF2 axis contributes to resistance to radiochemotherapy, positioning it as a novel therapeutic target in ESCC [151]. Targeting DNMT1 reshapes the global DNA hypomethylation pattern, enhancing anticancer efficacy while minimizing potential toxicity through balanced signaling synergy. DNMT1 thus acts as a key gatekeeper of fate and treatment outcomes in oral squamous cell carcinoma [152]. Overexpression of SOX17 increases radiosensitivity and chemosensitivity in esophageal cancer by transcriptionally downregulating DNA repair and damage response genes [153]. SOX7 contributes to apoptosis regulation via control of the MAPK/ERK-BIM signaling pathway [154,155]. PARP inhibitors enhance the antitumor effects of DNMT inhibitors in cholangiocarcinoma by promoting tumor cell senescence [156]. ARID1A modulates transcription and the epigenetic landscape through POLE and DMAP1; its deficiency or pharmacologic inhibition renders germ cell tumor cells sensitive to ATR inhibitors [157]. The methylation status of the Nanog promoter determines the transition between cancer cells and CSCS [158]. piRNA-823 contributes to the regulation of CSCS by altering DNA methylation associated with luminal breast cancer [159]. BEX1, regulated by DNMT1, is essential for the self-renewal and maintenance of liver CSCS via activation of Wnt/β-catenin signaling, making it a potentially valuable therapeutic target in HB and CSC-driven hepatocellular carcinoma (CSC-HCC) [160]. ABI2 has been identified as a novel therapeutic target for liver CSCS, promoting CSC activity and tumor recurrence through the MEOX2/KLF4-NANOG axis [161]. DNA methylation-mediated silencing of miR-7-5p enhances the invasiveness of gastric CSCS by upregulating Smo and Hes expression [162]. Lastly, DNMT3b-mediated hypermethylation of the SPAG6 promoter influences LSCC progression via the JAK/STAT signaling pathway [163].

Histone-Modifying Enzymes and ncRNAs in CSC Epigenetic Therapy EZH2, a key histone methyltransferase, has been identified as a CSC marker in clear cell renal cell carcinoma and a target of the antitumor effects of epigallocatechin gallate (EGCG) [164]. As a core component of the polycomb repressive complex, EZH2 plays a crucial role in maintaining CSC traits. In ovarian cancer, EZH2 activates CHK1 signaling, thereby sustaining CSC characteristics and contributing to chemoresistance [165]. UTMD-mediated shEZH2 silencing suppresses stemness and EMT of hepatic CSCs in vitro and in vivo by modulating the STAT3/PI3K/AKT pathway [166]. Capsaicin significantly inhibits stemness expression and metastatic potential in osteosarcoma. It suppresses stem-like properties and migration by downregulating SOX2 and EZH2, positioning capsaicin as a promising therapeutic agent against osteosarcoma metastasis [167]. lncRNA TUG1 inhibits the CSC-like properties of temozolomide-resistant glioma cells via interaction with EZH2 [168]. Downregulation of KDM2B and EZH2 reduces stemness in colorectal cancer cells, and their interaction may serve as a novel prognostic marker and therapeutic target in colorectal cancer [169]. In squamous cell carcinoma, EZH2 promotes the CSC phenotype by regulating the SETDB1/ΔNp63α axis through RUNX3 [170]. In HCC, inhibiting EZH2 mitigates sorafenib resistance by targeting NOTCH1-dependent CSCs via NOTCH1-associated miRNAs [171]. In PDAC, lncRNA HOTAIR promotes invasion and migration by sponging miR-34a and activating the JAK2/STAT3 pathway, thereby enhancing CSC-like features [172]. Ectopic expression of miR-34a/-328 sensitizes breast CSCs to γ-irradiation or doxorubicin by targeting BCL2/ABCG2, with miR-34a and miR-328 respectively downregulating Bcl-2 or ABCG2 to enhance apoptosis [173]. The DNMT1/miR-34a/FOXM1 axis contributes to hepatic CSC maintenance; thus, inhibiting DNMT1/miR-34a-mediated FOXM1 upregulation may suppress HCC via CSC targeting [174]. Cancer-associated fibroblasts promote CSC traits and chemoresistance in colorectal cancer by transferring exosomal lncRNA H19. H19 acts as a ceRNA for miR-141, thereby activating the β-catenin pathway, whereas miR-141 inhibits CSC properties in colorectal cancer cells [119]. Extracellular vesicles rich in lncRNA H19 derived from gastric CSCs promote tumorigenesis and metastasis by mediating intratumoral communication [175]. miR-675-3p, derived from lncRNA H19, directly targets SOCS5 to activate the STAT3 pathway, promoting EMT and stemness in human pancreatic cancer cells [176].

