Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Lipid availability determines fate of skeletal progenitor cells via SOX9

Abstract

The avascular nature of cartilage makes it a unique tissue1,2,3,4, but whether and how the absence of nutrient supply regulates chondrogenesis remain unknown. Here we show that obstruction of vascular invasion during bone healing favours chondrogenic over osteogenic differentiation of skeletal progenitor cells. Unexpectedly, this process is driven by a decreased availability of extracellular lipids. When lipids are scarce, skeletal progenitors activate forkhead box O (FOXO) transcription factors, which bind to the Sox9 promoter and increase its expression. Besides initiating chondrogenesis, SOX9 acts as a regulator of cellular metabolism by suppressing oxidation of fatty acids, and thus adapts the cells to an avascular life. Our results define lipid scarcity as an important determinant of chondrogenic commitment, reveal a role for FOXO transcription factors during lipid starvation, and identify SOX9 as a critical metabolic mediator. These data highlight the importance of the nutritional microenvironment in the specification of skeletal cell fate.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Fig. 1: Preventing vascular ingrowth during bone healing induces chondrogenesis.
Fig. 2: Lipid scarcity induces SOX9 in skeletal progenitors.
Fig. 3: SOX9 suppresses FAO in chondrocytes.
Fig. 4: Lipids regulate SOX9 through FOXO signalling.

Data availability

The bulk RNA-seq data that support the findings of this study have been deposited in ArrayExpress with the accession number E-MTAB-7564. The single-cell RNA-seq data were generated previously21 and are deposited in the Gene Expression Omnibus with accession number GSE128423. A portal for exploring the entire atlas is available at https://portals.broadinstitute.org/single_cell/study/mouse-bone-marrow-stroma-in-homeostasis. Source Data for Figs. 14 and Extended Data Figs. 18 are provided with the paper. All other data supporting the findings of this study are available within the paper.

Code availability

The full code used for the computational model of bone-graft healing is available from the authors upon request. More background information on the development of the model can be found in our previous publications10,11.

References

  1. 1.

    Kronenberg, H. M. Developmental regulation of the growth plate. Nature 423, 332–336 (2003).

    ADS  CAS  PubMed  Google Scholar 

  2. 2.

    Roberts, S. J., van Gastel, N., Carmeliet, G. & Luyten, F. P. Uncovering the periosteum for skeletal regeneration: the stem cell that lies beneath. Bone 70, 10–18 (2015).

    PubMed  Google Scholar 

  3. 3.

    Hallmann, R., Feinberg, R. N., Latker, C. H., Sasse, J. & Risau, W. Regression of blood vessels precedes cartilage differentiation during chick limb development. Differentiation 34, 98–105 (1987).

    CAS  PubMed  Google Scholar 

  4. 4.

    Yin, M. & Pacifici, M. Vascular regression is required for mesenchymal condensation and chondrogenesis in the developing limb. Dev. Dyn. 222, 522–533 (2001).

    CAS  PubMed  Google Scholar 

  5. 5.

    Maes, C. et al. Placental growth factor mediates mesenchymal cell development, cartilage turnover, and bone remodeling during fracture repair. J. Clin. Invest. 116, 1230–1242 (2006).

    CAS  PubMed  PubMed Central  Google Scholar 

  6. 6.

    Taylor, D. K. et al. Thrombospondin-2 influences the proportion of cartilage and bone during fracture healing. J. Bone Miner. Res. 24, 1043–1054 (2009).

    PubMed  PubMed Central  Google Scholar 

  7. 7.

    Miclau, K. R. et al. Stimulating fracture healing in ischemic environments: does oxygen direct stem cell fate during fracture healing? Front. Cell Dev. Biol. 5, 45 (2017).

    Google Scholar 

  8. 8.

    Tiyapatanaputi, P. et al. A novel murine segmental femoral graft model. J. Orthop. Res. 22, 1254–1260 (2004).

    PubMed  Google Scholar 

  9. 9.

    Stegen, S., van Gastel, N. & Carmeliet, G. Bringing new life to damaged bone: the importance of angiogenesis in bone repair and regeneration. Bone 70, 19–27 (2015).

    CAS  PubMed  Google Scholar 

  10. 10.

    Carlier, A. et al. MOSAIC: a multiscale model of osteogenesis and sprouting angiogenesis with lateral inhibition of endothelial cells. PLOS Comput. Biol. 8, e1002724 (2012).

    MathSciNet  CAS  PubMed  PubMed Central  Google Scholar 

  11. 11.

    Carlier, A., Geris, L., van Gastel, N., Carmeliet, G. & Van Oosterwyck, H. Oxygen as a critical determinant of bone fracture healing-a multiscale model. J. Theor. Biol. 365, 247–264 (2015).

    MathSciNet  CAS  PubMed  MATH  Google Scholar 

  12. 12.

    Zhao, L., Li, G., Chan, K. M., Wang, Y. & Tang, P. F. Comparison of multipotent differentiation potentials of murine primary bone marrow stromal cells and mesenchymal stem cell line C3H10T1/2. Calcif. Tissue Int. 84, 56–64 (2009).

    CAS  PubMed  Google Scholar 

  13. 13.

    van Gastel, N. et al. Engineering vascularized bone: osteogenic and proangiogenic potential of murine periosteal cells. Stem Cells 30, 2460–2471 (2012).

    PubMed  Google Scholar 

  14. 14.

    Debnath, S. et al. Discovery of a periosteal stem cell mediating intramembranous bone formation. Nature 562, 133–139 (2018).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  15. 15.

    Duchamp de Lageneste, O. et al. Periosteum contains skeletal stem cells with high bone regenerative potential controlled by Periostin. Nat. Commun. 9, 773 (2018).

    ADS  PubMed  PubMed Central  Google Scholar 

  16. 16.

    Chan, C. K. et al. Identification and specification of the mouse skeletal stem cell. Cell 160, 285–298 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  17. 17.

    Amarilio, R. et al. HIF1α regulation of Sox9 is necessary to maintain differentiation of hypoxic prechondrogenic cells during early skeletogenesis. Development 134, 3917–3928 (2007).

    CAS  PubMed  Google Scholar 

  18. 18.

    Robins, J. C. et al. Hypoxia induces chondrocyte-specific gene expression in mesenchymal cells in association with transcriptional activation of Sox9. Bone 37, 313–322 (2005).

    CAS  PubMed  Google Scholar 

  19. 19.

    Shapiro, I. M. & Srinivas, V. Metabolic consideration of epiphyseal growth: survival responses in a taxing environment. Bone 40, 561–567 (2007).

    PubMed  Google Scholar 

  20. 20.

    Stegen, S. et al. HIF-1α metabolically controls collagen synthesis and modification in chondrocytes. Nature 565, 511–515 (2019).

    ADS  CAS  PubMed  Google Scholar 

  21. 21.

    Baryawno, N. et al. A cellular taxonomy of the bone marrow stroma in homeostasis and leukemia. Cell 177, 1915–1932.e16 (2019).

    CAS  PubMed  Google Scholar 

  22. 22.

    Kim, C. et al. Studying arrhythmogenic right ventricular dysplasia with patient-specific iPSCs. Nature 494, 105–110 (2013).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  23. 23.

    Singh, R. et al. Autophagy regulates lipid metabolism. Nature 458, 1131–1135 (2009).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  24. 24.

    Rambold, A. S., Cohen, S. & Lippincott-Schwartz, J. Fatty acid trafficking in starved cells: regulation by lipid droplet lipolysis, autophagy, and mitochondrial fusion dynamics. Dev. Cell 32, 678–692 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  25. 25.

    Tsukamoto, S. et al. Autophagy is essential for preimplantation development of mouse embryos. Science 321, 117–120 (2008).

    ADS  CAS  PubMed  Google Scholar 

  26. 26.

    Imrichová, H., Hulselmans, G., Atak, Z. K., Potier, D. & Aerts, S. i-cisTarget 2015 update: generalized cis-regulatory enrichment analysis in human, mouse and fly. Nucleic Acids Res. 43 (W1), W57–W64 (2015).

    PubMed  PubMed Central  Google Scholar 

  27. 27.

    Shang, J., Liu, H., Li, J. & Zhou, Y. Roles of hypoxia during the chondrogenic differentiation of mesenchymal stem cells. Curr. Stem Cell Res. Ther. 9, 141–147 (2014).

    CAS  PubMed  Google Scholar 

  28. 28.

    Frey, J. L. et al. Wnt–Lrp5 signaling regulates fatty acid metabolism in the osteoblast. Mol. Cell. Biol. 35, 1979–1991 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  29. 29.

    Kim, S. P. et al. Fatty acid oxidation by the osteoblast is required for normal bone acquisition in a sex- and diet-dependent manner. JCI Insight 2, 92704 (2017).

    PubMed  Google Scholar 

  30. 30.

    Eijkelenboom, A. & Burgering, B. M. FOXOs: signalling integrators for homeostasis maintenance. Nat. Rev. Mol. Cell Biol. 14, 83–97 (2013).

    CAS  PubMed  Google Scholar 

  31. 31.

    Ashraf, S. & Walsh, D. A. Angiogenesis in osteoarthritis. Curr. Opin. Rheumatol. 20, 573–580 (2008).

    PubMed  Google Scholar 

  32. 32.

    Ratneswaran, A. et al. Peroxisome proliferator-activated receptor δ promotes the progression of posttraumatic osteoarthritis in a mouse model. Arthritis Rheumatol. 67, 454–464 (2015).

    CAS  PubMed  Google Scholar 

  33. 33.

    Zhong, L., Huang, X., Karperien, M. & Post, J. N. Correlation between gene expression and osteoarthritis progression in human. Int. J. Mol. Sci. 17, E1126 (2016).

    PubMed  Google Scholar 

  34. 34.

    Akasaki, Y. et al. Dysregulated FOXO transcription factors in articular cartilage in aging and osteoarthritis. Osteoarthritis Cartilage 22, 162–170 (2014).

    CAS  PubMed  Google Scholar 

  35. 35.

    Hadjantonakis, A. K., Gertsenstein, M., Ikawa, M., Okabe, M. & Nagy, A. Generating green fluorescent mice by germline transmission of green fluorescent ES cells. Mech. Dev. 76, 79–90 (1998).

    CAS  PubMed  Google Scholar 

  36. 36.

