Moesin controls cell-cell fusion and osteoclast function

Abstract

Cell-cell fusion is an evolutionarily conserved process that is essential for many functions, including the formation of bone-resorbing multinucleated osteoclasts. Osteoclast multinucleation involves dynamic interactions between the actin cytoskeleton and the plasma membrane that are still poorly characterized. We found that moesin, a cytoskeletal linker protein member of the Ezrin, radixin, and moesin (ERM) protein family, plays a critical role in both osteoclast fusion and function. Moesin inhibition favors osteoclast multinucleation as well as HIV-1- and inflammation-induced cell fusion. Accordingly, moesin depletion decreases membrane-to-cortex attachment and enhances the formation of tunneling nanotubes, F-actin-based intercellular bridges triggering cell-cell fusion. In addition, moesin regulates the formation of the sealing zone, a key structure determining osteoclast bone resorption area, and thus controls bone degradation via a β3-integrin/RhoA/SLK pathway. Finally, moesin-deficient mice have reduced bone density and increased osteoclast abundance and activity. These findings provide a better understanding of cell-cell fusion and osteoclast biology, opening new opportunities to specifically target osteoclasts in bone disease therapy.

Figure 1.

TNTs participate in the fusion of osteoclast precursors. (A) Human monocytes isolated from blood were differentiated into osteoclasts (hOC) and analyzed on days 3, 6, and 10. Representative super-resolution microscopy images: F-actin (phalloidin, white) and nuclei (DAPI, cyan). Scale bar, 20 µm. Image in Fig. 1 A (day 10, right) is reused in Fig. S3 B. (B) Same experiment as in A with osteoclasts derived from the murine HoxB8 immortalized cell line (mOC) on days 3, 5, and 7. (A and B) White arrowheads show TNTs and pink arrowheads show zipper-like structures. (C) Left panels: Super-resolution microscopy images of TNTs with colored-coded Z-stack of F-actin (phalloidin) staining of 3 day-hOC or 3 day-mOC from 0 µm (substrate, dark blue) to 15 µm (yellow). Scale bar, 20 µm. See Videos 1 and 2. Right panels: Quantification of the percentage of cells forming thick and thin TNTs in hOCs and mOCs after immunofluorescence analysis (see Materials and methods), from one representative differentiation out of 3. n > 250 cells per condition, means ± SEM are shown. (D) Representative immunofluorescence analysis showing thin (white arrowheads) and thick TNTs (orange arrowheads): F-actin (phalloidin, white), nuclei (DAPI, cyan), and microtubules (α-tubulin, orange). Scale bar, 10 µm. (E) Bright-field confocal images from a time-lapse movie of hOCs fusing from a TNT (hour:min). See also Videos 3, 4, and 5. Dashed green and red lines delineate the nuclei before cell fusion and dashed orange lines after fusion. Arrowhead shows a TNT-like protrusion. Scale bar, 10 µm.

Figure S1.

(Related to Fig. 2 ). (A and B) Representative western blot analysis of the level of ERM during murine (A, mOC) and human (B, hOC) osteoclast differentiation, actin was used as loading control. (C). Representative western blot analysis of the level of ERM and activated ERM (P-ERM) in the three individual KO mOC. The quantification of the expression level, normalized to actin, is shown under each band. (D) Quantification of fusion index in control (CTL) and ezrin and radixin KO mOC. Each circle represents an independent experiment, n = 4, SDs are shown. (E) Quantification of the area occupied by osteoclasts in control (CTL) versus moesin KO (MKO) mOC after microscopy analysis. Each circle represents an independent experiment, n = 6. (F) Flow cytometry analysis of the percentage of β3-integrin-positive cells in control (CTL) versus moesin KO (MKO) mOC. Each circle represents an independent experiment, n = 6, SDs are shown. (G) Quantification of mRNA expression of genes overexpressed in osteoclasts measured by RT-PCR in control (CTL, blue) versus moesin KO mOC (orange) on days 3, 5, and 7 of differentiation. Actin mRNA level was used as control. Each circle represents an independent experiment, n = 3 independent experiments, SDs are shown. (H and I) Western blot analysis of Scr (H) and cathepsin K (I). (left) Representative experiment and (right) quantification of expression level normalized to actin. Each circle represents a single donor, n = 3. Predicted molecular weight are indicated on western blots. Actin panel is the same in H and I. Statistical analyses: (D) Kruskal–Wallis and then Dunn’s multiple comparison tests. n.s., not significant. Source data are available for this figure: SourceData FS1.

Figure 2.

Moesin KO increases the fusion capacities of mOCs and hOCs. (A) Western blot analysis of activated ERM (P-ERM) expression level during mOC differentiation (days 3, 5, and 7); actin was used as loading control and quantification of P-ERM level was normalized to actin. Each circle represents an independent experiment, means ± SD are shown, n = 5. (B) Same experiment as in A during hOC differentiation (days 3, 6, and 10). Each circle represents a single donor, means ± SDs are shown, n = 9. (A and B) Predicted molecular weight are indicated. n.s. not significant. (C) Reprensentative bright-field microscopy images from a time-lapse movie of control (CTL) and moesin KO mOC (moesin KO) on day 4 of differentiation. Black arrowheads point to multinucleated giant osteoclasts. See Video 6. Scale bar, 100 µm. (D and E) Microscopy analysis of cell fusion in control (CTL) and moesin KO mOC. (D) Representative microscopy images: F-actin (phalloidin, white) and nuclei (DAPI, cyan). Scale bar, 200 µm. (E) Quantification of fusion index (each circle represents an independent experiment, n = 6); and nuclei number per multinucleated osteoclast (150–250 cells/condition, n = 3 independent experiments). (F) Representative western blot analysis of P-ERM expression level in control (CTL), ezrin KO, radixin KO, and moesin KO mOC; actin was used as loading control, n = 2. Predicted molecular weight is indicated. (G and H) Microscopy analysis of hOC fusion after treatment with nontargeting siRNA (siCTL) or siRNA targeting moesin (siMoesin). (G) Representative microscopy images: F-actin (phalloidin, white) and nuclei (DAPI, cyan). Scale bar, 100 µm. (H) Quantification of fusion index (each circle represents a single donor, n = 8) and nuclei number per multinucleated osteoclast (one representative experiment from 8 donors is shown, 100–200 cells/condition). Statistical analyses: (A and B) Friedman and then Dunn’s multiple comparison tests. *P ≤ 0.05; n.s., not significant. Source data are available for this figure: SourceData F2.

Figure S2.

(Related to Fig. 2 ). (A and B) Depletion of meosin by siRNA in hOC. Western blot analysis of moesin and activated ERM (P-ERM) expression level in hOC after treatment on day 0 with nontargeting siRNA (siCTL) or siRNA targeting moesin (siM), on days 3, 6, and 10 of differentiation. (A) Representative experiment and (B) quantification of moesin (left) and P-ERM (right) expression level normalized to actin. Each circle represents a single donor, n = 7–8, SDs are shown. (C–E) Effect of LPC treatment on hOC fusion and activated ERM expression level (P-ERM). On day 6 of differentiation, hOC were treated with LPC (+LPC, 17 h), or treated and then washed (90 min) (+/− LPC) or no treated (control, CTL). (C and D) Microscopy analysis of hOC fusion. (C) Representative microscopy images: F-actin (phalloidin, white) and nuclei (DAPI, cyan). Scale bar, 50 µm. (D) Quantification of fusion index (left) and area occupied by multinucleated OC (right). Each circle represents a single donor, n =3–4. (E) Western blot analysis of P-ERM expression level in each condition, normalized to actin. Representative blot (left panel) and quantification (right panel). Each circle represents a single donor, n = 4. (F) P-ERM signal by western blot analysis in murine inflammatory osteoclasts (DC-OC) versus control osteoclasts (MN-OC). See material and methods. Representative blot of P-ERM expression level in each condition (upper panel) and quantification of P-ERM expression level, normalized to actin (lower panel). Each circle represents a single mouse, n = 4, SDs are shown. (G–J) Effect of HIV infection on activated ERM expression level (P-ERM) in hOC (G and H) and macrophages (I and J). (G and H) On day 6 of differentiation, hOC were infected with the viral strain NLAD8-VSVG (+HIV-1) or not (CTL) and analyzed 8 days after infection. (G) Quantification of the fusion index (each circle represents a single donor, n = 4) and (H) representative western blot analysis of P-ERM expression level (left), and quantification of P-ERM expression level, normalized to actin (right, each circle represents a single donor, n = 5, SD is shown). (I and J) On day 6 of differentiation, macrophages were infected with the viral strain NLAD8-VSVG (+HIV-1) or not (CTL) and analyzed 7 days after infection. (I) Quantification of the fusion index (each circle represents a single donor, n = 4) and (J) representative western blot analysis of P-ERM expression level (left) and quantification of P-ERM expression level, normalized to actin (right). Each circle represents a single donor, n = 6, SDs are shown. Predicted molecular weights are indicated on western blots. Statistical analyses: (D and E) one-way ANOVA and then Tukey multiple comparison tests. P value is indicated on the graphs, *P ≤ 0.05. Source data are available for this figure: SourceData FS2.