Stem cell-mediated tumor therapy and epigenetic regulation

Human endometrial MSCs (eMSCs) have emerged as novel anticancer agents targeting breast CSCS [177]. MSCs are promising cellular vehicles for delivering anticancer agents into malignant tumors. For instance, tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) has been site-specifically integrated into iPSC-derived MSCs for targeted cancer therapy [178]. Tumor-intrinsic signals have also been utilized to induce MSC-mediated suicide gene therapy to combat malignant gliomas [179]. Glioma-associated MSCs (GA-MSCs) play a critical role in shaping the immunosuppressive microenvironment. As amplifiers of exosome-mediated immunosuppressive signaling in GBM, MSCs enhance the immunosuppressive activity of myeloid-derived suppressor cells (MDSCs) through a positive feedback loop involving miR-21/SP1/DNMT1 [180]. Extracellular vesicle (EV) mimetics derived from iPSC-MSCs have shown therapeutic benefits in metastatic prostate cancer. These EV-like nanovesicles, with theoretically unlimited scalability and consistent biological activity, represent a promising platform for targeted anticancer drug delivery [181]. Similarly, iPSC-derived neural stem cells (NSCs) have been used as cellular vehicles for treating gliomas and breast cancer [182]. Differentiation of MSCs from human iPSCs results in downregulation of c-Myc and DNA replication pathways and immunomodulation of CD4+ and CD8+ cells. iPSC-MSCs exhibit low tumorigenicity and strong immunomodulatory capacity, making them attractive therapeutic candidates regardless of the reprogramming method or original cell source [183]. PTEN, a frequently mutated tumor suppressor in various cancers, plays a pivotal role in stem cell transformation. PTEN deficiency reprograms human NSCs into GBM stem-like cells by interacting with CREB/CBP and repressing PAX7 transcription via promoter binding. PTEN loss leads to PAX7 upregulation, which in turn drives the oncogenic transformation of NSCs and restores the invasive phenotype of human GBM stem cells (GSCs). Notably, PTEN disruption transforms NSCs but not MSCs into tumorigenic cells [184]. Adipose-derived stem cell (ADSC) exosomes are effective carriers for in vivo small molecule delivery. miR-138-5p-loaded exosomes show therapeutic potential for prostate cancer by inhibiting migration, invasion, and proliferation of bladder cancer cells in vitro, and by penetrating tumor tissue and suppressing xenograft tumor growth in vivo [185]. Exosomes derived from MSCs have also been used as nano-carriers for vincristine sulfate (VCR) in treating breast CSCs Loading VCR into these exosomes enhances targeted delivery and reduces side effects [186]. The ExoDS platform, based on exosomes, serves as a multifunctional drug delivery system targeting both cancer cells and CSCs [187]. MSC-derived EVs delivering microRNA-34a-5p suppress colorectal tumorigenesis through the c-MYC/DNMT3a/PTEN axis [188]. In summary, stem cells offer multifaceted advantages in cancer therapy - not only enabling targeted delivery of antitumor agents but also modulating the epigenetic landscape of tumor cells to affect their proliferation, differentiation, and drug resistance. These strategies, combined with gene editing and epigenetic regulation, expand new avenues for precision oncology. Table 3 systematically summarizes the major stem cell-mediated therapeutic approaches and their corresponding epigenetic regulatory mechanisms, providing a reference for comparative analysis and future research.

Table 3.

