PPAR-g regulates osteoclastogenesis in mice - Nature

© 2007 Nature Publishing Group http://www.nature.com/naturemedicine

LETTERS

PPAR-g regulates osteoclastogenesis in mice Yihong Wan, Ling-Wa Chong & Ronald M Evans Osteoclasts are bone-resorbing cells derived from hematopoietic precursors of the monocyte-macrophage lineage. Regulation of osteoclast function is central to the understanding of bone diseases such as osteoporosis, rheumatoid arthritis and osteopetrosis1. Although peroxisome proliferator–activated receptor-c (PPAR-c) has been shown to inhibit osteoblast differentiation2,3, its role, if any, in osteoclasts is unknown. This is a clinically crucial question because PPAR-c agonists, ‘‘such as thiazolidinediones—’’ a class of insulin-sensitizing drugs, have been reported to cause a higher rate of fractures in human patients4,5. Here we have uncovered a pro-osteoclastogenic effect of PPAR-c by using a Tie2Cre/flox mouse model in which PPAR-c is deleted in osteoclasts but not in osteoblasts. These mice develop osteopetrosis characterized by increased bone mass, reduced medullary cavity space and extramedullary hematopoiesis in the spleen. These defects are the result of impaired osteoclast differentiation and compromised receptor activator of nuclear factor-jB ligand signaling and can be rescued by bone marrow transplantation. Moreover, ligand activation of PPAR-c by rosiglitazone exacerbates osteoclast differentiation in a receptor-dependent manner. Our examination of the underlying mechanisms suggested that PPAR-c functions as a direct regulator of c-fos expression, an essential mediator of osteoclastogenesis6. Therefore, PPAR-c and its ligands have a previously unrecognized role in promoting osteoclast differentiation and bone resorption. Bone is a dynamic tissue that undergoes constant remodeling, balancing bone formation by osteoblasts and bone resorption by osteoclasts. Osteoclasts are multinucleated cells of hematopoietic lineage; in contrast, osteoblasts are of mesenchymal origin. Defects in osteoclastic bone resorption cause osteopetrosis, a disease associated with an increased skeletal mass and abnormally dense bone. In severely affected individuals, the medullary cavity is filled with endochondral new bone, leaving little space for hematopoietic cells. This results in extramedullary hematopoiesis, which can be reversed by hematopoietic stem cell transplantation1. The nuclear receptor PPAR-g is an activator of adipogenesis7–9 and a repressor of osteoblastogenesis2,3. However, the specific role of PPAR-g in osteoclast function has not been fully explored. PPAR-g agonists, such as thiazolidinediones (TZDs), have been shown to cause bone loss in both mice and rats10–13, in part owing to increased bone resorption. In light of the increased rate of fractures reported in

