Osteoclasts recycle via osteomorphs during RANKL-stimulated bone resorption

Abstract

Osteoclasts are large multinucleated bone-resorbing cells formed by the fusion of monocyte/macrophage-derived precursors that are thought to undergo apoptosis once resorption is complete. Here, by intravital imaging, we reveal that RANKL-stimulated osteoclasts have an alternative cell fate in which they fission into daughter cells called osteomorphs. Inhibiting RANKL blocked this cellular recycling and resulted in osteomorph accumulation. Single-cell RNA sequencing showed that osteomorphs are transcriptionally distinct from osteoclasts and macrophages and express a number of non-canonical osteoclast genes that are associated with structural and functional bone phenotypes when deleted in mice. Furthermore, genetic variation in human orthologs of osteomorph genes causes monogenic skeletal disorders and associates with bone mineral density, a polygenetic skeletal trait. Thus, osteoclasts recycle via osteomorphs, a cell type involved in the regulation of bone resorption that may be targeted for the treatment of skeletal diseases.

Keywords: RANKL; cell fission; cellular recycling; denosumab; macrophage; osteoclast; osteomorph; osteoporosis; osteoprotegerin; skeletal dysplasia.

Graphical abstract

Figure 1

Intravital imaging of steady-state osteoclast dynamics (A) Schematic showing the method for tracking osteoclasts by intravital two-photon microscopy in mixed bone marrow chimeras. (B) 3D volume rendered image of the tibia showing the endosteal bone surface (blue, second harmonic generation, SHG) and the network of interconnecting osteoclasts (red pseudocolor). See also Video S1. Scale bars, 600 μm (tiled image, left) and 40 μm (inset image and bottom right). (C) Expression of LYSM (red), CSF1R (green), and uptake of OsteoSense 680 (yellow) by an osteoclast (dotted line) on the bone surface (blue). Scale bar, 50 μm. (D) Hoechst staining (cyan) showing multiple nuclei (yellow arrow heads) in the cell from (C). Scale bar, 50 μm. (E) Cathepsin K activity (magenta, right) in a large stellate osteoclast (red, left) on bone (blue). Scale bar, 50 μm. (F) Staining of in vitro generated osteoclast for Hoechst (cyan, white arrows), F-actin (red), cathepsin K activity (magenta), and OsteoSense 680 (blue). Scale bar, 25 μm. (G) Cell tracking using the surfaces feature in Imaris. Raw image (left panel) show LYSM⁺ (red) and CSF1R⁺ (green) cells. Time-lapse (right panels) show tracked double-labeled osteoclast. Scale bar, 25 μm. Timestamp hh:mm:ss. (H) Tracking morphology using the FilamentTracer plug-in in Imaris of osteoclast from (G). Cell processes shown by yellow lines. (I) Dynamic changes in cell volume (left), sphericity (middle), and total process length (right) of the cell tracked in (G and H). See also Video S2.

Figure 2

Intravital imaging of sRANKL-stimulated osteoclast dynamics and cell fusion (A) Cell tracking post-sRANKL stimulation. Raw images (left and right) show LYSM⁺ (red) and BLIMP1⁺ (green) expression. Timelapse (middle panels) shows tracked double-labeled osteoclasts. Scale bar, 30 μm. Timestamp hh:mm:ss. (B) Surface area, volume, and sphericity tracked in (A). (C) Tracking morphology of osteoclast from (A) using FilamentTracer. Cell processes shown by green lines. Scale bar, 30 μm. (D) Total process length in cells from (C). See also Video S3. (E) Cell tracking post-sRANKL stimulation. Raw images (left and right) show LYSM⁺ (red) and CSF1R⁺ (green) expression. Scale bar, 20 μm. Time-stamp hh:mm:ss. (F) Surface area, volume, and sphericity of cells tracked in (E). See also Video S4. (G) Live-cell imaging of osteoclast fusion post-sRANKL stimulation in vitro. Cells labeled with wheat germ agglutinin-AlexaFluor 488 (red pseudocolor) and Hoechst (cyan). Scale bar, 40 μm. Timestamp hh:mm:ss. See also Video S4. (H) Surface area, volume, and speed of cells tracked in (G).