Stem cell-mediated tumor therapy and epigenetic regulatory mechanisms

Stem Cell Type	Tumor Type	Epigenetic Regulatory Mechanism	Key Factors/Targets	Pathways or Signaling Mechanisms	Application Strategy	References
MSCs	Gastric Cancer	LncHCP5 sponges miR-3619-5p;	lncHCP5; lncRNA MACC1-AS1	HCP5/miR-3619-5p/PPARGC1A/PGC1α/CEBPB axis drives FAO;	Targeting lncRNA HCP5 enhances chemotherapy efficacy;	[106,107]
		lncRNA MACC1-AS1 antagonizes and regulates miR-145-5p		TGF-β1/SMAD2/3-induced MACC1-AS1 promotes FAO (FAO: Fatty Acid Oxidation)	FAO inhibitors combined with chemotherapy overcome gastric cancer drug resistance
CSCs	Colorectal Cancer	lncTUG1 enhances GATA6 protein stability;	lncTUG1, GATA6; HOTAIR, miR-211-5p, FLT-1	GATA6-BMP signaling pathway;	Targeting the TUG1/GATA6-BMP pathway for CRC treatment;	[252,253]
		lncRNA HOTAIR sponges miR-211-5p to regulate gene expression		HOTAIR/miR-211-5p/FLT-1 axis regulates cell stemness	Targeting the HOTAIR/miR-211-5p/FLT-1 axis as a potential therapeutic approach
CSCs	Breast Cancer	XIST sponges let-7a-2-3p;	XIST, let-7a-2-3p, IL-6, STAT3; LINC00589, miR-100, miR-452, DLG5, PRDM16	IL-6/STAT3 signaling pathway promotes CSC self-renewal;	Targeting the XIST/let-7a-2-3p/IL-6 axis inhibits breast CSCs;	[254,255]
		LINC00589 acts as a ceRNA to sponge miR-100/452		LINC00589-miR-100-DLG5 and LINC00589-miR-452-PRDM16-mucin4 axes	Serves as a diagnostic, prognostic biomarker and therapeutic target for HER2-positive breast cancer
CSCs	Liver Cancer	miR-22 inhibits CBL and indirectly upregulates SPRY2;	miR-22, CBL, SPRY2; DIO3OS, NONO, ZEB1	miR-22-3p/CBL/SPRY2/ERK axis inhibits EMT and related processes;	Targeting the miR-22/SPRY2/ERK axis for liver cancer treatment;	[256,257]
		lncRNA DIO3OS post-transcriptionally suppresses ZEB1 expression		DIO3OS-NONO-ZEB1 axis	Targeting the DIO3OS-NONO-ZEB1 axis for HCC therapy
					HCC
PCSC	Prostate Cancer	NUMB targets and inhibits the oncogenic effects of miR-9-5p;	NUMB, miR-9-5p; miR-34a	NUMB/miR-9-5p axis regulates PCSC characteristics;	Inhibiting miR-9-5p or overexpressing NUMB for prostate cancer (PCa) treatment;	[220,258]
		miR-34a mediates post-transcriptional suppression of target genes		TP53-miR-34a axis regulates PCSC-associated genes (PCSC: Prostate CSCs)	Developing ligand-conjugated miR-34a for targeted therapy of PCa
CSCs	Pancreatic Cancer	LINC00909 reduces SMAD4 mRNA stability;	LINC00909 and SMAD4; LncMIR4435-2HG, miR-1252-5p, STAT1	MAPK/JNK signaling pathway is activated;	Targeting LINC00909 inhibits pancreatic cancer metastasis; targeting MIR4435-2HG suppresses CSCs and enhances chemosensitivity.	[259,260]
		LncMIR4435-2HG sponges miR-1252-5p		MIR4435-2HG/miR-1252-5p/STAT1 axis
CSCs	Lung Cancer	lncMir100hg enhances H3K14 lactylation modification;	Mir100hg, HNRNPF, HNRNPA2B1, ALDOA; PGM5-AS1, miR-1247-5p, RSPO1	ALDOA drives glycolysis and histone lactylation;	Targeting HNRNPF/HNRNPA2B1 or the histone lactylation pathway;	[261,262]
		lncRNA PGM5-AS1 acts as a molecular sponge for miR-1247-5p		PGM5-AS1/miR-1247-5p/RSPO1 regulatory axis	Targeting the PGM5-AS1-miR-1247-5p-RSPO1 axis to inhibit tumor progression
CSCs	Melanoma	Exosomal miR-1268a regulation;	miR-1268a; lncRNA HotairM1, HOXA1, Nanog	Inhibits the autophagy pathway;	Block exosomal miR-1268a transfer or activate autophagy;	[263,264]
		HotairM1-mediated histone methylation and acetylation		HOXA1-Nanog signaling loop	Target HotairM1 or the HOXA1-Nanog loop to enhance HotairM1 expression
GSCs	GBM	lncCASCADES regulates SOX2 epigenetically;	lncCASCADES, SOX2; miR-370-3p, NEAT1, HMGA2, HIF1A	Promotes GSC stemness maintenance and inhibits neuronal differentiation;	Targeting CASCADES to disrupt GSC stemness for GBM treatment;	[265,266]
		miR-370-3p targets and regulates associated genes		Inhibits EMT and hypoxia signaling pathways	Restoring miR-370-3p expression to inhibit GBM progression
OCSCs	Ovarian Cancer	lncHOTAIR recruits EZH2 to catalyze H3K27 trimethylation;	lncHOTAIR; lncRNA LINC00665, CNBP	HOTAIR-EZH2-mediated chromatin modification and NF-κB pathway;	HOTAIR inhibitors combined with EZH2 is and platinum-based chemotherapy;	[6,267]
		LINC00665 targets CNBP mRNA for degradation		Wnt/Notch signaling pathways	Targeting LINC00665 to suppress ovarian cancer stemness

