Mandibular denervation impairs the healing of bone injury

To explore the response of Plp1-lineage cells to bone injury, we utilized the tooth extraction socket (TES) model in Plp1-creERT2; tdTomato reporter mice, where Plp1+ cells were marked with tdTomato fluorescence. In the non-injury regions of alveolar bones and inferior alveolar nerve (IAN) (Fig. 1b), tdTomato+ cells stained positive for SCs markers (Plp1 and Sox10) but not for endothelial cell marker CD31 or sensory neuron marker CGRP, confirming the specificity and efficiency of our lineage tracing (Fig. 1a). Noticeably, we also found that a small population of tdTomato+Sox10− cells were Gli1 positive (Fig. 1a, a2). With the increased bone formation and maturation from day 1 to day 28, tdTomato+ cells keep expanding from periodontal ligament (PDL) into the sockets and the quantification of cells density supported these observations (n = 3 for each time point) (Fig. S1A, B). However, no obvious tdTomato signal was observed in the healing TES from NC group (Plp1-creERT2; tdTomato mice without tamoxifen pre-treatment) (Fig. S1A).

Fig. 1

Proximal mandibular denervation impairs the healing of bone injury. a Representative images of Plp1, CD31, CGRP Sox10, and Gli1 immunostaining and tdTomato+ cells to evaluate the efficiency and specificity of recombination of the Plp1-creERT2 alleles. Scale bar: 100 μm. Ab alveolar bone, PDL periodontal ligament. b Representative images of Plp1 and CGRP immunostaining and tdTomato+ cells in IAN. Scale bar: 100 μm. IAN inferior alveolar nerve. c Schematic illustration of the lineage tracing approach and proximal mandibular denervation in the jaw bone healing model. d Representative images of CGRP immunostaining and tdTomato+ cells in healing sockets of Sham and Denervation mice at day 7 post tooth extraction and quantification of tdTomato+ cells in TES (n = 6). Dotted lines indicate the tooth sockets. TES: tooth extraction socket. Scale bar: 100 μm. e H&E staining of tooth sockets from Sham and Denervation mice at day 7 and day 14 post tooth extraction (n = 6). Dotted lines indicate the tooth sockets. Scale bar: 100 μm. f Dynamic histomorphometry of trabecular bone (Tb) with quantification of mineralization apposition rate (MAR) in tooth extracted sockets (n = 6). Sham and Denervation mice were injected with calcein 7 and 2 days before sacrifice, respectively. Scale bar: 5 μm. g Representative images of μCT reconstruction of the alveolar bone regeneration at day 7 post tooth extraction and quantitative analyses on bone volume/total volume (BV/TV) and bone mineral density (BMD) (n = 6). h Representative images of COL1 immunostaining and tdTomato+ cells and quantification of COL1+ areas per socket (n = 6). Scale bar: 100 μm. i Representative images of CD31 and EMCN immunostaining and quantification of CD31+EMCN+ type H vessels in TES (n = 6). Scale bar: 100 μm. Data were presented as mean ± SD; **P < 0.01, ***P < 0.001

To assess the impact of Plp1+ SCs invasion on bone healing, we employed a classical model of surgical denervation.21 TES model is a classic model for studying mandibular injury repair. Given the rich periosteal nerves and Plp1+ SCs distributed in the alveolar bone and the PDL surrounding the root, tooth extractions were performed with or without ipsilateral IAN transection (Fig. 1c). Mechanical hypoalgesia in denervated mice was confirmed through von Frey testing (P < 0.01, n = 6) (Fig. S1E) and the images of bone trabecula (Fig. S1C) and IAN (Fig. S1F) in mandible illustrated the degeneration of tdTomato+ cells in denervated mice (P < 0.001, n = 6). CD31 immunostaining results in non-injury regions revealed no significant differences, ensuring the removal of the nerve supply does not impact blood supply within mandibles (P > 0.05, n = 6) (Fig. S1D). The denervation significantly reduced tdTomato+ cell density (Fig. 1d) and bone volume at both day 7 and day 14 post-extraction, as shown by micro-CT scans and histological analysis (P < 0.01, n = 6) (Figs. 1e, g and S1G). Delayed bone formation in denervated mice was further evidenced by double-labeling analysis (P < 0.01, n = 6) (Fig. 1f). At day 7 post-injury, a reduced expression of type 1 collagen (COL1) was observed in the denervation group (P < 0.01, n = 6), with a subset also being tdTomato positive (Fig. 1h). Additionally, denervated mice displayed a lower density of type H vessels (P < 0.01, n = 6), characterized by high co-expression of CD31 and Endomucin (EMCN) (Fig. 1i). These findings collectively underscore the critical regulatory role of peripheral innervation in bone healing.

Identification of oeteoprogenitor cells with glial characteristics by scRNA-seq during bone repair

To elucidate the regulatory role of peripheral innervation in bone repair, we conducted scRNA-seq on both sham and denervated TES at day 3 post-extraction, a critical time point for the initiation of regeneration activities (Fig. 2a). A total number of 7 707 cells were obtained with 17 441 genes of each cell. We preprocessed the dataset with Seurat package. Only cells exhibiting between 200 and 6 000 expressed genes, and a mitochondrial unique molecular identifier (UMI) rate of less than 20%, were retained post cell-quality filtration (Fig. S2A). After quality control, 2 000 genes with the most variable value from 17 441 genes were selected for subsequent analysis. The UMAP dimensional reduction method identified nine distinct clusters (Fig. 2b), each characterized by unique cell markers including macrophages (Ccr2 and Csf1r), dendritic cells (Siglech), karyocytes (Ms4a2), hematopoietic stem and progenitor cells (Cd34, Kit, and Ly6a), myeloid progenitors (Mpo), B cells (Cd79a), T cells (Cd3e), neutrophils (Csf3r), and MSCs marked by Col1a2 and Gli1 (Figs. 2c and S2C). The top 5 expressed genes expressed in each defined cell type were identified and compared (Fig. S2B). Despite non-immune cells (MSCs) accounting for only 2.20% of all identified cells, we prioritized them due to their critical role in initiating bone regeneration and these MSCs could be divided into three subclusters (Fig. 2d). Cluster 3 was identified as OLCs for expression of Bmp2 and Bglap and enrichment in genes with established roles in osteoblast differentiation (Fig. S2E, F). Clusters 1 and 2 were identified as Gli1+ MSCs due to their high expression of Gli1 (Fig. 2d). Cluster 2, in particular, exhibited characteristics such as negative regulation of osteoblast differentiation and the capacity of cell migration, among others (Fig. S2D). Interestingly, while cluster 2 was equally represented in both sham and denervation groups, cluster 1 was enriched in sham TES and almost disappeared in denervation TES (Fig. 2e), indicating its relationship with inferior nerve and the healing of bone injury.

Fig. 2

Identification of oeteoprogenitor cells with glial characteristics by scRNA-seq during bone repair. a Flow chart of preparation of scRNA-seq samples from Sham and Denervation mice tooth extraction sockets. b Cell identities obtained through scRNA-seq were visualized via Uniform Manifold Approximation and Projection (UMAP). Various cell populations were demarcated and distinguished by color. Each point represented an independent cell. c The expression levels of stromal cell marker genes Col1a2 and Gli1 were projected onto UMAP atlas. d Subclustering of stromal cells reveals three cell clusters. Subclusters 1 and 2 express MSC marker Gli1. e The distribution of sham and denervation samples across stromal subcluster. f GO enrichment analysis of the biological functions of subcluster 1. g GSEA analysis indicating activation of a mesenchymal transition program of subcluster 1. h Violin plots showing increased co-expression of mesenchymal, neural glial cell markers, mesenchymal transition drivers, and Shh signaling associated markers in subcluster 1. i RNA velocities projected onto UMAP embedding of stromal cells subclusters. j, k Pseudotime lineage trajectory analysis of stromal cells subclusters. Inset: decreasing expression of glial marker Map1b and increasing expression of osteogenetic marker Sp7. T1: trajectory 1; T2: trajectory 2

To explore potential role of cluster 1, we performed Gene Ontology analysis and found several nervous system-related pathways were activated (Fig. 2f). Violin plots analysis consistently identified cluster 1 as a distinct subset of Gli1+ MSCs that was characterized by co-expression of mesenchymal transcripts (Gli1, Pdgfra), osteo-lineage cells marker (Cd200), neural glial markers (Plp1 and S100b), and transcription factors involved in epithelial-to-mesenchymal transition (EMT) regulation (Snai2 and Zeb2), opening the possibility of the existence of a “transitional” population expressing neural glial and mesenchymal transcripts indicative of the transition process (Fig. 2h). This notion was corroborated by striking enrichment of gene sets reflecting EMT (Fig. 2g, normalized enrichment score, 1.67; P = 0.035 51). To further investigate the transcriptional dynamics of MSCs differentiation, RNA velocity analysis was performed which predicts the future states of cellular subpopulations by taking into account the relative abundance of both nascent (unspliced) and mature (spliced) mRNA. Our findings revealed a differentiation trajectory originating from cluster 1, which was identified to be located in the upstream of the differentiation trajectory of OLCs (cluster 3), matching observations from above analyses (Fig. 2i). The experimental grouping information was projected onto the pseudotime trajectory, showing that cluster 1 has two potential differentiation trajectories. During normal bone healing, cluster 1 may differentiate into OLCs subcluster via trajectory 1 (T1) (Fig. 2j, k), with a stepwise decrease in expression of SCs marker Map1b and increase in expression of osteogenic marker Sp7 during the pseudotime (Fig. 2k, inset). However, it may switch to trajectory 2 (T2) and differentiate into cluster 2 after denervation, indicating the loss of osteogenic differentiation capacity. Collectively, these findings identified a subcluster of Gli1+ oeteoprogenitor cells with glial signature at the transcriptional level, which was computationally predicted to give rise to the formation of osteogenic niche during the healing of bone injury.

Plp1-lineage cells contribute to Gli1+ MSCs formation and promote bone regeneration

To define experimentally if Plp1-lineage cells contribute to the formation of Gli1+ MSCs via GMT process, we analyzed TES sections from Plp1-creERT2; tdTomato mice co-stained for Gli1. We observed a progressive increase in MSCs of Plp1 origin from day 0 to day 28 post-injury, signifying co-expression of tdTomato and Gli1(n = 6 for each time point). Before tooth extraction, tdTomato+ cells were detected in PDL and alveolar bone marrow, mostly being negative for Gli1 expression. From day 1 to day 28, tdTomato+ cells began to migrate from PDL into the sockets and keep expanding with the percentage of tdTomato+Gli1+ rising to (35.8 ± 7.9)% at day 28 (Fig. 3a), accompanied by the formation of type H vessels (Fig. 3b). Other mesenchymal markers CD105 and NG2 were also found to be co-expressed with tdTomato in TES (Fig. S3E). In addition, increasing tdTomato+ cells in TES were also positive for osteogenic marker osteopontin (OPN) at day 14 and day 28 (P < 0.001, n = 6), indicating the differentiation from Plp1-tdTomato+ cells to OLCs during the maturation stage of TES healing (Fig. S3A). Furthermore, Plp1-tdTomato+ cells in TES from 0 (alveolar bone), 3, 7 days after tooth extraction sorted by FACS revealed that the expression of the mature myelination marker MAG in SCs gradually decreased, while the expression of MSCs markers CD105, and Gli1 progressively increased (P < 0.05, n = 3) (Figs. 3c and S3C). RT-qPCR further confirmed this trend with a decline in Mag mRNA and an upregulation of Eng and Gli1 (P < 0.05, n = 3) (Fig. S3D). Notably, the presence of tdTomato+Gli1+ cells was significantly reduced in denervated TES at day 7 (P < 0.001, n = 6) (Fig. 3d). Compared to the Plp1-creERT2 negative littermate control mice, a significant proportion (50.92 ± 1.49%) of highly purified alveolar bone MSCs from day 7 TES, defined by CD45−Ter119−CD31−CD144−Sca1+CD51+ expression was found to be of glial origin as indicated by tdTomato+ in Plp1-creERT2; tdTomato mice. However, lower proportion [(37.57 ± 5.41)%] of downstream OLCs (CD45−Ter119−CD31−CD144−Sca1−CD51+) were also derived from Plp1-tdTomato+ cells (Figs. 3e and S3F, G), indicating a great portion (about 62.43%) of OLCs may be from other origins (n = 3). The immunostaining results showed that tdTomato+ cells were also associated with Runx2-labeled osteoprogenitors (Blue) in the healing region, suggesting its synergistic participation in the early osteogenesis process (Fig. S3B).

Fig. 3

Plp1-lineage cells contribute to Gli1+ MSCs formation and promote bone regeneration. a Representative images of tdTomato+ cells and Gli1 immunostaining during jaw bone regeneration and quantification of tdTomato+ Gli1+ cells in TES from Plp1-creERT2; tdTomato mice (n = 6). Scale bar: 100 μm. b Representative images of tdTomato+ cells combined with CD31 and EMCN immunostaining at day 7 post tooth extraction. c Western blot images of MAG, CD105, Gli1 expression of FACS-sorted tdTomato+ cells from tooth extraction sockets at day 0, 3, and 7 post tooth extraction. d Representative co-localization images of tdTomato+ cells and Gli1+ cells in healing sockets of Sham and Denervation mice at day 7 post tooth extraction and relative quantification (n = 6). Scale bar: 100 μm. e Representative flow cytometry plots depicting the gating strategy to identify MSCs and OLCs in TES of Plp1-creERT2; tdTomato and Plp1-creERT2 mice 7 days post tooth extraction. f Schematic illustration of the lineage tracing approach in the jaw bone healing model using Sox10-cre; tdTomato mice. g Representative images of tdTomato+ cells and Gli1 immunostaining during jaw bone regeneration in TES from Sox10-cre; tdTomato mice. Scale bar: 100 μm. h Schematic illustration of the lineage tracing approach in the jaw bone healing model using Gli1-creERT2; tdTomato mice. i Representative images of tdTomato+ cells and MAG immunostaining during jaw bone regeneration and quantification of tdTomato+ MAG+ cells in TES from Gli1-creERT2; tdTomato mice (n = 3). Scale bar: 100 μm. j Schematic of the experimental design to ablate SCs using an inducible diphtheria toxin allele (DTA). k H&E staining of tooth sockets from control and Plp1DTA mice at day 7 and day 14 post tooth extraction (n = 6). Scale bar: 100 μm. l Representative images of tdTomato+ cells and Gli1 immunostaining in TES of control and Plp1DTA mice at day 7 post tooth extraction and relative quantification per socket (n = 6). Scale bar: 100 μm. Data were presented as mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001

Furthermore, another SCs lineage tracing model Sox10-cre; tdTomato mice was used to add greater support to the GMT process during the healing of bone injury, showing the increasing co-expression of Sox10-tdTomato with Gli1 from d3 to d28 (n = 3 for each time point) (Figs. 3f, g and S3I). We also used Sox10-cre; tdTomato mice to trace the lineage differentiation and distribution of Sox10-positive cells in the development of the mandible. We observed that at postnatal day 3 (P3), Sox10+ cells were distributed around the tooth germ. As the jawbone and teeth developed, Sox10+ cells were widely distributed in pulp, PDL, and alveolar bone marrow space at P60, suggesting their potential involvement in both tooth and bone development (Fig. S3H). Gli1-creERT2; tdTomato mice were then used to detect potential multiple populations and the results showed that the number of tdTomato+MAG+ cells decreased during the healing of bone injury, providing additional evidence for the presence of GMT from a different perspective (P < 0.01, n = 3) (Fig. 3h, i).

To further substantiate the lineage tracing findings, we generated Plp1DTA mice and induced the selective ablation of Plp1+ cells followed by tooth extraction (Fig. 3j). Alveolar bone regions of tamoxifen-treated mice, harvested 3 days post-induction, exhibited widespread cell death among the Plp1+ cells population (Fig. S3L). von Frey filament stimulus testing showing no obvious mechanical hypoalgesia in Plp1DTA mice (P > 0.05, n = 5) (Fig. S3O) and the CGRP+ nerve fibers retained in TES after the deletion of SCs (P > 0.05, n = 6) (Fig. S3P). The volume of bone regenerated was severely suppressed in Plp1DTA mice at both day 7 and day 14 post-extraction (P < 0.01, n = 6) (Fig. S3M). Double-labeling analysis revealed defects in new bone formation rate in Plp1DTA mice by measuring the distance between the 2 fluorescent labels (P < 0.01, n = 6) (Fig. S3N). Further, markedly reduced Shh and HIF-1α expression were observed after selective ablation of Plp1+ cells (P < 0.01, n = 6) (Fig. S3Q, R). Histological analysis also revealed a reduced bone volume in TES of Plp1DTA mice at both day 7 (P < 0.01, n = 6) and day 14 (P < 0.05, n = 6) post-extraction (Fig. 3k). Of note, the number of total Gli1+ cells remarkably decreased after ablation of Plp1+ cells (P < 0.01, n = 6) (Fig. 3l). Likewise, the COL1+ bone area (P < 0.001, n = 6) and the density of type H vessels (P < 0.01, n = 6) also reduced in the TES of Plp1DTA mice (Fig. S3J, K). Collectively, these data corroborate our finding of a subset of Gli1+ MSCs within the TES, which originate from Plp1-lineage cells and play a pivotal role in bone regeneration.

Hh signaling driven Plp1-lineage cells transition is required for the healing of bone injury

Building upon the evidence that Plp1-lineage cells give rise to Gli1+ MSCs, we explored the mechanisms underlying GMT during bone healing. Initially, our scRNA-seq analysis highlighted the specific expression of Shh and downstream Hh signaling effectors, including Gli1, Gli2, and Smo, within cluster 1 (Fig. 2h). First, we detected the Shh mRNA expression in TES using FISH staining and found more co-expression of Shh with tdTomato than with CGRP, indicating that Shh is more likely secreted from tdTomato+ cells rather than neural fibers (P < 0.05, n = 3) (Fig. 4a). Shh expression was either co-localized with or in close proximity to tdTomato signals (n = 6 for each time point) (Fig. S4A, B), suggesting that injury-activated SCs secrete Shh to initiate their GMT. The Shh mRNA expression pattern also supports this conclusion (n = 6 for each time point) (Fig. 4b). Notably, Shh, recognized as a primary initiator of Gli1 expression and secreted by NVBs within the mandible, exhibited expression patterns that closely paralleled those of Gli1, showing a gradual increase from day 3 to day 7 post-tooth extraction before returning to baseline by day 14 (Fig. 4c). To directly test the role of Shh in driving GMT, we isolated primary tdTomato+ cells (SCs) from jaw bones sorted by FACS (Fig. 4d) and treated them with recombinant Shh (100 ng/mL). As expected, Shh significantly suppressed the expression of the SCs marker Mag and upregulated the mesenchymal markers Eng and Gli1 (P < 0.05, n = 3) (Fig. 4e).

Fig. 4

Hh signaling driven Plp1-lineage cells transition is required for the healing of bone injury. a Representative images of tdTomato+ cells, CGRP immunostaining in conjunction with in situ hybridization of Shh mRNA in TES from Plp1-creERT2; tdTomato mice. The Pearson’s correlation coefficient is presented as the quantification of Shh mRNA signals that were colocalized with tdTomato or CGRP (n = 3). Scale bar: 100 μm. b Representative images of tdTomato+ cells in conjunction with in situ hybridization of Shh mRNA and quantification of tdTomato+Shh+ cells in TES from Plp1-creERT2; tdTomato mice (n = 6). Scale bar: 100 μm. c Western blot images of Shh expression of FACS-sorted tdTomato+ cells from tooth extraction sockets at day 0, 3, and 7 post tooth extraction. d Schematic illustration of FACS assay and purified tdTomato+ cells from mandible. Scale bar: 20 μm. e RT-qPCR data of Eng, Gli1, and Mag mRNA expression of SCs treated with recombinant Shh (n = 3). f Shh converts SCs into osteolineage cells as demonstrated by alizarin red staining (ARS) of FACS-sorted tdTomato+ cells cultured in regular growth medium (GM) or osteogenic medium (OM). g The quantification of ARS in (f) (n = 3). h Western blot images of Runx2 and COL1 expression of FACS-sorted tdTomato+ cells cultured in GM or OM and treated with recombinant Shh. i RT-qPCR data of Runx2 and Alp mRNA expression of SCs cultured in GM or OM and treated with recombinant Shh (n = 3). j Experimental strategy for Hh signaling inhibition in Plp1-creERT2; tdTomato mice. k H&E staining of tooth sockets from DMSO and GDC mice at day 7 and day 14 post tooth extraction and the quantification analysis (n = 5). Scale bar: 100 μm. l, m Representative images of tdTomato+ cells and Gli1/COL1 immunostaining in TES at day 7 post tooth extraction and the relative quantification (n = 5). Scale bar: 100 μm. n Representative images of CD31 and EMCN immunostaining and quantification of CD31+EMCN+ type H vessels per socket (n = 5). Scale bar: 100 μm. Data were presented as mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001

Furthermore, to assess whether Plp1-lineage cells not only exhibit phenotypic changes indicative of GMT but also acquire cellular characteristics resembling stromal cell fates, we performed assays to evaluate the differentiation potential of Shh-treated Plp1-lineage cells into OLCs. The results demonstrated a dose-dependent increase in alizarin red staining, indicating that the addition of Shh initiated the differentiation of Plp1-tdTomato+ cells into OLCs (P < 0.01, n = 3) (Fig. 4f, g). WB and qRT-PCR results to test osteogenic markers also confirm this conclusion (P < 0.01, n = 3) (Fig. 4h, i). It is worth noting that cells treated with osteogenic medium alone were almost unable to form mineralized matrix, suggesting that Plp1+ cells do not have inherent osteogenic potential, but rather acquire the cellular characteristics resembling stromal cell fates to differentiate into OLCs under the activation of the shh signaling. To further investigate the role of Hh signaling in bone healing and GMT, we administered GDC-0449 (Hh signaling inhibitor) for three consecutive days following tooth extraction (Fig. 4j). The results indicated a decrease in tdTomato+Gli1+ cells (P < 0.01, n = 5) and no change in total tdTomato+ cells (P > 0.05, n = 5) (Figs. 4l and S4F). Micro-CT and double-labeling analyses revealed defects in bone regeneration at day 7 (P < 0.05, n = 5) (Fig. S4C) and new bone formation rate (P < 0.05, n = 5) (Fig. S4D, E) and the number of tdTomato+CD105+ or tdTomato+NG2+ cells also decreased after GDC-0449 treatment (P < 0.01, n = 5) (Fig. S4G, H), signifying an impaired GMT process. Additionally, there was a reduction in COL1+ areas in the GDC-0449 treated group (P < 0.05, n = 5) (Fig. 4m), and micro-CT analysis along with histological assessments revealed a delayed healing in TES on day 7, although no significant differences were observed by day 14 (P > 0.05, n = 5), suggesting a temporal effect of the inhibitor on healing (Figs. 4k and S4C). Furthermore, a decrease in CD31+EMCN+ type H vessels was observed following GDC-0449 administration at day 7 (P < 0.01, n = 5) (Fig. 4n). These findings collectively demonstrate that Shh signaling is indispensable for the GMT and the subsequent healing process of bone injuries.

Impaired GMT is associated with age-related healing delay of bone injury

Considering the effects of Hh signaling-driven GMT on the healing process of bone injury, we asked whether age-related healing delay was associated with impaired GMT and established the TES model with aged mice (Fig. 5a). Firstly, despite the comparable number of Plp1-tdTomato+ progenitors in alveolar bones and PDL before bone injury observed in both young (6 months) and old mice (18 months) (P > 0.05, n = 5) (Fig. S5B), aged mice exhibited a significant reduction in bone volume at day 7 and day 14 post-extraction compared to their younger counterparts (P < 0.01, n = 5) (Fig. S5A). This delay of bone regeneration was further characterized by decreased BMD and BV/TV in aged mice (P < 0.01, n = 5) (Figs. 5b and S5C), as well as declined bone formation rate (P < 0.01, n = 5) (Fig. S5D), reduced areas of COL1+ (P < 0.05, n = 5) (Fig. S5E) and CD31+EMCN+ type H vessels (Fig. 5e). Additionally, aged mice exhibited markedly lower levels of Shh expression and tdTomato+Gli1+ cells in the TES at day 7 (P < 0.001, n = 5) (Fig. 5c, d). Further investigation into SCs from aged mice, isolated via FACS from the alveolar bones (TES-0d) or TES (TES-7d), revealed a diminished ability to initiate GMT as evidenced by nearly invariant expression levels of Shh, the stromal markers CD105, Gli1, and the SCs marker MAG from day 0 to day 7 (Fig. 5f and Fig. S5F). The Shh mRNA expression also decreased markedly in SCs from aged mice (P < 0.01, n = 3) (Fig. 5g). These findings suggest that age-related delays in bone healing may be associated with an impaired GMT of SCs.

Fig. 5

Disrupted GMT of Plp1-lineage cells attenuates bone regeneration in aged mice. a Experimental strategy for aged mice lineage tracing and jaw bone healing model. b Representative images of μCT reconstruction of the alveolar bone regeneration at day 7 post tooth extraction from young/old mice and quantitative analyses (n = 5). c Representative images of Shh immunostaining in healing sockets at day 7 post tooth extraction and the quantification (n = 5). Scale bar: 100 μm. d Representative images of tdTomato+ cells and Gli1 immunostaining in healing sockets at day 7 post tooth extraction and relative quantification per socket (n = 5). Scale bar: 100 μm. e Representative images of CD31 and EMCN immunostaining and quantification of CD31+EMCN+ type H vessels per socket (n = 5). Scale bar: 100 μm. f Western blot images of CD105, Gli1, MAG, and Shh expression of FACS-sorted tdTomato+ cells from young/old mice tooth extraction sockets at day 0 and 7 post tooth extraction. g RT-qPCR data of Shh mRNA expression of SCs from young/old mice (n = 3). h NAD+ was measured in serum and supernatant medium of SCs from young and old mice (n = 5). Y, young mice. O, old mice. i RT-qPCR analysis of mRNA levels of Sirt1-7 in SCs (n = 3). j Western blot analysis of the knockdown efficiency of siRNAs for SIRT1, SIRT2, SIRT3, SIRT6, and relevant Shh expression. k Representative images of tdTomato+ cells and SIRT6 immunostaining in TES at day 7 post tooth extraction and the relative quantification (n = 5). Scale bar: 100 μm. l Western blot showing SIRT6, c-Jun, Shh, Ptch1, and Gli1 in SCs with aging or SIRT6 knockdown/overexpression. Data were presented as mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001

To unravel the upstream regulatory mechanisms responsible for age-related Shh downregulation in TES and the compromised GMT, we focused on NAD+ metabolism, a known hallmark of aging linked to skeletal diseases.32 Both serum and supernatant medium of SCs NAD+ levels in aged mice were substantially lower than those in younger mice (P < 0.01, n = 5) (Fig. 5h). Since sirtuins are NAD+-dependent protein deacetylases known to affect bone regeneration,33 we investigated their role in regulating Shh expression. Profiling of sirtuin genes in SCs during aging revealed significantly reduced expression of Sirt1, Sirt2, Sirt3, and Sirt6 (Fig. 5i). Notably, SIRT6 knockdown specifically resulted in a pronounced decrease in Shh expression, an effect not observed with knockdown of other sirtuins (Fig. 5j). Additionally, the reduced expression of SIRT6 in tdTomato+ cells further indicated a deficiency of this protein in SCs from aged mice (P < 0.01, n = 5) (Fig. 5k). To assess the impact of SIRT6 on Hh signaling, we conducted in vitro experiments treating primary SCs from young and old mice with SIRT6 knockdown (si-SIRT6) and overexpression vectors (SIRT6-OE). The results demonstrated that si-SIRT6 significantly reduced the expression of SIRT6, along with notable decreases in Shh, Ptch1, and Gli1 expression in young SCs. Conversely, SIRT6-OE induced the opposite effects in old SCs, enhancing the expression of these markers (Figs. 5l and S5G). Thus, our findings indicate that impaired GMT of SCs contributes to the age-related delay in bone injury healing, potentially attributed to a deficiency in SIRT6.

c-Jun/SIRT6/BAF170 complex binds to injury-specific enhancers and activates shh transcription

SIRT6, a histone deacetylase known for its role in chromatin condensation and gene suppression,34 has been identified as a positive regulator of Shh expression, as demonstrated in our studies (Fig. 5j, l). To elucidate the molecular mechanisms by which SIRT6 influences Shh transcription in SCs, we investigated its potential interacting partners. The c-Jun subunit of the AP-1 transcription factor complex has captured our attention due to its pivotal role in driving SCs’ transcriptional response following injury, controlling the transdifferentiation of myelin and Remak SCs into repair-specific cells.35 Notably, c-Jun binds to injury-induced enhancers to activate Shh transcription in SCs involved in repair processes.25 Our results showed that overexpression of c-Jun in SCs led to increased Shh expression, which was diminished by SIRT6 knockdown (Fig. 6a). Protein–protein docking predictions from AlphaFold 2 indicated that SIRT6 could bind directly to c-Jun through hydrogen bonds (Fig. 6b), and further co-immunoprecipitation (co-IP) assays confirmed the interaction between endogenous SIRT6 and c-Jun (Fig. 6c). Consistently, when Flag-tagged SIRT6 was introduced in HEK293T cells, this ectopic protein was efficiently precipitated by c-Jun antibodies (Fig. S6A). Immunofluorescence studies also validated this interaction (Fig. 6d).

Fig. 6

c-Jun/SIRT6/BAF170 complex binds to injury-specific enhancers and activates shh transcription in SCs. a RT-qPCR analysis of shh mRNA level in SCs treated with c-Jun overexpression or SIRT6 knockdown (n = 3). b 3D modeling of the interaction between c-Jun (blue) and SIRT6 (yellow) proteins elucidated through protein-protein docking predictions from AlphaFold 2. The stick representation delineates the amino acid residues, while yellow dashed lines illustrate the hydrogen bonds. c Western blot showing co-immunoprecipitation between SIRT6 and c-Jun. d Immunofluorescence images of the colocalization of SIRT6 and c-Jun in SCs. Scale bar: 10 μm. e RT-qPCR analysis of shh mRNA level in SIRT6 knockdown SCs rescued with different alleles of SIRT6 (n = 3). f Western blot showing co-immunoprecipitation between SIRT6 and BAF170 with downregulation or upregulation of SIRT6. g RT-qPCR analysis of Shh mRNA expression after BAF170 knockdown (n = 3). h Western blot showing the expression of BAF170 and Shh after BAF170 knockdown. i RT-qPCR analysis of shh mRNA level in SCs treated with SIRT6 knockdown and BAF170 overexpression (n = 3). j Western blot showing the expression of Shh, Ptch1, and Gli1 in SCs treated with SIRT6 knockdown and BAF170 overexpression. k RT-qPCR analysis of shh mRNA level in SCs after transfection with a plasmid expressing BAF170-K312A (n = 3). l Western blot showing the expression of Shh, Ptch1, and Gli1 in SCs treated with BAF170-K312A mutant. m Western blot showing co-immunoprecipitation between c-Jun and BAF170. n ChIP-qPCR for c-JUN and SIRT6 on three reported injury-specific shh enhancer sites and ChIP-qPCR for BAF170 on shh enhancer sites treated with SIRT6 knockdown (n = 3). o Luciferase reporter assay for shh enhancer 1 and 3 (n = 3). p Schematic showing the mechanism that c-Jun/SIRT6/BAF170 complex binds to injury-specific enhancers and activates SCs shh transcription. Data were presented as mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001

The SIRT6 protein possesses dual enzymatic functionalities: deacetylase and mono-ADP-ribosylase activities.31 To discern which of these enzymatic activities of SIRT6 is imperative for the augmentation of Shh transcription, we transfected SCs with functionally distinct SIRT6 mutants: the G60A variant (retaining deacetylase activity but deficient in mono-ADP-ribosylase function) and the R65A variant (competent in mono-ADP-ribosyl transferase activity but lacking deacetylase function). The catalytically dead (H133Y) and G60A alleles of SIRT6 significantly inhibited Shh expression while R65A alleles did not compare with WT plasmid, suggesting that ribosylation but not deacetylation activity of SIRT6 is involved in Shh transcriptional activation (Fig. 6e). It has been reported that SIRT6 recruits BAF170 to enhancer region of several genes locus and promotes transcriptional enhancement through mono-ADP ribosylation.31 We asked whether c-Jun/SIRT6 could recruit BAF170 to Shh enhancer regions and promote its mRNA transcription. A co-IP assay in the presence or absence of si-SIRT6 and SIRT6-OE plasmids was performed, verifying the interaction between endogenous SIRT6 and BAF170 (Fig. 6f). Immunofluorescence results also confirmed this conclusion (Fig. S6B). Upon BAF170 knockdown in SCs, Shh expression significantly decreased at both mRNA (P < 0.01, n = 5) and protein levels (P <