Skip to main content

Targeted Cell Recruitment During Organogenesis And Regeneration: Glia Makes The Tooth

Periodic Reporting for period 3 - STEMMING-FROM-NERVE (Targeted Cell Recruitment During Organogenesis And Regeneration: Glia Makes The Tooth)

Reporting period: 2018-08-01 to 2020-01-31

“Dental cell type atlas reveals new stem, intermediate progenitor and differentiated populations in self-renewing mouse incisor”

Mouse incisor is the most popular model system to study tooth development, growth, self-renewal, regeneration, hard matrix repair, stem cells and morphogenesis. However, the cellular composition of a tooth is poorly understood. The identity and developmental paths of stem cells are still enigmatic. Using a single cell transcriptomics approach in a combination with lineage tracing and functional studies, we generated an atlas of mature cell types, stem cells and transitory populations from the continuously growing mouse incisor. Our results revealed new subtypes of stem and progenitor cells providing insights into tooth maintenance and self-renewal processes. In the epithelial compartment, we revealed differentiation path of ameloblasts and validated the existence of novel Sox10+ and Fos+/Egr1+ stem-like populations. In the mesenchymal part, we discovered the differentiation trajectory of odontoblasts and pulp cells via a number of previously unknown stages. We provided the detailed knowledge of heterogeneity among regenerative perivascular cells as well as glial, smooth muscle and dental follicle populations. We showed the transition of glial cells into dental mesenchymal populatioins at gene expression level. The profiling of immune cells showed the presence of previously unknown macrophage subtypes and their site-specific spatial distribution conserved also in human teeth. The interactomics mapping suggested that CSF1 secreted by specific pulp cells is a key for hosting dental macrophages. The conditional knockout of Csf1 gene in the neural crest-derived pulp resulted not only in a loss of dental macrophages, but also in a failure of correct incisor development and shaping. Taken together, the unbiased analysis of dental cell types combined with predictions of communications between the cell populations resulted in new discoveries that proved catalyzing power of dental cell type atlas.

Summary of the context

Mammalian teeth are vital for feeding, fending and good quality of life. They are mineralized organs formed by the ectoderm of the first pharyngeal arch and ectomesenchyme of the neural crest. Sophisticated developmental interactions between these tissue types enable the construction of the morphogenetically complex dental structures with pre-defined shape and location. The epithelial part is responsible for the generation of enamel around the crown of the tooth, whereas the mesenchymal part produces the cells forming dentin and root cementum. Neural crest-derived cells also form the soft core, a pulp, which is mainly composed of connective tissue hosting vessels, nerves and immune cells (Zhang et al., 2005, Jussila and Thesleff, 2012; Balic and Thesleff, 2015; (Kapsimali, 2017; Krivanek et al., 2017). An adult mouse incisor is the major model system to study dental stem cell populations because, unlike the other teeth, this type of a tooth grows continuously throughout the entire life of an animal. Thus, the incisor undergoes constant renewal of all cell populations from the apical end to replenish both soft and mineralized tissue that is getting lost due to gnawing and biting at the incisal tip. This allows explorations of stem cell generation, cell differentiation, homeostasis, age-related apoptosis and injury-induced regeneration in the same organ. Despite the major cell types were identified in teeth long time ago, the question about heterogeneity and molecular characterization of rare cell subpopulations, stem cells and their developmental transitions stayed unanswered. Several recent studies expanded our knowledge about the molecular identities of some dental stem cells and their properties in vivo (Kaukua et al, 2014; Balic and Thesleff, 2015; Li et al., 2016; Sharpe, 2016)(Seidel et al., 2017). However, a number of recent publications suggested the existence of multiple types of dental stem cells waiting to be resolved including glia-derived mesenchymal stem cells.
High throughput single cell RNA-sequencing technology makes it feasible to analyze large numbers of individual cells simultaneously and classify them in an unbiased way (Hashimshony et al, 2012; Islam et al., 2014, Sandberg, 2014). We took advantage of this technique in combination with different validation strategies to generate an atlas of dental cell types and to produce multigenic signatures outlining differentiation trajectory from stem to mature cellular elements in epithelial and mesenchymal compartments. This resource further enables in-depth investigations of signaling mechanisms that integrate tooth self-renewal, homeostasis and protection against infection or physical damage. The knowledge of gene modules controlling stem and other cell identities will enable future tooth regrowth and regeneration strategies.

This is important for the society because the knowledge of tissue heterogeneity, interactions and inter-conversions between cell types is the key for the bright future of regenerative medicine. In this particular case, we created an extensive knowledge of all cell types inhabiting teeth and participating in self-renewal of this important organ. This knowledge is very important for developments in regenerative dentistry and enhaced dental hard matrix repair technologies especially given that we discovered new previously unknown cell types not seen in the tooth before. Also, we think this is a very interesting topics, and many curious people want to know what our teeth are made of.

Overall objectives are:

- To address how, when and to what extent peripheral glial cells residing in the sensory nerve are recruited to produce cells of pulp and odontoblast lineages in the tooth.

- To discover tooth initiating properties of peripheral glial cells and to explore their potential for regenerative odontology and medicine.

- To understand the contribution of peripheral glial cells to the odontoblast lineage in adult tooth under normal physiological conditions and during trauma-induced regeneration.

- To identify extrinsic and intrinsic molecular mechanisms controlling the transition from glia to odontoblast lineage.

We addressed the problem of a context for glial recruitment into mesenchymal populations. Thus, we unbisedly resolved dental tissue cell type heterogeneity and identified new types of stem cells including those related to nerve-associated glial cells according to the focus of ERC proposal. Mostly, we achieved a great progress in the last objective (solved it at individual cell level with single cell transcriptomics):

- To identify extrinsic and intrinsic molecular mechanisms controlling the transition from glia to odontoblast lineage.

Overview of the dental cell type atlas

We isolated dental pulp along the whole length of the adult mouse incisor together with the entire cervical loop area. Then we dissociated these tissues into single cells, which we individually dispensed using FACS into 384-well plates for the following sequencing of their transcriptomes with Smartseq2 protocol (reference). To enrich for epithelial stem cells and their immediate progeny, we took advantage of Sox2-GFP animals since Sox2 is expressed by epithelial stem cells as previously shown (reference).
Unsupervised clustering of 3054 cells that passed quality control (Supplementary Figure 1, Methods) revealed 17 major clusters that correspond to defined cell types, including ameloblasts, odontoblasts, pulp cells, endothelial cells, pericytes, smooth muscle cells of arterial walls, various immune cells and also elements of extra-dental tissues (Figure 1A-D; Figure S1A,B,D). We generated transcriptional signatures for every identified population, thus, providing extensive lists of new markers for all known and unknown cell subtypes (Supplementary data table X, Supplementary Figure 1). Large fraction of proliferating cells, that were identified using cell cycle signature, was associated with the specific sub-clusters inside of the epithelial and pulp populations pointing to the actively operating developmental processes (Figure 2A, 3B, Methods). We next separately re-analyzed every major population including epithelial, mesenchymal and immune cells (Figure 1E,F,G).

Differentiation dynamics and new cell subtypes in the epithelial compartment

The canonical knowledge states that epithelial compartment in continuously growing mouse incisor is represented by dental epithelial stem cells (DESCs), transiently amplifying cells (TACs) residing in the stellate reticulum; outer enamel epithelium, stratum intermedium and, finally, inner enamel epithelium formed by ameloblasts secreting enamel (reference). Separate analysis of the epithelial compartment identified 13 subpopulations that reflect listed cell types as well as the novel subpopulations (Figure 2A).
One of such novel subpopulations formed cluster 7 and expressed Egr1 and Fos transcription factors. Using immunohistochemistry for EGR1, we found positive cells residing inside of the cervical loop stellate reticulum, where the DESCs and TACs are dwelling (Figure 2B). The lineage tracing with FosCreERT2/R26ZsGreen1 demonstrated the presence of the epithelial progeny after 10 days of tracing predominantly inside of the cervical loop and in the outer enamel epithelium. Occasionally, we observed traced cells in the stratum intermedium and among the ameloblasts (Figure 2B). Thus, Egr1+/Fos+ epithelial cells (also co-expressing Fbn2, Ctgf, Pgf, Vrtn, etc.) are novel fate-restricted stem-like progenitors. The population of Egr1+/Fos+ cells appeared to be Sox2 negative, which opens up for the existence of Sox2 negative epithelial progenitor subtypes. At the same time, Sox2 demonstrated the broad expression in other stem and progenitor subtypes in the epithelial dataset (Figure SX). Most of the Sox2+ cells coincided with proliferative markers (Mki67, Pcna, Mcm2, etc.) and transited into Shh+ cells (clusters 2, 5, 11, 12 and 13 corresponding to stellate reticulum, TACs of stellate reticulum, preameloblasts, Lrig1+ TACs and a part of secretory ameloblasts) (Figure S2).
The analysis of expressed transcription factors showed the presence of few Sox10+ cells that did not form a cluster. However, one of these rare cells demonstrated extensive and unique signature (including Ebf1, Jph2, Sema3g, P2rx1, and other genes). To address whether any of the rare Sox10+ epithelial cells could possess stem cell properties, we took advantage of Sox10CreERT2/R26Confetti mouse line (reference) to perform lineage tracing. The results confirmed the existence of long-lasting Sox10+ stem cells and showed all types of the derived epithelial progeny including enamel-producing ameloblasts after 10 weeks of tracing (Figure 2C).
Other cells located in DESCs and TACs part of the dental epithelial tissue (stellate reticulum) formed three clusters including cluster 2 defined by Vat1l, Fam19a4, Fam19a1, Slc27a6, Hey2, Gcnt1; cluster 12 defined by Cdh6, Lrp11, Cpne5, Rrm1, Mcn4, Zfp367 and cluster 13 defined by Lrig1, Grp, Sfrp5, Rhoc, Ddit4l, Grs11, Tenc, Kcnip3. These clusters mostly included Sox2+/Mki67+ TA cells. Cluster 2 appeared proximal to the differentiating ameloblasts on the tSNE, cluster 12 - to stratum intermedium and cluster 13 - to outer enamel epithelium (Figure 1A).
Knowledge of the trajectory of ameloblasts differentiation containing all genes switching during ameloblasts formation process will be essential for the in vitro and in vivo attempts to regenerate enamel via inducing enamel-producing cells. Therefore, we analyzed the gene expression during the genesis of progenitor and mature ameloblasts populations (Figure 3D). The differentiation of ameloblasts revealed several distinct stages including transitions in pre-ameloblasts (cluster 11: Shh, Arrb1, Vwde, Vwa2, Fzd3, etc.), secretory (cluster 5: Enam, Amelx, Ckb, Chst1, Sox21, etc.), maturation (cluster 10: Klk4, Amtn, Gpr155, Slc1a1, Gad1, etc.) and protection (cluster 6: Gm17660, Dsg3, Ptpn22, Flt3l, Nat8l, etc.) phases. Also, we found previously unknown population of RYR2+ ameloblasts (cluster 3) located among the ameloblasts between the secretory and maturation phase and characterized by the multigenic signature based on Ryr2, Piezo2, Nrip3, Stab2, Selp, Adgb and other genes. The potential function inferred from the individual signature of this cluster (3) might be related to extracellular matrix turnover (Stab2), cation metabolism (Ryr1, Ryr2, Cngb1, Nrcam, etc.) and cell adhesion (Pcdh17, Fn1, Nrcam, etc.). Additionally, the presence of expressed mechanosensory genes Piezo2 together with Trpm2, Trpm3, Trpm6 and other cation channels and calcium-dependent genes (Itpr1, Ryr1, Ryr2 etc.) suggests the involvement in mechanosensation in the epithelial compartment (Figure 3E).
Each identified subpopulation of ameloblasts is specified by the expression of specific transcription factors (showed in Supplementary Figure X), e.g. for pre-ameloblasts: Camk4, Ank2; ameloblasts in a secretory phase: Zfp9, Sox21; ameloblasts in a maturation phase: Etv3, Creb5, Dbx2.
All non-ameloblastic epithelial populations were Cldn10+ according to the clustering and immunohistochemistry-based validation in the dental epithelial tissue (Figure SX). These were specified by the different sets of transcription factors including Heyl, Nrarp and Atf5 in stratum intermedium (clusters 8: Rab3il1, Pmch, Cyp2s1, Hlf, Clec11a, Fbxo44 and cluster 9: Psmb10, C1qb, Ibsp, Gnaz, Akap6) and Pthlh in outer enamel epithelium (cluster 4: Slco4a1, Th, Amer1, Trpm3, Pi15, Bdh2). To validate these results, we picked the predictions of expression of Acta2 expression and confirmed the protein product by the immunohistochemistry in the outer enamel epithelium, on the transition between stellate reticulum and enamel epithelim confirming its origin in the labial cervical loop region. In line with this, the lineage tracing with Acta2CreERT2/Ai9 mouse line revealed the presence of numerous traced cells in the mature outer enamel epithelium of the incisor (Figure SX).
One of the previously unknown populations (cluster 1) demonstrated a distinct multigenic signature including expression of Thbd, Gnrh1, Jph4, Nphs1, Chst2, Car9, Grtp1 and other genes. The immunohistochemistry revealed that these specific cells reside in what we called “cuboid layer” of stratum intermedium (Figure 3F). The gene expression signature revealed that these cells may perform functions important for the oxygen and ion transportation (e.g. Cygb, Nphs) between blood vessels and highly active ameloblasts at this developmental stage. Kirrel2, Jph4 and other structural genes could be important for the formation of junctional complexes, whereas Gnrh1, Paqr5 could play a supportive role in ameloblast differentiation and altogether might explain the specific localization of this new epithelial cell subtype.
Finally, we generated an interactomics map predicting the ligand-receptor interactions in the epithelial compartment of continuously growing incisor based on transcriptomics profiles of all identified subpopulations (Figure SX). For instance, this analysis revealed that ameloblasts in different stages produce various groups of ligands including Wnts (Wnt10a – cluster 5 secretory stage, cluster 11 – pre-ameloblasts; Wnt4 - cluster 6 protection and cluster 5 secretory phase; Wnt3 - cluster 11 pre-ameloblasts and cluster 5 secretory stage; Wnt6 - cluster 5 secretory stage). The generation of Wnts by the developing ameloblasts fits to the fact that dental epithelial stem cells express Lgr5 and are Wnt-sensitive according to the published data (Suomalainen M, Thesleff I. Dev Dyn 2010; Yang Z, Balic A, Michon F, Juuri E, Thesleff I. Stem Cells 2015). More examples include progenitor cells of outer enamel epithelium (cluster 7) producing Fgf1 that can potentially bind to Fgfr2 expressed in the outer enamel epithelium clusters 4 and 13 including progenitor TACs. Other FGF ligands are expressed in stellate reticulum (cluster 2) and in pre-odontoblasts (Fgf3, Fgf10) that might interact with the epithelium via these pathways. Thus, interactomics reliably corroborates the known data and provides further deep and detailed insight into the possible existence of previously unanticipated interaction between multiple cells types, including cross talks between dental mesenchyme and epithelium that are at the core of dental induction and morphogenesis.
The odontoblast specification from the bi-potent dental mesenchymal stem cells has always been enigmatic. Prediction of the epithelium-derived signals can explain how mesenchyme-derived odontoblast fate could be specified under the influence of the epithelium. According to our results, pre-odontoblasts express several receptors that can be available for the interplay with the epithelial compartment. For instance, they can potentially receive signals from pre-ameloblasts expressing Pdgfb and Mfge8 via Pdgfra and Pdgfrb receptors. Other inferred pathways include, as an example, pre-odontoblast-specific receptor Fgfr1 that can bind Fgf1 expressed by outer enamel epithelium progenitors or Fgf9 expressed by stratum reticulum and pre-ameloblasts (for the complete odontoblast-epithelial interactomics resource, please see Supplementary Figures X and Y).

Cell dynamics and cell subtypes in the mesenchymal compartment

As shown in our analysis in Figure 3, mesenchymal compartment turned out to be highly heterogeneous consisting of previously defined populations (general pulp and odontoblasts) and previously unknown mature subtypes of pulp cells and distinct progenitor states. The heterogeneity of pulp cells was previously investigated with various methods of gene expression analysis, including in situ hybridizations and mouse reporter lines (cite Irma and other people). Beyond proving the existence of known subpopulations and providing expanded signatures for them, we detected distinct groups of cells with yet unclear functional meaning. The discovered heterogeneity factors behind these populations correspond to the processes of continuous development of odontoblasts and pulp cells from dental mesenchymal stem cells (DMSCs) via TAC stage, regionalization of the pulp, maturation, apoptosis, interaction with spatially segregated neuro-vascular bundle (Honza, provide the references for every part of this statement).
The reconstruction of differentiation trajectory revealed three major branches corresponding to the odontoblast differentiation with specific markers including Sall1, Notum, Dkk1, Wisp1, Col24a2, Slc8a3, Dspp, Dmp1 and two separate direction of pulp differentiation generally outlined by Fgf10, Ccnd1, Bhle41, Sfrp2 and Igfbp5, Syt6, Irx5, Epas1 genes (Figure 3A,B,E,F).
One the pulp branches corresponded to the regionalized pulp around the neuro-vascular bundle (Igfbp5, Ildr2, Epas1, Tnfaip2, Irx5, Syt6, etc.) and another branch was outlining the peri-odontoblastic pulp (Fgf10, Ccnd1, Mfap4, Bhlhe41, Smoc2, Sfrp2, etc.) next to the cervical loops (Figure 3A,B,C). The pulp branches showed a transition bridge, where the populations could converge in a more mature state. The previously unknown populations of the pulp cells included transitory stages in pulp differentiation within both branches.
The differentiation trajectory analysis placed the fate split at the stage of TAC population. This fate split position demarcated the level on the trajectory where the gene expression programs started to diverge involving hundreds of genes relevant to the effector programs. The clonal lineage tracing experiments starting from the individual dental mesenchymal stem cells previously showed that a single mesenchymal stem cell can give rise to both pulp cells and odontoblasts (cite Kaukua at al., Nature 2014). This is consistent with bioinformatics analysis of trajectory that showed that the area of the pulp-odontoblast fate split contains highly heterogeneous populations of cells including TAC progenitors (outlined by the expression of Fgf3, Wnt10a, Etv4, Nmnat2, Smpd3, etc.). FoxD1 and other genes (Mdga1, Ecel1, etc.) expressed in this location on the trajectory, appear to demarcate potential dental mesenchymal stem cell populations coinciding in expression pattern with previously established markers for DMSCs such as Thy1 (CD90) (Figure 3C).
Early activation of a specific set of regulatory genes (Sall1, Gsc, Dkk1, Etv5, etc.) suggested minimal transcription factor code of pre-odontoblasts commitment and differentiation. Next, we bioinformatically isolated the cells forming the odontoblast branch and re-clustered them to perform the differentiation trajectory analysis. We observed fast commitment to the odontoblast fate already at the level of TACs (Figure 3E,F). In the end of the differentiation trajectory we observed the expression of classical odontoblast markers such as DMP1 and DSPP (Figure 3F). The TACs showing differentiation towards odontoblasts expressed numerous receptor genes that can potentially mediate their commitment towards the odontoblast lineage. For instance, pre-odontoblasts and peri-epithelial mesenchymal TACs can potentially receive signals from forming pre-ameloblasts (pair: Jag-Notch, Artn-Gfra1), stratum intermedium (pairs: Efnb2-Epha3, Efnb2-Rhbdl2) or even outer enamel epithelium (pairs: Efnb3-Epha44, Efnb33-Rhbdl22). In turn, differentiating odontoblasts can potentially send diffusing signals to stratum intermedium or outer enamel epithelium (pairs: FGF3-FGFR2, FGF3-FGFR3). They can also potentially communicate with different pulp cells forming peri-odontoblastic pulp and other cell types locally available (see Supplementary Figure X for interactomics mapping).
As we previously demonstrated (cite Kaukua at al., Nature 2014), during incisor development and growth, some peripheral glial cells can undergo a conversion into DMSCs. To accommodate for the rare events of such conversions we utilized lineage tracing using PLP1CreERT2/R26YFP mice with subsequent fluorescent sorting of individual cells to enrich the sequenced population for glia-derived, including transitory, cell types (Figure 4A,C). The population analysis of individual transcriptomes revealed 3 major clusters corresponding to glial, pulp and pre-odontoblasts populations that emerged among all PLP1CreERT2/R26YFP-traced cells (Figure 4E,G). The glial cluster appeared stretched towards the mesenchymal populations that included both pulp and pre-odontoblasts according to the bioinformatics predictions and lineage tracing experiments with PLP1CreERT2/R26YFP mouse line (Figure 4G).
In order to perturb the signaling between the nerve a
According to DoA and main objectives of this project, within 2.5 years of work we achieved the progress beyond the state of the art by creating the first of its kind atlas of all cell types inhabiting the tooth with the power of single cell transcriptomics to resolve the context and the process of glial-to-mesenchymal transition. Using this atlas, we demosntrated the transitions from glial into dental mesenchymal cells and odontoblasts, which is the key knowledge for the further development of the project. Additionally, we discovered novel, previously unknown subtypes of cells in all dental compartments.We are about to submit this manuscript to Nature.
We hope to expand the atlas by the end of the project and also to use the gained knowledge to enhance to transformation of glial cells into odontoblasts inside mouse incisors and molars.