Osteoclasts promote the formation of hematopoietic stem cell niches in the bone marrow

In oc/oc mice, the bone phenotype is dramatically altered by the absence of OCL activity, and BM cellularity is dramatically reduced (Seifert and Marks, 1985; Blin-Wakkach et al., 2006b; Wakkach et al., 2008). Despite the profound alterations in hematopoiesis described in these mice (Blin-Wakkach et al., 2004a,b), their HSC compartment has never been analyzed in the BM. Therefore, we analyzed the Lin^negSca1⁺cKit⁺ (LSK) hematopoietic stem/progenitor cell compartment by flow cytometry. The frequency and absolute number of LSK cells was decreased by 90 and 99.8%, respectively, in the BM of oc/oc mice compared with controls (Fig. 1, A and B). Interestingly, this near absence of LSK cells was accompanied by the accumulation of a Lin^negSca1⁺cKit^neg population in the BM of oc/oc mice (Fig. 1 A). Analysis of this population revealed that ∼80% of these cells did not express the panhematopoietic marker CD45, suggesting that they represent mesenchymal cells (Fig. 1 C). On the contrary, in the wild-type control mice, the Lin^negSca1⁺cKit^neg cells present in the BM belonged to the CD45⁺ hematopoietic cell population (Fig. 1 C).

Mesenchymal cells are key components of the HSC niche. Therefore, the dramatic decrease in the HSC pool size in the oc/oc BM could be related to modifications in mesenchymal cell populations, as suggested by the accumulation of the CD45^negLin^negSca1⁺cKit^neg population. To address this question, we further analyzed the phenotype of TER119^negCD45^neg mesenchymal cells present in the BM of oc/oc mice. Mesenchymal cells represented ∼20% of total BM cells in normal mice (Fig. 2 A). In oc/oc mice, TER119^negCD45^neg mesenchymal cells represent 80% of total BM cells, the majority of them coexpressing the progenitor cell marker Sca1 (Fig. 2, A and B). Indeed, flow cytometric analysis revealed a 10-fold increase in the proportion of TER119^negCD45^negSca1⁺ mesenchymal cells in the BM of oc/oc mice compared with the controls, whereas the percentage of TER119^neg CD45^neg Sca1^neg cells was equal in both mouse genotypes (Fig. 2 B). These data suggest that the mesenchymal cells present in oc/oc mice are mainly progenitor cells. The link between the decreased OCL activity and the alteration of the LSK and mesenchymal cell compartments was confirmed in newborn +/+ mice treated with zoledronic acid (ZA), a bisphosphonate that inhibits OCL activity (Russell et al., 2008). In ZA-treated mice, the LSK cell pool was decreased in association with an increased proportion of Lin^negSca1⁺cKit^neg CD45^neg mesenchymal progenitors (Fig. 2 C), as observed in the oc/oc BM (Fig. 1 C). Similar results were also observed in newborn +/+ mice treated with calcitonin, another OCL inhibitor (unpublished data). Altogether, these data strongly suggest that reduced OCL function affects the BM HSC niche.

To gain more insight into the phenotype of the oc/oc CD45^neg mesenchymal cells, we analyzed the expression of genes involved in regulation of HSC maintenance and homing in the BM. The expression of Ang-1, Jag-1, Sdf1, and Kit ligand (Kit-L) was dramatically decreased in CD45^neg cells from the oc/oc BM compared with the controls, whereas no significant difference was observed in N-Cad expression (Fig. 3 A). Interestingly, expression of Opn, a negative regulator of the HSC pool, was significantly higher in oc/oc mesenchymal cells compared with the control cells (Fig. 3 A). Altogether, these results demonstrate that the BM mesenchymal compartment was highly altered in oc/oc mice, and suggest its impairment in supporting hematopoietic stem/progenitor cell homing and retention.