Figure 3

Intravital imaging of sRANKL-stimulated osteoclast cell fission (A) Cell tracking showing osteoclast fission. Raw images (left and right) show LYSM⁺ (red) and CSF1R⁺ (green) expression. Timelapse (middle panels) shows tracked double-labeled osteoclast. Scale bar, 25 μm. Timestamp hh:mm:ss. See also Video S5. (B) Cell volume from (A). (C) Cell fate mapping from (A) showing fission of parent cell into five daughter cells. (D) Cumulative number of contacts between LYSM⁺CSF1R^neg macrophages and fission products. Dotted lines are fission events. (E) Live-cell imaging of cell fission in vitro. Scale bar, 40 μm. Timestamp is hh:mm:ss. (F) Surface area from (E). See also Video S5. (G) Cell tracking showing cell undergoing laser-induced apoptosis. Raw images (left and right) show LYSM⁺ (red) and BLIMP1⁺ (green) expression. Timelapse (middle panels) shows tracked double-labeled osteoclast and apoptotic fragments. Scale bar, 40 μm. Timestamp hh:mm:ss. See also Video S6. (H) Cell volume from (G). Red line identifies laser ablation. Grey window identifies apoptotic events. (I) Cell fate mapping of apoptotic cell fragments in (G). Circles represent relative cell volume on a log₁₀ scale. (J) Cumulative contacts between LYSM⁺CSF1R^neg macrophages and apoptotic cell fragments. Red line identifies laser ablation.

Figure 4

Osteoclasts recycle via osteomorphs (A) Cell tracking showing osteoclast recycling following sRANKL stimulation. Raw images (left and right) show LYSM⁺ (red) and CSF1R⁺ (green) expression. Timelapse (middle panels) shows tracked double-labeled osteoclast undergoing fission and fusion. Scale bar, 40μm. Timestamp hh:mm:ss. (B) Cell volume from (A). (C) Cell fate mapping from (A) showing a cell recycling (blue). See also Video S7. (D) Cell tracking showing osteoclast recycling following withdrawal of OPG:Fc treatment. Raw images (left and right) show LYSM⁺ (red) and BLIMP1⁺ (green) expression. Timelapse (middle panels) shows tracked double-labeled osteoclast undergoing recycling. Scale bar, 50 μm. Timestamp hh:mm:ss. (E) Cell volume from (D). (F) Cell fate mapping from (D) showing multiple fission, fusion, and cell recycling events. See also Video S8. (G) Number of cell fusion events and cell fission events in mice treated with vehicle (n = 8), sRANKL (n = 7), OPG:Fc (n = 8), or following withdrawal of OPG:Fc treatment (OPG:W) (n = 5). Each data point represents one mouse. Mean ± SEM are shown. (H) Percent of cells (undergoing fusion events and fission events in mice treated with vehicle, sRANKL, OPG:Fc, or following OPG:W). Each data point represents one mouse. Mean ± SEM are shown. See also Figure S1.

Figure S1

“Underground” cell fate maps depicting the fate of individual cells, related to Figures 3 and 4 (A) Underground maps of cells from mice treated with vehicle. (B) Underground maps of cells from mice treated with sRANKL. (C) Underground maps of cells from mice treated with OPG:Fc. (D) Underground maps of cells from mice following OPG:Fc withdrawal (OPG:W). Nodes represent cells and intersecting lines indicate fission or fusion events. Different colors represent different cells. Each map represents a replicate experiment. The number of experiments, the duration of imaging, the number of fission, fusion and recycling events and the number of events per hour are indicated.

