Gli1 identifies osteogenic progenitors for bone formation and fracture repair

Abstract

Bone formation in mammals requires continuous production of osteoblasts throughout life. A common molecular marker for all osteogenic mesenchymal progenitors has not been identified. Here, by lineage-tracing experiments in fetal or postnatal mice, we discover that Gli1⁺ cells progressively produce osteoblasts in all skeletal sites. Most notably, in postnatal growing mice, the Gli1⁺ cells residing immediately beneath the growth plate, termed here "metaphyseal mesenchymal progenitors" (MMPs), are essential for cancellous bone formation. Besides osteoblasts, MMPs also give rise to bone marrow adipocytes and stromal cells in vivo. RNA-seq reveals that MMPs express a number of marker genes previously assigned to mesenchymal stem/progenitor cells, including CD146/Mcam, CD44, CD106/Vcam1, Pdgfra, and Lepr. Genetic disruption of Hh signaling impairs proliferation and osteoblast differentiation of MMPs. Removal of β-catenin causes MMPs to favor adipogenesis, resulting in osteopenia coupled with increased marrow adiposity. Finally, postnatal Gli1⁺ cells contribute to both chondrocytes and osteoblasts during bone fracture healing. Thus Gli1 marks mesenchymal progenitors responsible for both normal bone formation and fracture repair.

Fig. 1

Embryonic Gli1⁺ cells give rise to multiple cell types in postnatal mice. Gli1-CreER^T2; Ai9 mice were administered tamoxifen (TM) at E13.5 and harvested at E14.5 a–d, E18.5 e–g, or 2 months of age h–l. Red: tdTomato⁺ cells; blue: nuclear staining by DAPI. Scale bars: 100 µm a, d, m, q, s or 500 µm e, g, h, l, n, p. a–c, e–f, h, k Representative confocal images from frozen sections of the femur. Boxed areas are shown at a higher magnification in corresponding panels to the right b, c, f, i–k. Rv Ranvier’s groove, SM skeletal muscle, GP growth plate. 1°: primary ossification center; 2°: secondary ossification center. Green arrow: chondrocytes in growth plate; white arrow: osteocyte; yellow arrow: osteoblast on periosteal surface. d, g, l H&E staining of sister sections corresponding to a, e, and h, respectively. m–o Representative confocal images from frozen sections through the lambdoid suture m or lumbar vertebrae n, o. Boxed area in n is shown at a high magnification in o. P parietal bone, Ip interparietal bone. p H&E staining of an adjacent section to n. q–t Immunofluorescence staining of Perilipin q, r or Lepr s, t on frozen sections of distal metaphyseal q, r or diaphyseal s, t bone marrow region of the femur. Boxed areas in q and s are shown at a high magnification in r and t, respectively. Arrowheads in r denote tdTomato⁺ adipocytes. Numbers in t indicate percentage (mean ± SD, n = 3) of tdTomato⁺Lepr⁺ over total tdTomato⁺ cells on two sections per mouse and three mice for each genotype

Fig. 2

Postnatal Gli1⁺ cells contribute to osteoblasts in both endochondral and intramembranous bones. Gli1-CreER^T2; Ai9 mice were treated with TM at 1 month of age and analyzed at 24 h a, b, k, l, q; 1 month c, d, m, n; 3 months e–g; or 9 months h–j, o, p, r after the last TM dosing. Representative images from three mice for each time point are presented. Red: tdTomato⁺ cells; blue: nuclear staining by DAPI. Scale bars: 500 µm a–o or 100 µm q, r. a–p Representative confocal images of frozen sections of the femur. Boxed areas are shown at a high magnification in corresponding panels to the right. GP growth plate. Green arrow: trabecular bone. CB cortical bone, M marrow. q, r Representative confocal image showing sections through the lambdoid suture of the skull. P parietal bone, Ip interparietal bone