    Ouyang, Z. et al. Prx1 and 3.2kb Col1a1 promoters target distinct bone cell populations in transgenic mice. Bone 58, 136–145 (2014).

    CAS  PubMed  Google Scholar 

  37. 37.

    Nuyts, J. et al. Iterative reconstruction for helical CT: a simulation study. Phys. Med. Biol. 43, 729–737 (1998).

    CAS  PubMed  Google Scholar 

  38. 38.

    Depypere, M. et al. An iterative dual energy CT reconstruction method for a K-edge contrast material. Proc. SPIE 7961, 79610M (2011).

    Google Scholar 

  39. 39.

    Vandersmissen, I. et al. Endothelial Msx1 transduces hemodynamic changes into an arteriogenic remodeling response. J. Cell Biol. 210, 1239–1256 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  40. 40.

    Maes, C. et al. Soluble VEGF isoforms are essential for establishing epiphyseal vascularization and regulating chondrocyte development and survival. J. Clin. Invest. 113, 188–199 (2004).

    CAS  PubMed  PubMed Central  Google Scholar 

  41. 41.

    Stiers, P. J., van Gastel, N., Moermans, K., Stockmans, I. & Carmeliet, G. An ectopic imaging window for intravital imaging of engineered bone tissue. JBMR Plus 2, 92–102 (2018).

    PubMed  PubMed Central  Google Scholar 

  42. 42.

    Stegen, S. et al. Osteocytic oxygen sensing controls bone mass through epigenetic regulation of sclerostin. Nat. Commun. 9, 2557 (2018).

    ADS  PubMed  PubMed Central  Google Scholar 

  43. 43.

    Stiers, P. J. et al. Inhibition of the oxygen sensor PHD2 enhances tissue-engineered endochondral bone formation. J. Bone Miner. Res. 34, 333–348 (2019).

    CAS  PubMed  Google Scholar 

  44. 44.

    McQuin, C. et al. CellProfiler 3.0: Next-generation image processing for biology. PLoS Biol. 16, e2005970 (2018).

    PubMed  PubMed Central  Google Scholar 

  45. 45.

    Daniëls, V. W. et al. Cancer cells differentially activate and thrive on de novo lipid synthesis pathways in a low-lipid environment. PLoS One 9, e106913 (2014).

    ADS  PubMed  PubMed Central  Google Scholar 

  46. 46.

    Eyckmans, J., Lin, G. L. & Chen, C. S. Adhesive and mechanical regulation of mesenchymal stem cell differentiation in human bone marrow and periosteum-derived progenitor cells. Biol. Open 1, 1058–1068 (2012).

    CAS  PubMed  PubMed Central  Google Scholar 

  47. 47.

    Ranall, M. V., Gabrielli, B. G. & Gonda, T. J. High-content imaging of neutral lipid droplets with 1,6-diphenylhexatriene. Biotechniques 51, 35–42 (2011).

    CAS  PubMed  Google Scholar 

  48. 48.

    Kimura, S., Noda, T. & Yoshimori, T. Dissection of the autophagosome maturation process by a novel reporter protein, tandem fluorescent-tagged LC3. Autophagy 3, 452–460 (2007).

    CAS  PubMed  Google Scholar 

  49. 49.

    Schoors, S. et al. Fatty acid carbon is essential for dNTP synthesis in endothelial cells. Nature 520, 192–197 (2015).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  50. 50.

    Lambrechts, D. et al. A causal relation between bioluminescence and oxygen to quantify the cell niche. PLoS ONE 9, e97572 (2014).

    ADS  PubMed  PubMed Central  Google Scholar 

  51. 51.

    Guo, W. et al. Slug and Sox9 cooperatively determine the mammary stem cell state. Cell 148, 1015–1028 (2012).

    CAS  PubMed  PubMed Central  Google Scholar 

  52. 52.

    Ran, F. A. et al. Genome engineering using the CRISPR–Cas9 system. Nat. Protoc. 8, 2281–2308 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  53. 53.

    Aubrey, B. J. et al. An inducible lentiviral guide RNA platform enables the identification of tumor-essential genes and tumor-promoting mutations in vivo. Cell Rep. 10, 1422–1432 (2015).

    CAS  PubMed  Google Scholar 

  54. 54.

    Trapnell, C., Pachter, L. & Salzberg, S. L. TopHat: discovering splice junctions with RNA-seq. Bioinformatics 25, 1105–1111 (2009).

    CAS  PubMed  PubMed Central  Google Scholar 

  55. 55.

    Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140 (2010).

    CAS  Google Scholar 

  56. 56.

    Stegen, S. et al. HIF-1α promotes glutamine-mediated redox homeostasis and glycogen-dependent bioenergetics to support postimplantation bone cell survival. Cell Metab. 23, 265–279 (2016).