Figure S3.

(Related to Figs. 1 and 3). (A–D) Super-resolution microscopy images showing moesin localization in hOC on glass coverslides (A–C, Scale bar, 10 µm) or on bone (D, Scale bar, 5 µm): F-actin (phalloidin, magenta), nuclei (DAPI, cyan), and moesin (green). F-actin structures: (A) TNTs, (B) zipper-like structures, (C) a podosome belt and (D) a sealing zone. Image in Fig. S3 B is reused from Fig. 1 A (day 10, right).

Figure S4.

(Related to Fig. 3 ). (A) Representative immunofluorescence images showing activated ERM (P-ERM) at TNTs in hOC. F-actin (phalloidin, magenta), nuclei (DAPI, cyan), and P-ERM (green). Scale bar, 10 µm. See also Video 7. (B) Representative microscopy images (F-actin, phalloidin, gray) illustrating the increase of TNT number after moesin depletion: (left) moesin KO versus CTL mOC and (right) siMoesin versus siCTL hOC. Scale bars, (left) 100 µm and (right) 50 µm. (C) Representative image of a 1:1 mixed culture of mOC control (CTL, transduced with mCherry-lifeact) and moesin KO (transduced with GFP-lifeact) seeded on glass coverslides on day 3 of differentiation. Arrowheads show TNT-like protrusions. Scale bar, 20 µm.

Figure 3.

Moesin depletion enhances the formation of TNTs and reduces MCA. (A–D) Effect of moesin depletion on TNT formation in mOCs (A and B) and hOCs (C and D). (A and C) Representative scanning electron microscopy images of mOCs (day 3) CTL versus moesin KO (A) and mononucleated hOCs (day 3) treated with nontargeting siRNA (siCTL) or targeting moesin (siMoesin) (C). White arrowheads show TNTs. (A) A giant mOC is colored in purple. Scale bar, 50 µm (A) and 20 µm (C). (B and D) Quantification of the percentage of cells forming thick and thin TNTs after immunofluorescence analysis in mOCs (B, n = 3 independent experiments) and hOCs (D, n = 4 donors) (see Fig. S1 B and Materials and methods), n > 250 cells per conditions, means ± SDs are shown. Statistical analysis is shown for thick TNTs. (E and F) Analysis of force by atomic force spectroscopy operated in dynamic tether pulling mode. (E) Force-velocity curve from dynamic tether pulling on CTL and moesin KO (MKO) mOCs. Data points are mean tether force ± SEM at 2, 5, 10 and 30 µm/s pulling velocity. At least 17 cells per condition were analyzed in 4 independent experiments. (F) Mean and SD of the MCA parameter Alpha obtained from fitting the Brochard-Wyart model (see Materials and methods for details).

Figure 4.

Moesin depletion boosts bone degradation in both mOCs and hOCs. (A–D) Effect of moesin KO on bone degradation (A and B) and sealing zone (SZ) formation (C and D) in mOCs. (A and B) mOC control (CTL) versus moesin KO (MKO) were cultured for 7 days on bone slices; after cell removal, bone was stained with toluidine blue. (A) Representative images of bone degradation, eroded bone surfaces are delineated by dashed white lines. Scale bar, 50 µm. (B) Quantification of bone eroded surface (%) using semiautomatic quantification. Each circle represents an independent experiment, n = 4, means ± SDs are shown. (C and D) mOC control (CTL) versus moesin KO (MKO) were cultured for 5 days on glass coverslips, detached and then seeded for additional 2 days on bone slices. (C) Representative microscopy images of sealing zones visualized by F-actin staining (phalloidin, white in upper panels and colored-coded intensity in lower panels). Scale bars, 20 and 5 µm. (D) Quantification of the number of sealing zones per bone surface (each circle represents an independent experiment, n = 4, means ± SDs are shown) and of sealing zone thickness (n = 3 independent experiments, 15–20 SZ/condition, 3 locations/SZ). (E–H) Effect of moesin depletion on bone degradation (E and F) and sealing zone formation (G and H) in hOCs. 6 day–differentiated hOCs on glass coverslips treated on day 0 with siCTL or siMoesin (siM) were detached and seeded for additional 24 h on bone slices. (E) Same legend as in A. Scale bar, 100 µm. (F) Same legend as in B (each circle represents a donor, n = 4, means ± SDs are shown). (G) Same legend as in C. (H) Quantification of the number of sealing zones (each circle represents a donor, n = 4) and of sealing zone thickness (n = 3 donors, 15–20 SZ/condition, 3 locations/SZ).

Figure S5.

(Related to Figs. 4 and 6). (A) Representative image from a culture with 1:1 ratio of mOC control (CTL, transduced with mCherry-lifeact) and moesin KO (transduced with GFP-lifeact) seeded on glass coverslips on day 4 of differentiation. Arrowheads show green podosome belts. Scale bar, 20 µm. (B) Effect of moesin KO on sealing zone formation in mOC. mOC control (CTL) versus moesin KO (MKO) were cultured for 5 days on glass coverslips, detached and then seeded for additional 2 days on bone slices. Quantification of the area occupied by sealing zones (left) and circularity of sealing zones (right). Each circle represents a single donor, n = 4, SDs are shown. (C) Effect of moesin depletion on sealing zone formation in hOC. 6 day–differentiated hOC on glass coverslips treated on day 0 with siCTL or siMoesin were detached and seeded for additional 24 h on bone slices. Quantification of the number of nuclei per cells forming sealing zones. Each circle represents an independent experiment, n = 4 donors. n.s., not significant. (A–C related to Fig. 4). (D) Effect of NSC23 and ML-141 on ERM activation. 6-day hOCs were treated or not (CTL) with NSC23 and ML 141, drugs targeting Rac1/2 and Cdc42 respectively. Representative western blot analysis (left) and quantification of P-ERM signal normalized to actin (right). Each circle represents a single donor (n = 5, SDs are shown). (E) Effect of Y27632 (ROCK kinase inhibitor) treatment on P-ERM signal. Representative western blot analysis (left) and quantification of P-ERM signal normalized to actin (right). Each circle represents a single donor, n = 6, SDs are shown. Predicted molecular weight are indicated on western blots. (D and E related to Fig. 6) Statistical analyses: (D and E) Friedman and Dunn’s multiple comparison tests. n.s. not significant. Source data are available for this figure: SourceData FS5.

Figure 5.

Moesin role in bone degradation is independent of its role in osteoclast fusion. (A) Experimental design of moesin depletion in late stages of hOC differentiation. (B) Effect of moesin depletion (siMoesin) on hOC fusion: quantification of fusion index after microscopy analysis of 10 day-differentiated hOCs on glass coverslips treated on day 6 with siCTL or siMoesin. Each circle represents a single donor, n = 7, SDs are shown. (C) Western blot analysis of moesin and P-ERM expression levels after moesin depletion in late stages of hOC differentiation. Representative western blot analysis (left) and quantification of moesin and P-ERM signals (right), normalized to actin. Predicted molecular weight are indicated. Each circle represents a single donor, n = 8, SDs are shown. (D–I) Effect of moesin depletion (siMoesin) in mature hOCs on bone degradation (D), morphology of the resorbed area (E), and sealing zone (SZ) formation (F–I). 10 day-differentiated hOCs on glass coverslips treated on day 6 with siCTL or siMoesin were detached and seeded for additional 24 h on bone slices. (D) Representative images of bone degradation (left, scale bar, 50 µm) and quantification of bone eroded surface (%) using semi-automatic quantification (right). Each circle represents a single donor, n = 5. (E) Quantification of the percentage of trenches (n = 2 independent experiments, SDs are shown). (F and G) (F) Representative microscopy images of sealing zone visualized by F-actin staining (phalloidin, white, scale bars, 20 µm); and (G) quantification of the number of sealing zones (number of SZ per bone surface (left) and the percentage of area covered by SZ (right). Each circle represents a single donor, n = 6. (H and I) Effect of moesin depletion (siMoesin) in mature hOCs on SZ organization (H) thickness (I). (H) Representative microscopy images of sealing zones visualized by F-actin and vinculin staining (phalloidin in pink and vinculin in green). Scale bars, 10 µm. (I) Quantification of sealing zone thickness (n = 3 donors, 15 SZ/condition, 2 locations/SZ). n.s., not significant. Source data are available for this figure: SourceData F5.

Figure 6.

The Rho/SLK axis downstream of β3-integrin controls ERM activation and sealing zone formation. (A) Effect of calyculin and staurosporine treatment on ERM activation (P-ERM), used as positive and negative control for western blot analysis of ERM activation, respectively. 6-day hOCs were treated or not (CTL) with calyculin and staurosporine. Representative western blot analysis (left) and quantification of P-ERM signal normalized to actin (right). Each circle represents a single donor, n = 6, SDs are shown. (B) RhoA inhibition reduces ERM activation. 6-day hOCs were treated or not (CTL) with TATC3, targeting the RhoGTPases RhoA. Representative western blot analysis (left) and quantification of P-ERM signal normalized to actin (right). Each circle represents a single donor, n = 6, means ± SDs are shown. (C) SLK suppression reduces ERM activation. hOCs were treated with non-targeting siRNA (siCTL) or siRNA targeting SLK kinase (siSLK). Representative western blot analysis (left) and quantification of P-ERM signal normalized to actin (right). Each circle represents a single donor, n = 7, means ± SDs are shown. (D) β3-integrin suppression favors ERM activation. hOCs were treated with nontargeting siRNA (siCTL) or siRNA targeting β3-integrin (si β3-integrin). Representative western blot analysis (left) and quantification of P-ERM signal normalized to actin (right). Each circle represents a single donor, n = 7, SDs are shown. Predicted molecular weights are indicated (A–D). (E and F) Effect of SLK and β3-integrin depletion on the formation of sealing zones in hOCs. (E) Representative microscopy images of sealing zones visualized by F-actin staining (phalloidin: white in upper panels and colored-coded intensity in lower panels). Scale bars, 20 and 5 µm. (F) Quantification of sealing zone thickness (n = 2 donors, 15–20 cells/condition and 3 locations/SZ). (G) Schematics showing the proposed Rho/SLK axis downstream of β3-integrin for ERM activation. Statistical analyses: Multiple comparison tests (A) Friedman and then Dunn’s, and (F) Kruskal–Wallis and then Dunn’s. ****P ≤ 0.0001. Source data are available for this figure: SourceData F6.

Figure 7.

Moesin is expressed in osteoclasts in vivo. (A) Representative immunofluorescence images of histological analysis of femurs from WT mice: nuclei (DAPI, blue) and moesin (section 1, green) or cathepsin K (section 2, green). Scale bar, 20 µm. Sections 1 and 2 are serial sections. White arrowheads show osteoclasts. (B) Representative confocal microscopy image (maximal projection of 25 images) of histological analysis of femurs from WT mice: nuclei (DAPI, blue) and moesin (green). Bone and BM (bone marrow) are shown. Scale bar, 20 µm. See also Video 8.

Figure 8.

Moesin deficiency translates in bone defects. (A) Representative x-ray images of the whole skeleton of WT and Msn^−/− mice. Scale bar, 5 cm. (B) Representative microcomputed tomography images of trabecular section of distal femurs from WT mice and Msn^−/− mice. Scale bar, 500 µm. (C) Histograms indicate means ± SD of trabecular bone volume per total volume (BV/TV), trabecular thickness (Tb.Th), number (Tb.N), and separation (Tb.Sp), analyzed by microcomputed tomography. (D) Histograms indicate means ± SD of cortical bone parameters, analyzed by microcomputed tomography. (B–D) Animal groups were composed of 6 mice of each genotype. In C, each mouse is represented by one color. 12 femora were analyzed in total for Msn^−/− mice and 10 femora for the WT mice group. n.s., not significant.

Figure 9.

Moesin deletion increases bone degradation and osteoclast activity in vivo. (A) Serum bone degradation marker analysis. CTX levels (ng/ml) in the serum of 11-wk-old male mice. Each circle represents a mouse, n = 8 WT and n = 7 Msn^−/− mice, means ± SDs are shown. (B–D) Histological analysis. (B and C) Histograms indicate the mean ± SD of bone surface (B) and TRAP+ surface per bone surface (C) for each condition. Animal groups were composed of four mice of each genotype. N ≥ 2–3 sections chosen among the most median part of the bone. Each circle represents a single bone section and each mouse is represented by one color. (D) Images from histological analysis using TRAP staining (osteoclasts in purple) and fast green (bone is green) on femurs from WT mice and Msn^−/− mice. Scale bars, 50 µm.

Figure S6.

(Related to Fig. 9 ). Serum bone formation marker analysis: PINP levels in the serum of 11-wk-old WT and Msn−/− male littermate mice was determined using the Rat/Mouse PINP EIA kit. Each circle represents a mouse, n = 8 WT and n = 7 Msn−/− mice, means ± SDs are shown.