Epigenetic Signatures of Dental Stem Cells: Insights into DNA Methylation and Noncoding RNAs

Abstract

Tooth development (odontogenesis) is regulated by interactions between epithelial and mesenchymal tissues through signaling pathways such as Bone Morphogenetic Protein (BMP), Wingless-related integration site (Wnt), Sonic Hedgehog (SHH), and Fibroblast Growth Factor (FGF). Mesenchymal stem cells (MSCs) derived from dental tissues—including dental pulp stem cells (DPSCs), periodontal ligament stem cells (PDLSCs), and dental follicle progenitor cells (DFPCs)—show promise for regenerative dentistry due to their multilineage differentiation potential. Epigenetic regulation, particularly DNA methylation, is hypothesized to underpin their distinct regenerative capacities. This study reanalyzed publicly available DNA methylation data generated with Illumina Infinium HumanMethylation450 BeadChip arrays (450K arrays) from DPSCs, PDLSCs, and DFPCs. High-confidence CpG sites were selected based on detection p-values, probe variance, and genomic annotation. Principal Component Analysis (PCA) and hierarchical clustering identified distinct methylation profiles. Functional enrichment analyses highlighted biological processes and pathways associated with specific methylation clusters. Noncoding RNA analysis was integrated to construct regulatory networks linking DNA methylation patterns with key developmental genes. Distinct epigenetic signatures were identified for DPSCs, PDLSCs, and DFPCs, characterized by differential methylation across specific genomic contexts. Functional enrichment revealed pathways involved in odontogenesis, osteogenesis, and neurodevelopment. Network analysis identified central regulatory nodes—including genes, such as PAX6, FOXC2, NR2F2, SALL1, BMP7, and JAG1—highlighting their roles in tooth development. Several noncoding RNAs were also identified, sharing promoter methylation patterns with developmental genes and being implicated in regulatory networks associated with stem cell differentiation and tissue-specific function. Altogether, DNA methylation profiling revealed that distinct epigenetic landscapes underlie the developmental identity and differentiation potential of dental-derived mesenchymal stem cells. This integrative analysis highlights the relevance of noncoding RNAs and regulatory networks, suggesting novel biomarkers and potential therapeutic targets in regenerative dentistry and orthodontics.

1. Introduction

Tooth development, or odontogenesis, is a highly regulated biological process driven by reciprocal interactions between epithelial and mesenchymal tissues. These interactions are mediated through molecular signaling pathways, including BMP, Wnt, SHH, and FGF, which control cell proliferation, differentiation, and morphogenesis [1]. A detailed understanding of these pathways is essential not only for elucidating normal craniofacial development but also for advancing regenerative strategies in dentistry.

In recent years, mesenchymal stem/stromal cells (MSCs) derived from dental tissues have emerged as valuable tools in tissue engineering. These cells, most notably dental pulp stem cells (DPSCs), periodontal ligament stem cells (PDLSCs), and dental follicle progenitor cells (DFPCs), exhibit the hallmarks of MSCs, including clonogenicity, multilineage differentiation, and immunomodulatory properties. As postnatal derivatives of neural crest cells, these stem cell populations are uniquely adapted to regenerate mineralized and soft tissues in the oral cavity [2,3].

Each of these odontogenic stem cell types displays distinct biological characteristics reflective of their developmental origin and niche [4,5]. DPSCs are particularly effective in dentin–pulp regeneration; PDLSCs demonstrate robust osteogenic and tenogenic differentiation; and DFPCs serve as progenitors for multiple periodontal structures. The accessibility and regenerative potential of these cell lines make them attractive candidates for clinical translation [6]. However, their differential performance in regenerative assays suggests that deeper molecular distinctions underlie their phenotypic variability. Among the mechanisms contributing to these functional differences, epigenetic regulation, particularly DNA methylation, plays a central role [7]. DNA methylation serves as a heritable but reversible regulatory mark that influences gene expression without altering the genetic code. In stem cells, DNA methylation is critical for maintaining multipotency and directing lineage-specific differentiation. CpG methylation in promoter regions, gene bodies, or enhancers can either silence or promote transcription depending on the genomic context. The landmark study by Ai et al. [8] provided compelling evidence that differential DNA methylation profiles among DPSCs, PDLSCs, and DFPCs are associated with their osteogenic potential. While the global methylation landscapes of the three cell types were largely similar, key differences emerged in genes related to bone formation, extracellular matrix organization, and signaling regulation. Notably, PDLSCs showed hypomethylation of osteogenic transcription factors, correlating with their enhanced in vitro and in vivo bone-forming ability.

Epigenomic profiling technologies, such as the array-based technologies, have enabled high-resolution interrogation of methylation states across the genome [9]. These platforms have facilitated the identification of differentially methylated CpG sites and their association with gene regulatory elements. When integrated with gene ontology and pathway analyses, methylation data can reveal functionally relevant epigenetic signatures that govern stem cell identity and behavior [10,11]. In this study, publicly available methylation data (GSE112933) [8] from DPSCs, PDLSCs, and DFPCs were reanalyzed. The objective is to identify both shared and lineage-specific epigenetic profiles using an updated bioinformatic pipeline. Special emphasis is placed on functionally annotated CpG regions, including those located in promoters, enhancers, and noncoding RNA loci. Through integrative analysis of methylation variability, clustering, and gene set enrichment, the aim is to elucidate the molecular logic that underpins odontogenic differentiation and provides insight into the epigenetic basis of dental stem cell diversity.

2. Materials and Methods

2.1. Data Collection

Illumina Infinium HumanMethylation 450K BeadChips (Illumina Inc., San Diego, CA, USA) data from human dental pulp stem cells (DPSCs), periodontal ligament stem cells (PDLSCs), and dental follicle progenitor cells (DFPCs) samples were obtained from Gene Expression Omnibus (GEO) under accession number GSE112933. Available average beta values, which are the ratio of the intensities of methylated and unmethylated alleles, were considered for the following analyses.

2.2. Preprocessing and Filtering

Detection p-values were used to retain only high-confidence probes (p < 0.01 in all samples). Probes mapping to sex chromosomes were excluded, and further filtering removed probes affected by known SNPs or with cross-reactivity, based on publicly available manifest files (Illumina HumanMethylation 450K, https://zwdzwd.github.io/InfiniumAnnotation [12], accessed on 10 April 2025). Probes with missing values, extreme beta values (outside the 0.05–0.95 range), or non-CpG annotations were removed. Only CpG probes (starting with the ‘cg#’ prefix) were retained.

2.3. Principal Component Analysis (PCA)

A PCA was conducted on the transposed β-value matrix using prcomp in R (ver. 4.4.2) to assess overall variation and clustering among the samples. The first two principal components were visualized to evaluate sample separation.

2.4. Probe Variance and Clustering

Probe-wise variance was computed across samples, and the elbow method was used to define the most variable subset of probes. This subset was subjected to hierarchical clustering using the R function pheatmap (ver. 1.0.12). Then, the silhouette function of the cluster (ver. 2.1.8.1) R package was used to determine the optimal number of probe clusters.

2.5. Cluster Annotation and Functional Enrichment

Probes within each cluster were annotated using the Illumina HumanMethylation450K manifest. Genes associated with each probe cluster were extracted based on RefGene annotations. Gene Ontology (GO) and KEGG pathway enrichment analyses were conducted using the clusterProfiler (ver. 4.16.0) R package. Enrichment results were filtered at an adjusted p-value threshold of 0.05 for each cluster.

2.6. Hyper- and Hypomethylated CpG Site Selection

The probes that indicated highly methylated or hypomethylated CpG sites consistently across all three samples (β ≥ 0.8 or ≤0.2, respectively) [13,14] were selected. The number of sites were further reduced, selecting only those whose minimum and maximum methylation levels differed by less than 0.005 across all three samples.

2.7. Genomic Context Profiling

The distribution of retained probes was analyzed across CpG contexts (Island, Shore, Shelf, Open Sea) and genic regions (TSS1500, TSS200, 5′UTR, 1st Exon, Gene Body, 3′UTR) using annotation data from the Illumina HumanMethylation450K manifest. Mean methylation levels were computed for all probes that differentiated the three samples and those that were consistently hyper- or hypomethylated in the samples.

2.8. Noncoding RNAs Analysis

As noncoding RNAs loci have been identified as frequent methylation targets, this study focused on noncoding RNAs, including small noncoding, long noncoding (lncRNA), and long noncoding RNAs for which an enhancer region around the transcription start site (TSS) was identified (eRNA). A dedicated functional annotation based on the Encyclopaedia of DNA Elements (ENCODE) and LNCipedia databases was used to re-annotate 450K data [15].

Hence, among the consistently hyper- and hypomethylated probes in the three cell lines, promoter-associated CpG sites were selected, and beta values from the three samples were extracted, along with transcript annotations. The HGNC symbols of genes presenting a TSS within the same ENCODE promoter regions were separated, cleaned, and deduplicated based on identical methylation profiles. Protein-coding genes were used for GO enrichment analysis with the clusterProfiler package. Biological processes were selected if Benjamini–Hochberg-corrected p-values were less than 0.05. Noncoding RNAs sharing the promoter sites of protein-coding genes that enriched selected GO terms were used to draw a network. The network linked GO terms, protein-coding genes, and their corresponding noncoding RNAs. Network connectivity and centrality analyses were performed using Pyntacle (ver. 1.3.1) [16]. In particular, the betweenness centrality metric was computed on nodes to assess their relative importance, in terms of the number of paths passing through a node over the total number of available paths in the network [17,18,19]. Sparseness of the network was computed using the completeness index [20,21]. Visualization was performed using igraph (ver. 2.1.4), with nodes colored by type.

3. Results

3.1. High-Quality Methylation Probe Selection

A comprehensive quality control pipeline was applied to retain only high-confidence CpG probes from the starting 485,577 probes, based on detection p-values (485,482), exclusion of sex chromosome probes (473,835), removal of probes affected by SNPs or cross-reactivity (412,997), filtering of probes with missing data (412,997), and extreme methylation values (246,824). The analysis of these values through principal component analysis showed that two components can capture 100% of the variance in methylation profiles across the three samples. The separation observed in the PCA space (Figure S1) suggests that all three samples exhibit distinct methylation profiles.

A balanced subset of probes that captures most of the biologically relevant variability, while reducing dimensionality and avoiding saturation by uninformative probes, was obtained using the elbow method. A total of 17,175 probes, ~4% of the full 450K array, were found to be the most informative and variable probes across the three samples (Figure S2). Probes beyond this point show, in fact, diminishing variance, likely contributing noise or background methylation.

3.2. Heatmap of Methylation Beta Values and Clustering

Hierarchical clustering revealed that DNA methylation profiling across dental pulp stem cells (DPSCs), dental follicle progenitor cells (DFPCs), and periodontal ligament stem cells (PDLSCs) is organized into six major clusters, encompassing the elbow-defined 17,175 probes as the most informative and variable across the three samples (Figure 1).

The methylation profiles across the six clusters reveal distinct epigenetic signatures for each of the three dental stem cell types. DPSCs are characterized by hypermethylation in Clusters 3 and 4, suggesting potential transcriptional repression of the loci that may be inactive in pulp-derived cells. Conversely, they are hypomethylated in Clusters 1 and 5, potentially indicating regions with greater epigenetic accessibility. Clusters 2 and 6 show intermediate methylation, suggesting a more balanced regulatory state.

DFPCs display a signature of hypermethylation in Clusters 5 and 6, suggesting potential silencing events, likely relevant to follicular identity. They are hypomethylated in Cluster 2 and slightly in Cluster 4, pointing to potential activation of DFPC-specific loci. Clusters 1 and 3 show intermediate methylation, reflecting a neutral or transitioning epigenetic state. PDLSCs exhibit hypermethylation in Clusters 1 and 2, implying stable repression of certain loci, while Clusters 3 and 6 are hypomethylated, suggesting transcriptionally active or poised regions. Clusters 4 and 5 fall into intermediate methylation, consistent with partial activity or plasticity in periodontal cells.

3.3. Gene Ontology and Pathway Analyses

Gene Ontology (GO) and KEGG pathway analyses enabled the identification of different biological processes associated with the different clusters, highlighting functional differences among the 3 cell lines, reflecting their embryological origins and differentiation potentials.

Cluster 1 (Table S1) displays hypomethylation in DPSCs, intermediate methylation in DFPCs, and hypermethylation in PDLSCs. This cluster contains genes involved in neurogenesis and neural development, multiorgan embryonic morphogenesis, cell adhesion and cytoskeleton organization, cell signaling (Wnt, PI3K-Akt, and TGF-β), mesenchymal, osteogenic, and chondrogenic differentiation. Key pathways include cell signaling pathways (as MAPK, Wnt, and TGF-β signaling pathways) that regulate critical cellular functions such as growth, differentiation, survival, and migration, cancer-related pathways, as well as cytoskeletal and adhesion mechanisms, regulating cellular shape, motility, and interactions with the extracellular matrix, all crucial for morphogenesis and differentiation, and already known or suggested to play a role in odontogenesis.

Cluster 2 (Table S2) shows intermediate methylation in DPSCs, hypomethylation in DFPCs, and hypermethylation in PDLSCs. The most representative biological processes highlight a strong convergence around embryonic morphogenesis, including Wnt signaling regulation, renal and neural development, and mesenchymal differentiation. Of note, several processes directly pertain to odontogenesis. The pathway analysis disclosed the occurrence of developmental and morphogenetic pathways (e.g., Wnt, Hippo, Adherens junctions), regulating tissue patterning, growth control, and morphogenetic processes in multicellular organisms; signal transduction and hormonal pathways (e.g., cAMP, hormonal pathways), which mediate hormone-driven cellular responses and metabolic regulation; and cancer-associated pathways, which reflect the dysregulation of core signaling networks such as Wnt, MAPK, PI3K-Akt, and TGF-β, common in both oncogenesis and development.

Cluster 3 (Table S3) is characterized by hypermethylation in DPSCs, intermediate methylation in DFPCs, and hypomethylation in PDLSCs, disclosing an opposite pattern compared to Cluster 1. The main biological processes include cytoskeletal dynamics and cell morphogenesis, developmental processes and organogenesis, signal transduction and cellular communication, as well as tissue repair. Pathways span crucial domains of development, metabolism, immunity, and disease. They include signal transduction pathways (e.g., PI3K-Akt, MAPK, mTOR, Wnt, Rap1, Ras, Hippo), which regulate cell proliferation, survival, metabolism, stress response, and plasticity; pathways involved in cell architecture and adhesion, which regulate structural organization and interactions with the cell environment; morphogenesis, hormonal, and endocrine pathways, which are involved in hormonal control and systemic homeostasis; and pathways regulating the pluripotency of stem cells.

Cluster 4 (Table S4) shows hypermethylation in DPSCs, slight hypomethylation in DFPCs, and intermediate levels in PDLSCs. The biological processes span several functional categories, including embryonic development and morphogenesis, Wnt and BMP signaling regulation, cell migration and adhesion, neurodevelopmental processes, and growth and differentiation. Pathway analysis disclosed a single enriched pathway, the PI3K-Akt signaling pathway, which may reflect a highly specific biological role.

Cluster 5 (Table S5) presents hypomethylation in DPSCs, hypermethylation in DFPCs, and intermediate methylation in PDLSCs. The processes span essential aspects of developmental biology, including BMP and TGF-β signaling and responses; cardiac and muscle development; cell morphogenesis, fusion, and adhesion; neurodevelopment and synaptic regulation; renal and limb development; and ion transport and electrophysiological regulation. The pathways involved collectively regulate structural organization, signaling, and mineral regulation. They include cytoskeletal and adhesion-related pathways that regulate the architecture, adhesion, motility, and mechanical properties of cells; cardiac-specific pathways are also enriched and involve genes and mechanisms such as ion channels, which are also active in developmental contexts, including muscular, epithelial, and connective tissues. In addition, developmental signaling pathways (e.g., the Notch signaling pathway), which are central to embryonic patterning and differentiation, are also represented. Further pathways include ion transport and hormonal regulation, which govern electrolyte balance and mineral homeostasis, processes that are essential in many tissues, especially mineralized ones, such as bone and teeth.

Cluster 6 (Table S6) features intermediate methylation in DPSCs, hypermethylation in DFPCs, and hypomethylation in PDLSCs. The biological processes describe a coordinated network of cytoskeletal, signaling, transport, and differentiation mechanisms that collectively guide tissue formation, including cytoskeleton dynamics and cell shape regulation; cell adhesion and migration; calcium and ion transport; signal transduction and hormonal responses; bone and tooth mineralization; and nervous system development and projection. The pathways define a complex interplay between intracellular signaling, hormone secretion, cytoskeletal organization, and neural development. They include signal transduction and hormonal pathways, which control intracellular communication, second messenger cascades, and endocrine responses; neurotransmission and synaptic function that influence neuronal function, circadian rhythm regulation, and sensory processing, as well as developmental pathways beyond the nervous system via guidance cues and neuropeptide activity; endocrine regulation and hormone secretion, which influence systemic homeostasis, including mineral balance, metabolism, and tissue remodeling; cytoskeleton and cell adhesion, which regulate cellular architecture, migration, and mechanical signaling, processes that are fundamental to tissue morphogenesis and structural integrity.

3.4. Genomic Probe Profiling Analysis

The distribution of probes was analyzed according to two genomic classifications. Distribution was analyzed relative to the CpG island and across genomic regions based on gene region features from UCSC, including distance from the transcription start site, 5′UTR, gene body, and 3′UTR.

The full set of probes included in the Illumina 450K array was analyzed, and the probes’ genomic distribution was compared to that of the filtered subset used in downstream analyses (Figure 2). In addition, the genomic profiles of probes within each of the six methylation clusters were examined, as well as those classified as consistently hypermethylated (β ≥ 0.8) or hypomethylated (β ≤ 0.2) across all samples (Figure 2).

The analysis of CpG site distribution across genomic regions for the six probe clusters revealed distinct profiles. Cluster 2 shows a strong enrichment in CpG islands, suggesting a regulatory role at gene promoters, which is associated with canonical transcriptional regulation, while others (e.g., Cluster 4) are linked to more dynamic regions (shores). In contrast, the shelf regions are consistently underrepresented across all clusters. The observed differences suggest that epigenetic variation between the cell lines is not randomly distributed, but is localized to specific regulatory regions.

The distribution of CpG sites across genomic regions for the six probe clusters was also evaluated. A clear peak is observed in the gene body region, which accounts for over 60% of the CpG sites in each cluster (Figure 3). The TSS1500 and 5′UTR regions display lower, but still detectable representation. In contrast, 1stExon, TSS 1500, and 3′UTR regions show consistently low percentages across all clusters. Minor differences among clusters were observed, suggesting a shared pattern of enrichment in gene body regions, with subtle variations in other regions.

Then, the probes that indicated highly methylated or hypomethylated CpG sites consistently across all three dental-derived cell lines were selected (β ≥ 0.8 or ≤0.2, respectively). The distribution across genomic regions revealed that hypomethylated probes were highly concentrated in CpG islands (~70%) and shore regions, while hypermethylated probes were more concentrated in shelf regions (Figure 4).

The distribution of hypermethylated and hypomethylated CpG probes across genomic regions revealed the enrichment of hypermethylated probes in the gene body, accounting and 3′UTR (Figure 5). In contrast, hypomethylated probes are more evenly distributed across the first exon, TSS200, TSS1500, and 5′UTR.

3.5. Noncoding RNAs Annotation and Network Analysis

Among the consistently hyper- or hypomethylated probes across the three dental-derived stem cell lines, a significant number of elements associated with noncoding transcripts was identified. These probes were annotated across different regulatory contexts based on ENCODE and LNCipedia data, including promoter, enhancer, and gene body regions. Among the noncoding elements identified, some noncoding RNAs (DANCR, MIR22HG, ZFAS1, and MALAT1) emerged as particularly relevant due to their known roles in tooth development and alveolar bone remodeling [22,23,24,25,26,27,28,29].

Next, this study focused on noncoding RNA loci that shared promoter-associated CpG sites—either consistently hypermethylated or hypomethylated across the three cell lines—with nearby protein-coding genes. GO enrichment analysis of these protein-coding genes identified 19 statistically significant biological process (BP) terms (Table S7). From these, three BPs (GO:0048568, embryonic organ development, fold-enrichment = 1.63, p-adj = 5.3 × 10⁻⁵; GO:0045165, cell fate commitment, fold-enrichment = 1.75, p-adj = 0.046; GO:0048863, stem cell differentiation, fold-enrichment = 1.81, p-adj = 0.047) were used to build a gene-concept network (Figure 6), as they capture key biological programs shared by dental stem cells, including their origin in embryonic craniofacial development, their lineage-specific fate commitment, and their capacity for regulated differentiation into odontogenic and periodontal cell types. The network contained 205 nodes and 235 edges and exhibited a compactness index of 0.033, which falls within the range classified as a sparse network.

Interestingly, several of these genes include elements in key developmental pathways regulating tooth formation, patterning, and mineralization (e.g., PAX6, TFAP2A, GATA2, FOXC2, SALL1, NR2F2, LHX1, HOXB2, JUNB, OSR2, BMP2, DLX5, GLI3, WNT2, BMP7, PITX2, JAG1, GATA4). In particular, PAX6, FOXC2, NR2F2, SALL1, BMP7, and JAG1 were among the top 10 most central nodes based on the betweenness centrality measure, indicating their involvement in multiple biological functions (Table S8).

4. Discussion

In this study, a comprehensive epigenomic analysis of three dental stem cell populations—dental pulp stem cells (DPSCs), dental follicle progenitor cells (DFPCs), and periodontal ligament stem cells (PDLSCs)—was conducted, by interrogating publicly available DNA methylation data profiles [8]. Unsupervised hierarchical clustering was based on 17,175 highly variable probes and disclosed the occurrence of six distinct clusters that clearly differentiate the three cell types and identify specific methylation profiles that reflect their developmental origin and functional role.

Each cluster was characterized by a specific methylation signature across the three samples. For example, DPSCs displayed hypermethylation in specific clusters (Clusters 3 and 4) and hypomethylation in others (Clusters 1 and 5), suggesting repression of some developmental pathways while maintaining accessibility in regions potentially relevant to neural or mesenchymal differentiation. DFPCs displayed hypermethylation in two clusters (Clusters 5 and 6), while hypomethylation in two other different clusters (Cluster 2 and slightly in Cluster 4), indicating distinct regulatory constraints that are potentially linked to their embryonic follicular identity. Conversely, PDLSCs showed hypermethylation in the first two clusters (Clusters 1 and 2) and hypomethylation in two other clusters (Clusters 3 and 6), which aligns with their periodontal niche and their potential role in structural remodeling and mechanical response.

The analysis of the distribution of methylation probes across genomic regions revealed consistent and functionally relevant patterns. Cluster 2, for instance, was highly enriched in CpG islands, typically located in gene promoters, suggesting a role in canonical transcriptional regulation. Other clusters were preferentially distributed in CpG shores, regions known to undergo dynamic methylation during development [30]. These patterns suggest that epigenetic regulation in dental stem cells is region-specific and modulates key loci involved in differentiation and tissue-specific functions. Furthermore, the gene region analysis disclosed a predominant occurrence of methylated CpGs in gene bodies in all clusters, with lower representation in promoters. This finding is consistent with prior evidence showing that gene body methylation may support transcriptional elongation and tissue-specific expression patterns through alternative splicing [31]. This suggests a potential role for epigenetic regulation in shaping isoform diversity during odontogenesis.

Gene Ontology and KEGG pathway enrichment analyses provided functional insights into the biological roles of each cluster. The clusters hypomethylated in DPSCs (e.g., Cluster 1) were enriched in neurogenesis, axon guidance, and Wnt signaling, which reflects the neurogenic potential of pulp-derived cells [32,33]. The clusters hypomethylated in PDLSCs (e.g., Cluster 3) instead showed enrichment in processes related to ECM remodeling, actin cytoskeleton, and MAPK signaling, all pathways being relevant to mechanotransduction and periodontal ligament function [34,35]. DFPCs’ hypomethylated Cluster 2 was associated with embryonic organ development, canonical Wnt signaling, and odontogenic pathways, consistent with their involvement in tooth root formation and follicular development [1,36]. Clusters with high hypermethylation in DPSCs, such as Cluster 4, showed suppression of PI3K-Akt and BMP signaling, suggesting the downregulation of pathways that could otherwise favor alternate differentiation trajectories.

This functional divergence between hypomethylated (poised or active loci) and hypermethylated (silenced) regions supports models in which epigenetic plasticity underlies stem cell multipotency and lineage commitment. These findings reinforce the notion that DNA methylation profiles and the related biological pathways play a significant role in defining the identity and functional activity of dental stem cells and the role of epigenetic regulation in cell fate determination.

This may provide significant insights into odontogenesis, as well as into orthodontic context.

For example, pathways involved in ECM–receptor interaction and actin cytoskeleton regulation—enriched in several clusters—are directly implicated in mechanotransduction, a process highly relevant in orthodontics. In this context, the response of periodontal ligament and surrounding tissues to mechanical force is fundamental for tooth movement. Epigenetic changes affecting focal adhesion and cytoskeletal genes may influence this response, potentially offering biomarkers for individual variability in treatment outcomes.

Further, methylation-mediated modulation of TGF-β and Wnt signaling—key pathways in craniofacial development and remodeling—suggests that these profiles could be leveraged to understand susceptibility to root resorption, a common side effect in orthodontic therapy. Moreover, Wnt and BMP pathways are activated during tooth movement [1,37]; therefore, the methylation profile of these pathways might help predict cellular responsiveness.

This study also focused on shared epigenetic features among all three cell types, evaluating probes that were consistently hypermethylated (β ≥ 0.8) or hypomethylated (β ≤ 0.2) across all samples. The analysis of the distribution of methylation probes across genomic regions revealed that hypomethylated probes were significantly enriched in CpG islands and shores, supporting the idea that active promoters are generally unmethylated to permit transcription initiation [38,39]. In contrast, hypermethylated probes were enriched in shelf regions, typically linked to repressive chromatin states. This distribution supports the idea that DNA methylation contributes to defining active versus silent regulatory domains, shaping cell-specific gene expression programs in DPSCs, DFPCs, and PDLSCs and are consistent with previous work showing the epigenetic plasticity of dental mesenchymal stem cells and their odontogenic potential [40,41].

The identification of consistently hyper- or hypomethylated CpG sites across all three stem cell types provides a useful resource for selecting stable epigenetic markers or potential targets for regenerative strategies, particularly in pulp–dentin complex regeneration and periodontal tissue engineering.

Although the profiles were derived from early-passage cells, existing evidence suggests that DNA methylation in mesenchymal stem cells is highly dynamic and responsive to differentiation cues. Studies have shown that osteogenic or odontogenic induction protocols can lead to demethylation or hypermethylation of lineage-specific regulators enhancing their expression during differentiation. Therefore, the patterns observed may represent baseline epigenetic states that predispose each cell type to specific differentiation trajectories.

The analysis was also extended to a class of genomic elements that are usually not fully considered in methylation analysis due to annotation limits, i.e., noncoding RNAs. Using an integrated functional annotation [15] based on dedicated databases (i.e., LNCipedia and ENCODE), the occurrence of several probes in noncoding related elements was revealed. Notably, among those, some noncoding RNAs (DANCR, MIR22HG, ZFAS1, and MALAT1) that are consistently hyper- or hypomethylated across all three dental-derived stem cell lines in their promoter regions and have been previously implicated in tooth development, periodontal regeneration, and alveolar bone remodeling were identified [22]. DANCR inhibits osteogenic differentiation and it enhances TNF-α/IL-6–mediated osteoclastogenesis in PDLSCs [23,24]. MIR22HG, a long noncoding RNA that also hosts the microRNA hsa-mir-22, was upregulated during the osteogenic differentiation of human PDLSCs. This upregulation closely mirrored the expression of osteogenic genes, suggesting their potential role in promoting osteogenesis [25]. Its role in alveolar bone regeneration has also been reported in in vivo studies [26]. ZFAS1 promotes osteogenesis by regulating the miR-499/EPHA5 axis and participates in the epigenetic control of osteogenic pathways [27]. MALAT1 is involved in both osteoblast and osteoclast regulation, modulates osteogenic differentiation through the miR-204/Smad4 axis [28], and regulates osteoclast activity by modulating the RANKL/OPG ratio and NF-κB signaling [29], which are mechanisms critical to alveolar bone turnover.

The consistent methylation patterns observed at these loci suggest a possible shared regulatory mechanism across dental stem cell types. This is particularly relevant in the context of dental cell identity and differentiation, where epigenetic regulation of noncoding elements is increasingly recognized as a critical layer of control [42]. These findings support emerging evidence that lncRNAs are key epigenetic and transcriptional regulators in odontogenesis [22,43,44] and orthodontic bone remodeling.

Furthermore, network analysis based on coding genes with shared promoters with noncoding elements revealed the occurrence of genes previously associated with odontogenesis and dental anomalies. These genes contribute to processes such as epithelial signaling (WNT2, PITX2) [45,46], tooth morphogenesis (GLI3) [47], tissue mineralization (BMP2) [48], BMP7 [49], and lamina positioning (PITX2) [50]. Others, such as GATA2 [51] and OSR2 [52], have been found to play a role in both normal odontogenesis and dental malformations.

This result reinforces the biological relevance of the methylation-defined pathways and highlights the potential of this network-based approach to uncover novel regulatory elements involved in tooth development, including those among noncoding elements. Moreover, the presence of known odontogenic regulators within the network supports its use as a discovery tool to prioritize additional candidates, both coding and noncoding, for functional validation in dental biology, regenerative medicine applications, and orthodontics applications.

Several previous studies focused on methylation changes during differentiation in cell types—such as DPSCs or PDLSCs—[40,41,53]. Compared to previous studies on DNA methylation in odontogenic stem cells, this work adopts a broad and integrative perspective. The study on which the present analyses are based [8] provided initial evidence that differential methylation profiles among dental pulp stem cells (DPSCs), periodontal ligament stem cells (PDLSCs), and dental follicle progenitor cells (DFPCs) are associated with their distinct osteogenic capacities, with PDLSCs exhibiting superior bone-forming potential. Their analysis focused primarily on identifying differentially methylated CpG sites linked to osteogenesis, followed by functional validation in vitro and in vivo, without exploring the broader regulatory epigenomic context or the contribution of noncoding RNAs [8]. In this study, a publicly available dataset (GSE112933), was reanalyzed, using an updated, stringent bioinformatic pipeline. A high-confidence probe filtering, CpG-context classification, and variance-based clustering were applied to uncover six distinct methylation clusters, each linked to specific biological processes via GO and KEGG enrichment. Crucially, a novel layer of analysis was incorporated, focusing on noncoding RNAs associated with consistently methylated promoter regions. By constructing a gene-concept regulatory network, central transcriptional regulators that are implicated in tooth development and dental anomalies were identified.

5. Limitations and Conclusions

This study presents a few limitations inherent to its in silico and retrospective design. The analyses were conducted on publicly available DNA methylation datasets and do not include functional validation or longitudinal assessment of methylation changes during differentiation. Consequently, the associations between DNA methylation, gene expression, and stem cell fate remain correlative. While the integrative network and enrichment analyses revealed biologically plausible candidates, the regulatory roles of many identified long noncoding RNAs (lncRNAs) remain hypothetical and require experimental confirmation. The constructed gene-lncRNA network was sparse, reflecting the conservative nature of the filtering pipeline, which prioritized robustness and reproducibility. However, sparse topologies may be more susceptible to stochastic associations, and centrality metrics should be interpreted with caution. Although several hub genes such as PAX6, BMP7, and JAG1 are well-established regulators of odontogenesis, the functional involvement of newly identified nodes remains to be experimentally validated. Future studies should aim to improve network resolution by integrating larger and more diverse methylomic or transcriptomic datasets, including time-course analyses under defined inductive protocols. Another limitation is that this study focused exclusively on dental-derived stem cells, without comparison to non-dental stem cells or fully differentiated dental tissues, limiting the broader generalizability of the identified methylation signatures to odontogenic identity. Moreover, probes on sex chromosomes were excluded to avoid confounding effects from X-inactivation and the lack of Y chromosome homology; as a result, potential sex-specific epigenetic differences were not addressed. Finally, although age-related variation is a known determinant of DNA methylation, this analysis was designed to characterize cell-type-specific patterns under standardized culture conditions, rather than inter-individual variability. Future research should explore the influence of age, sex, and differentiation status more explicitly to enhance the biological interpretability and translational relevance.

In summary, this work substantially extends the scope of existing studies by delivering a functionally annotated epigenomic map of dental stem cells and highlighting the role of noncoding RNA regulation in odontogenic differentiation. This integrated approach provides a valuable resource for future research in regenerative dentistry and orthodontics. Regarding the latter, this study can certainly shed light on crucial aspects of the biology of tooth movement and its common adverse effects, such as root resorption, in relation to craniofacial development and, ultimately, individual variability in treatment outcomes.