Skip to main content
PMC home page
eLife logoLink to eLife
. 2020 May 13;9:e54493. doi: 10.7554/eLife.54493

Dissecting the phenotypic and functional heterogeneity of mouse inflammatory osteoclasts by the expression of Cx3cr1

Maria-Bernadette Madel 1,2, Lidia Ibáñez 3, Thomas Ciucci 4, Julia Halper 1,2, Matthieu Rouleau 1,2, Antoine Boutin 1,2, Christophe Hue 5, Isabelle Duroux-Richard 6, Florence Apparailly 6, Henri-Jean Garchon 5,7, Abdelilah Wakkach 1,2, Claudine Blin-Wakkach 1,2,✉
Editors: Gordana Vunjak-Novakovic8, Kathryn Song Eng Cheah9
PMCID: PMC7220377  PMID: 32400390

Abstract

Bone destruction relies on interactions between bone and immune cells. Bone-resorbing osteoclasts (OCLs) were recently identified as innate immune cells activating T cells toward tolerance or inflammation. Thus, pathological bone destruction not only relies on increased osteoclast differentiation, but also on the presence of inflammatory OCLs (i-OCLs), part of which express Cx3cr1. Here, we investigated the contribution of mouse Cx3cr1+ and Cx3cr1neg i-OCLs to bone loss. We showed that Cx3cr1+ and Cx3cr1neg i-OCLs differ considerably in transcriptional and functional aspects. Cx3cr1neg i-OCLs have a high ability to resorb bone and activate inflammatory CD4+ T cells. Although Cx3cr1+ i-OCLs are associated with inflammation, they resorb less and have in vitro an immune-suppressive effect on Cx3cr1neg i-OCLs, mediated by PD-L1. Our results provide new insights into i-OCL heterogeneity. They also reveal that different i-OCL subsets may interact to regulate inflammation. This contributes to a better understanding and prevention of inflammatory bone destruction.

Research organism: Mouse

Introduction

Chronic inflammation and bone destruction are frequently associated due to complex interactions between activated immune cells and progenitors of bone-resorbing osteoclasts (OCLs). While they have long been attributed to the sole effect of immune cells on osteoclastogenesis, recent evidences demonstrated that these interactions are reciprocal (Buchwald et al., 2012; Ibáñez et al., 2016; Kiesel et al., 2009; Li et al., 2010). Indeed, OCLs are monocytic cells responding to immune signals and beside bone resorption, they present antigens and activate T cells (Madel et al., 2019). However, to date, the mechanisms that reciprocally link inflammation and bone destruction remain poorly understood.

OCLs are long-lived multinucleated cells arising from bone marrow (BM) progenitors (Arai et al., 1999; Jacome-Galarza et al., 2013; Jacome-Galarza et al., 2019). However, in inflammation, they also arise from inflammatory Ly6Chi monocytes (MNs) (Ammari et al., 2018; Seeling et al., 2013) and dendritic cells (DCs) including in vivo (Wakkach et al., 2008; Ibáñez et al., 2016; Mansour et al., 2012; Rivollier et al., 2004; Madel et al., 2019). Furthermore, OCLs permanently fuse with MNs and undergo fission to maintain their nuclei number (Jacome-Galarza et al., 2019). We and others showed that depending on their origin and environment, OCLs induce different T cell responses. In steady state or when derived from BM MNs of healthy mice, OCLs activate regulatory CD4+ and CD8+ T cells (tolerogenic-OCLs/t OCLs) (Ibáñez et al., 2016; Kiesel et al., 2009), whereas when derived from DCs or during inflammation, they induce TNFα-producing CD4+ T cells (inflammatory-OCLs/i-OCLs) (Ibáñez et al., 2016; Madel et al., 2019). Thus, according to their origin and environment, OCLs are heterogeneous and initiate different immune responses. However, the full contribution of i-OCLs to inflammatory bone loss remains to be elucidated.

Recently, we identified the fractalkine receptor Cx3cr1 as a marker identifying i-OCLs (Ibáñez et al., 2016). Cx3cr1 is expressed by OCL precursors and mediates their BM recruitment by Cx3cl1-producing endothelial cells and osteoblasts (Han et al., 2014; Hasegawa et al., 2019; Koizumi et al., 2009; Matsuura et al., 2017). During osteoclastogenesis, the level of expression of Cx3cr1 decreases and is very low in t-OCLs (Hoshino et al., 2013; Koizumi et al., 2009). However, in inflammatory conditions, Cx3cr1 remains expressed in about 25% of i-OCLs (Ibáñez et al., 2016). Nevertheless, the role of Cx3cr1 in i-OCLs as well as the function and contribution of Cx3cr1neg and Cx3cr1+ i-OCLs in inflammation have not yet been explored.

We addressed these questions using Cx3cr1GFP mutant mice in which the Gfp gene is inserted in the second exon of the Cx3cr1 gene (Jung et al., 2000). Consequently, while Cx3cr1 activity is normal in wild type (WT) Cx3cr1GFP/+ mice, this protein is non-functional in Cx3cr1GFP/GFP mice and in both mice, GFP expression allows to identify Cx3cr1-expressing cells (Jung et al., 2000). We evaluated the consequences of Cx3cr1 deficiency on inflammatory bone loss in vivo in ovariectomized (OVX) mice, a model where osteoclastogenesis is driven by TNF-α and RANK-L-producing CD4+ T cells (Cenci et al., 2000; Li et al., 2011; Weitzmann and Pacifici, 2006). These conditions are similar to those priming Cx3cr1+ OCLs previously described in inflammatory colitis (Ibáñez et al., 2016). We also compared Cx3cr1+ and Cx3cr1neg i-OCLs as well as Cx3cr1-deficient i-OCLs by RNA-sequencing (RNA-seq) analysis and in vitro functional assays. We showed that Cx3cr1 deficiency does not change the bone resorption and T cell activation capacity of mature Cx3cr1+ i-OCLs demonstrating that the Cx3cr1 protein per se does not play a major role in these functions. Furthermore, we showed that Cx3cr1 is a marker that identifies two distinct subsets of i-OCLs (Cx3cr1neg and Cx3cr1+) having different immune functions. Our findings unveil the heterogeneity of i-OCLs and their contribution to inflammation and bone loss. These new insights into osteoimmunology and OCL heterogeneity contribute to a better understanding of the bone microenvironment regulation and the underlying molecular processes during inflammatory bone destruction.

Results

Cx3cr1 deficiency protects against inflammatory bone loss in osteoporosis

We addressed in vivo the participation of Cx3cr1 in inflammatory osteoclastogenesis in ovariectomized (OVX) mice. We previously showed that Cx3cr1+ i-OCLs increased in inflammatory colitis associated with bone destruction (Ibáñez et al., 2016) and we confirmed here that they also increased after ovariectomy in WT OVX mice (Figure 1A). To evaluate the implication of Cx3cr1 in OVX-induced bone loss, we used Cx3cr1-deficient Cx3cr1GFP/GFP and WT Cx3cr1GFP/+ control mice. Micro-CT analysis revealed that after ovariectomy, Cx3cr1-deficient mice displayed moderate but significant higher BV/TV and trabecular number and lower trabecular separation than WT Cx3cr1GFP/+ mice (Figure 1B–C). They also had less TRAcP+ OCLs compared to OVX WT Cx3cr1GFP/+ mice (Figure 1D). Thus, Cx3cr1 deficiency partially protects against bone destruction in OVX mice.

Figure 1. Cx3cr1-deficient mice display reduced bone loss after ovariectomy.

(A) FACS analysis of Cx3cr1+ OCLs generated from BM cells from SHAM-operated and OVX WT mice. Histograms indicate the mean ± SD percentage of Cx3cr1+ i-OCLs (n = 4–5). (B) Representative images of femur µCT analysis from SHAM-operated and OVX WT Cx3cr1GFP/+ and Cx3cr1-deficient Cx3cr1GFP/GFP mice 6 weeks post-surgery. (C) Histograms show the mean ± SD percentage of bone volume fraction (BV/TV), trabecular number and trabecular separation (n = 8 mice per group). (D) Representative TRAcP staining on femora from OVX Cx3cr1GFP/+ and Cx3cr1GFP/GFP mice. Scale bar = 100 µm. (E) Ex vivo FACS analysis of CD11c+ DCs and (F) CD11b+Ly6Chi MNs present in the BM of SHAM and OVX mice. Cells were gated as in Figure 1—figure supplement 2 and histograms are showing the absolute cell number of these populations in the bone marrow. **p<0.01; ***p<0.001; n.s., no significant difference.

Figure 1.

Figure 1—figure supplement 1. Analysis of i-OCLs progenitors in OVX mice.

Figure 1—figure supplement 1.

(A) Schematic representation of the differentiation of OCLs from blood CD11b+Ly6Chi monocytes from WT Cx3cr1GFP/+ mice (B) FACS analysis on these CD11b+ Ly6Chi MNs-derived OCLs using Hoechst 33342-A staining revealed that GFP+ (Cx3cr1+) OCLs differentiate efficiently from CD11b+ Ly6Chi MNs. OCLs were gated as cells having three nuclei and more (≥3N) as described previously (Madel et al., 2018).
Figure 1—figure supplement 2. Gating strategy for ex vivo FACS analysis.

Figure 1—figure supplement 2.

Gating strategy for ex vivo FACS analysis of bone marrow DCs and inflammatory monocytes in SHAM and OVX mice.
Open in a new tab

As Cx3cr1 is involved in the recruitment of OCL progenitors to the BM particularly under inflammatory conditions (Charles et al., 2012), we examined i-OCL progenitors in the BM of OVX Cx3cr1-deficient Cx3cr1GFP/GFP mice and WT Cx3cr1GFP/+ mice. While both DCs and Ly6Chi MNs have been involved in the formation of OCLs in pathological conditions, only DCs were shown to give rise to Cx3cr1+ i-OCLs (Ibáñez et al., 2016). Thus, we analyzed the osteoclastogenic capacity of blood Ly6Chi MNs sorted from Cx3cr1GFP/+ mice. About 35% of the resulting OCLs expressed GFP indicating that Ly6Chi MNs are also progenitors of Cx3cr1+ OCLs (Figure 1—figure supplement 1A–B). FACS analysis in OVX mice revealed that both DCs and Ly6Chi MNs increased in the BM of OVX Cx3cr1GFP/+ but not Cx3cr1GFP/GFP mice (Figure 1E–F, Figure 1—figure supplement 2). These results suggest that the reduced bone loss observed in Cx3cr1-deficient Cx3cr1GFP/GFP mice is, at least partly, related to a decrease in BM progenitors of Cx3cr1+ OCLs, namely DCs and Ly6Chi MNs.

Cx3cr1 deficiency does not alter immune and bone resorption function of Cx3cr1+ i-OCLs

These results suggest a central role of Cx3cr1 in i-OCL differentiation. To further investigate underlying mechanisms, we performed the first comparative RNA-seq approach on GFP+ mature i-OCLs from WT Cx3cr1GFP/+ and Cx3cr1-deficient Cx3cr1GFP/GFP mice (Figure 2A). OCLs represent a small population of cells firmly attached to the bone that cannot be isolated ex vivo in sufficient number for transcriptomic and functional assays (Madel et al., 2018). Thus, i-OCLs were differentiated from BM-derived CD11c+ DCs of WT Cx3cr1GFP/+ and Cx3cr1-deficient Cx3cr1GFP/GFP mice as described (Ibáñez et al., 2016). Mature Cx3cr1+ i-OCLs were then sorted based on their GFP expression and multinucleation according to Madel et al. (2018) (Figure 2—figure supplement 1).

Figure 2. Cx3cr1 deficiency does not affect resorption and immune function in Cx3cr1+ i-OCLs.