diabetic individuals treated with TZDs in clinical studies including the recent A Diabetes Outcome Progression Trial (ADOPT)4,5, the potential role of PPAR-g in osteoclast function and bone resorption remains a clinically important issue. To achieve a genetic separation of the effect of PPAR-g deletion on the hematopoietic lineages from the one it has on the mesenchymal lineages, we used Tie2Cre mice14 and homozygous PPAR-g flox (Ppargflox/flox, or gf/f) mice15 to specifically delete the PPAR-g gene in hematopoietic and endothelial cells via excision of the loxP-flanked (floxed) Pparg alleles16,17. Tie2 is expressed in hematopoietic cells as early as 9.5 d postcoitum18. PCR analysis of genomic DNA showed that the floxed allele was efficiently deleted by Tie2Cre in all hematopoietic tissues, including the bone marrow, spleen, thymus and lymph node (Fig. 1a and Supplementary Fig. 1d online), but not in mesenchymal lineages, such as white adipose tissue. Consistent with these results, in the Tie2Cre/ROSA26-GFP Cre expression reporter mice, the flox deletion was virtually complete in each hematopoietic tissue, as evidenced by the shift of the entire FACS peak to GFP+ (Supplementary Fig. 1a). In addition, western blot analysis confirmed the reduction in PPAR-g protein abundance in both the bone marrow and the spleen in gf/f-Tie2Cre mutants (Supplementary Fig. 1b). Furthermore, quantitative RT-PCR analysis of individual cell types demonstrated that PPAR-g RNA expression was absent in osteoclasts and monocyte precursors, but remained normal in osteoblasts and adipocytes in the gf/f-Tie2Cre mutants (Supplementary Fig. 1c). Together, the genomic DNA, RNA, protein and FACS analyses show that PPAR-g expression was efficiently eliminated in the hematopoietic lineages but was unaltered in the mesenchymal lineages in gf/f-Tie2Cre mice. An examination of the gf/f-Tie2Cre mice revealed splenomegaly (Fig. 1b) and pale bones (Fig. 1c). The spleen–to–body weight ratio was increased by 1.4- to 2-fold in adult mice, as well as in 18- and 5-d-old pups (Fig. 1d). There was a pronounced accumulation of megakarocytes in the spleen, indicative of extramedullary hematopoiesis (Fig. 1b). Quantification of the nucleated cells from bone marrow and spleen indicated that the cellularity was decreased by 35% in bone marrow but increased by 63% in spleen (Fig. 1e). At the clonogenic progenitor cell level, both erythroid and granulocytic/monocytic progenitors were less abundant in bone marrow and more abundant in spleen (Fig. 1e). At the hematopoietic stem cell (HSC) level, the percentage of the Lin–cKit+Sca1+ stem cell population was reduced by 58% in bone marrow but elevated by 275% in spleen (Fig. 1e). When the changes in total cell numbers were taken into account, the absolute

Howard Hughes Medical Institute, Gene Expression Laboratory, Salk Institute for Biological Studies, La Jolla, California 92037, USA. Correspondence should be addressed to R.M.E. ([email protected]). Received 8 August; accepted 26 September; published online 2 December 2007; doi:10.1038/nm1672

1496

VOLUME 13

[

NUMBER 12

[

DECEMBER 2007

NATURE MEDICINE

LETTERS WBC differential count (Fig. 1h), suggesting that hematopoiesis was normal other than having shifted from bone to spleen. Adult hematopoiesis mainly occurs in the bone marrow and is regulated by the osseous environment1,20. Thus, we next examined bone volume and mineral density using microcomputed tomography (mCT) in vivo imaging (Fig. 2a). The bone volume fraction (BVF, or bone volume/tissue volume) was 40–70% higher in both the proximal and the distal regions of femur and tibia in the mutants (Fig. 2a),

c

Femur

Tibia

BM

Intestine

Brain

BAT

Muscle

LN

WAT

Ovary

Adrenal

Kidney

Spleen

Pancreas

Liver

Heart

Aorta

Thymus

Lung

a

flox f/– Tie2Cre+/–

Null

+Cre

flox f/– Null

b

–Cre

–Cre

+Cre

+Cre

0 PL(×100)

Spleen (×106)

BM (×106)

100 0 –Cre

1 0.8 0.6 0.4 0.2 0

15 Spleen (fold)

BM (fold) a2

200

+Cre

–Cre +Cre

10 5 0 E

GM

GM

at

a1

300

+Cre

E 0.05

100 –Cre +Cre

80 60 40 20 0

***

400

G

at G

a2 at

5

HGB

Spleen/body (%)

0

10

***

–Cre

1

h

RBC

+Cre

–Cre +Cre

G

at

a1

1 G

Lm

2

–Cre +Cre

WBC

0

100 80 60 40 20 0

Percentage of total cells

15

o2

0

0.2

1

1

0.4

Fo g

2

*

0.6

–Cre

Lm o2

–Cre +Cre

e

0.8

+Cre

WBC differential (%)

g

–Cre

Ta l1

3

Ta l1

Relative expression in BM

f

+Cre

Relative expression in Spleen

–Cre

***

1 0.8 0.6 0.4 0.2 0

Spleen/body (%)

***

0.5 0.4 0.3 0.2 0.1 0

Fo g

d

Spleen/body (%)

–Cre

CBC (×103/ul)


numbers of HSCs were decreased by 73% in bone marrow but increased by 511% in spleen. In addition, the expression of several transcription factors crucial for HSC differentiation and/or selfrenewal19 was also shifted from bone marrow to spleen (Fig. 1f). In total, these results show the presence of extramedullary hematopoiesis in the spleen of gf/f-Tie2Cre mice at both cellular and transcriptional levels. Notably, however, there were no significant differences in white blood cell (WBC), red blood cell and platelet counts (Fig. 1g) or in

***

0.04 0.03 0.02

–Cre +Cre

***

0.01 0

NE

LY

MO

EO

BA

BM

Spleen

Figure 1 Tie2Cre-mediated PPAR-g deletion leads to extramedullary hematopoiesis in the spleen. (a) Genotype PCR analysis of various tissues of flox/null mice with Tie2Cre (f/–Tie2Cre+/) or without Tie2Cre (f/–). The null allele was a recombined floxed allele and served as an internal DNA loading control. Tie2Cre converted the other floxed allele into a null allele in all hematopoietic tissues (yellow boxes) without affecting mesenchymal tissues. WAT, white adipose tissue; LN, lymph node; BAT, brown adipose tissue; BM, bone marrow. (b) Gross morphology (left) and histology (right) of the spleen. Arrows indicate megakarocytes; scale bar represents 50 mm. (c) Pale bones in the mutants (+Cre). (d) Spleen–to–body weight ratio in 10–18-week-old adults (left, n ¼ 13) and in 18-d-old (middle, n ¼ 6) and 5-d-old (right, n ¼ 6) pups. (e) Total cell number (top, n ¼ 9), progenitor cell number (middle, n ¼ 14) and hematopoietic stem cell percentage (bottom, n ¼ 8) in bone marrow (left) and spleen (right). BM, bone marrow; E, erythroid; GM, granulocyte/macrophage. (f) Expression of transcription factors regulating early hematopoiesis in bone marrow and spleen. Expression was normalized to that of b-actin. Tal1, T-cell acute lymphocytic leukemia 1; Lmo2, LIM domain only 2; Fog1, Friend of GATA-1; Gata1, GATA binding protein 1; Gata2, GATA binding protein 2. (g) Complete blood cell count. RBC, red blood cell; HGB, hemoglobin; PL, platelet. (h) White blood cell differential count. All changes in blood cell counts were not significant (n ¼ 9, P 4 0.05). NE, neutrophil; LY, lymphocyte; MO, monocyte; EO, eosinophil; BA, basophil. *P o 0.05; ***P o 0.005.

NATURE MEDICINE

VOLUME 13

[

NUMBER 12

[

DECEMBER 2007

1497

LETTERS indicating a markedly increased bone volume with a reduced medullary cavity space. A similar increase in bone mineral density (BMD; 30–70%) was also observed (Fig. 2a). Consistently with this, static bone histomorphometry showed increased trabecular volume, thickness and number, with a decreased trabecular separation (Fig. 2b).

*

*

0.8

0.4 0.2

0.2 Prox

n.s. Tibia BMD (mg/cc)

600 400 200

Dist

1,200

*

800

*

–Cre +Cre

400

X-ray 0

0 Prox 18 16 14 12 10 8 6 4 2 0

Dist 40

*

Prox

Tb.n (1/mm)

20 10

300

3 2 1

–Cre +Cre

*

200 150 100 50

H&E

0 –Cre +Cre

–Cre

d

250

0

0 –Cre +Cre

350

**

4

30 Tb.th (µm)

5

*

Dist

Tb.sp (µm)

Femur BMD (mg/cc)

µCT –Cre +Cre

0.4

Dist

*

800

BV / TV (%)

+Cre

0 Prox

b

–Cre

c

*

*

0.6

Tibia BVF

Femur BVF

0.6

0

–Cre +Cre

+Cre

–Cre

e

+Cre

B

B

BM

10

2

10 0

0

0 –Cre +Cre

4

–Cre +Cre

n.s.

100

2

1

0

0 –Cre +Cre

h

3 n.s.

200

–Cre +Cre

BM

B

B

150

90

*** 100

50

0

–Cre +Cre

60

30

0

–Cre +Cre

150

n.s. Osteocalcin (ng/ml)

20

6

300

BM

ALP (IU/L)

30

8

g

MAR (µm/d)

40

20

n.s.

nmol DPD/mmol creatinine

B

N.Ob/B.Pm

**

BM

B

Cortical

B

n.s. Ob.S/BS (%)

50

BM

10

30

60

BM

BFR / BS (µm3/µm2/y)

70

B

BM

Cortical

f

Trabecular

Trabecular

BM

N.Oc/mm2


a

Whole body X-ray analysis showed increased radio density in the skull, vertebrae and long bones (Fig. 2c). Furthermore, histological analyses confirmed the increased bone thickness and reduced marrow space (Fig. 2c). Moreover, in the wild-type (WT) controls, the bone had a uniform layered structure with parallel collagen fibers and

–Cre +Cre

n.s.

100

50

0

–Cre +Cre

Figure 2 Tie2Cre-mediated PPAR-g deletion leads to osteopetrosis represented by increased bone volume, decreased medullary cavity space and reduced bone resorption. (a) mCT analysis of the femur and tibia from 3-month-old wild-type (–Cre) and mutant (+Cre) mice. BVF (top) and BMD (bottom) were increased for both the proximal (Prox) and the distal (Dist) regions in the mutants (n ¼ 3). (b) Static histomorphometric analyses of the femur from 2-monthold mice. BV / TV, trabecular bone volume / total tissue volume; Tb.th, trabecular thickness; Tb.n, trabecular number; Tb.sp, trabecular separation (n ¼ 4; error bars, s.e.m.). (c) mCT images of the proximal region of femurs (top), whole-body X-ray analyses with enlarged images for femurs and knee joints (middle, arrow indicates the decreased size of the medullary cavity in the +Cre femur) and H & E stained femur sections imaged under transmitted polarized light (bottom). (d) TRAP staining of osteoclasts (arrows) in femur sections. (e) Alkaline phosphatase staining of osteoblasts (arrows) in femur sections. B, bone. (f) Quantification of osteoclast and osteoblast cells. N.Oc/mm2, number of osteoclasts per tissue area; Ob.S/BS%, osteoblast surface per bone surface; N.Ob/ B.Pm, number of osteoblasts per bone perimeter (mm–1). (g) Dynamic histomorphometry shows bone formation rate (BFR / BS) and mineral apposition rate (MAR) in 2-month-old mice, as measured by calcein double labeling (n ¼ 4; error bars, s.e.m.). (h) Detection of bone resorption and bone formation markers. Urine deoxypyridinoline (DPD) bone resorption marker (left, n ¼ 20) and serum alkaline phosphatase (middle) and osteocalcin (right) bone formation markers (n ¼ 17) were quantified and plotted as shown. Error bars, s.e.m. *P o 0.05; **P o 0.01; ***P o 0.005.

1498

VOLUME 13

[

NUMBER 12

[

DECEMBER 2007

NATURE MEDICINE

LETTERS

b

2

2

1

1

0 Ctsk Mmp9

3

3

2

2

Ctsk

Fos

Car2

1 0

– –

– +

+ +

– –

– +

+ +

– –

– +

+ +

– –

– +

+ +

3

P-c-Jun (Ser63)

1 0 0h

lκBα β-actin

f

CMX/veh CMX/BRL g+a/veh g+a/BRL

GL3B/BRL c-fos/BRL

500

PPR E/veh

400

PPR E/BRL

80

300 200 100 33 ng

100 ng

40

4.8 3.2 1.6

24 h 48 h 72 h

0 h 24 h 48 h 72 h

3

3

2

2

1

1 0

0h

24 h 48 h 72 h

0h

24 h

48 h 72 h

m-cfosA (–778) r-cfosA (–732) h-cfosA (–784)

AATC AGGTTA T TGTACA CTTA AAGC GGGTTA T TGAATA CTTA AAGT AGGTTA T CGTATT CTCT

(–758) (–712) (–764)

m-cfosB (–194) r-cfosB (–199) h-cfosB (–189)

TCAT AGGTGA A AGATGT ATGC TCAT AGGTGA A AGTTAC AGAC TCAC GAGATT A GGACAC GCGC

(–174) (–179) (–169)

GGGTGA GTGGTA AGGGCA AGGTCA

A A T A

ATGTGC ATTT AGGTCT TGTT CAGTCA AGGTCA

h

RANKL

0 GL3B

RANK

1.1 kb 0.48 kb 0.22 kb –Cre/veh

–Cre/veh

c-Fos

–Cre/BRL

–Cre/BRL 0.2 Anti-PPAR-γ (H100) (ChIP/input)

+Cre/veh +Cre/BRL

0.04 0.02 0 c-fos –5 kb

+Cre

1.1/1.0

PPAR-γ and RXR-α

0.06

–Cre

0

aP2 CTCT PEPCK AACT Acyl-CoA synthase DR-1 consensus

4.9/2.6

20 10 ng

1

2.4/1.4

0 0 ng

Veh GW9662 T0907

6.4

0

3.4/1.5

60 Luc/β-gal

Luc/β-gal

c-fos/veh 600

Tnfa

0

0h

0 h 24 h 48 h 72 h

GL3B/veh

g

2.5 2 1.5 1 0.5 0

Ccl2

+Cre

2.5 2 1.5 1 0.5 0

24 h 48 h 72 h

Ctsk

Acp5

c-Jun

Nfatc1 2

–Cre

–Cre +Cre

2

P-c-Jun (Ser73)

Fosl1

1

+Cre

Nfatc1

BRL RANKL

d

0 Fos

0 –Cre

+Cre 2

Fos

e

Anti-PPAR-γ (E8) (ChIP/input)


+Cre

+Cre 1

3

3 2 1 0

c

–Cre 2

+Cre/BRL

3

Acp5 Calcr

–Cre 1

+Cre/veh

–Cre/BRL

Acp5

–Cre

–Cre/veh 4

Pparg1

BRL

Mmp9

Veh

a

c-fosA

+Cre/BRL

JNK NFκB Fosl1 NFATc1

0.1

P-Jun

0

c-fosB

TRAF6

+Cre/veh

c-fos –5 kb

c-fosA

c-fosB

c-fos PPAR-γ /RXR-α

TRAP, Calcr, CAR2, CTSK Osteoclastogenesis

Figure 3 PPAR-g and ligand promote osteoclast differentiation by regulating c-fos expression. (a) In vitro RANKL-mediated osteoclast differentiation from splenocytes. Mature osteoclasts were identified as multinucleated TRAP+ cells; scale bar represents 50 mm. (b) mRNA expression of osteoclast-specific functional genes (left), RANKL-induced transcription factors (middle) and mature macrophage-related inflammatory genes (right). (c) mRNA expression of osteoclast genes were suppressed by PPAR-g antagonists GW9662 and T0070907 only in WT cells. (d) mRNA expression during a time course after RANKL treatment. The results are shown as relative mRNA levels to the ribosomal protein L19. BRL or antagonists were used at 1 mM. Acp5, TRAP; Calcr, calcitonin receptor; Car2, carbonic anhydrase 2; Ctsk, cathepsin K; Mmp9, matrix metallopeptidase-9; Fos, c-fos osteosarcoma oncogene; Fosl1, fos-like antigen 1; Nfatc1, nuclear factor of activated T-cells, cytoplasmic, calcineurin-dependent 1; Ccl2, monocyte chemoattractant protein-1; Tnfa, tumor necrosis factor-a. (e) RANKL-induced c-Jun phosphorylation or IkBa degradation was not decreased in the mutant cells, as detected by western blotting. (f) Identification of PPREs in the c-fos promoter. The c-fos promoter was induced by PPAR-g and ligand in transient transfection assays in RAW264.7 cells, as determined by luciferase reporter assays (left). Luciferase reporter assays with truncated c-fos promoters indicated the presence of two PPREs (middle). The two numbers for fold induction represent (g+a/BRL) / (CMX/veh) followed by (g+a/BRL) / (g+a/veh). Right, alignment of mouse (m), rat (r) and human (h) c-fosA and c-fosB PPRE regions, together with known PPREs in ap2 (adipocyte-specific fatty-acid binding protein), PEPCK (phosphoenolpyruvate carboxykinase), ACS (acyl-CoA synthase) and DR-1 (direct repeat-1) consensus sequence. b-gal, b-galactosidase. (g) ChIP analysis of PPAR-g binding to the c-fosA, c-fosB or upstream promoter region (c-fos –5 Kb, as negative control) in bone marrow–differentiated monocytes and macrophages. E8 and H100 denote the antibodies to PPAR-g described in the Methods. (h) A schematic diagram of RANKL signaling pathways, showing PPAR-g and its ligand promoting osteoclast differentiation by specifically regulating the c-fos pathway.

NATURE MEDICINE

VOLUME 13

[

NUMBER 12

[

DECEMBER 2007

1499

LETTERS

a

GFP

DAPI

Merge

Figure 4 In vitro and in vivo rescues of osteoclast differentiation and extramedullary hematopoiesis. (a,b) Osteoclast differentiation blockade can be rescued by ectopic expression of c-fos by retroviral gene transfer. (a) Number of multinucleated osteoclasts as a percentage of the number of GFP+ cells in mutant cells infected with c-fos virus (MIGR-cfos) or control virus (MIGR-F); scale bar represents 50 mm. (b) Ectopic expression of c-fos rescues the RANKL-mediated, but not BRL-stimulated, induction of osteoclast genes in the mutant cells. The results are shown as relative mRNA levels to the ribosomal protein L19. Tnfrsf11a, Tumor necrosis factor receptor superfamily, member 11a, or RANK. (c,d) Extramedullary hematopoiesis can be reversed by bone marrow transplantation. (c) Extramedullary hematopoiesis can be conferred by transplantation of mutant (+Cre) bone marrow into WT recipients (–Cre; n ¼ 5). (d) Extramedullary hematopoiesis can be rescued by transplantation of WT (–Cre) bone marrow into mutant recipients (+Cre; n ¼ 8). Extramedullary hematopoiesis was determined by spleen–to–body weight ratio, spleen cell number and bone marrow cell number. Transplantation efficiency was measured by FACS analysis of GFP+ population in blood cells. *P o 0.05; ***P o 0.005.

25 20 +

Oc/GFP (%)

MIGR-Φ

15 10 5

MIGR-cfos MIGR- MIGRcfos Φ

b Relative expression

4

–Cre/Φ +Cre/Φ +Cre/cfos

3 2 1 0

0.4 0.3 0.2 0.1 0 Donor Recipient

–Cre –Cre

–Cre –Cre

*

200

n.s.

600

–Cre –Cre

Veh

+Cre –Cre

n.s.

200 –Cre –Cre

–Cre +Cre

BRL Ctsk

120 80

*

40

0 Donor Recipient

–Cre –Cre

+Cre –Cre n.s.

