Itm2a expression marks periosteal skeletal stem cells that contribute to bone fracture healing

Abstract

The periosteum contains skeletal stem/progenitor cells that contribute to bone fracture healing. However, the in vivo identity of periosteal skeletal stem cells (P-SSCs) remains unclear, and membrane protein markers of P-SSCs that facilitate tissue engineering are needed. Here, we identified integral membrane protein 2A (Itm2a) enriched in SSCs using single-cell transcriptomics. Itm2a+ P-SSCs displayed clonal multipotency and self-renewal and sat at the apex of their differentiation hierarchy. Lineage-tracing experiments showed that Itm2a selectively labeled the periosteum and that Itm2a+ cells were preferentially located in the outer fibrous layer of the periosteum. The Itm2a+ cells rarely expressed CD34 or Osx, but expressed periosteal markers such as Ctsk, CD51, PDGFRA, Sca1, and Gli1. Itm2a+ P-SSCs contributed to osteoblasts, chondrocytes, and marrow stromal cells upon injury. Genetic lineage tracing using dual recombinases showed that Itm2a and Prrx1 lineage cells generated spatially separated subsets of chondrocytes and osteoblasts during fracture healing. Bone morphogenetic protein 2 (Bmp2) deficiency or ablation of Itm2a+ P-SSCs resulted in defects in fracture healing. ITM2A+ P-SSCs were also present in the human periosteum. Thus, our study identified a membrane protein marker that labels P-SSCs, providing an attractive target for drug and cellular therapy for skeletal disorders.

Keywords: Bone biology; Bone disease.

Figure 1. scRNA-Seq analysis shows that Itm2a can enrich SSCs in Prrx1 lineage periosteal cells.

(A) The isolation and scRNA-Seq workflow of Ai9⁺ periosteal cells from Prrx1-Cre R26-Ai9 mice. (B) t-SNE plot of collected Ai9⁺ cells. Cells were clustered into 13 subpopulations: stem/progenitor cells (clusters 0–3, 7, and 10), osteoblasts (clusters 4 and 12), ECs (clusters 5, 6, and 8), and muscle cells (clusters 9 and 11). (C–F) Feature plots showing the representative marker distribution in different cell types: periosteal stem/progenitor cells (Col3a1 and Itgbl1) (C), osteoblasts (Sp7 and Bglap) (D), muscle cells (Pax7 and Myob5) (E), and ECs (Emcn and Pecam1) (F). (G and H) Pseudotime analysis of the 3 states of Ai9⁺ cells. (I) Feature plots of CD105 and CD200 expression in Ai9⁺ cells. (J) t-SNE plot of the periosteal stem/progenitor cells (state 2, clusters 0, 1, 2, 3, 7, and 10). (K) Feature plots showing the distribution of Itm2a in P-SSCs. (L) Flow cytometric analysis of SSCs and progenitors in the murine periosteum. (M) Percentage of SSCs in Itm2a⁺ and Itm2a^– cell populations. Data are presented as the mean ± SD. n = 3. ***P < 0.001, by unpaired t test.

Figure 2. Functional characterization of Itm2a⁺ P-SSCs.

(A) Workflow of the kidney capsule transplantation and bone drill-hole transplantation experiment using actin-GFP mice. Itm2a^– and Itm2a⁺ GFP⁺ SSCs were sorted, cultured, and transplanted in a kidney capsule injection assay and a bone drill-hole model. (B) Representative μ-CT images of long bone defects at day 7 after injury and transplantation. (C) Bone volume/total volume (BV/TV) quantification of injury sites were assessed for the quantification of bone healing. n = 4. ***P < 0.001, by unpaired, 2-tailed t test. (D) Representative immunostaining images of long bone defects 7 days after injury and transplantation. Scale bar: 50 μm. (E) Representative μ-CT images of the renal grafts 2 months after transplantation of Itm2a^– and Itm2a⁺ SSCs. (F) Bone volume (BV) quantification of the renal grafts from G. n = 3. ***P < 0.001, by 2-tailed, unpaired t test. (G) Representative immunostaining images of the renal grafts 2 months after transplantation of Itm2a^– and Itm2a⁺ SSCs. Scale bar: 50 μm. (H and I) FACS analysis of Itm2a⁺ (H) and Itm2a^– (I) SSC-derived cells after transplantation. FSC-W, forward scatter width. Data are presented as the mean ± SD.

Figure 3. Itm2a-CreER labels P-SSCs.

(A) Strategy used to construct Itm2a-CreER-mCherry mice. KI, knockin. (B) Representative confocal images of Itm2a lineage cells (ZsGreen⁺) and Itm2a expression cell (mCherry⁺) distribution in periosteal (PO) and bone marrow (BM) regions in cells from Itm2a-CreER R26-Ai6 mice that had been tamoxifen treated at 4 weeks of age. Mice were analyzed 2 days after the treatment. Scale bars: 100 μm. (C) Representative confocal imaging of the osteoblast marker OPN and Itm2a lineage cells (ZsGreen⁺ cells) in femur sections from Itm2a-CreER R26-Ai6 mice that had been tamoxifen treated at 4 weeks of age. Mice were analyzed 2 days after the treatment. n = 3 mice per condition from 3 independent experiments. Scale bars: 50 μm. (D) Representative confocal images of the stem cell marker CD200 and Itm2a lineage cells (ZsGreen⁺ ) in femur sections from mice that had been tamoxifen treated at 4 weeks of age. Mice were analyzed 2 days after the treatment. n = 3 mice per condition from 3 independent experiments. Scale bars: 50 μm. (E) Flow cytometric analysis of SSCs in ZsGreen⁺ cells from long bone digests from mice that had been treated with tamoxifen at 4 weeks of age. Mice were analyzed 2 days after the treatment. n = 4.

Figure 4. Itm2a⁺ P-SSCs generate osteoblasts, chondrocytes, and marrow stromal cells upon injury.

(A) Experimental design for tamoxifen (Tam) induction, the bone fracture model, and tissue analysis. (B) Representative confocal images of Itm2a lineage cells (ZsGreen⁺) at the uninjured periosteum (left) and femoral fracture day-3 callus (right) in sections from Itm2a-CreER R26-Ai6 mice. Scale bar: 100 μm. (C) Representative 7 dpf confocal images of Itm2a lineage cells (ZsGreen⁺) colocalized with chondrocytes (COL2A1) at the fractured femur site in sections from Itm2a-CreER R26-Ai6 mice. Scale bars: 100 μm. (D) Representative 14 dpf confocal images of Itm2a lineage cells (ZsGreen⁺) colocalized with osteoblasts (OPN) at the fractured femur site in sections from Itm2a-CreER R26-Ai6 mice. Scale bars: 100 μm. (E) Representative 14 dpf confocal images of Itm2a lineage cells (ZsGreen⁺ ) colocalized with bone marrow stromal cells (LepR) at the fractured femur site in sections from Itm2a-CreER R26-Ai6 mice. Scale bars: 50 μm. (F) Percentage of Itm2a lineage cells (ZsGreen⁺) among COL2A1⁺ chondrocytes, OPN⁺ osteoblasts and LepR⁺ bone marrow stromal cells calculated from C, D, and E, respectively. n = 3 mice per condition. Data are presented as the mean ± SD. (G) Experimental design for tamoxifen induction, bone drill model, and tissue analysis. (H) Representative confocal images of femoral fracture sites in sections from Itm2a-CreER R26-Ai6 mice at 14 dpf after tamoxifen treatment. n = 4 mice per condition from 4 independent experiments. Scale bars: 100 μm. (I) Experimental design for tamoxifen induction, bone scratch model, and tissue analysis. (J) Representative confocal images of injury sites in tibia sections from Itm2a-CreER R26-Ai6 mice at 7 dpf and 14 dpf after tamoxifen administration. Scale bars: 100 μm. n = 4 mice per condition from 4 independent experiments.

Figure 5. Itm2a and Prrx1 lineage cells generate spatially separated subsets of chondrocytes and osteoblasts during fracture healing.

(A) Strategy used to generate Prrx1-CreER R26-LSL-Ai6 Itm2a-DreER R26-RSR-tdTomato mice. (B) Experimental design for tamoxifen induction, bone fracture model, and tissue analysis. dpf, days post fracture. (C and D) Representative confocal images of Itm2a lineage cells (tdTomato⁺) and Prrx1 lineage cells (ZsGreen⁺ cells) in femoral (C) and vertebral (D) sections from Prrx1-CreER R26-LSL-Ai6 Itm2a-DreER R26-RSR-tdTomato mice that had been tamoxifen treated at 4 weeks. Mice were analyzed 1 week after the treatment. n = 3 mice per condition from 3 independent experiments. Scale bars: 100 μm. (E) Representative 7 dpf confocal images of Itm2a lineage cells (tdTomato⁺) and Prrx1 lineage cells (ZsGreen⁺) colocalized with chondrocytes (COL2A1) at the fractured femur site in sections from Prrx1-CreER R26-LSL-Ai6 Itm2a-DreER R26-RSR-tdTomato mice. Scale bars: 100 μm. (F) Representative 14 dpf confocal images of Itm2a lineage cells (tdTomato⁺) and Prrx1 lineage cells (ZsGreen⁺) colocalized with osteoblasts (OSX⁺) at the fractured femur site in sections from Prrx1-CreER R26-LSL-Ai6 Itm2a-DreER R26-RSR-tdTomato mice. Scale bars: 100 μm. (G and H) Percentage of 3-cell populations among COL2A1⁺ chondrocytes and OSX⁺ osteoblasts calculated from E and F, respectively. n = 3 mice per condition. Data are presented as the mean ± SD.

Figure 6. BMP signaling is critical for the Itm2a⁺ P-SSC contribution to bone fracture healing.

(A and B) Bulk RNA-Seq data from Lin^–Itm2a^– and Lin^–Itm2a⁺ periosteal cells isolated from 4-week-old WT mice. GO plot (A) and heatmap (B) show that BMP signaling was enriched in Lin^–Itm2a⁺ periosteal cells. (C) Experimental design for tamoxifen induction, the bone fracture model, and tissue analysis of Itm2a-CreER Bmp2^fl/fl and control mice. (D) Analysis by x-ray of fractured femurs in Itm2a-CreER Bmp2^fl/fl mice and control mice at 7, 14, and 21 dpf. n = 4 mice per genotype per time point. (E) Callus index calculated from x-ray data showed callus formation at 7, 14, and 21 dpf in mutant and control mice. n = 4 mice per group. Data are presented as the mean ± SD (n = 4). (F) μ-CT (upper) and SO/fast green (SOFG) (lower) staining analysis of fractured femurs in Itm2a-CreER Bmp2^fl/fl mice and control mice at 14 dpf. n = 4 mice per genotype. Scale bar: 100 μm. (G) Bone volume quantification showed callus formation at 14 dpf in mutant and control mice. n = 4 mice per group. Data are presented as the mean ± SD (n = 4). ***P < 0.001, by unpaired, 2-tailed t test. (H) Experimental design for tamoxifen induction, the bone fracture model, and tissue analysis of Itm2a-CreER R26-DTA and R26-DTA mice. (I) Confocal images of Itm2a⁺ cell (anti-Itm2a staining) at the periosteal region from R26-DTA (left) and Itm2a-CreER R26-DTA (right) mice. Scale bar: 50 μm. (J) μ-CT (upper) and SOFG (lower) staining analysis of fractured femurs in R26-DTA and Itm2a-CreER R26-DTA at 14 dpf. Scale bar: 100 μm. (K) Non-union percentage calculated from R26-DTA and Itm2a-CreER R26-DTA mice at 28 dpf. *P = 0.025, by Fisher’s exact test.

Figure 7. ITM2A enriches P-SSCs in human periosteum samples.

(A) Representative image of a human periosteum specimen. (B) Immunostaining for ITM2A in human fibula sections. Arrows indicate ITM2A⁺ cells in the periosteal region. Scale bars: 100 μm. COR, cortical bone; PS, periosteum. (C) Flow cytometric analysis of the expression of human (H) SSC markers in ITM2A⁺ cells digested from human fibular periosteum. FSC-H, forward scatter height. (D) In vitro 3-lineage differentiation of sorted Lin^–ITM2A^– and Lin^–ITM2A⁺ human periosteal cells. Representative images of Alcian blue staining, Oil Red O staining, and alizarin red staining are shown. Original magnification, ×20. (E and F) A CFU-F assay of Lin^–ITM2A^– and Lin^–ITM2A⁺ cells from human fibular periosteum was performed. (E) Representative CFU-F image and (F) the number of CFU-U per well. n = 5 independent experiments per condition. Data are presented as the mean ± SD. ***P < 0.001, by unpaired t test. (G and H) Kidney capsule transplantation assay of Lin^–ITM2A^– and Lin^–ITM2A⁺ cells from human fibular periosteum. (Representative μ-CT images (G) and bone volume quantification (H). n = 4 independent experiment per condition. Data are presented as the mean ± SD. ***P < 0.001, by unpaired t test.