Decision letter: Bone marrow Adipoq-lineage progenitors are a major cellular source of M-CSF that dominates bone marrow macrophage development, osteoclastogenesis, and bone mass

Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Methods Data availability References Decision letter Author response Article and author information Metrics Abstract M-CSF is a critical growth factor for myeloid lineage cells, including monocytes, macrophages, and osteoclasts. Tissue-resident macrophages in most organs rely on local M-CSF. However, it is unclear what specific cells in the bone marrow produce M-CSF to maintain myeloid homeostasis. Here, we found that Adipoq-lineage progenitors but not mature adipocytes in bone marrow or in peripheral adipose tissue, are a major cellular source of M-CSF, with these Adipoq-lineage progenitors producing M-CSF at levels much higher than those produced by osteoblast lineage cells. The Adipoq-lineage progenitors with high CSF1 expression also exist in human bone marrow. Deficiency of M-CSF in bone marrow Adipoq-lineage progenitors drastically reduces the generation of bone marrow macrophages and osteoclasts, leading to severe osteopetrosis in mice. Furthermore, the osteoporosis in ovariectomized mice can be significantly alleviated by the absence of M-CSF in bone marrow Adipoq-lineage progenitors. Our findings identify bone marrow Adipoq-lineage progenitors as a major cellular source of M-CSF in bone marrow and reveal their crucial contribution to bone marrow macrophage development, osteoclastogenesis, bone homeostasis, and pathological bone loss. Editor's evaluation This fundamental work advances our understanding of the function of a subpopulation of bone marrow cells as an important source of M-CSF to regulate bone remodeling. The evidence supporting the conclusion is compelling, using Adipoq-Cre-driven conditional deletion of Csf1 and the analysis of the publicly available scRNAseq data. This paper is of interest to skeletal biologists studying bone marrow stem/progenitor cells and bone remodeling. https://doi.org/10.7554/eLife.82118.sa0 Decision letter Reviews on Sciety eLife's review process Introduction Macrophage colony-stimulating factor (M-CSF), encoded by the Csf1 gene, is a growth factor that plays a crucial role in the proliferation, differentiation, survival and function of myeloid lineage cells, including monocytes, macrophages, and osteoclasts (Stanley et al., 1997; Ross, 2006). The global absence of Csf1 in the Csf1op/ Csf1op mice, which carry an inactivating mutation in the coding region of Csf1, leads to tissue macrophage deficiency, growth retardation and skeletal abnormality. The severe osteopetrosis, shortened long bones and failure of dental eruption in Csf1op/ Csf1op mice attest to the essential role of M-CSF for osteoclast generation (Cecchini et al., 1994; Dai et al., 2002; Felix et al., 1990; Wiktor-Jedrzejczak et al., 1982; Wiktor-Jedrzejczak et al., 1990; Yoshida et al., 1990). Osteoclasts are giant multinucleated cells responsible for bone resorption. They are derived from the monocyte/macrophage lineage and are a specialized terminally differentiated macrophage. Osteoclasts play an important role not only in physiological bone development and remodeling, but also actively contribute to musculoskeletal tissue damage and accelerating the progression of postmenopausal osteoporosis and inflammatory arthritis. Osteoclasts and their progenitors express the M-CSF receptor c-Fms. M-CSF is essential for the entire process of osteoclast differentiation, from the generation of osteoclast precursors to the formation and survival of mature osteoclasts (Ross, 2006; Tanaka et al., 1993; Feng and Teitelbaum, 2013). M-CSF also acts together with β3 integrin to regulate actin remodeling in osteoclasts to enable osteoclasts to spread, migrate, fuse, and form actin rings to facilitate bone resorption (Ross, 2006; Ross and Teitelbaum, 2005; Insogna et al., 1997). These findings underscore the indispensable role of Csf1 in not only monocyte to macrophage development, but also osteoclast differentiation, function and survival, which directly influences bone mass. M-CSF is expressed by a variety of cells, such as endothelial cells, myoblasts, epithelial cells, and fibroblasts (Cecchini et al., 1994; Schrader et al., 1991; Clinton et al., 1992; Fibbe et al., 1988). This diversity of cellular sources of M-CSF is likely due to the need to support a widely distributed network of tissue-resident macrophages, which require M-CSF for their development, function, and homeostasis. Under physiologic conditions, circulating M-CSF is mainly produced by vascular endothelial cells (Clinton et al., 1992; Roth and Stanley, 1992; Ryan et al., 2001). The role of circulating M-CSF in tissue macrophage support appears to be highly context dependent. For example, macrophages in kidney and liver are highly dependent on circulating M-CSF. Bone marrow resident macrophages, however, appear to not require circulating M-CSF, indicating the importance of local M-CSF produced in bone marrow (Cecchini et al., 1994). Osteocytes embedded in bone express M-CSF that contributes to bone mass maintenance (Werner et al., 2020). Although in vitro studies show that M-CSF is expressed by an osteoblast cell line, calvarial osteoblastic cells and a cloned bone marrow stromal cell line (Tanaka et al., 1993; Elford et al., 1987; Zhao et al., 2002; Naparstek et al., 1986), there is no clear evidence supporting whether osteoblasts express M-CSF in vivo and the contributions of osteoblast-derived M-CSF to macrophage and bone homeostasis in vivo is unclear. Therefore, it is largely unknown what specific bone marrow cellular population mainly produces this critical cytokine. The rapid evolution of scRNAseq technology provides an opportunity to investigate transcriptomics at the individual cell level. In this study, we utilized bone marrow scRNAseq datasets (Zhong et al., 2020; Wolock et al., 2019; Tikhonova et al., 2020; Baryawno et al., 2019; Baccin et al., 2020; Dolgalev and Tikhonova, 2021), and identified a group of unique bone marrow cells featuring Adipoq expression that highly express Csf1. The level of Csf1 expressed by these bone marrow Adipoq-lineage progenitor cells is much higher than that produced by osteoblast lineage cells. We further demonstrated the functional importance of the Csf1 expressed by bone marrow Adipoq-lineage progenitors in macrophage development, osteoclastogenesis and bone mass maintenance. Results scRNAseq reveals Adipoq-lineage progenitors as a main cellular source expressing Csf1 in bone marrow Along with the rapid progress of single-cell RNAseq technology, single-cell transcriptomics provides an unprecedented assessment of tissue cellular composition and gene expression profile at individual cell resolution. We took advantage of a recently published dataset (Dolgalev and Tikhonova, 2021), which integrated three bone marrow scRNAseq datasets (Tikhonova et al., 2020; Baryawno et al., 2019; Baccin et al., 2020), and analyzed the expression profiles of non-hematopoietic bone marrow cells. Adipoq is found to be most highly expressed in the cluster MSPC-adipo (mesenchymal progenitor cells-adipo lineage)/Adipoq-lineage progenitors (Figure 1A, B and D) as a marker of this cell population among the clusters. This MSPC-Adipo cluster (Dolgalev and Tikhonova, 2021) was described as the adipo-primed mesenchymal progenitors (Tikhonova et al., 2020), adipocyte progenitors (Wolock et al., 2019), Adipo-CAR (Cxcl12-Abundant Reticular) cells (Baccin et al., 2020), Lepr-MSC (Baryawno et al., 2019), or marrow adipogenic lineage precursors (MALPs) (Zhong et al., 2020) identified in bone marrow. Since we utilized Adipoq Cre mice to investigate the function of this progenitor population, we used the nomenclature bone marrow Adipoq-lineage progenitors to designate these cells throughout this study. We found the Adipoq-lineage progenitors to be highly enriched for bone marrow stromal cell markers important for the hematopoietic niche, such as Lepr, Cxcl12 and Kitl, but also express unique genes, such as Lpl (Figure 1B, Figure 3A). These cells labeled by Adipoq Cre were relatively quiescent, with about 4% cells incorporating BrdU under basal conditions (Figure 1C). The Adipoq-lineage progenitors are not mature adipocytes, but express some common adipocyte lineage markers, such as Cebpa and Adipoq (Figure 1B). On the other hand, osteoblast lineage marker genes, such as Sp7, Alpl, Dmp1, and Bglap, are nearly undetectable in these cells (Figure 1B). Figure 1 Download asset Open asset Integrated analysis of the bone marrow niche datasets of scRNAseq shows that Adipoq-lineage progenitors (Adipoq+ MSPCs) express high level of Csf1. (A) UMAP plot of the integrated analysis of the bone marrow niche datasets of scRNAseq based on Dolgalev and Tikhonova, 2021. EC, endothelial cell; MSPC: mesenchymal progenitor cell. (B) Dot plot of several typical marker gene expression for bone marrow stromal cells, adipocyte lineage, osteoblast lineage and endothelial cells across the listed scRNAseq clusters. Cell clusters are listed on the y-axis. Features are listed along the x-axis. Dot size reflects the percentage of cells in a cluster expressing each gene. Dot color reflects scaled average gene expression level as indicated by the legend. (C) Flowcytometry images and quantification of the bone marrow Adipoq-lineage progenitors (Adipoq+) incorporating BrdU in 10-week-old female Adipoq Cre-mTmG mice. n=5. (D) UMAP plots of the expression of Adipoq (upper left panel), Csf1 (upper right panel) and the co-expression of these two genes (lower left panel) in bone marrow cells. The lower right panel shows a relative expression scale for each gene. (E) Dot plot of Adipoq and Csf1 expression across the listed scRNA-seq clusters. Cell clusters are listed on y-axis. Features are listed along the x-axis. Dot size reflects the percentage of cells in a cluster expressing each gene. Dot color reflects the scaled average gene expression level as indicated by the legend. (F) Violin plots of the expression of Adipoq and Csf1. Figure 1—source data 1 Integrated analysis of the bone marrow niche. https://cdn.elifesciences.org/articles/82118/elife-82118-fig1-data1-v2.zip Download elife-82118-fig1-data1-v2.zip When screening expression of known genes regulating skeletal homeostasis in the bone marrow, we found that these bone marrow Adipoq-lineage progenitors (MSPC-adipo cluster) express the highest level of Csf1 (Figure 1D, E and F), which was unexpected because the MSPC-osteo (mesenchymal progenitor cells-osteo lineage) cluster and osteoblasts were thought to be the cellular source expressing Csf1 in bone marrow based on previous studies of in vitro cultured osteoblast cell lines or calvarial osteoblastic cells (Tanaka et al., 1993; Elford et al., 1987). Further quantitative analysis showed that bone marrow Adipoq-lineage progenitors (70% of 6441 cells expressed with scaled average expression level at 2.6) express a markedly higher level of Csf1 than MSPC-osteo cluster cells (61% of 2247 cells expressed with scaled average expression level at 1.3) (Figure 1E, Figure 1—source data 1). Csf1 expression was negligible in osteo cluster and osteoblasts (scaled average expression level at 0, Figure 1E, Figure 1—source data 1). These results indicate that the Adipoq-lineage progenitors produce substantially more Csf1 than the osteoblast lineage cells. We next asked whether the Adipoq-lineage progenitors exist in human bone marrow. We analyzed a recently published scRNAseq dataset based on human femur bone marrow (Wang et al., 2021), and identified a similar cell population (cluster 1, Adipoq-lineage progenitors) that expressed high ADIPOQ, bone marrow stromal marker genes LEPR, CXCL12 and KITLG, LPL and CEBPA, but with nearly undetectable osteoblast lineage genes (Figure 2A and B, Figure 2—source data 1). This cell population was mostly highly enriched in CSF1 expression across the clusters (Figure 2C, D and E). In an additional human scRNAseq dataset (Li et al., 2022), we also found a similar human bone marrow stromal cluster (cluster 5, Adipo) that simultaneously expresses high levels of ADIPOQ and CSF1 (Figure 2—figure supplement 1). These results indicate that human bone marrow contains a cell population that is highly similar to the bone marrow Adipoq-lineage progenitors in mice with conserved robust expression of CSF1. Figure 2 with 1 supplement see all Download asset Open asset scRNAseq analysis of human bone marrow unveils the existence of ADIPOQ-lineage progenitors highly expressing CSF1. (A) UMAP plot analysis of the human bone marrow datasets of scRNAseq based on Wang et al., 2021. (B) Dot plot of several typical marker gene expression for bone marrow stromal cells, adipocyte lineage, osteoblast lineage and endothelial cells across the listed scRNA-seq clusters. Cell clusters are listed on the y-axis. Features are listed along the x-axis. Dot size reflects the percentage of cells in a cluster expressing each gene. Dot color reflects scaled average gene expression level as indicated by the legend. (C) UMAP plots of the expression of ADIPOQ (upper left panel), CSF1 (upper right panel) and the co-expression of these two genes (lower left panel) in bone marrow cells. The lower right panel shows a relative expression scale for each gene. (D) Dot plot of ADIPOQ and CSF1 expression across the listed scRNAseq clusters. Cell clusters are listed on y-axis. Features are listed along the x-axis. Dot size reflects the percentage of cells in a cluster expressing each gene. Dot color reflects the scaled average gene expression level as indicated by the legend. (E) Violin plots of the expression of ADIPOQ and CSF1. Figure 2—source data 1 scRNAseq analysis of human bone marrow. https://cdn.elifesciences.org/articles/82118/elife-82118-fig2-data1-v2.zip Download elife-82118-fig2-data1-v2.zip Figure 2—source data 2 Markers for cell type annotation (Baryawno et al., 2019; Baccin et al., 2020; Dolgalev and Tikhonova, 2021; Wang et al., 2021; Komori et al., 1997; Nakashima et al., 2002; Diegel et al., 2020; Wang et al., 2018; Meijer et al., 2012; Bajpai et al., 2018; Fuentes-Duculan et al., 2010; Ippolito et al., 2014; Esashi et al., 2008; Honda et al., 2005; Kapellos et al., 2019; Bian et al., 2020; Mathewson et al., 2021; Shi et al., 2019; Ledergor et al., 2018) in Figure 2. https://cdn.elifesciences.org/articles/82118/elife-82118-fig2-data2-v2.zip Download elife-82118-fig2-data2-v2.zip M-CSF is expressed in the bone marrow Adipoq-lineage progenitors but nearly undetectable in mature adipocytes in bone marrow or peripheral adipose As Adipoq is expressed in mature adipocytes (Berry and Rodeheffer, 2013), we wondered whether M-CSF is also produced by these cells. We sorted the Adipoq-lineage progenitors from the Adipoq Cre-mTmG mice, and isolated both white and brown lipid-laden mature adipocytes, as well as the lipid-laden mature bone marrow adipocytes (BMAd). We found that the mRNA expression of Csf1 in bone marrow Adipoq-lineage progenitor cells is 20–30 fold higher than that of mature adipocytes (Figure 3A, Figure 3—figure supplement 1). The Csf1 expression in bone marrow Adipoq-lineage progenitor cells is also much higher than that of the stromal vascular fraction (SVF) cells in white adipose (Figure 3A). With this striking difference in mRNA expression, we further performed immunofluorescence staining of M-CSF on bone slices. We observed that the majority of bone marrow Adipoq-expressing progenitor cells express M-CSF (Figure 3B, 1865 cells out of 2001 cells counted, n=3 mice, 93.2%). In contrast, M-CSF expression was not detected in mature bone marrow adipocytes (Perilipin1+) (Figure 3C, 0 cells out of 115 cells counted, n=3 mice, 0%), indicating that mature bone marrow adipocytes are unlikely to be a significant source of M-CSF. Moreover, we performed western blot to analyze M-CSF protein expression in peripheral adipose tissue. As shown in Figure 3D, the stromal vascular fraction (SVF) cells in adipose, which contain multiple cell populations including adipogenic progenitors, express M-CSF. On the contrary, M-CSF was nearly undetectable in the peripheral mature adipocytes isolated from inguinal and epididymal white adipose tissue (WAT) (Figure 3D). These data collectively support that mature adipocytes are not a significant source of M-CSF as evidenced by nearly undetectable M-CSF expression compared to the Adipoq-lineage progenitors. These results identify Adipoq-lineage progenitors residing in bone marrow as a new cell type highly expressing M-CSF. Figure 3 with 1 supplement see all Download asset Open asset M-CSF is mainly produced by the bone marrow Adipoq-lineage progenitors, but not by mature adipocytes in peripheral adipose or in bone marrow. (A) qPCR analysis of Csf1 and Lpl expression in bone marrow Adipoq-linage progenitors that were sorted from the bone marrow of Adipoq Cre-mTmG reporter mice, mature bone marrow adipocytes (BMAd), mature peripheral white and brown adipocytes and white stromal vascular fraction (SVF). E-adipocyte: mature adipocytes isolated from the epididymal white adipose tissue. I-adipocyte: mature adipocytes isolated from inguinal white adipose tissue. B-adipocyte: Brown adipocytes. n=5 for E-, I- and B-adipocytes from 12-week-old male mice. Five replicates, each with a pooled sample from 12-week-old male mice for BMAd (6–7 mice) and bone marrow Adipoq-lineage progenitors (3–4 mice). Error bars: Data are mean ± SD. ****p<0.0001 by one-way ANOVA analysis followed by post hoc Bonferroni’s correction for multiple comparisons. (B) Immunofluorescence staining of M-CSF (purple) on femur bone slices from 12-week-old male Adipoq Cre-mTmG reporter mice. DAPI: blue. Arrows: co-localization of M-CSF and Adipoq-GFP in GFP +Adipoq-lineage progenitors. n=3 mice. (C) Immunofluorescence staining of M-CSF (purple) and Perilipin1 (green, mature adipocyte marker) on femur bone slices from 12-week-old male mice. n=3. (D) Immunoblot analysis of M-CSF and Adiponectin expression in mature adipocytes and stromal vascular fraction (SVF) in peripheral adipose. (E) epididymal white adipose tissue. (I) the inguinal white adipose tissue. p38 was used as a loading control. Figure 3—source data 1 M-CSF is mainly produced by the bone marrow Adipoq-lineage progenitors, but not by mature adipocytes in peripheral adipose or in bone marrow. https://cdn.elifesciences.org/articles/82118/elife-82118-fig3-data1-v2.zip Download elife-82118-fig3-data1-v2.zip Adipoq Cre-driven Csf1 conditional knock out (Csf1∆Adipoq) mice exhibit osteopetrosis We next sought to investigate the contribution of the M-CSF produced by bone marrow Adipoq-lineage progenitors to bone development and homeostasis. We generated Csf1 conditional knock out (KO) mice, in which Csf1 is specifically deleted in Adipoq + cells by crossing Csf1flox/flox mice with Adipoq Cre mice (Csf1f/f;AdipoqCre; hereafter referred to as Csf1∆Adipoq). Their littermates with a Csf1f/f genotype were used as the controls. Compared to control mice, Csf1 expression was reduced by approximately 75% in a total bone marrow stromal cell culture derived from the Csf1∆Adipoq mice (Figure 4A). Given that bone marrow Adipoq-lineage progenitors constitute only about 0.08% of bone marrow cells (Figure 4—figure supplement 1), these results, together with the data shown in Figures 1—3, support that bone marrow Adipoq-lineage progenitors are the major cellular source of M-CSF expression in the bone marrow. Furthermore, immunofluorescence staining of bone slices showed a drastic decrease in M-CSF protein expression in bone marrow Adipoq-lineage progenitor cells in Csf1∆Adipoq mice compared to the control mice (Figure 4B). Although SVF cells express Csf1, these cells do not express Adipoq (Figure 3D). Thus, it is unlikely that Csf1 expression in SVF is changed in Csf1∆Adipoq mice. Indeed, our results showed that Csf1 expression in SVF was not altered in Csf1∆Adipoq mice (Figure 4—figure supplement 2). These results clearly demonstrate that adipoq-cre does not target SVF cells. We also examined the expression of a group of cytokines that often regulate macrophage function and osteoclastogenesis in bone marrow, including Il34, Il1b, Il6, Il10, Csf2, Tnf, Cxcl12, and found that the deficiency of Csf1 in Csf1∆Adipoq mice did not affect the expression of these genes (Figure 4—figure supplement 3). Csf1∆Adipoq mice did not display abnormalities in gross appearance, body weight, tooth eruption and long bone length (Figure 4C and D, Figure 4—figure supplement 4). In contrast, microcomputed tomographic (µCT) analyses showed that Csf1∆Adipoq mice exhibited a marked osteopetrotic phenotype, as indicated by a twofold increase in trabecular bone mass and marked increases in bone mineral density (BMD), connectivity density (Conn-Dens.), trabecular bone number and a decrease in trabecular bone spacing compared to the littermate control mice (Figure 4E and F). Cortical bone appeared normal in Csf1∆Adipoq mice (Figure 4G). In addition, heterozygous Csf1 conditional knockout mice (Csf1f/+;Adipoq Cre) did not show an abnormal bone phenotype compared to control mice (Csf1f/f) (Figure 4—figure supplement 5). There are no differences in vertebral BMD or bone mass between the control and Csf1∆Adipoq mice (Figure 4—figure supplement 6). Given that peripheral mature adipocytes (Adipoq + cells in peripheral adipose tissue) and mature bone marrow adipocytes (Adipoq +lipid-laden cells in bone marrow) express negligible levels of M-CSF, these data indicate that Csf1 in bone marrow Adipoq-lineage progenitors plays a key role in the bone mass maintenance of long bones. Figure 4 with 6 supplements see all Download asset Open asset Csf1 deficiency in Csf1ΔAdipoq mice increases bone mass. (A) Csf1 expression in BMSCs derived from Csf f/f and Csf1ΔAdipoq (n = 4/group). (B) Representative images of immunostaining of Adiponectin (red) and M-CSF (white) on femur bone slices from 12-week-old male Csf f/f and Csf1ΔAdipoq mice. DAPI: blue. n=3/group. (C) Gross appearance of the incisors from Csf f/f and Csf1ΔAdipoq mice. (D) Gross appearance of the femur (left panel), and the lengths of femur and tibia from Csf f/f and Csf1ΔAdipoq mice (right panels) (n = 6/group). (E) μCT images and (F) bone morphometric analysis of trabecular bone of the distal femurs isolated from 12-week-old male Csf f/f and Csf1ΔAdipoq mice (n = 5/group). (G) μCT images and bone morphometric analysis of cortical bone of the mid-shaft femurs isolated from 12-week-old male Csf f/f and Csf1ΔAdipoq mice (n = 5/group). BM, bone marrow; BMSC, bone marrow stromal cell; BV/TV, bone volume per tissue volume; BMD, bone mineral density; Conn-Dens, connectivity density; Tb.Th, trabecular thickness; Tb.Sp, trabecular separation; Tb.N, trabecular number. Ct.Th, cortical bone thickness; BA/TA: Bone area/Tissue area. A, D, F, G *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns: not statistically significant by two tailed unpaired Student’s t test. Error bars: Data are mean ± SD. Scale bars: B, 50 µm; E, 500 µm; G, 500 μm. Figure 4—source data 1 Csf1 deficiency in Csf1ΔAdipoq mice increases bone mass. https://cdn.elifesciences.org/articles/82118/elife-82118-fig4-data1-v2.zip Download elife-82118-fig4-data1-v2.zip Bone marrow macrophages and osteoclasts are suppressed in Csf1∆Adipoq mice The osteopetrotic phenotype in Csf1∆Adipoq mice implicated a defect in osteoclast function. Indeed, Csf1 deficiency in Adipoq+ cells impaired osteoclast formation in vivo evidenced by reduced osteoclast surface area and lower osteoclast numbers (Figure 5A). The TRAP level in serum was significantly lower in Csf1∆Adipoq mice than that in control mice (Figure 5B). On the other hand, mineral apposition rate (MAR) and bone formation rate (BFR) were not affected in Csf1∆Adipoq mice, indicating normal osteoblastic function in these mice (Figure 5—figure supplement 1). Since osteoclasts are derived from the myeloid macrophage lineage, we examined bone marrow macrophage populations. CD11b+Ly6 Chi monocytes were similar between control and Csf1∆Adipoq mice. CD11b+F4/80+macrophages were reduced by almost half in Csf1∆Adipoq mice (Figure 5C). This decrease in macrophages reflects the effects of M-CSF deficiency, as M-CSF is critical for macrophage development. We further stimulated bone-marrow-derived macrophages (BMMs) with LPS and found that the inflammatory response of BMMs, indicated by inflammatory gene induction (Figure 5—figure supplement 2A) and the activation of MAPK or NF-κB pathways in response to LPS (Figure 5—figure supplement 2B), was similar between the BMMs derived from the control and Csf1∆Adipoq mice. To further test the importance of M-CSF produced by bone marrow Adipoq-lineage progenitors for osteoclastogenesis, we cultured whole bone marrow ex vivo without exogenous M-CSF to test whether the Adipoq-lineage cell-produced M-CSF is sufficient to function together with RANKL to induce osteoclast differentiation. As shown in Figure 5D, RANKL can induce osteoclast differentiation in the control bone marrow cultures even without addition of M-CSF, but it failed to induce osteoclastogenesis in the Csf1∆Adipoq bone marrow culture. When recombinant M-CSF was added back to the bone marrow cultures, the osteoclast formation in Csf1∆Adipoq cell cultures was similar to that in the control cultures (Figure 5E). These results demonstrate that the M-CSF secreted by bone marrow resident Adipoq-lineage progenitors is critical for osteoclastogenesis. Figure 5 with 2 supplements see all Download asset Open asset Csf1 deficiency in Csf1ΔAdipoq mice suppresses the populations of bone marrow macrophages and osteoclasts. (A) TRAP staining and histomorphometric analysis of histological sections obtained from the metaphysis region of distal femurs of 12-week-old male Csf f/f and Csf1ΔAdipoq mice (n = 5/group). (B) ELISA analysis of serum TRAP levels in 12-week-old male Csf f/f and Csf1ΔAdipoq mice (n = 6/group). (C) Flowcytometry image (left) and quantification (right) of monocytes and macrophages in bone marrow. n=6/group. (D, E) Osteoclast differentiation directly from the cultures of bone marrows harvested from Csf f/f and Csf1ΔAdipoq mice stimulated with RANKL (40 ng/ml) but without recombinant M-CSF for ten days (D) or with both RANKL and recombinant M-CSF (20 ng/ml) for five days (E). TRAP staining (left panel) was performed and the area of TRAP-positive MNCs (≥3 nuclei/cell) per well was calculated (right panel). TRAP-positive cells appear red in the photographs. (n = 3/group). Oc.S/BS, osteoclast surface per bone surface; N.Oc/B.Pm, number of osteoclasts per bone perimeter. (A, B), C, D, E **p<0.01; ***p<0.001; ****p<0.0001; ns: not statistically significant by two tailed unpaired Student’s t test. Error bars: Data are mean ± SD. Scale bars: A, 100 µm; D, E, 200 µm. Figure 5—source data 1 Csf1 deficiency in Csf1ΔAdipoq mice suppresses the populations of bone marrow macrophages and osteoclasts. https://cdn.elifesciences.org/articles/82118/elife-82118-fig5-data1-v2.zip Download elife-82118-fig5-data1-v2.zip Csf1 deficiency in bone marrow Adipoq-lineage progenitors does not affect macrophage development in peripheral adipose tissue and spleen Besides bone marrow, peripheral adipose tissue contains a large amount of Adipoq +mature adipocytes. However, M-CSF expression is undetectable in these cells (Figure 3A and D). In addition, some organs, such as spleen, have many tissue macrophages but few Adipoq + cells. We then asked whether Csf1 expressed by bone marrow Adipoq-lineage progenitors affects macrophages outside of the bone marrow, such as in peripheral adipose tissue or the spleen. The gross appearance and weight of the spleen and peripheral adipose tissue, including the inguinal and epididymal adipose depots, were normal in Csf1∆Adipoq mice (Figure 6A and B). There was no difference in CD11b+Ly6 Chi monocytes in either the spleen, inguinal or epididymal adipose tissue between control and Csf1∆Adipoq mice, and the amount of CD11b+F4/80+macrophages was unchanged in these tissues (Figure 6C). Thus, in contrast to the effects on bone marrow macrophages, Csf1 deficiency in bone marrow Adipoq-lineage progenitors does not influence macrophage development in peripheral adipose tissue or the spleen. Figure 6 Download asset Open asset Csf1 deficiency in Csf1ΔAdipoq mice does not affect monocyte and macrophage populations in spleen and peripheral adiposes. (A) Gross appearance (left panel) and weight (right panel) of the spleen from Csf f/f and Csf1ΔAdipoq mice (n = 6/group). (B) Gross appearance and weight of the inguinal (left panels) and the epididymal (right panels) adipose from Csf f/f and Csf1ΔAdipoq mice (n = 6/group). (C) Flowcytometry quantification of monocytes and macrophages (gated on CD45 +Ly6G- cells) in the indicated tissues. n=6/group. (A, B, C) ns: not statistically significant by two tailed unpaired Student’s t test. Error bars: Data are mean ± SD. Figure 6—source data 1 Csf1 deficiency in Csf1ΔAdipoq mice does not affect monocyte and macrophage populations in spleen and peripheral adiposes. https://cdn.elifesciences.org/articles/82118/elife-82118-fig6-data1-v2.zip Download elife-82118-fig6-data1-v2.zip Lack of Csf1 in bone marrow Adipoq-lineage progenitors alleviates estrogen-deficiency induced osteoporosis We next investigated the significance of M-CSF secreted by Adipoq+ cells in pathological bone loss. We developed the ovariectomy (OVX) model in Csf1∆Adipoq mice to study the contribution of Adipoq+ cell-produced M-CSF to the estrogen-deficiency induced osteoporosis, which mimics postmenopausal bone loss. Ut