Gli1⁺ cell aggregates promote type H vessel formation and orchestrate bone defect regeneration Authors: Chao Ma, Yu-Ru Gao, Hao Wang, Bing-Dong Sui, Shan-Shan Huang, Yi Zhang, Lu Liu, Xiao-Hui Zhang, Shu-Juan Xing, Yuan-Yuan Li, Yi-Han Xiao, Ya-Nan Zheng, Li-Heng Ren, Yan Jin, Chen-Xi Zheng, Hao-Kun Xu, Ji Chen Published: 2026 ABSTRACT AbstractBackgroundRapid and stable regeneration of bone defects remains a pressing clinical challenge. We previously fabricated stem cell aggregates (CA) by mimicking developmental condensation and demonstrated their efficacy in promoting bone defect repair. Endogenous Gli1+ skeletal stromal/progenitor cells (SSPCs) are a pivotal SSPC subtype known to maintain bone homeostasis and enhance bone regeneration; however, the functional properties and translational potential of CA derived from these cells (Gli1+ CA) remain largely elusive.MethodsSingle-cell RNA sequencing was performed to characterize differential gene expression profiles between Gli1high and Gli1low SSPCs. The spatial relationship among Gli1+ cells, RUNX2+ cells, and type H vessels in vivo was further validated. Gli1-CreERT2;mT/mG transgenic mice were generated to enable the isolation of Gli1+ SSPCs and the subsequent fabrication of Gli1+ CA. The pro-angiogenic potential of Gli1+ CA was assessed in vitro, and the underlying regulatory mechanisms were further explored. Finally, Gli1+ CA were implanted into a mouse femoral defect model, and bone regenerative efficacy was evaluated by micro-CT and immunofluorescence staining.ResultsSingle-cell RNA sequencing revealed that, compared with Gli1low SSPCs, Gli1high SSPCs highly expressed genes associated with osteogenesis, angiogenesis, and extracellular matrix synthesis. In vivo validation demonstrated robust enrichment of Gli1+ cells in the metaphysis; these cells exhibited a tight spatial correlation with the osteogenic master transcription factor RUNX2 and type H vessels. We subsequently sorted Gli1+ SSPCs via flow cytometry and fabricated CA, and in vitro analysis confirmed that their expression profiles were consistent with the sequencing data. Functional assays further revealed that Gli1+ CA promoted endothelial tube formation through paracrine signaling. Ultimately, Gli1+ CA markedly accelerated bone regeneration in a mouse femoral defect model compared with unsorted CA and Gli1⁻ CA, likely by inducing type H vessel formation.ConclusionsThis study not only addresses a critical knowledge gap in Gli1+ CA-mediated bone regeneration, but also proposes a novel strategy termed “precision screening of endogenous SSPC subsets coupled with targeted aggregate fabrication”. This approach offers a more precise therapeutic direction for the regenerative treatment of bone defects.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13287-026-05109-y.Keywords: Gli1+ cell, Cell aggregates, Skeletal stromal/progenitor cells, Type H vessels, Regeneration, Bone Introduction Bone defect repair remains a critical clinical challenge in orthopedics and stomatology [1–3]. Traditional treatments such as bone grafting and prosthetic replacement are hampered by inherent limitations, including donor scarcity, immune rejection, and suboptimal long-term repair outcomes, which severely compromise patients’ quality of life [4, 5]. Notably, bone tissue regeneration shares striking similarities with embryonic bone development in terms of cell lineage differentiation and signal regulatory networks [6, 7]. Thus, developing regenerative strategies that recapitulate the key mechanisms of natural bone development is regarded as a promising approach to overcome the bottlenecks in bone defect repair [8]. The regenerative capacity of bone relies on the rapid activation and lineage-specific differentiation of skeletal stromal/progenitor cells (SSPCs). These cells can give rise to osteoblasts and play an essential role in osteogenesis during both physiological homeostasis and fracture repair [9, 10]. Our group has previously conducted a series of studies on the osteogenic potential of SSPCs, achieving favorable regenerative effects in bone disease models via exogenous stem cell transplantation [11, 12], which lays a foundation for subsequent investigations. However, accumulating evidence confirms significant heterogeneity within mesenchymal stromal/progenitor cell populations: distinct subpopulations differ substantially in proliferation, osteogenic capacity, and extracellular matrix (ECM) secretory capacity [13–15]. The inclusion of non-regenerative subpopulations may attenuate therapeutic efficacy and even induce adverse tissue reactions [15]. Therefore, precise screening of highly regenerative SSPC subpopulations and their targeted application in bone regeneration are of great theoretical value and clinical significance for enhancing the efficacy and safety of stem cell-based therapies. Developmental condensation is a critical initiating event in embryonic bone formation [16–20]. Through close intercellular adhesion and signal communication, it forms mesenchymal condensates with directional differentiation potential, laying a structural and functional foundation for subsequent bone matrix deposition and tissue maturation [21–23]. Based on this fundamental mechanism, our research group and counterparts worldwide have successfully constructed three-dimensional cellular structures such as stem cell sheets and stem cell aggregates (CA) [24–29]. By simulating the developmental condensation microenvironment, these structures significantly enhance the osteogenic efficiency of stem cells [30]. We further applied mesenchymal stem cell-derived CA in clinical trials, achieving the first successful regeneration of dental pulp tissue, as well as the functional revitalization of avulsed teeth, periodontal tissues and alveolar bone [31, 32]. It is worth noting that Gli1+ cells, as a widely reported endogenous SSPC subpopulation, play a key regulatory role in maintaining bone homeostasis, initiating post-injury bone regeneration, and regulating the repair process [33, 34]. Gli1+ SSPCs, acting via the Hedgehog signaling pathway, modulate osteoblast differentiation and angiogenesis to maintain bone homeostasis [35, 36]. Our group’s previous explorations on the distribution characteristics, differentiation potential, and functional regulation of Gli1+ SSPCs in maxillofacial bones have further confirmed their central position in tissue homeostasis and regeneration [37–40]. However, current research on CA largely relies on the isolation, primary culture, and fabrication of heterogeneous SSPC populations [39, 41, 42]. Limited by the difficulty in applying gene-traced mice and the technical bottleneck in the specific identification, isolation, and purification of endogenous SSPCs, there have been no reports on the direct construction of aggregates based on Gli1+ SSPCs and their application in bone regeneration, which also restricts the clinical transformation potential of this regenerative strategy. Recent years have witnessed a breakthrough in studying the coupling mechanism of osteogenesis and angiogenesis, thanks to the discovery of a specific capillary subtype in bone—type H vessels, marked by CD31hiEndomucin (EMCN)hi [43, 44]. Furthermore, combined with three-dimensional fluorescence imaging, tissue clearing techniques employed in a recent study revealed a close spatial correlation between endogenous SSPCs and type H vessels during bone development and defect repair [45]. Our team’s previous study further demonstrated specific knockout of Gli1+ SSPCs significantly reduces the density of type H vessels in the femoral metaphysis and markedly impairs bone regeneration [36]. Subsequent investigations revealed that CA can promote the regeneration of bone and skin defect tissues by inducing type H neovascularization [46–48]. Based on the above research foundation, we propose the core scientific question: compared with traditional mixed stem CA, can aggregates constructed from Gli1+ SSPCs more effectively promote type H vessel formation by enhancing the “Gli1+ cells-type H vessels” functional coupling, thereby improving bone defect repair outcomes? In this study, we first analyzed single-cell RNA sequencing data of long bone tissues and found that, compared with Gli1low SSPCs, genes upregulated in Gli1high SSPCs were significantly enriched in biological processes related to osteogenesis, angiogenesis, and ECM synthesis. In vivo validation demonstrated robust enrichment of Gli1+ SSPCs in the metaphysis; these cells exhibited a tight spatial correlation with the osteogenic master transcription factor RUNX2 and type H vessels. To further validate these transcriptomic findings, we sorted Gli1+SSPCs via flow cytometry, fabricated CA, and confirmed that their in vitro expression profiles were consistent with the sequencing data. To further validate these observations, which together suggest a coupled osteogenic and angiogenic potential, we sorted Gli1+ SSPCs, fabricated CA, and confirmed through comprehensive in vitro characterization that Gli1+ CA indeed exhibited the predicted osteogenic, angiogenic, and ECM-producing properties. Functional assays further revealed that Gli1+ CA promoted endothelial tube formation through paracrine signaling. In a mouse femoral defect model, Gli1+ CA markedly induced type H vessel formation and accelerated bone defect regeneration compared with unsorted CA and Gli1⁻ CA. Collectively, our findings not only address a critical knowledge gap in Gli1+ CA-mediated bone regeneration, but also propose a novel strategy termed “precision screening of endogenous SSPC subsets coupled with targeted aggregate fabrication”, offering a more precise therapeutic direction for the regenerative treatment of bone defects. Methods The work has been reported in line with the ARRIVE guidelines 2.0. Single-cell RNA sequencing To clarify gene expression differences between Gli1+ SSPCs and Gli1− SSPCs, we retrieved previously published single-cell RNA sequencing data (accession code: GSE128423) [49], which originally characterized stromal cells from mouse long bones and bone marrow. Data processing was performed using the Seurat R package. Low-quality cells were filtered out if they met any of the following criteria: UMI counts > 20,000, gene counts > 4000 or < 500, or mitochondrial UMI percentage > 10%. Doublets were subsequently removed using DoubletFinder (v2.0.3). The data were normalized via LogNormalize (scale factor = 10,000) and integrated using the Harmony algorithm to correct for batch effects. Cell clusters were identified with the “FindClusters” function (resolution = 0.4) after principal component analysis (PCA) based on the top 30 principal components, and visualized by Uniform Manifold Approximation and Projection (UMAP) with parameters n.neighbors = 30, metric = “correlation”, min.dist = 0.3. Differentially expressed genes (DEGs) were identified using the Wilcoxon rank-sum test (“FindAllMarkers”), with significance defined as |log2 fold change| ≥ 0.25 and a Bonferroni-adjusted P value < 0.01. Re-clustering of the integrated dataset yielded 24 distinct cell clusters (Fig. S1A). Cell type annotations were assigned based on canonical marker gene expression, and clusters exhibiting overlapping transcriptional profiles were merged, resulting in a refined classification of 11 major cell types (Fig. S1B; Fig. 1A). Finally, Gli1 expression levels were projected onto the UMAP embedding to identify clusters significantly enriched for Gli1 (Fig. 1G). Single-cell transcriptomic profiling of mouse long bones identifies Gli1high SSPCs with high expression of pro-angiogenic and pro-osteogenic genes. A UMAP plot showing unbiased graph-based clusters distribution of all cell populations in long bones from 6–8 weeks old mice. B Gli1 is predominantly enriched in SSPCs, fibroblasts and chondrocytes. C Cd44 is predominantly enriched in fibroblasts and erythrocytes, and is also expressed in cell subsets including SSPCs, chondrocytes, OLC, macrophages, monocytes and platelets. D Eng is widely expressed in almost all cell types (including SSPCs), with primary enrichment in BMEC. E Nt5e is predominantly enriched in SSPCs, fibroblasts and chondrocytes. F The negative SSPC marker Ptprc is predominantly enriched in erythrocytes, macrophages, monocytes, neutrophils and B cells, while its expression is negligible in SSPCs. G The differential expression of Gli1 was projected onto SSPC subsets. H Gli1+ SSPCs were stratified into Gli1high and Gli1low SSPC subsets based on the significant differences in Gli1 gene expression levels. I Heatmap showing DEGs between Gli1high and Gli1low SSPC subsets. n = 14. Statistical analyses were performed by Student’s t test. ***P < 0.001 Given the heterogeneous expression of Gli1 within the SSPC population, we further stratified Gli1+ SSPCs into two subgroups—Gli1high SSPCs and Gli1low SSPCs—based on the expression level of the Gli1 gene. Gli1+ SSPCs with the top 25% of Gli1 expression were defined as Gli1high SSPCs, while those with the bottom 25% of Gli1 expression in the Gli1+ SSPC population were designated as Gli1low SSPCs [50–53]. To elucidate the functional disparities between Gli1high SSPCs and Gli1low SSPCs, we performed differential gene expression analysis, generated a heatmap to visualize the significantly DEGs, and further conducted Gene Ontology (GO) enrichment analysis to delineate their functional differences in greater detail. The single-cell RNA sequencing data was analyzed using Omicsmart, a real-time interactive online platform for data analysis ( https://www.omicsmart.com). Animals All animal experiments in this study were conducted with the approval of the Institutional Animal Care and Use Committee of The Fourth Military Medical University. C57BL/6J (JAX no. 000664), Gli1-CreERT2 (JAX no. 007913) and ROSA26-mT/mG (JAX no. 007676) were purchased from the Jackson Laboratory, USA. Gli1-CreERT2 mice were crossed with ROSA26-mT/mG mice to generate Gli1-CreERT2;mT/mG mice. C57BL/6J (JAX no. 000664) were used as the wild type. All mice were housed in a pathogen-free condition, maintained on the standard 12-h light-dark cycle, fed with food and water at libitum, and used for analyses regardless of sex. Genotyping was performed by polymerase chain reaction (PCR) as indicated, using tail samples from mice and primer sequences provided by the Jackson Laboratory. The relevant primers are shown in Table S1. Tamoxifen (MCE, China) was injected intraperitoneally at 100 µg/g for three consecutive days to induce Cre-mediated DNA recombination [54]. At the experimental endpoint, mice were deeply anesthetized by intraperitoneal injection of 1% pentobarbital sodium to achieve a surgical plane of anesthesia prior to euthanasia, tissue collection and terminal procedures. For animals, the inclusion criteria were: (1) being healthy adult mice (6–8 weeks old, 18–25 g body weight) with no signs of infection, injury, or abnormal behavior; (2) successful genotyping confirmation via PCR using tail samples; (3) stable physiological status after 7-day acclimatization in a pathogen-free facility. Exclusion criteria for animals were: (1) failure to meet genotyping requirements; (2) severe adverse events during the experiment, including uncontrollable surgical site infection, severe weight loss, or inability to eat/drink independently; (3) death prior to the experimental endpoint. No animals were excluded during the experiment as all met the inclusion criteria and no severe adverse events leading to exclusion occurred. For data points during analysis, the exclusion criteria were: (1) data from animals that were excluded (none in this study); (2) incomplete or invalid data due to technical errors. No outliers or invalid data points were identified, so all collected data were included in the final analysis. Establishment of the femoral defect model C57BL/6J mice were anesthetized via intraperitoneal injection of 1% pentobarbital sodium until full anesthesia was achieved. Randomization was performed using Excel’s RANDBETWEEN function, and mice were randomly assigned to experimental groups according to the generated random numbers. The sample size was determined based on previous studies using similar femoral defect models in mice and our preliminary experimental data. Each mouse was considered as one experimental unit. Only the investigator performing group allocation was aware of group assignments, and this individual did not participate in subsequent experiments, outcome assessment, or data analysis, ensuring blinding during outcome evaluation and data processing. To minimize potential confounders, all surgical procedures, treatments, and outcome measurements were performed in a random order. Mice were housed in identical cages with standardized locations in the animal room to avoid environmental bias. The fascia and muscle tissues were bluntly separated along the long axis of the femur, and a circular defect with a diameter of 1 mm and a depth of approximately two-thirds of the femoral diameter (with the contralateral cortical bone remaining intact) was created using a dental round bur at the epiphysis, 0.5 mm inferior to the distal growth plate of the unilateral femur. After rinsing the bone defect with normal saline, unsorted CA, Gli1+ CA, and Gli1− CA were respectively implanted into the defect sites. Unsorted CA served as the control group for comparison with the other two groups. Finally, the muscle and skin were sutured layer by layer, and the surgical area was disinfected with 75% ethanol to prevent infection. The primary outcome measure consisted of the results from micro-CT analysis and immunofluorescence staining. Specifically, this included bone microstructural parameters (bone volume per tissue volume (BV/TV), bone surface/bone volume (BS/BV), trabecular thickness (Tb.Th), trabecular number (Tb.N), and trabecular separation/spacing (Tb.Sp)) and the expression levels of RUNX2, CD31, and EMCN in the regenerated femoral defect tissue. These indicators were used to determine the sample size and to test the study hypothesis regarding the efficacy of different cell populations in promoting bone regeneration and angiogenesis. Micro-CT analysis Following euthanasia, femurs were harvested and fixed in 4% PFA. The fixed specimens were scanned using micro-CT (Siemens Inveon, Germany; PerkinElmer, USA) at a resolution of 36 μm (50 kV, 80 µA). Three-dimensional reconstruction and quantification of bone microstructural parameters (BV/TV, BS/BV, Tb.Th, Tb.N, and Tb.Sp) were subsequently performed using VGStudio MAX software (Volume Graphics, Germany). Histological analysis After sacrificing the mice, the femurs were fixed in 4% paraformaldehyde solution (Sigma-Aldrich, USA) overnight at 4 °C. Subsequently, the fixed femurs were subjected to dehydration in a graded ethanol series, cleared in xylene, infiltrated with paraffin wax, and embedded in paraffin blocks. Paraffin sections with a thickness of 5 μm were prepared. Hematoxylin-eosin (HE) staining and Masson’s trichrome staining were performed using commercial staining kits (Baso Technology, China) following the protocols described in previous studies [36, 55]. Images were captured using the SLIDEVIEW VS200 slide scanner (Olympus, Japan). Immunofluorescence staining The femurs obtained were fixed overnight in 4% paraformaldehyde (Sigma-Aldrich, USA). After rinsing the tissues three times with phosphate-buffered saline (PBS) to completely remove residual paraformaldehyde, the samples were decalcified in 17% ethylenediaminetetraacetic acid (EDTA) decalcification solution for 3 weeks. Following another round of PBS rinsing, the tissues were dehydrated overnight in 30% sucrose solution. On the next day, the sinking of tissues to the bottom of centrifuge tubes indicated successful dehydration, after which the tissues were embedded in optimal cutting temperature (OCT) compound (Leica, Germany). Cryosections with a thickness of 15 μm were prepared using a cryostat (Model CM1950, Leica, Germany). The immunofluorescence staining protocol was as follows: the cryosections were rinsed with PBS and then permeabilized with 0.3% Triton X-100 (Sigma-Aldrich, USA) for 15 min at room temperature, followed by blocking with 5% goat serum (BOSTER, China). The sections were incubated with primary antibodies overnight at 4 °C. Details of the primary antibodies were as follows: anti-CD31 antibody (FAB3628G, R&D Systems, USA; diluted 1:100) conjugated with Alexa Fluor 488, anti-EMCN antibody (sc-65495, Santa Cruz Biotechnology, USA; diluted 1:100), anti-RUNX2 antibody (sc-390715, Santa Cruz Biotechnology, USA; diluted 1:200), anti-β-galactosidase antibody (ab9361, Abcam, UK; diluted 1:200), anti-F-actin antibody (ab130935, Abcam, UK; diluted 1:200), and FITC Phalloidin (RM02836, ABclonal, China; diluted 1:400). After rinsing the sections with PBS, they were incubated with secondary antibodies for 1 h at 37 °C. Details of the secondary antibodies were as follows: FITC-conjugated goat anti-mouse IgG (SA00003-1, Proteintech, China; diluted 1:200), Texas Red-conjugated goat anti-chicken IgG (ab6875, Abcam, UK; diluted 1:400), and Alexa Fluor 647-conjugated goat anti-rat IgG (ab150159, Abcam, UK; diluted 1:400). After secondary antibody incubation, the sections were rinsed with PBS and mounted using DAPI-containing anti-fluorescence quenching mounting medium (Abcam, UK). The process of image acquisition and statistical analysis was as follows: images were captured using a confocal laser scanning microscope (Model Nikon A1+, Nikon, Japan) and analyzed with ImageJ software (National Institutes of Health, USA) according to the method described in a previous study [36]. The brief image analysis protocol was as follows: ImageJ software was used to quantify RUNX2+ cells or CD31+EMCN+ vessels adjacent to Gli1+ cells; specifically, the area of each fluorescently stained region within a 20 μm range around Gli1+ cells was measured, and its percentage relative to the total area of the captured field of view was calculated (% total area). For the quantification of RUNX2+ cells, CD31+EMCN+ vessels, and F-actin+ cells, the area of their fluorescently stained regions was measured, and the corresponding percentage relative to the total area of the captured field of view was calculated (% total area). Cell isolation Gli1-CreER T2;mT/mG mice received consecutive intraperitoneal injections of tamoxifen for 3 days to induce membrane green fluorescent protein (mGFP) labeling of Gli1+ cells. According to the previously published protocol [56, 57], bone marrow SSPCs were isolated from the long bones of mice 1 week later. The detailed procedures were as follows: the femurs and tibias were harvested under sterile conditions. The proximal and distal epiphyses of the long bones were excised with scissors, and the bone marrow was flushed out using a 1 mL syringe until the flushing solution became clear. The bone marrow flushing fluid was cultured in a Petri dish together with the minced bone fragments. The culture was maintained in α-minimum essential medium (α-MEM, Invitrogen, USA) supplemented with 20% fetal bovine serum (FBS, Tianhang, China), 2 mmol/L L-glutamine (Invitrogen, USA), and 1% penicillin-streptomycin (Invitrogen, USA) at 37 °C with 5% carbon dioxide (CO₂). The culture medium was replaced every 3 days, and cell passaging was performed when the cell confluence reached 80%–90%. For flow cytometric analysis of the surface makers, the 3rd-passaged MSCs were collected, and no less than 1 × 105 cells were counted and resuspended with PBS. MSCs were incubated with FITC-conjugated anti-CD73 (127220, Biolegend, USA; diluted 1:200), FITC-conjugated anti-CD90 (MA1-80648, Invitrogen, USA; diluted 1:10), PE-conjugated anti-CD105 (12-1051-81, Invitrogen, USA; diluted 1:40), PE-conjugated anti-CD44 (553134, BD, USA; diluted 1:20), FITC-conjugated anti-CD11b (101206, Biolegend, USA; diluted 1:200), or PE-conjugated anti-CD45 (12-0451-82, Invitrogen, USA; diluted 1:150) at 4 °C for 30 min. The percentages of positively stained cells were analyzed by a CytoFLEX flow cytometer (Beckman Coulter, USA). Flow cytometric sorting of GFP-positive cells was conducted when the cells were cultured and passaged to the third generation. The cells were digested with 0.25% trypsin (MP Biomedicals, USA), collected by centrifugation at 500 × g for 5 min, resuspended in PBS containing 2% FBS at a density of 1 × 10⁶ cells/mL, and then subjected to flow cytometric sorting using a flow cytometer (BD, USA) to obtain GFP-positive cells. Tube formation assay We performed an in vitro tube formation assay to evaluate the capillary-like network formation capacity. The experiment was conducted on ibidi angiogenesis µ-Slides (ibidi, Germany) pre-coated with Matrigel (Shanghai NovaMedical Technology, China). Simian Virus 40-immortalized mouse lymph node endothelial cell line 4–10 (SVEC4-10)was purchased from the ATCC and endothelial cells (ECs) were seeded at a density of 2 × 10⁵ cells/mL (with 50 µL of cell suspension added per well), followed by treatment with conditioned media (CM) derived from unsorted CA, Gli1+ CA, and Gli1− CA, respectively. After incubation at 37 °C for 2 or 3 h, the tube formation morphology was observed and photographed using an inverted microscope (Leica, Germany). Three microscopic fields were randomly selected, and the total number of branching points and total tube length of the capillary-like networks were quantitatively analyzed using the Angiogenesis Analyzer plugin of ImageJ software (National Institutes of Health, USA). Induction of aggregates formation The protocol for constructing CA was performed according to the previously established method of our research group [56, 57]. The detailed procedures were briefly described as follows: unsorted SSPCs, sorted Gli1+ SSPCs, and sorted Gli1− SSPCs were seeded in 6-well plates at a density of 2 × 10⁵ cells per well. When the cell confluence reached 80%–90%, the culture medium was replaced with a CA induction medium. This induction medium was based on α-minimum essential medium (α-MEM, Gibco, USA) and supplemented with 10% FBS (Tianhang, China) and 50 µg/mL vitamin C (Sigma-Aldrich, USA), with the culture conducted in the dark throughout the entire process. The medium was refreshed every 2 days. After 14 days of induction culture, microscopic observation revealed that the CA exhibited a membrane-like structure, with thickened edges that detached from the bottom of the culture plate and underwent curling. We separately photographed the edges of the CA at 7 and 14 days post-aggregation induction using an inverted microscope (Leica, Germany). Three microscopic fields were randomly selected, the area of the edge curling regions of the CA was measured, and the corresponding percentage relative to the total area of the captured field of view was calculated (% total area). The CA were carefully detached from the well bottom in a stepwise manner starting from the peripheral area toward the center using sterile forceps, and then harvested for subsequent experiments. RNA extraction and quantitative real-time polymerase chain reaction (qRT-PCR) Total RNA was extracted using the MIsZol Kit (Mishushengwu, China), followed by further purification via phenol‒chloroform extraction. Complementary DNA (cDNA) was synthesized using TaqMan Reverse Transcription Reagents (Takara, Japan). The genes were amplified and analyzed via Bio-Rad SYBR Green Premix Ex Taq (Takara, Japan) on a Real-Time PCR System (CFX96, Bio-Rad, USA). GAPDH was used as the internal control, and the relative expression level of the target gene was calculated using the 2⁻ΔΔCt method, as described previously [58]. The primer sequences are provided in Table S2. Western blot analysis Western blot analysis was performed as previously reported [41]. Briefly, RIPA cell lysis buffer (Beyotime Biotechnology, China) was added to the CA samples, followed by lysis on ice for 30 min. The lysed cells were repeatedly scraped with a cell scraper and collected into 1.5 mL Eppendorf tubes. The lysates were centrifuged at 12,000 rpm for 15 min; the supernatants were collected, and the protein concentrations were determined using the BCA protein assay kit (Beyotime Biotechnology, China). Equal amounts of protein samples were mixed thoroughly with 5× loading buffer and PBS, then loaded onto precast sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels (Epizyme Biotechnologies Co., Ltd., China). After protein separation via electrophoresis, the proteins were transferred onto polyvinylidene fluoride (PVDF) membranes (Roche, Switzerland). The PVDF membranes were blocked with 5% skimmed milk (AR Co., China) at room temperature for 2 h, followed by incubation with primary antibodies at 4 °C overnight. After washing the PVDF membranes with TBST buffer, the membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, USA) at room temperature for 1 h. Finally, the membranes were immersed in enhanced chemiluminescence (ECL) substrate solution (AR Co., China), and the protein bands were visualized using a gel imaging system (Model 4600, Tanon Science & Technology Co., Ltd., China). The antibodies used in this study were as follows: anti-E-cadherin antibody (sc-52327, Santa Cruz Biotechnology, USA, diluted 1:1000), anti-Paxillin antibody (2542, Cell Signaling Technology, USA, diluted 1:1000), anti-Laminin antibody (ab11575, Abcam, USA, diluted 1:1000), anti-RUNX2 antibody (sc-390715, Santa Cruz Biotechnology, USA, diluted 1:1000) and anti-GAPDH antibody (30201ES20, Yeasen, China, diluted 1:1000). Statistical analysis All experimental data were presented as mean ± standard deviation (SD) of at least triplicate measurements. GraphPad Prism 9.4.1 (GraphPad Software, USA) was used for statistical analysis. Normality was assessed using the Shapiro–Wilk test, and homogeneity of variances was evaluated using Brown-Forsythe test. When data satisfied normality and equal variance assumptions, statistical significance was assessed by the Student’s t-test (two-tailed) for two group comparisons and one-way ANOVA followed by the Tukey’s post-hoc test for multiple group comparisons. P < 0.05 were considered statistically significant. All mice were randomly assigned, and the investigators who analyzed the samples were blinded to the group allocations. Results Single-cell transcriptomic profiling of mouse long bones identifies Gli1high SSPCs with high expression of pro-angiogenic and pro-osteogenic genes To further clarify the distribution and functional gene expression profile of Gli1+ cells in vivo, we reanalyzed a published single-cell RNA sequencing dataset of mouse long bones [49]. Stromal cells isolated from mouse long bones and bone marrow were classified into 11 distinct cell types, predominantly comprising SSPCs, fibroblasts, SSPC-descendent osteolineage cells (OLCs), bone marrow-derived endothelial cells (BMECs), chondrocytes, erythrocytes and immune cells (Fig. 1A). Given that Gli1+ cells have been characterized as SSPC-like cells [33, 59–61], we further examined the expression pattern of Gli1 within the SSPC subset. Gli1 was primarily enriched in SSPCs, fibroblasts, and chondrocytes (Fig. 1B), and the SSPC identity of Gli1+ cells was validated by the expression of SSPC-specific markers (Fig. 1C–F). To resolve the distribution of Gli1 across SSPC subsets, we performed additional dimensionality reduction and sub-clustering of the previously identified SSPC population (Fig. S2A), which yielded 10 transcriptionally distinct clusters. The expression profiles of canonical marker genes across these clusters were visualized by violin plots (Fig. S2B), and GO enrichment analysis was conducted on their DEGs to infer their biological identities (Fig. S3). Projecting Gli1 expression onto the SSPC population revealed marked heterogeneity across these cells (Fig. 1G), with substantial variation in Gli1 enrichment even among Gli1+ SSPCs (Fig. 1A, G). We therefore stratified Gli1+ SSPCs into two subgroups based on Gli1 expression level: Gli1high SSPCs (top 25% of Gli1 expression) and Gli1low SSPCs (bottom 25%), as shown in the violin plot (Fig. 1H). To investigate functional differences between these subsets, we performed differential gene expression analysis (Fig. 1I). Gli1high SSPCs highly expressed genes implicated in osteogenesis (e.g., Stat3, Runx2), ECM synthesis (Sod2, Chsy1), cytoskeletal regulation (Stat3, Actb), and angiogenesis (Hdgf, Vezf1). In contrast, Gli1low SSPCs exhibited elevated expression of Tnfrsf11b, Sirt2, and Bnip3, which collectively suppress osteogenic differentiation, cell proliferation, and angiogenesis (Fig. 1I). Gli1+ cells in the metaphysis of long bones exhibit spatial coupling with RUNX2+ cells and type H vessels Based on the distinct gene signatures of Gli1high and Gli1low SSPCs, we performed GO enrichment analysis to further delineate their functional differences. Compared with Gli1low SSPCs, genes upregulated in Gli1high SSPCs were significantly enriched for GO terms associated with “bone development,” “osteoblast differentiation,” “cartilage development,” and “bone mineralization,” highlighting their robust osteogenic and chondrogenic potential. Enrichment was also observed for terms related to angiogenesis, including “endothelium development,” “endothelial cell migration,” “cellular response to vascular endothelial growth factor stimulus,” and “vascular endothelial growth factor receptor signaling pathway,” indicative of the pro‑angiogenic properties of Gli1high SSPCs. Furthermore, terms such as “extracellular matrix organization,” “cell-substrate adhesion,” “cell adhesion,” and “regulation of cell‑matrix adhesion” were prominently enriched, underscoring the superior capacity of Gli1high SSPCs for ECM synthesis and cell adhesion. Notable enrichment of cell proliferation-related terms, including “cell cycle,” “cell proliferation,” and “cell division,” reflected their enhanced proliferative capacity. Finally, enrichment in “actin cytoskeleton organization”, “microtube-based process”, “microtubule cytoskeleton organization”, and “positive regulation of cytoskeleton organization” suggested an enhanced ability to regulate cytoskeletal remodeling (Fig. 2A). Gli1+ cells in the metaphysis of long bones are spatially closely associated with RUNX2+ cells and type H vessels. A GO enrichment analysis of genes highly expressed in Gli1high versus Gli1low SSPCs. B GO enrichment analysis of genes highly expressed in Gli1low versus Gli1high SSPCs. C Distribution of Gli1+ and RUNX2+ cells in the metaphysis of long bones. The lower panel shows a magnified view of the yellow boxed area in the upper panel, yellow arrow indicates Gli1+RUNX2+ double-positive cells. Dotted lines outline margins of growth plates in the metaphysis. Scale bar in the upper panel, 100 μm. Scale bar in the lower panel, 50 μm. D Quantitative analysis of the fluorescence area ratio of RUNX2+ cells adjacent to Gli1+ cells versus those non-adjacent to Gli1+ cells. n = 3 per group. E Distribution of Gli1+ cells and CD31+EMCN+ vessels in the metaphysis of long bones. The lower panel shows a magnified view of the yellow boxed area in the upper panel, yellow arrow indicates Gli1+CD31+EMCN+ triple-positive cells. Scale bar in the upper panel, 100 μm. Scale bar in the lower panel, 50 μm. F Quantitative analysis of the fluorescence area ratio of CD31+EMCN+ vessels adjacent to Gli1+ cells versus those non-adjacent to Gli1+ cells. n = 3 per group. Data are presented as Mean ± SD. Statistical analyses were performed by Student’s t test. *P < 0.05 Conversely, relative to Gli1high SSPCs, Gli1low SSPCs exhibited significant enrichment in terms associated with “programmed cell death”, “cell death”, “death”, “apoptotic process”, “negative regulation of cell proliferation”, “aging”, “negative regulation of stem cell proliferation” and “negative regulation of cell cycle”, reflecting their pro-apoptotic and pro-death functional characteristics. Enrichment for “negative regulation of cartilage development” indicated that Gli1low SSPCs function as inhibitor of osteogenesis. Terms including “negative regulation of intracellular signal transduction”, “negative regulation of metabolic process”, “negative regulation of response to stimulus”, “negative regulation of cytoplasmic transport” and “negative regulation of response to wounding” were also significantly enriched, which further confirmed that Gli1low SSPCs exert negative regulatory effects on intracellular signal transduction and cellular stress responses. Finally, enrichment of the term “negative regulation of cell adhesion” suggested an inhibitory capacity on cell adhesion (Fig. 2B). Furthermore, we stratified cells into Gli1high and Gli1low subgroups using a range of Gli1 expression thresholds. Consistent with our primary findings, the functional superiority of Gli1high SSPCs over Gli1low SSPCs—particularly in osteogenic differentiation, cell proliferation, and angiogenesis—remained robust across all threshold tested and was not substantially affected by the specific cutoff value (Fig. S4). Building on these single-cell RNA sequencing findings, we sought to explore the distribution of Gli1+ cells and their spatial relationship with osteogenic and angiogenic markers in vivo. Previous studies have demonstrated that Gli1+ cells are enriched in the metaphysis of long bones [35]; thus, we focused our analysis on the metaphysis below the growth plate. Immunofluorescence staining of this region in 8-week-old mice revealed abundant localization of both Gli1+ cells and RUNX2+ cells. In areas adjacent to the growth plate, Gli1+ cells and RUNX2+ cells were in close spatial proximity (Fig. 2C, D). Notably, the fraction of RUNX2+ area adjacent to Gli1+ cells was significantly greater near the growth plate than in regions distant from it (Fig. S5). Partial co-localization of these two populations was also observed (Fig. 2C). Type H vessels represent a specific capillary subtype recently identified in long bones, and previous studies, including our own, have established a functional coupling between type H vessels and osteogenesis in long bones [36, 43, 45]. Therefore, we further investigated the spatial relationship between type H vessels and Gli1+ cells. Immunofluorescence staining showed that type H vessels, identified by CD31 and EMCN co-expression, were densely distributed in the metaphysis, with the majority located in close association with Gli1+ cells (Fig. 2E). Quantitative analysis confirmed that the number of CD31+EMCN+ vessels adjacent to Gli1+ cells was significantly higher than that of vessels located farther away (Fig. 2F). Co-localization of Gli1+ cells and type H vessels was also detected in specific regions (Fig. 2E). Collectively, these results indicate a tight spatial coupling between type H vessels and Gli1+ cells in the femoral metaphysis. Isolation of Gli1+ cells and construction of CA To further characterize the functional differences between Gli1+ and Gli1− SSPCs, we utilized Gli1-CreERT2;mT/mG transgenic mice to label Gli1+ cells. Primary cells isolated from femurs were expanded in vitro. After identity confirmation by surface marker analysis (Fig. S6), fluorescence-activated cell sorting (FACS) was used to isolate GFP+ Gli1+ cells, while GFP− cells were defined as Gli1− cells (Fig. 3A). Flow cytometric analysis demonstrated that the primary SSPCs exhibited high viability (88.01%), and Gli1+ cells accounted for 2.62% of the total primary SSPC population (Fig. 3B). Prior to sorting, Gli1+ cells were sparsely distributed among the primary SSPCs, with Gli1− cells constituting the majority population. Post-sorting, Gli1+ cells became the predominant cell type in the sorted fraction (Fig. 3C, D). Isolation of Gli1+ cells and construction of CA. A Schematic diagram of primary culture, Gli1+ cell flow cytometric sorting, and CA construction. B Flow cytometric gating strategy for isolation and identification of Gli1+ cells in mouse bone marrow. C Immunofluorescence images of unsorted cell, Gli1+ cell, and Gli1− cell. Scale bar, 200 μm. D Quantitative analysis of GFP+ fluorescence area ratio in unsorted cell, Gli1+ cell, and Gli1− cell. n = 3 per group. E Photographs of unsorted CA, Gli1+ CA, and Gli1− CA at the edge of culture dishes after 7 or 14 days of aggregates induction. The area enclosed by the yellow dashed line is the curled region of the aggregates. F Quantitative analysis of edge curling in unsorted CA, Gli1+ CA, and Gli1− CA after 7 days of aggregates induction. n = 3 per group. G Quantitative analysis of edge curling in unsorted CA, Gli1+ CA, and Gli1− CA after 14 days of aggregates induction. n = 3 per group. H HE staining images of unsorted CA, Gli1+ CA, and Gli1− CA. Scale bar, 100 μm. I Masson’s trichrome staining images of unsorted CA, Gli1+ CA, and Gli1− CA. Scale bar, 100 μm. Data are presented as Mean ± SD. Statistical analyses were performed by one-way ANOVA. *P < 0.05, **P < 0.01, ns, not significant (P > 0.05) Our previous studies have demonstrated that CA possess unique regenerative advantages attributed to their abundant ECM deposition [62]. Although Gli1+ cells are widely recognized for their roles in maintaining tissue homeostasis, responding to injury, and promoting bone regeneration [33, 35, 59], the construction of CA using Gli1+ SSPCs has not been previously reported. Therefore, we induced and fabricated three types of CA: unsorted CA, Gli1+ CA, and Gli1− CA. We observed that with prolonged induction time, the thickness of CA cultured in vitro gradually increased, accompanied by marginal contraction and curling in some aggregates (Fig. 3E). Compared with unsorted CA, Gli1+ CA showed no significant difference in the edge-curling effect. However, after 14 days of aggregate induction, Gli1+ CA exhibited a significantly enhanced edge-curling effect relative to Gli1− CA (Fig. 3F, G). HE and Masson’s trichrome staining confirmed abundant ECM deposition within the CA (Fig. 3H, I). Gli1+ CA highly express osteogenic and ECM-related molecules Based on the aforementioned observation that Gli1+ CA were rich in ECM, we further investigated their potential to regulate cytoskeletal remodeling and osteogenic differentiation. Immunofluorescence staining revealed that Gli1+ CA exhibited significantly higher levels of the cytoskeletal protein F-actin compared with unsorted CA (Fig. 4A–B). Moreover, Gli1+ CA showed elevated expression of both F-actin and the osteogenic transcription factor RUNX2 relative to Gli1− CA (Fig. 4A, C). Additionally, Gli1+ CA displayed a slightly smaller cell size, denser cellular composition, and more uniform morphology than both unsorted and Gli1− CA, suggesting higher cellular homogeneity and stronger intercellular adhesion. qRT-PCR analysis demonstrated that the mRNA expression levels of Col1a1, Cdh1, Tuba1a, Actb, and Fn1 were significantly upregulated in Gli1+ CA relative to unsorted CA and Gli1− CA, suggesting that Gli1+ CA have enhanced capacity to promote osteogenesis, cytoskeletal remodeling, and ECM synthesis (Fig. 4D–H). These findings were further validated by western blot analysis, which confirmed that Gli1+ CA showed elevated protein expression of E-cadherin, Paxillin, Laminin, and RUNX2 compared with unsorted CA and Gli1− CA (Fig. 4I–M). Gli1+ CA highly express osteogenic and ECM-related molecules. A Expression of F-actin and RUNX2 in unsorted CA, Gli1+ CA, and Gli1− CA. Scale bar, 200 μm. B Quantitative analysis of F-actin+ fluorescence area ratio in unsorted CA, Gli1+ CA, and Gli1− CA. n = 3 per group. C Quantitative analysis of RUNX2+ fluorescence area ratio in unsorted CA, Gli1+ CA, and Gli1− CA. n = 3 per group. D–H qRT-PCR analysis of Col1a1, Cdh1, Tuba1a, Actb and Fn1 mRNA levels in unsorted CA, Gli1+ CA, and Gli1− CA. n = 3 per group. I Western blot detection of E-cadherin, Paxillin, Laminin and RUNX2 proteins in unsorted CA, Gli1+ CA, and Gli1− CA. GAPDH served as loading control. Full-length blots are presented in Fig. S7-11. J–M Quantitative western blot analysis of E-cadherin, Paxillin, Laminin and RUNX2 protein levels in unsorted CA, Gli1+ CA, and Gli1− CA. n = 3 per group. Data are presented as Mean ± SD. Statistical analyses were performed by one-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001, ns, not significant (P > 0.05) Gli1+ CA exhibit potent pro-angiogenic capacity Heatmap analysis of the single-cell RNA sequencing data presented in Fig. 1 indicated that Gli1high SSPCs highly express pro-angiogenic genes. GO enrichment analysis further suggested that Gli1high SSPCs possess enhanced pro-angiogenic potential. Additionally, immunofluorescence results in Fig. 2 demonstrated a tight spatial coupling between type H vessels and Gli1+ cells in the femoral metaphysis. Based on these findings, we conducted functional experiments to verify whether Gli1+ cells indeed highly express pro-angiogenic genes and are capable of promoting endothelial tube formation. qRT-PCR analysis revealed that the mRNA expression levels of Angpt1, Fgf2, Vegfa, and Apln were significantly increased in Gli1+ CA compared with unsorted CA and Gli1− CA, indicating that Gli1+ CA have robust paracrine activity and can secrete a variety of pro-angiogenic factors to facilitate angiogenesis (Fig. 5A–D). To evaluate the pro-angiogenic capacity of Gli1+ CA, we treated ECs with the CM collected from cultured CA and observed endothelial tube formation at different time points. We found that treatment with Gli1+ CA-CM for 2 h significantly increased the number of endothelial cell junctions (Fig. 5E, F) and the total tube length (Fig. 5E, G) compared with unsorted CA-CM and Gli1− CA-CM. Extending the treatment duration to 3 h further enhanced the pro-angiogenic effect of Gli1+ CA-CM, as evidenced by a more substantial increase in junction number (Fig. 5H, I) and total tube length (Fig. 5H, J) relative to the other two groups. Gli1+ CA exhibits pro-angiogenic capacity. A–D qRT-PCR analysis of Angpt1, Fgf2, Vegfa and Apln mRNA levels in unsorted CA, Gli1+ CA, and Gli1− CA. n = 3 per group. E Representative images of tube formation assay of ECs treated 2 h by CM from unsorted CA, Gli1+ CA, and Gli1− CA. The lower panel shows a magnified view of the yellow boxed area in the upper panel. Scale bar in the upper panel, 500 μm. Scale bar in the lower panel, 250 μm. F Quantification of the formed junctions of ECs in (E), as indicated by red circles. n = 3 per group. G Quantification of the total tube length of ECs in (E), as indicated by yellow and green lines. n = 3 per group. H Representative images of tube formation assay of ECs treated 3 h by CM from unsorted CA, Gli1+ CA, and Gli1− CA. The lower panel shows a magnified view of the yellow boxed area in the upper panel. Scale bar in the upper panel, 500 μm. Scale bar in the lower panel, 250 μm. I Quantification of the formed junctions of ECs in (H), as indicated by red circles. n = 3 per group. J Quantification of the total tube length of ECs in (H), as indicated by yellow and green lines. n = 3 per group. Data are presented as Mean ± SD. Statistical analyses were performed by one-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001 Implantation of Gli1+ CA promotes bone defect regeneration Finally, from a therapeutic perspective, we explored whether Gli1+ CA could exert superior efficacy in promoting bone defect regeneration compared with unsorted CA and Gli1− CA. The pre-induced and fabricated CA were rolled into an appropriate size and implanted into the femoral metaphyseal defect region of mice. Samples were collected on day 14 for micro-CT analysis and immunofluorescence staining, and on day 21 for additional immunofluorescence staining. Micro-CT-based 3D reconstruction and quantitative analysis revealed varying degrees of bone regeneration across all groups. The Gli1+ CA group exhibited the most robust regenerative response, with extensive de novo bone formation throughout the defect site (Fig. 6A–C). Quantitative analysis demonstrated that, compared with unsorted CA and Gli1− CA, the Gli1+ CA group displayed significantly higher BV/TV and Tb.N, significantly lower BS/BV, and no significant differences in Tb.Th or Tb.Sp (Fig. 6D–H). Immunofluorescence staining revealed that at 14 days post-implantation, Gli1+ CA implantation resulted in a significant increase in the number of RUNX2+ cells within the bone defect area compared with the unsorted CA and Gli1− CA groups (Fig. 6I, J). In addition, the proportion of CD31+EMCN+ vessels in the regenerated bone tissue was markedly higher in the Gli1+ CA group than in the other two groups (Fig. 6I, K). To further evaluate the impact of CA implantation on the late stage of bone defect repair, we analyzed samples harvested at 21 days post-implantation. The results exhibited a similar trend to those observed at 14 days: the Gli1+ CA group still contained the highest number of RUNX2+ cells (Fig. 6L, M) and the largest proportion of CD31+EMCN+ vessels (Fig. 6L, N) within the bone defect area compared with the unsorted CA and Gli1− CA groups. Collectively, these results suggest that during the middle and late stages of bone defect repair, Gli1+ CA may exert their pro-osteogenic and pro-angiogenic effects by paracrine secretion of osteogenic, ECM-related, and angiogenic factors, thereby promoting type H vessel formation and osteogenesis. Implantation of Gli1+ CA promotes bone defect regeneration. A–C Representative micro-CT images of femoral defects 14 days after implantation of unsorted CA, Gli1+ CA, and Gli1− CA. (Left) Sagittal sections; (Right) 3D reconstructed images of the defective area, illustrating the overall healing. Scale bar, 200 μm. D–H Quantitative analysis of BV/TV, BS/BV, Tb.Th, Tb.N and Tb.Sp in the femoral defects. n = 6 per group. I Immunofluorescence staining of RUNX2, CD31, and EMCN in femoral defects 14 days after implantation of unsorted CA, Gli1+ CA, and Gli1− CA. Scale bar, 100 μm. J Quantitative analysis of RUNX2+ area ratio in femoral defects 14 days post-implantation of unsorted CA, Gli1+ CA, and Gli1− CA. n = 3 per group. K Quantitative analysis of CD31+EMCN+ area ratio in femoral defects 14 days post-implantation of unsorted CA, Gli1+ CA, and Gli1− CA. n = 3 per group. L Immunofluorescence staining of RUNX2, CD31, and EMCN in femoral defects 21 days after implantation of unsorted CA, Gli1+ CA, and Gli1− CA. Scale bar, 100 μm. M Quantitative analysis of RUNX2+ area ratio in femoral defects 21 days post-implantation of unsorted CA, Gli1+ CA, and Gli1− CA. n = 3 per group. N Quantitative analysis of CD31+EMCN+ area ratio in femoral defects 21 days post-implantation of unsorted CA, Gli1+ CA, and Gli1− CA. n = 3 per group. Data are presented as Mean ± SD. Statistical analyses were performed by one-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001, ns, not significant (P > 0.05) Discussion Developmental condensation, the core initial mechanism of embryonic bone formation, establishes the structural and functional basis for ordered bone tissue construction by mediating tight cell-cell adhesion, coordinated signal transduction, and directional differentiation [16, 21–23]. This mechanism also provides a critical biomimetic paradigm for regenerative therapies targeting adult bone defects. Guided by the concept of “developmental condensation-facilitated regeneration”, our research group has applied mesenchymal stromal/progenitor cell-derived CA in clinical trials, achieving the first successful regeneration of dental pulp tissue, as well as the functional revitalization of avulsed teeth, periodontal tissues and alveolar bone [31, 32]. However, emerging evidence highlights the profound heterogeneity of mesenchymal stromal/progenitor cell populations, with distinct subpopulations differing markedly in proliferation, osteogenic potential, and ECM secretory capacity [13, 14, 63, 64]. Traditional unsorted CA, which inevitably incorporate non-regenerative subpopulations, are limited by suboptimal regenerative efficiency and imprecise functional regulation [13–15]. Therefore, precise sorting of SSPC subsets, identification of pro-regenerative subpopulations, and subsequent fabrication of subpopulation-specific CA represent an inevitable strategy to advance bone regenerative therapy [15, 65]. In this study, we overcame this limitation through genetic lineage tracing and successfully isolated endogenous Gli1+ SSPCs with pro-regenerative potential, from which we fabricated Gli1+ CA. We further demonstrated that Gli1+ CA exhibited superior regenerative capacity both in vitro and in vivo, thus providing a novel, targeted strategy for bone defect regeneration. Beyond addressing the knowledge gap in Gli1+ CA-mediated bone regeneration, this study proposes an innovative paradigm termed “precision screening of endogenous SSPC subsets coupled with targeted aggregate fabrication,” thereby offering a more precise therapeutic direction for bone regenerative therapy. Gli1+ SSPCs have been reported to be involved in bone homeostasis maintenance and defect repair [59, 61]. The present study further demonstrated, through single-cell RNA sequencing analysis, that compared with Gli1high SSPCs, the Gli1low subpopulation highly expressed genes predominantly enriched in biological processes related to promoting cellular senescence and death, inhibiting osteogenesis and angiogenesis, suppressing intracellular signal transduction, and regulating cellular stress responses. This observation suggests that Gli1low and Gli1high SSPCs may exert opposing functional effects during bone regeneration. This finding raises a critical question: why does the body retain such a seemingly detrimental Gli1low SSPC subpopulation within the bone tissue microenvironment? Given the dynamic equilibrium of bone development and regeneration, we hypothesize that Gli1low SSPCs may function as a “brake” to exert key regulatory effects. Specifically, during the regenerative phase, their highly expressed cell death-related genes may help regulate the initial post-injury inflammatory response by programmatically inducing apoptosis or death of cells, including immune cells, at different stages of regeneration, thereby creating favorable conditions for subsequent Gli1high SSPCs-mediated bone regenerative repair. Nevertheless, the specific molecular pathways underlying this regulatory mechanism, as well as the interplay between Gli1high and Gli1low SSPCs, require further investigation using approaches such as single-cell RNA sequencing, cell co-culture, and conditional knockout experiments. As a specialized capillary subtype that has garnered considerable attention in recent years, type H vessels play pivotal roles in bone development, homeostasis, and regeneration [43, 45]. Our prior work further identified their broader involvement in tissue repair, establishing them as regeneration-associated vasculature [46, 66]. By orchestrating a “vascular–stem cell” synergistic microenvironment, type H vessels directly govern stem cell proliferation, differentiation, and bone regeneration, thereby serving as the central nexus that couples angiogenesis with osteogenesis [44, 67–69]. Consistently, genetic or pharmacological ablation of CD31+EMCN+ vessels impairs normal tissue development and compromises defect healing [36, 43, 66, 69]. In the present study, we observed that type H vessels were preferentially enriched in the vicinity of Gli1+ SSPCs at bone defect sites, suggesting a potential functional interplay between these two components. Building on this observation, we further demonstrated, through aggregate engineering and in vivo functional assays, that Gli1+ CA promote type H vessel formation and accelerate bone defect regeneration. These findings are consistent with our earlier lineage-tracing studies, which showed that targeted ablation of Gli1+ cells significantly suppresses both type H vessel formation and bone regeneration during femoral defect repair [36]. Mechanistically, we previously demonstrated that Gli1+ SSPCs maintain type H vessel integrity and promote neovascularization through coordinated Gli1 and HIF-1α signaling [36]. Complementing this, the PDGF/PDGFRβ pathway has been shown to drive type H vessel formation via paracrine secretion of PDGF-BB [70]. Similarly, the Slit3/Robo1 signaling axis acts as a potent regulator of type H angiogenesis and bone formation, playing a pivotal role in osteogenesis-angiogenesis coupling [71, 72]. In further support, a recent study revealed that occlusal loading preserves bone homeostasis by upregulating both type H vessels and osteoprogenitors through the PIEZO1/Ca2+/HIF-1α/SLIT3 signaling pathway [73]. Genetic evidence from transgenic mouse models further substantiates this functional crosstalk: endothelial Notch activation induces Noggin expression in osteogenic progenitor cells, thereby facilitating their differentiation; conversely, endothelial-specific ablation of Notch signaling impairs angiogenesis and leads to progressive bone loss [44]. Collectively, these findings indicate that Gli1+ SSPCs, as representative cells of the osteogenic lineage, promote type H vessel formation and, consequently, bone tissue regeneration through paracrine signaling and the integrated regulation of multiple molecular pathways. Given the high evolutionary conservation of skeletal development, bone regeneration mechanisms, and Gli1-dependent skeletal stem cell function across mammals, our findings are likely generalizable across species [34, 59, 61]. However, because Gli1 functions as a transcription factor rather than a cell-surface marker, its translational potential for clinical application in humans warrants cautious interpretation [74]. We therefore propose that identifying or enriching equivalent Gli1+ SSPCs from clinical tissues is a critical long-term goal requiring systematic investigation. This goal may be advanced through the following integrated strategies: First, candidate cell-surface markers specifically expressed by Gli1+ cells will be identified and validated using well-characterized transgenic mouse models. Second, the mechanistic role of Gli1 signaling in regulating the expression of these surface markers will be elucidated. Third, the validated markers will be used to isolate Gli1+ cells by flow cytometry. Finally, the specificity and clinical utility of these markers will be rigorously validated in human tissues. This study systematically validated the superior therapeutic efficacy of Gli1+ CA in repairing bone defects. The core conclusions are as follows. First, we clarified the functional differences between Gli1+ and Gli1− SSPCs and verified that, compared with Gli1low SSPCs, Gli1high SSPCs highly expressed genes related to osteogenesis, angiogenesis, and ECM synthesis. Second, using genetic lineage tracing, we specifically identified and isolated endogenous Gli1+ SSPCs and subsequently generated high-purity Gli1+ CA, thereby overcoming the limitations inherent to conventional heterogeneous stem CA. Third, in vivo experiments demonstrated that Gli1+ CA markedly enhanced type H vessel formation and accelerated bone defect regeneration relative to both unsorted CA and Gli1− CA. Nevertheless, this study has several limitations. First, although we showed that Gli1+ CA promote bone defect regeneration alongside type H vessel neovascularization, the precise molecular mechanisms governing this process remain incompletely understood. Importantly, without specific blockade experiments targeting type H vessels, we cannot definitively conclude that the pro-regenerative effects of Gli1+ CA are mediated exclusively by type H vessel formation. Second, the long-term biosafety profile of Gli1+ CA transplantation remains to be fully characterized. Addressing these knowledge gaps will be a key focus in our future investigations. Conclusions In summary, guided by the concept of “developmental condensation-facilitated regeneration”, our findings not only address a critical knowledge gap in Gli1+ CA-mediated bone regeneration, but also proposes a novel strategy termed “precision screening of endogenous SSPC subsets coupled with targeted aggregate fabrication”, thereby offering a more precise therapeutic direction for the regenerative treatment of bone defects. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Acknowledgements The authors declare that they have not use AI-generated work in this manuscript. Abbreviations Bone volume per tissue volume Bone surface/bone volume Cell aggregates Differentially expressed genes Extracellular matrix Endothelial cells Ethylenediaminetetraacetic acid Endomucin Fluorescence-activated cell sorting Fetal bovine serum Gene Ontogeny Hematoxylin–eosin Green fluorescent protein Optimal cutting temperature Phosphate-buffered saline Polyvinylidene fluoride Quantitative real-time polymerase chain reaction Runt-related transcription factor 2 Skeletal stromal/progenitor cells Trabecular number Trabecular separation/spacing Trabecular thickness Uniform manifold approximation and projection Author contributions Chao Ma, Hao-Kun Xu and Ji Chen designed the study, performed single-cell RNA sequencing and drafted the manuscript. Yu-Ru Gao and Hao Wang established the femoral defect model, performed histological experiments and immunofluorescence staining. Bing-Dong Sui, Shan-Shan Huang, Yi Zhang, Lu Liu and Xiao-Hui Zhang performed cell isolation, tube formation assay and induction of aggregate formation. Shu-Juan Xing performed flow cytometry for cell identification, contributed to experimental data analysis, and revised the manuscript. Yuan-Yuan Li, Yi-Han Xiao, Ya-Nan Zheng and Li-Heng Ren performed qRT-PCR, western blot analysis and statistical analysis. Yan Jin and Chen-Xi Zheng conceived the project and supervised the experiments. All authors reviewed and approved the final manuscript. Funding This work was supported by grants from the National Natural Science Foundation of China (82401201, 82371020, 82301028), Project of State Key Laboratory of Oral & Maxillofacial Reconstruction and Regeneration (2024MS04), the China Postdoctoral Science Foundation (BX20230485) and Partner Laboratory Cooperation and Exchange Program Project of Fourth Military Medical University (2024HB014). Data availability Information required to reanalyze the data reported in this work is available from the corresponding authors upon reasonable request. Declarations Ethics approval and consent to participate This research does not contain clinical experiments. The laboratory animals were handled in accordance with the “Guidelines for the Care and Use of Laboratory Animals” and the “Animal Welfare Act in China”. The animal experiments were approved by the Institutional Animal Care and Use Committee of The Fourth Military Medical University (Title: Mechanisms underlying sympathetic nerve activation in cranial sutures drives Gli1+ MSCs mobilization to guide cranial bone defect regeneration; Approval No. K9-2024-030; Date of approval: 08/03/2024). Consent for publication All authors confirm their consent for publication. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Chao Ma, Yu-Ru Gao, and Hao Wang contributed equally to this study. Contributor Information Yan Jin, Email: yanjin@fmmu.edu.cn. Chen-Xi Zheng, Email: chenxizheng@fmmu.edu.cn. Hao-Kun Xu, Email: xuhaokun@fmmu.edu.cn. Ji Chen, Email: jim117cj@sohu.com. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement Information required to reanalyze the data reported in this work is available from the corresponding authors upon reasonable request.