Osteoclasts are dispensable for hematopoietic stem cell maintenance and mobilization

To determine whether loss of the BM niche and osteoclasts influences hematopoiesis, we analyzed osteopetrotic op/op mice, which carry a loss-of-function mutation in M-CSF, a cytokine essential for osteoclast differentiation (Yoshida et al., 1990). These mice lack a BM cavity and exhibit impaired osteoclastogenesis (Fig. 1 A). Previously, the cell cycle–specific cytotoxic agent 5-fluorouracil (5-FU), which is used in chemotherapy, has been used to evaluate HSC function because HSCs are resistant to 5-FU–induced cell death as a result of their quiescence (Cheng et al., 2000; Miyamoto et al., 2007). op/op and wild-type mice were serially treated with 5-FU, and we found that op/op mice were lethally susceptible to 5-FU treatment compared with control op/+ mice (Fig. 1 B), suggesting that osteoclasts and BM cavities are required to maintain HSCs.

To confirm roles of osteoclasts and BM cavities in the hematopoietic system, op/op mice were serially treated with G-CSF and HSPCs mobilization was analyzed (Fig. 1, C–E). Because a functional BM niche or osteoclasts are reportedly required to mobilize HSPCs to the periphery from the BM (Heissig et al., 2002; Katayama et al., 2006; Kollet et al., 2006), we predicted that op/op mice would show inhibited HSPC mobilization after G-CSF injection. However, we found that mobilization of the HSC-enriched fraction (lineage negative, c-Kit positive, and Sca1 positive [LSK]) to peripheral blood, as analyzed by flow cytometry, was induced to levels even higher in op/op mice than in control littermate mice (op/+; Fig. 1, C and D). HSPCs mobilized in op/op mice were functional: colony-forming assays showed that greater numbers of colonies formed from the peripheral blood of G-CSF–treated op/op mice than from that of wild-type mice (Fig. 1 E). This observation suggests that op/op mice accumulate a greater number of HSCs than do wild-type mice, and that neither BM cavities nor osteoclasts are required for HSC maintenance and mobilization and may even be inhibitory to the process.

To confirm these findings, c-Fos–deficient mice, which are also osteoclast deficient and lack BM cavities, were analyzed (Fig. S1 A). We observed that c-Fos–deficient mice were less susceptible to serial 5-FU treatment than were op/op mice (Fig. S1 B), suggesting that M-CSF, rather than BM cavities and osteoclasts, is required to resist 5-FU treatment. Indeed, injection of M-CSF into op/op mice rescued mice from a lethal response to 5-FU (Fig. S1 C). M-CSF regulates osteoclast formation as well as macrophage differentiation and innate immunity (Miyamoto et al., 2001; Chitu and Stanley, 2006). In wild-type mice, M-CSF concentration in sera was significantly up-regulated after 5-FU treatment, suggesting a protective role for M-CSF against myelosuppression induced by chemotherapy (Fig. S1 D). In fact, both Gram-positive and -negative bacteria were detected in multiple organs of 5-FU–treated op/op mice (Fig. S1 E and not depicted). These results indicate that by inducing macrophages, M-CSF likely plays a critical role in resistance to hematopoietic suppression after 5-FU treatment. They also suggest that serum M-CSF concentrations should be carefully monitored during myelosuppressive chemotherapies to prevent opportunistic infections.

RANKL-deficient mice are also osteoclast deficient and lack BM cavities (Fig. S1 F). Interestingly, increased HSPC mobilization after serial G-CSF treatment was observed in both c-Fos–deficient and RANKL-deficient mice (Fig. 2), as it was in op/op mice, supporting the idea that osteoclasts and BM cavities are not required to maintain HSCs. Young female PTPε-knockout mice reportedly show a mild loss of osteoclast function but exhibit a BM cavity and impaired HSPC mobilization by G-CSF (Kollet et al., 2006). The three osteoclast- and BM cavity–deficient animal models evaluated in this study did not show defective HSPC mobilization after G-CSF treatment, suggesting that the function of osteoclasts in driving HSCs to the periphery is biphasic or BM cavity dependent. In wild-type mice, the number of osteoclasts reportedly does not increase during G-CSF treatment, but only after cessation of treatment (Takamatsu et al., 1998; Christopher and Link, 2008; Winkler et al., 2010), also suggesting that osteoclast loss likely does not affect HSPC mobilization in response to G-CSF treatment. All three osteopetrotic animal models show increased mobilization of HSPCs to the periphery without G-CSF administration (Figs. 1 and 2), suggesting that steady-state levels of circulating HSPCs are increased in these animals.

To further explore the role of BM cavities and osteoclasts in HSC maintenance, HSPC mobilization in OPG-deficient mice was analyzed by serial G-CSF injection (Fig. 3). OPG is a decoy receptor for RANKL and therefore acts as a RANKL inhibitor (Bucay et al., 1998; Mizuno et al., 1998). OPG-deficient mice show severe osteoporosis as a result of accelerated osteoclastogenesis (Fig. 3, A and B). As expected, in contrast to osteoclast-deficient osteopetrotic mice, reduced HSPC mobilization was seen in OPG-deficient mice, as indicated by flow cytometry and colony-forming assays, compared with wild-type mice (Fig. 3, C–E). Furthermore, long-term competitive repopulation assays using mobilized cells from CD45.2⁺ op/op, c-Fos-deficient, or RANKL-deficient mice versus CD45.1⁺ BM competitors showed higher chimerism relative to that induced using mobilized cells from control mice (Fig. 3 F). In contrast, mobilized cells from OPG-deficient mice induced lower chimerism than those from wild-type mice (Fig. 3 F). CD45.2 reconstitution was multilineage; CD45.2 reconstitution from mobilized donor HSPCs was observed in CD11b⁺ myeloid, B220⁺ B cell, and CD3⁺ T cell recipient lineages (Fig. S2). Thus, osteoclasts and BM cavities appear to antagonize HSC maintenance.

Recently, it was reported that F4/80⁺ macrophages, termed osteomacs, reside in the endosteal (Chang et al., 2008) where they support osteoblast function, thereby contributing to retain HSCs in the BM niche (Winkler et al., 2010). Thus, increased HSPC mobilization seen in op/op mice after G-CSF injection could be a result of reduced levels of osteomacs. Indeed, we observed fewer F4/80⁺ osteomacs in op/op mice, although their levels in c-Fos–deficient and RANKL-deficient mice were similar to those observed in wild-type mice (Fig. 4). These observations suggest that osteomac reduction is not a common property of osteopetrotic mice, all of which show increased HSPC mobilization after G-CSF administration. CD11b⁺F4/80⁺Ly6-G⁺ macrophages, which also reportedly inhibit HSPC mobilization into the peripheral blood, are reduced after G-CSF treatment in wild-type mice (Winkler et al., 2010). However, like osteomacs, the CD11b⁺F4/80⁺Ly6-G⁺ macrophage population was reduced in op/op mice compared with wild-type mice, but such a reduction was not seen in c-Fos–deficient mice (Fig. S3 A). In OPG-deficient mice, osteomac and CD11b⁺F4/80⁺Ly6-G⁺ macrophage levels were similar to those seen in wild-type mice (Fig. 4 and Fig. S3 A). These results suggest that both of these cell types are M-CSF dependent and are deficient in the op/op model of osteopetrotic mice, but that this deficiency is not likely be the cause of increased mobilization of HSPCs seen in osteopetrotic animals. A more likely explanation is that HSPC mobilization seen in all osteopetrotic models is a result of loss of osteoclasts and BM cavities. Because F4/80⁺ osteomacs and CD11b⁺F4/80⁺Ly6-G⁺ macrophages were detected in osteoclast-deficient c-Fos–deficient mice, we do not consider F4/80⁺ cells to be osteoclasts. Indeed, F4/80 is reportedly not expressed in osteoclasts (Quinn et al., 1999).

Hematopoiesis in osteopetrotic mice reportedly occurs in extramedullary spaces, such as spleen (Lowell et al., 1996). Indeed, a greater proportion of LSK cells was seen in the spleen of op/op mice than in control (op/+) mice (Fig. S3, B and C). Thus, we removed the spleen from op/op mice and treated those mice with G-CSF serially to analyze HSPC mobilization (Fig. 5). Interestingly, HSPC mobilization was induced or even significantly higher in splenectomized op/op mice compared with sham-operated op/op mice (Fig. 5). These data suggest that HSCs are maintained in tissues other than spleen in osteoclast-deficient osteopetrotic mice. Interestingly, we found that bone tissues in osteoclast-deficient mice contained small spaces where c-Kit–positive hematopoietic cells are located (Fig. S4 A). Thus, hematopoiesis may be maintained in bone, even in osteopetrotic mice. In fact, flow cytometry analysis revealed an HSC-enriched LSK population in osteopetrotic bones of op/op mice (Fig. S4 B), and these cells were shown to be functional by an LTC-IC assay (not depicted). These results indicate that HSCs are maintained in bones of osteopetrotic mice, that increased bone mass as a result of impaired osteoclastogenesis may increase functionality of the HSC niche, and that osteoclasts could be a therapeutic target to expand HSCs as well as bone mass. Bones are formed by osteoblasts, and osteoblasts reportedly express various niche factors such as angiopoietin 1, osteopontin, Cxcl12, and KitL (Arai et al., 2004; Stier et al., 2005; Adams and Scadden, 2006; Katayama et al., 2006). We found that these niche factors were expressed in osteopetrotic bones of op/op, c-Fos–deficient, and RANKL-deficient mice (unpublished data).

Finally, we asked whether blocking osteoclast function or differentiation pharmacologically would inhibit or enhance hematopoietic activity in wild-type mice. Wild-type mice were pretreated with bisphosphonate (alendronate) or vehicle (PBS), and HSPC mobilization after serial G-CSF injection was analyzed (Fig. 6). Bisphosphonate treatment significantly increased bone mass (Fig. 6, A and B), and, interestingly, mobilization of HSPCs as determined by a colony-forming assay was highly induced in bisphosphonate- compared with vehicle-treated mice (Fig. 6 C). Thus, bisphosphonate therapy increased bone mass and up-regulated HSPC mobilization. Similar phenotypes were detected in mice treated with neutralizing RANKL antibody compared with isotype-matched control antibody-treated mice (Fig. 6, D–F). Thus our results demonstrate that HSCs can be expanded pharmacologically. Treatment with zoledronate, another bisphosphonate, also reportedly increased HSPC mobilization in response to G-CSF (Winkler et al., 2010), further suggesting that bisphosphonate, neutralizing RANKL antibody, or other osteoclast inhibitors might serve as adjuvants to increase HSPC mobilization.

Collectively, our results challenge the dogma that BM cavity–deficient animals cannot maintain HSCs and raises the question of why mammalian and avian species developed BM cavities. Although the density of osteopetrotic bones in op/op mice was greater than that seen in controls (Fig. S5 A), strength tests indicated that bones were more fragile than control bones (Fig. S5 B). Fractures were induced in osteopetrotic bones earlier and by lower energy than in control bones, based on constant displacement tests (Fig. S5, B and C). Osteopetrotic bone also showed reduced elasticity (Fig. S5 D). H/E staining indicated that op/op mouse bones were filled with trabecular bone and that mutants exhibited thinner cortical bones than did control mice (Fig. S5, E and F). Indeed, osteopetrotic patients suffer from frequent fractures (Landa et al., 2007). Thus bones likely developed not only to maintain HSCs but to optimize bone strength and body support. To overcome the reduced ability to support HSCs in an open BM cavity created by osteoclasts, BM niches were likely developed by osteoblastic cells, vascular endothelial cells, reticular cells, or aggregates of these cells (Adams and Scadden, 2006; Kiel and Morrison, 2008), or by the products of these cells such as angiopoietin 1, osteopontin, Cxcl12, and thrombopoietin 1 (Calvi et al., 2003; Zhang et al., 2003; Arai et al., 2004; Kiel et al., 2005; Stier et al., 2005; Adams and Scadden, 2006; Sugiyama et al., 2006; Kiel and Morrison, 2008; Lymperi et al., 2010).

Manipulating HSCs in vitro to increase their number likely results in reduced stemness because quiescence is essential to maintain stem cell function (Cheng et al., 2000; Miyamoto et al., 2007). Thus, there is likely an antagonistic relationship between expansion and quiescence in HSC expansion. Bone mineral density (BMD) decreases with age, as osteoporosis is increased by osteoclastic activity (Manolagas and Jilka, 1995; Teitelbaum, 2007). Hematopoietic activity also decreases with age (Geiger and Rudolph, 2009; Waterstrat and Van Zant, 2009), suggesting that bone aging can cause both reduced bone mass and reduced hematopoiesis, and that antagonizing osteoclasts pharmacologically could promote increased bone mass and stimulate an induced niche in vivo to increase the HSC pool.

op/op, c-Fos–deficient, and RANKL-deficient mice were generated by crossing heterozygotes of respective genotypes. c-Fos-deficient mice were provided by E. Wagner (Institute of Molecular Biotechnology of the Austrian Academy of Sciences, Vienna, Austria). Animals were maintained under pathogen-free conditions in animal facilities certified by the Keio University School of Medicine animal care committee. To analyze blood counts, peripheral blood from the postorbital vein was collected in heparinized microtubes (Drummond Scientific). 5-FU (Kyowa Hakko Kirinn) was administered intravenously at 150 mg/kg to 8–12-wk-old op/op, c-Fos–deficient, or littermate control mice once a week, and mouse survival was monitored daily. Animal protocols were approved by the Keio University School of Medicine animal care committee.

8–12-wk-old mice were necropsied, and their hindlimbs were removed, fixed in 70% ethanol, and subjected to dual energy x-ray absorptiometry analysis to measure BMD and for microradiographic analysis. BMD was shown in 20 longitudinal divisions of femurs.

To induce HSPC mobilization, mice were injected with recombinant human G-CSF (Kyowa Hakko Kirin Co.; 250 µg/kg/d for 5 d, subcutaneous injection). Peripheral blood was harvested 3 h after the last G-CSF injection and used for blood counts, determination of colony-forming units in culture (CFU-C), and flow cytometric (FACS) analysis. Blood counts were analyzed using CellTac (Nihon Kohden). For CFU-C assays, cells were cultured in methylcellulose medium containing recombinant mouse (rm) stem cell factor, rm IL-3, recombinant human (rh) IL-6, and rh erythropoietin (MethoCult GF M3434; STEMCELL Technologies) in 35-mm culture dishes (Falcon) and incubated at 37°C in 5% CO₂. The number of CFU-Cs was determined on day 7 using a microscope (Olympus). FACS analysis was undertaken as described. For some experiments, Alendronate or RANKL neutralizing antibody was administered. 5 µg alendronate/mouse/d was injected subcutaneously for 14 d before G-CSF injection. 5 mg/kg RANKL neutralizing antibody (Oriental Yeast Co.) or isotype control antibody (BD) at 14 d and 2.5 mg/kg at 1 d before G-CSF injection was injected into the peritoneal cavity. For competitive repopulation assays, peripheral blood from individual CD45.2⁺ mice was pooled, and 20 µl of pooled blood was mixed with 500,000 competitive BM cells from untreated CD45.1⁺ mice and injected intravenously into lethally irradiated CD45.1⁺ recipient mice. After 3 mo of transplantation, peripheral blood reconstitution in multiple lineages was determined by FACS using CD45.1 (A20), CD45.2 (104), Mac1 (M1/70), CD3 (500A2), and B220 (RA3-6B2) antibodies (all BD).

mAbs (all BD) recognizing the following markers were used for flow cytometric analyses and cell sorting: c-Kit (2B8), Sca-1 (E13-161.7), CD3 (500A2) B220 (RA3-6B2), TER-119 (Ly-76), Gr-1 (RB6-8C5), and Mac-1 (M1/70). A mixture of mAbs recognizing CD3, B220, TER-119, Mac-1, or Gr-1 was used to identify lineage⁺ cells. Mature myeloid cells were stained with PEcy7-conjugated anti-CD11b (M1/70; eBioscience), Alexa Fluor 488–conjugated anti-F4/80 (CI:A3-1; BioLegend), and PE-conjugated anti–Ly6-G antibody (1A8; BD). Stained cells were analyzed on a FACS Aria2 machine (BD).

Serum M-CSF levels were assayed using the Mouse M-CSF ELISA kit (R&D Systems), according to the manufacturer’s instructions.

Frozen sections of op/op mice femoral bone that had not been decalcified were stained with goat anti–c-Kit antibody (AF1356; R&D Systems), followed by Alexa Fluor 488–conjugated donkey anti–goat antibody (Invitrogen) to detect HSCs and with TOTO3 (Invitrogen) to detect nuclei. Samples were observed under a fluorescence microscope (IX70; Olympus). Paraffin sections of wild-type, op/op, c-Fos–deficient, and OPG-deficient mice femoral bone were stained with Alexa Fluor 488–conjugated anti-F4/80 antibody (BioLegend) and TOTO3 as a nuclear stain and then observed under a confocal microscope (FV1000; Olympus).

The femur was removed from mice and placed on the lower supports of a three point bending fixture with the anterior side facing up using an apparatus called the Bone Strength Tester MZ-500S (Maruto) installed with a load of 50 N. The span between the two lower supports was set at 6 mm. For all femoral specimens, the upper loading device was aligned to the center of the femoral shaft and the load was applied at a constant displacement rate of 10 mm/min until breakage occurred.

Fig. S1 demonstrates the sensitivity of c-Fos–deficient and op/op mice to 5-FU, and the bone phenotype of RANKL-deficient mice. Fig. S2 shows multilineage reconstitution of donor-derived cells in recipient mice. Fig. S3 demonstrates the frequency of the CD11b⁺F4/80⁺Ly6-G⁺ macrophage population in wild-type, op/op, c-Fos–deficient, and OPG-deficient mice, and the frequency of LSK cells in op/op mouse spleen. Fig. S4 demonstrates the presence of HSCs in osteopetrotic bone. Fig. S5 shows osteopetrotic bone strength.

We thank Professor Wagner for providing c-Fos–deficient mice and Drs. Momoshima, Ishikawa, and Mogoe, Mr. Yamazaki, and Mrs. K. Fukushima for technical support.

K. Miyamoto was supported by a grant-in-aid for Young Scientists (B). T. Miyamoto was supported by a grant-in-aid for Young Scientists (A) and by Precursory Research for Embryonic Science and Technology (PREST), the Takeda Science Foundation, the Nakatomi Foundation, the Inamori Foundation, and the Keio Kanrinmaru project, Japan.

The authors have no conflicting financial interests.