Figure 5

Regulation of osteoclast recycling by RANKL signaling (A) Raw images of LYSM⁺BLIMP1⁺ osteoclasts from mice treated with vehicle (n = 9), sRANKL (n = 6), OPG:Fc (n = 8), or following OPG:W (n = 6). Scale bar, 60 μm. (B) Pseudocolor density plot of volume and sphericity for double-labeled cells that have taken up fluorescent bisphosphonate analyzed in (A). (C) Network analysis using FilamentTracer of cells in (A). Percentage of cells in each quadrant are shown. (D) Sum of network lengths and the proportion of networks >1 mm in length from (C). Each data point represents one mouse. Mean ± SEM are shown. (E) Longitudinal changes in serum TRAP 5b in mice treated with vehicle (n = 9) and following OPG:W (n = 8). Mean ± SEM are shown. (F) Longitudinal changes in BMD in mice treated with vehicle and OPG:W. Mean ± SEM are shown. (G) Micro-computed tomography (CT) images of cancellous bone in the femur of mice treated with vehicle (n = 9), sRANKL (n = 7), OPG:Fc (n = 7), and OPG:W at 5 (n = 6), 8 (n = 8), 12 (n = 6), and 17 (n = 7) weeks (w). Scale bar, 200 μm. (H) Trabecular bone volume/tissue volume (BV/TV) calculated from (G). (I) Histological images of the femur stained for TRAP (red) showing osteoclasts (arrow heads) in mice treated with vehicle (n = 8), sRANKL (n = 5), OPG:Fc (n = 6), and OPG:W at 5 (n = 7), 8 (n = 7), 12 (n = 6), and 17 (n = 7) w. BM, bone marrow; GP, growth plate. Scale bar, 50 μm. (J) Number of TRAP⁺ osteoclasts per mm of bone surface. Each data point represents one mouse. Mean ± SEM are shown. See also Figure S2.

Figure S2

Bone structure and activity in mice treated with vehicle, sRANKL, OPG:Fc, and following OPG:Fc withdrawal, related to Figure 5 (A) Micro-CT data showing effect on bone microarchitecture. (B) Enumeration of osteoclast cell surface (left) and numbers (right) per unit bone surface. Vehicle (n = 8), sRANKL (n = 5), OPG:Fc (n = 6) and OPG:W at 5 (n = 7), 8 (n = 7), 12 (n = 6) and 17 (n = 7) weeks (w). (C) Histomorphometry showing effect on osteoblast cell surface (far left) and number (left), osteoblast activity (middle, right, far right panels). Vehicle (n = 9-10) and OPG:W (n = 6-7).

Figure 6

Osteomorphs reassemble into osteoclasts capable of resorbing bone (A) Number of TRAP⁺ multinucleated cells generated from bone marrow mononuclear cells isolated from mice treated with vehicle (n = 2) or OPG:Fc (n = 2) and cultured in the presence of increasing doses of sRANKL. Mean ± SEM are shown. (B) Cells isolated from mice in (A) directly seeded onto dentine slices and cultured in the presence of M-CSF and sRANKL. TRAP⁺ osteoclasts (black arrowheads) and resorption pits on dentine slices (white arrowheads) as shown by scanning electron microscopy (SEM). Scale bar, 100 μm. (C) FACS analysis showing LYSM and CSF1R expression in Lysm^Cre/+.Tdtomato^LSL/LSL, Csf1r^Egfp/+, and Lysm^Cre/+.Tdtomato^LSL/LSL:Csf1r^Egfp/+ mixed bone marrow chimeras. (D) Histogram (left panel) showing uptake of zoledronic acid (ZOL) by cells in the indicated gates from Lysm^Cre/+.Tdtomato^LSL/LSL:Csf1r^Egfp/+ mixed bone marrow chimeras in (C). Right panel shows MFI for ZOL for the identified cell populations for individual mice (n = 6). Mean ± SEM are shown. (E) FACS-sorted LYSM⁺CSF1R⁺ cells cultured with non-fluorescent osteoclasts in vitro (four examples). Individual LYSM⁺ (red), CSF1R⁺ (green), and Hoechst (cyan) channels and the overlay are shown. Dotted lines identify osteoclasts. Scale bars, 50 μm. (F) FACS-sorted LYSM⁺CSF1R⁺ cells cultured on isolated calvarial bone (two examples). LYSM (red), CSF1R (green), and OsteoSense 680 (yellow) channels with Hoechst (cyan) and the overlay are shown. Dotted lines identify osteoclasts. Note nuclei of endogenous non-fluorescent cells. Scale bar, 20 μm.

Figure S3

Identification and isolation of triple-positive putative osteomorphs from marrow and osteoclasts from bone by FACS analysis and cell sorting, related to Figures 6 and 7 (A) Pseudocolor density plot shows the presence of circulating TOM⁺GFP⁺ cells in the blood in mixed bone marrow chimeras. (B) Pseudocolor density plot (left panel) shows the presence of TOM⁺GFP⁺ cells in the bone compartment in mixed bone marrow chimeras. Histogram overlay (middle panel) shows double-labeled cells selectively take up fluorescent bisphosphonate and are also ZOL⁺. Graph of ZOL MFI (right panel) for each of the indicated populations. (C) Gating strategy for FACS sorting of single-positive and triple-positive cells from marrow and bone. Pseudocolor density plot of single cell suspension from marrow showing single cell gates for doublet exclusion, gating for red and green single-positive cells (green box) and double-labeled cells that are also ZOL⁺ (red box). (D) Expression of the Egfp and Tdtomato genes in single cells sorted from marrow and bone for scRNA-seq. (E) FACS analysis comparing the expression of a number of osteomorph markers on single- and triple-positive cells from the marrow. Overlay histograms (top panels) show AXL, CD11B, CCR3, VCAM1, CD74 and CADM1 with mean ± SEM of the respective MFIs. Individual data points are plotted in bottom panels.

Figure 7

Osteomorph genes control bone structure, function, and disease LYSM^negCSF1R⁺ (TOM^negGFP⁺) and LYSM⁺CSF1R^neg (TOM⁺GFP^neg) single-positive cells (sPos) and LYSM⁺CSF1R⁺Zol⁺ (GFP⁺TOM⁺ZOL⁺) triple-positive (tPos) cells were index sorted from marrow and bone by FACS and analyzed by scRNA-seq. (A) Consensus plot showing cells in the marrow fraction decomposed into two clusters made up predominantly of tPos cells in one group (cluster 1) and GFP⁺ and TOM⁺ sPos (cluster 2) in another. (B) Consensus plot showing cells in both the marrow and bone fraction can be decomposed into two made up predominantly of tPos cells in one group (cluster 1) and sPos (cluster 2) in another irrespective of their tissue compartment. (C) Expression of canonical osteoclast genes by sPos and tPos cells isolated from bone marrow (top) and the bone fraction (bottom). (D) Consensus plot showing the marrow and bone fraction can be further decomposed into three clusters made up of tPos cells from bone (cluster 1), tPos cells from marrow (cluster 2), and sPos cells from both marrow and bone (cluster 3). (E) Visualization of three cell clusters based on their metagene expression. (F) Venn diagram showing overlap of genes upregulated in tPos cells isolated from bone (enriched with osteoclasts) and marrow (enriched with osteomorphs) compared to sPos macrophages isolated from the bone and marrow. (G) Genes upregulated in osteomorphs, with (light blue) and without (gray) outlier bone phenotypes, in 40 knockout mouse lines from the OBCD database. Eleven genes (blue box) have not been previously reported to be involved in skeletal structure or function. Two genes (asterisks) were uniquely upregulated in osteomorphs but not osteoclasts. (H) Representative quantitative X-ray microradiographic images of the femurs (top) and vertebrae (bottom) of adult, female wild-type (WT), Ddx56^+/−, Myo7a^−/−, and Wdr89^−/− mice. Scale bar, 1 mm. Dot plots illustrate summary data for individual parameters. For each variable, the mean (center line), ± 1 SD (middle lines), and ± 2 SD (gray boxes) for WT mice (n = 320) are shown. Mean values (colored dots) for each parameter in Ddx56^+/− (orange), Myo7a^−/− (green), and Wdr89^−/− (pink) mouse lines are shown. Genes were considered outliers if the mean was >2 SD from the WT reference mean and are denoted by an asterisk (^∗) and colored according to the individual mouse line. (I) Representative micro-CT images of cancellous (top) and cortical (bottom) bone of adult, female WT, Ddx56^+/−, Myo7a^−/−, and Wdr89^−/− mice. Scale bar, 100 μm. Dot plots illustrate summary data for bone volume as a proportion of tissue volume (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), cortical thickness (Ct.Th), internal endosteal diameter, and bone mineral density (BMD). Data are described, and outlier phenotypes identified, as in (H). (J) Model load displacement curves (left panel) and summary displacement curves (right panel) from three-point bend testing of the femur of adult, female WT, Ddx56^+/−, Myo7a^−/−, and Wdr89^−/− mice. Dot plots illustrate summary data for yield load, maximum load, fracture load, and stiffness. Data are described, and outlier phenotypes identified, as in (H). (K) 71 of the 520 osteomorph upregulated genes with human orthologs are significantly (P_MULTI < 2.4E-6) associated with eBMD in the UK Biobank study. Chromosomes number and genome coordinates of each human ortholog are shown.