    CAS  PubMed  Google Scholar 

Download references

Acknowledgements

We thank K. Moermans, I. Stockmans, C. MacGillivray and H. Soled for technical assistance, A. Nagy for the CAG-eGFP mice, S. Murakami for the Col1a1-cre/ERT2 and DsRed mice, T. Yoshimori for the RFP-GFP-LC3 plasmid, R. A. Weinberg for the pLKO.1-sh-mSOX9-5 lentiviral plasmid, M. Mazzone for hypoxic glove box use, the histology core of the Harvard Department of Stem Cell and Regenerative Biology for histology services, the FACS cores of the KU Leuven and the Harvard Department of Stem Cell and Regenerative Biology for access to the flow cytometers, and the Cell Imaging Core and the Molecular Imaging and Photonics division of the KU Leuven and the Harvard Center for Biological Imaging for access to the confocal microscopes. This work was supported by grants from the Research Fund Flanders (FWO; G096414, G0A4216N and G0B3418N to G.C.), KUL grant C24/17/07 (G.C.), grants from the European Research Council (ERC 308223 to H.V.O., ERC 279100 to L.G. and ERC 269073 to P.C.), and long-term structural Methusalem funding by the Flemish Government (P.C.). N.v.G. is funded by BOF-KU Leuven GOA project 3M120209. P.J.S. is a fellow from the Agency for Innovation by Science and Technology in Flanders (IWT). S. Stegen, A.C. and Dennis L. are postdoctoral fellows of the FWO. V.W.D. is a fellow of the FWO and the Flemish League against Cancer (VLK). This work is part of Prometheus, the KU Leuven R&D Division of Skeletal Tissue Engineering.

Author information

Affiliations

Authors

Contributions

N.v.G. and G.C. conceived the study. N.v.G., S. Stegen, G.E., S. Schoors, P.-J.S., Dennis L., S.T. and A.S. performed the in vitro experiments. N.v.G. performed the in vivo experiments. A.C. performed the in silico experiments. V.W.D. and J.V.S. contributed to the design and execution (V.W.D.) of lipid-rescue experiments. N.B. and D.P. performed and analysed the single-cell RNA-seq experiment. M.D. and F.M. contributed to the design and execution (M.D.) of microCT analyses. R.V.L. and A.S. performed histology. P.A. contributed to the design and interpretation of autophagy experiments. N.v.G., A.C., L.G. and H.V.O. contributed to the design and interpretation of in silico experiments. Diether L. and B.T. contributed to the design, execution and interpretation of RNA-seq experiments. P.C. contributed to the design and interpretation of metabolic analyses. D.T.S. contributed to the design and interpretation of in vivo experiments. P.A., J.V.S., P.C. and D.T.S. provided reagents. N.v.G., S. Stegen and G.C. designed the experiments and interpreted data. N.v.G. and G.C. wrote the manuscript. All authors agreed on the final version of the manuscript.

Corresponding author

Correspondence to Geert Carmeliet.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature thanks Thomas Clemens, Michael T. Longaker and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Removal of periosteum reduces bone formation and callus vascularization.

a, Histological characterization of the mouse bone-autograft healing model. At the host–graft junction cartilage (safranin O+) is formed at PFD7. Note absence of CD31+ blood vessels in these regions. Near the graft centre new woven bone (bright pink on H&E staining) is deposited, cartilage is absent and blood vessels are abundant. By PFD14, the cartilage at the host–graft junction is gradually being replaced by bone, while the woven bone near the graft centre appears mature (representative images of four mice). Scale bars, 200 μm in host–graft junction images, 100 μm in graft-centre images, 50 μm in magnifications. b, MicroCT-based visualization and quantification of newly formed bone around control autografts, autografts from which the periosteum was removed or devitalized allografts (no living cells) at PFD28 (n = 3 mice). Coverage ratio represents percentage of graft surface covered by new bone. c, Dual-energy microCT-based visualization and quantification of vascularization in a 250-μm-wide region around autografts and allografts at PFD14 (n = 5 mice for autograft, n = 6 mice for devitalized allograft). d, CD31 immunohistochemical visualization and quantification of vascularization in a 250 μm-wide region around autografts and allografts at PFD14 (n = 3 mice). Scale bars, 500 μm. b, bone; c, cartilage; ft, fibrous tissue; g, graft; h, host; m, muscle; p, periosteum. Data are mean ± s.e.m.; one-way ANOVA with Bonferroni post hoc test (b), two-tailed Student’s t-test (c, d).

Source data

Extended Data Fig. 2 Reducing vascularization alters but does not prevent bone healing.

a, Histological visualization and quantification of apoptotic cells (TUNEL+; n = 4 mice for control, n = 5 mice for filter 0.2) in the callus of grafts with or without a filter (0.2 μm pore size) at PFD7. Scale bars, 50 μm. b, Histological visualization and quantification of proliferating (BrdU+; n = 3 mice) cells in the callus of grafts with or without a filter (0.2 μm pore size) at PFD7. Scale bars, 100 μm. c, MicroCT-based visualization and quantification of newly formed bone around control grafts or grafts surrounded by a filter (0.2 μm pore size) at PFD14 (n = 4 mice for control, n = 6 mice for filter 0.2). Coverage ratio represents percentage of graft surface covered by new bone. d, Cell tracing of donor periosteal cells during healing of bone grafts, derived from CAG-eGFP mice, with or without filter (0.2 μm pore size) at PFD14 showing equal contribution of donor cells to cartilage in both conditions, but reduced contribution of donor cells to bone near the graft ends. Arrows, GFP+ osteoblasts; arrowheads, GFP+ osteocytes; representative images of three mice. Scale bars, 50 μm. e, MicroCT-based visualization and quantification of newly formed bone around control grafts or grafts surrounded by a filter (0.2 μm pore size) at PFD28 (n = 3 mice). f, Histological analysis of autografts with or without a filter (0.2 μm pore size) at PFD28 showing comparable callus morphology and composition, although remaining cartilage islands (detail image) were seen when a filter was present but not in the callus of control grafts (representative images of three mice). Scale bars, 500 μm. Data are mean ± s.e.m.; two-tailed Student’s t-test.

Source data

Extended Data Fig. 3 In silico modelling supports a role for nutritional stress in chondrogenic commitment.

Application of a previously described computational model of bone repair10,11 to the bone-graft healing setup. In this model, the behaviour (survival, proliferation, differentiation and tissue formation) of skeletal progenitor cells, chondrocytes, osteoblasts and fibroblasts is dependent on the local supply of nutrients by blood vessels, in addition to the presence of growth factors, extracellular matrix and the cell density. a, Schematic overview (top) of the modelled region shown in green. The hatched area represents the graft callus. At the start of the simulation the modelled region was filled with loose fibrous tissue matrix, growth factors, stem cells, osteoblasts, fibroblasts and nutrients, representing the fracture haematoma. Overview of the Dirichlet boundary conditions (bottom) showing the starting points of blood vessels and the sites of release of cells and growth factors (and nutrients for the condition with filter) during the healing process. b, Application of the model to the normal bone graft (that is, blood vessels can come from the muscle side). Heat map-based visualization of blood vessel, nutrient, cartilage and bone distribution in the modelled region at different time points shows that the model correctly predicts the spatiotemporal progression of the bone-healing process. Nutrients and tissue fractions are expressed on a non-dimensional scale ranging from 0 (absence) to 1 (saturation). c, d, Application of the model to bone-graft healing in the presence of a filter placed in between graft and muscle (that is, blood vessels cannot come from the muscle side) with visual representation (c) and quantification (d) of the different tissue fractions in the modelled region. Quantification was performed only in the left rectangle of the modelled region, as indicated by the hatched area in a, representing the graft callus. The amount of nutrients that can pass through the filter (the boundary condition (BC)) was varied between 100% (the maximum amount that can be supplied by the vasculature, applied to the whole filter length, resulting in similar nutrient distributions as in the control) and 0%. When nutrient supply through the filter is set at 20–40%, the model correctly recapitulates the chondrogenic switch in the central region of the graft as observed in vivo. When nutrient supply through the filter was >40%, the cells in the central graft region differentiated directly into osteoblasts, and a supply of nutrients <20% induced massive cell death and completely prevented tissue formation and graft healing. e, Visual representation of the effect of additional growth factor (gf) diffusion and/or progenitor cell (prog) migration from the filter side on cartilage and bone fractions at day 14. The control situation (no filter) is shown on the top and the filter situation with a boundary condition for nutrients of 40% is shown on the bottom. No large effect of these additional boundary conditions on the healing response was observed.

Source data

Extended Data Fig. 4 Skeletal progenitors resist nutritional stress via induction of SOX9.

a, Immunoblot detection of nuclear SOX9 in C3H10T1/2 cells and periosteal cells exposed for 24 h to control or CND medium, with lamin A/C as loading control (n = 2 independent experiments). b, mRNA levels of Sox9 and Col2a1 in periosteal cells exposed for the indicated times to control or CND medium (relative to control; n = 3 biologically independent samples). c, mRNA levels of runt-related transcription factor 2 (Runx2; osteogenic lineage), peroxisome proliferator-activated receptor γ (Pparg; adipogenic lineage) and Myod (myogenic lineage) in periosteal cells exposed for 48 h to control or CND medium (relative to control; n = 3 biologically independent samples). d, mRNA levels of Sox9 in C3H10T1/2 cells exposed for the indicated times to control or SD medium (relative to control, n = 3 independent experiments). e, Immunoblot detection of total SOX9 in C3H10T1/2 cells exposed for different durations to control or SD medium, with β-actin as loading control (n = 2 independent experiments). f, Immunoblot detection of nuclear and cytoplasmic SOX9 in C3H10T1/2 cells exposed for 6 h to control or SD medium, with lamin A/C or β-actin as loading control (n = 2 independent experiments). g, Immunoblot detection of SOX9 in total cell protein extracts of C3H10T1/2 cells exposed for 6 h to control medium, SD medium or SD medium supplemented with different concentrations of the transcription inhibitor actinomycin D (Act. D) or the translation inhibitor cycloheximide (CHX). Detection of β-actin was used as loading control (n = 2 independent experiments). h, mRNA levels of Runx2, Pparg and Myod in C3H10T1/2 cells exposed for the indicated times to control or SD medium (relative to control, n = 3 independent experiments). i, Immunoblot detection of nuclear SOX9 in periosteal cells exposed for 24 h to control or SD medium with lamin A/C as loading control (n = 3 biologically independent samples). j, Osteogenic differentiation of periosteal cells in control or SD medium, assessed by visualization of mineral deposits (alizarin red staining) and quantification of Ocn mRNA levels (relative to Actb, n = 3 biologically independent samples). k, Immunoblot detection of SOX9 in total cell protein extracts of C3H10T1/2 cells (in control or SD medium), periosteal cells and growth plate-derived chondrocytes transduced with shSox9 or shScr, with β-actin as loading control. A longer exposure time was used for SOX9 detection in C3H10T1/2 cells and periosteal cells compared with chondrocytes in order to visualize any remaining protein in the shSOX9 conditions (n = 2 independent experiments for C3H10T1/2 cells, n = 3 biologically independent samples for periosteal cells, growth plate-derived chondrocytes). l, Quantification of cell viability of C3H10T1/2 cells, periosteal cells and growth plate-derived chondrocytes transduced with shSox9 or shSCR, after 72 h of exposure to control, SD or CND medium (n = 3 independent experiments for C3H10T1/2 cells, n = 3 biologically independent samples for periosteal cells, growth plate-derived chondrocytes). Data are mean ± s.e.m.; two-way ANOVA with Bonferroni post hoc test (b, d, h, l), two-tailed Student’s t-test (c, j). For gel source data, see Supplementary Fig. 1.

