Next Article in Journal
Behavioral Modeling of RF Power Amplifiers with Carrier-Frequency Generalization Using Interpolated Memory Polynomials
Previous Article in Journal
The Bioremediation of Nutrients and Heavy Metals in Watersheds: The Role of Floating Treatment Wetlands
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Applied Sciences
Volume 15
Issue 18
10.3390/app15189900
applsci-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

A Single-Nucleus Transcriptomic Atlas of Human Supernumerary Tooth Pulp Reveals Lineage Diversity and Transcriptional Heterogeneity Using PCA-Based Analysis

by
Eungyung Lee
1,2,3 and
In-Ryoung Kim
2,4,*
1
Department of Pediatric Dentistry, School of Dentistry, Pusan National University, Busandaehak-ro, 49, Mulguem-eup, Yangsan-si 50612, Gyeongsangnam-do, Republic of Korea
2
Dental and Life Science Institute, School of Dentistry, Pusan National University, Busandaehak-ro, 49, Mulguem-eup, Yangsan-si 50612, Gyeongsangnam-do, Republic of Korea
3
Dental Research Institute, Pusan National University Dental Hospital, 20 Geumo-ro, Mulgeum-eup, Yangsan-si 50612, Gyeongsangnam-do, Republic of Korea
4
Department of Oral Anatomy, School of Dentistry, Pusan National University, Busandaehak-ro, 49, Mulguem-eup, Yangsan-si 50612, Gyeongsangnam-do, Republic of Korea
*
Author to whom correspondence should be addressed.
Appl. Sci. 2025, 15(18), 9900; https://doi.org/10.3390/app15189900
Submission received: 20 August 2025 / Revised: 9 September 2025 / Accepted: 9 September 2025 / Published: 10 September 2025
(This article belongs to the Section Applied Dentistry and Oral Sciences)
Browse Figures
Versions Notes

Abstract

(1) Background: Supernumerary teeth are developmental anomalies, and their pulp tissue may harbor unique cellular and molecular features. However, the biology of this rare tissue remains poorly understood. This study aimed to characterize the cellular diversity and regenerative potential of supernumerary pulp at single-nucleus resolution. (2) Methods: Human supernumerary tooth pulp samples were analyzed using single-nucleus RNA sequencing. Gene expression profiles were processed and reduced to their main patterns of variation using principal component analysis (PCA), supported by clustering, pathway analysis, and lineage-specific scoring. (3) Results: The analysis suggested two dominant biological programs: a vascular–immune/stress axis and an extracellular matrix (ECM)/contractile remodeling axis. Vascular lineages were closely linked to immune and stress responses, while mesenchymal and perivascular populations were enriched in ECM-related pathways. Neural and glial contributions were relatively minor. (4) Conclusions: These findings suggest that supernumerary pulp appears to preserve key regenerative features similar to normal pulp, but with potential reinforcement of vascular–immune coupling and ECM remodeling. This work represents the first single-nucleus transcriptomic reference for supernumerary pulp, offering a foundation for future studies on dental pulp regeneration.

1. Introduction

Tooth development begins around the sixth week of embryogenesis with the formation of the dental lamina [1], which subsequently gives rise to the deciduous tooth germs, followed by the formation of permanent tooth germs in mid-gestation [2,3]. Supernumerary teeth can arise at either stage and have been attributed primarily to localized hyperactivity of the dental lamina or, more rarely, to dichotomy of the tooth bud [4,5]. Consequently, deciduous supernumeraries originate during the bud stage of primary tooth development, whereas permanent supernumeraries are associated with the initiation of permanent tooth germs [5].
Although supernumerary teeth are relatively common in pediatric populations, their developmental and biological properties remain poorly understood. Previous investigations have largely been limited to epidemiological surveys [6,7] or clinical case reports [5,8,9], providing only fragmentary evidence regarding their origins or functional significance. Given their unique developmental context, the pulp of supernumerary teeth may harbor distinct cellular compositions and molecular programs beyond those observed in normal pulp [10]. While the transcriptomic architecture of normal pulp has been increasingly characterized by recent single-cell studies [11,12], supernumerary pulp has yet to be comprehensively analyzed at the single-cell or single-nucleus level, leaving critical gaps in understanding whether it merely recapitulates normal pulp biology or instead possesses unique lineage features and functional potential.
Among computational strategies for single-cell transcriptomic analysis, principal component analysis (PCA) is particularly suited for identifying major axes of transcriptional variance [13,14]. More than a mathematical tool, PCA captures biologically meaningful programs that structure tissue heterogeneity [15], thereby offering insights into lineage-specific interactions and pathways governing pulp biology. Applying PCA to supernumerary pulp therefore presents an opportunity to delineate transcriptional features that distinguish it from normal pulp and to uncover not only regenerative potential but also developmental attributes.
Single-nucleus RNA sequencing (snRNA-seq) further provides a powerful means of systematically profiling human tissue at high resolution [16]. The pulp tissue of supernumerary teeth constitutes a rare biological resource [10], making the construction of its transcriptomic atlas of substantial academic value and providing a reference framework for comparative studies with normal pulp and other dental tissues.
The aim of this study was to generate the first single-nucleus transcriptomic atlas of human supernumerary tooth pulp and to delineate its cellular and molecular heterogeneity. Specifically, we sought to identify and classify heterogeneous cell populations, define major axes of transcriptional variance through PCA, provide baseline data for future comparative studies with normal pulp, and explore lineage-specific programs and signaling pathways that may illuminate its developmental characteristics and functional potential.

2. Materials and Methods

2.1. Supernumerary Tooth Pulp Collection and Nuclei Isolation

Human supernumerary tooth pulp tissues were obtained from pediatric patients treated at the Department of Pediatric Dentistry, Pusan National University Dental Hospital. All procedures involving human-derived materials were approved by the Institutional Review Board (IRB No. PNUDH 2025-03-006), and written informed consent was obtained from the legal guardians of all donors prior to sample collection. In total, pulp tissues were collected from two independent donors, three extracted supernumerary pulp (three anterior supernumerary teeth), from which nuclei were isolated for snRNA-seq analysis. Extractions were performed under routine clinical procedures, and pulp tissues were dissected under aseptic conditions, snap-frozen in liquid nitrogen, and stored at −80 °C until further processing.
For single-nucleus RNA sequencing, cryopreserved pulp tissues were processed using a protocol adapted from the BD Rhapsody™ single-nucleus RNA-seq workflow, with modifications optimized for dental pulp. Briefly, tissues were minced on ice in EZ Lysis Buffer (Sigma-Aldrich, St. Louis, MO, USA) supplemented with RNase inhibitor (Enzynomics, Daejeon, Republic of Korea) to lyse cell membranes while preserving nuclear integrity. Following sequential incubations on ice and centrifugation (500× g, 4 °C), nuclei were washed, passed through a 40 µm strainer (BD Falcon, Franklin Lakes, NJ, USA), and resuspended in cold 1% BSA/DPBS. Nuclear concentration and integrity were assessed using AO/PI staining with a LUNA-FX7™ automated cell counter (Logos Biosystems, Anyang, Republic of Korea), and only preparations with intact morphology were used for library construction.

2.2. mRNA Capture and Library Preparation

Isolated nuclei were loaded into BD Rhapsody™ microwells pre-coated with oligo(dT) beads to capture polyadenylated nuclear transcripts. Reverse transcription incorporated unique molecular identifiers (UMIs) and cell-specific barcodes. cDNA libraries were generated with the BD Rhapsody™ HT Xpress System using the Whole Transcriptome Analysis (WTA) Kit (BD Biosciences, San Jose, CA, USA). Library quality was assessed with the Agilent Bioanalyzer High Sensitivity DNA Kit (Agilent, Santa Clara, CA, USA), and quantification was performed using the KAPA Library Quantification Kit (Kapa Biosystems, Wilmington, MA, USA). Indexed libraries were size-selected for 300–800 bp fragments and sequenced on an Illumina NovaSeq X Plus platform (Illumina, San Diego, CA, USA) (2 × 150 bp, paired-end), with an average depth of ~50,000 reads per nucleus.

2.3. Preprocessing and Quality Control

Raw sequencing data were processed with the BD Rhapsody WTA Pipeline to generate digital gene expression (DGE) matrices. Downstream analyses were performed in Seurat (v4.3.0, R v4.3.3). Nuclei with fewer than 200 detected genes, more than 5000 detected genes, or >10% mitochondrial transcripts were removed. Gene expression values were log-normalized, and the top 2000 highly variable genes were retained. Counts were scaled and corrected for library size and mitochondrial content.

2.4. Principal Component Analysis, Clustering, and t-SNE Visualization

Dimensionality reduction was carried out using principal component analysis (PCA). A scree plot of the top 20 PCs indicated an elbow at PC5, and the first five PCs were retained for downstream analyses. Cells were projected into PC1–PC2 space, and unsupervised k-means clustering (k = 5) distinguished major populations. PC1 and PC2 gene loadings were extracted, and the top 30 contributors for each component were used for interpretation. In addition, t-distributed stochastic neighbor embedding (t-SNE) was performed on the same PCA-reduced space (first 20 PCs) to provide an independent visualization of cellular heterogeneity. t-SNE was implemented using the RunTSNE function in Seurat (v4.3.0) with default parameters, and results were visualized to confirm clustering consistency with PCA.

2.5. Pathway Enrichment and Module Scoring

Genes with high positive loadings on PC1 and PC2 were analyzed by over-representation analysis (ORA) and gene set enrichment analysis (GSEA) using clusterProfiler (v4.8.3) with MSigDB hallmark gene sets. Enrichment was considered significant at adjusted p < 0.05. To evaluate lineage-level contributions, PC1- and PC2-associated gene sets were applied in Seurat’s AddModuleScore function, and module scores were visualized with violin plots.

2.6. Visualization and Interpretation

Representative marker expression for PC1- and PC2-associated genes was displayed with dot plots and heatmaps (pheatmap v1.0.12). Cluster-level PCA centroids were calculated and projected to illustrate dominant axes of heterogeneity. Figures were generated using ggplot2 (v3.4.4) and ragg (v1.3.0) to ensure publication-quality output.

2.7. Data Availability

The raw and processed sequencing data have been deposited in the NCBI Gene Expression Omnibus (GEO) under accession number GSE306937.

3. Results

3.1. Principal Component Analysis Reveals Major Axes of Transcriptomic Variation in Supernumerary Tooth Pulp Cells

Principal component analysis (PCA) was conducted to reduce dimensionality and identify the dominant sources of transcriptional variation in supernumerary tooth pulp–derived cells. Scree plot analysis revealed that the first five principal components (PCs) accounted for the majority of explained variance, with a sharp decline observed beyond PC5, consistent with an “elbow” pattern (Figure 1A). Accordingly, PCs 1–5 were retained for downstream analyses. Projection of cells onto the PC1–PC2 space followed by k-means clustering (k = 5) distinguished five transcriptionally distinct clusters, highlighting the substantial heterogeneity among pulp-derived cell populations (Figure 1B). To gain further biological insights, we extracted the top positive and negative loading genes for PCs 1–5, which are comprehensively summarized in Table 1. In particular, PC1 contrasted extracellular matrix–related genes on the positive axis (e.g., COL27A1, COL1A1, ST8SIA1, COL24A1) with canonical vascular–endothelial markers on the negative axis (VWF, PECAM1, CD34, IGFBP4), suggesting an ECM versus endothelial program. PC2 integrated both neural/synaptic-related genes (NRXN1, LSAMP, SLIT2) and ECM/contractile markers (POSTN, COL3A1, VCAN), indicating a hybrid neural–matrix remodeling program rather than a purely neuronal axis. These patterns are visualized by the top 30 loading genes along PC1 and PC2 (Figure 1C,D), illustrating the major gene drivers that define each axis.

