Plap-1 lineage tracing and single-cell transcriptomics reveal cellular dynamics in the periodontal ligament

Mammalian teeth are supported by periodontal tissue, which consists of the periodontal ligament (PDL), gingiva, alveolar bone and cementum. As periodontal tissue is highly adaptive to normal and excess occlusal and orthodontic forces, it undergoes extensive remodeling (Beertsen et al., 1997). Periodontal tissue also possesses some regenerative capacity, as demonstrated in human and animal models (Sallum et al., 2019; Kitamura et al., 2016; Nagayasu-Tanaka et al., 2015). The plasticity of periodontal tissue is thought to be underpinned by the mesenchymal stem/progenitor population in the PDL (Zhao et al., 2020). This population presumably gives rise to three stromal lineages: periodontal fibroblasts, cementoblasts and osteoblasts (Luan et al., 2009; Iwayama et al., 2022). The PDL has also gained attention as a potent source of mesenchymal stem/stromal cells (MSCs) (Sharpe, 2016) or multipotent stromal cells (Soliman et al., 2021). The PDL contains various other cell types, including epithelial cell rests of Malassez (Xiong et al., 2013), hematopoietic cells (Wilson et al., 2018) and neurovascular cells (Men et al., 2020). However, evidence regarding the lineage characteristics and cellular hierarchies of the adult mammalian PDL in vivo remains limited.

One reason for this ambiguity is the lack of clear phenotypic markers that distinguish the PDL fibroblast lineage [periodontal ligament cells (PDLCs), including stem/progenitor cells and fibroblasts)] from other lineages of the periodontium. Based on recent lineage-tracing analyses, many mesenchymal stem/progenitor markers have been proposed, including GLI-Kruppel family member GLI1 (Gli1)-, alpha-smooth muscle actin (αSMA)- and Axin2-positive cells in the adult PDL (Roguljic et al., 2013; Yuan et al., 2018; Men et al., 2020). However, none of these markers is exclusively expressed in the PDL. Rather, they are also expressed in other components of the periodontium, as these markers are readouts of essential signaling pathways (sonic hedgehog, transforming growth factor β and Wnt signaling, respectively) with dynamic expression in various tissues. Another challenge in PDL biology is the lack of efficient methods for isolating cells from tissues. Owing to strong fiber attachments, dissociating the adult PDL is difficult (De Jong et al., 2017). Recent single-cell transcriptome analyses of the periodontium have revealed that isolated cells predominantly comprise epithelial cells (Caetano et al., 2021) and constitute only a small number of PDL-derived stromal cells (Krivanek et al., 2020; Pagella et al., 2021). In this regard, a method for isolation and evaluation is required to decipher the cellular hierarchy of the PDL.

The PDL expresses a large number of extracellular matrix (ECM) components, including collagen (Naveh et al., 2018) and proteoglycans (Chen et al., 2021). Among these, several genes are candidate PDL-specific genes (Takimoto et al., 2015; Horiuchi et al., 1999). We previously identified periodontal ligament associated protein-1 (PLAP-1; also known as ASPN) from a human cDNA library and demonstrated that it is highly expressed in the PDL at the transcript and protein levels (Yamada et al., 2001, 2007). PLAP-1 is a 43-kDa ECM protein that is a subtype of the small leucine-rich family of proteoglycans (SLRPs) (Kalamajski and Oldberg, 2010). PLAP-1 functions as a negative regulator of cytodifferentiation and mineralization of PDLCs through interactions with BMP2 and FGF2 (Yamada et al., 2006, Awata et al., 2005). Its transcript levels in the maxilla, including the PDL, have been reported to be more than tenfold greater than levels in adipose tissue, brain, heart, lung, liver, stomach, intestine, kidney, spleen, bone marrow and muscle (Sakashita et al., 2021). Compared with dissected gingiva, dissected PDLs contain 60 times more transcripts (Ueda et al., 2021). Nevertheless, the expression of PLAP-1 in the PDL at the single-cell resolution has yet to be investigated.

In this study, we identified mouse Plap-1 as a molecule that clearly distinguishes the PDL fibroblast lineage from other lineages, including osteoblasts and cementoblasts. Using knock-in mice expressing GFP and CreER^T2 under the endogenous transcriptional control of Plap-1, we demonstrated that PDL cells differentiated into osteoblasts and cementoblasts in vivo. Furthermore, we developed a method for highly efficient isolation of PDL-derived cells based on GFP expression in knock-in mice. A single-cell transcriptome atlas of the PDL revealed its cellular heterogeneity including Plap-1⁺Nes⁺ mural cells, and Schwann cells and other non-stromal cells. Platelet-derived growth factor receptor alpha (PDGFRα)⁺ stromal cells in PDL could be divided into Plap-1⁺ PDLCs and bone sialoprotein (Ibsp)⁺ cemento-/osteoblastic cells. Among Ibsp⁺ cells, SPARC-like 1 (Sparcl1) and collagen type XI alpha 2 (Col11a2) were exclusively expressed in cementoblasts and osteoblasts, respectively. RNA velocity analysis suggested that Plap-1^highLy6a (also known as Sca-1)⁺ cells were the source of adult PDLCs. Upon tissue injury, PDLCs expanded to repair the periodontal tissue. These results advance our understanding of the PDL and its mechanisms of regeneration in both healthy and diseased states.

The PDL is the ECM-rich connective tissue between the cementum and alveolar bone. Histological staining of this tissue revealed spindle-shaped fibroblastic cells aligned along the horizontal axis of the fiber bundles (Fig. 1A). To identify exact cell types according to gene expression profiles at single-cell resolution, we performed fluorescence in situ hybridization (FISH). We confirmed that both lineages expressed ECM components, such as collagen type I alpha 1 (Col1a1), biglycan (Bgn) and secreted protein acidic and rich in cysteine (Sparc) (Fig. 1B,C, Fig. S1A). Cementoblasts and osteoblasts were labeled with secreted phosphoprotein 1 (Spp1) and Ibsp (Fig. 1D,F) as previously reported (Foster et al., 2018). We then found that only fibroblast-like cells in the PDL in the horizontal direction were positive for Plap-1, whereas the layer of cells on the surface of the alveolar bone and cementum was negative for Plap-1 (Fig. 1E). FISH using both Plap-1 and Ibsp probes showed that these cells are mutually exclusive in PDL (Fig. 1F). As Plap-1 is a secreted protein, its protein distribution was widespread throughout the PDL and could be seen around osteoblasts and cementoblasts in addition to the Plap-1-producing PDLCs (Fig. 1E versus Fig. S1B). Having identified Plap-1 as a molecular marker that specifically labels the periodontal fibroblast lineage (including stem/progenitor cells), we defined the three previously proposed zones of the PDL (Lee et al., 2015) using Plap-1 and Ibsp (Fig. 1G).

Next, we devised a strategy for CreER-2A-GFP knock-in to faithfully reproduce Plap-1 expression in mice (Fig. S1C-F). GFP expression in Plap-1-GFP-2A-CreER mice was consistent with the FISH results, and only cells in the fibroblastic zones were labeled (Fig. 2A). Furthermore, using the tamoxifen-induced Cre expression system, genetic lineage tracing of Plap-1⁺ cells can be conducted with the knock-in mice (Fig. 2B). Short-term lineage tracing revealed that they were restricted to the fibroblastic zone 2 days after treatment with tamoxifen (Fig. 2C), as expected. The tight regulation of CreER was further confirmed by the absence of tdTomato expression without tamoxifen treatment (Fig. S1G) and limited tdTomato expression by a lower (one-tenth) dose tamoxifen treatment (Fig. S1H). The PDLCs were traced for a further 3 months, when fibroblastic turnover of the tissue was completed (Sodek, 1977). The PDL still consisted of tdTomato⁺ cells, suggesting that the PDL homeostasis was maintained by Plap-1⁺ stem/progenitor cells (Fig. 2D). TdTomato⁺ cells were also observed both in cemento-/osteoblastic zones at 3 months post-treatment (Fig. 2D), suggesting that Plap-1⁺ cells gave rise to cemento-/osteoblast lineage cells. This was also confirmed by the existence of Ibsp-expressing PDLC lineage cells in the middle and furcation areas of PDL (Fig. 2E). In addition, we investigated Plap-1 expression in other tissues using the knock-in mice (Fig. S2A-L) and confirmed the specificity of Plap-1 expression in the PDL as previously reported (Sakashita et al., 2021).

To decipher the cell types in the PDL that contribute to the homeostasis and repair of periodontal tissue, an efficient method for isolation and analysis of PDL cells is required. We previously established a method for isolating PDL-derived cells by scraping the root surface with a PDL explant (Ueda et al., 2021). However, this method can only be used to analyze migratory cells. In this study, we used GFP expression in Plap-1-GFP-2A-CreER mice to evaluate the efficacy of PDL cell isolation. We adopted our previous protocol in which the gingiva was removed from the maxilla before tooth extraction to avoid any contamination from the gingival cells to extracted molars (Ueda et al., 2021). We first digested the scraped PDL with enzymes. Nevertheless, we did not observe any clear cell populations in the forward scatter (FSC)/side scatter (SSC) plot (Fig. S3A). We then modified the protocol to digest extracted whole teeth (Fig. 3A) and observed a clear population in the FSC/SSC plot (Fig. S3A). Using flow cytometry and microscopy, we compared two enzymes, Liberase and Collagenase II. We identified a more evident population in the FSC/SSC plot with the use of Liberase (Fig. S3B). This was in agreement with the observation that the PDL was completely removed from the root surface with the use of Liberase (Fig. S3C). Histological analysis of the alveolar socket after tooth extraction showed that PDL tissue remained on the alveolar bone surface, but cells in the furcation area close to the alveolar crest were mostly removed (Fig. S3D). We further confirmed that Liberase treatment was superior in terms of the percentage and the absolute number of live cells (Fig. S3E). Efficient PDL detachment from the tooth surface with Liberase was confirmed histologically (Fig. 3B). The isolated cells were analyzed using flow cytometry, and a gating strategy was established according to GFP expression. GFP⁺ cells from live singlets were backgated onto the FSC/SSC plot, and the gating was defined as a PDL-enriched population (Fig. 3C). This population was then used to determine the relative proportion of each cell type. PDL cell-enriched populations yielded more than 10,000 cells per animal, of which 20%, 6% and 10% comprised CD45 (Ptprc)⁺ immune cells, CD31 (Pecam1)⁺ vascular endothelial cells and CD326 (Epcam)⁺ epithelial cells, respectively (Fig. 3D). In addition, we investigated whether a stromal cell marker, CD51, also known as integrin αV, can label stromal cells, including Plap-1⁺ cells. CD51⁺ cells included most GFP⁺ cells, with an average of 30% single-positive cells. GFP⁺ cells consistently accounted for approximately 30% of the population. We also tested the workflow using a Col1a1-GFP mouse line that expresses GFP in both fibroblastic and cemento-/osteoblastic cells in periodontal tissue (Fig. S3F). As expected, more GFP⁺ cells were observed in the Col1a1-GFP PDL than in the Plap-1-GFP PDL (Fig. S3G). A comparison of gingiva with the PDL revealed that Col1a1-GFP expression levels were higher in the PDL than in gingiva, consistent with a previous report that collagen turnover is more active in the PDL than in gingiva (Sodek, 1977). Additionally, we examined the culture conditions suitable for PDLCs and developed a method that enabled stable PDL cell expansion in vitro under hypoxic conditions (Fig. 3E). GFP⁺ cells exhibited significantly higher fibroblast colony-forming unit (CFU-F) formation compared with all cells from the PDL-enriched population (Fig. 3F) and demonstrated osteogenic differentiation capacity in vitro (Fig. 3G), confirming that PDLCs contain stem/progenitor cells. However, the cellular composition and heterogeneity of the PDL were not determined from this analysis.