Source data

Extended Data Fig. 5 Reduced lipid availability favours chondrogenesis over osteogenesis.

ac, Immunoblot detection of total SOX9 in C3H10T1/2 cells exposed for 6 h to control medium, SD medium or SD medium supplemented with increasing concentrations of palmitate (a), VLDL (b) or PUFA (c). Detection of β-actin was used as loading control. EtOH was used as a vehicle control in a and c (n = 2 independent experiments). d, Histological visualization (by immunofluorescence for COL2) of chondrogenic differentiation of periosteal cells in pellet cultures in control, SD or LRS medium supplemented with vehicle (EtOH), oleate or PUFA (representative images of n = 2 independent experiments). Scale bars, 100 μm. e, Osteogenic differentiation of periosteal cells in control, SD or LRS medium, assessed by visualization of mineral deposits (alizarin red staining) and quantification of Ocn mRNA levels (relative to Actb, n = 3 biologically independent samples). f, Flow cytometric detection and quantification of the percentage of SOX9high cells and total SOX9 levels in C3H10T1/2 cells, periosteal cells and skeletal stem cells exposed for 24 h to control, SD or LRS medium (n = 4 independent experiments for C3H10T1/2 cells, n = 4 biologically independent samples for periosteal cells, skeletal stem cells). Gating for SOX9high cells was set to have approximately 10% SOX9high cells in control conditions in each cell type. g, h, Flow cytometric quantification of cell cycle (g) and apoptosis (h) in SOX9low and SOX9high subpopulations of C3H10T1/2 cells, periosteal cells and skeletal stem cells exposed for 24 h to control, SD or LRS medium (n = 3 independent experiments for C3H10T1/2 cells, n = 3 biologically independent samples for periosteal cells, skeletal stem cells). i, Histological visualization and quantification of early chondrogenic (SOX9+) and osteogenic (Col1a1-DsRed+) cells in metatarsals cultured for one week in control medium, SD medium or SD medium supplemented with PUFA or vehicle (EtOH) (n = 6 biologically independent samples for control, SD and SD + vehicle, n = 7 biologically independent samples for SD + PUFA). Scale bars, 50 μm. j, Histological visualization of mineralization by Von Kossa staining in metatarsals cultured for one week in control medium, SD medium or SD medium supplemented with vehicle or PUFA (representative images of n = 6 biologically independent samples for control, SD and SD + vehicle, n = 7 biologically independent samples for SD + PUFA). Scale bars, 100 μm. k, Histological visualization (safranin O staining) and quantification of cartilage and woven bone in the callus at PFD7 of mice treated daily with free fatty acids (FFA; 20 μl corn oil) or sham injection (saline) at the fracture site (n = 5 mice). Scale bars, 500 μm. l, Flow cytometric quantification of total SOX9 levels in C3H10T1/2 cells or skeletal stem cells exposed for 24 h to control, SD or LRS medium supplemented with 100 μM GW9508 or vehicle (DMSO) (n = 3 independent experiments for C3H10T1/2 cells, n = 3 biologically independent samples for skeletal stem cells). m, Visualization and quantification of diffusion of a fluorescent fatty acid (FL-C16) and fluorescent glucose (2-NBDG) in collagen gels seeded with periosteal cells (5 × 106 per ml) (n = 3 biologically independent samples for FL-C16, n = 5 biologically independent samples for 2-NBDG). Scale bars, 500 μm. n, o, Visualization of alcian blue staining (n) and visualization and quantification of Sox9 expression (o) in micromass co-cultures of periosteal cells from Sox9-GFP mice and sorted cell populations from skeletal muscle of CAG-DsRed mice, after nine days in chondrogenic SD medium (n = 4 biologically independent samples). Addition of oleate was used as positive control. Scale bars, 100 μm. EC, endothelial cell, MΦ, macrophage. Data are mean ± s.e.m.; one-way ANOVA (e, f, i, o), two-way ANOVA (h, l) or three-way ANOVA (g) with Bonferroni post hoc test, two-tailed Student’s t-test (k, m). For gel source data, see Supplementary Fig. 1.

Source data

Extended Data Fig. 6 Chondrocytes do not depend on FAO.

a, Quantification of glycolytic rate, oxygen consumption and palmitate oxidation in periosteal cells (PC, n = 5 biologically independent samples), skeletal stem cells (SSC, n = 3 biologically independent samples), growth plate-derived chondrocytes (GCH, n = 3 biologically independent samples for oxygen consumption, n = 4 biologically independent samples for glycolysis and palmitate oxidation), rib chondrocytes (RCH, n = 5 biologically independent samples for oxygen consumption, n = 4 biologically independent samples for glycolysis and palmitate oxidation), calvarial osteoblasts (COB, n = 5 biologically independent samples) and trabecular osteoblasts (TOB, n = 5 biologically independent samples). b, t-Distributed stochastic neighbour embedding (t-SNE) plot of 20,896 non-haematopoietic cells (mixed bone and bone marrow fractions, n = 6 mice) based on single-cell RNA-seq data, annotated post hoc and coloured by clustering (top) or by expression (ln(TP10K)) of selected genes (bottom). c, Expression (row-wide Z score of ln of average TP10K; single-cell RNA-seq) of FAO- and glycolysis-related genes (rows) in the cells of each cluster (columns). d, RT–qPCR analysis of genes involved in glycolysis (Glut1 (also known as Slc2a1), Pfkfb3 and Ldha; n = 6 independent samples for Glut1 and Pfkfb3 in cartilage, n = 9 independent samples for Glut1 and Pfkfb3 in bone, n = 8 independent samples for Ldha) and FAO (Cpt1a, Acadm and Acadl; n = 8 independent samples) in mouse growth plate cartilage and cortical bone biopsies (relative to Actb). e, Analysis of adjacent histological sections of a growth plate and fracture callus (PFD7) of mice injected intravenously with a fluorescent fatty acid (Red-C12) or glucose (2-NBDG) (representative images of n = 3 mice). Scale bars, 100 μm in growth plate images, 50 μm in fracture callus images. f, Immunofluorescence analysis of a fracture callus (PFD7) of a mouse injected intravenously with a fluorescent fatty acid (Red-C12) and stained for SOX9 (left; cartilage area shown) or COL1 (right; trabecular bone area shown) (representative images of n = 3 mice). Scale bars, 50 μm. g, Histological visualization and quantification at PFD7 of CAG-DsRed+ skeletal stem cells (SSC), transduced with shCpt1a or shScr and transplanted at the fracture site on PFD0 (n = 3 mice). Dotted lines delineate cortical bone ends. h, Quantification of number of live and dead cells in cultures of periosteal cells, growth plate-derived chondrocytes and calvarial osteoblasts after 48 h of exposure to etomoxir (n = 3 biologically independent samples). Data are mean ± s.e.m.; one-way (a) or two-way (h) ANOVA with Bonferroni post hoc test, two-tailed Student’s t-test (d, g).

Source data

Extended Data Fig. 7 Changes in FAO and autophagy after lipid deprivation.

a, Oxidation of extracellularly added palmitate by periosteal cells in control medium or at different times in LRS medium (n = 4 biologically independent samples). b, Quantification of FAO-linked OCR in periosteal cells in control medium or at different times in LRS medium (n = 4 biologically independent samples). c, Confocal microscopy of periosteal cells labelled with Red-C12 (fluorescent fatty acid, red) and stained with MitoTracker (mitochondria, green) and DPH (lipid droplets, blue) shows increased colocalization (as quantified by Pearson’s correlation coefficient) of MitoTracker and Red-C12 after exposure of cells for 6 h to SD medium (n = 4 biologically independent samples). Scale bars, 20 μm. d, Immunoblot detection of LC3 in total cell protein extracts of C3H10T1/2 cells and periosteal cells exposed for different times to control or SD medium, with β-actin as loading control. Note increased conversion of LC3-I to LC3-II at early time points, indicative of activation of autophagy (n = 2 independent experiments). e, f, Confocal microscopy of C3H10T1/2 cells (e; n = 3 independent experiments) or periosteal cells (f; n = 3 biologically independent samples), expressing an RFP–GFP–LC3 tandem construct, shows activation of autophagy with time upon serum deprivation, evidenced by increased total number of LC3 puncta per cell and higher percentage of RFP+GFP puncta. Scale bars, 20 μm. g, Confocal microscopy-based visualization (top) and quantification (bottom) of C3H10T1/2 cells, stained with the neutral lipid dye DPH to reveal lipid-droplet dynamics at different time points after SD. Cells were transduced with shAtg5 to inhibit autophagy or shScr as a control (n = 6 independent experiments). Scale bars, 20 μm. h, Quantification of FAO-linked OCR in periosteal cells in control medium or at different times after serum deprivation, treated with 10 μM chloroquine (CQ) or vehicle (n = 3 biologically independent samples). i, Quantification of cell viability of C3H10T1/2 cells and periosteal cells after 72 h of exposure to control or SD medium in the presence or absence of 50 μM (C3H10T1/2 cells) or 10 μM (periosteal cells) CQ (n = 3 independent experiments for C3H10T1/2 cells, n = 3 biologically independent samples for periosteal cells). j, Immunoblot detection of total SOX9 in C3H10T1/2 cells and nuclear SOX9 in periosteal cells exposed for 6 h (C3H10T1/2 cells) or 24 h (periosteal cells) to control medium (with DMSO as vehicle control) or medium supplemented with 100 μM etomoxir (Eto), with β-actin or lamin A/C as loading control (n = 2 independent experiments for C3H10T1/2 cells, n = 3 biologically independent samples for periosteal cells). k, Cell morphology of growth plate-derived chondrocytes transduced with shSox9 or shScr (representative images of six biologically independent samples). Scale bar, 100 μM. l, RT–qPCR analysis of genes involved in chondrogenesis (Sox9, Col2a1 and Acan) and FAO (Cpt1a, Acadm and Acadl) in growth plate-derived chondrocytes transduced with shSox9 or shScr (relative to shScr, n = 6 biologically independent samples). Data are mean ± s.e.m.; one-way ANOVA (a, b, e, f) or two-way ANOVA (gi) with Bonferroni post hoc test, two-tailed Student’s t-test (c, l). For gel source data, see Supplementary Fig. 1.