Stem cells have garnered increasing attention in recent years as a promising therapeutic strategy in cancer treatment. Owing to their self-renewal capacity and multipotent differentiation potential, stem cells have been employed not only in targeted and gene therapy for tumors but also show significant promise in cancer immunotherapy. Among them, MSCs and iPSCs have been extensively studied due to their tumor-tropic properties within the TME. These cells can serve as carriers for drugs and genes, enabling targeted delivery of anti-tumor agents, therapeutic genes, or immune cells directly to tumor sites, thereby enhancing therapeutic efficacy while minimizing damage to normal tissues. MSCs, for instance, can carry anticancer genes or immunomodulatory factors and migrate autonomously to tumor regions for precise therapeutic action [189].

Moreover, stem cells exhibit great potential in tumor immunotherapy by homing to tumor sites and activating immune cells such as T cells and macrophages, thus amplifying the anti-tumor immune response [190]. They may also serve as vehicles for tumor vaccines to induce tumor-specific immune clearance. Beyond these applications, stem cells have shown the ability to reverse tumor drug resistance by modulating the epigenetic landscape of tumor cells or secreting cytokines, thereby alleviating chemoresistance [191].

Despite the multiple advantages of stem cells in cancer therapy, several challenges remain in their clinical translation. Notably, iPSCs may carry a risk of tumorigenicity due to genomic instability, and the long-term efficacy and safety of stem cell-based therapies require further extensive validation. Nevertheless, advancements in gene-editing technologies (e.g., CRISPR/Cas9) and immunomodulatory strategies are expected to significantly enhance the safety and precision of stem cell therapies. These developments position stem cell-based interventions as promising candidates for future applications in personalized medicine and multimodal combination therapies [192].

Frontier research and future perspectives

Application of single-cell epigenetic analysis technologies in stem cell and tumor research

In recent years, breakthroughs in single-cell epigenomics have opened up new dimensions for dissecting the heterogeneity of the TME and the mechanisms of dynamic cellular interactions. By constructing frameworks for intercellular differences in DNA methylation landscapes, chromatin accessibility maps, and histone modification profiles, researchers have systematically unveiled the epigenetic regulatory networks governing heterogeneous self-renewing populations in GBM. Single-cell analysis of primary human GBM samples has revealed significant epigenetic diversity within patient-derived GSC populations. Invasive subpopulations acquire heightened therapeutic resistance through epigenetic reprogramming, and this phenotype is strongly associated with poor patient prognosis [193]. In AML, the integration of multi-omics at the single-cell level - such as scTEM-seq - has demonstrated that decitabine treatment induces global DNA methylation heterogeneity, which in turn directly mediates the remodeling of immune-related gene expression. A significant inverse correlation between the expression of transposable elements and DNA methylation levels further highlights the subclonal specificity of epigenetic regulation, offering novel epigenetic insights into the mechanisms underlying chemotherapy resistance [194].