After establishing an efficient PDLC isolation technique, we obtained cell suspensions from 20 adult wild-type mice. All viable cells in the PDL were collected by cell sorting and subjected to single-cell transcriptome analysis to generate a ‘PDL single-cell atlas’ (Fig. 4A,B, Fig. S4A). After the removal of low-quality cells, 17 clusters were identified from the PDL single-cell RNA-sequencing (scRNA-seq) data (7318 cells) (Fig. 4C). The cell types in each cluster were annotated using known cell marker genes (Fig. 4D). The most abundant populations were stromal cells expressing collagens, matrix metalloproteinases and platelet-derived growth factor receptors (Fig. 4D, Fig. S4B). Consistent with the histological analysis (Fig. 1F), collagen-expressing stromal cells were mutually exclusively defined by the expression of Plap-1 or Ibsp (Fig. 4E). The stromal cell cluster was subjected to in-depth analysis (Fig. 5). Non-stromal cell clusters, including immune, epithelial and vascular endothelial cells, were also identified (Fig. 4D, Fig. S4C,E). Flow cytometry analysis of isolated immune cells from PDL showed the presence of macrophages, proerythroblasts, B cells and T cells in the PDL and no overlap with Plap-1-GFP⁺ cells (Fig. S4D). Two epithelial cell clusters were distinguished by differential expression of keratin 15 (Krt15) and keratin 17 (Krt17). FISH analysis confirmed that Krt17-positive epithelial cells were derived from the inner and outer epithelia, whereas Krt15-positive epithelial cells were derived from the basement membrane (Fig. S4F). Notably, a cluster of Plap-1⁺ cells exhibited distinct gene expression in stromal cell clusters. This cluster expressed Nes, regulator of G-protein signaling 5 (Rgs5) and myosin 11 (Myh11) (Fig. 4E,F), which are markers of vascular mural cells (vascular smooth muscle cells and pericytes), suggesting that the PDL also contains mural cell populations that are less heterogeneous than the fibroblast lineage (Muhl et al., 2020). To identify Plap-1⁺Nes⁺ cell clusters in PDL histology, we used Nes-GFP transgenic mice, in which stem/progenitor populations are labeled in other tissues (Méndez-Ferrer et al., 2010; Iwayama et al., 2015). Nes-GFP⁺ cells exhibited a perivascular appearance, with long cellular processes along the blood vessels in the PDL (Fig. 4G). The cell bodies were covered by the basement membrane of vascular endothelial cells, suggestive of pericytes (Fig. 4H).

To identify the heterogeneity in the PDL that maintains tissue homeostasis, 3675 stromal cells expressing collagens, matrix metalloproteinases and platelet-derived growth factor receptors (Fig. S4B) were subjected to subclustering analysis. We identified 11 clusters (Fig. 5A) and confirmed that they were expressed either Plap-1 or Ibsp, except cluster 8 (Fig. 5B). Intriguingly, Scx, Mkx and Tnmd, which have been previously reported to be PDLC specific, were expressed only in a subset of both Plap-1⁺ and Ibsp⁺ cells (Fig. S5A). It should be noted that the master regulators of osteogenic differentiation, Runt-related transcription factor 2 (Runx2) and Sp7 transcription factor 7 (Sp7, also known as osterix), were also globally expressed in both lineages, suggesting that PDLCs are readily differentiated into osteoblasts and cementoblasts (Fig. S5B). This skewed differentiation is also supported by the expression of other bone-related proteins and SLRPs in PDLCs (Fig. 1B,C, Fig. S1A). Based on the representative gene expression profile (Fig. 5C), cluster 8 was annotated as Schwann cells, and S100 calcium-binding protein B (S100B)-expressing cells were identified in the PDL (Fig. S5C). In addition to the annotation of each cell type within stromal cells, we conducted RNA velocity analysis, which can predict the future state of individual cells by distinguishing between unspliced and spliced mRNA (Fig. 5D). The results suggested that the cells in cluster 9 are the source of differentiated cells in the PDL environment. This cluster showed higher expression and positive velocity of Plap-1 (Fig. S5D). In search of marker genes, lymphocyte activation protein 6a (Ly6a) and leucine rich repeat containing 15 (Lrrc15) were identified. The protein product of Ly6a, also known as stem cell antigen (Sca-1), has been used for identifying stem cells in many tissues in combination with another antigen, such as PDGFRα (Houlihan et al., 2012). The relative expression level of the Ly6a was higher in cluster 9 than in any other cluster (Fig. 5E), and we further characterized the population as Ly6a-positive Plap-1^high cells. Histological analysis revealed that they were located in the apical part of the PDL (Fig. 5F), and flow cytometry analysis showed they were 5.4% of the PDL live singlets (Fig. S5E). Lepr-, Gli1-, Axin2- and Acta2-positive cells have been implicated in stem/progenitor cells in the PDL. However, in this dataset, they were not associated with a specific cluster, with the exception of Lepr⁺ cells, which were enriched in cluster 5 (Fig. S5F). Crucially, none of these cells were observed at the top of the lineage hierarchy (Fig. 5D). Among Ibsp⁺ clusters, Col11a2 expression was specific to cluster 2 (Fig. 5G), and this gene was exclusively expressed in osteoblasts but not in cementoblasts (Fig. 5H). Compared with the osteoblast cluster, the other Ibsp⁺ cluster (cluster 6) expressed higher levels of Sparcl1 (Fig. 5I). Sparcl1 expression was observed in cementoblasts and cementocytes, but not in osteoblasts (Fig. 5J, Fig. S5G). However, the cluster did not specifically express recently reported markers (Nagata et al., 2021), including parathyroid hormone-related protein (Pthlh), class III β-tubulin (Tubb3) and Wnt inhibitory factor 1 (Wif1) (Fig. S5H). To assess the putative differentiation pathways, we also performed a pseudotime analysis of clusters 0, 2, 4 and 6. Gene expression kinetics of Plap-1, Ibsp, Col3a1, Col1a1, Sparcl1, Col11a2, Lrrc15 and Ly6a showed unique transient changes during differentiation (Fig. S6A-F).

To investigate the role of PDLCs in the repair of injured periodontal tissue, we utilized a ligature-induced periodontitis model (Fig. 6A). In this widely used model (Abe and Hajishengallis, 2013), a suture is placed around the second molar of the maxilla for 7 days to induce periodontal tissue loss. Upon ligature removal, the periodontal tissue can be repaired to the baseline level 7-14 days after removal. On day 1, after ligature removal, a layer of cementum on the surface remained intact, although alveolar bone resorption was observed (Fig. 6B, left). On day 7, recovered alveolar bone levels and fibrous attachment between the bone and cementum were observed (Fig. 6B, right). Plap-1 mRNA expression was restricted to the fibroblastic zone between the remaining alveolar bone and cementum (Fig. 6C), which was defined as the PDL. Notably, Ibsp expression was dramatically downregulated in the remaining cementum (Fig. 6D). To elucidate the contribution of Plap-1⁺ cells in the remaining PDL to tissue repair, Plap-1-GFP-2A-CreER; Rosa26-tdTomato mice were administered tamoxifen 1 day before ligature removal (Fig. 6E). In the early stages of periodontal wound healing, the granulation tissue beneath the gingival epithelial layer consisted of Plap-1 lineage cells and αSMA-positive myofibroblasts (Fig. 6F,G). Of the αSMA-stained cells in granulation tissue, 20.2±6.7% were Plap-1 lineage cells (n=6; mean±s.d.), which was 22.4±6.7% of the total Plap-1 lineage cells in the granulation tissue (n=6; mean±s.d.). These data suggest the partial contribution of Plap-1⁺ cells to granulation tissue formation. On day 7, in addition to most PDLCs in repaired PDL, 56.5% of osteoblasts and cementoblasts on the surface of repaired alveolar bone and cementum were derived from Plap-1⁺ lineage cells (Fig. 6H-J). These data confirm the long-standing concept that periodontal tissue is maintained and repaired by the PDL and emphasize the importance of the PDL.

Long-term lineage-tracing analysis with Nes-Cre revealed that Nes⁺ mural cells were quiescent and remained perivascular for up to a year without expansion during periodontal tissue homeostasis (Fig. S7A). Based on this observation, we performed a ligature experiment with Nes-Cre lineage tracing to determine whether Plap-1⁺Nes⁺ cells contribute to periodontal tissue repair (Fig. S7B). On day 7, after ligature removal, we did not detect the Nes lineage cells in repaired PDL, alveolar bone or cementum (Fig. S7C). These results suggest that Nes⁺ mural cells in the PDL do not contribute to the periodontal tissue repair in a cell-autonomous manner.

Periodontitis is a chronic inflammation of periodontal tissues caused by bacterial biofilms, which eventually cause irreversible destruction of periodontal tissues. Periodontitis has a high prevalence worldwide and is the most common cause of tooth loss in adults in developed countries. In addition to attaching teeth to the alveolar bone, the PDL interferes with and senses the occlusal force and transmits it to the central nervous system (Beertsen et al., 1997). Dental implants, however, lack these functions because they form a direct connection to the alveolar bone (osseointegration) without the PDL. Fibrous attachment of the PDL is one of the most clinically important indicators of periodontal disease and regeneration. It has long been proposed that the PDL contains stem/progenitor populations based on the fact that periodontal tissue can partially regenerate (Kitamura et al., 2016; Sallum et al., 2019) and has a large adaptive capacity for orthodontic treatment and occlusal forces in addition to remodeling capacity (Beertsen et al., 1997). PDL tissue attached to extracted human teeth has been analyzed extensively, and PDL stem cells (PDLSCs) with minimal stemness criteria (Mao et al., 2012), including in vitro differentiation capacity to osteoblasts and ectopic bone formation by transplantation assays, have been widely used (Li et al., 2021). However, as with MSCs in other tissues, the definition of PDLSCs is ambiguous, and rigorous in vivo analyses are required to define these cell types. In this study, we established a PDL single-cell atlas and demonstrated that the PDL comprises heterogeneous cell types. Notably, the stromal compartment was collectively divided into Plap-1⁺ fibroblastic cells or Ibsp⁺ osteo-/cementoblastic cells in a mutually exclusive manner. PLAP-1 is also highly expressed in the human PDL and used as a cultured PDLC marker (Yamada et al., 2001; Loo-Kirana et al., 2021; Mun et al., 2022). RNA velocity analysis indicated that cluster 9, Plap-1^high, with a positive velocity of Plap-1 mRNA, constituted a potential cellular source of other PDL clusters. This population expressed Ly6a, the protein product of which is known to be expressed in the stem cell population and in epithelial and endothelial cells. This Plap-1^highLy6a⁺ population could be PDLSCs that maintain and repair the entire periodontal tissue and overlap with the recently defined skeletal stem cell-like population in periodontal tissue (Liang et al., 2022). To clarify the cell fate of the putative PDLSC in vivo, a definitive single marker gene or an intersectional lineage-tracing approach will be needed in future studies.

In this study, we also harnessed single-cell analysis in combination with Nes-GFP and Nes-Cre transgenic mice to demonstrate the presence of strictly defined mural cells in the PDL for the first time. Although the association of pericytes with MSCs in other tissues has been described (Soliman et al., 2021), our findings suggest that Nes⁺ cells are quiescent mural cells near capillaries, and they did not directly contribute to PDL regeneration. Further study is needed to clarify whether they have an indirect contribution, including the release of paracrine signals and exosomes (Wagoner and Zhao, 2021). Our single-cell analysis revealed the presence of mesenchymal cells and many immune cells, especially macrophages and monocytes, in the PDL. The presence of CD45-positive immune cells has been reported (Soliman et al., 2021), but this remains inconclusive owing to the lack of efficient methods to isolate PDLCs and other cells from the PDL. Future studies should elucidate how pericytes and PDLCs repair tissues by interacting with immune cells at each stage of wound healing. Other cell types, such as plasma cells, which were not analyzed in this study (Fig. S4C), might be identified by specific isolation of immune cells and deeper sequencing.

Regeneration of the cementum-PDL-alveolar bone complex is a major goal of periodontal therapy. In contrast to bone regeneration, the mechanism of cementum regeneration remains unclear. Without regeneration of the cementum, new attachment formation by the PDL cannot be obtained even if the bone has regenerated. In the ligature model, Ibsp⁺ cementoblasts disappeared from the root surface of destroyed periodontal tissue. In vivo lineage tracing of PDLCs demonstrated that Ibsp⁺ cells reappeared in the regenerated tissue and more than half of Ibsp⁺ cells originated from the PDLCs (Fig. 6H-J). However, the exact cellular dynamics of the cementoblasts remain to be investigated with definitive markers. Recent analyses of cementum development in mice have highlighted several molecules that are specifically expressed in cementoblasts (Nagata et al., 2021). In this study, we defined cementoblasts as Plap-1⁻Ibsp⁺Sparcl1⁺ and alveolar osteoblasts as Plap-1⁻Ibsp⁺Col11a1⁺. Sparcl1 is an ECM protein related to Sparc and is secreted by nerves and blood vessels (Naschberger et al., 2016; Fan et al., 2021). In the future, single-cell epigenetic profiling of periodontal tissue may be harnessed to identify transcriptional regulators of cementoblasts.

In conclusion, our work identified Plap-1 as a PDLC-specific molecule, and cell-lineage analysis using knock-in mice confirmed that PDLCs differentiate into osteoblasts and cementoblasts and play a role in periodontal tissue homeostasis. In addition, we developed a novel cell isolation method and performed scRNA-seq to generate a PDL single-cell atlas that clearly defined osteoblasts and cementoblasts. Furthermore, lineage tracing revealed that injured periodontal tissues were repaired by PDLCs. These findings will contribute to the development of efficient regenerative therapy for periodontitis.

All animal experiments in this study were approved by the Animal Experimentation Committee of Osaka University Graduate School of Dentistry (approval numbers: R-02-012-0 and 31-001-0). Plap-1-GFP-2A-CreER mice were generated using a conventional gene-targeting strategy. R26-LSL-tdTomato (stock #007909) and Nes-Cre (stock #003771) were obtained from The Jackson Laboratory. Nes-GFP mice were provided by Dr Grigori Enikolopov of Stony Brook University, NY, USA. Col1a1-GFP mice were provided by Dr David Brenner, University of California, CA, USA. PCR genotyping was performed using the primers listed in Table S1. Mice were provided sterile food and water under specific pathogen-free conditions. All animal experiments complied with ARRIVE guidelines.

A targeting vector harboring the PGK-DTA-long homology arm-GFP-2A-CreER-WPRE-pA-short homology arm sequence was generated. Correctly targeted embryonic stem cell clones were identified by PCR and confirmed by Southern blotting using the primers and probes listed in Table S1. Chimeric males were generated using standard aggregation of clumped embryonic stem cells with eight-cell embryos and transplantation procedures and were used to establish germline transmission.

Mice mice were euthanized with CO₂ gas, perfused and fixed with 4% paraformaldehyde/phosphate buffer (PFA; Fujifilm Wako Pure Chemicals, 163-20145). Maxillary bones and other tissues were harvested and, after overnight immersion fixation in PFA, the hard tissue was demineralized in demineralizing solution B (Fujifilm Wako Pure Chemicals, #041-22031) for 3-5 days with three exchanges of solution. For paraffin sections, the tissue was embedded in paraffin and sectioned at 7 μm. Hematoxylin & Eosin (HE) staining was performed as previously described (Ueda et al., 2021). For frozen sections, the tissue was immersed in 20% sucrose (Fujifilm Wako Pure Chemicals, #196-00015) solution and then embedded in TissueTech O.C.T. compound (Sakura Finetech Japan, 4583), frozen in ethanol cooled on dry ice. Tissue sectioning at 14 μm was performed using a cryostat CM3050S (Leica Microsystems). A cryofilm [type 2C(9), Section-lab, Hiroshima, Japan] was used as section support.

The prepared frozen sections were washed with PBS and blocked with Blocking One Histo (Nacalai Tesque, 06349-64) for 30 min at room temperature. Sections were incubated with rabbit anti-mouse laminin polyclonal primary antibody (1:1000, Sigma-Aldrich, L9393) at 4°C overnight followed by incubation with Alexa Fluor 647-conjugated goat anti-rabbit IgG secondary antibody (1:200, Thermo Fisher Scientific, A32733) at room temperature for 30 min. For myofibroblast identification, sections were incubated with Alexa Fluor 647-conjugated anti-mouse αSMA monoclonal antibody (1:100, clone 1A4, Santa Cruz Biotechnology, sc-32251) at 4°C overnight. Nuclear and vascular staining was performed with DAPI (1:50,000, Sigma-Aldrich, D9542) and Alexa Fluor 647-conjugated Isolectin GS-IB4 (1:200, Thermo Fisher Scientific, I32450), respectively. Fluorescently labeled tissue sections were imaged using a fluorescence microscope (ECLIPSE Ti, Nikon) or Leica TCS SP8 confocal microscope (Leica Microsystems). For confocal microscopy, 10-14 z-stack images (total thickness of approximately 5 µm) were acquired, and 3D images were obtained using the maximum value projection method.

Cryosections were hydrated and hybridized (Advanced Cell Diagnostics) using the Multiplex Fluorescent Reagent v2 kit according to the manufacturer's instructions. Nuclei were counterstained with DAPI (1:50,000, Sigma-Aldrich, D9542), sealed with ProLong Diamond Antifade Mountant (Thermo Fisher Scientific, P36961), and observed under a Leica TCS SP8 confocal microscope, as described above. The probes used in this study are listed in Table S1.

Tamoxifen (Sigma-Aldrich, T5648) was dissolved in corn oil at a concentration of 20 mg/ml and mixed overnight. Mice (6-8 weeks old) received an intraperitoneal injection of 100 μl of tamoxifen (2 mg). In low-dose experiments, 10 µl of tamoxifen (0.2 mg) was administered intraperitoneally and analyzed after 16 days. To detect any leaky expression of tdTomato, 14-week-old Plap-1-GFP-2A-CreER; Rosa26-tdTomato mice without tamoxifen treatment were used.

To induce experimental periodontitis, a 5-0 silk ligature was tied around the left maxillary second molar of 8-week-old mice as described previously (Ueda et al., 2021). After 7 days, the ligature was removed. Tissue recovery was analyzed on day 7 after removal. The contralateral molar tooth in each mouse was left unligated as intact periodontal tissue.

After euthanization with CO₂ gas, 8- to 12-week-old mice were perfused with PBS, and the maxilla and mandible were separated. Gingival tissue was peeled and carefully removed. Subsequently, the molars were extracted and collected into PBS in a 12-well plate. These procedures were performed under a Leica S6 stereomicroscope (Leica Microsystems). The teeth were washed three times and transferred to fresh wells. Liberase (Merck, 05401119001) or collagenase type II (Worthington Biochemical Corporation, CLS-2) solution was used to digest PDL, as described in the Results section. Liberase powder was reconstituted with 2 ml of distilled water and stored at −20°C. The working solution was prepared in digestion buffer (PBS containing 2% fetal calf serum) at a final concentration of 2 Wünsch units/ml. The teeth were then incubated in 2 ml digestion buffer in 15 ml low-binding tubes (STEMFULL; Sumitomo Bakelite, MS-90150) for 20 min at 37°C with shaking at 200 rpm. Subsequently, 20 µl of 500 mM EDTA was added, and the teeth were further incubated for 10 min at 37°C and 200 rpm. After digestion, 10 ml of buffer was added, and cells were centrifuged at 400 g for 8 min at 4°C. The pellet was resuspended in 2 ml of buffer and passed through a 70-µm cell strainer (BMS, BC-AMS-17001). After centrifugation at 320 g for 5 min at 4°C, cells were finally resuspended in 400 µl of buffer and subjected to downstream analyses. In some experiments, PDL was scraped from the surface of the mesial root of the first molar using micro-instruments, including a micro-curette (0.5 mm) (Fine Science Tools). Scraped tissue was digested as described above.

The isolated cells were resuspended in FACS buffer (PBS containing 5% fetal bovine serum). The cell pellet was incubated for 30 min on ice with fluorescent protein-conjugated antibodies (Table S1). Flow cytometry analysis was performed on a FACSCanto II (BD) flow cytometer using FCS Express 7 (De Novo Software) or FlowJo v10.8 (BD Biosciences) software.

Cell sorting was performed using an SH800Z Cell Sorter (Sony Biotechnology) with 100-μm microfluidic sorting chips. The isolated cells were expanded and maintained at 37°C, 5% O₂ and 10% CO₂ with a MesenCult Expansion Kit (STEMCELL Technologies, ST-05513) according to the manufacturer's protocol. The culture medium contained MesenPure, L-glutamine (Wako, 073-05391), and penicillin-streptomycin (Wako, 168-23191). For the CFU-F assay, 500-1000 sorted cells were maintained in a 6-well plate. On day 14, the cells were fixed with 100% methanol and stained with Giemsa staining solution (Nacalai Tesque, 37114-64). Fibroblastic colonies with more than 20 cells were quantified as CFU-F.

To induce osteogenic differentiation, confluent cells were cultured in osteogenic medium comprising cell culture medium supplemented with 50 µg/ml l-ascorbic acid phosphate magnesium salt n-hydrate (Fujifilm Wako Pure Chemical, 013-19641) and 10 mM glycerol 2-phosphate disodium salt n-hydrate (Fujifilm Wako Pure Chemical, 193-02041). The medium was replaced with fresh medium every 3 days. For mineral staining, cells were fixed with cold 100% ethanol for 5 min and stained with 1.0% Alizarin Red S solution (pH adjusted to 6.4; Wako Pure Chemical Industries).

scRNA-seq was performed based on a previous report (Picelli et al., 2013; Ho et al., 2021) with the following modifications. Primer mix composed of 5 μl of lysis buffer comprising 3.1375 μl of Buffer EB (QIAGEN, 19086), 0.5 μl of 10 mM dNTP (GenScript, C01582), 0.05 μl of Phusion HF buffer [New England BioLabs (NEB), B0518S], 0.3125 μl of Proteinase K (Nacalai Tesque, 15679-64) and 1 μl of 1 μM barcoded oligo-dT primer (5′-ACGACGCTCTTCCGATCT[Barcode]NNNNNNNNTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVN-3′, where ‘N’ is any base and ‘V’ is either ‘A’, ‘C’ or ‘G’; IDT) was aliquoted into 384-well plates. The isolated PDL-derived cells were stained with SYTOX™ Blue dead cell stain (Thermo Fisher Scientific, S34857). Unstained cells were considered live cells. Live cells were sorted into plates using a BD FACSAria III instrument (BD Biosciences; 100-µm chip) in single-cell purity mode. The plates were immediately centrifuged (500 g for 5 s at room temperature) and frozen at −80°C. Plates were incubated at 50°C for 10 min and then at 80°C for 10 min. A volume of 5 μl of first-strand reaction mix containing 2 μl of 5× Superscript IV First-Strand Buffer (Thermo Fisher Scientific, 18090050B), 0.5 μl of 100 mM DTT (Thermo Fisher Scientific), 0.0125 μl of SuperScript IV reverse transcriptase (200 U/μl, Thermo Fisher Scientific, 18090200), 0.05 μl of SUPERase In RNase Inhibitor (Thermo Fisher Scientific, AM2696) and 2.4375 μl of water was aliquoted into each well. The plates were then incubated at 55°C for 10 min, and the reaction was inactivated by incubation at 80°C for 10 min. To remove unincorporated oligos, 2 μl of exonuclease I mix containing 0.0625 μl of exonuclease I (Thermo Fisher Scientific, EN0582), 1.2 μl of 10× reaction buffer and 0.7375 μl of water was added, and the mixture was incubated at 37°C for 20 min. Samples were pooled and purified using a DNA Clean & Concentrator Kit-100 (Zymo Research, D4030), concentrated using a DNA Clean & Concentrator Kit-5 (Zymo Research, D4014), and eluted in 12 μl of Buffer EB. Eluted cDNA was denatured at 95°C for 2 min and immediately placed on ice for 2 min. The cDNA was pre-amplified using the Accel-NGS 1S Plus DNA Library Kit (10096; Swift Biosciences). A volume of 10 μl of Adaptase reaction mix containing 1.25 μl of Buffer EB, 2 μl of Buffer G1, 2 μl of Reagent G2, 1.25 μl of Reagent G3, 0.5 of μl Enzyme G4, 0.5 μl of Enzyme G5 and 0.5 μl of Enzyme G6 was added, and the mixture was incubated at 37°C for 15 min and at 95°C for 2 min. A volume of 23.5 μl of Extension reaction mix containing 9.25 μl of Buffer EB, 1 μl of Reagent Y1, 3.5 μl of Reagent W2, 8.75 μl of Buffer W3 and 1 μl of Enzyme W4 was added, and the mixture was incubated at 98°C for 30 s, 63°C for 15 s and 68°C for 5 min. Amplified cDNA was purified using 26.1 µl of AMPure XP beads (Beckman Coulter Diagnostics, A63881) and eluted in 19.5 μl of Buffer EB. To amplify cDNA libraries, each well was mixed with 3 μl of 10 μM i5 primer (5′-AATGATACGGCGACCACCGAGATCTACAC[i5]ACA CTCTTTCCCTACACGACGCTCTTCCGATCT-3′; IDT), 2.5 μl of 12 μM D7 primer (CAAGCAGAAGACGGCATACGAGATCGAGTAATGTGACTGGAGTTCAGACGTGTGCTCTTCCGATC-3′; IDT) and 25 μM KAPA HiFi ReadyMix (KAPA Biosystems, KK2602). Amplification was performed using the following program: 98°C for 3 min; 14 cycles of 98°C for 20 s, 67°C for 15 s and 72°C for 2 min; and a final hold at 72°C for 5 min. Each well was purified using 30 μl of AMPure XP beads, eluted in 30 μl of Buffer EB and transferred to a new PCR tube. Each well was then purified using 18 μl of AMPure XP beads and eluted in 10 μl of buffer EB. Amplified cDNA (1 ng) was mixed with water to a total volume of 5 μl. The following reactions were performed using the Nextera XT DNA Library Prep Kit (Illumina, FC-131-1096). Each well was mixed with 10 μl of Nextera TD buffer and 5 μl of Amplicon Tagment enzyme and then incubated at 55°C for 5 min for tagmentation. After tagmentation, samples were mixed with 20 μl of DNA Binding Buffer (Zymo Research, D4004−1-L), purified using 32 μl of AMPure XP beads and eluted in 16 μl of Buffer EB. The eluted DNA was mixed with 2 μl of 10 μM i5 primer, 2 μl of 10 μM P7 primer (5′-CAAGCAGAAGACGGCATACGAGAT[i7]GTCTCGTGGGCTCGG-3′) and 20 μM NEBNext High-Fidelity 2× PCR Master Mix (NEB, #M0541L). Amplification was performed using the following program: 72°C for 3 min, 98°C for 30 s, eight cycles of 98°C for 10 s, 66°C for 30 s and 72°C for 1 min, and a final hold at 72°C for 5 min. After amplification, the samples were purified using 32 μl of AMPure XP beads and eluted in 10 μl of Buffer EB. Purified samples were quantified for concentration using the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Q32854), and size was measured using a High Sensitivity D5000 Reagent Kit (Agilent Technologies, 5067-5593) and High Sensitivity D5000 Screen Tape (Agilent Technologies, 5067-5592). Sequencing libraries were sequenced on a NextSeq2000 platform (Illumina). The read length was set to 20 (read 1), eight (i7), eight (i5) and 51 (read 2) bases.

BCL files were obtained from NextSeq (Illumina) and then demultiplexed and converted into FASTQ files using bcl2fastq2 (https://jp.support.illumina.com/downloads/bcl2fastq-conversion-software-v2-20.html). FASTQ reads were mapped onto mouse mm10 reference genomes using STARv2.7.6a with solo options (Dobin et al., 2013). The count tables from STAR were loaded as a Seurat object (Hao et al., 2021). Low-quality samples with extremely high (>3500) or low (<200) UMI counts or high percentages of mitochondrial genes (>10%) were removed. Downstream analyses, including normalization, scaling and clustering, were conducted using Seurat v4 functions following a standard analytic procedure. For sub-cluster analysis, stromal cells were extracted from whole cells, and normalization and scaling were performed. A phenograph was employed in the sub-cluster analysis to run more detailed clustering (Levine et al., 2015). Gene expression dynamics along pseudotime were calculated using Palantir Technologies software (Setty et al., 2019).

Spliced and unspliced transcripts were quantified with velocyto (La Manno et al., 2018) using BAM files from STARv2.7.6, with the ‘--soloFeatures Gene GeneFull SJ Velocyto’ option. The output loom files were combined with the Seurat object and then converted into an h5ad file using SeuratDisk (https://mojaveazure.github.io/seurat-disk/). Further RNA velocity analyses were performed using the Scanpy (Wolf et al., 2018) and scVelo (Bergen et al., 2020) packages.

Data are presented as mean±s.d. in Figs 3F, 6J and Fig. S6D or mean±s.e.m. in Fig. 3D, Figs S3E and S4D; otherwise, raw data values were plotted. Statistical analysis was performed using GraphPad Prism v9.3 (GraphPad Software; www.graphpad.com). Statistical significance was assumed at P<0.05. Quantification of tdTomato, αSMA and Ibsp mRNA was performed by counting positive cells in the targeted area (day 3: granulation tissue upon ligature was defined as the tissue between gingival epithelium and repaired alveolar crest; day 7: PDL attached to coronal half of the root) of histological sections. One field of view each from two sections from three individual mice was used for cell quantification.

We thank Ms Kotomi Aso and Ms Minako Mori for their technical assistance. The Nes-GFP mouse line was a gift from Dr Grigori Enikolopov of Stony Brook University. The Col1a1-GFP mouse line was a gift from Dr David Brenner of the University of California, San Diego. Confocal microscopy was performed at the Center for Frontier Oral Science (Osaka University Graduate School of Dentistry). Flow cytometry was performed at the Center for Medical Research and Education (Graduate School of Medicine, Osaka University). Figures were created using Biorender.com.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201203.reviewer-comments.pdf.