Abstract

   Glucocorticoid (GC) induced osteoporosis (GIOP) and osteonecrosis
   remain a significant health issue with few approved therapies. Here, we
   investigate the cellular and molecular processes by which GCs affect
   osteogenesis and angiogenesis. We find that GC treatment reduces bone
   mass through decreased bone formation by skeletal stem cells (SSCs).
   Concomitantly, endothelial cells increase in number but display
   distorted phenotypical features. Transplantation studies of SSCs
   combined with molecular analysis by single cell RNA-sequencing and
   functional testing of primary human cells tie GC-induced skeletal
   changes to altered stem cell differentiation dynamics. This in turn
   perpetuates reduced osteogenesis and vascular malformation through
   direct SSC-endothelial crosstalk mediated at least in part by Basigin.
   The genetic deletion of Basigin in the skeletal lineage as well as
   antibody-mediated blockade of Basigin during GC treatment prevents bone
   loss. Intriguingly, when administered to 2-year-old mice, anti-Basigin
   therapy reinstates bone remodeling to significantly improve bone mass.
   These findings provide therapeutic vantage points for GIOP and
   potentially other conditions associated with bone loss.

   Subject terms: Bone, Stem-cell differentiation, Bone development
     __________________________________________________________________

   Glucocorticoid (GC) induced osteoporosis (GIOP) remains a significant
   clinical problem. Here, the authors find that GCs disrupt skeletal stem
   cell–endothelial crosstalk via Basigin that can be targeted as a
   potential therapy for GIOP.

Introduction

   Glucocorticoids (GCs) are potent anti-inflammatory compounds; however,
   continued exposure results in bone loss and osteonecrosis^[58]1. Since
   skeletal stem cells (SSCs) are crucial for maintaining skeletal
   homeostasis, it is tempting to speculate that GCs might affect their
   function. Recently, detrimental changes to angiogenesis have also been
   implicated in GC-induced bone loss^[59]2,[60]3. Nevertheless, the
   mechanism by which GCs alter vascularity remains unknown. Our study
   aimed to evaluate GC effects on both osteogenesis and angiogenesis, as
   well as whether treatment with parathyroid hormone (hPTH 1-34) modifies
   the effects in a mouse model of GC-induced bone loss.

   GCs are frequently prescribed for both acute and chronic medical
   conditions to reduce inflammation and immune system activity. GCs are
   known to alter bone remodeling by reducing osteoblast activity,
   stimulating osteoclast maturation and activity and reducing gonadal
   hormone production, which indirectly activates
   osteoclastogenesis^[61]1. Bone loss induced by GCs tends to have two
   phases, with an initial rapid loss of bone mass, particularly
   trabecular, followed by a sustained, gradual reduction that results in
   the loss of cortical bone. Independent of the GC-induced loss of bone
   mass, subjects treated with GCs often rapidly reduce the mechanical
   strength of bone such that many subjects experience fractures while
   treated^[62]4.

   After the initial stages of GC-induced bone loss, a continued
   suppression of bone formation has been reported. This is mainly due to
   reduced activity of osteoblasts, the differentiated cells that are
   responsible for bone formation derived from a pool of SSCs. hPTH 1-34,
   also known as teriparatide, is FDA-approved to treat GC-induced
   osteoporosis (GIOP) as it stimulates osteoblasts to generate bone in
   the presence of GCs^[63]5,[64]6, decreases sclerostin production by
   osteocytes and prevents osteoclast maturation in the presence of
   GCs^[65]7.

   GC exposure can also induce osteonecrosis, the result of damaged
   skeletal vasculature that provides essential nutrients for normal cell
   function and tissue homeostasis. This adverse event is commonly
   associated with higher cumulative doses and longer treatment courses of
   systemic GCs; however, it has also been described after intra-articular
   injections, topical administration or low-dose, short-term oral
   GCs^[66]3. No major gender differences have been identified. While the
   exact etiology of osteonecrosis is not clear, a reduction in the
   vascular supply to the proximal femur is frequently observed.
   Angiogenesis is required for the osteogenic process, as endothelial
   cell invasion is needed to transition hypertrophic chondrocytes to
   osteoblasts at the growth plate, and angiogenesis through
   macrophage-initiated signaling is needed for the formation of a bone
   remodeling unit. However, the influence of GCs on angiogenesis is
   unclear^[67]8.

   Therefore, given the observation that GCs reduce bone formation and may
   alter vascular supply directly or indirectly to bone, the purpose of
   this study was to increase our understanding of how GCs affect the
   cellular and molecular components of the bone marrow environment. We
   performed a detailed analysis of the cellular composition, SSC activity
   and single cell gene expression in mice treated with GCs, after
   recovery from GCs and GC with hPTH 1-34 treatment. We determined that
   bone loss through GC exposure is initially associated with a reduced
   number of SSCs and vascular cells, while with continued GC exposure,
   skeletal precursor cells accumulated but lost their ability to
   differentiate into osteogenic and chondrogenic cell types.
   Interestingly, these SSCs secrete specific factors, including Basigin,
   that alter endothelial morphology and function through direct
   crosstalk. Concomitant hPTH 1-34 during GC exposure or genetic and
   antibody-mediated Basigin blockade abrogated that detrimental signaling
   axis and reversed the GC-mediated phenotype. Excitingly, anti-Basigin
   treatment also improved bone parameters in aged mice, suggesting a
   potential broader clinical utility. Altogether, these findings reveal a
   previously unappreciated connection between stem and endothelial cell
   interaction that can be targeted to prevent GC-induced bone loss.

Results

GC-induced bone loss is reversed by concurrent hPTH treatment

   To assess the cellular and molecular changes of continuous GC exposure
   on bone tissue, we subcutaneously transplanted 4-month-old, male
   Balb/cJ mice with 5 mg Methylprednisolone (GC) or placebo release
   pellets. Randomly grouped mice were housed for 28 days before analysis.
   To assess the consequences of prolonged GC treatment, additional
   animals were kept for another 28 days and compared to two separate
   groups either having GC pellets removed after 28 days or animals
   receiving hPTH 1-34 at 40 μg/kg, 5 days/week during GC exposure
   throughout the experiment (Fig. [68]1a). As expected from our previous
   study using a similar model^[69]9, micro-CT analyses and mechanical
   testing showed that 28- and 56-day GC exposure significantly reduced
   femoral trabecular bone parameters and bone strength, respectively,
   compared to placebo control (Fig. [70]1b–d, Supplementary Fig. [71]1a,
   b). Interestingly, removing GC pellets after 28 days did not reverse
   trabecular bone loss after an additional 28 days, while hPTH 1-34 in
   the presence of GC exposure normalized trabecular parameters to placebo
   levels. Cortical thickness and area were significantly reduced upon
   acute GC exposure but were unaltered in all groups monitored over a
   56-day period (Supplementary Fig. [72]1c, d). To test how these
   observations were tied to changes in bone remodeling, we conducted
   dynamic histomorphometry and found that bone formation and osteoclast
   activity were inversely controlled by GCs. Mineral apposition and bone
   formation rate were significantly reduced by GC treatment at both
   investigated timepoints, and only hPTH 1-34 treatment improved those
   parameters to placebo control levels (Supplementary Fig. [73]2a–c). In
   contrast, osteoclast activity was strongly increased upon GC exposure
   as measured by TRAP staining and bone marrow-derived in vitro
   osteoclastogenesis; however, GC removal alone was sufficient to reduce
   bone resorption activity to control levels (Supplementary Fig. [74]2c,
   d). Osteoclast surface per bone surface was elevated in GC mice that
   received hPTH 1-34, suggesting stimulation of increased bone remodeling
   with high bone formation rates outweighing increased bone resorption.
   Altogether, these data provide insight into the complex effects of GC
   and hPTH 1-34 actions on bone parameters of our mouse model.

Fig. 1. A GC-mediated anti-osteogenic, pro-endothelial, pro-myeloid shift is
reversed by hPTH.

   [75]Fig. 1
   [76]Open in a new tab

   a Experimental schematic for exposing mice to continuous
   glucocorticoids through subcutaneous Methylprednisolone treatment at
   two timepoints and varying interventions. Created in BioRender.
   Ambrosi, T. (2025) [77]https://BioRender.com/8pmjxe4. b Mechanical
   strength test of femurs by 3-point bending. c Quantification of
   trabecular bone volume per total volume (Tb.BV/TV) shown as percentage
   change compared to day-0 age-matched controls. Analysis of femur bones
   from n = 8 biologically independent mice for Placebo and GC removal
   groups, n = 7 for GC d-28, n = 6 for GC d-56 and n = 9 for GC+hPTH. d
   Representative microCT images of distal femur trabecular bone at day-28
   and day-56. Data shown as mean ± SEM. Statistical testing between
   Placebo and other group by two-sided unpaired Student’s t-test.
   *p < 0.05, **p < 0.01 ***p < 0.001, ****p < 0.0001. e UMAP of
   femur-derived single cells from all experimental groups and their
   distinct clustering by Leiden. f Specific markers of Leiden clusters
   determining cellular identity. g UMAP plot showing cellular clustering
   labeled by experimental group. h Bar graphs showing the relative
   abundance of each cell type captured for each experimental group. i
   Dotplot showing selected osteochondrogenic gene expression in
   mesenchymal cell subsets. j Global pathway enrichment analysis of the
   top 200 differentially expressed genes in mesenchymal cells of the GC
   group using EnrichR. k Dotplots of endothelial and l myeloid gene
   expression in vascular and hematopoietic cell types, respectively,
   between different experimental groups.

GC exposure drives distinct cellular and molecular changes in bones

   Having established our model of GC-induced bone loss, we next sought to
   derive a more detailed view of the cellular changes of bone tissue. We
   conducted 10X Chromium single-cell RNA-sequencing (scRNAseq) of
   dissected femurs from the four experimental groups on day 56. Given the
   low representation of non-hematopoietic cells in single cell
   preparations of bone tissue, we sorted equal numbers of CD45-positive
   and CD45-negative cell populations into the same collection tube for
   each group. Unbiased Leiden clustering analysis of stringently
   quality-filtered single cells established the cellular composition of
   captured cells for each experimental group (Fig. [78]1e, f). This
   approach covered the broad heterogeneous composition of the bone/bone
   marrow (BM) composition, including hematopoietic, mesenchymal and
   endothelial cell types. When we investigated specific differences in
   the cellular makeup of each group, we observed that GCs increased
   stromal cell populations in the BM, including CXCL12-expressing
   reticular (CAR) cells (Fig. [79]1g, h). There were also slight changes
   to committed bone-forming cell types, indicating potential alterations
   to mesenchymal lineage allocation commitment. Specifically, expression
   of genes associated with osteogenesis and chondrogenesis was reduced
   upon GC exposure but showed strong improvement to placebo levels if
   mice were simultaneously treated with hPTH 1-34 (Fig. 1i). Conducting
   global pathway enrichment analysis with the top 200 differentially
   expressed genes for each group, revealed that GC exposure led to
   increases in angiogenesis-related signaling in the BM environment (Fig.
   [80]1j). Specific gene expression patterns of endothelial genes showed
   that GC removal and hPTH 1-34 treatment normalized the expression to
   placebo levels (Fig. [81]1k). Similarly, while the immune cell
   compartment composition was not strongly altered by GC exposure, we
   observed an increase in pro-myeloid and pro-osteoclastic signaling that
   was partially reversed by GC removal and hPTH 1-34 treatment (Fig.
   [82]1l). In sum, scRNAseq of the bone tissue from the different
   experimental groups revealed distinct cellular and molecular changes of
   the BM compartment supportive of alterations at the tissue level.
   Strikingly, the strongest changes were associated with alterations in
   angiogenic signaling.

GCs alter skeletal stem and progenitor function and blood vessel
characteristics

   Since we observed bone loss and reduced expression of osteochondrogenic
   genes in mice exposed to GC, we wondered if skeletal stem and
   progenitor activity might be altered. Therefore, we analyzed the
   frequency of phenotypic SSCs
   (CD45^−Ter119^−Tie2^−CD90^−6C3^−CD105^−CD51^+[CD200^+]) and the
   directly downstream transient multipotent bone-cartilage-stromal
   progenitors (BCSPs; CD45^−Ter119^−Tie2^−CD90^−6C3^−CD105^+CD51^+) using
   flow cytometry^[83]10. Results showed that GCs initially (day 28)
   decreased SSC and BCSP numbers, while extended exposure (day 56) drove
   an increase in SSCs (Fig. [84]2a, Supplementary Fig. [85]3a, b).
   Similarly, GC removal did not change the significant accumulation of
   SSCs and BCSP compared to placebo controls. Again, only GC with
   concomitant hPTH 1-34 normalized skeletal progenitor abundance to
   control levels. To further assess the functional properties of SSCs, we
   freshly purified them from the bone tissue of each experimental group
   on day 56 and seeded them for standard in vitro osteogenic and
   chondrogenic assays. In alignment with micro-CT and transcriptomic
   data, GCs significantly reduced the osteogenic and chondrogenic
   potential of skeletal progenitors compared to placebo controls, which
   was also not altered upon GC removal assessed after 28 days (Fig.
   [86]2b, c). SSCs from bones of mice exposed to GC but treated with hPTH
   1-34 injections showed strong osteochondrogenic activity. These results
   indicate that skeletal stem and progenitor function is negatively
   affected by GCs, directly associating it with negative structural and
   mechanical changes observed for bones.

Fig. 2. GCs drive impaired skeletal stem and progenitor function and altered
bone marrow blood vessel characteristics.

   [87]Fig. 2
   [88]Open in a new tab

   a Flow cytometry-based quantification of skeletal stem cell (SSC,
   CD45-Ter119-Tie2-CD90-6C3-CD105-CD51+[CD200+]) in femurs of
   experimental groups. n = 8 biologically independent mice for Placebo
   d-28 group, n = 7 for GC d-28 group, n = 5 for Placebo d-56 group,
   n = 3 for GC d-56 group, n = 6 for GC removal and GC+hPTH groups. b In
   vitro osteogenesis (Alizarin Red S stain) and c chondrogenesis (Alcian
   Blue stain) assays on purified, primary SSCs. Spectrophotometric
   quantification of staining (right). Cells from n = 3 biologically
   independent mice per group. d Flow cytometry-based quantification of
   endothelial cell populations (CD31+) in femurs of experimental groups.
   n = 4 biologically independent mice for Placebo d-28, n = 3 for GC d-28
   and d-56 groups, n = 5 for Placebo d-56 and GC+hPTH groups, n = 6 for
   GC removal group. e Immunohistochemistry staining for Endomucin (Emcn)
   in bone marrow of day-56 placebo and GC-treated mice, as well as in
   24-month-old wild-type mice. f Representative immunohistochemistry
   staining for Endomucin (Emcn) in bone marrow of day-56 experimental
   groups and quantification of blood vessel area (right top) and size of
   sinusoid lumen (right bottom) based on immunohistochemistry staining
   for Endomucin. n = 4. Arrows: sinusoids. All data shown as mean ± SEM.
   Statistical testing between Placebo and other groups by two-sided
   unpaired Student’s t-test. *p < 0.05, **p < 0.01 ***p < 0.001,
   ****p < 0.0001. Scale bars, 20 µm.

   Based on scRNAseq results, we also wondered how endothelial numbers and
   composition may be altered. Flow cytometric analysis revealed a similar
   pattern as seen with SSCs. While after 28 days, endothelial cell
   numbers were reduced with GCs compared to the placebo group, longer GC
   exposure drove an accumulation of endothelial cell presence that was
   only normalized to placebo levels when mice received hPTH 1-34 with GCs
   (Fig. [89]2d, Supplementary Fig. [90]3c). When we stained histological
   sections of femur bones after 56 days of GC exposure with the
   endothelial marker Endomucin, we observed an increase of blood vessel
   abundance as well as distinct morphological changes that resembled the
   endothelial compartment of aged (24-month-old) mice (Fig. [91]2e).
   Previous work has reported that cellular stress in the BM environment
   is connected to increased numbers of sinusoids with dilated
   lumina^[92]11. Quantitative analyses showed that at day 56 of GC
   exposure, there were significantly more Endomucin-positive sinusoidal
   blood vessels with increased lumen area in the BM compared to placebo
   controls (Fig. [93]2f, Supplementary Fig. [94]3d). The blood vessel
   area and dilation of lumens remained at placebo levels when mice were
   concomitantly treated with hPTH 1-34. In summary, GCs mediate specific
   changes to bone-forming lineage cells and angiogenic processes that can
   be reversed by co-stimulation with hPTH 1-34 treatment, suggesting a
   direct connection between SSCs and endothelial cells.

GCs alter SSC-mediated ossicle formation

   To determine if GC-mediated alterations were connected through changes
   observed in SSC and endothelial cell activity, we assessed SSC-specific
   bone formation characteristics via renal capsule
   transplantation^[95]12. To that end, we freshly FACS-purified equal
   numbers of SSCs from ubiquitous GFP-reporter mice and transplanted them
   beneath the renal capsule of wild-type C57BL/6 mice. We randomly
   assigned mice to four groups, i.e., a control group receiving placebo
   pellets at the time of surgery, a group with hPTH 1-34 treatments but
   no GCs as well as two groups receiving GC release pellets. One of the
   GC groups received 5x weekly hPTH 1-34 over the 21 days of the
   experiment (Fig. [96]3a). At that time, we isolated tissue grafts and
   processed them for scRNAseq. As expected, generated ossicles of all
   groups contained mesenchymal, endothelial and hematopoietic cell types
   recognized by distinct single cell clustering based on overall gene
   expression profiles (Fig. [97]3b–e). Given limitations of scRNAseq
   approaches, including transcriptomic coverage, preventing faithful
   separation of rare, homogeneous SSC populations using unbiased
   clustering analysis, we identified the SSC-enriched cluster by known
   but not unique marker expression (e.g., Sox9, Col2a1, Pthlh) within a
   broader pool of skeletal stem and progenitor cells (SSPCs). These SSPCs
   also showed high expression of the gene Nr3c1 encoding the
   glucocorticoid receptor compared to other mesenchymal lineage cells
   (Supplementary Fig. [98]3e). Mice treated with GCs alone presented with
   a higher number of undifferentiated SSPCs than placebo-treated mice.
   The expression of osteogenic and chondrogenic gene programs also
   decreased with GC treatment alone but was rescued by hPTH 1-34
   treatment (Fig. [99]3f). Analyzing gene expression patterns of all
   cells revealed that GC exposure led to high expression of genes related
   to cellular stress including those associated with the
   senescence-associated secretory phenotype (SASP) and oxidative stress,
   supporting the idea that GCs appear to drive an aging-like and
   pro-inflammatory microenvironment (Fig. [100]3g). Interestingly, genes
   involved with blood vessel recruitment and formation were highly
   increased in the presence of GCs which also was associated with
   increased oxidative stress gene expression (Fig. [101]3h). Once again,
   hPTH 1-34 added to the GC treatment was able to reverse gene expression
   patterns toward placebo control levels. Of note, higher circulating
   levels of hPTH 1-34 elevated expression of SASP/cellular stress-related
   genes compared to the placebo group. Slight increases in those markers
   might be the reflection of a more activated state of hPTH 1-34 exposed
   cells rather than increased inflammation. Since SSC-derived ossicle
   formation strongly depends on host endothelial blood vessel
   recruitment, these results support a co-regulation of SSC and
   endothelial lineages during GC exposures and provide a list of
   molecular interaction partners.

Fig. 3. GC alters de novo in vivo bone formation by SSCs and the niches they
generate.

   [102]Fig. 3
   [103]Open in a new tab

   a Experimental schematic of renal capsule transplants of freshly
   purified wild-type, GFP-labeled SSCs in mice exposed to placebo, GC,
   hPTH alone, or GC+hhPTH (hPTH 1-34). Created in BioRender. Ambrosi, T.
   (2025) [104]https://BioRender.com/r8dp0jv. b Light microscopic images
   (top) and GFP signal of grafts formed beneath the renal capsule of
   transplanted mice. c Single-cell transcriptomic analysis of formed
   tissue grafts displayed as UMAP of single cells clustered by Leiden for
   cell type separation and by group (top left). SSPCs, skeletal stem and
   progenitor cells. d Tracksplot of specific markers of Leiden clusters
   determining cellular identity. e Bar graphs showing the relative
   abundance of each cell type captured for each experimental group. f
   Dotplot showing selected osteochondrogenic (left) genes in SSPC cluster
   between different experimental groups. g Dotplot showing SASP
   (senescence-associated secretory phenotype)/cellular stress-related
   gene expression in all cells between different experimental groups. h
   Dotplot showing pro-endothelial and oxidative stress-related gene
   expression between different experimental groups.

GC-induced SSC-derived Basigin impairs bone-forming and vascular network
properties

   Among the multiple pro-endothelial genes upregulated upon GC exposure
   and rescued by hPTH 1-34 treatment (Kdr/Vegfr, Vegfa, Pdgfa, Wnk1), we
   identified Basigin (Bsg) as well as its known intracellular interaction
   partners of the monocarboxylate transporter (Mct1-3) family to be
   highly expressed in accumulated SSPCs (Figs. [105]1k, [106]3h, [107]4a,
   Supplementary Fig. [108]3f)^[109]13. We therefore hypothesized that
   Basigin, a known activator of cell proliferation^[110]14,[111]15, might
   drive GC-mediated aberrations in the BM environment through a distinct
   SSC-endothelial signaling axis. To further explore this concept and its
   translational relevance, we conducted functional tests in primary human
   SSCs (hSSCs, CD45^-CD235a^-CD31^-TIE2^-CD146^-PDPN^+CD164^+CD73^+) and
   in the human VeraVec HUVEC endothelial cell line. We collected 48 h
   supernatant from hSSCs either lentivirally overexpressing Basigin or
   vector-controlled to directly tie SSPC-derived Basigin expression to
   changes in endothelial cell activity. Then, we assessed the effect of
   the supernatants of both groups during in vitro tube formation and
   wound scratch assays of VeraVec endothelial cells. Strikingly, in the
   presence of high levels of Basigin in the supernatants, blood vessel
   architecture was significantly impaired, as indicated by reduced mesh,
   tube and node numbers compared to the control (Fig. [112]4b, c). We
   also found that Basigin impaired endothelial cell migration in a wound
   scratch assay (Fig. [113]4d). In addition, elevated Basigin levels
   increased ROS generation in cultured endothelial cells (Fig. [114]4e),
   confirming scRNAseq readouts (Fig. [115]3g, h). Furthermore,
   overexpression of Basigin in hSSCs drove increased colony-forming
   ability, which is a measure of proliferative activity, while it
   impaired in vitro osteogenic and chondrogenic differentiation compared
   to controls (Fig. [116]4f–h). Finally, we subcutaneously transplanted
   equal numbers of hSSCs either overexpressing Basigin or vector control
   into immunodeficient NOD scid gamma (NSG) mice to determine if ossicles
   formed displayed differences in osteogenesis and angiogenesis in vivo
   (Fig. [117]4i, j). In line with a recent report on the effect of GCs on
   fracture healing^[118]16, we observed reduced bone remodeling dynamics
   in grafts derived from Basigin-overexpressing cells that were less
   mineralized and showed low bone resorption activity by host-derived
   osteoclasts (Fig. [119]4k, l). Interestingly, the shortened morphology
   of recruited blood vessels of the Basigin-SSC grafts mirrored the
   GC-induced phenotype observed in femurs (Fig. [120]4m). These results
   establish a direct connection between SSC-derived Basigin and
   detrimental effects of skeletogenesis and blood vessel architecture
   caused by GCs that are transferable to human cells.

Fig. 4. Basigin overexpression alters human skeletal lineage dynamics that
impair endothelial function.

   [121]Fig. 4
   [122]Open in a new tab

   a Single cell transcriptomic analysis of tissue grafts from mouse SSCs
   either exposed to placebo, GC, human PTH or GC+human PTH (hPTH 1-34)
   displayed as UMAP showing high expression of Basigin in SSPCs. SSPCs,
   skeletal stem and progenitor cells. Left bottom: Violin plot showing
   increased expression of Bsg in GC group across all cells. b
   Representative image of tube formation assay by VeraVec endothelial
   cells exposed to control supernatant or supernatant of human SSCs
   overexpressing Basigin. c Quantification of ImageJ-based tube formation
   analysis. n = 6 independent samples from two independent experiments. d
   Representative brightfield images of the endothelial scratch assay
   (left) and its quantification (right). n = 6 independent samples from
   three independent experiments. e Measurement of reactive oxygen species
   in cultured endothelial cells after 24 h supernatant exposure. Left:
   representative fluorescence images. Right: Quantification of
   fluorescence signal. n = 6. All experiments with supernatant from at
   least two donors and two independent experiments. f Colony-forming unit
   ability of primary human SSCs either overexpressing control vector or
   Basigin. CFU-Fs stained by Crystal violet. n = 6 independent replicates
   from cells of two biologically independent donors. g In vitro
   osteogenesis (Alizarin Red S stain) by hSSCs of the same groups. n = 3
   independent replicates from cells of one biological donor. h In vitro
   chondrogenesis (Alcian Blue stain) assays of the same groups. n = 6
   independent replicates from cells of two biologically independent
   donors. i Schematic of subcutaneous transplant approach. Created in
   BioRender. Ambrosi, T. (2025) [123]https://BioRender.com/4wl26h3. j
   Immunohistochemistry of Basigin (green) expression in SSC-generated
   grafts with corresponding quantification. n = 3 biologically
   independent donor cells in three biologically independent mice per
   group. k Representative Alizarin Red S staining of sectioned grafts and
   quantification of mineralized tissue in grafts containing transplanted
   human SSCs as assessed by Alizarin Red S staining. n = 6 sections from
   three biologically independent grafts. l Representative TRAP staining
   of sectioned grafts and quantification of TRAP-positive area of grafts.
   n = 6 sections from three biologically independent grafts. m
   Representative immunohistochemistry staining for Endomucin and DAPI of
   sectioned graft. Small insert shows Endomucin staining without DAPI.
   All data shown as mean ± SEM. Statistical testing between Placebo and
   other groups by two-sided unpaired Student’s t-test. *p < 0.05,
   **p < 0.01 ***p < 0.001, ****p < 0.0001. Scale bars, 50 µm.

Antibody-mediated blockade of hSSC-derived Basigin reverses impaired
endothelial function and bone formation in vitro

   Next, we investigated whether the negative effects of Basigin
   overexpression in human SSCs could be pharmacologically reversed. To
   that end, we treated human SSCs with a monoclonal antibody against
   Basigin (aBSG) and collected supernatant to test its effect on vascular
   modeling. Indeed, the detrimental paracrine effect on human endothelial
   tube formation in the presence of supernatant from
   Basigin-overexpressing SSCs was mostly returned to control levels with
   antibody treatment (Fig. [124]5a). Similarly, VeraVec cells exposed to
   Basigin and treated with aBSG performed similarly to controls in wound
   scratch assays (Fig. [125]5b). Strikingly, the Basigin induced
   osteogenesis-impairing effects due to overexpression in SSCs were
   rescued when we either exposed the cells to hPTH 1-34 or aBSG (Fig.
   [126]5c), supporting Basigin antibody blockade as a strategy to prevent
   or reverse GC-induced SSC dysfunction mediated bone loss.

Fig. 5. Antibody blockade of Basigin rescues GC-induced endothelial and
skeletal impairments in vitro.

   [127]Fig. 5
   [128]Open in a new tab

   a Tube formation assay of human VeraVec endothelial cells treated with
   supernatant from different experimental groups. n = 6 independent
   replicates per group from two independent experiments. b Representative
   brightfield images of endothelial scratch assay (top) and its
   quantification displayed as boxplots (bottom) after 12 h. Endothelial
   cells were treated with supernatant of cultured hSSCs treated as shown.
   n = 6 independent replicates from two independent experiments for
   Ctrl+aBsg and Bsg-OE+aBsg groups. n = 12 independent replicates from
   three independent experiments for Bsg-OE group. Boxplots with box and
   whiskers Min to Max. c In vitro osteogenesis of patient-derived human
   SSCs. Left: Boxplots showing quantification of Alizarin Red S staining.
   Right: Representative images of Alizarin Red S staining. Ctrl: control
   media only; Ctrl+PTH: control media with PTH 1-34 treatment 6 h before
   media change; Ctrl+aBsg: control media with Basigin antibody treatment.
   Bsg-OE: Lentivirally Basigin-overexpressing human SSCs with control
   media. n = 10 independent replicates of cells from two independent
   donors. Data shown as mean ± SEM. Statistical testing in a,c by one-way
   ANOVA with Fisher-LSD test, in b by Wilcoxon signed-rank test.
   *p < 0.05, **p < 0.01 ***p < 0.001, ****p < 0.0001. Scale bars, 100 µm.

Pharmacological and genetic ablation of Basigin in vivo prevents detrimental
bone loss induced by GCs

   To confirm our in vitro findings, we next tested whether antibody
   blockade of Basigin can prevent GC-induced bone loss in mice. We
   compared placebo-treated control mice with mice that were exposed to
   GCs for 28 days. Of these GC-treated mice we looked at mice receiving
   no additional therapy or treated with hPTH 1-34 or aBSG throughout that
   time period (Fig. [129]6a). Histological analyses of femoral bone
   sections at day 28 showed that hPTH 1-34 and aBSG treatments reduced
   Basigin expression seen in the GC only group (Fig. [130]6b). Also, the
   reduction in Basigin levels in these treatment groups correlated with
   trabecular bone volume, osteoclast numbers and bone marrow endothelial
   morphology of placebo controls (Fig. [131]6c–e), prevented the
   GC-induced short-term reduction in SSC frequency (Fig. [132]2a) and
   maintained their in vitro osteogenic potential (Fig. [133]6f, g). While
   we observed a GC-driven shift toward an increase in circulating myeloid
   cell types in blood, which was reversed by aBSG treatment, we did not
   detect any significant differences in the hematopoietic stem and
   progenitor compartment in the bone marrow (Fig. [134]6h–i,
   Supplementary Fig. [135]4).

Fig. 6. Pharmacological blockade and genetic knockout of Basigin reverses
GC-induced skeletal changes.

   [136]Fig. 6
   [137]Open in a new tab

   a Schematic of experimental setup. Created with BioRender.com. b
   Representative immunohistochemistry images and quantification of
   Basigin expression in the different experimental groups. n = 3
   biologically independent mice per group. c Representative H&E stain of
   distal femurs of experimental groups with quantification of trabecular
   bone area below the growth plate region. n = 5 biologically independent
   mice per group. d TRAP staining and quantification of femur bones from
   all experimental groups. n = 3 biologically independent mice per group.
   e Representative images of Endomucin-positive endothelium in the bone
   marrow environment. Arrows: small sinusoids. f Flow cytometric analysis
   of femur bones of experimental groups for skeletal stem cells (SSCs).
   n = 5 biologically independent mice per group. g In vitro osteogenic
   differentiation of bone marrow stromal cells derived from experimental
   groups stained with Alizarin Red. n = 3 biologically independent mice
   per group. h Flow cytometric analysis of blood from mice of different
   experimental groups showing lymphoid to myeloid ratio. i Flow
   cytometric analysis of bone marrow (BM) common myeloid progenitor cells
   (CMPs) from mice of different experimental groups. n = 5 biologically
   independent mice per group. j Experimental schematic for GC experiments
   in heterozygous Basigin knockout mice. Control (Ctrl) and
   tamoxifen-treated conditional knockout (cKO) mice were compared for
   their skeletal response to GC exposure. Created in BioRender. Ambrosi,
   T. (2025) [138]https://BioRender.com/35tbvy1. k Co-staining of Basigin
   (green) and endothelial marker CD31 (red) in bones of experimental
   groups after GC exposure. l Micro-CT analysis of trabecular bone
   parameters at endpoint. n = 3 mice per group. m TRAP staining for
   osteoclast activity at endpoint. Average from n = 3 biologically
   independent mice per group. All data shown as mean ± SEM. Statistical
   testing by one-way ANOVA with Fisher-LSD test for pharmacological
   approach (b–i) or by two-sided unpaired Student’s t-test for comparison
   of genetic knockout groups (k–m). *p < 0.05, **p < 0.01, ***p < 0.001,
   ****p < 0.0001. Scale bars, 50 µm.

   To strengthen our findings and tie GC-induced bone loss specifically to
   skeletal stem and progenitor cell-derived Basigin, we crossed Collagen
   2a1 (Col2a1)-Cre^ERT2 with Basigin^flox/flox (Bsg^fl/fl) mice to
   generate conditional knockout mice lacking Basigin expression in SSPCs.
   Our previous studies^[139]17,[140]18 and current results (Fig. [141]3d)
   indicated that SSPCs, but not endothelial or immune cell types, express
   Col2a1. Female Col2a1-Cre^ERT2 x Bsg^fl/^+ transgenic mice were
   tamoxifen-induced when they reached 8 weeks of age and implanted with
   4-week GC release pellets 1 week later (Fig. [142]6j). Compared to
   vehicle treated (corn oil) control mice, heterozygous SSPC-specific
   knockout mice presented with significantly reduced Basigin levels upon
   GC exposure (Fig. [143]6k). This reduction was accompanied by
   significantly higher trabecular bone parameters and reduced osteoclast
   numbers suggesting that a reduction in expression of Basigin from SSPCs
   was sufficient to prevent GC-induced bone loss (Fig. [144]6l, m,
   Supplementary Fig. [145]4e–g). Thus, administration of aBSG during GC
   treatment or conditional genetic ablation specifically in the skeletal
   stem and progenitor lineage prior to GC exposure prevents skeletal
   maladaptation.

Basigin antibody therapy improves bone mass in aged mice independent of sex

   Since Basigin has been reported to be a therapeutic target for a number
   of pro-inflammatory and pro-fibrotic conditions, we asked whether aBSG
   could also reverse age-related bone loss, i.e.,
   osteoporosis^[146]19,[147]20. Indeed, immunohistochemistry analysis
   showed higher expression of Basigin in bone marrow of old mice, in
   particular in females (Fig. [148]7a). When we treated 2-year-old female
   and male mice with aBSG three times/week at 1 mg/kg for 4 weeks we
   found a significant improvement in volumetric trabecular bone of long
   bones compared to IgG control treated mice (Fig. [149]7b). Vertebral L5
   bone mineral density at 14 days and BV/TV at 28 days of treatment
   measured by DEXA and micro-CT, respectively, demonstrated an anabolic
   effect on the skeleton by aBSG independent of sex (Fig. [150]7c, d).
   Interestingly, SSCs of the same mice showed higher in vitro osteogenic
   potential, whereas histological analysis suggested an increase in bone
   resorptive activity with aBSG (Fig. [151]7e, f). This was accompanied
   by endothelial restoration but no detectable changes in the frequency
   of skeletal and hematopoietic stem and progenitor cells (Supplementary
   Fig. [152]5). Collectively, these results indicated that aBSG treatment
   boosts bone remodeling in aged mice in a sex-independent manner,
   leading to a net gain in bone mass despite elevated osteoclast activity
   likely through higher bone-forming activity.

Fig. 7. Antibody blockade of Basigin in aging mice reinvigorates skeletal
remodeling improving bone mass.

   [153]Fig. 7
   [154]Open in a new tab

   a IHC of Basigin expression in tibias of 24-month-old mice. Right:
   Histology-based quantification Basigin-expression cells per analyzed
   area. n = 3 biologically independent mice per sex and group. b H&E
   staining of proximal tibia regions below the growth plate. Right:
   Histology-based quantification of tibial trabecular BV/TV. n = 12
   biologically independent female mice per group. n = 9 biologically
   independent male mice per group. c DEXA bone mineral density (BMD)
   measurements of vertebral L5 at treatment starting day-0 and at day-14
   of treatment. d Micro-CT measurement of vertebral L5 BV/TV at 4 weeks
   of treatment. n = 5 biologically independent mice per sex and group. e
   Representative Alizarin Red S staining of in vitro osteogenesis. Cells
   from n = 5 independent mice per group. f TRAP quantification in tibia
   sections of experimental groups. N.Oc/BS: number of osteoclasts per
   bone surface. Analysis of n = 4 biologically independent female mice
   per group. n = 3 biologically independent male mice per group. g
   Graphical summary of experimental findings. Created in BioRender.
   Ambrosi, T. (2025) [155]https://BioRender.com/03daips. All data shown
   as mean ± SEM. Statistical testing between IgG control and other group
   by two-sided unpaired Student’s t-test. *p < 0.05, **p < 0.01,
   ***p < 0.001, ****p < 0.0001. Scale bars, 50 µm.

Discussion

   In this study, we identified a previously unknown connection between
   SSC function and angiogenesis in GIOP. While it has been established
   that GCs increase osteoclastogenesis and alter osteogenesis and
   angiogenesis, a mechanism related to vascular changes has not been
   appreciated. Our findings demonstrate that continuous treatment with
   GCs promotes SSC proliferation, leading to their accumulation at the
   expense of differentiation required for bone formation. We uncovered a
   signaling axis by which Basigin, a transmembrane protein, specifically
   expressed by skeletal stem and progenitor cells when exposed to GCs,
   alters endothelial function and remodels BM vasculature, driving
   cellular stress and promoting bone loss through reduced stem cell-based
   bone formation and increased resorption.

   Basigin has been considered an innovative marker in the context of
   cancer stem cells^[156]21, but little is known about Basigin in the
   realm of skeletal biology. Two studies report that high levels of
   Basigin promote osteoclastogenesis through NFATc1 signaling^[157]22 and
   its involvement in alveolar bone remodeling and soft tissue
   degradation^[158]23. It also acts as a potent stimulator of
   Interleukin-6 secretion in multiple cell lines that include
   monocytes^[159]24, altogether supporting our observation of increased
   pro-osteoclastic signaling and bone resorption activity in GC-treated
   animals with high stem and progenitor cell-specific Basigin expression.
   In line with SSC accumulation as well as their altered cytokine and
   elevated extracellular matrix factor expression upon GC exposure,
   Basigin has also been found to contribute to tumor progression by
   stimulation of proliferation, elevated growth factor secretion and its
   matrix metalloprotease activity^[160]25. More specifically, previous
   studies have shown that Basigin acts as a chaperone for the shuttling
   of monocarboxylate transporters (MCTs) to the cell surface, where MCTs
   facilitate the movement of metabolites such as short-chain fatty acids
   and lactate. Lactate can directly regulate the cell cycle to promote
   cellular proliferation^[161]26, for example, by modifying histones
   through lactylation changing epigenetic states^[162]27. Indeed, in
   Basigin-expressing SSPCs, we found concomitant upregulation of Mct1-3
   needed for lactate influx but downregulation of Mct4 used for lactate
   efflux, suggesting that the altered metabolic state of SSPCs with high
   intracellular lactate could contribute to the functional changes
   observed. Follow-up studies will have to experimentally address this
   and potential other mechanisms, such as altered mechano-sensing of
   SSPCs due to Basigin-mediated ECM remodeling^[163]28.

   At the endothelial cell level, we found that GC-induced vascular
   impairments are closely tied to SSC-derived Basigin. Supporting our
   findings connecting Basigin to blood vessel formation is its known role
   as a coreceptor for vascular endothelial growth factor receptor 2
   (Kdr/Vegfr2) in endothelial cells, enhancing their Vegfr-mediated
   activation and downstream signaling^[164]29. Continuous GC treatment in
   mice increased endothelial cell number and blood vessel area. However,
   Basigin exposure to endothelial cells resulted in irregularities in the
   connectivity and thickness of the nascent blood vessels. Flow
   cytometric and histological readouts, single-cell gene expression data
   and functional experiments connect these vascular changes to enhanced
   Vegfr signaling with a pathological blood vessel phenotype and
   increased oxidative stress. This is corroborated by studies that have
   shown that overactivation of Vegfr signaling is a driver of cellular
   stress in part through the stimulation of cell-damaging and
   inflammatory reactive oxygen species that negatively affect formation,
   migration and permeability of blood vessels^[165]30,[166]31. One of the
   most perplexing questions in medicine is how glucocorticoids induce
   osteonecrosis (ON). While epidemiologic studies find GCs increase the
   risk of ON, the pathology of the ON lesion is that of reduced blood
   vessel density and increased adipose tissue within the bone
   marrow^[167]8. Therefore, our observations may be relevant to
   GC-induced ON, as the altered morphology of these nascent blood vessels
   may reduce the ability to deliver hemoglobin/oxygen, and likely other
   crucial nutrients, to the bones. In certain areas of specific skeletal
   compartments, like the femoral head, this may lead to increased cell
   death. Potentially connected to this, low hemoglobin is independently
   associated with increased fracture risk in non-vertebral bones of older
   adults^[168]32. Altogether, it will be worthwhile to investigate if the
   pronounced changes in angiogenic signaling mediated by Basigin may
   initiate the development of GC-induced ON and higher fracture risk.

   Excitingly, the observation that this pathological signaling mechanism
   is reversible presents therapeutic vantage points to counter GC-induced
   bone loss and maybe ON. Concomitant administration of hPTH or aBSG
   prevented bone loss and other pathological skeletal phenotypes observed
   with GC exposure. Counterintuitively to the known anti-inflammatory
   actions of GCs, one of the bone marrow-specific changes we observed
   with GC exposure was an increase in myeloid signaling. This result is
   consistent with the bone cell activity changes observed in GC-treated
   subjects, as GC treatment reduces bone formation and increases the
   number and activity of osteoclasts^[169]8,[170]9. The locally
   bone-restricted pro-inflammatory marker upregulation, often also seen
   with cellular stress and SASP, is likely indirect and related to
   changes in bone formation dynamics that favor increased
   osteoclastogenesis and pathological remodeling of blood vessel
   architecture that drive disruption in oxygen supply, favoring
   pro-necrotic states at least in part through Basigin-mediated
   mechanisms. It remains to be determined if strategies preventing
   GC-induced bone loss have any impact on the intended clinical effects
   of GC therapy.

   We observed that the withdrawal of GCs on day 28 did not restore bone
   mass to baseline levels by day 56. Elevated GCs result in a reduction
   in the release of GCs from the hypothalamus^[171]33, and our 28-day GC
   treatment may have sufficiently suppressed the
   hypothalamic-pituitary-adrenal (HPA) axis such that the 28-day
   treatment period was insufficient to see recovery of the skeletal
   tissues. In support of this observation, both osteogenesis and
   angiogenesis measurements for this group were similar to those of the
   GC-only group at 56 days. New studies will need to investigate the HPA
   axis recovery in this treatment group to contextualize our current
   findings. Clinical studies have reported that GIOP can be treated with
   hPTH 1-34, leading to an increase in trabecular bone mass and
   biochemical markers of increased bone formation. Furthermore,
   meta-analyses of multiple studies have reported improved bone
   strength^[172]6,[173]34. However, the effect of hPTH 1-34 on
   angiogenesis in the presence of GCs has not been investigated. Our
   study demonstrated that hPTH 1-34 treatment during GC exposure reduced
   Basigin expression, restored blood vessel number and morphology, as
   well as osteochondrogenic differentiation of SSCs. The role of hPTH
   1-34 in angiogenesis is complex and might be further affected by GC
   exposure. While hPTH 1-34 has previously been shown to be osteoanabolic
   and to direct blood vessels toward the bone-forming surface, it has not
   been found to be pro-angiogenic^[174]35. In contrast, direct beneficial
   actions of hPTH 1-34 on endothelial cells have been reported in
   fracture regeneration settings^[175]36,[176]37. Even though GCs and
   hPTH 1-34 likely act on multiple skeletal lineage cell populations, our
   results provide evidence for their crucial role in regulating SSC
   activity.

   Intriguingly, we could show that the antibody treatment against Basigin
   and heterozygous conditional deletion of SSPC-derived Basigin mitigate
   both its detrimental autocrine and paracrine effects on skeletogenesis
   and vascular modeling, respectively. Recognizing that many phenotypic
   changes in response to GC treatment resembled age-associated skeletal
   changes, i.e., dysfunctional SSCs, pro-osteoclastic signaling and loss
   of bone marrow blood vessel integrity, we deduced that GC-induced bone
   loss might underlie overlapping mechanistic processes^[177]17,[178]38.
   Fittingly, we found Basigin to be highly abundant in the bone marrow of
   2-year-old mice. When we tested treatment with aBSG in aged male and
   female mice, bone parameters significantly improved, suggesting
   antibody treatment against Basigin could also serve as an
   anti-osteoporotic drug independent of sex. Additional work with
   variations in dosing and longer treatment regimens is needed to
   validate this. Similar to hPTH 1-34, aBsg might restore proper SSC
   function, thereby preventing elevated Basigin levels and their negative
   downstream effects. Future work will have to inquire if aBSG elicits
   overlapping molecular pathways in SSCs or if it might be a superior
   therapeutic approach to hPTH 1-34 to prevent GC-mediated bone loss.
   Given that GC treatment is often prescribed to patients for many years
   and hPTH 1-34 can only be administered for a few years to restore bone
   mass, aBSG might become a valuable alternative. In summary, our work is
   the first to report the detrimental role of Basigin signaling in
   osteo-angiogenic coupling of bones and provides a therapeutic target to
   prevent or potentially reverse skeletal health decline. Future studies
   are needed to determine the specific autocrine and paracrine signaling
   cascades Basigin elicits on skeletal and endothelial cell populations
   in more detail, and whether its blockade might also be able to reverse
   established GIOP and prevent GIOP-induced ON.

Methods

Animals

   All mouse experiments complied with all relevant ethical regulations
   and were conducted under approved protocols by the Institutional Animal
   Care and Use Committee of Stanford University and the University of
   California, Davis. Mice were maintained at animal facilities in
   accordance with institutional guidelines. Mice were given food and
   water ad libitum and housed in temperature-, moisture- and
   light-controlled (12-h light-dark cycle) micro-insulators. Unless
   otherwise specified, experiments were conducted using 3-month-old
   Balb/cJ male mice, purchased from Jackson Laboratories
   (Strain#:000651). To study the effects of glucocorticoids on bone
   biology, Methylprednisolone 5 mg 60-day release pellets or vehicle
   (cat# SG-241, Innovative Research of America, Sarasota, FL, USA) were
   implanted under the skin near the lumbar spine at day 0, and pellets
   were maintained for 56 days or removed after 28 days as indicated.
   Human parathyroid hormone 1-34 (hPTH 1-34) was purchased from Sigma
   (cat# P3796), and the treatment was administered subcutaneously at
   40 μg/kg, 5x/week for four consecutive weeks. For 28-day studies with
   concomitant hPTH (40 μg/kg/day, subcutaneously 5x/week) or aBSG
   (1 mg/kg, i.v. 3x/week) treatment mice were implanted with
   Methylprednisolone 2.5 mg 21-day release pellets or vehicle control
   pellets. Aged (24-months) male and female C57BL/6 from were ordered
   from NIA and treated i.v. with aBSG or IgG controls (1 mg/kg, 3x/week
   for 4 weeks; ThermoFisher, cat#: 16-1471-82 and 16-4321-82). For renal
   capsule transplantation experiments, cells from 3-month-old male GFP
   reporter mice (C57BL/6-Tg(CAG-EGFP)1Osb/J; JAX: 003291) were
   transplanted into male B6 mice (C57BL/Ka-Thy1.1-CD45.1; JAX: 000406).
   On the same day, mice were transplanted with GC pellets (2.5 mg 21-day
   release pellets or vehicle) and hPTH treatment was initiated
   (40 μg/kg/day, 5x/week for 3 weeks). Three-month-old immunodeficient
   NSG mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ; JAX: 005557) were used for
   human cell transplantation studies. Bsg^flox/flox mice were obtained
   from the lab of Dr. Romana Nowak at the University of Illinois at
   Urbana-Champaign and as previously published^[179]39. Tamoxifen
   inducible Col2a1-Cre^ERT2 mice were crossed with Bsg^flox/flox mice to
   generate Col2a1-Cre^ERT2 x Bsg^flox/+ (conditional heterozygous Basigin
   knockout, cKO) female offsprings. At 8 weeks of age, induction of the
   Cre-recombinase was achieved by daily oral gavage of tamoxifen
   (75 mg/kg) for 5 consecutive days. For the control group, transgenic
   mice were injected with vehicle (corn oil). One week after the first
   dose, mice were implanted with 21-day Methylprednisolone release
   pellets.

Human primary cells and cell lines

   Human skeletal stem cells (hSSCs) were obtained from fracture callus
   (day 3–5 after injury) tissues during open reduction and internal
   fixation procedures. Procurement and handling were in accordance with
   the guidelines set by the Stanford University Institutional Review
   Board (IRB-35711) and the UC Davis Institutional Review Board
   (IRB-1997852). Informed consent was not required as samples were
   considered biological waste. No restrictions were made regarding the
   race, gender, or age of the specimen’s donor. Following excision, all
   specimens were placed on ice, and hSSCs were isolated as described
   below. Early passage (3–5) VeraVec HUVEC cells were used for
   endothelial assays as further explained below.

Micro-computed tomography and dual-energy x-ray absorptiometry (DEXA)
analysis

   Soft tissue-free femurs were scanned within 2 h of dissection using a
   Bruker Skyscan 1276 (Bruker Preclinical Imaging) with a source voltage
   of 85 kV, a source current of 200 μA, a filter setting of AI 1 mm, and
   a pixel size of 17.5 μm at 2016 × 1344. Reconstructed samples were
   analyzed using CT Analyser (CTan) v.1.17.7.2 and CTvox v.3.3.0 software
   (Bruker). Anatomical landmarks as per ASBMR guidelines^[180]40 were
   used to set the region of interest for analyzing trabecular (200
   consecutive sections) and cortical (100 consecutive sections) bone
   parameters. Subsequent mechanical strength testing was conducted using
   3-point bending. The maximum load (N) sustained prior to fracture was
   recorded^[181]41. Whole-body DEXA imaging was performed at baseline
   (day of surgery) and 14 days post-surgery to determine lumbar spine
   (L5) BMD. Mice were anesthetized with isoflurane and placed in a
   cabinet x-ray system (Mozart®, Kubtec Medical Imaging) for analysis.

Histology

   Soft-tissue-free specimens were fixed in 4% PFA at 4 °C overnight.
   Samples were decalcified in 400 mM EDTA (EDTA) in PBS (pH 7.2) at 4 °C
   for 2 weeks with a change of EDTA every other day. The specimens were
   then dehydrated in 30% sucrose at 4 °C overnight. Specimens were
   embedded in optimal cutting temperature compound (OCT) and sectioned at
   5 µm. Representative sections were stained with freshly prepared
   hematoxylin and eosin (H&E) or Tartrate-resistant acid phosphatase
   (TRAP). Immunofluorescence on sections of cryopreserved long bone and
   ectopic bone specimens was incubated with 3% Bovine Serum Albumin in
   Tris Buffered Saline (TBS) for 1 h. Then, samples were probed with
   primary antibody (Endomucin 1:200, cat# sc-65495, Santa Cruz; CD31
   1:200: cat# AF3628, ThermoFisher; Basigin 1:200, cat# 16-1471-82,
   ThermoFisher and cat# NB500-430, Novus Biologicals) diluted in 1%
   BSA/PBS and incubated in a humidified chamber at 4 °C overnight. The
   specimens were washed with PBS three times. Secondary antibody (AF488
   dk anti-rat 1:500: Donkey anti-Rat IgG (H + L) Highly Cross-Adsorbed
   Secondary Antibody, Alexa Fluor™ 488, cat# A21208; Donkey anti-Goat IgG
   (H + L) 1:500 Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647, cat#
   A21447, both ThermoFisher) was applied for 15 min at room temperature
   in the dark. Specimens were also incubated with 1 µg/mL of DAPI for
   10 min and then washed twice. The specimens were then mounted with a
   coverslip using Fluoromount-G and imaged. ImageJ
   ([182]http://wsr.imagej.net/distros/osx/ij152-osx-java8.zip;
   RRID:SCR_003070) was used to quantify Endomucin-positive blood
   vessel/sinusoid lumen area. H&E stains were analyzed using image
   deconvolution according to Landini et al.^[183]42. The bone marrow area
   spanning 1 mm below the growth plate between the cortical bone region
   was selected, and the image deconvolution plugin, Colour Deconvolution
   2, was run on the selected region. The H&E 2 filter within the plug-in
   was applied. The color threshold on panel “Colour_2” was adjusted to
   represent the Eosin-stained bone regions. To measure the BV/TV, the
   partial (bone) area and total area were analyzed.

Bone histomorphometry analysis

   Mice were injected with 20 mg/kg of Calcein (Sigma-Aldrich, St. Louis,
   MO, USA) 8 and 2 days before euthanasia. Histological sections of bones
   were prepared as described above. A standard sampling site in the
   secondary spongiosa of the distal metaphysis was established. Mineral
   apposition rate (MAR), bone formation rate (BFR) and osteoclast
   number/bone surface (Oc/BS) were calculated.

Flow cytometric isolation of skeletal progenitor cells

   SSC lineage populations were isolated using cell-surface-marker
   profiles as previously described^[184]10,[185]12. In brief, femurs were
   dissected, cleaned of soft tissue and crushed using a mortar and
   pestle. Then, the tissue was digested in M199 (cat# 11150067,
   ThermoFisher Scientific) with 2.2 mg/mL collagenase II buffer (cat#
   C6885, Sigma-Aldrich) at 37 °C for 60 min. Dissociated cells were
   strained through a 100-μm nylon filter, washed in staining medium (10%
   fetal bovine serum (FBS) in PBS) and pelleted at 200 g at 4 °C. The
   cell pellet was resuspended in staining medium, and red blood cells
   were depleted via ACK lysis for 5 min. The cells were washed again in
   staining medium and pelleted at 200 g at 4 °C. Then, the cells were
   prepared for flow cytometry with fluorochrome-conjugated antibodies.
   For mouse SSC lineages: CD90.1 (1:200, ThermoFisher, 47–0900), CD90.2
   (1:200, ThermoFisher, 47–0902), CD105 (1:200, ThermoFisher, 13–1051),
   CD51 (1:200, BD Biosciences, 551187), CD45 (1:400, BioLegend, 103110),
   Ter119 (1:400, ThermoFisher, 15–5921), Tie2 (1:100, ThermoFisher,
   14–5987), 6C3 (1:200, BioLegend, 108312), streptavidin PE-Cy7 (1:400,
   ThermoFisher, 25–4317), Sca-1 (1:200, ThermoFisher, 56-5981), CD45
   (1:400, ThermoFisher, 11-0451), CD31 (1:200, ThermoFisher, 12-0311) and
   CD24 (1:200, ThermoFisher, 47–0242). For human SSC lineages: CD45
   (1:200, BioLegend, 304029), CD235a (1:200, BioLegend, 306612), CD31
   (1:200, ThermoFisher Scientific, 13-0319), CD202b (TIE-2) (1:100,
   BioLegend, 334204), streptavidin APC-AlexaFlour750 (1:400,
   ThermoFisher, SA1027), CD146 (1:200, BioLegend, 342010), PDPN (1:200,
   ThermoFisher Scientific, 17-9381), CD164 (1:200, BioLegend, 324808) and
   CD73 (1:200, BioLegend, 344016). Flow cytometry was conducted on a FACS
   Aria II Instrument (BD BioSciences) using a 70-μm nozzle in the Shared
   FACS Facility in the Lokey Stem Cell Institute (Stanford). The skeletal
   stem-cell lineage gating strategy was determined using appropriate
   isotype and fluorescence-minus-one controls. Propidium iodide staining
   was used to determine cell viability. Cells were sorted for purity.
   Flow cytometric analysis was conducted using FlowJo (FLOWJ LLC,
   v10.10). Blood panels and bone marrow analysis were conducted on a
   CYTEK Aurora Sorter. Blood was stained with Ter119-PE-Cy5 (1:400,
   116210, BioLegend), CD45-FITC (1:400, 11-0451, Invitrogen),
   B220-APC-Cy7 (1:200, 103224, BioLegend), CD11b-PE-Cy7 (1:200, 101216,
   BioLegend), CD3-APC (1:200, 100236, BioLegend), Gr1-BV711 (1:200,
   108443, BioLegend). CD11b^+ and Gr1^+ cells within CD45+ were
   considered myeloid lineage, while CD45 + CD11b-Gr-CD3+ and B220+ were
   marked as lymphoid lineage. HSC lineage populations were stained with
   Lineage cocktail-Pacific Blue (1:50, 133310, BioLegend), CD127-BV711
   (1:200, 135035, BioLegend), CD117-APC-Cy7 (1:200, 105826, BioLegend),
   Sca1-APC (1:200, 160904, BioLegend), CD16/32-PE (1:200, 156606,
   BioLegend), CD34-BV786 (1:100, 742971, BD Biosciences), CD135-PE-Cy5
   (1:200, 135312, BioLegend), CD150-BV510 (1:200, 115929, BioLegend).
   DAPI was used as live stain.

Mouse cell culture

   Cells were cultured in α-MEM with 10% FBS and 1%
   penicillin–streptomycin (ThermoFisher Scientific; 15140-122). For in
   vitro osteogenic and chondrogenic differentiation assays, 12,000
   FACS-purified mouse SSCs were cultured to sub-confluency in expansion
   media. Cells were washed in PBS, trypsinized and transferred to a
   96-well plate with osteogenic differentiation medium containing 10%
   FBS, 100 μg/mL ascorbic acid and 10 mM β-glycerophosphate in α-MEM for
   14 days. Alternatively, for chondrogenic differentiation micromass
   cultures were generated by a 5-μl droplet of cell suspension with
   approximately 1.5 × 10^7 cells per mL pipetted in the center of a
   24-well plate and cultured for 2 h in the incubator before adding warm
   chondrogenic medium consisting of α-MEM (high glucose) with 10% FBS,
   100 nM dexamethasone, 1 μM L-ascorbic acid-2-phosphate and 10 ng/mL
   TGF-β1 (Invitrogen, cat# PHG9204). The micromass was maintained for 21
   days with media changes every other day. At the end of differentiation,
   cells were washed in PBS, fixed in 4% PFA and stained with Alizarin Red
   S (cat# A5533-25G, Sigma-Aldrich) solution (osteogenesis) or Alcian
   Blue (cat# A3157, Sigma-Aldrich) solution (chondrogenesis). For
   osteogenic potential, Alizarin Red S stain was dissolved in 300 µL of
   20% Methanol/10% Acetic Acid solution. After complete liberation of
   staining, 80 µL of each well was transferred into a new 96-well plate
   in duplicates and absorbance was measured at 450 nm. Chondrogenic
   potential was assessed by spectrophotometrically measuring absorption
   at 595 nm. For osteoclastogenesis assays flushed bone marrow cells from
   long bones were plated in 24-well plates at a density of 200,000 cells
   per well with α-MEM without phenol red, 1% GlutaMAX supplement (cat#
   35050061, Gibco), 10% FBS, 1% Penicillin-Streptomycin 10,000 U/mL, 1 µM
   prostaglandin E2 (cat# P0409, Sigma), and 10 ng/mL Csf1 recombinant
   murine protein (cat# 315-02, Peprotech) for 3 days. Starting on day
   three, the media was changed daily to also include 10 ng/mL recombinant
   mouse RANKL (cat# 315-11, Peprotech). Osteoclast culture continued for
   10 days until large, multinucleated osteoclasts appeared. Plates were
   stained for osteoclasts using the TRAP kit (cat# 387A, Sigma-Aldrich).

Human SSC (hSSC) culture

   Freshly sorted, primary hSSCs were cultured in α-MEM with 10% human
   platelet-derived lysate (HPL; cat# 06962, STEMCELL technologies), 1%
   Penicillin-Streptomycin, 0.01% heparin and maintained at 37 °C with 5%
   CO[2]. For in vitro differentiation assays, hSSCs were treated and
   supplemented with the same osteogenic and chondrogenic differentiation
   protocols as described for mouse cells above. For viral overexpression
   experiments, the coding region of Basigin (269aa) was cloned into a
   pCCLc backbone for manufacture of second-generation lentivirus using
   Lenti-X 293T as packaging cells, as previously described^[186]43. The
   viral titer was determined functionally, based on the amount of virus
   necessary to reach 90–95% eGFP expression. Cells were transduced using
   Lipofectamin3000 (cat# L3000001, ThermoFisher). For rescue experiments
   media was either supplemented with hPTH 1-34 by adding a final
   concentration of 2.5 nM 6 h before each media change (to resemble
   intermittent exposure) or with 1 µg/mL of monoclonal Basigin antibody
   (cat# NB55-430, Novus Biologicals) during media change. Reactive oxygen
   species measurement was performed according to the manufacturer’s
   instructions (cat# ab113851, Abcam).

Cell line experiments

   Low passage VeraVec endothelial cells were thawed in a 10 cm dish with
   10 mL of EGM-2 medium (cat# CC-3156, Lonza Bioscience). VeraVecs were
   grown at 37 °C with 5% CO[2] until confluent, with media changes every
   3 days. Before the assays, VeraVec culture was starved with low serum
   medium. For migration (wound scratch assay)^[187]44, VeraVecs were
   lifted with 0.05% trypsin, spun down and resuspended in supernatant
   from 48 h cultured SSCs. They were then plated at 1.5 × 10^5 cells/well
   in a 24-well plate. To induce a gap for the migration, a scratch assay
   was generated using a 1000 µL micropipette tip, scratching vertically
   from one side of the well to the other. The gap was imaged, and area
   measured at times 0 h and 12 h. The area healed was calculated by (area
   at 0 h − area at 12 h)/(area at 0 h) × 100%. For tube formation, a
   matrix (cat# A1413201, Geltrex, ThermoFisher) was deposited in wells of
   a 48-well plate at an amount of 50 µL/cm^2, for a total of 50 µL per
   well. As for migration, VeraVecs were lifted and resuspended in
   supernatant from SSCs. VeraVecs were plated at 1.5 × 10^4 cells/well.
   Cells were imaged at 12 h.

Subcutaneous xenografts

   Primary hSSCs were transplanted into the dorsum of 3-month-old
   immunodeficient NSG mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ; JAX:
   005557) as described previously^[188]45. Briefly, freshly sorted
   patient-derived hSSCs were sorted, expanded to confluency and virally
   transduced to either overexpress Basigin or control. 1 × 10^6 cells
   were mixed with 5 µL Matrigel and seeded on macroporous composite
   scaffolds formed of hydroxyapatite (HA) and poly(lactide-co-glycolide)
   (PLG) (HA-PLG) on ice. Scaffolds were fabricated using a gas
   foaming/particulate leaching method as previously described^[189]46.
   Microspheres composed of PLG (85:15; DLG 7E; Lakeshore Biomaterials,
   Birmingham, AL) were prepared using a double-emulsion process and
   lyophilized. Lyophilized microspheres (7.1 mg) were combined with
   17.8 mg of synthetic HA (particle size, <200 nm; Aldrich Chemistry, St.
   Louis, MO) and 134.9 mg of NaCl (300 to 500 μm in diameter) to yield a
   2.5:1:19 mass ratio of ceramic:polymer:salt. The powdered mixture was
   compressed under 2 metric tons for 1 min to form solid disks (final
   dimensions, 8 mm in diameter and 1.5 mm in height) using a Carver Press
   (Carver Inc., Wabash, IN). Compressed disks were exposed to
   high-pressure CO[2] gas (5.5 MPa) for at least 24 h, followed by rapid
   pressure release to prompt polymer fusion. Salt particles were leached
   from scaffolds in distilled H[2]O for 24 h to generate HA/PLG composite
   scaffolds. HA/PLG composite scaffolds were cut with a biopsy punch to
   produce scaffolds with final dimensions of 4 mm in diameter and 1.5 mm
   in height. Scaffolds were sterilized in 70% ethanol in 24-well plates
   for 20 min, followed by two rinses in sterile PBS. Sterile scaffolds
   were dried and kept until use. To promote cell adhesion, scaffolds were
   incubated in culture media at 37 °C with 5% CO[2] for 30 min directly
   before cell seeding. A small skin incision was made in the dorsum of
   anesthetized NSG mice, and the cell containing scaffold was slid under
   the skin. Interrupted sutures were applied to close the incision, and
   transplants were allowed to engraft and grow for 6 weeks.

Renal capsule transplantation

   Renal capsule transplantations were conducted as previously
   described^[190]10,[191]12. Briefly, in the anaesthetized mouse, a 5-mm
   dorsal incision was made, and the kidney was exposed manually. Then, a
   2-mm incision was created in the renal capsule using a needle bevel,
   and 5000 FACS-purified mouse SSCs resuspended in 2 μl of Matrigel were
   transplanted beneath the capsule. The kidney was re-approximated
   manually, and incisions were closed using sutures and staples. At the
   time of surgery, mice were randomly divided into four groups. Mice
   either received subcutaneous placebo, hPTH or Methylprednisolone
   pellets as described above. One additional Methylprednisolone pellet
   group was treated with hPTH (see above). Grafts were collected after 21
   days.

Single-cell RNA-sequencing

   Femurs were collected from all the treatment groups, and each was
   processed, digested and prepared for FACS as described above. Single
   cell solutions of each treatment group were then pooled (n = 5 per
   group), and 1 × 10^6 PI-Ter119-CD45+ and 1 × 10^6 PI-Ter119-CD45- cells
   were sorted into one collection tube containing FACS buffer for each
   experimental group. Cells were then processed with 10X Chromium Next
   GEM Single Cell 3’ GEM kit (10X Genomics, v.3.1) according to the
   manufacturer’s instructions to target 5000 cells per group, sequenced
   on a partial lane Illumina NovaSeq platform. Barcoded samples were
   demultiplexed, aligned to the mouse genome (GRCm39.104), and
   UMI-collapsed with the Cellranger toolkit with standard settings
   (v.7.1.0, 10X Genomics). We used the Scanpy package (v.1.9.1.) to
   explore the data. First, quality filtering to exclude multiplets and
   cells of poor quality was conducted by only keeping cells with a gene
   count of more than 250 and fewer than 3000, with less than 15%
   mitochondrial and 15% ribosomal gene content, leaving 5735 cells. Genes
   expressed in fewer than three cells across all cells were also removed
   from downstream analysis. Data were log-normalized, cell
   cycle-regressed and scaled for analysis. Dimensionality reduction and
   Leiden clustering, as well as subclustering, were conducted, choosing
   parameters based on PCA elbow plots. For single-cell RNA-sequencing of
   SSC-derived grafts, the tissue was dissected out and processed as
   described for femur bone processing to isolate SSCs above. Single cell
   solution was then stained with PI and Ter119, and living (PI-negative),
   non-red blood cells (Ter119-negative) were sorted into FACS buffer for
   each group. Based on yield, cells were then processed with 10X Chromium
   Next GEM Single Cell 3’ GEM kit (10X Genomics, v.3.1) according to the
   manufacturer’s instructions to target 1000 cells for Placebo, hPTH and
   GC + hPTH groups, as well as 500 cells for GC group. All other steps
   were conducted as described above. A total of 852 graft-derived cells
   across all groups passed stringent quality filtering. Data are
   available under GEO accession number [192]GSE253044.

Statistical analysis

   Statistical significance between placebo and treatment groups was
   determined using two-tailed, unpaired Student’s t-test or One-way ANOVA
   for multiple groups comparison with LSD Fisher test unless stated
   otherwise in the figure legend (GraphPad Prism; version 10).
   Statistical significance was defined as p < 0.05. All data points refer
   to biological replicates and are presented as mean ± standard error of
   the mean (SEM) unless otherwise stated in the figure legend.

Reporting summary

   Further information on research design is available in the [193]Nature
   Portfolio Reporting Summary linked to this article.

Supplementary information

   [194]Supplementary Information^ (17.7MB, pdf)
   [195]Reporting Summary^ (104.5KB, pdf)
   [196]Transparent Peer Review file^ (5.4MB, pdf)

Source data

   [197]Source Data^ (40.2KB, xlsx)

Acknowledgements