The innovative integration of epigenetic clocks with lineage-tracing technologies has significantly enhanced our capacity for spatiotemporal dynamic analysis. The EpiTrace analytical framework, developed based on scATAC-seq data, reconstructs developmental trajectories across multi-lineage cell populations by quantifying the proportion of open chromatin sites. Its strong concordance with DNA methylation clocks confirms the central role of epigenetic regulation in determining cellular age. In GBM studies, this analytical system revealed that neural progenitor-like cells and outer radial glia-like cells share a common epigenetic regulatory compartment. Their dynamic chromatin accessibility patterns recapitulate gene regulatory features characteristic of early neurodevelopment, despite pronounced epigenetic heterogeneity among individual tumors [195,196]. Of particular interest, single-nucleus epigenetic profiling (snATAC-seq) of IDH1-mutant gliomas has, for the first time, identified non-coding RNA regions as major sources of epigenetic heterogeneity. Rare nuclear subpopulations exhibiting features of both IDHmut-codel and IDHmut-noncodel subtypes were discovered, and their distinct epigenetic signatures may serve as novel biomarkers for predicting treatment response [197].

Significant progress has been made in elucidating the mechanisms linking epigenetic dysregulation to tumor aggressiveness. In invasive GBM, regions of locally elevated DNA methylation abnormalities are strongly associated with transcriptomic disruption and aberrant activation of environmental stress response pathways. In vitro functional assays have confirmed that epigenetic instability accelerates disease progression by promoting cellular state plasticity. This phenomenon of genetic-epigenetic synergistic dysregulation has been validated in longitudinal clinical cohorts, providing critical theoretical support for targeting epigenetic vulnerabilities [198]. In GSC models, heterogeneous epigenomic features show strong correlations with aberrant activation of DNA damage repair pathways and cell proliferation regulatory networks. Dysregulated expression of TET enzymes leads to failure in enhancer landscape remodeling, directly accounting for the defective epigenetic reprogramming observed in GSCs compared to normal NSCs. Furthermore, imbalanced dynamic regulation of chromatin accessibility persists throughout the differentiation process [199]. Looking forward, the integration of spatial omics with single-cell epigenetic analysis is expected to enable in situ dissection of epigenetic regulatory mechanisms within stem cell-tumor cell interaction networks. This advancement holds promise for developing novel, precision-targeted therapeutic strategies.

Application of AI and big data in stem cell and tumor epigenetics research

The deep integration of AI and big data is reshaping the research paradigm of stem cell and tumor epigenetics, enabling full-spectrum breakthroughs from molecular mechanism exploration to clinical translation through multidimensional technological integration. In the field of single-cell multi-omics analysis, a global-local dual-layer interpretability framework based on the random forest algorithm has successfully decoded pan-cancer DNA methylation landscapes, accurately pinpointing the hypomethylated promoter region of the CHD8 gene characteristic of GBM [200]. The EpiTrace method integrates temporal dynamics of chromatin accessibility with convolutional neural network models, enabling - for the first time - cross-species developmental trajectory analysis. This approach has validated the functional conservation of H3K4me3 histone modifications during evolution [195]. To address the challenge of data heterogeneity in tumor epigenomics, the crossNN deep learning framework employs transfer learning strategies to harmonize datasets across different sequencing platforms. This model has achieved breakthrough diagnostic performance in the molecular subtyping of lung adenocarcinoma, while simultaneously identifying IDH1 mutation-associated methylation anomalies [201]. The scMeFormer model, which innovatively incorporates a Transformer architecture, effectively alleviates the sparsity of single-cell sequencing data and has uncovered heterogeneous methylation patterns at the SOX2 enhancer region in glioma stem cell research [202].

At the level of multimodal data integration, the MGRL framework has pioneered the fusion of DNA methylation profiles, chromatin accessibility maps, and protein-protein interaction network data. Using graph attention network modeling, it quantitatively elucidated the regulatory relationship between methylation levels at ribosomal protein gene promoters and stem cell proliferation capacity [203]. These technological breakthroughs have already demonstrated significant clinical translational value. For example, the stemness index (mDNAsi), constructed using the OCLR algorithm, successfully enabled prognostic stratification of GBM patients and systematically elucidated the synergistic mechanisms of epigenetic regulatory drugs [204]. Additionally, a prognostic model based on random forest integration of single-cell multi-omics data has provided a clinical decision-support system for combination cancer therapies, incorporating novel biomarkers such as POU5F1B [205].

Looking ahead, the deep integration of federated learning frameworks with multimodal knowledge graphs is expected to effectively overcome the challenge of data silos. Recent studies have demonstrated that combining dynamic features of H3K4me3 modifications with AI-driven causal inference models enables the systematic construction of epigenetic regulatory networks underlying CSC stemness maintenance [206]. This technological convergence not only promises to accelerate the clinical validation of epigenetic biomarkers but also holds great potential for advancing epigenetic drug discovery through virtual screening platforms. Ultimately, it paves the way for a new era of precision medicine and opens intelligent and innovative pathways for both stem cell therapy and tumor immunotherapy.

Future directions and challenges in stem cell and tumor epigenetics research

Combined analyses of single-cell and spatial transcriptomics have revealed that SCs drive tumor initiation and maintain stemness through interaction networks with inflammation-associated cancer-associated fibroblasts (iCAFs), highlighting the pivotal role of this crosstalk in tumor progression [207]. Multi-omics integration has further clarified the pathways linking genomic variation to phenotypic regulation: integrating genomic and transcriptomic data enables causal inference between genotype and tumor-associated phenotypes, while joint analysis of epigenomic and transcriptomic data systematically reveals regulatory networks by which epigenetic modifications modulate target gene expression [208,209]. Technological innovations are propelling the field toward a deeper integration of dynamic regulatory mechanism analysis and clinical translation. Single-cell multi-omics approaches - by integrating DNA, RNA, and proteomics data - have significantly improved the resolution of tumor heterogeneity. However, the bisulfite treatment commonly used for methylome sequencing still induces DNA damage, limiting accuracy [209]. Recently developed enzyme-based base conversion combined with nanopore sequencing (ONT) now enables long-read, phased epigenetic analysis at single-molecule resolution, offering a new paradigm for studying epigenetic heterogeneity [210]. A representative application is the nanoNOMe platform, which combines GpC methyltransferase labeling with nanopore sequencing to simultaneously detect genomic DNA methylation and chromatin accessibility - substantially improving data quality while reducing DNA damage [211]. This technology has been successfully applied in human cancer cell line studies, where integration with scRNA-seq and scATAC-seq revealed that dynamic changes in chromatin accessibility and extrachromosomal DNA (ecDNA) jointly drive intracellular heterogeneity [212].

However, current technological approaches still face limitations in accurately analyzing protein phosphorylation modifications and controlling biases introduced by cell fixation. Future research must focus on three major breakthrough directions: ① Optimizing experimental protocols or developing novel multi-omics combinations (such as integrated genome-proteome analyses) to improve detection sensitivity; ② Combining CRISPR-dCas9-based epigenetic editing tools with organoid models to enable precise validation of therapeutic targets in tumor stem cells; ③ Constructing multidimensional data integration models (e.g., the TMODINET framework) to address the challenge of therapy resistance driven by epigenetic heterogeneity [213]. Although real-time in vivo monitoring of epigenetic dynamics presents ethical challenges, interdisciplinary collaboration and long-term clinical follow-up will be essential for elucidating the mechanisms of stem cell-tumor epigenetic interactions. Current technological advances mark a transition in epigenetic research - from static correlation analysis toward the dissection of dynamic regulatory networks - ultimately paving the way for clinical translation based on epigenetic reprogramming.

Conclusion

Stem cells, with their unique self-renewal and multipotent differentiation capabilities, hold broad application prospects in regenerative medicine and tissue repair. CSCs maintain heterogeneity and drive tumor progression and drug resistance through epigenetic mechanisms. DNA methylation, histone modifications, and non-coding RNA regulation form the core link between stem cells and tumorigenesis. Embryonic and adult stem cells regulate fate decisions via dynamic epigenetic networks, while iPSCs achieve epigenetic reprogramming. Epigenetic abnormalities in tumors promote genomic instability and treatment resistance, and the TME exacerbates tumor heterogeneity and drug resistance through multidimensional regulation. Epigenetic memory and non-coding RNA networks confer dormancy and resistance traits to CSCs, providing a molecular basis for recurrence and metastasis. DNA methylation inhibitors and HDACi demonstrate clinical potential, and precise regulation combined with stem cell carriers and exosomes expands novel anti-tumor therapeutic avenues. Single-cell omics, spatial omics, and AI technologies drive in-depth analysis of regulatory networks. Future integration of CRISPR editing and multimodal data promises to overcome therapeutic bottlenecks, advancing regenerative medicine and precision oncology toward a new level of personalized medicine.

Disclosure of conflict of interest

None.