To address this hypothesis, we performed an in vitro migration assay. Using transwell chambers, we showed that the CD45^neg cells from oc/oc mice displayed a fourfold-reduced ability to attract LSK cells compared with those from the wild-type mice (Fig. 3 B). This result was further confirmed by in vivo assay where LSK cells from actin-GFP mice were injected into 2-d-old irradiated oc/oc and wild-type mice. 18 h after this transfer, the localization of GFP⁺ progenitors was investigated by flow cytometry, in the BM, spleen, and blood. As expected, in +/+ mice, the transferred GFP⁺ cells localized mainly in the BM compared with the spleen (Fig. 3 C). In contrast, homing of transferred GFP⁺ LSK cells into the oc/oc BM was >10-fold reduced compared with the control mice and the majority of injected GFP⁺ cells were located in the spleen or in the blood (Fig. 3 C). Moreover, analysis of endogenous LSK cells revealed that their proportion was increased in the spleen (Fig. 3 D) and liver (not depicted) of oc/oc mice compared with the controls. Collectively, these data confirmed that the oc/oc BM environment does not support LSK cell homing and that other perinatal hematopoietic organs such as the spleen and liver can represent alternative sites for hematopoiesis in oc/oc mice, as previously suggested (Blin-Wakkach et al., 2004b).

In the BM, HSCs reside in endosteal niches, in close association with OBLs, and also in perivascular niches, in proximity to BM sinusoidal vessels (Nilsson et al., 2001; Calvi et al., 2003; Zhang et al., 2003; Visnjic et al., 2004; Morikawa et al., 2009; Méndez-Ferrer et al., 2010). Therefore, we examined both components of the niche in the oc/oc BM. Flow cytometric analysis demonstrated that the TER119^negCD45^negSca1⁺ population from the oc/oc BM contained the same proportion of CD44⁺ cells, whereas the proportion of cells expressing integrin α5 (CD49e) was reduced (Fig. 4 A). Interestingly, the percentage of TER119^neg CD45^neg mesenchymal cells expressing CD51 was also reduced (Fig. 4 A). Because CD49e expression has been associated with OBL lineage commitment (Hamidouche et al., 2009) and CD51 is a marker for OBLs (Lundberg et al., 2007), these results indicate that the OBL differentiation is impaired in oc/oc mice. This was confirmed by real time RT-PCR analysis of the expression of the main OBL markers in the CD45^neg cells purified from the BM of oc/oc and control mice. Expression of Runx2, a master gene of OBL differentiation, as well as alkaline phosphatase (Alp), osteocalcin, and Bone sialoprotein (Bsp) was significantly reduced in the CD45^neg mesenchymal cells from oc/oc mice, compared with the controls (Fig. 4 B). These results confirmed recently published data showing an increased number of colony forming unit–fibroblast (CFU-F)–representing mesenchymal cells, but a decreased CFU-ALP–representing OBLs, in the BM of oc/oc mice (Mansour et al., 2011). Altogether, these findings confirmed a reduced OBL commitment and differentiation in oc/oc mice.

We also analyzed the perivascular primitive mesenchymal cells present in oc/oc BM marrow using the recently described marker platelet-derived growth factor receptor α (PDGFRα; Morikawa et al., 2009). The frequency of PDGFRα⁺ cells was significantly higher in mesenchymal progenitors from oc/oc mice compared with the controls (Fig. 4 A). The function of perivascular mesenchymal cells and OBLs in maintaining the HSC niche is regulated by macrophages (Winkler et al., 2010; Chow et al., 2011; Christopher et al., 2011). As macrophages arise from the same progenitor as OCLs, we wondered if the increase in OCL numbers observed in oc/oc mice was detrimental to the differentiation of macrophages, resulting in a reduced retention of HSCs in the BM. The percentage of CD11b⁺F4/80⁺Ly6G⁺ and F4/80⁺CD115⁺Gr1^negCD169⁺ macrophages, which are both involved in HSC retention in their niche (Winkler et al., 2010; Chow et al., 2011), was increased in the BM of oc/oc mice compared with control (Fig. 4, C and D). These results indicate that the decrease in HSC numbers in the oc/oc BM is not related to a reduction of perivascular mesenchymal cells or macrophage populations.

The Tcirg1 gene mutated in oc/oc mice is not expressed in mesenchymal cells and in HSCs (unpublished data; Schinke et al., 2009). Thus, the profound changes observed both in mesenchymal and hematopoietic cell populations are likely to be caused by the absence of OCL activity rather than by a cell autonomous effect of the oc mutation on mesenchymal and LSK cells.

We have therefore assessed whether restoring OCL activity in oc/oc mice could reestablish both the bone phenotype and the formation of the HSC niche in the BM. OCLs derive from the monocyte lineage, and the transfer of hematopoietic cells restores OCL activity in oc/oc mice (Seifert and Marks, 1987; Frattini et al., 2005; Johansson et al., 2006; Wakkach et al., 2008). Thus, we corrected OCL activity in oc/oc mice by adoptive transfer of Sca1⁺ hematopoietic progenitors from actin-GFP mice. To avoid any contamination by CD45^neg mesenchymal cells, the injected hematopoietic progenitors were sorted as the CD45⁺Sca1⁺GFP⁺ cells (Fig. 5 A). Analysis of the mouse phenotype was performed at day 45 in the wild-type and the treated oc/oc mice, caused by the time required to reach restoration of the bone turnover (Wakkach et al., 2008; Mansour et al., 2011), and at day 17 in the untreated oc/oc mice because of their short life span up to 21 d (Blin-Wakkach et al., 2006a). As expected and already described (Frattini et al., 2005; Johansson et al., 2006), this treatment induced a restoration of bone resorption as supported by the formation of a BM cavity that was absent in oc/oc mice (Fig. 5 B). No GFP⁺CD45^neg cells were present in the treated mice, confirming that no donor mesenchymal cells were transferred (Fig. 5 C). The proportion of LSK cells in the BM of treated oc/oc mice (Fig. 5 D) dramatically increased compared with that of the untreated mice (Fig. 5, D and E), eventually reaching the level observed in the wild-type mice. Simultaneously, the proportion of LSK cells in the spleen of treated oc/oc mice decreased to the level found in the wild-type mice (Fig. 5 E). Importantly, the changes in the localization of LSK cells were associated with a decrease in the Lin^negSca1⁺cKit^neg cells that also became equivalent to the level observed in the wild-type mice (Fig. 5, D and F). A similar restoration of the LSK and mesenchymal cell compartments was also observed in oc/oc mice treated with +/+ dendritic cells (Fig. 5, G and H) that have been recently shown to efficiently differentiate into functional OCLs in this model (Wakkach et al., 2008). This observation confirmed that this restoration is specifically linked to the recovering of OCL activity and not to another hematopoietic cell modification.

Analysis of the LSK cells present in the BM of oc/oc mice treated with CD45⁺Sca1⁺GFP⁺ cells revealed that they were both from the donor (LSK GFP⁺) and the recipient (LSK GFP^neg) mice (Fig. 5 I). This result reveals that endogenous oc/oc LSK cells have a normal capacity to home to the BM, and that their dramatic decrease was caused by the defective niche and not by a cell autonomous impairment in their homing capacity. Furthermore, both the donor and recipient hematopoietic progenitors were able to generate cells of the main hematopoietic lineages in the treated oc/oc mice, indicating that LSK cells in these mice have normal differentiation potential (Fig. 5 J). Altogether, these data clearly demonstrated that the restoration of OCL function has corrected the deficient LSK cell homing in oc/oc BM.

To determine whether this restoration of LSK cell homing in the BM was caused by the restoration of the mesenchymal cell phenotype, we have further analyzed the TER119^negCD45^neg cells in the oc/oc mice treated with CD45⁺Sca1⁺GFP⁺. The percentage of the TER119^negCD45^neg cells in the treated oc/oc mice was decreased compared with the untreated oc/oc mice and returned to the level observed in the wild-type mice (Fig. 6 A). Moreover, this decrease was found within the CD45^neg Sca1⁺ fraction (Fig. 5, D and F; and Fig. 6 B) suggesting that the mesenchymal cells in the TER119^neg population from the treated-oc/oc mice have differentiated into a more mature phenotype.

To further evaluate the recovery of the mesenchymal cell phenotype and OBL differentiation and function, histological and gene expression analysis were performed. Histological analysis using Von Gieson’s staining showed that the oc/oc BM is filled with bone tissue containing large zones of cartilage and type I collagen (Fig. 6 C, blue and pink staining, respectively) as previously described (Seifert and Marks, 1985; Wakkach et al., 2008). This altered bone phenotype revealed an impaired endochondral ossification that probably participated in the alteration of the HSC niche in the BM. In contrast, in the treated oc/oc mice, the bone structure was similar to the one observed in the wild-type mice with a normal BM cavity and a reduced amount of cartilage compared with the untreated oc/oc mice, indicating a restoration of a normal bone formation (Fig. 6 C). This finding was confirmed by RT-PCR analysis showing that the expression of OBL markers also returned to a normal level (Fig. 6 D). Furthermore, the expression of the main factors important for the niche function was also equivalent to the levels observed in the wild-type control mice. These results strongly supported an interaction between OCLs and OBL precursors that contribute to the formation of the HSC niche in normal mice. Therefore, we performed double staining of tartrate-resistant acid phosphatase (TRAP), a marker of OCLs, and ALP, a marker of early OBLs and osteoid formation. In the trabecular zone of +/+ mice, most of OCLs were juxtaposed with early ALP⁺ OBLs (Fig. 6 E). On the contrary, in oc/oc mice, the ALP labeling was dramatically decreased (Fig. 6 F), as previously described (Mansour et al., 2011), because of the decreased OBL commitment. However, early ALP⁺ OBLs were detected in some regions of the BM; most of them were isolated from OCLs (Fig. 6 G). This result confirmed a potential interaction between OCLs and early OBLs, which is abolished in oc/oc mice. Interestingly, the juxtaposition between OCLs and early OBLs was restored in treated oc/oc mice, confirming that the interaction between these cells had been corrected by the restoration of OCL activity (Fig. 6 H). Together with the increased proportion of LSK cells (Fig. 4, D and E), these data demonstrate that restoring OCL activity not only rescued the commitment and maturation of OBLs but also restored their function as providers of hematopoietic niches, probably through cellular interactions between OCLs and early OBLs.

The present studies establish a novel aspect of the role of bone cells in the initial formation of the HSC niche in the BM and demonstrated that OCLs play a critical role in this formation. We demonstrate for the first time that in the absence of OCL activity, the BM HSC niche formation is severely affected and mesenchymal cells have a highly reduced capacity to attract hematopoietic progenitors, leading to an impaired homing of these progenitors. Furthermore, we show that restoration of OCL activity reverses this phenotype and allows normal hematopoietic progenitor homing in the BM.

The shift of hematopoiesis to the BM occurs just before birth and is preceded by endochondral ossification and OCL activity (Aguila and Rowe, 2005). During this process, the initial unmineralized and mineralized cartilage is gradually degraded by OCLs and replaced by the bone matrix produced by OBLs (Mackie et al., 2008). In oc/oc mice, this process is altered because of the absence of OCL activity, and long bones display a defect in bone matrix formation characterized by the persistence of large areas of mineralized and unmineralized cartilage (Banco et al., 1985; Seifert and Marks, 1985; Wakkach et al., 2008). Here, we showed that this alteration in endochondral ossification is caused by an impaired OBL differentiation from mesenchymal progenitors in the oc/oc BM, as revealed by the dramatic decrease in the expression of the OBL markers Runx2, Alp, Osteocalcin, and Bsp and the reduced proportion of cells expressing CD51 and the integrin α5 (CD49e), both of which are traits associated with OBL commitment (Lundberg et al., 2007; Hamidouche et al., 2009). This is in agreement with our recent studies demonstrating that the generation of CFU-ALP (representing OBL-committed cells) from oc/oc BM is very low compared with controls and that the number of ALP-producing cells is significantly reduced in vivo in the oc/oc bone (Mansour et al., 2011).

Different mesenchymal cells have been implicated in the HSC niche, including immature and perivascular primitive mesenchymal cells (Morikawa et al., 2009; Méndez-Ferrer et al., 2010), osteoprogenitors (Raaijmakers et al., 2010), and OBLs (Nilsson et al., 2001; Calvi et al., 2003; Zhang et al., 2003; Visnjic et al., 2004), but their relative role in the HSC niche is still a matter of debate (Askmyr et al., 2009; Ehninger and Trumpp, 2011; Lévesque and Winkler, 2011). In oc/oc mice, the percentage of Lin^negCD45^negSca1⁺ mesenchymal progenitor cells is dramatically increased in the BM, a result that is in agreement with our previous finding of a high number of CFU-F (representing mesenchymal progenitors) generated from oc/oc BM cells (Mansour et al., 2011). Furthermore, the proportion of TER119^negCD45^negSca1⁺PDGFRα⁺ cells described as perivascular primitive mesenchymal cells being part of the HSC niche in adults (Morikawa et al., 2009) is also increased. Thus, the impaired HSC niche formation observed in oc/oc mice is not caused by a defect in mesenchymal progenitors and perivascular mesenchymal cells, indicating that these cells are not sufficient for this formation and that another essential actor is altered in these mice. In this sense, the correlation between the impaired OBL commitment and the dramatic reduction in BM HSC number strongly indicated that osteoblastic cells are required for the formation of the HSC niche. This was further supported by the effect of treatment with ZA on the HSC niche in newborn mice. Interestingly, we previously reported that, in adults, blockade of OCL activity by ZA reduced OBL commitment and increased mesenchymal progenitors without affecting the LSK cell compartment (Mansour et al., 2011). These observations suggest that the OBLs are less important for the maintenance of the HSC niche in adults than for its formation in newborns, and that the OCL effect on the HSC niche reported here is specific to the initial formation of the niche, but is probably not involved in latter stages.

The reduced expression of the key regulators of HSC expansion, quiescence, homing, and maintenance (such as Jag-1, Ang-1, and SDF1) found in oc/oc CD45^neg mesenchymal cells, as well as their reduced ability to attract hematopoietic progenitors, are most likely involved in the impaired progenitor homing and responsible for the loss of HSCs in the oc/oc BM. In addition, because OPN was described as a negative regulator of the niche and was shown to down-regulate the expression of Jag-1 and Ang-1 (Stier et al., 2005), the high expression of OPN in the oc/oc mesenchymal cells may also contribute to the reduced HSC number in the oc/oc BM.

Importantly, the alteration of both the HSC niche and OBL commitment is not self-autonomous, as the mutated Tcirg1 gene in oc/oc mice is not expressed in mesenchymal cells (Schinke et al., 2009) and the restoration of OCL activity allows recovery of the OBL commitment and normal bone and hematopoietic niche formation. Collectively, these data point out that OCLs act as the key regulators of all these processes.

Previous studies have identified OCLs and phagocytic cells from the same lineage (macrophages) as regulators of the mobilization of HSCs outside of their BM niches in adult mice. Stimulation of OCL activity induces HSC mobilization and hematopoietic progenitor expansion mediated by the production of proteolytic enzymes that degrade components of the HSC niche (such as SDF-1 or SCF) required for stem cell anchorage and retention (Kollet et al., 2006). The same effect has also been observed when OCL differentiation and function are inhibited, without reduction of OBL number and function (Lymperi et al., 2011; Miyamoto et al., 2011). Moreover, the depletion of macrophages induces the mobilization of HSCs outside of the BM. This mobilization is associated with a reduction in the expression of SDF1, Ang-1, and stem cell factor (SCF) and involves modification in the phenotype of either perivascular mesenchymal stem cells (Chow et al., 2011) or OBLs (Winkler et al., 2010; Christopher et al., 2011). All these data indicate that once the HSC niche is formed, OCLs and phagocytic cells play an important role in regulating HSC maintenance in this niche. This role is probably dependent on both the proteolytic activity of these cells and the crosstalk signals with OBLs, but the precise mechanisms involved remained to be determined. In oc/oc mice, the BM macrophages were not decreased, ruling out a possible effect of macrophage depletion in this model. Our results point out a novel function of active OCLs that is clearly different from those mentioned at the beginning of the paragraph. Indeed, the previously reported role of OCLs deals with the modulation of HSC egress from a “mature” HSC niche in adults, whereas our data demonstrate the importance of OCLs in the initial niche formation and in its colonization by LSK cells through the stimulation of OBL commitment.

Recently, analysis in osteopetrotic models, in particular in op/op mice defective in M-CSF and in OCL function, reported that LSK mobilization in response to G-CSF is unaffected or even increases (Miyamoto et al., 2011). But the maintenance of LSK cells in the BM of these mice was not clearly established. Indeed, despite the observation of HSCs in the BM of op/op mice (Miyamoto et al., 2011), their number remains 10-fold lower than in control mice (Begg et al., 1993) and no data were reported in the other osteopetrotic models (Miyamoto et al., 2011). Moreover, despite HSC maintenance in op/op mice (Miyamoto et al., 2011), the authors have excluded the spleen as a site for the high number of stem cells observed in their livers suggests that hematopoiesis may be maintained in this organ after birth (Wiktor-Jedrzejczak et al., 1982). Therefore, together with our results, all these data suggest that in osteopetrotic models, the colonization of the BM by HSCs may be impaired, explaining why OCLs have no effect on their mobilization.

In oc/oc mice, OCLs are present but inactive, thus their effect on the formation of the HSC niche is clearly dependent on their bone resorbing activity; our results demonstrated that their effect is linked to their capacity to support OBL commitment. Therefore, the mean by which OCLs control the formation of the HSC niche is likely to depend on coupling or cell adhesion factors produced during bone resorption, and on participating in the interaction between OCLs, OBLs, and their progenitors (Martin and Sims, 2005; Tang et al., 2009). The cross talk between bone resorption and bone formation may be achieved through various mechanisms. Indeed, coupling factors such as insulin-like growth factor, basic fibroblast growth factor, TGF-β, bone morphogenetic proteins, and platelet-derived growth factor are released from the bone matrix during bone resorption and are known to induce bone formation, thereby coupling bone resorption and formation (Sims and Gooi, 2008). Active OCLs also produce acid and proteases that activate the coupling factors, e.g., TGF-β1. Recently, it has been shown that the release and activation of TGF-β1 in response to osteoclastic bone resorption induces migration of mesenchymal precursor cells to the resorptive sites and their differentiation into OBLs (Tang et al., 2009). It was also reported that blockage of OCL activity by bisphosphonates inhibits the release of active TGF-β1 and the recruitment of Sca-1⁺ skeletal stem cells for the bone formation (Tang et al., 2009). In this sense, we observed in the oc/oc BM a reduced juxtaposition between OCLs and early OBLs that is corrected after restoration of OCL activity. Such a phenotype probably involves reduced expression of molecules involved in the contact between OCLs and osteoblastic cells. Interestingly, the expression of c-kit and its ligand Kit-L is dramatically reduced in the oc/oc BM. The c-Kit–Kit-L pathway is a potent stimulator of cell adhesion involved not only in the maintenance of HSCs, but also in the adhesion between OCLs and OBLs (Gattei et al., 1996; Lotinun et al., 2005). Therefore, the decreased expression of both c-kit and its ligand could participate in a reduced communication between OCLs and osteoblastic cells that could be involved in the defect in the HSC niche observed in oc/oc mice.

Although the specific mechanisms involved in this cross talk between OCLs and OBL progenitors leading to HSC niche formation remains to be elucidated, a reduced production of coupling factors or adhesion factors in oc/oc mice may result in an impaired OBL commitment and HSC niche formation. Lastly, in addition to their effect on OBLs, OCLs may also act directly on HSC homing through calcium released from the bone matrix, which is also a potent regulator of HSC retention in close physical proximity to the endosteal surface, interacting with the calcium-sensing receptors expressed on HSCs (Adams et al., 2006).

In conclusion, we show here for the first time that active OCLs are necessary for the initial formation of the HSC niche. The loss of OCL-resorbing activity blocks the cellular and molecular interactions in the BM and ultimately disrupts the niche formation through the loss of endosteal OBLs, thereby inhibiting the colonization of the BM by HSCs. These findings extend our knowledge of the processes participating in the establishment of the HSC niche, which is critical in understanding normal and pathological hematopoiesis.

Osteopetrotic oc/oc mice (C57BL/6J X C3HheB/FeJ genetic background) were obtained by crossing heterozygous oc/+ mice. Genotyping was performed as previously described (Blin-Wakkach et al., 2004b). Analyses were performed on 17-d-old oc/oc mice and control +/+ littermates. Transgenic mice expressing GFP under the control of the actin promoter (actin-GFP mice) were obtained from C. Mueller (Institut de Biologie Moléculaire et Cellulaire, Strasbourg, France). Pregnant C57BL/6 females received an intraperitoneal injection of 100 µg/kg ZA (Novartis) in 100 µl PBS or 100 µl PBS alone (control mice) 3 d a week for the duration of their pregnancy. From the day of birth, the newborn mice issued from ZA-treated females also received intraperitoneal injection of 100 µg/kg ZA in 50 µl PBS, whereas those from PBS-treated females received PBS alone, 3 d a week. Analyses were performed on 15-d-old PBS- or ZA-treated mice. All animals were maintained in our central animal facility in accordance with the general guidelines of the Direction des Services Vétérinaires. Approval for the use of mice in this study was obtained from the animal facility committee from the Faculty of Medicine at the University of Nice Sophia Antipolis.

BM cells from 17-d-old oc/oc mice and control +/+ littermates were collected by crushing the femora into small pieces and vigorous pipetting, as previously described (Blin-Wakkach et al., 2004a,b). Splenocytes were isolated by filtration through a 40-µm cell strainer (BD). Blood cells were obtained by intracardiac puncture, as previously described (Augier et al., 2010). After red blood cell lysis, HSCs were analyzed after exclusion of lineage (Lin)-positive cells using FITC-conjugated antibodies against Ly76 (Ter-119), CD3 (17A2), CD11b (M1/70), and B220 (RA3-6B2) and after labeling with Sca1 (D7) and c-kit (288) antibodies. For mesenchymal cell analysis, cells were labeled with antibodies against CD45 (30F11), Ly76 (Ter-119), Sca1, CD44 (IM7), CD49e (5H1Q-275), CD51 (RMV-7; all from BD), and PDGFRα (APA5; eBioscience). For macrophage analysis, cells were labeled with antibodies against CD11b (M1/70), F4/80 (Cl:A3-1), Ly6G (1A8), Gr1 (RB6-8C5), CD115 (AFS98; all from BD), and CD169 (3D6.112; AbD Serotec). After washes, the labeled cells were analyzed on a FACSCanto (BD). For cell purification, CD45^neg mesenchymal cells were sorted on a cell sorter (FACSAria; BD) with high purity (>99%).

Total RNA was extracted from the purified CD45^neg cells by adsorption onto silica membranes (Macherey-Nagel) as previously described (Blin-Wakkach et al., 2004a,b). Total RNA (1 µg) was reverse transcribed with random primers according to the manufacturer’s protocol (Invitrogen). Real-time PCR analysis was performed on an ABI Prism 7000 (Applied Biosystems), in a 20-µl volume containing 10-fold diluted cDNA, 10 µl SYBR Green Master Mix, and 300 nM of each primer. Samples were treated according to the following program: 5°C for 2 min, 94°C for 10 min, and 40 cycles of 95°C for 15 s and 60°C for 1 min. Analyses were performed in triplicate. For each sample, the cycle threshold (Ct) values were determined. Test cDNA results were normalized to the ribosomic 36B4 RNA on the same plate. Differences in gene expression between oc/oc and control mice were calculated using the 2ΔCt method. At the end of the PCR assay, generating a melting curve of the PCR product and analysis by gel electrophoresis controlled the specificity of amplification. The following primers were used: Sdf1, 5′-GAGCCAACGTCAAGCATCTG-3′ and 5′-CGGGTCAATGCACACTTGT-3′; N-Cad: 5′-TCCCTTATCTCCAGTGACTATTAAAAGA-3′ and 5′-TCTTTAAATTGAGCTGAGCAAAATCA-3′; Ang-1, 5′-GAAGCAACAACTGGAGCTCATG-3′ and 5′-TCCTCCCTTTAGCAAAACACCTT-3′; Jag-1, 5′-AACGACCGTAATCGCATCGT-3′ and 5′-TATCAGGTTGAATAGTGTCATTACTGGAA-3′; Kit-l, 5′-TCCGAAGAGGCCAGAAACTAG-3′ and 5′-CTAAGGGAGCTGGCTGCAA-3′; Opn, 5′-CTGTGTCCTCTGAAGAAAAGGATG-3′ and 5′-GCTTTCATTGGAATTGCTTGG-3′; 36B4, 5′-TCCAGGCTTTGGGCATCA-3′ and 5′-CTTTATCAGCTGCACATCACTCAGA-3′; Osteocalcin, 5′-CCACCCGGGAGCAGTGT-3′ and 5′-CTAAATAGTGATACCGTAGATGCGTTTG-3′; Runx2, 5′-TTTAGGGCGCATTCCTCATC-3′ and 5′-TGTCCTTGTGGATTAAAAGGACTTG-3′; Bone sialoprotein (BSP), 5′-CCAAGCACAGACTTTTGAGTTAGC-3′ and 5′-CTTTCTGCATCTCCAGCCTTCT-3′; Alkaline phosphatase (ALP), 5′-TCAGGGCAATGAGGTCACATC-3′ and 5′-CACAATGCCCACGGACTTC-3′.

BM cells from 3–4-wk-old actin-GFP mice were enriched in Sca1⁺ cells using Sca1 microbeads (Miltenyi Biotech) according to the manufacturer protocol. From these Sca1⁺-enriched cells, CD45⁺ LSK GFP⁺ cells were sorted (FACSAria; BD) with high purity sorting (>99%). For in vivo homing analysis, these cells were injected intraperitoneally (2.5 × 10⁵ cells in 50 µl PBS) into 2-d-old oc/oc or +/+ mice after nonlethal total body irradiation (3Gy). Animals were sacrificed after 18 h, and the presence of GFP⁺ progenitor cells was analyzed by flow cytometry (FACSCanto; BD). For in vitro migration assay, CD45^neg cells were sorted from the BM of oc/oc and +/+ mice on a cell sorter (FACSAria; BD). 10⁶ CD45^neg cells were placed in MEMα medium containing 5% fetal calf serum (HyClone) and incubated at 37°C and 5% CO₂ in the bottom chamber of 5-µm pore size transwells (Costar). After 24 h, 5 × 10⁴ LSK GFP⁺ cells were added in the top chamber in MEMα medium. After 3 h of incubation at 37°C and 5% CO₂, cells that had migrated into the bottom chamber were collected and the number of GFP⁺ cells was determined by flow cytometry.

Sca1⁺CD45⁺ cells were sorted from the BM of 3–4-wk-old actin-GFP mice (FACSAria; BD), with high purity (100%). These cells were injected intraperitoneally (5 × 10⁶ cells in 50 µl PBS) into 2-d-old oc/oc or +/+ irradiated mice (3Gy) as previously described (Wakkach et al., 2008) and animals were sacrificed at day 45. The untreated oc/oc mice were sacrificed at day 17, because of their short life span (Blin-Wakkach et al., 2006a), and used as the control. For treatment with DCs, CD11c⁺MHC-II⁺ DCs were sorted from +/+ mice and were transferred (5 × 10⁶ cells) intraperitoneally into 2-d-old oc/oc irradiated mice (3Gy), as previously described (Wakkach et al., 2008).

For histological analysis of the tibia, undecalcified bones were fixed in 4% paraformaldehyde, dehydrated, and embedded in methylmethacrylate, as previously described (Wakkach et al., 2008). 7-μm sections were stained with a 0.3% Alcian blue in 3% acetic acid solution and, subsequently, with Van Gieson’s solution. Acquisitions were performed using a microscope (model DMLB; Leica), and a 3CCD color video DXC-390 camera (Sony). For OCL and early OBL analysis, femora were fixed in 4% paraformaldehyde for 24 h at 4°C, decalcified for 10 d in 10% EDTA at 4°C and incubated overnight in a 30% glucose solution. 8-µm bone sections were stained for ALP activity with the leukocyte alkaline phosphatase kit (fast blue BB salt; Sigma-Aldrich), and then for TRAP activity with the leukocyte acid phosphatase kit (Sigma-Aldrich), according to the manufacturer’s protocol. After hematoxylin staining, images were acquired on an Observer Z1 inverted microscope and an AxioCam lCc1 8-Bit Color camera (both from Carl Zeiss, Inc.).

We thank M. Topi for animal care and technical assistance. We thank C. Mueller (IBMC, Strasbourg, France) for the kind gift of actin-GFP mice. We thank R. Dacquin (LBMC, Centre National de la Recherche Scientifique UMR 5161, Lyon, France) for Van Gieson’s analysis. We also greatly acknowledge the cell imaging facility of the Centre Méditérranéen de Médecine Moléculaire, Nice, France (MICA plateform), the flow cytometry facility, and the PETC transcriptomic facility both from Institut National de la Santé et de la Recherche Médicale UMR 576, Nice.

This work was supported by the Ministère de l’enseignement et de la Recherche (A. Mansour).

The authors have no competing financial interests.