Osteoclasts degrade endosteal components and promote ... - Nature

© 2006 Nature Publishing Group http://www.nature.com/naturemedicine

ARTICLES

Osteoclasts degrade endosteal components and promote mobilization of hematopoietic progenitor cells Orit Kollet1, Ayelet Dar1, Shoham Shivtiel1, Alexander Kalinkovich1, Kfir Lapid1, Yejezkel Sztainberg1, Melania Tesio1, Robert M Samstein1, Polina Goichberg1, Asaf Spiegel1, Ari Elson2 & Tsvee Lapidot1 Here we investigated the potential role of bone-resorbing osteoclasts in homeostasis and stress-induced mobilization of hematopoietic progenitors. Different stress situations induced activity of osteoclasts (OCLs) along the stem cell–rich endosteum region of bone, secretion of proteolytic enzymes and mobilization of progenitors. Specific stimulation of OCLs with RANKL recruited mainly immature progenitors to the circulation in a CXCR4- and MMP-9–dependent manner; however, RANKL did not induce mobilization in young female PTPe-knockout mice with defective OCL bone adhesion and resorption. Inhibition of OCLs with calcitonin reduced progenitor egress in homeostasis, G-CSF mobilization and stress situations. RANKL-stimulated boneresorbing OCLs also reduced the stem cell niche components SDF-1, stem cell factor (SCF) and osteopontin along the endosteum, which was associated with progenitor mobilization. Finally, the major bone-resorbing proteinase, cathepsin K, also cleaved SDF-1 and SCF. Our findings indicate involvement of OCLs in selective progenitor recruitment as part of homeostasis and host defense, linking bone remodeling with regulation of hematopoiesis.

Bone remodeling is a tightly regulated process involving bone-resorbing OCLs and bone-forming osteoblasts. OCLs are derived from hematopoietic stem cells and require RANKL signals from osteoblasts for their proliferation, differentiation and bone-resorbing activity1,2. OCL-osteoblast interactions are carried out on the bone surface including the endosteum region, in close proximity to specialized niches harboring hematopoietic stem and progenitor cells. In mice, endosteal bone-lining osteoblasts were shown to physically support hematopoietic stem cells, transferring signals that preserve undifferentiated stem cells with self-renewal potential. These signals comprise interplay between cytokines, chemokines, proteolytic enzymes and adhesion molecules, which anchor the stem cells to their specialized osteoblast-supportive microenvironment (reviewed in ref. 3). The chemokine SDF-1 is highly expressed by endosteal osteoblasts and bone marrow endothelial cells3,4. This ligand and its receptor CXCR4 have a pivotal role in multiple checkpoints of stem cell biology, including survival, proliferation and anchorage to the endosteal and endothelial microenvironment3,5,6. Whereas the vast majority of the stem cells are located within the bone marrow, a small progenitor population is constitutively released to the circulation; however, the mechanisms of this release are undefined. Stress situations such as inflammation, injury, chemotherapy or other clinical protocols such as treatment with granulocyte colony-stimulating factor (G-CSF) trigger the imbalance of steady-state homeostasis and induce massive stem cell mobilization. These events involve activation of proteolytic enzymes, which cleave the adhesive interactions between stem cells and their bone marrow microenvironment. Consequently, maturing

and immature leukocytes proliferate and are mobilized from the bone marrow reservoir to the circulation as part of host defense and organ repair (reviewed in ref. 5). Mobilization with repeated stimulations of G-CSF triggers a transient increase in secretion of SDF-1, massive progenitor proliferation and differentiation, activation of neutrophils, secretion of membrane-bound SCF5,7, and increased secretion of proteolytic enzymes including cathepsin G, elastase and MMP-9. These are followed by degradation of SDF-1, increased expression of CXCR4 and mobilization of maturing leukocytes, progenitors and stem cells8,9. Neutropenic individuals receiving G-CSF have increased OCL activity. G-CSF–treated mice have higher numbers of active OCLs and increased bone resorption; however, no direct effect of OCLs on progenitor mobilization has been found10,11. Active OCLs are identified by expression of the phosphatase TRAP, and they secrete the mobilizing cytokine interleukin (IL)-8 (ref. 12) and the proteolytic enzyme MMP-9, which are also involved in progenitor mobilization7,13,14 and migration of OCL precursors to bone-resorption sites in the developing fetal endosteum15. In addition, MMP-9–knockout mice suffer delayed ossification of growth-plate cartilage16, suggesting a role for this enzyme in bone remodeling. OCLs secrete other proteases, amongst them the major bone-resorbing enzyme cathepsin K (CTK), which is essential for bone remodeling and degradation of type I collagen17. The bone-related protein osteopontin is involved in regulation of stem cell migration, homing, anchorage, proliferation and quiescence in the endosteum region18,19 and is expressed by OCLs as well20.

1Department of Immunology and 2Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, 76100, Israel. Correspondence should be addressed to T.L. ([email protected]).

Received 1 February; accepted 25 April; published online 21 May 2006; doi:10.1038/nm1417

NATURE MEDICINE

VOLUME 12

[

NUMBER 6

[

JUNE 2006

657

We hypothesized that because of their location in close proximity to the stem cell–rich endosteum region and high degradation activity, OCLs may be physiologically involved in stress-induced mobilization of progenitors. Strain-dependent mouse knockout models with defective OCL function show a smaller bone marrow cavity associated with lower bone marrow cellularity and splenomegaly21, which indicate altered steady-state hematopoiesis. We therefore used young PTPe-knockout female mice, which have a temporary and mild phenotype of defective OCL bone adhesion and resorption22 with no identified hematopoietic disorders. Our results indicate a unique role and direct contribution of OCLs to progenitor mobilization, involving reduction in the stem cell niche components osteopontin, SCF and SDF-1. RESULTS Bleeding and LPS promote OCL formation and mobilization Bleeding results in accelerated production of blood cells, aimed at replenishing the lost cells. We used controlled bleeding as a mild stress. A considerable number of endosteal TRAP+ OCLs appeared 3–14 d after bleeding, peaking on day 7 (Fig. 1a and Supplementary Methods online). The frequency of circulating progenitors doubled (Fig. 1b) and bone marrow SDF-1 was significantly reduced in comparison to control mice (Fig. 1c). Next, we tested whether treatment with lipopolysaccharide (LPS), applied to mimic bacterial infection, also affects hematopoietic progenitors. Five days after a single injection of LPS, the number and size of TRAP+ active OCLs in the femur metaphysial endosteum region were increased (Fig. 1d). OCL precursors were also expanded in the bone marrow (Fig. 1e). In line with previous reports23, we documented higher numbers of circulating leukocytes (twofold increase; data not shown). The frequency of colony-forming cells in the peripheral blood was significantly increased (eightfold; Fig. 1f), indicating extensive progenitor mobilization from the bone marrow to the circulation. SDF-1 expressed in the bone marrow of LPS-treated mice was reduced by 66% (Fig. 1g). Together, these results suggest that various stress conditions share a common pathway: extensive OCL development accompanied by reduced SDF-1 levels in the bone marrow, in parallel with

progenitor recruitment to the circulation. We next sought to identify candidate mediators involved in the potential cross-talk between these pathways. SDF-1 and HGF promote OCL formation and mobilization SDF-1 and CXCR4 are expressed by primary human OCLs24,25 and the mouse pre-OCL cell line RAW26. Similarly, we found that primary mouse multinucleated TRAP+ OCLs express CXCR4 and SDF-1 (Fig. 2a,b), suggesting autocrine and paracrine regulation by SDF-1–producing cells. Hepatocyte growth factor (HGF), which increases surface expression of CXCR4 and migration of human progenitors toward SDF-1 (ref. 14), is highly expressed after G-CSF– induced mobilization27, and is involved in osteoblast and OCL proliferation and activation28. We therefore injected SDF-1 or HGF for five consecutive days. Both factors induced a substantial increase in the number of OCL precursors in the bone marrow (Fig. 2c). In parallel, we observed a significant increase in the frequency of circulating progenitors on day 5 (Fig. 2d). In addition, we documented considerable formation of TRAP+ OCLs along the endosteum of femoral bone trabecules (Fig. 2e–g).

658

SDF-1 (ng/mg total protein)

PB progenitors (colonies/2 × 105 cells)


PB progenitors (colonies/2 × 105 cells)

RANKL-stimulated OCLs promote progenitor mobilization For more selective stimulation of OCLs, we administered the specific OCL-differentiating cytokine RANKL in vivo. RANKL administration for 3 d followed by 2 d rest led to extensive development of TRAP+ OCLs along the trabecular bone surface. We saw abundant TRAP+ OCLs in the endosteal bone shaft in RANKL-treated but not in control mice (Fig. 3a). We found that RANKL-induced mobilization was different from the widely used G-CSF treatment, known to induce massive proliferation, differentiation and mobilization of maturing bone marrow neutrophils. Differential counts of peripheral blood cells from wild-type mice treated with RANKL showed only a slight increase in the numbers of total white blood cells (WBCs) with no increase in neutrophil numbers (Fig. 3b). In contrast, RANKL specifically and preferentially increased the levels of circulating colony-forming progenitors in a time- (Fig. 3c) and dose- (data not shown) dependent manner. Notably, the levels of bone marrow progenitors were significantly increased as well (Fig. 3c). The more primitive Lin–Sca1+c-kit+ progenitor cells, previously shown to a BM b c B 1.4 30 contain most of the long-term repopulating 1.2 * 25 activity and are therefore enriched with stem 1.0 20 * 0.8 cells, showed similar increase in response to 15 0.6 B BM 10 RANKL stimulations both in the bone mar0.4 5 0.2 row and the peripheral blood, confirming 0 0 Ctrl Bleed Ctrl Bleed mobilization (Fig. 3d). We next tested whether inhibition of OCLs could induce the d e 250 f 120 g 1.4 B B * opposite effect. We injected mice with calci1.2 * 100 200 1.0 80 tonin, which is known to inhibit formation of BM 250 0.8 60 0.6 the OCL ruffle border and production of BM 200 * 40 0.4 50 TRAP29 and is used clinically to treat osteo20 0.2 0 0 0 porotic individuals. Although injections of Ctrl LPS Ctrl LPS Ctrl LPS calcitonin did not reduce the total WBC count of treated mice, this drug significantly Figure 1 Bleeding and LPS trigger mobilization of progenitors and formation of OCLs. (a) TRAP staining of femoral metaphysis of untreated mouse (left) and a mouse 7 d after bleeding (right). reduced the numbers of circulating progeniArrowheads indicate active TRAP+ OCLs stained in red (original magnification, 200). BM, bone tors under homeostatic conditions and, moremarrow; B, bone. (b) Circulating mobilized progenitors. PB, peripheral blood. (c) SDF-1 in bone over, was able to reduce G-CSF–induced marrow fluids. Data are mean ± s.e.m., *P o 0.05, comparing bled to control mice. (d) TRAP mobilization and LPS-induced recruitment staining of femoral metaphysis of control (left) and LPS-injected mouse (right). (e) TRAP+ of colony-forming cells (Fig. 3e). To establish multinucleated cells developed from total bone marrow cells reflect OCL precursors in the bone these findings, we further applied a genetic marrow. (f) Circulating mobilized progenitors. (g) SDF-1 in bone marrow fluids. Data are approach, using female PTPe-knockout mice. mean ± s.e.m. *P o 0.05, comparing LPS treatment to control mice. OCL precursors (/2 × 105 BM cells)


ARTICLES

VOLUME 12

[

NUMBER 6

[

JUNE 2006

NATURE MEDICINE

ARTICLES

c SDF-1

d

*

e

*

f

*

80

g

B

60

B B

*

40

BM

20

BM

0

Ctrl SDF-1 HGF

BM

Ctrl SDF-1 HGF

Figure 2 SDF-1 and HGF induce formation of TRAP+ OCLs and mobilization of progenitors. (a,b) Primary OCLs immunolabeled (red) for CXCR4 (a) and SDF-1 (b), and stained for polymerized actin (green) and nuclear DNA (blue). Pink results from merging of red and blue. Bone marrow (BM) OCL precursors (c) and circulating progenitors (d). PB, peripheral blood. Data are mean ± s.e.m., *P o 0.05, comparing treated to control mice. TRAP staining of femoral metaphysis of control (e), SDF-1–injected (f) and HGF-injected (g) mice. BM, bone marrow; B, bone. Arrowheads indicate TRAP+ active OCLs (original magnification, 200).

young female PTPe-knockout mice, we observed no mobilization (Fig. 3f). A modest but significantly lower frequency of circulating progenitors in untreated female knockout mice (Fig. 3f) suggests reduced release of progenitors to the circulation of these mice in steady-state homeostasis. Notably, the numbers of Lin–Sca-1+c-kit+ primitive progenitors in the bone marrow of the young female knockout mice increased in response to stimulation with RANKL; however, no increase was seen in the circulation (Fig. 3g). Comparing G-CSF–induced mobilization in wild-type versus knockout females or knockout males (with a normal phenotype) versus knockout females provided additional confirmation of our hypothesis. In both cases, wild-type females and knockout males with normal OCL function had

b 60

a

50

B

BM

B

B

40 30

10

104

103

103

102

102

101

101

10

RANKL 26

PB

102

101

101

*

6 4 2 0

100 100 100 101 102 103 104 100 101 102 103 104 104 104 0.37% 1.1% 103 103 102

8

f

BM

100 100 100 101 102 103 104 100 101 102 103 104

Sca-1

Ctrl 40 35 30 25 20 15 10 5 0

CT

*

G-CSF WT KO

**

Ctrl

*

G-CSF + CT

h

RANKL

90 80 70 60 50 40 30 20 10 0

Ctrl

RANKL

*

LPS

LPS+CT

150

*

*

50

*

BM

160 140 120

*

100 80 60 40 20

3d

5d –

0

3d

RANKL

104

*

200

100

G-CSF

50 PB 45 40 35 30 25 20 15 10 5 0 –

g

250

PB progenitors (/2 × 105 cells)

9

12


Ctrl

50 45 40 35 30 25 20 15 10 5 0

14


c-kit

104

e

**

20

0

d

c *

c-kit

BM

PB cells

BM

WBC (× 103/µl) Neut (percent)


Reduced mobilization in mice with defective OCLs Young female PTPe-knockout mice have defective OCL function owing to impaired adhesion, which leads to reduced bone resorption, a mild phenotype limited to young females22. Because the number of total WBCs and the percentage of immature Lin– cells in the bone marrow of these mice were similar to their wild-type counterparts (data not shown), with no other reported defects relevant to hematopoiesis, we selected this model to address the potential role of OCL activity in progenitor mobilization. In wild-type counterparts, RANKL treatment increased the total numbers (1.5-fold; data not shown) and more profoundly the frequency (2.5-fold) of mobilized, circulating progenitors. Notably, when we administered RANKL to

BM progenitors (/1.5 × 104 cells)


CXCR4

300 250 200 150 100 50 0


b

OCL precursors (/2 × 105 BM cells)

a

Ctrl 6

104

103

103

102

102

101

101

RANKL 8

PB

100 100 100 101 102 103 104 100 101 102 103 104 104 104 0.38% 0.9% 103 103 102

102

101

101

BM

100 100 100 101 102 103 104 100 101 102 103 104

Sca-1

0 WT KO

M

F

Figure 3 RANKL-mediated mobilization of progenitors. (a) TRAP staining of femur metaphysis of control mouse (left), and a mouse injected with RANKL (middle and right). TRAP+ OCLs are on trabecules of the metaphysis (left, middle) and along the cortical bone shaft (right). Original magnification, 200. (b) Total WBC counts and neutrophils in the blood of wild-type mice left untreated (Ctrl), injected with RANKL, or treated for 3 d with G-CSF. (c) Circulating (left) and bone marrow (BM; right) progenitors of control (–) and RANKL-treated mice. Data are mean ± s.e.m., *P o 0.04. (d) Flow cytometry analysis of mononuclear cells derived from peripheral blood (top) or bone marrow (bottom) of control (left) or RANKL-treated (right) wild-type mice. Values represent the number (blood) or percentage (bone marrow) of Lin–Sca-1+c-kit+ cells per 106 total acquired cells. (e) Circulating progenitors of wild-type mice untreated or injected with calcitonin (CT) alone (left), in combination with G-CSF (middle) or LPS (right). (f) Circulating progenitors in control or RANKLtreated female PTPe-knockout mice (white; KO) or wild-type mice (black; WT). (g) Lin–Sca-1+c-kit+ staining of female PTPe-knockout mice, as described in d. (h) Circulating progenitors in G-CSF–treated wild-type versus female PTPe-knockout mice (left), or male versus female PTPe-knockout mice (right). PB, peripheral blood.

NATURE MEDICINE

VOLUME 12

[

NUMBER 6

[

JUNE 2006

659

b

1.5 1 0.5 Ctrl

1.2 1

40 30 20 10 0

Ctrl RANKL

e

d

*

SCF

B

im

OPN

B

*

120 80

0.2 0

Ctrl RANKL

40

80

*

60 40 20 0

Ctrl RANKL

Ctrl CTK

RANKL

known to block SDF-1 signaling and reduce G-CSF–induced progenitor mobilization8 were also coadministered with RANKL. These inhibitors considerably reduced RANKL-mediated mobilization (Fig. 4c). In vivo stimulation with RANKL led to accumulation of pyridinoline (PYD), a bone collagen degradation product released to the circulation during bone resorption (Fig. 4d). This OCL-resorbing activity prompted us to evaluate several components of the stem cell niche, shown to anchor stem cells to the endosteum, and control their quiescence, self-renewal and survival (reviewed in ref. 3). Firstly, substantial reductions of osteopontin expression by endosteal bone-lining osteoblasts accompanied this RANKL treatment of mice (Fig. 4e). Secondly, SCF expression by these osteoblasts was also reduced (Fig. 4e), even though we found that the plasma levels of SCF are increased after RANKL treatment (Fig. 4f)—a result that is in line with a previously published study7. This differential

OCL-secreted enzymes reduce endosteal niche components Mice treated with RANKL showed increased expression of both MMP-9 (Fig. 4a) and the major OCL bone-resorbing enzyme cathepsin K (CTK; Fig. 4b). To evaluate the effect of these enzymes on OCL-mediated progenitor mobilization, we coinjected RANKL into mice either with a MMP-2/9–specific inhibitor or with the cysteine inhibitor E-64, which inhibits CTK. CXCR4-neutralizing antibodies

b

RANKL

Ctrl

Ctrl

c

RANKL

d

SDF-1

B

B

B

TRAP

e

f

g

N-term. K15C Poly αSDF-1 7.5

15

30

60 (hours) 0

5

10

14 16

16 +E64

SDF-1–FITC (percent of control)

TRAP

1.2 1.0

*

0.8 0.6 0.4 0.2 0

h 100 80 60 40 20 0

40

*

30

*

20 10

0 SDF-1 – +

Ctrl RANKL

+ + + Ctrl CTK CTK CTK + E64 1 µg/ml 2 µg/ml

40

a 30 Counts

SDF-1


B

B

CTK (min)

0.4

100

better mobilization of progenitors in response to stimulation with G-CSF (Fig. 3h). These results prompted us to search for mechanisms whereby OCLs may facilitate progenitor mobilization, in addition to the known mobilization-attributed function of IL-8 and MMP-9 cleavage of membrane-bound SCF, both secreted by active OCLs (reviewed in ref. 5).

a

0.6

120 160

0

B

Ctrl

Figure 4 Mechanisms of mobilization induced by bone-resorbing OCLs. (a) Bone marrow (BM) MMP-9 secretion after RANKL injections. Pro–MMP-9 activity. (b) Bone marrow Ctsk mRNA (encoding CTK). (c) Circulating progenitors in untreated mice, mice treated with RANKL alone or in combination with CXCR4-specific antibody, MMP-2/9 inhibitor or E-64. (d) Plasma PYD in RANKL-treated mice, indicative of bone resorption. (e) Immunoreactivity (brown) of SCF (top) and osteopontin (bottom) in femoral metaphysial trabecules (B) of control (left) and RANKL-treated (right) mice. Arrowheads indicate endosteal bone-lining osteoblasts. Original magnification, 400. SCF levels assayed by ELISA in plasma from control and RANKL-treated mice (f) and of SCF incubated in vitro with CTK (g). Data are mean ± s.e.m., *P o 0.05.

*

g Plasma SCF (pg/ml)

f B

0.8

– MMP E-64 αR4 inh RANKL

Ctrl

CTK 18S

Pro MMP-9


1.4

0.8

RANKL

*

SCF (percent of control)

0

c

1.6

Migration (percent)

*

2

Ctsk mRNA (fold of control)

BM MMP-9 (fold increase)

2.5


a

Plasma PYD (nmol/L)

ARTICLES

*

c

20

b

10

0

30

60

0 100

180

CTK incubation time (min)

101 102 103 SDF-1–FITC

104

Figure 5 CTK cleaves SDF-1. (a) Top panels, serial sections of SDF-1 staining (brown) in cortical femur of control (left) or RANKL-treated (right) mice. Endosteal bone-lining osteoblasts (black arrowheads, left), retracted from the endosteum in a RANKL-treated mouse (white arrowheads, right). Bottom panels, TRAP staining (red, arrows) indicate OCL formation and localization in the endosteum of femoral cortical bone of control (left) or RANKL-treated (right) mice. Original magnification, 200. (b) Serial sections of femoral trabecules of control (original magnification, 400) and RANKL-treated (original magnification, 200) mice, stained as described in a. Arrowheads indicate SDF-1–expressing bone-lining osteoblasts. Arrows indicate TRAP+ OCLs. (c) Bone marrow SDF-1 of control or RANKL-treated mice. (d) Migration of G2 cells toward a gradient of intact SDF-1, SDF-1 incubated with CTK or with CTK preincubated with E-64. *P o 0.05. (e,f) Western blot analyses of intact SDF-1, SDF-1 incubated with CTK or CTK preincubated with E-64. N-terminal–specific monoclonal antibody (top), polyclonal antibody (bottom). (g) Fluorescence intensity of primary osteoblasts incubated with FITCconjugated SDF-1 and then with CTK for indicated time points. Data present Geo Mean fluorescence intensity of positive cells (B40%) compared to untreated cells. (h) Representative flow cytometry analysis. a, unlabeled; b, SDF–1-FITC at time 0; c, 180-min incubation with CTK.

660

VOLUME 12

[

NUMBER 6

[

JUNE 2006

NATURE MEDICINE

ARTICLES

Blood vessel

Functional SDF-1 Cleaved SDF-1 HGF

OPN

SCF

Bone marrow OCL precursor

Nonadhesive PTPε-deficient OCL

Permeabilized endothelium

Mobilized stem cells


MMP-9 IL-8

CTK

HGF SDF-1 RANKL

Anchored stem cells

Functional OCL Bone-lining osteoblasts

Endosteum

Bone

Figure 6 A proposed model for the role of bone-resorbing OCLs in stressinduced stem cell mobilization. Stress signals increase HGF and SDF-1 in the endosteum region. Responding to these signals as well as to RANKL stimulation, bone marrow OCL precursors differentiate, proliferate and develop into TRAP+ active OCL. Bone-lining osteoblasts are retracted from the endosteum region occupied now by TRAP+ functional OCLs, which secrete various enzymes including MMP-9 and CTK. Cleavage of SDF-1, osteopontin (OPN) and SCF on the osteoblast cell surface and the surrounding extracellular matrix weakens stem cell anchorage and alters signals provided by the niche. Preliminary in vitro results suggest that increased permeability of the endothelium induced by adjacent active OCLs in a MMP-9–dependent manner (data not shown) supports transmigration and mobilization of neighboring stem cells to the circulation.

expression can be explained by the potential of CTK to cleave SCF in vitro (Fig. 4g). These results strongly suggest proteolytic activity and imbalance of CXCR4–SDF-1 interactions as a major axis in progenitor mobilization resulting from enhanced OCL activity. Proteolytic processing and inactivation of SDF-1 by degrading enzymes such as elastase, cathepsin G, CD26, MT1-MMP and MMP-9 were previously reported as part of the mobilization process8,9,30,31; however, a potential role for CTK (primarily secreted by bone-resorbing OCLs) in regulation of SDF-1 expression in the bone marrow has not been investigated. OCL-secreted CTK cleaves and inactivates SDF-1 Femoral bone sections of control mice had bone-lining osteoblasts expressing SDF-1 along the endosteum of the cortical bone shaft, whereas no positive staining was observed at the same region of RANKL-treated mice (Fig. 5a). TRAP staining of serial sections showed that RANKL-induced OCLs along the endosteum (Fig. 5a) do not colocalize with either bone-lining osteoblasts or SDF-1– positive staining, suggesting retraction of osteoblasts owing to bonesurface occupancy by active OCLs. Metaphysial trabecules of control mice contained SDF-1–expressing endosteal osteoblasts (Fig. 5b) and TRAP+ OCLs (Fig. 5b) in a distinct pattern: SDF-1 was not expressed in close proximity to TRAP+ OCLs. We saw the same pattern in RANKL-treated mice, although with higher levels of OCLs (Fig. 5b). Of note, expression of SDF-1 in the bone marrow is higher in steadystate homeostasis compared to RANKL-stimulated mice (Fig. 5a,b). RANKL-treated mice also had slightly but significantly reduced levels of SDF-1 in their bone marrow fluids on day 5 (Fig. 5c). We therefore hypothesized a possible role for CTK in expression and function of SDF-1, in addition to its degradation by MMP-9. SDF-1 was incubated with CTK, and its chemotactic potential was tested. Although

NATURE MEDICINE

VOLUME 12

[

NUMBER 6

[

JUNE 2006

CTK inactivated SDF-1 and abrogated its chemotactic activity in a dose-dependent manner, preincubation of CTK with the cysteine protease inhibitor E-64 suppressed its enzymatic activity, enabling similar levels of cell migration as induced by untreated SDF-1 (Fig. 5d). G2 cells pretreated directly with CTK showed normal migration levels, excluding a direct effect of the enzyme on the migrating cells (data not shown). Western blot analyses using N terminus–specific monoclonal antibody (clone K15C) and a polyclonal antibody detecting the entire SDF-1 peptide showed that CTK cleaves the signaling N terminus of SDF-1 within the short time period of 30 min (Fig. 5e). Prolonged incubation periods resulted in fragmentation and total degradation of the chemokine, which could be prevented by preincubation of CTK with E-64 (Fig. 5f). SDF-1 is expressed and presented by various bone marrow cell types, including human and mouse endosteal bone-lining osteoblasts3,4. We evaluated the potential of CTK to cleave membranebound SDF-1 by incubating CTK with primary mouse osteoblasts, previously incubated with SDF-1 conjugated to FITC via its N terminus. Incubation with CTK gradually reduced the signal of FITC–SDF-1 bound to the cell membrane in a time-dependent manner (Fig. 5g,h). Together, our results suggest a regulatory role for OCLs and their proteolytic enzymes in the expression and membrane presentation of SCF, osteopontin and SDF-1 in the endosteum region during steadystate homeostasis and stress situations. In addition, CTK, the hallmark of bone degradation, also has a role in regulation of niche factors such as SDF-1 and SCF turnover and primitive progenitor mobilization (Fig. 6). DISCUSSION Here, we identified a unique role for bone-resorbing OCLs in selective egress of immature hematopoietic progenitors from the bone marrow to the circulation during homeostasis, G-CSF–induced mobilization and physiological stress–induced recruitment. First, we showed that stimulation of OCLs by two different stress signals induces mobilization of progenitors, whereas, when we impaired OCL function by genetic manipulation or by OCL inhibitors there was reduced homeostatic and stress-induced egress of primitive progenitors to the circulation. Second, our findings show that OCL-secreted MMP-9 and CTK are important regulators of progenitor mobilization. Finally, we document reduced expression of major components of the endosteal stem cell niche, previously shown to have a crucial role in stem cell anchorage, survival and quiescence3,6,7,18,19, after stimulation with RANKL. Mild bleeding exerts direct stress on the hematopoietic system and induces OCL proliferation32. Likewise, LPS-induced inflammation, mimicking bacterial infection, triggers OCL differentiation mediated by secretion of tumor necrosis factor (TNF)33 and robust progenitor egress to the circulation23. We found increased OCL precursors in the bone marrow and appearance of a considerable number of TRAP+ OCLs in the endosteum, a region of the stem cell niche. Meanwhile, enhanced HGF production is facilitated in G-CSF–induced mobilization of hematopoietic progenitors27, as well as with stimulation by LPS34 and liver injury35, two stress situations that are also associated with progenitor mobilization. This point is notable because HGF can directly upregulate expression of CXCR4 and migration of human CD34 progenitors14, affect both OCL and osteoblast differentiation28 and regulate SDF-1 production by bone marrow stromal cells36. SDF-1 can also induce progenitor mobilization, as previously reported37,38. Elevation of SDF-1 levels enhances OCL-mediated bone resorption and secretion of MMP-9 (ref. 39), which in turn

661


ARTICLES deactivates this chemokine. In contrast to bone marrow CD34 cells, circulating human CD34 progenitors secrete MMP-2 and MMP-9, suggesting that inhibitory signals are provided by the bone marrow and, moreover, that these enzymes have a role in progenitor cell egress or mobilization40. A transient increase in bone marrow SDF-1 in response to G-CSF is followed by its proteolytic degradation8. Thus, SDF-1 is required for enhanced secretion of MMP-9 and other proteolytic enzymes in the bone marrow, which later deactivate the anchoring potential of this chemokine as part of the mobilization process. Of note, G-CSF–treated bone marrow cells increased expression of CXCR4, and neutralization of the ligand or its receptor reduced mobilization8. Coinjection of CXCR4-neutralizing antibodies with RANKL may thus cause an imbalance of SDF-1 signals required for secretion of MMP-9 and progenitor mobilization. Others have shown that immature OCL precursors express CXCR4 and respond to SDF-1 by increasing secretion of MMP-9. In addition, OCL precursors follow SDF-1 signals, which are required for their recruitment from the blood to the bone marrow25,26. Thus, CXCR4 neutralization may also block OCL recruitment, leading to reduced bone-resorbing activity, lower levels of MMP-9 and decreased OCL-mediated progenitor mobilization. SDF-1 levels are also regulated by endothelialmediated transcytosis of functional SDF-1 from the blood to the bone marrow41 and apparently to tissues of other organs as well, which may mediate progenitor recruitment to the injured liver responding to signals mediated by this chemokine14. Direct stimulation of OCLs by RANKL or inhibition of OCLs by calcitonin, E-64 and young female PTPe-knockout mice provide strong evidence for the central role of bone-resorbing OCLs in homeostatic and stress-induced progenitor mobilization. RANKL also affects other cell types42 including tumor cells metastasizing to the bone marrow43. RANKL did not induce progenitor mobilization in female PTPe-knockout mice, showing the dominant effect of RANKL on OCL-induced mobilization. Calcitonin, clinically used to treat osteoporotic individuals, impairs formation of the ruffled border and production and secretion of TRAP in mature OCLs29. Lower levels of steady-state circulating progenitors in young female PTPeknockout mice and further decreases in progenitor levels induced by calcitonin in wild-type mice indicate involvement of OCLs in steadystate release of progenitors. Moreover, this inhibitor was able to reduce G-CSF– and LPS-mediated mobilization of progenitors, indicating the need for bone-resorbing OCLs for optimal progenitor mobilization. Bone-resorbing OCLs secrete enzymes, including MMP-9 (ref. 44) and CTK17, that allow them to have bone-resorbing activity, but also give them SDF-1–degradation capacity. Notably, CTK and MMP-9 are also needed for optimal progenitor mobilization, as inhibition of each enzyme reduced RANKL-induced mobilization. But because these enzymes independently contribute to the process of bone remodeling, we assume that both participate in OCL-mediated progenitor mobilization in a nonexclusive manner. Immunostaining of the endosteum region show no expression of SDF-1 in close proximity to TRAP+ OCLs, suggesting local proteolytic degradation. Osteoblast retraction resulting from surface occupancy by OCLs, shown by others in vitro45, may explain the disappearance of SDF-1–expressing, bone-lining osteoblasts and the appearance of TRAP+ OCLs in the bone-shaft endosteum of RANKL-treated mice. Expression of other components of the endosteal stem cell niche, such as membrane-bound SCF and osteopontin, previously shown to have a crucial role in stem cell anchorage, survival and quiescence in the bone marrow3,7,18,19 is reduced after stimulation with RANKL. Notably, G-CSF treatment reduces SDF-1 transcription and the number of endosteal osteoblasts46,47, a process that may involve

662

OCL activity. These findings imply that the stem cell niche and its signals are functionally altered. Endosteal osteoblasts secrete osteopontin. This glycoprotein is thought to be a key regulator of the hematopoietic stem cell endosteal niche and is thought to be involved in stem cell self-renewal and localization to the endosteum. Additionally, osteopontin negatively regulates and limits the number of endosteal stem cells3, which may explain the increased numbers of progenitors and primitive Lin–Sca-1+c-kit+ cells we observed in the bone marrow of RANKL-treated mice. Thus, the suggested cleavage of osteopontin by CTK21 in the endosteum may participate in progenitor mobilization. Cleavage of membrane-bound SCF from the endosteum and its increase in the plasma were shown to mediate progenitor mobilization7. Extramedullar hematopoiesis resulting from reduced bone marrow cavities in CTK-knockout mice21 indicate the crucial role of OCLs and their proteolytic enzymes for bone marrow hematopoiesis. This process, which is amplified by stress signals, links bone remodeling and progenitor recruitment to the immune system as part of host defense and repair. Notably, the calcium-sensing receptor (CaR) is essential for stem cell migration and lodgment in their endosteal niche, suggesting preferential localization of homing stem cells in close proximity to calcium-releasing OCLs48. The major bone-resorbing protease CTK cleaves bone matrix proteins such as type I collagen17. We identified N-terminal cleavage and functional inactivation of SDF-1 by CTK, similar to cathepsin G and elastase, previously shown to be involved in G-CSF–mediated mobilization8,9, and thus propose this enzyme as a new player involved in progenitor mobilization. CTK, which also cleaves SCF, as we show here, is colocalized with TRAP in transcytotic vesicles of the active OCLs, transporting bone degradation products from the bone surface to the extracellular space facing the bone marrow49,50. We could partially block RANKL-induced mobilization by coinjection of a MMP-2/9–specific inhibitor, indicating a role for MMP-9 in the mobilization process. Other stress conditions resulting from DNA damage induced by chemotherapy treatment37 or liver injury14 induce SDF-1 secretion, which in turn triggers secretion of MMP-9. These results suggest that MMP-9 is involved in facilitating both bone remodeling16 and increased migration and mobilization of bone marrow progenitors. In summary, we propose unique roles for bone-resorbing OCLs in the regulation of homeostatic and stress-induced recruitment and localization of primitive progenitors. Our results suggest that boneresorbing OCLs trigger proliferation of immature and primitive bone marrow progenitors, resulting in their constitutive release to the circulation. We link bone remodeling with regulation of stress-induced mobilization of progenitors for the first time and propose that the delicate osteoblast-OCL balance is also a major regulator of hematopoiesis and turnover of the endosteal stem cell niche. In addition to the potential role of OCL stimulation for treatment of increased bone mass in osteopetrotic disorders, RANKL should be considered together with other mobilizing agents aimed at selective mobilization of primitive stem and progenitor cells for a broad range of clinical transplantation protocols, in particular for poorly-mobilizing elderly or chemotherapy-treated individuals. METHODS Mice and mobilization protocols. Experiments with Balb/c, C57BL/6 and PTPe-knockout female22 (Supplementary Methods online) mice were approved by the Weizmann Institutional Animal Care and Use Committee. We bled mice or injected them with mouse SDF-1, human HGF, LPS, G-CSF or RANKL to induce mobilization (Supplementary Methods). We used the following inhibitors in this study: calcitonin, CXCR4-specific antibody, MMP2/9 inhibitor and E-64 (Supplementary Methods). We harvested peripheral

VOLUME 12

[

NUMBER 6

[

JUNE 2006

NATURE MEDICINE

ARTICLES blood and subjected it to Ficoll separation. We either fixed bones for histology or flushed them either with PBS to obtain live bone marrow cells or with Trireagent for RNA extraction. We repeated experiments at least three times, with three mice per group.


TRAP staining. We fixed mouse femurs (4% formaldehyde in PBS, 48 h, 24 1C), decalcified them (10% EDTA, pH 7.5, 7–10 d) and embedded them in paraffin. We deparaffinized 5 mm sections, rehydrated and stained them for TRAP activity (Sigma) according to the manufacturer’s instructions. Colony forming assay. See Supplementary Methods. Differential WBC counts. We collected blood samples immediately after CO2 inhalation to standard BD Vacutainer tubes (Becton Dickinson). Blood cell analyses were performed by ADVIA 120 (Bayer). OCL precursor quantification. We seeded total bone marrow cells obtained from untreated or treated mice (2 105 cells/200 ml, in a 96-well plate) in a-MEM (Sigma) supplemented with recombinant mouse macrophage colonystimulating factor (M-CSF, 20 ng/ml; PeproTech) and mouse RANKL (20 ng/ml, R&D Systems). We replaced culture medium every other day and stained plates for TRAP activity (Sigma) according to the manufacturer’s instructions on the fifth day. We scored TRAP+ osteoclasts containing 44 nuclei. Mouse SDF-1 ELISA. We flushed bones with PBS. Total protein content in bone marrow fluids was quantified by the Bradford assay and equal protein amounts were assayed for SDF-1 (ref. 8). Mouse SCF ELISA. We performed ELISA (R&D Systems) for mouse SCF to quantify protein levels in frozen plasma samples and of SCF cleaved by CTK. We incubated SCF (1 ng/ml) with CTK (2 mg/ml, 2 h, 37 1C) and subjected the mixture to ELISA. CTK did not cleave the ELISA capture and detection antibodies (data not shown). Plasma PYD. We tested plasma PYD (Metra, Quidel) on frozen plasma samples according to the manufacturer’s instructions. Chemotaxis assay. See Supplementary Methods. Flow cytometry. We used triple staining with FITC-conjugated antibodies indicating Lin+ phenotype (CD4, NK, CD8 purchased from eBiosciences; B220, CD11b, Gr-1 from Pharmingen), together with PE-conjugated Sca-1 (Pharmingen) and APC-conjugated c-kit (eBiosciences) to determine the levels of Lin–Sca-1+c-kit+ stem cells in peripheral blood and bone marrow mononuclear cells, acquiring 1 106 cells (FACSCalibur, BD Biosciences). Membrane-bound SDF-1 cleavage. We incubated SDF-1 conjugated to FITC via its N terminus (provided by N. Fujii, Kyoto University) with mouse calvaria-derived osteoblasts (1 mg/ml, 30 min., 37 1C). We washed cells, exposed them to CTK (2 mg/ml, 37 1C, for indicated time points) and analyzed levels of FITC–SDF-1 by flow cytometry (FACSCalibur). Semiquantitvative RT-PCR for CTK. We prepared cDNA from mouse bone marrow cells using standard protocols. We performed semiquantitative PCR analysis for CTK expression for 25 cycles: 95 1C for 30 s, 60 1C for 30 s, 72 1C for 1 min. We used the following primer sequences: 5¢-TCTGCTGCACGT ATTGGAAG-3¢, 5¢-GGCCTCTCTTGGTGTCCATA-3¢, 388 bp. MMP-9 zymography. We performed MMP-9 zymography as previously described14 with the following modifications: we loaded bone marrow fluids (0.5 mg protein as measured by Bradford assay) on 10% SDS-PAGE containing 1 mg/ml gelatin. Western blot analysis. We performed western blot analysis for SDF-1 as previously described8 with the following modifications: we incubated SDF-1 (20 ng; 37 1C) for indicated time points with 1.7 mg/ml CTK, in 30 ml reaction volume completed with PBS. When indicated, we preincubated CTK with E-64 (0.13 mM, 1.5 h, 37 1C), before we added SDF-1. We blotted the membrane with N-terminal SDF-1–specific monoclonal antibody (clone K15C) or a polyclonal antibody (R&D Systems).

NATURE MEDICINE

VOLUME 12

[

NUMBER 6

[

JUNE 2006

Immunohistochemistry. We immunolabeled fixed (4% formaldehyde in PBS, 48 h, 24 1C), decalcified (10% EDTA, pH 7.5, 7–10 d) and paraffin-embedded femoral bone sections using goat polyclonal antibody against human SDF-1 (20 mg/ml), goat antibody against mouse osteopontin (10 mg/ml) and goat antibody against mouse SCF (5 mg/ml), all from R&D Systems. We detected immunoreactivity using biotinylated goat-specific (DAKO) and LSAB2 avidin-biotin-DAB detection kit (DAKO) according to the manufacturer’s instructions. Immunocytochemical staining. We obtained bone marrow–derived primary mouse OCLs by coculture with mouse osteoblasts on cover slips as previously described22. On day 6, we gently removed osteoblasts after a 15-min incubation with aMEM supplemented with collagenase (0.1%)/dispase (0.2%) at 37 1C. We fixed OCLs with 3% paraformaldehyde (Merck) and permeabilized them with 0.5% Triton X-100 (Sigma). We indirectly immunolabeled samples at 24 1C in a humidified chamber with mouse antibody against human CXCR4 (12G5, R&D Systems) and rabbit antibody against human SDF-1 (R&D Systems), both of which are cross-reactive with mouse. We added the following secondary antibodies: Cy3-conjugated mouse-specific goat IgG or rabbitspecific goat IgG (Jackson ImmunoResearch Labs). FITC-phalloidin and DAPI (Sigma) were added. Images were acquired using scientific-grade CCD camera and processed by the DeltaVision system using Resolve3D software (Applied Precision). Statistical analysis. Significance levels were determined by two-tailed Student t-test analyses, using Microsoft Excel. Note: Supplementary information is available on the Nature Medicine website. ACKNOWLEDGMENTS The author extend a special thanks to A. Globerson, D. Zipori, R. Alon and S. Yung for critically reviewing the manuscript and to D. Rashkovan for her assistance. This work was supported in part by the Israel Science Foundation (to T.L. and O.K.) and Ares Serono (to T.L.). T.L. holds The Edith Arnoff Stein Professorial Chair in Stem Cell Research. AUTHOR CONTRIBUTIONS O.K. designed and performed the research, collected and analyzed data, and wrote the manuscript. A.D. designed and performed immunohistochemical staining and microscopic analysis. S.S. performed in vivo experiments, and collected and analyzed data. A.K. designed, performed and analyzed ELISAs. K.L. helped with TRAP staining. Y.S. helped with designing and conducting experiments. M.T. helped with mouse injections and colony assay. R.S. designed and carried out western blots and helped to define SDF-1 cleavage. P.G. contributed fluorescent staining and microscopy. A.S. helped with conducting and analyzing SDF-1 cleavage. A.E. contributed his PTPE-knockout mice. T.L. designed the research and wrote the manuscript. COMPETING INTERESTS STATEMENT The authors declare that they have no competing financial interests. Published online at http://www.nature.com/naturemedicine/ Reprints and permissions information is available online at http://npg.nature.com/ reprintsandpermissions/

1. Teitelbaum, S.L. Bone resorption by osteoclasts. Science 289, 1504–1508 (2000). 2. Boyle, W.J., Simonet, W.S. & Lacey, D.L. Osteoclast differentiation and activation. Nature 423, 337–342 (2003). 3. Adams, G.B. & Scadden, D.T. The hematopoietic stem cell in its place. Nat. Immunol. 7, 333–337 (2006). 4. Ponomaryov, T. et al. Induction of the chemokine stromal-derived factor-1 following DNA damage improves human stem cell function. J. Clin. Invest. 106, 1331–1339 (2000). 5. Cottler-Fox, M.H. et al. Stem cell mobilization. Hematology (Am. Soc. Hematol. Educ. Program) 419–437 (2003). 6. Lapidot, T., Dar, A. & Kollet, O. How do stem cells find their way home? Blood 106, 1901–1910 (2005). 7. Heissig, B. et al. Recruitment of stem and progenitor cells from the bone marrow niche requires mmp-9 mediated release of kit-ligand. Cell 109, 625–637 (2002). 8. Petit, I. et al. G-CSF induces stem cell mobilization by decreasing bone marrow SDF-1 and up-regulating CXCR4. Nat. Immunol. 3, 687–694 (2002).

663


ARTICLES 9. Levesque, J.P., Hendy, J., Takamatsu, Y., Simmons, P.J. & Bendall, L.J. Disruption of the CXCR4/CXCL12 chemotactic interaction during hematopoietic stem cell mobilization induced by GCSF or cyclophosphamide. J. Clin. Invest. 111, 187–196 (2003). 10. Takamatsu, Y. et al. Osteoclast-mediated bone resorption is stimulated during short-term administration of granulocyte colony-stimulating factor but is not responsible for hematopoietic progenitor cell mobilization. Blood 92, 3465–3473 (1998). 11. Watanabe, T. et al. Effect of granulocyte colony-stimulating factor on bone metabolism during peripheral blood stem cell mobilization. Int. J. Hematol. 77, 75–81 (2003). 12. Rothe, L. et al. Human osteoclasts and osteoclast-like cells synthesize and release high basal and inflammatory stimulated levels of the potent chemokine interleukin-8. Endocrinology 139, 4353–4363 (1998). 13. Pruijt, J.F. et al. Prevention of interleukin-8-induced mobilization of hematopoietic progenitor cells in rhesus monkeys by inhibitory antibodies against the metalloproteinase gelatinase B (MMP-9). Proc. Natl. Acad. Sci. USA 96, 10863–10868 (1999). 14. Kollet, O. et al. HGF, SDF-1, and MMP-9 are involved in stress-induced human CD34+ stem cell recruitment to the liver. J. Clin. Invest. 112, 160–169 (2003). 15. Blavier, L. & Delaisse, J.M. Matrix metalloproteinases are obligatory for the migration of preosteoclasts to the developing marrow cavity of primitive long bones. J. Cell Sci. 108, 3649–3659 (1995). 16. Vu, T.H. et al. MMP-9/gelatinase B is a key regulator of growth plate angiogenesis and apoptosis of hypertrophic chondrocytes. Cell 93, 411–422 (1998). 17. Goto, T., Yamaza, T. & Tanaka, T. Cathepsins in the osteoclast. J. Electron Microsc. (Tokyo) 52, 551–558 (2003). 18. Stier, S. et al. Osteopontin is a hematopoietic stem cell niche component that negatively regulates stem cell pool size. J. Exp. Med. 201, 1781–1791 (2005). 19. Nilsson, S.K. et al. Osteopontin, a key component of the hematopoietic stem cell niche and regulator of primitive hematopoietic progenitor cells. Blood 106, 1232–1239 (2005). 20. Merry, K., Dodds, R., Littlewood, A. & Gowen, M. Expression of osteopontin mRNA by osteoclasts and osteoblasts in modelling adult human bone. J. Cell Sci. 104, 1013–1020 (1993). 21. Gowen, M. et al. Cathepsin K knockout mice develop osteopetrosis due to a deficit in matrix degradation but not demineralization. J. Bone Miner. Res. 14, 1654–1663 (1999). 22. Chiusaroli, R. et al. Tyrosine phosphatase epsilon is a positive regulator of osteoclast function in vitro and in vivo. Mol. Biol. Cell 15, 234–244 (2004). 23. Molendijk, W.J., van Oudenaren, A., van Dijk, H., Daha, M.R. & Benner, R. Complement split product C5a mediates the lipopolysaccharide-induced mobilization of CFUs and haemopoietic progenitor cells, but not the mobilization induced by proteolytic enzymes. Cell Tissue Kinet. 19, 407–417 (1986). 24. Grassi, F. et al. Human osteoclasts express different CXC chemokines depending on cell culture substrate: molecular and immunocytochemical evidence of high levels of CXCL10 and CXCL12. Histochem. Cell Biol. 120, 391–400 (2003). 25. Wright, L.M. et al. Stromal cell-derived factor-1 binding to its chemokine receptor CXCR4 on precursor cells promotes the chemotactic recruitment, development and survival of human osteoclasts. Bone 36, 840–853 (2005). 26. Yu, X., Huang, Y., Collin-Osdoby, P. & Osdoby, P. Stromal cell-derived factor-1 (SDF-1) recruits osteoclast precursors by inducing chemotaxis, matrix metalloproteinase-9 (MMP-9) activity, and collagen transmigration. J. Bone Miner. Res. 18, 1404–1418 (2003). 27. Fujii, K. et al. Elevation of serum hepatocyte growth factor during granulocyte colonystimulating factor-induced peripheral blood stem cell mobilization. Br. J. Haematol. 124, 190–194 (2004). 28. Grano, M. et al. Hepatocyte growth factor is a coupling factor for osteoclasts and osteoblasts in vitro. Proc. Natl. Acad. Sci. USA 93, 7644–7648 (1996).

664

29. Zaidi, M., Inzerillo, A.M., Moonga, B.S., Bevis, P.J. & Huang, C.L. Forty years of calcitonin–where are we now? A tribute to the work of Iain Macintyre, FRS. Bone 30, 655–663 (2002). 30. Christopherson, K.W., Cooper, S. & Broxmeyer, H.E. Cell surface peptidase CD26/ DPPIV mediates G-CSF mobilization of mouse progenitor cells. Blood 101, 4680–4686 (2003). 31. McQuibban, G.A. et al. Matrix metalloproteinase activity inactivates the CXC chemokine stromal cell-derived factor-1. J. Biol. Chem. 276, 43503–43508 (2001). 32. Johnell, O. & Hulth, A. Proliferation of osteoclasts in rat bone following bleeding and femoral fractures. Calcif. Tissue Res. 23, 241–244 (1977). 33. Abu-Amer, Y., Ross, F.P., Edwards, J. & Teitelbaum, S.L. Lipopolysaccharide-stimulated osteoclastogenesis is mediated by tumor necrosis factor via its P55 receptor. J. Clin. Invest. 100, 1557–1565 (1997). 34. Crestani, B. et al. Differential role of neutrophils and alveolar macrophages in hepatocyte growth factor production in pulmonary fibrosis. Lab. Invest. 82, 1015– 1022 (2002). 35. Armbrust, T., Batusic, D., Xia, L. & Ramadori, G. Early gene expression of hepatocyte growth factor in mononuclear phagocytes of rat liver after administration of carbon tetrachloride. Liver 22, 486–494 (2002). 36. Matsuda-Hashii, Y. et al. Hepatocyte growth factor plays roles in the induction and autocrine maintenance of bone marrow stromal cell IL-11, SDF-1 alpha, and stem cell factor. Exp. Hematol. 32, 955–961 (2004). 37. Hattori, K. et al. Plasma elevation of stromal cell-derived factor-1 induces mobilization of mature and immature hematopoietic progenitor and stem cells. Blood 97, 3354– 3360 (2001). 38. Shen, H. et al. CXCR-4 desensitization is associated with tissue localization of hemopoietic progenitor cells. J. Immunol. 166, 5027–5033 (2001). 39. Grassi, F. et al. CXCL12 chemokine up-regulates bone resorption and MMP-9 release by human osteoclasts: CXCL12 levels are increased in synovial and bone tissue of rheumatoid arthritis patients. J. Cell. Physiol. 199, 244–251 (2004). 40. Janowska-Wieczorek, A., Matsuzaki, A. & Marquez, L.A. The hematopoietic microenvironment: matrix metalloproteinases in the hematopoietic microenvironment. Hematology 4, 515–527 (2000). 41. Dar, A. et al. Chemokine receptor CXCR4-dependent internalization and resecretion of functional chemokine SDF-1 by bone marrow endothelial and stromal cells. Nat. Immunol. 6, 1038–1046 (2005). 42. Lee, S.K., Gardner, A.E., Kalinowski, J.F., Jastrzebski, S.L. & Lorenzo, J.A. RANKLstimulated osteoclast-like cell formation in vitro is partially dependent on endogenous interleukin-1 production. Bone 38, 678–685 (2005). 43. Jones, D.H. et al. Regulation of cancer cell migration and bone metastasis by RANKL. Nature 440, 692–696 (2006). 44. Okada, Y. et al. Localization of matrix metalloproteinase 9 (92-kilodalton gelatinase/ type IV collagenase ¼ gelatinase B) in osteoclasts: implications for bone resorption. Lab. Invest. 72, 311–322 (1995). 45. Perez-Amodio, S., Beertsen, W. & Everts, V. (Pre-)osteoclasts induce retraction of osteoblasts before their fusion to osteoclasts. J. Bone Miner. Res. 19, 1722–1731 (2004). 46. Semerad, C.L. et al. G-CSF potently inhibits osteoblast activity and CXCL12 mRNA expression in the bone marrow. Blood 106, 3020–3027 (2005). 47. Katayama, Y. et al. Signals from the sympathetic nervous system regulate hematopoietic stem cell egress from bone marrow. Cell 124, 407–421 (2006). 48. Adams, G.B. et al. Stem cell engraftment at the endosteal niche is specified by the calcium-sensing receptor. Nature 439, 599–603 (2005). 49. Vaaraniemi, J. et al. Intracellular machinery for matrix degradation in bone-resorbing osteoclasts. J. Bone Miner. Res. 19, 1432–1440 (2004). 50. Kiviranta, R. et al. Impaired bone resorption in cathepsin K-deficient mice is partially compensated for by enhanced osteoclastogenesis and increased expression of other proteases via an increased RANKL/OPG ratio. Bone 36, 159–172 (2005).

VOLUME 12

[

NUMBER 6

[

JUNE 2006

NATURE MEDICINE