3.2. Pathway Enrichment Analysis of Principal Components

Enrichment analysis clarified the biological pathways underlying PCA axes. ORA of PC1-positive loadings revealed enrichment for KRAS signaling, interferon response, apoptosis, TGF-β signaling, and angiogenesis (Figure 2A). GSEA confirmed enrichment of interferon-α/γ response, hypoxia, apoptosis, and TNFα–NFκB signaling, supporting the interpretation of PC1 as an endothelial–immune/stress axis rather than a purely vascular signature (Figure 2C). PC2-positive loadings revealed enriched for epithelial–mesenchymal transition (EMT), angiogenesis, and KRAS signaling (Figure 2B), consistent with remodeling processes, although GSEA revealed no significant hallmark enrichments (Figure 2D). Overall, PC1 appeared to capture vascular–inflammatory variance, while PC2 seemed to reflect ECM- and remodeling-associated mesenchymal programs (Figure 2E).

3.3. PC1 and PC2 Module Scores Distinguish Vascular and Mesenchymal Lineages

Module scoring reinforced lineage-specific contributions. PC1 scores were elevated in endothelial clusters (arterial/tip, venous/activated), whereas PC2 scores were highest in mesenchymal/ECM-rich and pericyte/VSMC-like clusters (Figure 3A). Neural and Schwann/glial populations displayed low module scores for both components, suggesting limited contribution to principal variance axes. Marker expression was consistent with these results: PC1 genes (ADGRL4, ABCB1, ENG, FLT4, CAV1, DUSP6, ETS1, EMP1, ACE, BTG2) were characteristic of endothelial subsets, while PC2 genes (POSTN, ACTA2, VCAN, COL1A1/4A1/5A2, ADAM12, NRP1) were enriched in mesenchymal and perivascular lineages (Figure 3B). Notably, ERBB3 expression in glial populations was consistent with Neuregulin–ERBB signaling, while NRP1/JAG2 expression across both endothelial and perivascular lineages suggested angiogenic cross-talk.

3.4. PCA Reveals Vascular–Mesenchymal Dual Axes of Heterogeneity

Principal component loadings–based heatmap further supported the biological interpretation of PCA axes (Figure 4A). Genes with high loadings on PC1, including PECAM1, FLT1, EMCN, and CD34, were predominantly expressed in endothelial-enriched clusters, underscoring a vascular/angiogenic program. In contrast, PC2 loading genes such as COL12A1, POSTN, IGFBP4, and MMP20 were preferentially upregulated in mesenchymal/ECM-producing clusters, consistent with a contractile and matrix-remodeling axis. PCA centroid mapping further clarified these dominant axes of heterogeneity (Figure 4B). Cluster 2 was positioned along the positive PC1 axis, consistent with vascular/angiogenic enrichment, and Cluster 3 also occupied the upper-right quadrant with strong positive loadings on both PC1 and PC2, consistent with vascular/ECM programs. By contrast, Cluster 4 localized near the PCA origin with minimal displacement along either axis, indicating limited contribution to the primary variance structure. Endothelial-related clusters such as 5 and 9 were positioned on the negative PC1 axis, consistent with alternative endothelial signatures. By contrast, clusters 0, 1, and 10 localized near the PCA origin with significantly lower dispersion compared to other lineages (median r = 7.29 vs. 12.5; 1.7-fold difference, Wilcoxon P < 2.2 × 10−16), suggesting that neural/glial populations contributed minimally to the principal variance axes. Together, these findings suggest that transcriptional heterogeneity in supernumerary pulp is likely governed by vascular–mesenchymal programs: PC1 was enriched for endothelial and immune/stress-related gene loadings, which may reflect a vascular–immune/stress axis, whereas PC2 showed strong association with ECM and contractile gene programs, suggesting a role in matrix remodeling. To confirm that these findings were not restricted to PCA, we additionally performed t-SNE visualization (Figure 4C). The t-SNE map recapitulated the discrete clustering structure observed with PCA, supporting the potential robustness of the vascular–mesenchymal dual axis as a major source of heterogeneity in supernumerary pulp.

4. Discussion

This study structured the single-nucleus transcriptome of supernumerary tooth pulp through principal component analysis (PCA) and suggested that a vascular/mesenchymal dual axis may govern the variance structure of this tissue. The loading patterns in Figure 1C,D, and Table 1 suggest that PC1 strongly encompasses canonical endothelial markers (VWF, PECAM1, CD34, IGFBP4). However, because both ORA (Figure 2A) and GSEA (Figure 2C) consistently enriched interferon, TNFα–NFκB, hypoxia, and apoptosis pathways, we interpret PC1 as a vascular–immune/stress coupled axis rather than a simple “endothelial axis,” while acknowledging that this interpretation is based on computational inference and requires further validation. This interpretation was supported by module scores (Figure 3A), representative markers (Figure 3B), the loading heatmap (Figure 4A), and cluster centroid positioning (Figure 4B), where endothelial-enriched clusters occupied the positive direction of PC1. In other words, one of the main axes of variance in supernumerary pulp may show that the vascular program is tightly intertwined with immune and stress responses. This finding is consistent with reports that endothelial cell–derived IL-6 activates STAT3 signaling in DPSCs, enhancing self-renewal via Bmi-1 induction while promoting vasculogenic capacity [17]. Additionally, recent reviews have highlighted endothelial-mediated immune–angiogenic crosstalk as a regulatory hub in the pulp stem cell niche [18,19]. Notably, whereas previous single-cell studies of normal dental pulp described PC1 primarily as an endothelial–stromal axis, our data suggest that in supernumerary pulp this variance may be more closely associated with immune and stress programs, although further comparative validation with normal pulp will be required.
In contrast, PC2 initially appeared to include neural/synaptic-related genes such as NRXN1, LSAMP, and SLIT2 (Figure 1D, Table 1). However, integrating top-down pathway analyses (Figure 2B,D) with bottom-up lineage mapping (Figure 3A,B and Figure 4A) redirected interpretation toward a tentative, hybrid neural–ECM remodeling program rather than a purely neuronal axis. Specifically, EMT, angiogenesis, and KRAS-related pathways were prominent in ORA, PC2 module scores peaked in mesenchymal/ECM-rich and pericyte/VSMC-like clusters (Figure 3A), and markers such as POSTN, COL12A1, ACTA2, VCAN, and COL1A1/4A1/5A2 emphasized matrix production and contractility (Figure 3B). Further, PC2 high-loading genes dominated mesenchymal/ECM-producing clusters in the heatmap (Figure 4A), and clusters skewed toward the positive PC2 axis represented the apex of ECM/contractile programs (Figure 4B). Conceptually, this ECM–contractile axis corresponds to fibrotic and matrix-remodeling responses characteristic of injured pulp [20]. Recent single-cell studies of carious pulp similarly revealed increased ECM component expression (COL1A1, FN1) and enrichment of myofibroblast-like fibroblasts [21]. Moreover, PDGFRβ+ DPSCs have been shown to remodel extracellular matrix by secreting fibronectin, laminin, and collagen, thereby promoting angiogenesis and pulp-like tissue regeneration [22]. Nevertheless, given that PC2 also contained neural/synaptic-related genes in its loadings and that GSEA enrichment was limited, we interpret this axis as a tentative, hybrid neural–ECM remodeling program that remains hypothesis-generating and requires future validation (e.g., fibroblast differentiation assays, gel contraction assays).
Interestingly, neural/glial lineages contributed only minimally to variance in the PCA space of this study. Their low module scores (Figure 3A) and clustering near the origin (Figure 4B) suggest that, despite functional relevance, neural elements were not major drivers of transcriptional dispersion in supernumerary pulp. This contrasts with normal pulp, where glial populations are instrumental in sensory functions, homeostatic regulation, and repair, as shown in recent studies [23,24,25], and where neural–pulp signaling via mechanosensitive Piezo/PANX-1 pathways has been established [24,26]. Single-cell profiling of normal pulp further underscores immune and stromal interactions as dominant, with neural/glial lineages contributing less to variance [25]. In supernumerary pulp, however, the vascular axis, potentially associated with stronger immune and stress coupling, outweighed neural contributions, but we again emphasize that this conclusion remains hypothesis-generating and will require side-by-side validation with normal pulp datasets. To further ensure that these axes were not restricted to PCA, we additionally performed t-SNE visualization (Figure 4C). The t-SNE map reproduced the discrete clustering structure observed with PCA, supporting the potential robustness of the vascular–mesenchymal dual axis across dimensionality-reduction methods.
These findings provide two complementary insights: the preservation of the vascular/mesenchymal dual axis, which suggests that supernumerary pulp may share with normal pulp a regenerative blueprint of angiogenic support and stromal remodeling [27], and the immune/stress coupling of PC1, which suggests that supernumerary pulp harbors a vascular microenvironment particularly sensitive to inflammatory stimuli. This view is supported by studies showing that low-dose IFN-γ enhances DPSC proliferation and migration while suppressing odonto/osteogenic differentiation [28], and by evidence that NF-κB is pathologically overactivated in caries and pulpitis, driving aberrant matrix degradation and altered cell fate [29]. Importantly, the inhibition of NF-κB, for example through agents such as TSG-6, can restore DPSC differentiation even under inflammatory conditions [30]. Under acute injury, such immune–vascular coupling may prime angiogenesis for regeneration, whereas chronic stimulation could promote fibrosis and functional decline. The ECM/contractile axis represented by PC2 aligns with this duality.
In summary, supernumerary pulp may preserves the vascular/mesenchymal axis shared with normal pulp and thus retains regenerative potential, while at the same time exhibiting suggestive biases—namely potential reinforcement of immune/stress coupling and ECM remodeling—that may define its molecular profile. We now describe it as a rare and underexplored tissue resource that may provide complementary insights into vascular–immune–matrix interactions. Future work will be essential to validate these computationally derived axes through direct comparative single-nucleus RNA-seq analyses with normal pulp and functional assays such as in vitro tube formation, gel contraction, and co-culture systems, in order to clarify causal relationships within the vascular–immune–matrix triad we propose.
A limitation of this study is the restricted sample size, which may limit the generalizability of the identified transcriptional programs and introduce donor-specific variation. Nevertheless, this work provides the first single-nucleus transcriptomic atlas of supernumerary pulp, offering a valuable foundation for future studies incorporating larger cohorts and biological replicates.

5. Conclusions

This study suggests that transcriptional heterogeneity in supernumerary tooth pulp is primarily structured by a vascular/mesenchymal dual axis. PC1 appears to reflect a vascular–immune/stress program and PC2 seems to represent an ECM/contractile remodeling program, whereas neural/glial lineages contributed only minimally to major axes of variance. These findings suggest that supernumerary pulp may preserve the regenerative blueprint characteristic of dental pulp while exhibiting subtle biases toward immune/stress coupling and matrix remodeling. We emphasize that this work should not be interpreted as establishing direct translational or clinical relevance; rather, it provides the first single-nucleus transcriptomic atlas of this rare tissue and offers a foundational resource that may inform and guide future comparative and functional investigations.

Author Contributions

Conceptualization, E.L. and I.-R.K.; methodology, I.-R.K.; software, I.-R.K. and E.L.; validation, E.L. and I.-R.K.; formal analysis, E.L. and I.-R.K.; investigation, E.L.; resources, E.L.; data curation, I.-R.K.; writing—original draft preparation, E.L. and I.-R.K.; writing—review and editing, E.L. and I.-R.K.; visualization, I.-R.K.; supervision, I.-R.K.; project administration, I.-R.K.; funding acquisition, E.L. All authors have read and agreed to the published version of the manuscript.

Funding

This study was supported by 2025 Clinical Research Grant, Pusan National University Dental Hospital.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Pusan National University Dental Hospital (protocol code PNUDH 2025-03-006, 30 April 2025).

Informed Consent Statement

Written informed consent was obtained from the legal guardians of all pediatric donors involved in the study. The assent of children aged 6 years or older was additionally obtained and documented.

Data Availability Statement

Data are unavailable due to ethical and privacy restrictions.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
PCAPrincipal Component Analysis
PCPrincipal Component
snRNA-seqSingle-nucleus RNA sequencing
ORAOver-representation analysis
GSEAGene Set Enrichment Analysis
DPSCDental Pulp Stem Cell
ECMExtracellular Matrix
EMTEpithelial–Mesenchymal Transition
VSMCVascular Smooth Muscle Cell
NF-κBNuclear Factor kappa-light-chain-enhancer of activated B cells
STAT3Signal Transducer and Activator of Transcription 3
IL-6Interleukin-6
IFNInterferon
TNFαTumor Necrosis Factor alpha
RUNX2Runt-related transcription factor 2
NRP1Neuropilin-1
JAG2Jagged canonical Notch ligand 2
GEOGene Expression Omnibus
IRBInstitutional Review Board
UMIUnique Molecular Identifier
BSABovine Serum Albumin
DPBSDulbecco’s Phosphate-Buffered Saline
AO/PIAcridine Orange/Propidium Iodide

References

  1. Som, P.; Miletich, I. Review of the Embryology of the Teeth. Neurographics 2018, 8, 369–393. [Google Scholar] [CrossRef]
  2. Maas, R.; Bei, M. The genetic control of early tooth development. Crit. Rev. Oral Biol. Med. 1997, 8, 4–39. [Google Scholar] [CrossRef]
  3. Liu, M.; Goldman, G.; MacDougall, M.; Chen, S. BMP signaling pathway in dentin development and diseases. Cells 2022, 11, 2216. [Google Scholar] [CrossRef] [PubMed]
  4. Tworkowski, K.; Gasowska, E.; Baryla, D.; Gabiec, K. Supernumerary teeth-literature review. J. Pre-Clin. Clin. Res. 2020, 14, 18–21. [Google Scholar] [CrossRef]
  5. Rajab, L.; Hamdan, M. Supernumerary teeth: Review of the literature and a survey of 152 cases. Int. J. Paediatr. Dent. 2002, 12, 244–254. [Google Scholar] [CrossRef]
  6. Kumar, D.K.; Gopal, K.S. An epidemiological study on supernumerary teeth: A survey on 5000 people. J. Clin. Diagn. Res. 2013, 7, 1504–1507. [Google Scholar] [CrossRef]
  7. Demiriz, L.; Durmuşlar, M.C.; Mısır, A.F. Prevalence and characteristics of supernumerary teeth: A survey on 7348 people. J. Int. Soc. Prev. Community Dent. 2015, 5, S39–S43. [Google Scholar] [CrossRef]
  8. Alarcón, J.; Guzmán, J.; Masuko, T.S.; Cáceres, P.N.; Fuentes, R. Non-Syndromic Familial Mesiodens: Presentation of Three Cases. Diagnostics 2022, 12, 1869. [Google Scholar] [CrossRef] [PubMed]
  9. Dias, G.F.; Hagedorn, H.; Maffezzolli, M.D.L.; Silva, F.d.F.d.; Alves, F.B.T. Diagnosis and treatment of supernumerary teeth in the pediatric clinic-case report. Rev. Cefac. 2019, 21, e16319. [Google Scholar] [CrossRef]
  10. Lertruangpanya, K.; Roytrakul, S.; Surarit, R.; Horsophonphong, S. Comparative proteomic analysis of dental pulp from supernumerary and normal permanent teeth. Clin. Oral Investig. 2024, 28, 321. [Google Scholar] [CrossRef]
  11. Ren, H.; Wen, Q.; Zhao, Q.; Wang, N.; Zhao, Y. Atlas of human dental pulp cells at multiple spatial and temporal levels based on single-cell sequencing analysis. Front. Physiol. 2022, 13, 993478. [Google Scholar] [CrossRef]
  12. Yang, Y.; Alves, T.; Miao, M.; Wu, Y.; Li, G.; Lou, J.; Hasturk, H.; Van Dyke, T.; Kantarci, A.; Wu, D. Single-cell transcriptomic analysis of dental pulp and periodontal ligament stem cells. J. Dent. Res. 2024, 103, 71–80. [Google Scholar] [CrossRef]
  13. Lall, S.; Sinha, D.; Bandyopadhyay, S.; Sengupta, D. Structure-aware principal component analysis for single-cell RNA-seq data. J. Comput. Biol. 2018, 25, 1365–1373. [Google Scholar] [CrossRef]
  14. Chen, G.; Ning, B.; Shi, T. Single-cell RNA-seq technologies and related computational data analysis. Front. Genet. 2019, 10, 317. [Google Scholar] [CrossRef]
  15. Gough, A.; Stern, A.M.; Maier, J.; Lezon, T.; Shun, T.-Y.; Chennubhotla, C.; Schurdak, M.E.; Haney, S.A.; Taylor, D.L. Biologically relevant heterogeneity: Metrics and practical insights. SLAS Discov. 2017, 22, 213–237. [Google Scholar] [CrossRef]
  16. Deleersnijder, D.; Callemeyn, J.; Arijs, I.; Naesens, M.; Van Craenenbroeck, A.H.; Lambrechts, D.; Sprangers, B. Current methodological challenges of single-cell and single-nucleus RNA-sequencing in glomerular diseases. J. Am. Soc. Nephrol. 2021, 32, 1838–1852. [Google Scholar] [CrossRef]
  17. Oh, M.; Zhang, Z.; Mantesso, A.; Oklejas, A.E.; Nör, J.E. Endothelial-Initiated Crosstalk Regulates Dental Pulp Stem Cell Self-Renewal. J. Dent. Res. 2020, 99, 1102–1111. [Google Scholar] [CrossRef]
  18. Zaw, S.Y.M.; Kaneko, T.; Zaw, Z.C.T.; Sone, P.P.; Murano, H.; Gu, B.; Okada, Y.; Han, P.; Katsube, K.I.; Okiji, T. Crosstalk between dental pulp stem cells and endothelial cells augments angiogenic factor expression. Oral Dis. 2020, 26, 1275–1283. [Google Scholar] [CrossRef]
  19. Yushkov, B.; Chereshnev, V.; Korneva, E.; Yushkova, V.; Sarapultsev, A. Stem-Cell Niches in Health and Disease: Microenvironmental Determinants of Regeneration and Pathology. Cells 2025, 14, 981. [Google Scholar] [CrossRef]
  20. Goldberg, M.; Smith, A.J. Cells and Extracellular Matrices of Dentin and Pulp: A Biological Basis for Repair and Tissue Engineering. Crit. Rev. Oral Biol. Med. 2004, 15, 13–27. [Google Scholar] [CrossRef]
  21. Balic, A.; Perver, D.; Pagella, P.; Rehrauer, H.; Stadlinger, B.; Moor, A.E.; Vogel, V.; Mitsiadis, T.A. Extracellular matrix remodelling in dental pulp tissue of carious human teeth through the prism of single-cell RNA sequencing. Int. J. Oral Sci. 2023, 15, 30. [Google Scholar] [CrossRef]
  22. Di, T.; Wang, L.; Feng, C.; Xu, J.; Shao, B.; Cheng, B.; Wang, L.; Zhang, X.; Zhang, L.; Chen, Y. ECM remodeling by PDGFRβ+ dental pulp stem cells drives angiogenesis and pulp regeneration via integrin signaling. Stem Cell Res. Ther. 2025, 16, 283. [Google Scholar] [CrossRef]
  23. Wang, C.; Liu, X.; Zhou, J.; Zhang, Q. The Role of Sensory Nerves in Dental Pulp Homeostasis: Histological Changes and Cellular Consequences after Sensory Denervation. Int. J. Mol. Sci. 2024, 25, 1126. [Google Scholar] [CrossRef]
  24. Duan, Y.; Liang, Y.; Yang, F.; Ma, Y. Neural Regulations in Tooth Development and Tooth–Periodontium Complex Homeostasis: A Literature Review. Int. J. Mol. Sci. 2022, 23, 14150. [Google Scholar] [CrossRef]
  25. Opasawatchai, A.; Nguantad, S.; Sriwilai, B.; Matangkasombut, P.; Matangkasombut, O.; Srisatjaluk, R.; Charoensawan, V. Single-Cell Transcriptomic Profiling of Human Dental Pulp in Sound and Carious Teeth: A Pilot Study. Front. Dent. Med. 2022, 2, 71–80. [Google Scholar] [CrossRef]
  26. Ohyama, S.; Ouchi, T.; Kimura, M.; Kurashima, R.; Yasumatsu, K.; Nishida, D.; Hitomi, S.; Ubaidus, S.; Kuroda, H.; Ito, S.; et al. Piezo1-pannexin-1-P2X3 axis in odontoblasts and neurons mediates sensory transduction in dentinal sensitivity. Front. Physiol. 2022, 13, 891759. [Google Scholar] [CrossRef]
  27. Gronthos, S.; Mankani, M.; Brahim, J.; Robey, P.G.; Shi, S. Postnatal human dental pulp stem cells (DPSCs) in vitro and in vivo. Proc. Natl. Acad. Sci. USA 2000, 97, 13625–13630. [Google Scholar] [CrossRef]
  28. He, X.; Jiang, W.; Luo, Z.; Qu, T.; Wang, Z.; Liu, N.; Zhang, Y.; Cooper, P.R.; He, W. IFN-γ regulates human dental pulp stem cells behavior via NF-κB and MAPK signaling. Sci. Rep. 2017, 7, 40681. [Google Scholar] [CrossRef]
  29. Chen, Z.; Lang, G.; Xu, X.; Liang, X.; Han, Y.; Han, Y. The role of NF-kappaB in the inflammatory processes related to dental caries, pulpitis, apical periodontitis, and periodontitis—A narrative review. PeerJ 2024, 12, e17953. [Google Scholar] [CrossRef]
  30. Wang, Y.; Xie, Y.; Xue, N.; Xu, H.; Zhang, D.; Ji, N.; Chen, Q. TSG-6 Inhibits the NF-κB Signaling Pathway and Promotes the Odontogenic Differentiation of Dental Pulp Stem Cells via CD44 in an Inflammatory Environment. Biomolecules 2024, 14, 368. [Google Scholar] [CrossRef]
Applsci 15 09900 g001
Figure 1. Principal component analysis of supernumerary tooth pulp single-nucleus transcriptomes. (A) Scree plot of variance explained by the top 20 principal components, showing an elbow at PC5. (B) PCA score plot of nuclei on PC1 versus PC2, with clusters defined using the top five PCs. (C,D) Top 30 genes with the highest absolute loadings contributing to PC1 (C) and PC2 (D), illustrating the major gene drivers along each axis.
Figure 1. Principal component analysis of supernumerary tooth pulp single-nucleus transcriptomes. (A) Scree plot of variance explained by the top 20 principal components, showing an elbow at PC5. (B) PCA score plot of nuclei on PC1 versus PC2, with clusters defined using the top five PCs. (C,D) Top 30 genes with the highest absolute loadings contributing to PC1 (C) and PC2 (D), illustrating the major gene drivers along each axis.
Applsci 15 09900 g001
Applsci 15 09900 g002
Figure 2. Pathway enrichment analysis of principal components. (A) ORA of PC1-positive loadings showing enrichment for KRAS signaling, interferon response, apoptosis, TGF-β signaling, and angiogenesis. (B) ORA of PC2-positive loadings highlighting epithelial–mesenchymal transition (EMT), angiogenesis, and KRAS signaling. (C) GSEA of PC1-positive loadings confirming enrichment of immune, hypoxia, and apoptotic programs. (D) GSEA of PC2-positive loadings, showing no significant enrichment (weak signals in metabolism-related sets). (E) Summary barplot comparing top representative pathways, illustrating immune–stress programs in PC1 and EMT/angiogenesis programs in PC2.
Figure 2. Pathway enrichment analysis of principal components. (A) ORA of PC1-positive loadings showing enrichment for KRAS signaling, interferon response, apoptosis, TGF-β signaling, and angiogenesis. (B) ORA of PC2-positive loadings highlighting epithelial–mesenchymal transition (EMT), angiogenesis, and KRAS signaling. (C) GSEA of PC1-positive loadings confirming enrichment of immune, hypoxia, and apoptotic programs. (D) GSEA of PC2-positive loadings, showing no significant enrichment (weak signals in metabolism-related sets). (E) Summary barplot comparing top representative pathways, illustrating immune–stress programs in PC1 and EMT/angiogenesis programs in PC2.
Applsci 15 09900 g002
Applsci 15 09900 g003
Figure 3. PC1/PC2 module scores and representative marker expression. (A) Violin plots showing PC1- and PC2-associated module scores across major cell types. PC1 is enriched in endothelial clusters, while PC2 is high in mesenchymal/ECM-rich and pericyte/VSMC-like clusters. (B) Dot plot of representative genes. PC1 markers (ADGRL4, ABCB1, ENG, FLT4, CAV1, DUSP6, ETS1, EMP1, ACE, and BTG2) highlight endothelial subsets, whereas PC2 markers (POSTN, ACTA2, VCAN, COL1A1/4A1/5A2, ADAM12, and NRP1) are enriched in mesenchymal and perivascular populations.
Figure 3. PC1/PC2 module scores and representative marker expression. (A) Violin plots showing PC1- and PC2-associated module scores across major cell types. PC1 is enriched in endothelial clusters, while PC2 is high in mesenchymal/ECM-rich and pericyte/VSMC-like clusters. (B) Dot plot of representative genes. PC1 markers (ADGRL4, ABCB1, ENG, FLT4, CAV1, DUSP6, ETS1, EMP1, ACE, and BTG2) highlight endothelial subsets, whereas PC2 markers (POSTN, ACTA2, VCAN, COL1A1/4A1/5A2, ADAM12, and NRP1) are enriched in mesenchymal and perivascular populations.
Applsci 15 09900 g003
Applsci 15 09900 g004
Figure 4. PCA-based interpretation of transcriptomic heterogeneity. (A) Heatmap of top loading genes for PC1 and PC2, showing endothelial enrichment for PC1-associated genes and mesenchymal/ECM expression for PC2-associated genes. (B) PCA centroid mapping of clusters highlights a vascular–mesenchymal dual axis as the dominant source of variance, while neural/glial clusters exhibited low scores on both axes. (C) t-SNE embedding of the same nuclei recapitulates discrete cluster structure observed with PCA, supporting the robustness of the vascular–mesenchymal axis across dimensionality-reduction methods. Cluster identities were annotated as follows: Cluster 0, Neural-like (synaptic neuron); Cluster 1, Neural-like (glutamatergic neuron); Cluster 2, Odontoblast-like/dentinogenic; Cluster 3, Neural-like; Cluster 4, Mesenchymal/ECM, metabolic; Cluster 5, Endothelial/vascular; Cluster 6, Endothelial; Cluster 7, Neural progenitor-like; Cluster 8, Pericyte/VSMC-like; Cluster 9, Endothelial (angiogenic); Cluster 10, Schwann/Glial.
Figure 4. PCA-based interpretation of transcriptomic heterogeneity. (A) Heatmap of top loading genes for PC1 and PC2, showing endothelial enrichment for PC1-associated genes and mesenchymal/ECM expression for PC2-associated genes. (B) PCA centroid mapping of clusters highlights a vascular–mesenchymal dual axis as the dominant source of variance, while neural/glial clusters exhibited low scores on both axes. (C) t-SNE embedding of the same nuclei recapitulates discrete cluster structure observed with PCA, supporting the robustness of the vascular–mesenchymal axis across dimensionality-reduction methods. Cluster identities were annotated as follows: Cluster 0, Neural-like (synaptic neuron); Cluster 1, Neural-like (glutamatergic neuron); Cluster 2, Odontoblast-like/dentinogenic; Cluster 3, Neural-like; Cluster 4, Mesenchymal/ECM, metabolic; Cluster 5, Endothelial/vascular; Cluster 6, Endothelial; Cluster 7, Neural progenitor-like; Cluster 8, Pericyte/VSMC-like; Cluster 9, Endothelial (angiogenic); Cluster 10, Schwann/Glial.
Applsci 15 09900 g004
Table 1. Top positive and negative loading genes for the first five principal components (PC1–PC5) in supernumerary tooth pulp transcriptomes.
Table 1. Top positive and negative loading genes for the first five principal components (PC1–PC5) in supernumerary tooth pulp transcriptomes.
PCDirectionGenes
PC1PositiveCOL27A1, COL1A1, ERBB4, MMP20, ST8SIA1, COL24A1, SEMA3E, SORBS2, SPOCK3, LTBP1, DTNBP1, GRAMD1B, PCDH7, SLC2A13, GRIK1, DOK6, MAP1B, LRP1B, GPC3, CD36, SLIT3, CYFIP2, WDR72, B3GALT1, PPFIA2, RANBP3L, COBL, DIAPH3, NRXN1, TMEM117
NegativePTPRB, HLA-E, TXNIP, SHANK3, NFIB, CD34, IGFBP7, PECAM1, CD74, FLT1, VWF, IGFBP4, EMCN, SPARCL1, KLF2, HLA-B, DIPK2B, LDB2, CYYR1, PLCB4, EGFL7, MMRN2, MYRIP, ADGRF5, B2M, HLA-A, EPAS1, ADGRL4, FZD4, PLCB1
PC2PositivePTPRK, ST8SIA1, MMP20, SLC12A2, SEMA3E, SPOCK3, SLC2A13, COL27A1, GRIK1, DOK6, CD36, GPC3, HDAC9, MAP1B, ERBB4, WDR72, SORBS2, UCK2, CYFIP2, FAM107B, DTNBP1, SLIT3, COBL, ABLIM2, SYBU, ENSG00000289986, LTBP1, PHEX, PPFIA2, ZNF385B
NegativeNRXN1, CPA6, LSAMP, LRP1B, SLIT2, ENSG00000255595, BMPR1B, POSTN, TF, SLC4A4, COL3A1, VCAN, NRXN3, ENSG00000225096, CTNNA2, CHSY3, PRICKLE1, PTN, EFNA5, ITGA8, GALNTL6, IGFBP5, CP, NGF, ARL15, CXCL14, ENSG00000239268, PDZRN3, CHRM3, GRM7
PC3PositiveCOL14A1, CDH6, CDH19, ITIH5, LGI4, ERBB3, MCAM, PLP1, OLFML2A, PLCE1, SOX5, MPZ, GUCY1A1, ASPA, DMD, ACTA2, XKR4, SOD3, SOX10, NGFR, NR2F2-AS1, TAGLN, EDNRA, PMP22, CRYAB, SLC35F1, GJC1, PARM1, EGFLAM, GFRA3
NegativeLRP1B, TSHZ2, CPA6, ENSG00000255595, CHRM3, SLCO2A1, POSTN, SLC4A4, LINC01515, ZNF521, ARL15, CTNNA2, ITGA9, ENSG00000225096, LSAMP, RAPGEF4, BMPR1B, ADAMTSL1, CADPS2, PRKCH, DNM3, PRICKLE1, ABCB1, ST6GALNAC3, LINC02147, ITGA8, DIPK2B, NOSTRIN, EMCN, TLL1
PC4PositiveCDH19, ERBB3, PLP1, XKR4, ASPA, SOX10, MPZ, COL28A1, GFRA3, L1CAM, GAS2L3, CADM2, TRPM3, ZNF536, SCN7A, CHL1, HSPA12A, ADAM23, SORCS1, NKAIN3, ITGB8, CADM3, SLITRK5, MATN2, FIGN, MAL, ADGRB3, GRIK3, NCAM2, NGFR
NegativeGUCY1A1, NOTCH3, ACTA2, EDNRA, SOD3, FRZB, TAGLN, PARM1, MYH11, RCAN2, NDUFA4L2, IRAG1, GJC1, CARMN, STEAP4, RGS5, MCAM, THY1, ANGPT2, LINGO1, CASQ2, SMOC2, SUSD2, CACNA1H, TEX41, SGIP1, INPP4B, ABCC9, CPE, GUCY1B1
PC5PositiveNR2F2-AS1, INPP4B, LETR1, EDNRA, PARM1, ABCC9, TEX41, CCDC102B, TRPC6, NEURL1B, CDH6, SMOC2, STEAP4, RGS6, DACH1, CARMN, LINC02147, GUCY1A1, RCAN2, GJC1, RAPGEF4, ARHGAP6, SGIP1, POU6F2, IRAG1, UBE2E2, DOCK8, SOX5, NR2F2, ARHGAP15
NegativeMT-ATP6, MT-CO3, MT-ND4, MT-ND2, MT-ND1, MT-ND3, MT-CO2, MT-CYB, MT-CO1, CLU, CXCL14, MT-ND5, PTN, CRABP1, MMP2, EEF1A1, ACTB, TF, IGFBP5, TMSB4X, TPT1, IFITM3, IGFBP3, RPL34, RPS6, RPL32, RPL23, KCTD12, RPL39, RPS8
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Lee, E.; Kim, I.-R. A Single-Nucleus Transcriptomic Atlas of Human Supernumerary Tooth Pulp Reveals Lineage Diversity and Transcriptional Heterogeneity Using PCA-Based Analysis. Appl. Sci. 2025, 15, 9900. https://doi.org/10.3390/app15189900

AMA Style

Lee E, Kim I-R. A Single-Nucleus Transcriptomic Atlas of Human Supernumerary Tooth Pulp Reveals Lineage Diversity and Transcriptional Heterogeneity Using PCA-Based Analysis. Applied Sciences. 2025; 15(18):9900. https://doi.org/10.3390/app15189900

Chicago/Turabian Style

Lee, Eungyung, and In-Ryoung Kim. 2025. "A Single-Nucleus Transcriptomic Atlas of Human Supernumerary Tooth Pulp Reveals Lineage Diversity and Transcriptional Heterogeneity Using PCA-Based Analysis" Applied Sciences 15, no. 18: 9900. https://doi.org/10.3390/app15189900

APA Style

Lee, E., & Kim, I.-R. (2025). A Single-Nucleus Transcriptomic Atlas of Human Supernumerary Tooth Pulp Reveals Lineage Diversity and Transcriptional Heterogeneity Using PCA-Based Analysis. Applied Sciences, 15(18), 9900. https://doi.org/10.3390/app15189900

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop