New insights into the mechanism of microvascular rarefaction via endothelial-mesenchymal transition in steroid-induced osteonecrosis of femoral head Authors: Xiaoqiang Yang, Peng Yang, Jiaru Liao, Fan Yang, Mincong He, Ziqi Li, Qiushi Wei, Wei He Published: 2025 ABSTRACT AbstractThis study aimed to investigate the potential role of endothelial-to-mesenchymal transition (EndMT) in steroid-induced osteonecrosis of the femoral head (SONFH), evaluate its association with microvascular rarefaction and tissue damage, and preliminarily explore whether glucocorticoid-induced oxidative stress may be involved in this process. A combination of single-cell RNA sequencing (scRNA-seq), in vitro experiments, and animal models were employed. scRNA-seq was performed on SONFH cystic lesions to identify endothelial subpopulations and trace developmental trajectories. In vitro, human umbilical vein endothelial cells (HUVECs) were treated with methylprednisolone and antioxidant (N-acetylcysteine) intervention to assess EndMT features and endothelial function. In vivo, a glucocorticoid-induced SONFH mouse model was established with or without antioxidant intervention to evaluate changes in microvascular density, EndMT occurrence, and oxidative stress levels. scRNA-seq analysis revealed endothelial subpopulations with EndMT-like features in the cystic areas of SONFH. In vitro experiments demonstrated that glucocorticoid treatment downregulated endothelial markers and upregulated mesenchymal markers in HUVECs, accompanied by impaired angiogenic function, and antioxidant therapy improves these conditions. In vivo, glucocorticoid administration led to significant microvascular rarefaction and increased EndMT marker expression in the femoral head, along with elevated oxidative stress; these changes were partially alleviated by antioxidant treatment. This study is the first to identify a potential involvement of EndMT in the pathogenesis of SONFH, highlighting its close association with microvascular rarefaction. Targeting EndMT and its upstream regulators, such as oxidative stress, may offer novel strategies for the early intervention and treatment of SONFH.Keywords: EndMT, Microvascular rarefaction, SONFHSubject terms: Cell biology, Diseases, Pathogenesis Introduction Steroid-induced osteonecrosis of femoral head (SONFH) is a bone damaging disease caused by prolonged use of glucocorticoids, primarily characterized by necrosis of the femoral head, bone collapse, and hip joint dysfunction1,2. The femoral head is a crucial load-bearing structure in the human body, with a unique anatomical structure that provides it with distinctive blood supply characteristics3. Due to the limited collateral circulation in the femoral head, any disruption to blood flow can easily result in local ischemia, necrosis, and bone collapse. Therefore, vascular damage, particularly the destruction of the microvascular network, is a significant pathological feature of femoral head necrosis4. Although various mechanisms of SONFH have been revealed in prior research, the precise pathophysiological mechanisms remain incompletely understood, especially with regard to the mechanisms of vascular injury. In recent years, increasing attention has been focused on the role of microvascular damage in SONFH5–7. Microvascular rarefaction refers to the reduction in the density of the microvascular network, a decrease in the number of blood vessels, and restricted blood flow8,9. This phenomenon is commonly observed in various pathological conditions, such as diabetes10,11, hypertension12, atherosclerosis13, and aging. Microvascular rarefaction not only exacerbates tissue hypoxia and metabolic disturbances but may also promote bone necrosis by impairing bone metabolism and repair functions14. A substantial body of research8,15,16 suggests that microvascular rarefaction is closely linked to endothelial-mesenchymal transition (EndMT), a process that plays an important role in the development of numerous diseases. In addition, in the study of SONFH, we used single-cell RNA sequencing(scRNA-seq) to explore the cell types in cystic in SONFH since the cystic have an unusual repair response as we observed, and also found some unusual endothelial cell phenotypes, which have a significant impact on the prognosis of femoral head necrosis. EndMT is a cellular trans-differentiation process in which vascular endothelial cells lose their endothelial characteristics and acquire features of mesenchymal cells, such as fibroblasts or smooth muscle cells17,18. EndMT has been reported in various pathological conditions, including cardiovascular diseases19, cancer metastasis20, and fibrosis21. Multiple signaling cascades, including TGF-β/Smad, Notch, Wnt/β-catenin and other protein molecule like CD4522, have been implicated in regulating EndMT under pathological stress. These pathways are often activated by oxidative stimuli, linking redox imbalance to vascular remodeling and fibrosis. However, its involvement in the reconstruction of the femoral head vascular network and in diseases related to bone metabolism remains underexplored. The occurrence of EndMT may exacerbate ischemic bone damage and marrow adiposity. Nonetheless, whether glucocorticoids induce microvascular rarefaction in the femoral head via EndMT remains inadequately verified. Oxidative stress is considered one of the key factors in glucocorticoid-induced bone damage and antioxidant therapy can relieve the damage caused by glucocorticoids23,24. Prolonged use of glucocorticoids can lead to the accumulation of reactive oxygen species, thereby inducing damage to endothelial and other cells25. Studies26–28 have shown that oxidative stress is closely associated with the occurrence of EndMT. While several studies have investigated the mechanisms underlying SONFH, the specific mechanisms by which glucocorticoids induce microvascular rarefaction through oxidative stress and EndMT remain unclear. Therefore, the aim of this study is to explore whether oxidative stress induced by glucocorticoids triggers EndMT, contributing to microvascular rarefaction and promoting femoral head necrosis. Through this study, we aim to elucidate the pathological mechanisms of microvascular rarefaction induced by glucocorticoids, providing new theoretical foundations and experimental data to support early diagnosis and therapeutic strategies for SONFH. Materials and methods Single-cell RNA sequencing Process of tissue samples In this study, we collected a total of four clinical samples from female patients aged 39, 43, 45, and 49 years. All patients had a documented history of long-term glucocorticoid therapy for systemic lupus erythematosus and were clinically diagnosed with SONFH with cystic at ARCO stage III. All patients subsequently underwent total hip arthroplasty, from which the femoral head specimens were obtained for scRNA-seq. All procedures were performed in compliance with relevant laws and institutional guidelines and have been approved by the institutional ethics review board at the third Affiliated Hospital of Guangzhou University of Chinese Medicine (PJ-KY-20220420-013, 2022-04-20). The privacy rights of human subjects have been observed and that informed consent was obtained for experimentation with human subject. The tissue was dissociated into a single-cell suspension using mechanical and enzymatic dissociation methods. We prepared four independent single-cell suspensions from four SONFH patients and one suspension per patient (libraries prepared separately and integrated with batch correction during analysis). Surgically resected cystic tissues were trimmed in ice-cold PBS to remove clots and fat and minced to ~ 1 mm3 fragments. If trabecular bone fragments were present, an optional mild pre-treatment with 0.5 mM EDTA for 10 min (RT, gentle inversion) was used to facilitate dissociation. Tissues were digested in HBSS + 0.04% BSA containing Collagenase II 2.0 mg/mL, Dispase II 1.0 mg/mL, and DNase I 100 U/mL at 37 °C for 30–45 min with gentle agitation (~ 120 rpm), with brief trituration every 10 min. Digestion was quenched with DMEM + 10% FBS, followed by sequential filtration through 70 µm and 40 µm strainers. Red blood cells were lysed with ACK buffer for 1–2 min, and cells were washed 2–3 times with PBS + 0.04% BSA. When viability was suboptimal or debris was excessive, a commercial dead-cell removal kit was used as needed. Cell counts were obtained using Trypan Blue or AO/PI. Post-dissociation viability typically ranged from 85 to 92% (median ~ 88%); samples with viability < 80% were not taken forward for library preparation. Cells were resuspended to ~ 1,000 cells/µL in 0.04% BSA/PBS and loaded following the 10 × Genomics 3’ v3.1 workflow, targeting 8000–10,000 cells per sample. During the single cell capture and library construction process, cells were distributed into individual microreactor chambers via microfluidic chips, followed by cell lysis, mRNA capture, reverse transcription, and amplification, culminating in the generation of sequencing libraries. The library quality was assessed using Qubit and Agilent chips to ensure that the library concentration and fragment length met the required standards. Sequencing was conducted on the Illumina NovaSeq platform. All raw data have been transferred to the GEO (accession numbers: GSE290411). Cell annotation Data normalization and batch effect removal were performed using seurat software. Principal component analysis (PCA) was applied to reduce the dimensionality of the data. The first 20 principal components were selected based on the PCA results, followed by further visualization of the cell population distribution using t-SNE and UMAP. Annotation of each cell population was carried out based on known marker genes, in combination with the cellmarker database and literature, to identify the types of each cell cluster. Pseudotime analysis To explore the developmental processes and state transitions of different cell populations in the sample, pseudotime analysis was performed. Monocle 3 was used to perform pseudotime analysis on endothelial cells. By selecting variant genes and constructing developmental trajectories, cells were ordered along the developmental timeline. Dynamic changes in each cell type during development were analyzed to uncover key genes and their roles in the developmental process. Immunofluorescence Samples of the femoral head of patients who required total hip arthroplasty due to SONFH were collected, and the area that had not yet been completely necrotic was selected for tissue cutting, and the bone tissue in this area was made into paraffin sections for immunofluorescence staining. Paraffin-embedded femoral head tissue Sects. (5 μm thick) were deparaffinized in xylene and rehydrated through a graded ethanol series. Antigen retrieval was performed in citrate buffer (pH 6.0) using a microwave for 15 min, followed by cooling to room temperature. Sections were then washed with PBS and incubated with 0.3% Triton X-100 for 15 min to increase permeability. After blocking with 5% bovine serum albumin (BSA) for 1 h at room temperature to reduce nonspecific binding, the sections were incubated overnight at 4 °C with primary antibodies against CD31(Affinity, #BF0611), S100A4(Affinity, #DF6516), and COL1A1(Affinity, #AF7001). The following day, sections were washed three times with PBS and incubated with corresponding fluorescent secondary antibodies (FITC and CY3) for 1 h in the dark. Nuclei were counterstained with DAPI for 10 min, and the slides were mounted with an anti-fade mounting medium. Images were captured using a confocal laser scanning microscope. EndMT in Vitro Steroid-Induced EndMT in human umbilical vein endothelial cells (HUVECs) To more comprehensively verify whether glucocorticoids induce EndMT, HUVECs were treated with methylprednisolone. HUVECs was purchased from Shanghai Zhong Qiao Xin Zhou Biotechnology (DFSC-EC-01) . HUVECs were cultured in an incubator at 37 °C with 5% CO2, and the methylprednisolone (0.1 mM) group was treated for 7 days (GCs group). The control group received no treatment. Western blot and PCR were used to detect endothelial cell markers and EndMT markers in HUVECs from both the control and methylprednisolone groups. To investigate the temporal dynamics of EndMT, HUVECs were divided into three groups: control group (Ctrl), GCs-1 group (intervention for 3 days), GCs-2 group (intervention for 7 days). Following the interventions, cells were harvested using 0.25% trypsin–EDTA, washed with PBS, and resuspended in staining buffer. Cells were incubated with the following primary antibodies for 30 min at 4 °C in the dark: Anti-CD31-APC (endothelial marker, [BD Pharmingen, 564089]), Anti-S100A4-PE (mesenchymal marker, [Biolegend, 370010]). After staining, cells were washed, resuspended in PBS, and analyzed using a BD FACSCanto II flow cytometer (BD Biosciences). Data were processed and analyzed with FlowJo software (Version X.0.7, Tree Star Inc.). To further evaluate whether glucocorticoid treatment leads to functional alterations of endothelial cells, we performed migration assay and matrigel tube formation assays to assess the migratory and angiogenic capacity of HUVECs. Migration assay HUVECs were seeded into 6-well plates and cultured until reaching over 90% confluence. A uniform scratch was made in the cell monolayer using a sterile 200μL pipette tip. Detached cells were gently washed away with PBS, and serum-free medium containing 0.1 mM methylprednisolone (GCs group) was added. Images of the scratched area were taken at 0 h and 24 h using a phase-contrast microscope. The migration distance was measured using ImageJ software to evaluate the migration rate. Tube formation assay Matrigel (Corning, 356234) was thawed on ice and coated (50μL/well) onto pre-chilled 96-well plates, then polymerized at 37 °C for 30 min. Treated HUVECs from the Ctrl and GCs groups (0.1 mM methylprednisolone, 5 days) were resuspended and seeded at a density of 2 × 104 cells/well onto the matrigel surface. After 6 h of incubation, tubular structures were observed under an inverted microscope, then images of the 12th and 24th hours were recorded. Tube formation was quantified by ImageJ (Angiogenesis Analyzer plugin), including the number of segments and nodes. Animal model establishment and grouping All procedures were approved by the Animal Ethics Review Committee of Guangzhou University of Chinese Medicine (approval number: [20240717013]). This research followed the principles outlined in the Guide for the Care and Use of Laboratory Animals (National Institutes of Health, 8th edition, 2011)29 and the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines30. C57BL/6 male mice (24 weeks old, weighing 30–35 g) were used as experimental subjects. All mice were acclimated for 7 days prior to the experiment. The animals were randomly divided into three groups, with 8 mice in each group: Control group (Ctrl) Mice received no treatment and were injected with physiological saline for both intraperitoneal and subcutaneous injections. The volume of all mice is 0.2 mL and the injection was administered once daily. Each injection of saline was synchronized with the model group. Control group (Ctrl) Mice received no treatment and were injected with physiological saline for both intraperitoneal and subcutaneous injections. The volume of all mice is 0.2 mL and the injection was administered once daily. Each injection of saline was synchronized with the model group. SONFH model group (GCs) To simulate the clinical features of SONFH, lipopolysaccharide (LPS, Sigma-Aldrich, L2880-10MG) and methylprednisolone (MPS, MCE, HY-B0260) were administered by injection. The experimental schedule is outlined in Fig. 1. The total experimental period was 6 weeks. In the first week, LPS was injected intraperitoneally for the first 2 days at a dose of 0.25 mg/kg, followed by MPS injection subcutaneously for 3 days at a dose of 100 mg/kg. No treatment was given on the last 2 days. Starting from the second week, LPS was injected intraperitoneally for the first 2 days at a dose of 0.50 mg/kg, followed by MPS injection for 3 days at 200 mg/kg, with no treatment on the last 2 days. The total period of LPS and MPS injections lasted for 5 weeks. SONFH model group (GCs) To simulate the clinical features of SONFH, lipopolysaccharide (LPS, Sigma-Aldrich, L2880-10MG) and methylprednisolone (MPS, MCE, HY-B0260) were administered by injection. The experimental schedule is outlined in Fig. 1. The total experimental period was 6 weeks. In the first week, LPS was injected intraperitoneally for the first 2 days at a dose of 0.25 mg/kg, followed by MPS injection subcutaneously for 3 days at a dose of 100 mg/kg. No treatment was given on the last 2 days. Starting from the second week, LPS was injected intraperitoneally for the first 2 days at a dose of 0.50 mg/kg, followed by MPS injection for 3 days at 200 mg/kg, with no treatment on the last 2 days. The total period of LPS and MPS injections lasted for 5 weeks. Antioxidant intervention group (GCs + NAC) Mice in this group were treated with N-Acetylcysteine (NAC) (MCE, #HY-B0215, 300 mg/kg/day, intraperitoneal injection, 4–6th week) in addition to the MPS treatment for antioxidant intervention. Antioxidant intervention group (GCs + NAC) Mice in this group were treated with N-Acetylcysteine (NAC) (MCE, #HY-B0215, 300 mg/kg/day, intraperitoneal injection, 4–6th week) in addition to the MPS treatment for antioxidant intervention. SONFH model induction protocol. Throughout the experiment, the mice were regularly monitored for appearance, behavior, and food and water intake. After completion of the experiments, mice were euthanized by inhalation of a high concentration of isoflurane in a sealed chamber, and death was confirmed by the cessation of heartbeat and respiration, in accordance with institutional animal care guidelines. Trabecular architecture observation Micro-CT was used to observe the trabecular bone structure and to quantitatively analyze changes in the trabecular architecture, including Bone volume fraction (BV/TV), connective density (Conn. D.), structure model index (SMI), trabecular number (Tb. N), trabecular thickness (Tb. Th) and trabecular separation (Tb. Sp). Immunofluorescence double staining was used to assess the proportion of H-type blood vessels in femoral head samples from different groups. Femoral head tissues were harvested and fixed in 4% paraformaldehyde, followed by dehydration, paraffin embedding, and sectioning (5 µm thick). The sections were stained with hematoxylin and eosin (H&E) to observe morphological changes in the femoral head, particularly the evaluation of bone empty lacunae and bone marrow adiposity. Additionally, immunofluorescence staining of the femoral head in mice was performed to observe the distribution of H-type vessels using CD31 (Affinity, #BF0611) and EMCN (Affinity, #DF13357) in the Ctrl and GCs groups (the specific immunofluorescence procedure is the same as described in sect. Immunofluorescence"). EndMT detection in mice Immunofluorescence staining was employed to detect microvascular rarefaction and the expression of EndMT markers in the femoral head tissue. Blood vessels were labeled using CD31(Affinity, #BF0611), and EndMT markers mainly included ACTA2(Affinity, #AF1032), S100A4(Affinity, #DF6516), COL1A1(Affinity, #AF7001). After staining, the slides were examined under a confocal microscope to record the number and distribution of positive cells. The microvascular density and the occurrence of EndMT were analyzed and compared among the different groups. Oxidative stress level detection The expression of Nrf2 and HO-1 can be detected by immunohistochemistry to reflect the occurrence of oxidative stress and the cellular response, thereby comparing the levels of oxidative stress across different groups. In the meanwhile, testing EndMT by using Immunofluorescence staining in the group of GCs and GCs + NAC. Then, we further utilize NAC intervention methylprednisolone-induced(0.1 mM) HUVECs to observe their angiogenic capacity and changes in EndMT markers. The GCs group was treated with methylprednisolone(0.1 mM) for 7 days and the GCs + NAC group was treated with methylprednisolone(0.1 mM) on the 4th and 7th day. Data statistics and analysis All data were analyzed using GraphPad Prism (Version 10.4.1) software. An independent t-test was used to compare the mean differences between the two groups. Data are presented as the mean ± standard deviation (Mean ± SD), and a p-value of < 0.05 was considered statistically significant. Results Transcriptional features of cystic in SONFH We performed single-cell transcriptome sequencing on 37,561 cells from 4 cases of SONFH with cystic area, including 8943 endothelial cells. After dimensionality reduction and clustering, a total of 18 clusters were identified, as shown in Fig. 2a. Clusters2, 6, 10, and 15 highly expressed endothelial cell-specific genes (PECAM1, CLDN5, FLT1, RAMP2, EGFL7, PLVAP, CDH5, TIE1), and were annotated as endothelial cells, as shown in Fig. 2b. The specific expression or high expression of related genes in other cells is shown in Fig. 2c. These specific genes were combined with the biological functions of various cell types to facilitate cell identification. The cell annotation results are presented in Fig. 2d. Notably, many immune cells, endothelial cells, and chondrocytes were observed within cystic. Single-cell transcriptome analysis. (a) Dimensionality reduction and clustering identified 18 clusters. (b) Annotation of endothelial cells. (c) Specific or high expression of genes in other cell types. (d) Cell annotation results. (e) Endothelial cell subpopulations: cluster0: vein; cluster1: vein; cluster2: angiogenesis; cluster3: artery; cluster4: lymphatic; cluster5: DNA related damage and repair; cluster6: lymphatic. (f) endothelial cell subpopulations express both endothelial and mesenchymal cell genes. Based on cell types and biological functions, we performed subpopulation analysis of endothelial cells. The results revealed the presence of venous endothelial cells, arterial endothelial cells, lymphatic endothelial cells, as well as subpopulations related to angiogenesis and DNA repair, as shown in Fig. 2e. Furthermore, we found that the endothelial cell subpopulations simultaneously expressed genes related to both endothelial (PECAM1, TIE2, CDH5, FLT1, RAMP2) and mesenchymal (FAP, VIM, ACTA2, S100A4, COL1A1) cells, as shown in Fig. 2f. The results of the pseudotime analysis presented in Fig. 3a based on single-cell transcriptomic data. In the plot, the x-axis and y-axis represent the reduced dimensions (Component 1 and Component 2), with each point corresponding to a single cell. The color of the cells transitions from dark blue to light blue, indicating their pseudotime points. It is evident from the plot that the cells are distributed into distinct clusters, with a progression from the center to the periphery. This suggests dynamic changes in the cell populations during development, reflecting the transition from early to late stages. Node numbers generally indicate key time points or developmental stages in the pseudotime analysis. These nodes are critical turning points where cells transition from one state to another, often reflecting the differentiation of various cell populations. the pseudotime analysis. (a) The developmental trajectories of the cells over time, while the colorful Figure below highlights the different biological states of cells along these trajectories using color coding. (b) The increase in the expression of fibrosis-related genes. (c) Expression changes of markers related to endothelial cell function and angiogenesis in pseudotim analysis. The figure with multi-color in Fig. 3a further illustrates the distribution of different cell states. The multi-color in figure represents distinct cell states (State), with each color corresponding to a unique cell state, ranging from state 1 to state 7. The cells of each state occupy different positions along the pseudotime trajectory, indicating that different cell populations are at various biological stages during development or state transitions. For example, certain cell clusters are concentrated at earlier time points along the pseudotime trajectory, while others are more prominent at later time points. These differences suggest that the cell populations undergo distinct biological changes or transitions during their developmental process. In the pseudotime analysis, the gradual increase in the expression of fibrosis-related genes (such as COL1A1, COL1A2 and S100A4) in endothelial cells in Fig. 3b reflects the transition of cells from early migration and morphological changes to the later stages of fibrosis. These genes show enhanced expression in the later pseudotime stage, indicating their close association with the EndMT phase. Combined with Fig. 3c, we observe that endothelial cell-related genes are expressed at higher levels during the early stages, with expression gradually decreasing in the later stages. In summary, the expression patterns of fibrosis and angiogenesis-related marker genes are intertwined, suggesting that endothelial cells are in a partial EndMT phase. EndMT testing of femoral head samples The results in Fig. 4 indicate that in the regions of the femoral head of SONFH patients that have not yet undergone complete necrosis, co-expression of endothelial cell-related genes and mesenchymal-related genes can be observed (yellow arrow). This suggests that endothelial cells are undergoing partial EndMT. Additionally, some endothelial cells remain in their normal state and have not yet initiated the process of mesenchymal transition (red arrow). Co-expression of endothelial cell markers (CD31) and mesenchymal cell markers (COL1A1, S100A4) in SONFH samples. Increase of EndMT in Vitro In vitro PCR experiments demonstrated that after methylprednisolone treatment, HUVECs exhibited a decreased expression of CD31, while CDH5 expression remained largely unchanged. Meanwhile, an increase in ACTA2 and S100A4 expression was observed, with S100A4 showing a particularly significant upregulation. Western blot analysis further confirmed these findings, showing that compared to the Ctrl group, the GCs group exhibited a significant reduction in CD31 protein expression and an increase in S100A4 protein expression as shown in Fig. 5b, consistent with the PCR results as shown in Fig. 5a. Flow cytometry results demonstrated that with the prolongation of glucocorticoid intervention, CD31 expression exhibited a slight decrease, whereas S100A4 expression gradually increased, particularly after 7 days of treatment (Fig. 5c). Specifically, compared to the control group, HUVECs exposed to glucocorticoids for 3 days showed a mild reduction in CD31-positive cells and a corresponding slight increase in S100A4-positive cells. Notably, after 7 days of treatment, a significant elevation in S100A4 expression was observed, accompanied by a more pronounced decline in CD31 expression. In vitro evidence of GCs-induced EndMT and functional alterations (a) The results of PCR. (b) Western blot. (c) Flow cytometry analysis showed that the fluorescence intensity of the glucocorticoid-treated groups (GCs-1, GCs-2) in the APC-A channel was lower than that of the control group, indicating that glucocorticoids reduced the expression of endothelial cell markers. In contrast, the fluorescence intensity of the glucocorticoid-treated groups in the PE channel was significantly higher than that of the control group, suggesting that glucocorticoids significantly increased the expression of mesenchymal cell markers. (d) Representative images and quantification of HUVEC migration assay at 0 h and 24 h. (e) Tube formation assay. number of segments and nodes were analyzed. ns: p > 0.05, *p < 0.05, **p < 0.001. To evaluate the functional changes of endothelial cells upon glucocorticoid exposure, we performed migration assay and matrigel tube formation assays. As shown in Fig. 5d, control group HUVECs effectively migrated to close the scratch area within 24 h, whereas the GCs-treated cells displayed increase migratory ability. Quantitative analysis using ImageJ revealed higher migration rates in the GCs group compared to the control group (p < 0.05). The tube formation assay further demonstrated (Fig. 5e) that control HUVECs formed well-organized, interconnected capillary-like structures on matrigel within 12 and 48 h, while the GCs-treated group showed a substantial reduction in tube formation capacity. The number of segments, and nodes were significantly decreased in the GCs group (p < 0.01). These results indicate that glucocorticoid treatment impairs angiogenic functionality in endothelial cells, supporting the occurrence of EndMT. Disorders of trabecular architecture Except for the Ctrl group, all other mice exhibited a noticeable reduction in daily activity and food intake starting from the second week. No mortality was observed throughout the experimental period. Micro-CT was used to assess trabecular bone structure, and the morphological changes in each group are shown in Fig. 6a. The trabecular bone under stress (the main part that bears the weight of the body) in the GCs group has significantly thinned (red area). Quantitative analysis of trabecular architecture revealed that the GCs intervention group had significantly lower BV/TV, Conn. D, Tb. N, and Tb. Th compared to the Ctrl group, while Tb. Sp was higher. There was no significant difference in SMI between the two groups. Additionally, histological analysis showed that, compared to the Ctrl group, the GCs group exhibited disorganized and fractured trabecular structure, along with numerous empty lacunae. Furthermore, bone marrow fat was most prominent in the GCs group, as illustrated in Fig. 6b. Immunofluorescence double staining (EMCN + CD31) indicated a marked reduction in H-type vessels in the GCs group, primarily localized in the proximal region of the femoral head, as shown in Fig. 6c. (a) Micro-CT images and quantitative analysis of trabecular architecture. Red area: trabecular of stress bone. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (b) Empty lacunae: green regions. (c) Type H blood vessels. Increase of EndMT in SONFH mice By observing EndMT between the Ctrl group and the GCs group using immunofluorescence, we observed that a greater number of femoral head microvascular endothelial cells underwent mesenchymal transition in the GCs group as shown in Fig. 7. EndMT between the group of Ctrl and GCs. Oxidative stress levels in the different group As shown in Fig. 8a, immunohistochemical results indicated the presence of extensive Nrf2 and HO-1 positive expression areas in the femoral heads of mice in the GCs group. This suggests that, compared to the control group, the GCs group exhibited numerous brown positive expression areas and a more intense oxidative stress response, which could be alleviated by antioxidant therapy. Besides, the trabecular bone was significantly sparser, and most of the bone marrow cells were replaced by adipocytes. (a) Immunohistochemical results of femoral head samples from the Ctrl, GCs, and GCs + NAC groups (Nrf2, HO-1). (b) Comparison of EndMT among femoral head samples from the GCs, and GCs + NAC groups. To further explore the potential correlation between elevated oxidative stress and increased EndMT in the GCs group, we analyzed the GCs + NAC group. The results demonstrated that in SONFH, antioxidant therapy not only mitigated oxidative stress levels in the femoral head but also alleviated EndMT, as illustrated in Fig. 8b. In the in vitro HUVEC model, after glucocorticoid induction, cells were treated with NAC could improve tube formation capacity (Fig. 9a). The Western blot and PCR results demonstrated that NAC partially reversed the downregulation of endothelial markers and the upregulation of mesenchymal markers induced by glucocorticoids as shown in Fig. 9b,c. (a) tube formation; (b) PCR; (c) Western blot. Discussion This study used single-cell sequencing, femoral head specimens, cell cultures, and animal experiments to analyze the potential role of EndMT in the pathogenesis of SONFH. Our data suggest that EndMT may play a critical role in the progression of SONFH, particularly in the development of microvascular rarefaction and bone necrosis. However, despite providing valuable insights, our results still have certain limitations that require further validation and supplementation. ScRNA-seq seRNA-seq analysis revealed gene expression changes associated with EndMT, particularly the upregulation of mesenchymal markers such as COL1A1 and S100A4. Further pseudotime analysis demonstrated a gradual loss of endothelial characteristics and acquisition of mesenchymal features in endothelial cells, suggesting that EndMT is a dynamic and progressive process in the pathogenesis of SONFH rather than merely a late-stage consequence. Our data support the potential role of EndMT in microvascular dysfunction; however, whether EndMT directly contributes to microvascular rarefaction requires further verification. The role of oxidative stress in SONFH Oxidative stress induced by glucocorticoids appears to be a mediator of vascular and bone injury25,31,32. and this study demonstrated that oxidative stress levels were elevated in the SONFH model. Immunohistochemical analysis revealed extensive positive expression of Nrf2 and HO-1 in the femoral heads of the GCs group, suggesting that glucocorticoid administration enhanced local oxidative stress responses. This finding is consistent with previous studies reporting that glucocorticoids induce ROS accumulation and oxidative tissue injury, indicating that oxidative stress may play an important role in the development of SONFH33–35. Besides, we initiated NAC treatment at week 4 based on our observations in the animal model. H&E staining at week 4 revealed partial bone lacunae and bone marrow fat vacuoles, indicating that the damage was still in its early stages and could benefit from conservative intervention. In contrast, by week 6, large-scale necrosis had occurred, making conservative treatments ineffective. Clinically, most patients present with hip pain after partial femoral head necrosis, and if the necrotic area is still small and has not collapsed, hip-preserving treatments can be effective. Thus, initiating NAC at week 4 aligns with both the experimental evidence and clinical practice, offering potential therapeutic benefits before extensive necrosis develops. Morphologically, the trabecular bone in the GCs group appeared markedly sparser, and most of the bone marrow cavity was replaced by adipocytes, further supporting the notion that glucocorticoid exposure leads to structural deterioration and metabolic imbalance in bone tissue. Notably, antioxidant treatment with NAC partially alleviated these changes. Compared with the GCs group, the GCs + NAC group showed reduced Nrf2 and HO-1 (markers commonly associated with oxidative damage) expression, indicating that antioxidant therapy could decrease local oxidative stress levels. In addition, EndMT levels were alleviated in the GCs + NAC group, suggesting a potential association between oxidative stress and EndMT. This study provides experimental evidence supporting the role of oxidative stress in the pathogenesis of SONFH. Antioxidant intervention may partially protect against bone and vascular injury; however, the precise mechanisms require further investigation. The impact of EndMT on microvascular rarefaction Recent studies have shown that the EndMT process has shown potential in tissue engineering and regenerative medicine, particularly in the field of vascular regeneration and repair.36,37 Studies have shown that EndMT plays a key role in diseases such as atherosclerosis38, pulmonary fibrosis39and kidney damage21. ScRNA-seq provided high-resolution insights into microvascular injury and EndMT process in SONFH. Microvascular rarefaction is one of the key pathological features of SONFH, characterized by a reduction in the density of the microvascular network, a decrease in the number of blood vessels, and restricted blood flow. Studies have shown that microvascular rarefaction not only exacerbates local hypoxia and metabolic disturbances but may also promote further bone necrosis by inhibiting bone metabolism and interfering with bone repair processes14. EndMT is a form of endothelial cell reprogramming, where vascular endothelial cells lose their typical phenotype and transform into cells with fibroblast or smooth muscle cell characteristics8,40. This process is often accompanied by changes in extracellular matrix components such as collagen and fibronectin41–43. Especially under the influence of glucocorticoids, these transformed cells may deposit excessive collagen and other matrix components, causing thickening of the microvascular walls, narrowing of the vascular lumen, increased fibrosis, and loss of elasticity and permeability, which further restricts blood flow and exacerbates the degree of microvascular rarefaction21,44. Endothelial reprogramming can also inhibit angiogenesis. O'Riordan and colleagues45 found that in cells treated with NOS inhibitors, the synthesis of collagen XVIII increased, along with the overproduction of its anti-angiogenic fragments, leading to capillary rarefaction. This process was associated with an increase in the expression of transforming growth factor-β and connective tissue growth factor in the kidneys. Furthermore, inhibition of NOS triggers downstream signaling factors from the endostatin and TGF-beta pathways, reprogramming endothelial cells to adopt a myofibroblast phenotype and initiating the fibrotic process46. We demonstrated through single-cell sequencing, human specimen validation, cell experiments, and animal studies that glucocorticoids can promote the upregulation of mesenchymal-related markers in endothelial cells. These findings confirm the presence of EndMT in SONFH, as demonstrated by EndMT detection in regions of the femoral head samples from SONFH patients that had not yet undergone complete necrosis. Clinical implications and future directions This study enhances our understanding of the pathophysiological mechanisms behind SONFH. The findings suggest that managing oxidative stress may play a role in preventing or mitigating the vascular injury that drives the disease. Clinically, antioxidant therapies could potentially be employed to reduce the severity of SONFH, especially in patients who require long-term glucocorticoid therapy. This study also underscores the need for early diagnosis and intervention in patients at risk of SONFH, particularly those on chronic steroid treatments. Limitations However, there are limitations that should be addressed in future research. First, scRNA-seq was not performed directly on femoral head bone tissue. This decision was based on the technical challenges of isolating sufficient numbers of viable vascular endothelial cells from osseous tissue during single-cell dissociation. Moreover, most femoral head specimens available for research are from advanced-stage SONFH, which often fail to meet the stringent quality control requirements of single-cell transcriptomic analysis. Therefore, we instead analyzed cystic regions within SONFH lesions, which are known to exhibit atypical reparative responses. Although indirect, this approach provided valuable insights into the potential role of EndMT in SONFH pathogenesis. Besides, HUVECs were used for the in vitro investigation of EndMT in SONFH. Due to the inherent heterogeneity between HUVECs and bone microvascular endothelial cells, the results may not fully reflect the pathological changes occurring in the bone microenvironment. Furthermore, while we observed the activation of EndMT and oxidative stress in animal models, the precise molecular pathways, particularly involving the signaling pathway, were not fully explored. Further studies focusing on the detailed signaling mechanisms of EndMT could offer deeper insights into potential drug targets for preventing vascular injury in SONFH. Additionally, while NAC demonstrated some promise in reducing oxidative stress and EndMT, its long-term efficacy and safety in human clinical settings need further investigation. Future research should also consider exploring other antioxidant strategies or direct inhibitors of EndMT as potential therapeutic options. Moreover, clinical trials involving patients with SONFH are necessary to validate these preclinical findings and assess the clinical relevance of targeting oxidative stress and EndMT in preventing or reversing the disease. Conclusion This study integrated single-cell sequencing, human femoral head specimens, cell culture, and animal experiments to explore the role of EndMT in SONFH. The findings suggest that EndMT is involved in the pathological process of SONFH and may contribute to microvascular rarefaction and subsequent bone necrosis. In addition, oxidative stress appears to be a mediator linking glucocorticoid exposure to vascular and bone injury, and antioxidant intervention with NAC partially mitigated these effects in the animal model. Our results provide experimental evidence supporting the interaction between oxidative stress and EndMT in the progression of SONFH. However, further mechanistic studies are needed to confirm these findings and clarify the causal relationships involved. Overall, this work offers new insights into the vascular pathology underlying SONFH and suggests that targeting oxidative stress and endothelial dysfunction may hold potential for future therapeutic strategies. Author contributions All authors contributed to the study conception and design. Material preparation and all experiment were performed by Xiaoqiang Yang, Peng Yang and Jiaru Liao. Data collection and analysis were performed by Fan Yang and Mincong He. The first draft of the manuscript was written by Ziqi Li and Xiaoqiang Yang. Qiushi Wei and Wei He checked and revised the manuscript. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding This work was supported by the National Natural Science Foundation of China [grant numbers: 8227150353]; State Key Laboratory of Traditional Chinese Medicine Syndrome [grant number: SKLKY2024A003]. Data availability All sc-RNAseq raw data have been transferred to the GEO (accession numbers: GSE290411). Declarations Competing interests The authors declare that they have no competing interests. Ethics approval All procedures were performed in compliance with relevant laws and institutional guidelines and have been approved by the institutional ethics review board at the third Affiliated Hospital of Guangzhou University of Chinese Medicine. (PJ-KY-20220420-013, 2022-04-20). The privacy rights of human subjects have been observed and that informed consent was obtained for experimentation with human subjects. Animal experiments are approved by the Animal Experimentation Center of Guangzhou University of Chinese Medicine (20240717013). Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Xiaoqiang Yang and Peng Yang have equally contribution to this work. References Associated Data Data Availability Statement All sc-RNAseq raw data have been transferred to the GEO (accession numbers: GSE290411).