Source data

Extended Data Fig. 8 Lipids regulate SOX9 through FOXO signalling.

a, Heat map showing differential expression of cartilage-related genes in C3H10T1/2 cells exposed for different times to SD versus control medium, as determined by RNA-seq (n = 3 replicates). b, Volcano plot showing significantly enriched and depleted mRNAs in C3H10T1/2 cells exposed for 3 or 6 h to SD versus control medium, as determined by RNA-seq (n = 3 replicates). c, Top 10 most significantly enriched transcription factor motifs with normalized enrichment scores (NES) in C3H10T1/2 cells exposed for 3 h (left) or 6 h (right) to SD versus control medium, as determined by i-cisTarget analysis on the 100 most-significantly increased mRNAs (n = 3 replicates). Motif shown on top is the Hmga1 motif for 3 h and the Atf4 motif for 6 h. d, Confocal microscopy of C3H10T1/2 cells stained for FOXO1 after exposure of cells for 3 h to SD or LRS medium in the presence of vehicle (EtOH), oleate (60 μM) or PUFA (representative images of two independent experiments). Scale bars, 20 μm. e, Nuclear FOXO activity in C3H10T1/2 cells exposed for 3 h to control, SD or LRS medium (n = 5 independent experiments). f, Nuclear FOXO activity in skeletal stem cells exposed for 3 h to control medium, LRS medium or LRS medium supplemented with PUFA (n = 3 biologically independent samples). EtOH was used as vehicle control. g, Occupancy of FOXO1 at the Sox9 promoter of Cas9-expressing C3H10T1/2 cells transduced with sgFoxo1, sgFoxo3a or sgScr, exposed for 3 h to control or SD medium, as determined by ChIP–qPCR (n = 3 independent experiments). h, Flow cytometric quantification of total SOX9 levels in C3H10T1/2 cells (n = 4 independent experiments for control and serum deprivation, n = 3 independent experiments for LRS) and skeletal stem cells (n = 3 biologically independent samples) exposed for 24 h to control, SD or LRS medium supplemented with 1 μM AS1842856 or vehicle (DMSO). i, Immunoblot detection of total SOX9 in Cas9-expressing C3H10T1/2 cells transduced with inducible sgFoxo1 and sgFoxo3a (sgFoxo1/3a) or with sgScr, exposed for 6 h to control, SD or LRS medium in the presence or absence of doxycycline (dox; 250 ng ml−1), with β-actin as loading control (n = 2 independent experiments). j, Flow cytometric quantification of total SOX9 levels in skeletal stem cells transduced with shFoxo1 and shFoxo3a (shFoxo1/3a) or with shScr, exposed for 24 h to control, SD or LRS medium (n = 5 biologically independent samples). k, Histological visualization and quantification of FOXO3a-expressing cells in the fracture callus at PFD7 of mice treated daily with GW9508 (10 nmol) or vehicle (0.2% DMSO in saline) at the fracture site (n = 5 mice). Scale bars, 500 μm. Dotted lines delineate cortical bone ends. l, Histological visualization and quantification in the fracture callus at PFD7 of CAG–DsRed+ skeletal stem cells (SSC), transduced with shFoxo1/3a or shScr and transplanted at the fracture site on PFD0 (n = 5 mice). Dotted lines delineate cortical bone ends. Data are mean ± s.e.m.; one-way ANOVA (e, f), two-way ANOVA (g, h, j) with Bonferroni post hoc test, two-tailed Student’s t-test (k, l). For gel source data, see Supplementary Fig. 1.

Source data

Extended Data Fig. 9 Flow cytometry gating for cell sorting.

a, Contour plots showing the gating strategy for the identification and isolation of skeletal stem cells from long bones of newborn mice. b, Contour plots showing the gating strategy for the identification and isolation of macrophages, endothelial cells and pericytes from skeletal muscle of adult mice.

Supplementary information

Supplementary Figure

Uncropped gel images for results obtained by gel-based electrophoretic separation.

Reporting Summary

Source data

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

van Gastel, N., Stegen, S., Eelen, G. et al. Lipid availability determines fate of skeletal progenitor cells via SOX9. Nature 579, 111–117 (2020). https://doi.org/10.1038/s41586-020-2050-1

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links