Fig. 3

Gli1⁺ cells are critical for cancellous bone formation in postnatal mice. a–j Postnatal Gli1⁺ cells produce osteoblasts. Gli1-CreER^T2; Ai9; ColI-GFP mice were treated with TM at 1 month of age and analyzed at 24 h a, b, g, h; 1 month c, d, i, j; or 3 months e, f after the last TM dosing. Representative confocal images are shown for tibia a–f and lumbar vertebrae g–j. Boxed areas are shown at a higher magnification in corresponding panels to the right. Percentage (mean ± SD, n = 3) of tdTomato⁺GFP⁺ over total GFP⁺ cells a–f was calculated from cancellous bone region extending 300 µm from growth plate on two sections each of three mice. Scale bar: 500 µm. k–q Postnatal Gli1⁺ cells are necessary for cancellous bone formation. DTA (Gli1-CreER^T2; Ai9; Rosa-DTA) or Ctrl (Gli1-CreER^T2; Ai9) littermates were treated with TM once daily for 3 consecutive days starting at 1 month of age and harvested at 3 weeks after last dosing. k, l Loss of tdTomato⁺ cells in the distal femur of DTA (l) compared to Ctrl mice (k) at the time of harvest. Scale bar: 500 µm. m, n Representative μCT images from cancellous bone of distal femur. o Quantification of cancelluous bone mass (BV/TV) by µCT in distal femur. p Serum levels of P1NP (left) and CTX-I (right) at the time of harvest. q In vivo microCT analyses of cancellous bone parameters in distal femur of mice immediately before TM (white bar, “Before”) or immediately before harvest (black bar, “After”). BV/TV bone volume over tissue volume, Tb.N trabeculae number, Tb.Th trabeculae thickness, Tb.Sp trabeculae spacing. n = 3, *p < 0.05, paired Student's t-tests

Fig. 4

Gli1 marks MMPs in postnatal growing mice. a–f Confocal images of immunofluorescence staining for Endomucin a, b, Aggrecan c, d, or Lepr e, f on frozen sections of the proximal tibia. Boxed areas shown in magnified images to the right. GP growth plate. Arrowhead in f denotes a typical Lepr⁺tdTomato⁺ cell. Percentage (mean ± SD, n = 3) of double positive cells over total tdTomato⁺ cells was calculated from cancellous bone region extending 300 µm from growth plate in two sections each of three mice. Same below. Scale bars: 100 µm (a, e) or 500 µm c. g, h Confocal images of direct fluorescence on frozen sections of proximal tibia. Boxed area shown at a higher magnification to the right. Arrowheads in h denote Pdgfra⁺tdTomato⁺ cells. Scale bar: 100 µm g. i FACS analyses of Lepr (left) or Pdgfra (right) expression among CD31⁻CD45⁻Ter119⁻tdTomato⁺ cells from the cancellous bone region beneath the growth plate of either distal femur or proximal tibia. Red and blue lines represent control IgG and specific antibody, respectively. j–o Confocal images of Runx2 j, k or Osx l–o immunofluorescence staining on frozen sections of the proximal tibia. Boxed areas are shown in magnified images to the right k, m, o. DAPI signal was omitted in k, m, o for better visualization. Scale bar: 100 µm. p FACS analyses of Lepr (left) or Pdgfra (right) expression among CD31⁻CD45⁻Ter119⁻tdTomato⁺ from the bone marrow. Red and blue lines represent the control and specific antibody, respectively. Quantification is presented as mean ± SD, n = 3. q–t Confocal images of immunofluorescence staining for perilipin on frozen sections of distal femur. Boxed areas are shown at higher magnification to the right. Percentage (mean ± SD, n = 3) of perilipin⁺tdTomato⁺ over perilipin⁺ adipocytes was acquired from cancellous bone region extending 400 µm from the growth plate in two sections each of three mice. Scale bar: 500 µm. u–z Confocal images showing Gli1⁺ cells (red) and immunofluorescence signal of Aggrecan (green) in distal femur of Gli1-CreER^T2;Ai9 mice treated with TM at different ages and harvested after 24 h. Scale bar: 500 µm. Boxed areas are shown at higher magnification to the right v, x, z. (AA) Differential expression of MSC marker genes in Gli1⁺ MMPs vs. Gli1⁻ mesenchymal cells. Heatmap generated from RNA-seq data of three paired biological replicates. Scale derived from voom-transformed data indicating relative increase (red) or decrease (blue) in MMPs

Fig. 5

Hh signaling is required for normal MMP proliferation and differentiation. Gli1-CreER^T2;Ai9;Smo^c/c (SmoCKO) or Gli1-CreER^T2;Ai9 (Ctrl) mice treated with TM at 1 month of age and harvested after 10 days. GP growth plate. Scale bar: 100 µm. a, b Representative images a or quantification of bone parameters b from µCT analyses of cancellous bone region of distal femur. c EdU staining on sections of proximal tibia. Arrowheads indicate EdU⁺tdTomato⁺ cells. d Percentage (mean ± SD) of EdU⁺tdTomato⁺ over tdTomato⁺ cells quantified within cancellous bone region extending 300 µm from the growth plate of proximal tibia in two sections per mouse and three mice per genotype. e Immunofluorescence staining of Osx on sections of the proximal tibia. Arrowheads denote Osx⁺tdTomato⁺ cells. f Percentage (mean ± SD) of Osx⁺tdTomato⁺ over tdTomato⁺ cells quantified within cancellous bone region extending 300 µm from the growth plate of proximal tibia in two sections per mouse and three mice per genotype. *p < 0.05, n = 3, paired Student's t-tests

Fig. 6

β-Catenin deletion in MMPs leads to osteopenia and fate switch toward adipogenesis. Gli1-CreER^T2; Ai9 (Ctrl), or Gli1-CreER^T2;Ai9;β-catenin^c/c (βcatCKO) littermate mice were treated with TM at 1 month of age and analyzed after 1 month. a µCT images of cancellous bone in distal femur. b Trabecular bone parameters from µCT analyses of distal femur. c Cortical bone parameters from µCT analysis of femoral diaphysis. Tt.A, total area, Ct.Ar cortical area, Ct.Th cortical thickness. d H&E staining for distal femur. e Serum CTX-I levels. f Immunofluorescence staining of Perilipin (red) on sections of distal femur. Boxed areas shown in magnified images to the right. Scale bar, 100 µm. GP growth plate. g Number of Perilipin⁺ adipocytes within cancellous bone region extending 400 µm from the growth plate of distal femur in two sections per mouse and three mice per genotype. Data represent mean ± SD. h Area of Perilipin⁺ adipocytes (meansured by imageJ) from the same area as g. Data was collected from two sections per mouse and three mice per genotype. A total of 80 adipocytes in each genotype (each represented by one data point) were quantified. i Confocal images showing tdTomato⁺ cells (red) and immunofluorescence staining of Perilipin staining (green) on sections of distal femur. Boxed areas are shown in magnified images to the right. Scale bar: 100 µm. Arrowheads indicate Perilipin⁺tdTomato⁺ adipocytes. Percentage (mean ± SD) of Perilipin⁺tdTomato⁺ over Perilipin⁺ adipocytes was calculated within cancellous bone region extending 400 µm from the growth plate in two sections per mouse and three mice per genotype. *p < 0.05, n = 3, paired Student's t-tests

Fig. 7

Gli1⁺ cells produce chondrocytes and osteoblasts during fracture healing. Gli1-CreER^T2; Ai9 mice were administered TM at 1 month of age and subjected to fractures 6 weeks later. Mice were harvested at 10 days after fracture. Scale bar: 500 µm. a Representative confocal image of contralateral (unfractured) femoral sections at the time of harvest. Note relatively little tdTomato signal (red) in periosteum. b–d Representative confocal image of femoral sections through fracture site, with immunofluorescence staining of Aggrecan (green) to demarcate chondrocytes. Hash symbol indicates position of pin used for fracture stabilization. Boxed areas of cartilage callus (box 1) and bony region (box 2) are shown at higher magnification in c and d, respectively. Percentage (mean ± SD, n = 3) of tdTomato⁺Aggrecan⁺ over Aggrecan⁺ cells in c was calculated from three mice (one representative section per mouse). e, f Immunofluorescence staining of Ocn on bony callus. Boxed area in e is shown at higher magnification in f. Percentage (mean ± SD, n = 3) of tdTomato⁺;Ocn⁺ over Ocn⁺ osteoblasts was calculated from three mice (one representative section per mouse)