Secreted frizzled related-protein 2 (Sfrp2) deficiency decreases adult skeletal stem cell function in mice

Abstract

In a previous transcriptomic study of human bone marrow stromal cells (BMSCs, also known as bone marrow-derived "mesenchymal stem cells"), SFRP2 was highly over-represented in a subset of multipotent BMSCs (skeletal stem cells, SSCs), which recreate a bone/marrow organ in an in vivo ectopic bone formation assay. SFRPs modulate WNT signaling, which is essential to maintain skeletal homeostasis, but the specific role of SFRP2 in BMSCs/SSCs is unclear. Here, we evaluated Sfrp2 deficiency on BMSC/SSC function in models of skeletal organogenesis and regeneration. The skeleton of Sfrp2-deficient (KO) mice is overtly normal; but their BMSCs/SSCs exhibit reduced colony-forming efficiency, reflecting low SSC self-renewal/abundancy. Sfrp2 KO BMSCs/SSCs formed less trabecular bone than those from WT littermates in the ectopic bone formation assay. Moreover, regeneration of a cortical drilled hole defect was dramatically impaired in Sfrp2 KO mice. Sfrp2-deficient BMSCs/SSCs exhibited poor in vitro osteogenic differentiation as measured by Runx2 and Osterix expression and calcium accumulation. Interestingly, activation of the Wnt co-receptor, Lrp6, and expression of Wnt target genes, Axin2, C-myc and Cyclin D1, were reduced in Sfrp2-deficient BMSCs/SSCs. Addition of recombinant Sfrp2 restored most of these activities, suggesting that Sfrp2 acts as a Wnt agonist. We demonstrate that Sfrp2 plays a role in self-renewal of SSCs and in the recruitment and differentiation of adult SSCs during bone healing. SFRP2 is also a useful marker of BMSC/SSC multipotency, and a factor to potentially improve the quality of ex vivo expanded BMSC/SSC products.

Fig. 1

Analysis of scRNA-seq data from previously published datasets that assessed non-hematopoietic cell lineages of mouse bone marrow in steady state conditions. a tSNE plot of cre-induced osteoblasts (Col2.3⁺; red), mesenchymal stromal cells (Lepr⁺; green), and vascular cells (VE-Cad+; blue) and (b) associated clusters. c, d Sfrp2-expressing cells were found to be significantly enriched in cluster 2, a cluster containing the Lepr⁺ skeletal stem/progenitors (termed “mesenchymal stromal” cell population by the authors) that also exhibited the greatest number of cells with a high magnitude of Lepr expression

Fig. 2

Sfrp2 KO mice do not exhibit radiological or histological skeletal differences compared with WT mice. a Representative X-ray images of 2-month-old male mice femora and full skeletons. b, c Micrographs (top row) of mid-sagittal H&E-stained sections of the knee joint, with femora on the right and tibiae on the left, and high magnification micrographs of the sub-metaphyseal area of femora (bottom row), of male mice at 3 months (b), and 11 months (c)

Fig. 3

Sfrp2-deficient BMSCs/SSCs have lower colony-forming efficiency and reduced osteogenic differentiation in ectopic bone transplants. a Colony-forming efficiency assay showing colony-forming units-fibroblasts (CFU-F) per 100 000 nucleated bone marrow cells from WT and Sfrp2 KO littermates. b Gelatin sponges were loaded with 2 × 10⁶ BMSCs/SSCs depleted of hematopoietic cells and transplanted into subcutaneous pockets in immunocompromised mice for 8 weeks. c H&E-stained sections of the transplants. WT BMSCs/SSCs formed a complete bone/marrow organ, with cortical and trabecular bone, stroma that supports hematopoiesis of mouse origin (black asterisk). Sfrp2 KO BMSCs/SSCs were less osteogenic; transplants showed residual gelatin sponge scaffold material (black arrowheads, far right panel). d µCT reconstructions of transplants and their trabecular compartments. e µCT analysis of trabecular bone. Compared with WT transplants, Sfrp2 KO transplants had fewer, thinner trabeculae that were more separated and had abnormal geometry (SMI). (Tb.BV/TV)/% Trabecular bone volume/Total volume (%), Tb.Th Trabecular Thickness, Tb.Sp Trabecular Separation, Tb.N Trabecular Number, SMI Structure Model Index. Individual values are shown as five white (WT) or six black circles (Sfrp2 KO), and blue bars represent the mean. *P < 0.05; **P < 0.01

Fig. 4

Sfrp2 KO mice exhibit a bone regeneration defect in vivo. a Drilled bone defect assay. b Mid-transverse sections of the bone defect stained with H&E. Periosteal regeneration was evident in WT mice, while Sfrp2 KO mice exhibited little periosteal response (black arrows). Tb trabecular bone, Cb cortical bone. c µCT reconstructions showing in gray the drilled volume evaluated. d Compared with WT, the defects in Sfrp2 KO mice had less bone, fewer trabeculae with less connectivity that were more widely spaced. (Tb.BV/TV)/% Trabecular bone volume/Total volume (%), Tb.N Trabecular number, Tb.Conn.D Trabecular connectivity density, Tb.Sp Trabecular separation. Individual values are shown as six white (WT) or five black dots (Sfrp2 KO), and blue bars represent the mean. *P < 0.05 vs. WT

Fig. 5

Sfrp2 deficiency leads to a reduced osteogenic differentiation capacity of BMSCs/SSCs in vitro. a After 3 days of osteogenic differentiation with dexamethasone, β-glycerol phosphate and L-ascorbic acid 2-phosphate (D/P/A), Sfrp2 KO BMSCs/SSCs showed less expression of Runx2 and Osterix (vs. Gapdh) than WT BMSCs/SSCs. Addition of exogenous rm-Sfrp2 (1 μg·mL⁻¹) normalized Runx2 expression, and increased Osterix expression above WT levels. b After 21 days of osteogenic differentiation (D/P/A), Sfrp2 KO BMSC/SSC cultures were less mineralized than WT cultures, as measured by alizarin red S staining and absorptiometry at 405 nm. *P < 0.05; **P < 0.01; ***P < 0.01 vs. WT BMSCs/SSCs; ^#P < 0.05; ^##P < 0.01; ^###P < 0.01 vs. Sfrp2 KO BMSCs/SSCs^. c–f DM-5, an immortalized BMSC/SSC line that retains the ability to form an ectopic ossicle upon transplantation, was transfected with scrambled (Con-siRNA) or Sfrp2-siRNA. After 3 days of osteogenic differentiation with either D/P/A in (c), or BMP-2, β-glycerol phosphate and L-ascorbic acid 2-phosphate (B/P/A) in (e), cells silenced for Sfrp2 expression showed decreased expression of Runx2 and Osterix (vs. Gapdh). After 7 days of osteogenic differentiation with D/P/A or B/P/A, cells silenced for Sfrp2 expression showed decreased calcium accumulation as shown in alizarin red S stains in (d, f). Data represent mean ± SD. *P < 0.05; **P < 0.01 vs. Con-siRNA-treated cells

Fig. 6

Sfrp2 deficiency reduces Wnt signaling in murine BMSC/SSCs. a, b BMSCs/SSCs from WT and Sfrp2 KO mice were cultured, depleted of hematopoietic cells and treated with 1 μg·mL⁻¹ rm-Sfrp2 or vehicle. C,D DM5 cells were transfected with scrambled (Con-siRNA) or Sfrp2-siRNA and treated with 1 μg·mL⁻¹ rm-Sfrp2 or vehicle. a, c Lrp6-Ser1490 phosphorylation, as measured by western blot densitometry versus HSP90, was decreased in Sfrp2-deficient cultures and trended to normalize 5 h after the addition of rm-Sfrp2. It is possible that the reported levels of pLrp6 (Ser1490) also include pLrp5 (ser1493), as per the manufacturer’s description. The antibody used is expected to also detect this highly similar epitope. b, d Expression of Wnt target genes Axin2, C-myc and Cyclin D1 (vs. Gapdh) was decreased in Sfrp2-deficient cultures and partially normalized 24 h after the addition of rm-Sfrp2. Data are presented as the mean of three or more independent experiments ± SD. ^aP < 0.05 vs. WT BMSCs/SSCs, ^bP < 0.05 vs. Sfrp2 KO BMSCs/SSCs. *P < 0.05; **P < 0.01 vs. Con-siRNA treated cells; ^#P < 0.05, ^##P < 0.01 vs. Sfrp2-siRNA treated cells