(A) Schematic representation of the differentiation of i-OCLs from BM-derived DCs of WT Cx3cr1GFP/+ and Cx3cr-deficient Cx3cr1GFP/GFP mice. (B) RNA-seq analysis on sorted GFP+ i-OCLs (representing Cx3cr1+ i-OCLs) differentiated from BM-derived DCs of WT Cx3cr1GFP/+ and Cx3cr-1deficient Cx3cr1GFP/GFP mice (gated as in Figure 2—figure supplement 1). Volcano-plot indicates –Log10 adjusted p value versus log2 fold-change comparing gene expression in i-OCL subsets. (C) Matrix dissolution activity of sorted GFP+ i-OCLs from WT Cx3cr1GFP/+ and Cx3cr1-deficient Cx3cr1GFP/GFP mice seeded at the same cell density on a calcified matrix was evidence by red alizarin staining of the mineralized matrix. Unstained areas correspond to the resorbed areas. Left panel: representative images of resorbed area. Scale bar = 100 µm. Right panel: quantification of resorbed areas presented as mean ± SD percentage of three independent biological replicates each in triplicates. (D) FACS analysis of fluorescent-OVA uptake among GFP+ i-OCLs from Cx3cr1GFP/+ and Cx3cr1GFP/GFP mice. Left panel: representative density plots and right panel: percentage of OVA+ cells from five independent experiments. (E) FACS analysis of GFP+ i-OCLs from Cx3cr1GFP/+ and Cx3cr1GFP/GFP mice in at least four independent experiments. (F) T cell proliferation assay on CD4+ T cells labelled with CFSE and cultured in the presence of OVA-challenged GFP+ i-OCLs differentiated from Cx3cr1GFP/+ and Cx3cr1GFP/GFP mice and analyzed by FACS after 4 days of coculture. (G) CD4+ T cells activated by OVA-challenged GFP+ i-OCLs from Cx3cr1GFP/+ and Cx3cr1GFP/GFP mice were analyzed for their expression of TNFα and IFNγ by FACS after intracytoplasmic staining of these cytokines. n.s., no significant difference.

Figure 2.

Figure 2—figure supplement 1. Gating strategy for mature Cx3cr1+ and Cx3cr1neg i-OCLs.

Figure 2—figure supplement 1.

(A) Gating strategy for the analysis and sorting of mature GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs using flow cytometry. After gating on total cells, doublets were excluded and cells with ≥3 nuclei (≥3N) cells were considered as OCLs and distinguished from 1 to 2 nucleated (1–2N) cells based on their Hoechst 33342 staining level. OCLs were analyzed based on their GFP expression and GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs were sorted.
Figure 2—figure supplement 2. Comparative transcriptomic analysis of Cx3cr1+ i-OCLs from Cx3cr1GFP/GFP and Cx3cr1GFP/+ mice.

Figure 2—figure supplement 2.

(A) Comparative RNA-seq analysis between GFP+ (Cx3cr1+) i-OCLs from KO Cx3cr1GFP/GFP and WT Cx3cr1GFP/+ mice. Heatmap visualization of the z-scored expression for selected genes involved in OCL differentiation, T cell stimulation and antigen presentation. (B) Relative mRNA expression of GFP+ (Cx3cr1+) i-OCLs from KO Cx3cr1GFP/GFP and WT Cx3cr1GFP/+ mice. Results are represented as the mean with 95% confidence interval of three independent biological replicates conducted in triplicates. n.s., no significant difference.
Open in a new tab

Except for their Cx3cr1 expression, no differences were detected between GFP+ i-OCLs from Cx3cr1-deficient Cx3cr1GFP/+ and WT Cx3cr1GFP/GFP mice suggesting that Cx3cr1 deficiency did not change the identity of i-OCLs (Figure 2B, Figure 2—figure supplement 2A–B). Similarly, functional in vitro assays revealed that regardless if they derived from Cx3cr1-deficient Cx3cr1GFP/+ or WT Cx3cr1GFP/GFP mice, GFP+ i-OCLs had equivalent resorption activity, antigen-presentation capacity or expression of MHC-II and co-stimulatory molecules (Figure 2C–E). In addition, we performed a T cell activation assay using carboxyfluorescein succinimidyl ester (CFSE)-labelled CD4+ T cells from OT-II mice bearing a T cell receptor (TCR) specific for the immunodominant ovalbumin (OVA) peptide (Barnden et al., 1998). No differences between GFP+ i-OCLs from Cx3cr1GFP/GFP versus Cx3cr1GFP/+ mice were found in their T cell activation and polarization capacity (Figure 2F–G). Thus, our results indicate that the function of the Cx3xr1 protein per se is not essential for the resorption and immune function of mature Cx3cr1-expressing i-OCLs.

Cx3cr1+ and Cx3cr1neg i-OCLs are distinct populations

Although Cx3cr1 is important for i-OCL progenitors, its deficiency does not affect bone resorption or T cell activation by mature Cx3cr1-expressing i-OCLs. However, being expressed in part of i-OCLs, Cx3cr1 remains a marker for the heterogeneity of these i-OCLs. We explored this heterogeneity and in particular the specific role of Cx3cr1-expressing and Cx3cr1neg i-OCL subsets, comparing them using an RNA-seq approach. Mature Cx3cr1+ and Cx3cr1neg i-OCLs were generated in vitro from BM-derived CD11c+ DCs of WT Cx3cr1GFP/+ mice and sorted based on their GFP expression and multinucleation (Figure 3A, Figure 2—figure supplement 1; Madel et al., 2018). A total of 1771 genes were significantly differentially expressed (adj.pVal <0.05, Log2FC ≥ 1) between the 2 populations. Principal Component Analysis (PCA) of all detectable genes revealed that Cx3cr1+ (GFP+) and Cx3cr1neg (GFPneg) i-OCLs were clustered into 2 distinct populations (Figure 3B), which was reinforced by volcano plot and hierarchical clustering analysis of the top 107 significantly differently expressed genes (adj.pVal <0.05 and Log2FC ≥ 1) (Figure 3C–D). In addition to Cx3cr1, genes interacting with its functional pathway were also differentially expressed (Figure 3—figure supplement 1A). In particular, Nr1d1, a negative regulator of Cx3cr1 (Song et al., 2018) was reduced whereas Tlr4 that mediates Cx3cr1 upregulation by LPS (Panek et al., 2015) was increased in Cx3cr1+ versus Cx3cr1neg i-OCLs (Figure 3—figure supplement 1B–C). These data suggest that Cx3cr1 expression is tightly controlled during i-OCL differentiation. This was further confirmed by showing that Cx3cr1+ OCLs could be generated from either Cx3cr1+ or Cx3cr1neg BM-derived DCs (Figure 3—figure supplement 1D). Thus, the expression of Cx3cr1 in i-OCLs is not only inherited from their progenitors but may be acquired through a dynamic up-regulation during the differentiation of a subset of i-OCLs having a specific function. In particular, genes and functional pathways differentially expressed between Cx3cr1+ and Cx3cr1neg i-OCLs (Figure 3D–E) strongly suggested that the function of Cx3cr1+ i-OCLs is related to the immune system. Therefore, we further characterized Cx3cr1+ and Cx3cr1neg i-OCLs focusing on their bone resorption capacity and their immune potential.

Figure 3. Transcriptomic profiling of Cx3cr1+ and Cx3cr1neg i-OCLs reveals two distinct populations of i-OCLs.

(A) Schematic representation of the differentiation of GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs from BM-derived DCs of WT Cx3cr1GFP/+ mice and their sorting (gated as in Figure 2—figure supplement 1). (B) Principal component analysis of Cx3cr1+ and Cx3cr1neg i-OCLs clusters samples in two groups (red: GFP+ (Cx3cr1+) OCLs and blue GFPneg (Cx3cr1neg) OCLs, both from Cx3cr1GFP/+ mice). Data shows the two first components on the top 500 most differentially expressed genes after batch correction. Each dot represents the expression profile of one sample. (C) Volcano-plot showing representative differentially expressed genes for Cx3cr1+ and Cx3cr1neg i-OCLs. Cut-off values were defined by a fold change (FC) >2 and an adjusted p-value<0.05. (D) Heatmap visualization of the top 107 genes significantly differentially expressed between GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs from Cx3cr1GFP/+ mice (adjusted p-value<0.05, FC ≥2). (E) Graphical representation of gene ontology analysis associated with differentially expressed genes between Cx3cr1+ and Cx3cr1neg i-OCLs.

Figure 3.

Figure 3—figure supplement 1. Cx3cr1+ and Cx3cr1neg i-OCLs differ in their expression of Cx3cr1-interacting genes.

Figure 3—figure supplement 1.

(A) Heatmap visualization of selected genes involved in the Cx3cr1 pathway that are significantly differentially expressed (p<0.05, FC ≥2) between GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs differentiated from WT Cx3cr1GFP/+ mice. (B) Relative expression levels of Nr1d1 and (C) Tlr4 in Cx3cr1+ and Cx3cr1neg i-OCLs. (D) BM-derived DCs from Cx3cr1GFP/+ mice were sorted based on their expression of GFP and were differentiated into i-OCLs. Both GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) DCs gave rise to approx. 20% of Cx3cr1+ i-OCLs. *p<0.05; ****p<0.0001.
Open in a new tab

Transcriptomic profiling revealed that Cx3cr1neg i-OCLs expressed higher levels of genes responsible for bone resorption (Figure 4A), which was confirmed by RT-qPCR (Figure 4B). We compared Cx3cr1neg and Cx3cr1+ i-OCLs subsets to CD169+CD68+ macrophages that also express Cx3cr1 (Figure 4—figure supplement 1A–B). Our results revealed that both Cx3cr1neg and Cx3cr1+ i-OCL subsets differed from BM-derived macrophages as they did not express CD68 and CD169 (Figure 4—figure supplement 1C–D). Moreover, the 2 i-OCLs subsets shared the main characteristics of OCLs, e.g. TRAcP expression and resorption activity (Figure 4C–F) while BM-macrophages did not express TRAcP (Figure 4—figure supplement 1E). However, when sorted mature Cx3cr1neg and Cx3cr1+ i-OCLs were seeded at the same cell concentration, Cx3cr1neg i-OCLs showed a higher matrix dissolution activity compared to Cx3cr1+ i-OCLs (Figure 4E–F). These results demonstrate that Cx3cr1neg and Cx3cr1+ subsets correspond to 2 subsets of bona fide OCLs that differ in their resorption activity.

Figure 4. Cx3cr1neg and Cx3cr1+ i-OCLs differ in their resorbing activity.

(A) Heatmap visualization of the z-scored expression for selected genes involved in bone resorption, osteoclast fusion and differentiation that are differentially expressed between GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs both differentiated from WT Cx3cr1GFP/+ mice (adjusted p-value<0.05, FC ≥2). (B) RT-qPCR analysis on GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs. Results are represented as the mean with 95% confidence interval of 3 independent biological replicates in triplicates. (C) Flow cytometry analysis of TRAcP expression using ELF95 substrate in the 2 i-OCLs subsets. Percentage of positive cells (dark curve) compared to the negative control (light curve) is indicated. (D) Quantification of the mean fluorescence intensity (MFI) of TRAcP in GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs (n = 3). (E) Representative images of matrix dissolution activity of sorted GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs seeded at the same cell density. Red alizarin staining of the mineralized matrix revealed the resorbed areas as unstained. Scale bar = 100 µm. (F) Quantification of resorbed areas from three independent experiments in triplicates are indicated as percentage. ****p<0.0001; n.s., no significant difference.

Figure 4.

Figure 4—figure supplement 1. Cx3cr1+ i-OCLs differ from Cx3cr1+ BM-derived macrophages.

Figure 4—figure supplement 1.

(A) Schematic representation of the differentiation of macrophages from the BM of Cx3cr1GFP/+ mice. (B) FACS analysis of macrophage markers on BM-derived macrophages and (C) on GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs (n = 3). (D) Quantification of CD68+CD169+ cells in Cx3cr1+ BM-derived macrophages and in Cx3cr1neg and Cx3cr1+ i-OCL subsets. (E) Representative images of TRAcP staining on i-OCLs and BM-derived macrophages showing that contrasting to i-OCLs that all express TRAcP, BM-derived macrophages are negative for TRAcP expression. Scale bar = 50 µm.
Open in a new tab

Besides these differences in bone resorption, RNA-seq data revealed significant differences in genes involved in antigen-processing and presentation (Figure 5A). Further in vitro analysis confirmed that whereas both i-OCL subsets engulfed OVA, this capacity was significantly decreased in Cx3cr1neg i-OCLs (Figure 5B–C). In line with the RNA-seq data, FACS analysis confirmed that MHC-II molecules were more expressed in Cx3cr1neg than Cx3cr1pos i-OCLs (Figure 5D), while no significant differences were observed in the expression of CD80 and CD86 (Figure 5E–F). Moreover, Cx3cr1neg i-OCLs showed higher capacity to induce T cell proliferation than Cx3cr1+ i-OCLs (Figure 5G). Nevertheless, both i-OCL subsets were able to induce TNFα and IFNγ-producing CD4+ T cells but not FoxP3+CD4+ regulatory T (Treg) cells (Figure 5—figure supplement 1), in accordance with the previously described immune properties of i-OCLs (Ibáñez et al., 2016).

Figure 5. Cx3cr1neg and Cx3cr1+ subsets differ in their antigen uptake and presentation.

(A) Heatmap visualization of the z-scored expression for selected genes involved in antigen uptake, processing and presentation that are differentially expressed between GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs from WT Cx3cr1GFP/+ mice (adjusted pVal <0.05, FC ≥2). (B) Representative FACS plots of OVA uptake in Cx3cr1+ and Cx3cr1neg i-OCLs. (C) Quantification of FACS analysis for OVA uptake in GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs (n = 5). (D–F) FACS analysis of GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs for (D) MHC-II, (E) CD80 and (F) CD86 (n = 6). (G). Representative FACS histograms for T cell proliferation assay of CFSE-labelled CD4+ T cells from OT-II mice cultured with OVA-loaded GFP+ (Cx3cr1+) or GFP neg (Cx3cr1neg) i-OCLs after 5 days of coculture. **p<0.01; n.s., no significant difference.

Figure 5.

Figure 5—figure supplement 1. CD4+ T cell polarization analysis CD4+ T cells from OT-II mice were cocultured with OVA-challenged GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs from WT Cx3cr1GFP/+ mice, respectively, and analyzed by intracytoplasmic staining for their (A) TNFα, (B) IFNγ and (C) FoxP3 expression using flow cytometry. n.s., no significant difference.

Figure 5—figure supplement 1.

Open in a new tab

Cx3cr1+ i-OCLs express immunosuppressive factors and control the immune function of Cx3cr1neg i-OCLs in vitro

RNA-seq data also revealed that both i-OCLs subsets expressed inflammation-associated genes in line with their capacity to induce TNFα and IFNγ-producing CD4+ T cells (Figure 6A). However, Cx3cr1+ i-OCLs had significant higher expression of immunosuppressive genes (Figure 6A) that could contribute to their lower inflammatory potential. Further RT-qPCR of the immunosuppressive genes CD274 (PD-L1), Lgals9 (Galectin-9) and Tnfrsf14 (HEVM) confirmed these findings (Figure 6B). Interestingly, blocking the PD-L1/PD-1 axis with an anti-PD-1 antibody in co-cultures of Cx3cr1+ i-OCLs and CD4+ T cells increased the capacity of Cx3cr1+ i-OCLs to induce T cell proliferation (Figure 6C). These findings reveal that PD-L1 plays a central role in the immunosuppressive potential of Cx3cr1+ i-OCLs. They also suggest that Cx3cr1+ i-OCLs could have an immunosuppressive effect on Cx3cr1neg i-OCLs. Therefore, we set up an immunosuppressive assay where OVA-loaded Cx3cr1neg i-OCLs were co-cultured with CFSE-labelled CD4+ T cells from OT-II mice. Different ratios of Cx3cr1+ i-OCLs not challenged with OVA (unable to directly activate OT-II T cells) were added to the culture (Figure 6D). Cx3cr1neg i-OCLs alone (ratio 1:0) highly stimulated CD4+ T cell proliferation, which was dramatically decreased by the addition of Cx3cr1+ i-OCLs (ratio 1:2; Figure 6E). These data demonstrate that Cx3cr1+ i-OCLs are able to control the inflammatory activity of Cx3cr1neg i-OCLs in vitro, suggesting the existence of a novel mechanism of interaction between OCLs in inflammatory conditions that remains to be addressed in vivo.

Figure 6. Cx3cr1neg and Cx3cr1+ subsets differ in their T cell activation capacity.

Figure 6.

Open in a new tab

(A) Heatmap visualization of the z-scored expression for selected genes involved in T cell stimulation and inhibition that are differentially expressed between GFP+ (Cx3cr1+) and GFPneg (Cx3cr1neg) i-OCLs from WT Cx3cr1GFP/+ mice (adjusted p<0.05, FC ≥2). (B) RT-qPCR analysis of immunosuppressive molecules. Graphs show three independent experiments conducted in triplicates. (C) Representative FACS histograms and quantification of the proliferation of CSFE-labelled CD4+ T cell from OT-II mice cocultured with OVA-loaded Cx3cr1+ i-OCLs from WT Cx3cr1GFP/+ mice and an isotype antibody (left panel) or an anti-PD-1 antibody (right panel). (D) Schematic representation of the experimental setup. Sorted GFPneg (Cx3cr1neg) i-OCLs from WT Cx3cr1GFP/+ mice were loaded with OVA for 3 hr and incubated with CFSE+ CD4+ T cells in the presence of different rations of non-OVA loaded GFP+(Cx3cr1+) i-OCLs from WT Cx3cr1GFP/+ mice. (E) FACS analysis and quantification of CFSE+ CD4+ T cells of OT-II mice cocultured in the presence of different ratios (1:0; 1:1; 1:2) between OVA-loaded Cx3cr1neg i-OCLs and Cx3cr1+ i-OCLs (non-loaded with OVA) for 5 days. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; n.s., no significant difference.

Discussion

This study provides further evidence of OCL heterogeneity. We previously demonstrated the existence of t-OCLs priming CD4+ Treg cells and i-OCLs inducing TNFα+CD4+ T cells that emerge under different conditions (Ibáñez et al., 2016). Here, we show that i-OCLs encompass two distinct populations of bona fide OCLs that can be distinguished by their Cx3cr1 expression. Although these 2 i-OCL subsets prime TNFα-producing CD4+ T cells and resorb, they display different capacities for bone resorption and immunosuppression. Moreover, they do not activate Treg cells, and thus, they differ considerably from t-OCLs (Ibáñez et al., 2016). We also show that Cx3cr1+ i-OCLs control the inflammatory activity of Cx3cr1neg i-OCLs in vitro, suggesting that different OCL subsets could interact with each other to control their immune activity, an hypothesis that remains to be investigated in vivo (Figure 7).

Figure 7. Heterogeneity of osteoclasts and underlying molecular mechanisms.

Figure 7.

Open in a new tab

In steady state, BM progenitors differentiate into tolerogenic OCLs (t-OCLs) that are able to present antigens and induce CD4+ Treg cells. During inflammation, inflammatory MNs and DCs are recruited to the BM and differentiate into i-OCLs. Approx. 25% of these i-OCLs can be characterized by their expression of Cx3cr1 while the majority of i-OCLs does not express this marker. Cx3cr1neg i-OCLs show significantly higher bone resorption activity in vitro compared to Cx3cr1+ i-OCLs. Both i-OCL subsets act as antigen-presenting cells but differ in their T cell activation capacity that is higher for the Cx3cr1neg i-OCL subset. Both i-OCL subsets are able to induce TNFα-producing CD4+ T cells, however the Cx3cr1+ OCL subset express high levels of co-inhibitory molecules such as PD-L1 that reduce their capacity to activate T cells. Moreover, Cx3cr1+ i-OCLs have an immune suppressive effect on Cx3cr1 neg i-OCLs by reducing their T cell proliferation capacity in vitro.

Cx3cr1 is involved in the development and progression of chronic inflammatory diseases, such as colitis and arthritis (Kotani et al., 2013; Rossini et al., 2014; Tarrant et al., 2012) while Cx3cr1 deficiency reduces the severity of these diseases (Niess and Adler, 2010; Rossini et al., 2014; Tarrant et al., 2012). Our results extend the protective effect of Cx3cr1 deficiency to OVX-induced bone destruction. As in inflammatory bowel disease, this model is driven by TNFα and RANK-L-producing CD4+ T cells leading to the development of Cx3cr1+ OCLs (Cenci et al., 2000; Ibáñez et al., 2016; Weitzmann and Pacifici, 2006). As a chemokine receptor, Cx3cr1 is required for the migration of monocytic cells and was suggested to participate in the maintenance of OCL progenitors particularly in inflammatory conditions (Charles et al., 2012; Han et al., 2014; Hasegawa et al., 2019). Its ligand (Fractalkine/Cx3cl1) increases in OVX mice and osteoporotic patients and correlates with disease severity (Chen et al., 2016; Wang et al., 2017). Fractalkine-producing mesenchymal cells attract OCL precursors and support their differentiation (Goto et al., 2016; Matsuura et al., 2017). In agreement with these data, we demonstrate that the reduced bone loss in OVX Cx3cr1-deficient mice is related to a decrease in BM i-OCL precursors (DCs and Ly6Chi MNs) and a reduced osteoclastogenesis. Although Cx3cr1-deficient mice were reported to have slightly reduced bone density (Hoshino et al., 2013), we only observed moderate but significant differences in OVX mice, highlighting the importance of Cx3cr1 in pathological bone destruction. Compared to previous reports (Hoshino et al., 2013), we gained advantage from working on purified mature i-OCLs from Cx3cr1-deficient mice and we did not observe any significant variations when comparing Cx3cr1+ i-OCLs from Cx3cr1-deficient Cx3cr1GFP/GFP and WT Cx3cr1GFP/+ mice. Overall, our results demonstrate that the Cx3cr1/Cx3cl1 axis is essential for the increased BM recruitment and differentiation of i-OCL progenitors observed in OVX mice but that the Cx3cr1 protein per se is not involved in the resorption and immune function of fully mature i-OCLs.

This raises the question of the role of Cx3cr1 in a subset of mature i-OCLs. As OCLs permanently fuse with new monocytic cells (Jacome-Galarza et al., 2019), Cx3cr1 could be considered as a marker for some of these progenitor cells that remains expressed in the resulting OCLs. However, Cx3cr1+ OCLs are induced by inflammatory cytokines such as IL-17 and TNFα and arise in conditions associated with inflammatory bone destruction (Ibáñez et al., 2016; Madel et al., 2019). Furthermore, Cx3cr1+ and Cx3cr1neg i-OCLs significantly differ in the expression of genes involved in Cx3cr1-interacting pathways. For instance, Cx3cr1+ OCLs show lower expression of Nr1d1, a Cx3cr1 inhibitor (Song et al., 2018) and higher levels of Tlr4 that mediates the positive effects of LPS on Cx3cr1 (Panek et al., 2015). Furthermore, we observed that Cx3cr1+ OCLs arise from both Cx3cr1+ and Cx3cr1neg progenitors (Figure 3—figure supplement 1D), indicating that Cx3cr1 expression by some mature OCLs is inducible and not solely inherited from their progenitors.

In line with our observations in OCLs, Cx3cr1 is also a marker for MN and DC diversity (Ginhoux et al., 2009; Varol et al., 2009; Yona et al., 2013). Depending on their environment, BM progenitors can differentiate into immunogenic Cx3cr1low/negMHC-IIhi DCs or Cx3cr1+MHC-IIlow macrophages with high antigen-uptake and a PD-L1-associated immunosuppressive function (Zhang et al., 2012). In the gut mucosa, Cx3cr1hi macrophages efficiently take up antigens but induce lower T cell proliferation compared to CD103+Cx3cr1neg DCs (Schulz et al., 2009). The balance between immunosuppressive Cx3cr1hi and inflammatory Cx3cr1low phagocytes in the gut is essential in controlling inflammatory responses in colitis (Regoli et al., 2017; Yin et al., 2012). Accordingly, Cx3cr1+ and Cx3cr1neg i-OCLs exert different immune functions. Cx3cr1+ cells were more efficient in antigen uptake but less potent in T cell activation than Cx3cr1neg i-OCLs. In addition, as Cx3cr1hi macrophages, Cx3cr1+ i-OCLs decrease T cell proliferation, notably via immunosuppressive molecules such as PD-L1, Galectin-9 and HVEM. These factors are major immune checkpoints involved in autoimmune disorders and tumor-induced immune suppression associated with T cell dysfunction, exhaustion, and low activation (Bjordahl et al., 2013; Cai and Freeman, 2009; Francisco et al., 2009; Tai et al., 2018). Thus, their high expression in Cx3cr1+ i-OCLs further strengthens the hypothesis that Cx3cr1+ i-OCLs act as immunosuppressive cells emerging upon inflammatory signals to regulate inflammation, which was confirmed in vitro by a functional immunosuppressive assay (Figure 6).

The contribution of OCLs to an immunosuppressive BM microenvironment has been recently explored in the context of multiple myeloma (MM). Bone destruction in MM is supported by malignant plasma cells that favor OCL differentiation, including from DCs, through high RANK-L, IL-17 and TNFα production (Mansour et al., 2017; Noonan et al., 2010; Tucci et al., 2011). In these conditions, OCLs suppress T cell activation via PD-L1, Galectin-9 and HVEM (An et al., 2016; Mansour et al., 2017; Tai et al., 2018). Although OCL heterogeneity was not investigated in MM, osteoclastogenic conditions are similar to those supporting the differentiation of Cx3cr1+ i-OCLs (Ibáñez et al., 2016). Thus, our results provide new insights into the immunosuppressive property of OCLs and suggest that Cx3cr1+ i-OCLs are likely to be involved in pathological conditions such as MM.

Although Cx3cr1+ and Cx3cr1neg i-OCLs both induce TNFα-producing CD4+ T cells and are resorbing cells, we showed that mature Cx3cr1neg i-OCLs are more efficient to induce CD4+ T cell proliferation and have a higher resorption capacity in vitro than Cx3cr1+ i-OCLs. This suggests that Cx3cr1neg i-OCLs play a major role in inflammatory bone destruction such as in osteoporosis and chronic inflammatory diseases. Interestingly, when comparing Cx3cr1+ i-OCLs from Cx3cr1-deficient Cx3cr1GFP/GFP and WT Cx3cr1GFP/+ mice no differences were observed. Thus, the Cx3cr1 protein as such is not essential for the immunomodulatory and bone resorbing function of Cx3cr1+ i-OCLs. However, it remains a membrane marker allowing to distinguish between 2 i-OCLs subsets (namely Cx3cr1+ and Cx3cr1neg i-OCLs) exerting different immunomodulatory roles. Apart from the lack of Cx3cr1 expression and their immunogenic effect, there is currently no specific marker for the identification of Cx3cr1neg i-OCLs. Thus, an in-depth characterization of these cells is necessary to fully address their specific role. Interestingly, Cx3cr1neg i-OCLs induce high proportions of TNFα and IFNγ-producing CD4+ T cells in vitro. These cytokines are potent inducers of Cx3cr1 and PD-L1 (Bjordahl et al., 2013; McGrath and Najafian, 2012; Tsukamoto et al., 2019) suggesting that TNFα-producing T cells induced by Cx3cr1neg i-OCLs may participate in the regulation of Cx3cr1+ i-OCLs. This observation supports the existence of a regulatory loop between pro-inflammatory Cx3cr1neg i-OCLs and immunosuppressive Cx3cr1+ i-OCLs that negatively regulate Th1 cell activation to control inflammation and that remains to be confirmed in vivo. Therefore, our findings suggest that the balance between the 2 i-OCL subsets would be an important aspect of the immune regulation during inflammatory bone loss.

In summary, we describe 2 OCL subsets that arise during inflammation and that are characterized by their Cx3cr1 expression (Figure 7). While Cx3cr1neg i-OCLs mediate inflammatory bone destruction, Cx3cr1+ i-OCLs are more prone to immunosuppression. This might contribute to the regulation of inflammatory processes but also to maintain an immunosuppressive BM microenvironment. Therefore, our in vitro data suggest the existence of a negative feedback mechanism by a gatekeeper subset of Cx3cr1-expressing i-OCLs during inflammation. This strongly emphasizes the heterogeneity of i-OCLs and provides new insights in their inflammatory function. Our study uncovered a hitherto neglected OCL diversity that contributes to a deeper understanding of the balance between inflammation and immunosuppression in the BM. Interestingly, OCL-related reduction of T cell activation can be restored using anti-PD-1 antibodies. Thus Cx3cr1+ i-OCLs might represent a novel promising target to overcome the OCL-mediated immunosuppression and restore an anti-tumorigenic BM microenvironment. On the other hand, our results suggest that Cx3cr1neg i-OCLs are promising targets against inflammatory bone destruction. A deeper comprehension of OCL diversity is therefore indispensable to identify more specific markers and to evaluate the therapeutic interest of targeting specific OCL subsets.

Materials and methods

Key resources table.

Reagent type
(species) or
resource		 Designation Source or
reference		 Identifiers Additional
information
Strain, strain background (Mus musculus) Cx3cr1GFP/+ Jung et al., 2000 CDTA, CNRS, Orléans, France
Sequenced-based reagent Calcr PCR primers CTTCCATGCTGATCTTCTGG andCAGATCTCCATTGGGCACAA
Sequenced-based reagent Acp5 PCR primers TGCCTACCTGTGTGGACATGA andCACATAGCCCACACCGTTCTC
Sequenced-based reagent Mmp9 PCR primers TGAGTCCGGCAGACAATCCT and CGCCCTGGATCTCAGCAATA
Sequenced-based reagent Ctsk PCR primers CAGCAGAGGTGTGTACTATG and GCGTTGTTCTTATTCCGAGC
Sequenced-based reagent Atp6V0a3/Tcirg1 PCR primers CGCTGCGAGGAACTGGAG and AGCGTCAGACCTGCCCG
Sequenced-based reagent Tnfrsf11a PCR primers CTTGGACACCTGGAATGAAGAAG and AGGGCCTTGCCTGCATC
Sequenced-based reagent Nfatc1 PCR primers TGAGGCTGGTCTTCCGAGTT and CGCTGGGAACACTCGATAGG
Sequenced-based reagent Dcstamp PCR primers GGGCACCAGTATTTTCCTGA and CAGAACGGCCAGAAGAATGA
Sequenced-based reagent Oscar PCR primers GTAACGGATCAGCTCCCCAG and CGCGGTACAGTGCAAAACTC
Sequenced-based reagent CD274 PCR primers CAAGCGAATCACGCTGAAAG and GGGTTGGTGGTCACTGTTTGT
Sequenced-based reagent Lgals9 PCR primers TCAAGGTGATGGTGAACAAGAAA and GATGGTGTCCACGAGGTGGTA
Sequenced-based reagent Lightr/Tnfrsf14 PCR primers TGTCCATTCTTTTGCCACTTG and CCTGATGGTGTTCTCCTGTTGTT
Sequenced-based reagent 36B4 PCR primers TCCAGGCTTTGGGCATCA and CTTTATCAGCTGCACATCACTCAGA
Antibody Monoclonal anti-mouse CD80 BD Biosciences clone 16-10A1 APC-conjugated
(1:100)
Antibody Monoclonal anti-mouse CD86 BD Biosciences clone GL1 PE-conjugated
(1:100)
Antibody Monoclonal anti-mouse I-A[b] BD Biosciences clone AF6-120.1 Biotin-conjugated
(1:200)
Antibody Monoclonal anti-mouse/human CD11b BD Biosciences clone M1/70 PE-conjugated
(1:500)
Antibody Monoclonal anti-mouse CD4 BD Biosciences clone RM4-5 PECy7-conjugated
(1:1000)
Antibody Monoclonal anti-mouse CD68 ThermoFisher Scientific clone FA-11 PECy7-conjugated
(1:200)
Antibody Monoclonal anti-mouse CD11c ThermoFisher Scientific clone N418 PECy7-conjugated
(1:200)
Antibody Monoclonal anti-mouse Ly6C ThermoFisher Scientific clone HK1.4 PerCP-conjugated
(1:1000)
Antibody Monoclonal anti-mouse CD169 ThermoFisher Scientific clone SER-4 APC-conjugated
(1:100)
Antibody Monoclonal anti-mouse IFNγ ThermoFisher Scientific clone XMG1.2 APC-conjugated
(1:400)
Antibody Monoclonal anti-mouse TNFα ThermoFisher Scientific clone MP6-XT22 BV421-conjugated
(1:400)
Chemical compound, drug streptavidin ThermoFisher Scientific PercP-conjugated
(1:200)
Chemical compound, drug ovalbumin ThermoFisher Scientific Alexa Fluor 647-conjugated
Commercial assay or kit Acid Phosphatase Leucocyte (TRAcP) kit Sigma-Aldrich 387A
Commercial assay or kit ELF 97 Endogenous Phosphatase Detection Kit Molecular probes E-6601
Open in a new tab

Mice and ovariectomy-induced osteoporosis

OT-II mice were purchased from Charles River Laboratory and C57BL/6 Cx3cr1GFP/+ mice were kindly provided by F. Laurent (INRA, Nouzilly, France). Animals were maintained under a 12 hr light/12 hr dark cycle with free access to water and standard mouse diet. For ovariectomy, female mice were randomly divided into 2 groups, which were then sham-operated (SHAM) or ovariectomized (OVX). Ovariectomy and sham surgery were performed on 6 weeks old Cx3cr1GFP/+ and Cx3cr1GFP/GFP female mice after anesthesia with isoflurane. The animals were intensively monitored for 72 hr after the procedure. Animals were weighed at the beginning of the study and twice a week thereafter. Six weeks after ovariectomy, mice were sacrificed. Uteri were weighed to control the quality of ovariectomy (not shown). Bones were used either for bone marrow cell isolation and further flow cytometry analysis or for subsequent µCT and histological analysis. Approval for animal experiments was obtained from the Institutional Ethics Committee on Laboratory Animals (CIEPAL-Azur, Nice Sophia-Antipolis, France) and experiments were conducted in compliance with ethical regulations for animal testing and research.

Bone analyses

Femora were fixed in 4% paraformaldehyde (PFA) overnight. Subsequently, microcomputed tomography (µCT) was performed using the Skyscan 1176 µCT system (Bruker µCT, Belgium) at the preclinical platform ECELLFRANCE (IRMB, Montpellier, France). Scans were performed using isotropic voxels size of 18 µm, voltage of 50 kV, current of 500 mA, 0.5 mm aluminum filter, 180 degrees with a 0.7-degree rotation step and 210 ms exposure time. Data 3D reconstructions were generated for visual representation using NRecon software (Bruker µCT, Belgium).

For histological analysis, fixed femora were decalcified in 10% EDTA for 10 days, embedded in paraffin and tartrate-resistant acid phosphatase (TRAcP) staining was performed on 7 µm sections following manufacturer’s recommendations (Sigma).

Primary osteoclast culture

As OCLs cannot be isolated directly ex vivo in sufficient number for subsequent functional analysis, inflammatory OCLs (i-OCLs) were differentiated in vitro from BM-derived CD11c+ DCs or from blood CD11b+ Ly6Chi MNs of 6 week old WT Cx3cr1GFP/+ and deficient Cx3cr1GFP/GFP mice as previously described (Ibáñez et al., 2016). Briefly, bone marrow was flushed out of the long bones and after red blood cell lysis, BM cells were cultured in 24-well plates at 0.5 × 106 cells/well in RPMI medium (ThermoFisher Scientific) supplemented with 5% serum (Hyclone, GE Healthcare), 1% penicillin-streptomycin, 50 µM β-mercaptoethanol (both from ThermoFisher Scientific), 10 ng/ml IL-4 and 10 ng/ml GM-CSF (both from PeproTech). CD11c+ DCs were sorted using anti-CD11c antibody (1:200; HL3; BD Biosciences) and anti-biotin microbeads (Miltenyi Biotec). Alternatively, when indicated, i-OCLs were obtained from blood SSClo CD11b+ Ly6Chi MNs of Cx3cr1GFP/+ OVX mice sorted on a FACS-Aria II (BD Biosciences) using anti-CD11b and Ly6C antibodies. For i-OCLs culture, 2 × 104 CD11c+ DCs or CD11b+ Ly6Chi MNs were seeded/well on 24-well plates in MEM-alpha (ThermoFisher Scientific) supplemented with 5% serum (Hyclone, GE Healthcare), 1% penicillin-streptomycin, 50 µM β-mercaptoethanol, 25 ng/ml M-CSF and 30 ng/ml RANK-L (both R and D). Cells were cultured at 37°C and 5% CO2. Medium was changed every 3–4 days. Fully differentiated i-OCLs were stained with TRAcP (Sigma-Aldrich) according to manufacturer’s instructions or collected from the culture plates using Accutase (Sigma-Aldrich) and used for subsequent FACS analysis or FACS sorting based on their nuclei number using H33342 (Sigma-Aldrich) as described previously (Madel et al., 2018). When indicated, macrophages were differentiated in vitro from total BM cells (0.5 × 106 BM cells/well in 24-well plates) in MEM-alpha supplemented with 5% serum (Hyclone, GE Healthcare), 1% penicillin-streptomycin, 50 µM β-mercaptoethanol and 25 ng/ml M-CSF for a total of 6 days.

FACS and cell sorting on osteoclasts

Mature in vitro differentiated i-OCLs were analyzed and sorted based on their multinucleation and GFP expression as described previously (Madel et al., 2018) using flow cytometry analysis. Briefly, cells were incubated with 5 µg/ml H33342 (Sigma-Aldrich) in PBS 1X supplemented with 1% FBS and 2 mM EDTA (PSE) for 30 min at 37°C. For cell sorting, cells were filtered through a 100 µm nylon mesh and sorted on a FACS Aria IIu (BD Bioscience) using a 100 µm nozzle at a flow rate of 2000 events/s. Sorted cells were collected in FBS. Hoechst-stained i-OCLs from Cx3cr1GFP mutant mice were labeled with antibodies specific for murine anti-CD80, CD86 and IAb (MHC-II, biotinylated) antibodies with a PercP-conjugated streptavidin in PSE for 15 min on ice. For antigen-uptake, mature OCLs were collected and incubated in the presence of 25 µg/ml ovalbumin (ThermoFisher Scientific) for 2.5 hr at 37°C before adding 5 µg/ml H33342 staining for another 30 min and subsequent FACS acquisition. Gating strategies for Cx3cr1+ and Cx3cr1neg i-OCLs are shown in Figure 2—figure supplement 1. For FACS analysis of TRAcP expression, OCLs were stained using the ELF97 substrate according to the manufacturer's protocol (endogenous phosphatase detection kit; Molecular Probes).

For ex vivo FACS analysis of OCL progenitors, after red blood cell lysis (Sigma-Aldrich) BM cells were labelled with anti-CD11c, CD11b and Ly6C antibodies. For FACS analysis of macrophages, BM-derived macrophages were labeled with anti-CD68 and CD169 antibodies.

For intracellular cytokine analysis, CD4+ T cells were stimulated with PMA, ionomycin and Brefeldin A, labelled with an anti-CD4 antibody and fixed with 4% formaldehyde overnight as described (Ciucci et al., 2015). Cells were stained with anti-IFNγ and TNFα antibodies in Saponin 1X. Data were acquired using a FACS Canto-II (BD Biosciences).

Resorption assay

Mature in vitro differentiated i-OCLs were sorted based on their multinucleation and GFP (Cx3cr1) expression as described above (Figure 2—figure supplement 1; Madel et al., 2018). A total of 103 Cx3cr1+ or Cx3cr1neg i-OCLs was seeded/well on a 96-well osteoassay plate (Corning) in MEM-alpha containing 10% FBS (Hyclone, GE Healthcare) and 30 ng/ml RANK-L (R and D). After 48 hr, matrix dissolution activity was evaluated by removing the cells using distilled sterile water and staining with 2% Alizarin Red (Sigma-Aldrich) for 1 min. Imaging was performed using an Axio Observer D1 microscope (Zeiss) and images were taken using AxioVision Rel. 4.8 software (Zeiss). Resorbed areas were quantified using Fiji/ImageJ software (Schindelin et al., 2012).

T cell proliferation and immune suppression assay

Mature in vitro differentiated Cx3cr1+ and Cx3cr1neg i-OCLs were sorted based on their multinucleation and GFP expression as mentioned above (Figure 2—figure supplement 1) and seeded on 96-well plates with 103 cells/well in the presence of 600 nM ovalbumin (OVA) peptide (ThermoFisher Scientific). Spleens of 8 weeks old OT-II mice were used to isolate CD4+ T cells that express transgenic OVA-specific αβ-TCRs by using the Dynabeads Untouched Mouse CD4 Cell Isolation Kit (ThermoFisher Scientific) according to manufacturer’s specifications. To analyze T cell proliferation, CD4+ T cells were labelled with 250 µM CFSE for 10 min at 37°C. OVA-loaded i-OCLs and CD4+ T cells were seeded in a ratio 1:20 in IMDM medium supplemented with 5% FCS (Hyclone, GE Healthcare), 1% penicillin-streptomycin (Gibco) and 50 µM β-mercaptoethanol (Gibco). T cell proliferation was assessed by FACS after 3–5 days.

Implication of the PD-1/PD-L1 axis was evaluated by adding 5 µg/ml isotype control or anti-PD-1 (clone Rpm 14–1; kindly provided by V. Vouret (IRCAN, Nice, France)) antibodies at the beginning to the coculture. To validate the immune suppressive effect of Cx3cr1+ i-OCLs, sorted Cx3cr1neg i-OCLs were stimulated with 600 nM OVA for 3 hr at 37°C, washed 3 times with IMDM medium and cultured together with CFSE-labelled CD4+ T cells from transgenic OT-II mice on 96-well plates as described above. Cx3cr1+ i-OCLs were sorted and added to the coculture without ovalbumin challenge in different ratios between Cx3cr1neg vs Cx3cr1+ i-OCLs (1:1; 1:2; 2:1). T cell proliferation was assessed using FACS after 3–5 days of coculture.

RNA sequencing on sorted osteoclasts

Total RNA (100 ng) from 4 biological replicates (from 4 different mice) in each group was extracted from in vitro differentiated i-OCLs (after sorting according to their multinucleation and GFP expression as described above and shown in Figure 2—figure supplement 1) with the RNeasy kit (Qiagen) and processed for directional library preparation using the Truseq stranded total RNA library kit (Illumina). Libraries were pooled and sequenced paired-ended for 2 × 75 cycles on a Nextseq500 sequencer (Illumina) to generate 30–40 million fragments per sample. After quality controls, data analysis was performed with 2 different approaches. For the first one, reads were ‘quasi’ mapped on the reference mouse transcriptome (Gencode vM15) and quantified using the SALMON software with the mapping mode and standard settings (Patro et al., 2017). Estimates of transcripts counts and their confidence intervals were computed using 1000 bootstraps to assess technical variance. Gene expression levels were computed by aggregating the transcript counts for each gene. Gene expression in biological replicates (n = 4) was then compared using a linear model as implemented in Sleuth (Pimentel et al., 2017) and a false discovery rate of 0.01. Lists of differentially expressed genes were annotated using Innate-DB and EnrichR web portals. For the second approach, raw RNAseq fastq reads were trimmed with Trimmomatic and aligned to the reference mouse transcriptome (Gencode mm10) using STAR (v. 2.6.1 c) (Dobin et al., 2013) on the National Institutes of Health high-performance computing Biowulf cluster. Gene-assignment and estimates counts of RNA reads were performed with HTseq (Anders et al., 2015). Further analyses were performed with R software and gene expression in biological replicates (n = 4) was compared between the different conditions to identify differentially expressed genes using DESeq2 (Love et al., 2014) with the Wald test (FDR < 0.01). Batch removal was performed using limma (Ritchie et al., 2015). Gene ontology (GO) pathway analyzes were performed using Goseq (Young et al., 2010). Both approaches gave equivalent results (not shown).

Gene expression analyses

Total RNA of sorted in vitro differentiated Cx3cr1+ and Cx3cr1neg i-OCLs was extracted using TRIzol reagent with subsequent isopropanol precipitation following manufacturer’s instructions. RT-qPCR was performed after reverse transcription (Superscript II, Life Technologies) as described previously (Mansour et al., 2012) using SYBR Green and the primers indicated in the Key ressources table. Samples of 3 biological replicates were run in triplicates and results were normalized to the reference gene 36B4 using the 2−ΔCt method as described (Mansour et al., 2011).

Statistical analyses

All data were analyzed using Graph Pad Prism 7.0 software using an appropriate two tailed student’s t-test with Bonferroni adjustment when comparing two groups. When more than two groups were compared two-way analysis of variance (ANOVA) was used. Statistical significance was considered at p<0.05. Experimental values are presented as mean ± standard deviation (SD) of at least three biological replicas. Error bars for gene expression analysis of humans and mice using RT-qPCR show the mean value with 95% confidence interval. All experiments were repeated with a minimum of three biological replicates and at least two technical replicates.

Acknowledgements

We acknowledge M Topi (LP2M, Nice, France) for her technical assistance, the Genomic Facility of the UFR Simone Veil, (Université Versailles-Saint-Quentin, France) for the RNA sequencing, the IRCAN animal core facility (Nice, France) and the preclinical platform of ECELLFRANCE for µCT analysis (IRMB, Montpellier, France). This work utilized the computational resources of the NIH-HPC-Biowulf cluster (http://hpc.nih.gov).

The work was supported by the Agence Nationale de la Recherche (ANR-16-CE14-0030), the French government, managed by the ANR as part of the Investissement d’Avenir UCAJEDI project (ANR-15-IDEX-01), the Fondation Arthritis, Société Française de Biologie des Tissus Minéralisés, European Calcified Tissue Society and American Society of Bone and Mineral Research. M-B M is supported by the Fondation pour la Recherche Médicale (FRM, ECO20160736019) and TC is supported by the Intramural Research Program of the National Cancer Institute, Center for Cancer Research, National Institutes of Health.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Claudine Blin-Wakkach, Email: claudine.BLIN@unice.fr.

Gordana Vunjak-Novakovic, Columbia University, United States.

Kathryn Song Eng Cheah, The University of Hong Kong, Hong Kong.

Funding Information

This paper was supported by the following grants:

  • Agence Nationale de la Recherche ANR-16-CE14-0030 to Florence Apparailly, Henri-Jean Garchon, Claudine Blin-Wakkach.

  • Agence Nationale de la Recherche ANR-15-IDEX-01 to Maria-Bernadette Madel, Claudine Blin-Wakkach.

  • Fondation pour la Recherche Médicale ECO20160736019 to Maria-Bernadette Madel.

  • Fondation Arthritis to Maria-Bernadette Madel, Claudine Blin-Wakkach.

  • Société Française de Biologie des TissusMinéralisés to Maria-Bernadette Madel.

  • European Calcified Tissue Society to Maria-Bernadette Madel.

  • American Society of Bone and Mineral Research to Maria-Bernadette Madel.

  • National Cancer Institute Intramural Research Program to Thomas Ciucci.

Additional information

Competing interests

No competing interests declared.

Author contributions

Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing.

Formal analysis, Validation, Investigation, Methodology, Writing - review and editing.

Formal analysis, Visualization, Writing - review and editing.

Formal analysis, Investigation, Methodology.

Visualization, Writing - review and editing, contributed to the discussion of results and provided helpful advices throughout the study.

Formal analysis, Validation, Investigation.

Data curation, Validation, Investigation, Methodology.

Formal analysis, Writing - review and editing.

Supervision, Writing - review and editing, Contributed to the discussion of results and provided helpful advices throughout the study.

Formal analysis, Validation, Visualization, Methodology, Writing - review and editing.

Conceptualization, Supervision, Visualization, Methodology, Writing - review and editing.

Conceptualization, Formal analysis, Supervision, Funding acquisition, Validation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing.

Ethics

Animal experimentation: The protocol for animal experiments was approved by the Institutional Ethics Committee on Laboratory Animals (CIEPAL-Azur, France) (#2016121216457153 (V3). Experiments were conducted in compliance with ethical regulations for animal testing and research.

Additional files

Source data 1. Raw counts-HTSeq.

Raw counts (HTseq) per gene and per sample from the RNAseq analysis of Cx3cr1+ and Cx3cr1neg i-OCLs from WT Cx3cr1GFP/+ mice and Cx3cr1+ i-OCLs from deficient Cx3cr1GFP/GFP mice. In each group, the 4 i-OCL samples have been obtained from 4 different mice.

elife-54493-data1.xlsx (3.3MB, xlsx)
Source data 2. Counts_Deseq NormaBatch Corrected.

Counts per gene and per sample after normalization and batch correction from the RNAseq analysis of Cx3cr1+ and Cx3cr1neg i-OCLs from WT Cx3cr1GFP/+ mice and Cx3cr1+ i-OCLs from deficient Cx3cr1GFP/GFP mice. In each group, the 4 i-OCL samples have been obtained from 4 different mice.

Source data 3. Cx3cr1+ i-OCLs from WT Cx3cr1GFP/+ mice versus Cx3cr1+ i-OCLs from deficient Cx3cr1GFP/GFP mice _Deseq.

Statistics (Deseq) for the comparison of expressed genes in Cx3cr1+ i-OCLs from WT Cx3cr1GFP/+ mice versus Cx3cr1+ i-OCLs from deficient Cx3cr1GFP/GFP mice. In each group, the 4 i-OCL samples have been obtained from 4 different mice.

elife-54493-data3.xlsx (1.2MB, xlsx)
Source data 4. Cx3cr1+ i-OCLs from WT Cx3cr1GFP/+ mice versus Cx3cr1neg i-OCLs from WT Cx3cr1GFP/+ mice mice _Deseq.

Statistics (Deseq) for the comparison of expressed genes in Cx3cr1+ versus Cx3cr1neg i-OCLs from WT Cx3cr1GFP/+ mice. In each group, the 4 i-OCL samples have been obtained from 4 different mice.

elife-54493-data4.xlsx (1.2MB, xlsx)
Transparent reporting form

Data availability

All the RNA sequencing data are included in the submitted manuscript as data source files. RNA-Sequencing data have been deposited in ENA (European Nucleotide Archive) under accession number PRJEB36092.

The following dataset was generated:

Madel MB, Garchon HJ, Wakkach A, Blin-Wakkach C. 2020. Heterogeneity of inflammatory osteoclasts based on CX3CR1 expression. European Nucleotide Archive. PRJEB36092

References

  1. Ammari M, Presumey J, Ponsolles C, Roussignol G, Roubert C, Escriou V, Toupet K, Mausset-Bonnefont AL, Cren M, Robin M, Georgel P, Nehmar R, Taams L, Grün J, Grützkau A, Häupl T, Pers YM, Jorgensen C, Duroux-Richard I, Courties G, Apparailly F. Delivery of miR-146a to Ly6C high monocytes inhibits pathogenic bone erosion in inflammatory arthritis. Theranostics. 2018;8:5972–5985. doi: 10.7150/thno.29313. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. An G, Acharya C, Feng X, Wen K, Zhong M, Zhang L, Munshi NC, Qiu L, Tai YT, Anderson KC. Osteoclasts promote immune suppressive microenvironment in multiple myeloma: therapeutic implication. Blood. 2016;128:1590–1603. doi: 10.1182/blood-2016-03-707547. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Anders S, Pyl PT, Huber W. HTSeq--a Python framework to work with high-throughput sequencing data. Bioinformatics. 2015;31:166–169. doi: 10.1093/bioinformatics/btu638. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Arai F, Miyamoto T, Ohneda O, Inada T, Sudo T, Brasel K, Miyata T, Anderson DM, Suda T. Commitment and differentiation of osteoclast precursor cells by the sequential expression of c-Fms and receptor activator of nuclear factor kappaB (RANK) receptors. The Journal of Experimental Medicine. 1999;190:1741–1754. doi: 10.1084/jem.190.12.1741. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Barnden MJ, Allison J, Heath WR, Carbone FR. Defective TCR expression in transgenic mice constructed using cDNA-based alpha- and beta-chain genes under the control of heterologous regulatory elements. Immunology and Cell Biology. 1998;76:34–40. doi: 10.1046/j.1440-1711.1998.00709.x. [DOI] [PubMed] [Google Scholar]
  6. Bjordahl RL, Steidl C, Gascoyne RD, Ware CF. Lymphotoxin network pathways shape the tumor microenvironment. Current Opinion in Immunology. 2013;25:222–229. doi: 10.1016/j.coi.2013.01.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Buchwald ZS, Kiesel JR, DiPaolo R, Pagadala MS, Aurora R. Osteoclast Activated FoxP3+ CD8+ T-Cells Suppress Bone Resorption in vitro. PLOS ONE. 2012;7:e38199. doi: 10.1371/journal.pone.0038199. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cai G, Freeman GJ. The CD160, BTLA, LIGHT/HVEM pathway: a bidirectional switch regulating T-cell activation. Immunological Reviews. 2009;229:244–258. doi: 10.1111/j.1600-065X.2009.00783.x. [DOI] [PubMed] [Google Scholar]
  9. Cenci S, Weitzmann MN, Roggia C, Namba N, Novack D, Woodring J, Pacifici R. Estrogen deficiency induces bone loss by enhancing T-cell production of TNF-alpha. Journal of Clinical Investigation. 2000;106:1229–1237. doi: 10.1172/JCI11066. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Charles JF, Hsu LY, Niemi EC, Weiss A, Aliprantis AO, Nakamura MC. Inflammatory arthritis increases mouse osteoclast precursors with myeloid suppressor function. Journal of Clinical Investigation. 2012;122:4592–4605. doi: 10.1172/JCI60920. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chen YD, Huang CY, Liu HY, Yao WF, Wu WG, Lu YL, Wang W. Serum CX3CL1/fractalkine concentrations are positively associated with disease severity in postmenopausal osteoporotic patients. British Journal of Biomedical Science. 2016;73:121–128. doi: 10.1080/09674845.2016.1209897. [DOI] [PubMed] [Google Scholar]
  12. Ciucci T, Ibáñez L, Boucoiran A, Birgy-Barelli E, Pène J, Abou-Ezzi G, Arab N, Rouleau M, Hébuterne X, Yssel H, Blin-Wakkach C, Wakkach A. Bone marrow Th17 tnfα cells induce osteoclast differentiation, and link bone destruction to IBD. Gut. 2015;64:1072–1081. doi: 10.1136/gutjnl-2014-306947. [DOI] [PubMed] [Google Scholar]
  13. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras TR. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29:15–21. doi: 10.1093/bioinformatics/bts635. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Francisco LM, Salinas VH, Brown KE, Vanguri VK, Freeman GJ, Kuchroo VK, Sharpe AH. PD-L1 regulates the development, maintenance, and function of induced regulatory T cells. The Journal of Experimental Medicine. 2009;206:3015–3029. doi: 10.1084/jem.20090847. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Ginhoux F, Liu K, Helft J, Bogunovic M, Greter M, Hashimoto D, Price J, Yin N, Bromberg J, Lira SA, Stanley ER, Nussenzweig M, Merad M. The origin and development of nonlymphoid tissue CD103+ DCs. The Journal of Experimental Medicine. 2009;206:3115–3130. doi: 10.1084/jem.20091756. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Goto Y, Aoyama M, Sekiya T, Kakita H, Waguri-Nagaya Y, Miyazawa K, Asai K, Goto S. CXCR4 + CD45 −cells are niche forming for osteoclastogenesis via the SDF-1, CXCL7, and CX3CL1 signaling pathways in bone marrow: cxcr4 + CD45 −cells in osteoclastogenesis. Stem Cells. 2016;34:2733–2743. doi: 10.1002/stem.2440. [DOI] [PubMed] [Google Scholar]
  17. Han KH, Ryu JW, Lim KE, Lee SH, Kim Y, Hwang CS, Choi JY, Han KO. Vascular expression of the chemokine CX3CL1 promotes osteoclast recruitment and exacerbates bone resorption in an irradiated murine model. Bone. 2014;61:91–101. doi: 10.1016/j.bone.2013.12.032. [DOI] [PubMed] [Google Scholar]
  18. Hasegawa T, Kikuta J, Sudo T, Matsuura Y, Matsui T, Simmons S, Ebina K, Hirao M, Okuzaki D, Yoshida Y, Hirao A, Kalinichenko VV, Yamaoka K, Takeuchi T, Ishii M. Identification of a novel arthritis-associated osteoclast precursor macrophage regulated by FoxM1. Nature Immunology. 2019;20:1631–1643. doi: 10.1038/s41590-019-0526-7. [DOI] [PubMed] [Google Scholar]
  19. Hoshino A, Ueha S, Hanada S, Imai T, Ito M, Yamamoto K, Matsushima K, Yamaguchi A, Iimura T. Roles of chemokine receptor CX3CR1 in maintaining murine bone homeostasis through the regulation of both osteoblasts and osteoclasts. Journal of Cell Science. 2013;126:1032–1045. doi: 10.1242/jcs.113910. [DOI] [PubMed] [Google Scholar]
  20. Ibáñez L, Abou-Ezzi G, Ciucci T, Amiot V, Belaïd N, Obino D, Mansour A, Rouleau M, Wakkach A, Blin-Wakkach C. Inflammatory osteoclasts prime TNFα-Producing CD4+ T Cells and Express CX3 CR1. Journal of Bone and Mineral Research. 2016;31:1899–1908. doi: 10.1002/jbmr.2868. [DOI] [PubMed] [Google Scholar]
  21. Jacome-Galarza CE, Lee S-K, Lorenzo JA, Aguila HL. Identification, characterization, and isolation of a common progenitor for osteoclasts, macrophages, and dendritic cells from murine bone marrow and periphery. Journal of Bone and Mineral Research. 2013;28:1203–1213. doi: 10.1002/jbmr.1822. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jacome-Galarza CE, Percin GI, Muller JT, Mass E, Lazarov T, Eitler J, Rauner M, Yadav VK, Crozet L, Bohm M, Loyher P-L, Karsenty G, Waskow C, Geissmann F. Developmental origin, functional maintenance and genetic rescue of osteoclasts. Nature. 2019;568:541–545. doi: 10.1038/s41586-019-1105-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Jung S, Aliberti J, Graemmel P, Sunshine MJ, Kreutzberg GW, Sher A, Littman DR. Analysis of fractalkine receptor CX(3)CR1 function by targeted deletion and green fluorescent protein reporter gene insertion. Molecular and Cellular Biology. 2000;20:4106–4114. doi: 10.1128/MCB.20.11.4106-4114.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kiesel JR, Buchwald ZS, Aurora R. Cross-presentation by osteoclasts induces FoxP3 in CD8+ T cells. Journal of Immunology. 2009;182:5477–5487. doi: 10.4049/jimmunol.0803897. [DOI] [PubMed] [Google Scholar]
  25. Koizumi K, Saitoh Y, Minami T, Takeno N, Tsuneyama K, Miyahara T, Nakayama T, Sakurai H, Takano Y, Nishimura M, Imai T, Yoshie O, Saiki I. Role of CX3CL1/fractalkine in osteoclast differentiation and bone resorption. The Journal of Immunology. 2009;183:7825–7831. doi: 10.4049/jimmunol.0803627. [DOI] [PubMed] [Google Scholar]
  26. Kotani M, Kikuta J, Klauschen F, Chino T, Kobayashi Y, Yasuda H, Tamai K, Miyawaki A, Kanagawa O, Tomura M, Ishii M. Systemic circulation and bone recruitment of osteoclast precursors tracked by using fluorescent imaging techniques. The Journal of Immunology. 2013;190:605–612. doi: 10.4049/jimmunol.1201345. [DOI] [PubMed] [Google Scholar]
  27. Li H, Hong S, Qian J, Zheng Y, Yang J, Yi Q. Cross talk between the bone and immune systems: osteoclasts function as antigen-presenting cells and activate CD4+ and CD8+ T cells. Blood. 2010;116:210–217. doi: 10.1182/blood-2009-11-255026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Li JY, Tawfeek H, Bedi B, Yang X, Adams J, Gao KY, Zayzafoon M, Weitzmann MN, Pacifici R. Ovariectomy disregulates osteoblast and osteoclast formation through the T-cell receptor CD40 ligand. PNAS. 2011;108:768–773. doi: 10.1073/pnas.1013492108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology. 2014;15:550. doi: 10.1186/s13059-014-0550-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Madel MB, Ibáñez L, Rouleau M, Wakkach A, Blin-Wakkach C. A novel reliable and efficient procedure for purification of mature osteoclasts allowing functional assays in mouse cells. Frontiers in Immunology. 2018;9:2567. doi: 10.3389/fimmu.2018.02567. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Madel M-B, Ibáñez L, Wakkach A, de Vries TJ, Teti A, Apparailly F, Blin-Wakkach C. Immune function and diversity of osteoclasts in normal and pathological conditions. Frontiers in Immunology. 2019;10:01408. doi: 10.3389/fimmu.2019.01408. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Mansour A, Anginot A, Mancini SJ, Schiff C, Carle GF, Wakkach A, Blin-Wakkach C. Osteoclast activity modulates B-cell development in the bone marrow. Cell Research. 2011;21:1102–1115. doi: 10.1038/cr.2011.21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Mansour A, Abou-Ezzi G, Sitnicka E, Jacobsen SE, Wakkach A, Blin-Wakkach C. Osteoclasts promote the formation of hematopoietic stem cell niches in the bone marrow. The Journal of Experimental Medicine. 2012;209:537–549. doi: 10.1084/jem.20110994. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Mansour A, Wakkach A, Blin-Wakkach C. Emerging roles of osteoclasts in the modulation of bone microenvironment and immune suppression in multiple myeloma. Frontiers in Immunology. 2017;8:954. doi: 10.3389/fimmu.2017.00954. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Matsuura T, Ichinose S, Akiyama M, Kasahara Y, Tachikawa N, Nakahama KI. Involvement of CX3CL1 in the migration of osteoclast precursors across osteoblast layer stimulated by Interleukin-1ß. Journal of Cellular Physiology. 2017;232:1739–1745. doi: 10.1002/jcp.25577. [DOI] [PubMed] [Google Scholar]
  36. McGrath MM, Najafian N. The role of coinhibitory signaling pathways in transplantation and tolerance. Frontiers in Immunology. 2012;3:47. doi: 10.3389/fimmu.2012.00047. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Niess JH, Adler G. Enteric flora expands gut Lamina propria CX3CR1+ dendritic cells supporting inflammatory immune responses under normal and inflammatory conditions. Journal of Immunology. 2010;184:2026–2037. doi: 10.4049/jimmunol.0901936. [DOI] [PubMed] [Google Scholar]
  38. Noonan K, Marchionni L, Anderson J, Pardoll D, Roodman GD, Borrello I. A novel role of IL-17-producing lymphocytes in mediating lytic bone disease in multiple myeloma. Blood. 2010;116:3554–3563. doi: 10.1182/blood-2010-05-283895. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Panek CA, Ramos MV, Mejias MP, Abrey-Recalde MJ, Fernandez-Brando RJ, Gori MS, Salamone GV, Palermo MS. Differential expression of the fractalkine chemokine receptor (CX3CR1) in human monocytes during differentiation. Cellular & Molecular Immunology. 2015;12:669–680. doi: 10.1038/cmi.2014.116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Patro R, Duggal G, Love MI, Irizarry RA, Kingsford C. Salmon provides fast and bias-aware quantification of transcript expression. Nature Methods. 2017;14:417–419. doi: 10.1038/nmeth.4197. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Pimentel H, Bray NL, Puente S, Melsted P, Pachter L. Differential analysis of RNA-seq incorporating quantification uncertainty. Nature Methods. 2017;14:687–690. doi: 10.1038/nmeth.4324. [DOI] [PubMed] [Google Scholar]
  42. Regoli M, Bertelli E, Gulisano M, Nicoletti C. The multifaceted personality of intestinal CX 3 CR1 + macrophages. Trends in Immunology. 2017;38:879–887. doi: 10.1016/j.it.2017.07.009. [DOI] [PubMed] [Google Scholar]
  43. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Research. 2015;43:e47. doi: 10.1093/nar/gkv007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Rivollier A, Mazzorana M, Tebib J, Piperno M, Aitsiselmi T, Rabourdin-Combe C, Jurdic P, Servet-Delprat C. Immature dendritic cell transdifferentiation into osteoclasts: a novel pathway sustained by the rheumatoid arthritis microenvironment. Blood. 2004;104:4029–4037. doi: 10.1182/blood-2004-01-0041. [DOI] [PubMed] [Google Scholar]
  45. Rossini V, Zhurina D, Radulovic K, Manta C, Walther P, Riedel CU, Niess JH. CX3CR1⁺ cells facilitate the activation of CD4 T cells in the colonic Lamina propria during antigen-driven colitis. Mucosal Immunology. 2014;7:533–548. doi: 10.1038/mi.2013.70. [DOI] [PubMed] [Google Scholar]
  46. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, Rueden C, Saalfeld S, Schmid B, Tinevez JY, White DJ, Hartenstein V, Eliceiri K, Tomancak P, Cardona A. Fiji: an open-source platform for biological-image analysis. Nature Methods. 2012;9:676–682. doi: 10.1038/nmeth.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Schulz O, Jaensson E, Persson EK, Liu X, Worbs T, Agace WW, Pabst O. Intestinal CD103+, but not CX3CR1+, antigen sampling cells migrate in lymph and serve classical dendritic cell functions. The Journal of Experimental Medicine. 2009;206:3101–3114. doi: 10.1084/jem.20091925. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Seeling M, Hillenhoff U, David JP, Schett G, Tuckermann J, Lux A, Nimmerjahn F. Inflammatory monocytes and fcγ receptor IV on osteoclasts are critical for bone destruction during inflammatory arthritis in mice. PNAS. 2013;110:10729–10734. doi: 10.1073/pnas.1301001110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Song C, Tan P, Zhang Z, Wu W, Dong Y, Zhao L, Liu H, Guan H, Li F. REV-ERB agonism suppresses osteoclastogenesis and prevents ovariectomy-induced bone loss partially via FABP4 upregulation. The FASEB Journal. 2018;32:3215–3228. doi: 10.1096/fj.201600825RRR. [DOI] [PubMed] [Google Scholar]
  50. Tai YT, Cho SF, Anderson KC. Osteoclast immunosuppressive effects in multiple myeloma: role of programmed cell death ligand 1. Frontiers in Immunology. 2018;9:1822. doi: 10.3389/fimmu.2018.01822. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Tarrant TK, Liu P, Rampersad RR, Esserman D, Rothlein LR, Timoshchenko RG, McGinnis MW, Fitzhugh DJ, Patel DD, Fong AM. Decreased Th17 and antigen-specific humoral responses in CX₃ CR1-deficient mice in the collagen-induced arthritis model. Arthritis & Rheumatism. 2012;64:1379–1387. doi: 10.1002/art.34320. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Tsukamoto M, Imai K, Ishimoto T, Komohara Y, Yamashita YI, Nakagawa S, Umezaki N, Yamao T, Kitano Y, Miyata T, Arima K, Okabe H, Baba Y, Chikamoto A, Ishiko T, Hirota M, Baba H. PD ‐L1 expression enhancement by infiltrating macrophage‐derived tumor necrosis factor‐α leads to poor pancreatic cancer prognosis. Cancer Science. 2019;110:310–320. doi: 10.1111/cas.13874. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Tucci M, Ciavarella S, Strippoli S, Brunetti O, Dammacco F, Silvestris F. Immature dendritic cells from patients with multiple myeloma are prone to osteoclast differentiation in vitro. Experimental Hematology. 2011;39:773–783. doi: 10.1016/j.exphem.2011.04.006. [DOI] [PubMed] [Google Scholar]
  54. Varol C, Vallon-Eberhard A, Elinav E, Aychek T, Shapira Y, Luche H, Fehling HJ, Hardt WD, Shakhar G, Jung S. Intestinal Lamina propria dendritic cell subsets have different origin and functions. Immunity. 2009;31:502–512. doi: 10.1016/j.immuni.2009.06.025. [DOI] [PubMed] [Google Scholar]
  55. Wakkach A, Mansour A, Dacquin R, Coste E, Jurdic P, Carle GF, Blin-Wakkach C. Bone marrow microenvironment controls the in vivo differentiation of murine dendritic cells into osteoclasts. Blood. 2008;112:5074–5083. doi: 10.1182/blood-2008-01-132787. [DOI] [PubMed] [Google Scholar]
  56. Wang L, Hu Y-Q, Zhao Z-J, Zhang H-Y, Gao B, Lu W-G, Xu X-L, Lin X-S, Wang J-P, Jie Q, Luo Z-J, Yang L. Screening and validation of serum protein biomarkers for early postmenopausal osteoporosis diagnosis. Molecular Medicine Reports. 2017;16:8427–8433. doi: 10.3892/mmr.2017.7620. [DOI] [PubMed] [Google Scholar]
  57. Weitzmann MN, Pacifici R. Estrogen deficiency and bone loss: an inflammatory tale. Journal of Clinical Investigation. 2006;116:1186–1194. doi: 10.1172/JCI28550. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Yin N, Xu J, Ginhoux F, Randolph GJ, Merad M, Ding Y, Bromberg JS. Functional specialization of islet dendritic cell subsets. The Journal of Immunology. 2012;188:4921–4930. doi: 10.4049/jimmunol.1103725. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Yona S, Kim K-W, Wolf Y, Mildner A, Varol D, Breker M, Strauss-Ayali D, Viukov S, Guilliams M, Misharin A, Hume DA, Perlman H, Malissen B, Zelzer E, Jung S. Fate Mapping Reveals Origins and Dynamics of Monocytes and Tissue Macrophages under Homeostasis. Immunity. 2013;38:79–91. doi: 10.1016/j.immuni.2012.12.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Young MD, Wakefield MJ, Smyth GK, Oshlack A. Gene ontology analysis for RNA-seq: accounting for selection Bias. Genome Biology. 2010;11:R14. doi: 10.1186/gb-2010-11-2-r14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Zhang X, Yu S, Hoffmann K, Yu K, Förster R. Neonatal lymph node stromal cells drive myelodendritic lineage cells into a distinct population of CX3CR1+CD11b+F4/80+ regulatory macrophages in mice. Blood. 2012;119:3975–3986. doi: 10.1182/blood-2011-06-359315. [DOI] [PubMed] [Google Scholar]

Decision letter

Editor: Gordana Vunjak-Novakovic1

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The authors studied the heterogeneity of inflammatory osteoclasts (i-OCL) using Cx3cr1 as a marker of osteoclast heterogeneity in a mouse model of OVX-induced bone resorption. Cx3cr1-neg OCLs showed higher capacity for bone matrix dissolution, phagocytosed less OVA antigen, and had higher capacity to induce T cell proliferation in vitro, compared to Cx3cr1+ OCLs.

Decision letter after peer review:

Thank you for submitting your article "Dissecting the phenotypic and functional heterogeneity of inflammatory osteoclasts by the expression of CX3CR1" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Kathryn Cheah as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Essential revisions:

1) The conclusions rely on (1) gene expression profiling of sorted CX3CR1+ and CX3CR1- cells, and (2) in vitro culture followed by functional analysis of said sorted cells. However, for these in vitro experiments, the culture conditions employed prior to analysis are not clear. The Materials and methods section suggests that there was extensive culture and stimulation prior to analysis following sorting, but these conditions are not explicitly stated in the main text. If cells were extensively cultured with stimulating cytokines prior to functional analysis, then this would be expected to drastically change their functional profile compared to cells sorted directly from the in vivo environment. This would affect the results in Figures 2, 4, 5, and 6.

2) While Cx3cr1+ and Cx3cr1- cells have different effects on T cell proliferation, that doesn't necessarily mean that one regulates the activity of the other in vivo. To make this conclusion it would be necessary to deplete selectively one population in vivo and observe the effects on the other population.

3) Given that macrophages can fuse into multinucleated cells and that Cx3cr1 is a marker of macrophage heterogeneity, it has not been demonstrated that the cells were indeed OCLs and not macrophages. Evidence of the cell identity, OCL or macrophage should be provided.

4) The absence of Cx3cr1 in Cx3cr1 gfp/gfp mice resulted in reduced recruitment of dendritic cells and Ly6C-hi monocytes to the sites of bone resorption in OVX animals. This result would be expected because Cx3cr1 is well known to be involved in leukocyte trafficking. The authors concluded that this result shows that Ly6C-hi monocytes are progenitors of inflammatory OCLs. This conclusion is not supported by the results.

5) The authors conclude that Cx3cr1+ cells express higher levels of immunosuppressive genes compared to Cx3cr1- cells. Because the list of genes clearly includes many genes that are immunostimulatory (IL1b, CCL5, Nfkb1, Ccl2, Cd80, CD86), the conclusions about immunosuppressive gene expression profile in Cx3cr1- cells are not clearly supported.

6) In Figure 1 the authors claim that CXC3R1 positive osteoclasts are needed for the full bone resorbing activity of osteoclasts after ovariectomy. The claim is based on only minor differences and whether CXC3R1 positive osteoclasts were preset or not, the osteoporosis phenotype develops in an equally severe manner. Moreover, these data contradict a statement made later in the paper that CXC3R1 positive osteoclasts are not needed for bone resorption.

7) Data shown in Figure 2 (no difference in the bone resorption capacities of CX3CR1+ cells isolated from GFP/+ mice and CX3CR1- cells isolated from GFP/GFP mice) and those shown in Figure 4 (large differences in bone resorptive capacities of CX3CR1+ and CX3CR1-negative cells derived from dendricytes) are contradictory. This inconsistency calls into question the validity of using cells derived from the dendricytes in culture if they behave differently from the mature cells isolated from the genetically modified mice.

eLife. 2020 May 13;9:e54493. doi: 10.7554/eLife.54493.sa2

Author response

Essential revisions:

1) The conclusions rely on (1) gene expression profiling of sorted CX3CR1+ and CX3CR1- cells, and (2) in vitro culture followed by functional analysis of said sorted cells. However, for these in vitro experiments, the culture conditions employed prior to analysis are not clear. The Materials and methods section suggests that there was extensive culture and stimulation prior to analysis following sorting, but these conditions are not explicitly stated in the main text. If cells were extensively cultured with stimulating cytokines prior to functional analysis, then this would be expected to drastically change their functional profile compared to cells sorted directly from the in vivo environment. This would affect the results in Figures 2, 4, 5, and 6.

The reviewers raised an important point. We generated in vitro inflammatory OCLs in the presence of RANK-L and M-CSF, the 2 main osteoclastogenic factors, according to the gold standard protocol for in vitro osteoclastogenesis and starting from their known progenitors, BM-derived dendritic cells (Ibanez et al., 2016). Of course, we agree that investigating osteoclast function on ex-vivo isolated osteoclasts would have been the best approach. However, up to now, there are no publications investigating osteoclast function on murine ex vivo-isolated osteoclasts. These cells are rare and firmly attached to the bone and to date there's no procedure in the literature to isolate them in sufficient number to perform subsequent analysis. Consequently, functional assays on osteoclasts always use in vitro-generated cells and much of osteoclast biology described so far in the literature has been investigated in such conditions (Madel et al., 2018). Moreover, at the moment there is no marker to distinguish between inflammatory and tolerogenic Cx3cr1neg osteoclasts, which make it impossible to sort them ex vivo. Up to know, the only way to obtain these cells is to differentiate them from their respective known progenitors. Of course, we agree with the reviewers that this may affect their function, but to the same extend as in all studies investigating osteoclast activity and published so far and we don't have any other option.

To make the procedure of osteoclast preparation clearer as requested, we revised the main text and included more details in the Results and Materials and methods sections as well as in the figure legends. We also added in Figure 2 and Figure 3 a schematic representation of osteoclast preparation.

2) While Cx3cr1+ and Cx3cr1- cells have different effects on T cell proliferation, that doesn't necessarily mean that one regulates the activity of the other in vivo. To make this conclusion it would be necessary to deplete selectively one population in vivo and observe the effects on the other population.

The reviewers are perfectly right and demonstrating the interaction between the 2 i-OCLs subsets in vivo would require specific depletion in vivo of one of these subsets. We are currently investigating in more details the heterogeneity of the different OCL subsets in order to identify suitable factors that would allow to specifically target one or the other OCL subset. This study, and in particular the validation of markers, requires time and the data will be published in a future manuscript.

However, to address the reviewer's concern and to avoid any over-interpretation of our results, in the revised manuscript we softened our conclusion and clearly stated that the mechanisms we described are only shown in vitro and remain to be demonstrated in vivo.

3) Given that macrophages can fuse into multinucleated cells and that Cx3cr1 is a marker of macrophage heterogeneity, it has not been demonstrated that the cells were indeed OCLs and not macrophages. Evidence of the cell identity, OCL or macrophage should be provided.

To prove the identity of the cells we are analyzing, we performed additional experiments as requested. In particular, because in our protocol osteoclasts are generated in the presence of RANK-L but also M-CSF, a potent inducer of macrophages, we compared these osteoclasts to macrophages generated from bone marrow cells in the presence of M-CSF. In these conditions, we didn't observed cell fusion in macrophages. Moreover, our results showed that while BM-derived macrophages express the 2 BM macrophage markers CD68 and CD169 but are negative for TRAcP expression (Figure 4—figure supplement 1), both Cx3cr1+ and Cx3cr1neg subsets of OCLs do not express these macrophage markers but share osteoclast hallmarks, such as TRAcP expression and resorption capacity (Figure 4C-F). These data were included in the revised manuscript and confirm that the cells we are analyzing are bona fide OCLs, whether they express or not Cx3cr1.

4) The absence of Cx3cr1 in Cx3cr1 gfp/gfp mice resulted in reduced recruitment of dendritic cells and Ly6C-hi monocytes to the sites of bone resorption in OVX animals. This result would be expected because Cx3cr1 is well known to be involved in leukocyte trafficking. The authors concluded that this result shows that Ly6C-hi monocytes are progenitors of inflammatory OCLs. This conclusion is not supported by the results.

Ly6Chi monocytes are involved in osteoclastogenesis in particular in conditions related to inflammatory bone destruction (Seeling et al., 2013; Ammari et al., 2018; Madel et al., 2019), which suggest that they could be progenitors of Cx3cr1+ osteoclasts that are also associated to the same conditions (Ibanez et al., 2016). To address the capacity of Ly6Chi monocytes to differentiate into Cx3cr1+ osteoclasts, we sorted Ly6Chi monocytes from the blood of WT Cx3cr1GFP/+ mice and cultured them with M-CSF and RANKL for 6 days as shown in Figure 1—figure supplement 1A. We analyzed the resulting osteoclasts by FACS and found that about 35% of these osteoclasts are GFP/Cx3cr1+ (Figure 1—figure supplement 1B) demonstrating that Ly6Chi monocytes are indeed progenitors of Cx3cr1+ osteoclasts, which support our conclusion. To make this point more visible and clearer in the revised version of the manuscript, we modified the text in the Results section and added a schematic representation of the experiment (Figure 1—figure supplement 1A).

5) The authors conclude that Cx3cr1+ cells express higher levels of immunosuppressive genes compared to Cx3cr1- cells. Because the list of genes clearly includes many genes that are immunostimulatory (IL1b, CCL5, Nfkb1, Ccl2, Cd80, CD86), the conclusions about immunosuppressive gene expression profile in Cx3cr1- cells are not clearly supported.

There might be a typo error (Cx3cr1- instead of Cx3cr1+) in the last sentence of the reviewer's comment because we only discussed the immunosuppressive capacity of Cx3cr1+ cells, not of Cx3cr1neg cells.

The reviewers are perfectly right and Cx3cr1+ osteoclasts do not only express high levels of immunosuppressive genes but also of genes involved in inflammation (Figure 6A), which is in line with their capacity to stimulate to some extend the proliferation of inflammatory CD4+ T cell (Figure 5G, Figure 5—figure supplement 1). Thus, we agree that just the RNA expression profile of these OCLs is not fully supporting their immunosuppressive role. In the revised manuscript, the comment on the RNAseq analysis for this figure in the Results section has been rephrased accordingly. However, our conclusions do not only rely on the expression profile of osteoclasts, but also on in vitro functional analysis that clearly show:

1) that the T cell activation capacity of Cx3cr1+ OCLs is increased when blocking the immunosuppressive checkpoint protein PD1 in vitro (Figure 6C), 2) that in an immunosuppressive assay, Cx3cr1+ OCLs reduce the capacity of Cx3cr1neg OCLs to stimulate T cell proliferation (Figure 6D-E) which is a clear functional demonstration of their immunosuppressive effect, at least in vitro as mentioned in the revised version.

6) In Figure 1 the authors claim that CXC3R1 positive osteoclasts are needed for the full bone resorbing activity of osteoclasts after ovariectomy. The claim is based on only minor differences and whether CXC3R1 positive osteoclasts were preset or not, the osteoporosis phenotype develops in an equally severe manner. Moreover, these data contradict a statement made later in the paper that CXC3R1 positive osteoclasts are not needed for bone resorption.

We agree that the differences observed when comparing OVX WT mice and OVX Cx3cr1-deficient mice (Figure 1B-C) for the microscanner analysis are minor but they are significant and are further supported by a reduced osteoclast number (Figure 1D). But to follow the reviewer concern, we rephrased the comment of this figure in the Results section to indicate this low degree of variation.

However, there might be a misunderstanding as we didn't claim that Cx3cr1+ OCLs are required for bone loss in OVX mice. Figure 1 is not addressing the role of Cx3cr1+ osteoclasts but the implication of the Cx3cr1 protein in bone destruction associated to ovariectomy, which are 2 different questions. In Figure 1, we showed that Cx3cr1, as a protein, is important for the recruitment of OCL progenitors in the bone marrow of OVX mice (Figure 1E-F). Moreover, we showed that i-OCLs from Cx3cr1-deficient and from WT mice have the same resorption activity (Figure 2C), which demonstrates that the protein Cx3cr1 is not important for this activity.

Moreover, we didn't claim that Cx3cr1+ OCLs are not needed for bone resorption, we only showed that they resorb less than the Cx3cr1neg OCLs (Figure 4E-F).

But to avoid any misunderstanding, we modified the text to make it clearer and to describe in more details the mouse model we used. We also added in Figure 2 and 3 a schematic representation of the preparation of the cells to avoid any confusion.

7) Data shown in Figure 2 (no difference in the bone resorption capacities of CX3CR1+ cells isolated from GFP/+ mice and CX3CR1- cells isolated from GFP/GFP mice) and those shown in Figure 4 (large differences in bone resorptive capacities of CX3CR1+ and CX3CR1-negative cells derived from dendricytes) are contradictory. This inconsistency calls into question the validity of using cells derived from the dendricytes in culture if they behave differently from the mature cells isolated from the genetically modified mice.

As explained in the first point above, due to methodological issues, in particular the absence of a protocol to isolate sufficient osteoclasts or to sort ex vivo specific subsets of OCLs, there is no other possibility to investigate the function of these OCLs than the use of in vitro differentiated cells. In particular in our study, all inflammatory osteoclasts used in functional assays were derived from dendritic cells in exactly the same culture conditions, as already described (Ibanez et al., 2016). Thus, even if we cannot exclude that the culture conditions could affect osteoclast function compared to in vivo, it would be to the same extend in all our analysis. It would also be to the same extend as in all studies investigating osteoclast activity published so far.

Regarding the reviewer's concern on potential inconsistency in our results, there might be a misunderstanding. In Figure 2, we addressed the role of Cx3cr1, as a protein, in bone resorption and T cell activation by osteoclasts. This was achieved comparing Cx3cr1+ osteoclasts coming either from WT (Cx3cr1GFP/+) mice or from mice baring a non-functional Cx3cr1 gene (Cx3cr1GFP/GFP). There are no data on Cx3cr1neg cells in Figure 2. And indeed, we showed that Cx3cr1+ cells have equivalent resorption capacity whether Cx3cr1 is functional or not (Figure 2), which demonstrates that the protein Cx3cr1 is not important for this activity.

In contrast, in Figure 4, we addressed, in WT mice, the specific function of the 2 distinct subsets of osteoclasts that are identified using Cx3cr1 as a membrane marker. We showed that Cx3cr1+ and Cx3cr1neg i-OCLs from WT mice differ in their resorption and T cell activation capacity (Figures 3-6).

Thus, Figures 1-2 and Figures 3-6 explore 2 completely distinct aspects that could not be compared.

Importantly, Cx3cr1+ OCLs having a non-functional Cx3cr1 gene could not be considered equivalent to Cx3cr1neg OCLs from WT mice. Indeed, as shown in Author response image 1, principal component analysis of our transcriptomic data clearly demonstrated that while Cx3cr1+ i-OCLs from WT and mutant mice are closely related, both differ considerably from Cx3cr1neg cells from WT mice as they cluster in 2 completely distinct groups (panel A). This was further confirmed by matrix similarity analysis of our transcriptomic data that showed again that the 2 populations are distinct (panel B). Therefore none of the conclusions made for one of these population can be applied to the other.

Author response image 1. Cx3cr1neg i-OCLs from Cx3cr1GFP/+ mice are distinct from Cx3cr1+ i-OCLs from deficient Cx3cr1GFP/GFP mice.

Author response image 1.

Open in a new tab

A) Principal component analysis of Cx3cr1+ from deficient Cx3cr1GFP/GFP mice (green) and Wt Cx3cr1GFP/+ mice (red) and Cx3cr1neg i-OCLs from Cx3cr1GFP/+ mice (blue). Data shows the two first components after batch correction. Each dot represents the expression profile of one sample. B) Similarity matrix analysis on the same samples.

However, we recognize that switching from the analysis of the implication of the Cx3cr1 protein in bone loss to the analysis of the respective role of Cx3cr1+ and Cx3cr1neg i-OCL could be confusing. Thus, in the revised manuscript we modified the text to make it clearer and we added schematic representations of the preparation of OCLs in the figures.

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Data Citations

    1. Madel MB, Garchon HJ, Wakkach A, Blin-Wakkach C. 2020. Heterogeneity of inflammatory osteoclasts based on CX3CR1 expression. European Nucleotide Archive. PRJEB36092

    Supplementary Materials

    Source data 1. Raw counts-HTSeq.

    Raw counts (HTseq) per gene and per sample from the RNAseq analysis of Cx3cr1+ and Cx3cr1neg i-OCLs from WT Cx3cr1GFP/+ mice and Cx3cr1+ i-OCLs from deficient Cx3cr1GFP/GFP mice. In each group, the 4 i-OCL samples have been obtained from 4 different mice.

    elife-54493-data1.xlsx (3.3MB, xlsx)
    Source data 2. Counts_Deseq NormaBatch Corrected.

    Counts per gene and per sample after normalization and batch correction from the RNAseq analysis of Cx3cr1+ and Cx3cr1neg i-OCLs from WT Cx3cr1GFP/+ mice and Cx3cr1+ i-OCLs from deficient Cx3cr1GFP/GFP mice. In each group, the 4 i-OCL samples have been obtained from 4 different mice.

    elife-54493-data2.xlsx (2MB, xlsx)
    Source data 3. Cx3cr1+ i-OCLs from WT Cx3cr1GFP/+ mice versus Cx3cr1+ i-OCLs from deficient Cx3cr1GFP/GFP mice _Deseq.

    Statistics (Deseq) for the comparison of expressed genes in Cx3cr1+ i-OCLs from WT Cx3cr1GFP/+ mice versus Cx3cr1+ i-OCLs from deficient Cx3cr1GFP/GFP mice. In each group, the 4 i-OCL samples have been obtained from 4 different mice.

    elife-54493-data3.xlsx (1.2MB, xlsx)
    Source data 4. Cx3cr1+ i-OCLs from WT Cx3cr1GFP/+ mice versus Cx3cr1neg i-OCLs from WT Cx3cr1GFP/+ mice mice _Deseq.

    Statistics (Deseq) for the comparison of expressed genes in Cx3cr1+ versus Cx3cr1neg i-OCLs from WT Cx3cr1GFP/+ mice. In each group, the 4 i-OCL samples have been obtained from 4 different mice.

    elife-54493-data4.xlsx (1.2MB, xlsx)
    Transparent reporting form
    elife-54493-transrepform.pdf (271.9KB, pdf)

    Data Availability Statement

    All the RNA sequencing data are included in the submitted manuscript as data source files. RNA-Sequencing data have been deposited in ENA (European Nucleotide Archive) under accession number PRJEB36092.

    The following dataset was generated:

    Madel MB, Garchon HJ, Wakkach A, Blin-Wakkach C. 2020. Heterogeneity of inflammatory osteoclasts based on CX3CR1 expression. European Nucleotide Archive. PRJEB36092

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES