A single-cell atlas of human teeth

Summary

Teeth exert fundamental functions related to mastication and speech. Despite their great biomedical importance, an overall picture of their cellular and molecular composition is still missing. In this study, we have mapped the transcriptional landscape of the various cell populations that compose human teeth at single-cell resolution, and we analyzed in deeper detail their stem cell populations and their microenvironment. Our study identified great cellular heterogeneity in the dental pulp and the periodontium. Unexpectedly, we found that the molecular signatures of the stem cell populations were very similar, while their respective microenvironments strongly diverged. Our findings suggest that the microenvironmental specificity is a potential source for functional differences between highly similar stem cells located in the various tooth compartments and open new perspectives toward cell-based dental therapeutic approaches.

Graphical abstract

Highlights

• •Dental atlas of the pulp and periodontal tissues of human teeth
• •Identification of three common MSC subclusters between dental pulp and periodontium
• •Dental pulp and periodontal MSCs are similar, and their niches diverge

Cell Biology; Stem Cells Research; Omics; Transcriptomics

Introduction

Teeth are composed of a unique combination of hard and soft tissues. Enamel, the hardest tissue of the human body, covers the crown of the tooth, and it is supported by a second less mineralized tissue, the dentin. The central portion of the tooth is occupied by the dental pulp, a highly vascularized and innervated tissue that is lined by odontoblasts, the cells responsible for dentin formation. The tooth is anchored to the surrounding alveolar bone via the periodontium, which absorbs the various shocks associated with mastication and provides tooth stability by continuously remodeling its extracellular matrix, the periodontal ligament (Nanci, 2013). The development of the tooth results from sequential and reciprocal interactions between cells of the oral epithelium and the cranial neural crest-derived mesenchyme (Kollar, 1986; Mitsiadis and Graf, 2009; Nanci, 2013). Oral epithelial cells give rise to ameloblasts that produce enamel. Dental mesenchymal cells give rise to odontoblasts that form the dentin, as well as to the dental pulp (Mitsiadis and Graf, 2009; Nanci, 2013). Dental pulp and periodontal tissues contain mesenchymal stem cells (MSCs), namely the dental pulp stem cells (DPSCs) and periodontal stem cells (PDSCs) (Gronthos et al., 2000; Roguljic et al., 2013). The epithelial cell remnants in the periodontal space upon dental root completion form an additional tooth-specific epithelial stem cell population (Athanassiou-Papaefthymiou et al., 2015). DPSCs and PDSCs are multipotent and respond to a plethora of cellular, chemical, and physical stimuli to guarantee homeostasis and regeneration of dental tissues. Isolated DPSCs and PDSCs are the subject of intense investigation as possible tools for the regeneration of both dental and non-dental tissues (Chen et al., 2020; Iohara et al., 2011; Lei et al., 2014; Orsini et al., 2018; Ouchi and Nakagawa, 2020; Trubiani et al., 2019; Xuan et al., 2018). In vivo studies aiming at the regeneration of dental pulp and periodontal tissues were however not completely successful (Chen et al., 2020; Xu et al., 2019; Xuan et al., 2018). Indeed, the behavior of these and other stem cell populations is regulated by molecular cues produced in their microenvironment by stromal cells, neurons, vascular-related cells, and immune cells, as well as by physical factors such as stiffness, topography, and shear stress (Chacon-Martinez et al., 2018; Machado et al., 2016; Oh and Nor, 2015; Oh et al., 2020; Pagella et al., 2015; Rafii et al., 2016; Scadden, 2014; Yang et al., 2017). Much effort has been spent in the last decades to understand the fine composition of tissues and the cellular and molecular mechanisms that mediate the cross talk between stem cells and their environment to drive regenerative processes (Blache et al., 2018; Chakrabarti et al., 2018; Lane et al., 2014; Mitsiadis et al., 2017a; Oh et al., 2020; Rafii et al., 2016). Concerning teeth, one recent article reported the single-cell RNA sequencing analysis of mouse dental tissue and the human dental pulp, focusing mostly on the continuously growing mouse incisor and on the conservation between species of cellular populations and features that underlie tooth growth (Krivanek et al., 2020). A second single-cell RNA sequencing analysis study in the continuously erupting mouse incisor identified dental epithelial stem cells subpopulations that are important upon tooth injury and contribute to enamel regeneration (Sharir et al., 2019). Despite the great clinical relevance, the cellular composition of the two main human dental tissues, the dental pulp and the periodontium, has not been investigated in deeper detail.

Results

We used single-cell profiling to elucidate the cellular and molecular composition of human teeth and shed light on fundamental biological questions concerning dental stem cell behavior. We first characterized the cell populations that form the dental pulp and periodontal tissues in human teeth, and thus, we focused on the MSC populations.

Single-cell RNA sequencing analysis of the dental pulp of human teeth

We first analyzed the cellular and molecular composition of the dental pulp of human teeth. For this purpose, we isolated dental pulps from five extracted third molars, dissociated them into single-cell suspensions and proceeded with droplet-based encapsulation (using the 10x Genomics Chromium System) and sequencing. Our analyses yielded a total of 32′378 dental pulp cells (Figure S1). We identified 15 clusters of cells using the graph clustering approach implemented in Seurat v3 (Hafemeister and Satija, 2019) and visualized them using uniform manifold approximation and projection (McInnes et al., 2018) (Figures 1A–1D). Our analysis identified a variety of cell populations including MSCs, fibroblasts, odontoblasts, endothelial cells (ECs), Schwann cells (ScCs), immune cells, epithelial-like cells, and erythrocytes (Figure 1B). MSCs were characterized by the higher expression of FRZB, NOTCH3, THY1, and MYH11 (Figures 1C, 1D, and 1G) (log-fold change of 2.07, 1.24, 1.54, and 1.4, respectively, and adjusted p value < 0.001, compared to other cell types in the pulp) and represented on average 12% of the dental pulp tissue (mean proportion = 0.12 and standard deviation (sd) = 0.05; Figure 1E). MSCs were localized around the vessels (Figure 1G), where the perivascular niches are formed (Lovschall et al., 2007; Shi and Gronthos, 2003), as well as in the sub-odontoblastic area, which is another potential stem cell niche location in the dental pulp (Mitsiadis and Rahiotis, 2004; Mitsiadis et al., 2003). The fibroblastic compartment composed the bulk of the dental pulp tissue (mean proportion = 0.38 and sd = 0.1; Figure 1E). Different fibroblastic clusters could be identified. Fibroblasts were characterized by the expression of collagen-coding genes (e.g., COL1A1; logFC = 0.91 and adjusted p value <0.001) and MDK (logFC = 1.44 and adjusted p value <0.001), a gene whose expression is restricted to the dental mesenchyme during mouse odontogenesis (Mitsiadis et al., 1995), as well as by the reduced expression of FRZB (logFC = −0.76 and adjusted p value <0.001; Figures 1B–1D). One cluster, characterized by the high expression of osteomodulin/osteoadherin (Figure S3), represented an intermediate state between MSCs and fibroblasts, with shared gene expression from these two groups. Odontoblasts were characterized by the expression of dentin sialophosphoprotein (DSPP) (Figure 1I) and dentin matrix acidic phosphoprotein 1 (DMP1), genes encoding for phosphoproteins that constitute essential components of the dentin matrix (D'Souza et al., 1997; Liang et al., 2019). ECs, which constitute important components of the MSC microenvironment (Rafii et al., 2016), showed a significant degree of heterogeneity (Figures 1B and 1I, and S4). Three well-defined clusters of ECs were detected. A first cluster was characterized by the expression of EDN1/CLDN5 and represented arterial ECs (Figure S4). A second endothelial cluster was characterized by the expression of ACKR1/CD234 (Figures 1I and S4) and represented postcapillary and collecting venules. The third main endothelial cluster was characterized by the expression of the insulin receptor (INSR) and RGCC (Figure S4). Immune cells are part of all healthy tissues and organs (Senovilla et al., 2013) and were also consistently detected in the healthy dental pulp tissues. This cluster mostly consisted of T cells and macrophages, characterized by the expression of PTPRC, CD3E, and CSF1R (Figures 1B and S5). Nerve fibers are crucial elements of stem cell niches, as they regulate MSC functions and fates (Pagella et al., 2015). ScCs formed two clearly distinct clusters of SOX10⁺ cells, identified as myelinating MBP⁺-ScCs and non-myelinating GFRA3⁺-ScCs (Figures 1B, 1C, 1J and S6A). MBP⁺-ScCs were mostly localized around major nerve fibers entering the dental pulp, while GFRA3⁺-ScCs were detected at a distance from nerve fibers and mostly within the sub-odontoblastic regions (Figure 1J), where NOTCH3-expressing MSCs were localized. We further identified an epithelial-like cell population within the human dental pulp tissue (Figures 1B and 1C), in accordance with previous reports in human deciduous teeth (Nam and Lee, 2009). These epithelial cells express keratin-coding genes such as KRT14 and KRT5, as well as stratifin (SFN) (Figures 1C and S6B). We validated the presence of the epithelial cluster within the dental pulp with an immunofluorescent staining against keratin14 (Figure 1K). We finally identified a population of erythrocytes that is characterized by the presence of the beta-hemoglobin-coding transcript HBB.

Figure 1.

Single-cell RNA sequencing analysis of adult healthy human dental pulps

(A) Schematic representation of the experimental setup.

(B) UMAP visualization of color-coded clustering of the dental pulp (n > 32′000 cells).

(C) Expression of example key genes used for the annotation and the characterization of the clusters.

(D) Heatmap showing expression of most differentially expressed genes between each cluster and all others.

(E) Boxplot of relative abundance of main cell types composing the dental pulp from each patient. Boxes illustrate the interquartile range (25th to 75th percentile), the median is shown as the middle band, and the whiskers extend to 1.5 times the interquartile range from the top (or bottom) of the box to the furthest datum within that distance. Any data points beyond that distance are considered outliers.

(F) Hematoxylin-eosin staining showing the structure of the dental pulp.

(G) Immunofluorescent staining showing localization of FRZB-expressing (green color, white arrows) MSCs around blood vessels (laminin positive, red color).

(H) Immunofluorescent staining showing localization of DSPP-expressing odontoblasts.

(I) Immunofluorescent staining showing CD234-expressing (red color) endothelial cells (CD31⁺, green color). CD31⁺CD234⁺ cells are marked by arrows; CD31⁺CD234^- cells are marked by arrowheads.

(J) Immunofluorescent staining showing localization of MBP-expressing, myelinating Schwann cells (green color), and GFRA3-expressing, non-myelinating Schwann cells (M, red color).

(K) Immunofluorescent staining showing localization of KRT14-expressing epithelial cells (red color). Blue color: DAPI. d, dentin; nf, nerve fibers; o, odontoblasts; p, pulp; v, vessels. Scale bars: (F), 300 μm; (G) and (K), 25 μm; (H), 50 μm; (I), 200 μm; (J), 250 μm.

Single-cell RNA sequencing analysis of the periodontium of human teeth

We then set out to identify and characterize the cell populations that compose the periodontium of human teeth. We obtained the periodontal tissue by scraping the surface of the apical two-thirds of the roots of five extracted third molars. We dissociated the isolated periodontal tissue to single-cell suspensions and processed them for single-cell RNA sequencing (Figure 2A). We obtained a total of 2′883 periodontal cells (Figure S2A) and identified 15 clusters of cells (Figure 2B). MSCs, fibroblasts, ECs, ScCs, immune cells, epithelial-like, cells and erythrocytes composed the human periodontal tissue (Figure 2B). Similar to the dental pulp tissue, MSCs represented a large fraction of the periodontium (mean proportion = 0.19, sd = 0.11 and se = 0.05). We detected a cluster of MSCs expressing FRZB, NOTCH3, MYH11, and THY1 (logFC of 1.83, 1.24, 1.47, and 1.61, respectively, and adjusted p value <0.001, compared to other cells in the periodontium; Figures 2B and 2C). The fibroblastic compartment was defined by cells expressing MDK (logFC = 1.25 and adjusted p value < 0.001; Figure 2C) and collagen-coding genes such as COL1A1 (Figures 2B and 2C; logFC = 3.42 and adjusted p value < 0.001). This cluster represented a small fraction of the periodontium (mean proportion = 0.11, sd = 0.08 and se = 0.03; supplemental information Appendix, Figure S2). ECs were more abundant than fibroblasts and represented a big proportion of the periodontal tissues (mean proportion = 0.19, sd = 0.17 and se = 0.07; Figure S2B). We distinguished two main separate ECs clusters, which were characterized by the expression of EDN1/CLDN5/CXCL12 and ACKR1/CD234 (Figure 2B), similar to what observed in the dental pulp (Figure 1B). A cluster of INSR/RGCC-expressing ECs was observed as an intermediate state between the EDN1/CLDN5/CXCL1 and ACKR1/CD234 ECs clusters. ScCs represented a minor population within the periodontium. ScCs expressed SOX10, GFRA3, NGF, and NGFR (Figures 2B and 2C; Dataset). The periodontium was characterized by the presence of PTPRC⁺ immune cells, including T cells (CD3E⁺/CD3D⁺), B cells (MZB1⁺), monocytes, and macrophages (CSF1R⁺) (Figure 2C; Dataset). Unexpectedly, we found that the most abundant population of the periodontium consisted of epithelial cells (mean proportion = 0.28, sd = 0.27 and se = 0.12) (Figures 2B–2E and S2B). Epithelial cells formed different subclusters, characterized by the expression of epithelial genes such as KRT14 and ODAM, signaling molecules such as WNT10A, and specific sets of interleukin-coding genes such as IL1A and IL1B (Figure S7). Using immunofluorescent staining, we showed that the epithelial cells are organized in discrete islets along the entire periodontium (Figures 2H and 2I). Finally, we identified a small cluster of erythrocytes expressing HBB (Figure 2B).

Figure 2.

Single-cell RNA sequencing analysis of the periodontium

(A) Schematic representation of the experimental setup.

(B) UMAP visualization of color-coded clustering of the periodontium (n > 2.800 cells).

(C) Expression of example key genes used for the annotation and the characterization of the clusters.

(D) Heatmap showing expression of the gene most differentially expressed between each cluster and all other ones.

(E) Boxplot of relative abundance of main cell types composing the dental periodontium from each patient. Boxes illustrate the interquartile range (25th to 75th percentile), the median is shown as the middle band, and the whiskers extend to 1.5 times the interquartile range from the top (or bottom) of the box to the furthest datum within that distance. Any data points beyond that distance are considered outliers.

(F) Hematoxylin-eosin staining showing the structure of the periodontium.

(G) Immunofluorescent staining showing localization of FRZB-expressing MSCs (green color). Red color: laminin, marking blood vessels; blue color: DAPI.

(H) Immunofluorescent staining showing localization of KRT14-expressing epithelial cells (green color) and vimentin-expressing (VIM) mesenchymal cells (red color) along the periodontium. Blue color: DAPI.

(I) Higher magnification of (H). ab, alveolar bone; lam, laminin; pe, periodontium; rd, root dentin; v, vessel. Scale bars: (F), 100 μm; (G), 40 μm; (I), 50 μm; (H), 150 μm.

Comparison of dental pulp and periodontal stem cell populations

The establishment of the single-cell atlas of the dental pulp and periodontium of human teeth allows further analyses and comparisons at the molecular level between these two tissues (Figures 3 and 4). Therefore, we first proceeded with the comparison between the stem cell clusters detected in these two dental components. In both tissues, MSCs were characterized by the expression of FRZB and NOTCH3 (logFC = 2.05 and 1.27 and p values < 0.001; Figures 1C, 1D, 2C, 2D, and 3E–3G). We then analyzed the composition of the dental pulp and periodontal MSC clusters in deeper detail. Upon separate subclustering of the NOTCH3⁺FRZB⁺ pulp and periodontal MSCs, we identified three major MSC subpopulations (Figures 3A–3D). Unexpectedly, the main dental pulp and periodontal MSC populations exhibited very similar molecular signatures. Both compartments contained two main MSC clusters characterized by increased expression of MYH11 (logFC = 2.00 and p value <0.001) and THY1 (logFC = 1.63 and p value <0.001), respectively, when compared to all other clusters (Figures 3B and 3D). We detected a second THY1-positive (and MYH11-negative) MSC cluster, with increased expression of CCL2 (logFC = 3.46 and p value <0.001 when compared to other clusters; Figures 3B and 3D). The CCL2⁺ MSC cluster also expressed genes associated with the remodeling of the extracellular matrix, such as TNC (tenascin C) (Figure 3E).

Figure 3.

Comparative analysis of the MSC compartment in the pulp and the periodontium

(A) UMAP visualization of pulp clusters, highlighting the MSC compartment.

(B) Feature plots showing genes that characterize the main MSC subclusters within the pulp.

(C) UMAP visualization of periodontium clusters, highlighting the MSC compartment.

(D) Feature plots showing genes that characterize the main MSC subclusters within the periodontium.

(E and F) Feature plots showing the distribution of the expression of common genes characterizing dental pulp (E) and periodontal (F) MSCs. FRZB is expressed by all MSCs, both in the dental pulp and in the periodontium. ACTA2, RERGL, and PLN (phospholamban) are particularly enriched in the MYH11⁺ MSC subcluster, while DCN (decorin) and STEAP4 are highly expressed in the THY1⁺ MSC subcluster. TNC (tenascin) is highly expressed in the CCL2⁺ MSC subcluster. Previous studies have shown that TNC is expressed during odontogenesis in the dental mesenchyme (Vainio et al., 1989) as well as in the mature periodontium at the interface with cementum and with the alveolar bone (Lukinmaa et al., 1991; Midwood et al., 2016).

(G) Dot plot showing the top 40 genes that characterize both dental pulp and periodontal MSCs against other dental cell types. Light yellow highlights genes of particular interest. MSCs in the dental pulp and the periodontium shared the expression of many stem cell markers and genes associated with stem cell function. MYH11 codes for a myosin heavy chain and its expression has been primarily observed in perivascular smooth muscle cells and pericytes, a common source of MSCs (Murgai et al., 2017). Similarly, ACTA2 is often expressed in pericytes. THY1 codes for CD90, a cell surface protein used as a classical marker for MSCs (An et al., 2018; Balic et al., 2010). MCAM/CD164 is a classical marker of MSCs in dental and non-dental tissues (Shi and Gronthos, 2003). RGS5 expression marks the perivascular NOTCH3⁺ MSCs in the dental pulp (Lovschall et al., 2007). Most MSCs populations express ID4, which codes for a transcription factor that inhibits cell differentiation (Junankar et al., 2015; Patel et al., 2015).

(H) Heatmap showing differential gene expression between the periodontium and pulp MSCs. See Table S1, for logFC and adjusted p value.

Figure 4.

Comparison of the dental pulp and periodontal microenvironment

(A) UMAP plot showing the clusters distribution in the merged dental pulp/periodontium data set.

(B) Comparison of the relative abundance of the different cell types composing the pulp and the periodontium. Epithelial cells are the most abundant cell type in the periodontium, while fibroblasts constitute the most abundant cluster in the dental pulp.

(C) Jaccard similarity plot between the various periodontium and pulp clusters using the top three thousand differentially expressed genes. See supplemental information Appendix, Table S2, for Jaccard similarity ranking.

Next, we merged the dental pulp and periodontium data sets and jointly clustered them to compare the transcriptomes of their MSCs (Figures 3H and 4A). We detected gene expression log-fold changes higher than 0.25 in only 333 genes and as few as 33 genes with a logFC higher than 1 (p values < 0.05, Figure 3H; Table S1). MSCs from the two tissues showed no significant differences in the expression of the already mentioned NOTCH3, FRZB, THY1, and MYH11, as well as the other stem cell markers MCAM/CD146, RGS5, ACTA2, and ID4 (Figure 3G). Some genes were significantly more expressed in periodontal MSCs than in the pulp, such as CCL2 (logFC = 0.78 and p < 0.001), and those coding for collagens (e.g., COL3A1, COL1A1, COL6A1, COL6A3, COL4A1) (logFC = 1.65, 1.59, 1.02, 0.70, 0.86, respectively, and adjusted p values < 0.001; Figure 3H; Tables S1 and S3; Figure S8). Periodontal MSCs were also characterized by higher expression of SPARC/osteonectin, a secreted molecule fundamental for the regulation of periodontal homeostasis and collagen content (logFC = 1.00 and p value < 0.001; Figure 3H; Tables S1 and S6). In contrast to the periodontal MSCs, dental pulp MSCs expressed higher levels of CXCL14 and RARRES1 (logFC = 2.04 and 1.00, respectively, and p values < 0.001; Figure 3H; Table S1). Surprisingly, dental pulp MSCs strongly expressed KRT18, a gene previously reported to be exclusively expressed in cells of single-layered and pseudostratified epithelia (logFC = 1.46, p value < 0.001; Table S1, Figure S10).

Comparative analysis of the MSC microenvironment in the dental pulp and periodontium of human teeth

We then compared the two specific MSC niches in the dental pulp and the periodontium (Figures 4 and S2). We observed that their cell compositions diverged in relative proportion for certain cell types, mainly the fibroblastic and epithelial compartments. Fibroblasts represented the most abundant cell population within the dental pulp, while in the periodontium, the proportion of fibroblasts was considerably lower (mean dental pulp = 0.38 and se = 0.04; mean periodontium: 0.11 and se = 0.03. Figures 4B and S2). Likely, due to the high variability of scRNA-seq, it is not possible to statistically confirm this difference using our data set. Genes coding for collagens and matrix metalloproteases (MMPs) were highly expressed by periodontal fibroblasts and MSCs (Figures S8 and S9 and Tables S3 and S5) when compared to their pulp counterparts. Interestingly, genes coding for bone-specific proteins, such as osteonectin (SPARC), osteocalcin (BGLAP), and bone sialophosphoprotein (BSP), were expressed by the periodontal fibroblasts (Figure S11 and Table S6). Periodontal fibroblasts also expressed MGP (matrix Gla protein), a potent inhibitor of mineralization (Figure S11, Table S6). The periodontium was characterized by a larger proportion of cells expressing epithelial cell markers such as KRT5 and KRT14 (Figure S2B). As in the case of fibroblasts, it was not possible to statistically confirm this difference in proportion. These periodontal epithelial-like cells expressed different sets of keratin-coding genes when compared to those of the dental pulp (Figure S10, Table S4). In the periodontium, keratin-coding genes such as Krt14, Krt17, and Krt19 were not exclusively expressed by epithelial cells but also significantly enriched in fibroblasts and ScCs (Figure S10 and Table S4). The periodontal epithelial-like cells also expressed genes encoding for signaling molecules such as FDCSP (follicular dendritic cell-secreted protein) and WNT10A (Figures 2D and S7). We also found that in the periodontium, the MSCs expressed significantly higher levels of collagen-coding genes (e.g., COL1A1, COL3A1, COL6A1; Figure S8A). We further estimated the pairwise extended Jaccard similarity for all cell types present in the periodontium and dental pulp and ranked these pairwise similarities. This analysis revealed that the three most similar cell types between the periodontium and the dental pulp were, in order, endothelial cells, erythrocytes, and MSCs (Figure 4C and Table S2).

We analyzed the overall dynamics and differentiation trajectories of dental pulp and periodontal MSCs by velocity (Figure S12). We did not identify major differentiation trajectories between different cell types neither in the dental pulp nor in the periodontium. In the dental pulp, endothelial cells showed the most dynamic behavior, while only minor differentiation trajectories were identified within most dental pulp cell populations (Figure S12). In the periodontium, epithelial-like cells, fibroblasts, and MSCs displayed dynamic behaviors (Figure S12). Periodontal MSCs showed a directional gene expression trajectory from the MYH11⁺ to the THY1⁺ sub-cluster (Figure S12). Periodontal THY1⁺ MSCs co-expressed genes that characterize the fibroblastic compartment, such as the collagen-coding genes, MMP14, and SPARC (Figures S8 and S11). MYH11⁺ cells might thus constitute the most undifferentiated pool of MSCs within the periodontal tissue, while the THY1⁺ sub-cluster could represent MSCs directed toward the fibroblastic fate.

Discussion

Understanding the fine composition of human organs is of paramount importance to develop regenerative therapies. In particular, unraveling the composition of stem cell populations and their niches is fundamental to drive regenerative processes toward the reconstitution of fully functional tissues and organs. This study revealed that MSCs in the human dental pulp and periodontium are characterized by the expression of FRZB, NOTCH3, THY1, and MYH11. Frzb has already been shown to mark periodontal ligament cells from very early developmental stages (Mitsiadis et al., 2017b), while its expression in the dental pulp has not yet been reported. Previous studies have also shown that Notch3 is expressed in perivascular MSCs both in dental and non-dental tissues (Jamal et al., 2015; Lovschall et al., 2007; Wang et al., 2014). Both dental pulp and periodontium MSCs can be subdivided into subpopulations characterized by the expression of the same specific markers MYH11, THY1, and CCL2. THY1/CD90 is a general marker of human mesenchymal stem cells, and it is vastly used to sort human dental pulp stem cells (Dominici et al., 2006; Ledesma-Martinez et al., 2016; Sharpe, 2016). MYH11 is mostly known to be expressed in smooth muscle cells, and in our data sets, it was generally co-expressed with ACTA2 (α-smooth muscle actin), recently found to play an important role in MSC cell fate specification (Talele et al., 2015). CCL2 codes for the chemokine ligand 2, and its expression in MSCs was shown to be a key mediator of their immunomodulatory properties (Giri et al., 2020). Expression of these markers in the dental pulp stem cells is in accordance with the data sets reported in a recent work (Krivanek et al., 2020), while the existence of three distinct MSC subclusters, both in the dental pulp and periodontium, was not reported before. Beyond the expression of these markers, dental MSCs show an overall striking homogeneity, in contrast to current assumptions (Hakki et al., 2015; Lei et al., 2014; Otabe et al., 2012). Indeed, previous studies have shown that although dental pulp and periodontal stem cells possess similar differentiation potentials in generating adipoblasts, myoblasts, chondroblasts, and neurons, their efficacies in forming bone tissues differ (Bai et al., 2010; d'Aquino et al., 2011; Schiraldi et al., 2012; Yagyuu et al., 2010). Human dental pulp and periodontal stem cells do not differ in their specific migratory behavior when cultured separately in vitro (Schiraldi et al., 2012). However, when these two cell types are co-cultured, the periodontal MSCs quickly spread and directionally migrate toward the dental pulp stem cells, which exhibit limited proliferative and migratory capabilities (Schiraldi et al., 2012). MSC proliferation and directional migration cues are generally produced by the target tissue, as well as by direct contacts established through the interactions of MSCs with cells composing their niches (Schiraldi et al., 2012; Shellard and Mayor, 2019). The divergent behavior of these MSCs, both in migration and in differentiation, could be due to their interaction with different environments rather than due to intrinsic differences. Our results support that dental MSC homogeneity is counteracted by a great divergence in their niches. In our samples, the dental pulp was composed mostly by fibroblasts, while epithelial cells constituted the most abundant cluster in the periodontium. Fibroblasts and epithelial cells within the dental pulp and the periodontium also expressed very different sets of molecules that could modulate MSC behavior. Genes coding for collagens and MMPs, as well as genes encoding for regulators of mineralization such as osteonectin, were highly expressed by periodontal fibroblasts and MSCs when compared to their dental pulp equivalent. Osteonectin is known to regulate Ca²⁺ deposition during bone formation, but in the periodontium, its function is essential for proper collagen turnover and organization (Luan et al., 2007). Periodontal fibroblasts also expressed MGP, a potent inhibitor of mineralization (Kaipatur et al., 2008). The most abundant periodontal cell type is represented by epithelial-like cells. These periodontal epithelial-like cells expressed genes encoding for signaling molecules such as FDCSP and WNT10A, which exert fundamental roles in the modulation of periodontal MSC proliferation and differentiation (Wei et al., 2011; Xu et al., 2017; Yu et al., 2020). Epithelial cells from the periodontium have been long proposed to constitute a dental epithelial stem cell population, with potential to generate tooth-associated hard tissues such as enamel, dentin, and alveolar bone (Athanassiou-Papaefthymiou et al., 2015; Tsunematsu et al., 2016). We showed that these cells also have signaling properties that could influence the behavior of periodontal MSCs. Overall, the cellular and molecular signature of the periodontium identified in this study was indicative of its continuous and dynamic remodeling, which is tightly linked to the masticatory function of the teeth, and that requires continuous collagen secretion, extracellular matrix remodeling, and inhibition of mineralization (Takimoto et al., 2015). Taken together, these significant cellular and molecular differences in the microenvironment of the dental pulp and periodontium constitute strong tissue-specific traits. These traits can be indicative of a microenvironment that privileges MSC differentiation toward a fibroblastic-like fate in the periodontium, as opposed to the dental pulp microenvironment, which favors the osteogenic fate of MSCs. Both dental pulp and periodontal MSCs derive from cranial neural crest cell populations, and this common origin provides a developmental basis for the observed similarities in gene expression patterns (Luan et al., 2009). Dental pulp and periodontal precursors display however divergent behaviors from very early developmental stages. Such differences were proposed to be induced from the interaction of similar neural crest cells with different microenvironments (Diekwisch, 2002; Luan et al., 2009; Svandova et al., 2020). These interactions would thus be the basis for the generation of tissues as diverse as the dental pulp, periodontium, and alveolar bone, from common neural crest-derived cell populations (Svandova et al., 2020). Subpopulations of periodontal MSCs indeed maintain for long time a highly migratory behavior, which has been hypothesized to depend as well on the peculiar periodontal microenvironment (Diekwisch, 2002; Luan et al., 2009). Microenvironmental cues would then result in the generation of different mesenchymal cell and stem cell populations via induction of vast epigenetic alterations (Gopinathan et al., 2019; Luan et al., 2009), thus modulating MSC behavior and determining their identity in the dental pulp and periodontium both during development and in adult life.

Two recent articles described the single-cell RNA sequencing analysis of dental tissues (Krivanek et al., 2020; Sharir et al., 2019). One study identified the main cell types that compose the dental pulp and compared their behavior in mice and humans and between human adult and erupting teeth (Krivanek et al., 2020). This work showed that basic features underlying tooth growth, such as lineage hierarchy between Smoc2^- and Smoc2⁺ cells, are conserved between mice and humans (Krivanek et al., 2020). The data sets concerning the human dental pulp presented in this work are in general agreement with our data. Our results provide a significantly more resolved analysis, in which we identified not only the major cell types present within the dental pulp and the periodontium but also their heterogeneity. In a second study, the authors performed single-cell RNA sequencing analysis of the epithelium of the continuously growing mouse incisor and revealed the role of Notch1-expressing stem cells showing that these cells are responsive to tooth injury and contribute to enamel regeneration (Sharir et al., 2019). Overall, these studies are complementary to our work, as they focused mostly on mouse teeth, while they did not investigate in detail the cell types that compose the human dental pulp and periodontium.

Taken together, our findings provide a thorough investigation of the human pulp and periodontal tissues at single-cell resolution, thus representing the basis for future research involving cell-based regenerative treatments.

Limitations of the study

This is the first complete single-cell atlas of human teeth that allows a comparative single-cell RNA analysis of human dental pulp and periodontium. In our data sets, we identified great variability between patients, which was particularly pronounced in the periodontium. The latter could be due to the highly dynamic nature of the periodontium (Luan et al., 2007) and to the peculiar experimental procedure needed to isolate periodontal cells, i.e., scraping them from the surface of the tooth roots. Since our atlas represents cells that survive experimental procedures, the number of odontoblasts in the dental pulp might be underestimated, due to possible damages induced to some of them during the extraction of the dental pulp from the tooth. With our analysis, we observed little differences between dental pulp and periodontal MSCs, which were counteracted by a great divergence in the composition of their niches. We hypothesized that such divergence could be the basis for the observed differences in the behavior of otherwise similar MSCs in the dental pulp and periodontium. This hypothesis requires nevertheless further experimental validation.

Resource availability

Lead contact

Information and requests for resources should be directed to the lead contact, Thimios A. Mitsiadis (thimios.mitsiadis@zzm.uzh.ch).

Materials availability

This study did not generate new unique reagents.

Data and code availability

The accession number for all sequencing data reported in this paper is GEO: GSE161267. All code is publicly available at: https://github.com/TheMoorLab/Tooth.

Methods

All methods can be found in the accompanying transparent methods supplemental file.

Published: May 21, 2021