120

400

0 Donor Recipient

BRL Acp5

400

0 Donor Recipient

Veh

80 40

0 Donor Recipient

–Cre –Cre

–Cre +Cre

smooth endosteal surface, typical of a normal lamellar bone. In contrast, in the mutants, the bone structure appeared irregular and the endosteal surface was uneven, reminiscent of woven bone (Fig. 2c). These results show that Tie2Cre-mediated PPAR-g deletion leads to increased bone mass, reduced medullary cavity space and altered bone remodeling, which is manifested as pale bones (Fig. 1c). Because PPAR-g was deleted in the hematopoietic but not the mesenchymal lineages, this phenotype is a probable result of osteoclastic rather than osteoblastic defects. Indeed, tartrate-resistant acid phosphatase (TRAP) staining of femur sections revealed that the numbers of mature osteoclasts were lower in the mutants both in the trabecular bone and along the cortical bone shaft, whereas the numbers of alkaline phosphatase+ osteoblasts were unaffected (Fig. 2d–f). Consistently with this, dynamic histomorphometry by calcein double labeling showed that the bone formation rate and mineral apposition rate were unchanged (Fig. 2g). Furthermore, urine deoxypyridinoline levels were significantly lower in the mutants (Fig. 2h), suggesting decreased bone resorption; in contrast, serum alkaline phosphatase and osteocalcin levels (Fig. 2h) were normal, suggesting unaffected bone formation. The decrease in urine deoxypyridinoline at this time point assessed was 24%, but the cumulative effect of a continued decrease in bone resorption on bone volume and remodeling over time may be more profound. These results indicate that the osteopetrosis observed in the mutants was mainly due to decreased osteoclast number and bone resorption. We next assessed the effect of PPAR-g deletion on receptor activator of nuclear factor-kB ligand (RANKL)-induced osteoclast differentiation from spleen progenitor cells in an in vitro culture system21. After 6 d, many mature osteoclasts, represented by multinucleated TRAP+ cells, developed in the WT cultures (Fig. 3a). In contrast, multinucleated or TRAP+ cells were rarely observed in the mutant

1500

Veh

BRL

Tfrsf11a 100 75 50 25 0 Donor Recipient

n.s.

Blood cells + GFP (%)

600

+Cre –Cre

–Cre +Cre

BRL Calcr

100 75 50 25 0 Donor Recipient

–Cre –Cre

+Cre –Cre n.s.

Blood cells + GFP (%)

Veh

BM cells (×106)

0.1

Spleen cells (×106)

0.2

0 Donor Recipient

d

***

0.3

Spleen cells (×106)

Spleen/body weight (%)

c

BRL Fos

BM cells (×106)

Veh

Spleen/body weight (%)


0

–Cre –Cre

–Cre +Cre

cultures (Fig. 3a). Because the numbers of stem and progenitor cells were actually higher in the mutant spleen (Fig. 1), the decreased osteoclast number was due not to a lack of progenitor cells but rather to intrinsic defects in the RANKL-mediated osteoclast differentiation. In addition, treatment with a PPAR-g agonist, BRL 49653 (BRL, also known as rosiglitazone), stimulated osteoclast differentiation in WT cells, but not in mutant cells (Fig. 3a), indicating that the effect was PPAR-g–dependent. These results show a pro-osteoclastogenic role for PPAR-g and its ligand. Mature osteoclasts are characterized by the expression of several genes that are crucial for extracellular matrix degradation and bone resorption22. Their gene products include TRAP, calcitonin receptor, carbonic anhydrase-2, cathepsin K and matrix metalloproteinase-9. In the RANKL-treated, spleen-derived macrophages, the expression of these genes was reduced by 35–80% in the mutants (Fig. 3b) and was induced by BRL only in the WT cells. These results confirm that PPAR-g and its ligand stimulate osteoclast differentiation at the transcriptional level. Similar and often more noticeable results were observed in bone marrow–osteoclast differentiation cultures (Supplementary Fig. 2 online). Binding of RANKL to its receptor RANK triggers intricate and distinct signaling cascades that control lineage commitment and osteoclast activation23,24. c-fos is an important mediator of osteoclastogenesis, and mice lacking c-fos develop osteopetrosis as a result of a block in osteoclast differentiation6. RANKL induces c-fos expression, which is required for the induction of nuclear factor of activated T cells, cytoplasmic, calcineurin-dependent-1 (NFATc1) and fos-like antigen-1 (refs. 24,25). The expression of all three transcription factors was substantially reduced in mutant cells and was induced by BRL only in WT cells (Fig. 3b). In contrast, mature macrophage–related inflammatory genes such as monocyte chemoattractant protein-1 and tumor necrosis factor-a had a reciprocal pattern, with increased expression in the mutant cells and suppression by BRL in WT cells

VOLUME 13

[

NUMBER 12

[

DECEMBER 2007

NATURE MEDICINE


LETTERS (Fig. 3b), which is consistent with the previously reported antiinflammatory role of PPAR-g (ref. 26). Moreover, the expression of several genes that are common to macrophages and osteoclasts, including transcription factor PU.1, receptors MCSFR and RANK, and adaptor protein TRAF6, was unaffected by PPAR-g deletion (Supplementary Fig. 3 online). These results suggest that the stimulatory effect of PPAR-g and its ligand is specific to osteoclast genes. The pro-osteoclastogenic role of PPAR-g was further illustrated by the observation that two PPAR-g antagonists, GW9662 (ref. 27) and T0070907 (ref. 28), inhibited osteoclast gene expression in a receptor-dependent manner by 40% and 70%, respectively (Fig. 3c). Examination of gene expression during a time course after RANKL treatment showed that the induction of c-fos (encoded by Fos) and its target genes was highly impaired in the mutants (Fig. 3d). In contrast, RANKL-induced phosphorylation of c-Jun and degradation of IkBa, two other downstream signaling events, were intact (Fig. 3e). This suggests that PPAR-g deficiency selectively blocks the c-fos arm of the RANKL signaling pathways. Indeed, in a transient transfection assay in RAW264.7 cells, transcription from a 1.1-kilobase mouse c-fos promoter was upregulated with increasing amounts of PPAR-g and its heterodimer partner RXR-a, and it was further induced by BRL only in the presence of these receptors (Fig. 3f). Using both luciferase reporter assays of truncated c-fos promoters (Fig. 3f) and electrophoretic mobility shift assays (Supplementary Fig. 4 online), we identified two conserved PPAR-g response elements (PPREs) in the c-fos promoter (Fig. 3f). The ligand-dependent binding of PPAR-g to these two PPREs in vivo was shown by chromatin immunoprecipitation (ChIP) assays with bone marrow–differentiated monocytes and macrophages (Fig. 3g). Notably, PPAR-g regulation of c-fos occurred both before (time 0) and after RANKL stimulation (Fig. 3d), suggesting that PPARg promotes both osteoclast lineage commitment and osteoclast maturation by maintaining the levels of c-fos in monocyte precursors and osteoclasts. In contrast with these results, osteoblast differentiation from bone marrow cells was unaffected, as evidenced by the unaltered expression of the regulators of osteoblastogenesis20 (Supplementary Fig. 5a online). Moreover, BRL inhibited osteoblast differentiation in cells from both WT and mutant mice, confirming that PPAR-g was not deleted in osteoblasts (Supplementary Fig. 5b). Together, these results show that PPAR-g and its ligand promote osteoclast differentiation by directly regulating c-fos expression (Fig. 3h). We next evaluated whether ectopic expression of c-fos can rescue the osteoclast differentiation blockade in the PPAR-g–deleted cells by using retrovirus-mediated gene transfer. The number of osteoclasts differentiated from mutant cells infected with c-fos virus was increased by fivefold compared to cells infected with control virus (Fig. 4a). Consistently with these results, quantitative RT-PCR analyses showed that c-fos overexpression effectively rescued the RANKL-mediated induction of osteoclast genes in the mutant cells without altering the expression of RANK (Fig. 4b). Moreover, BRL had no effect on the osteoclast gene expression rescued by the c-fos virus in the mutant cells (Fig. 4b). These results further demonstrate that PPAR-g deficiency selectively blocks the c-fos arm of the RANKL signaling pathways. We next tested whether the extramedullary hematopoiesis phenotype could be reversed by bone marrow transplantation. We used ubiquitin-GFP (Ub-GFP) transgenic bone marrow to monitor the reconstitution efficiency. First, we transplanted WT or mutant bone marrow into irradiated WT recipient mice (Fig. 4c). The mutant bone marrow recipients developed extramedullary hematopoiesis, as evidenced by increases in both spleen–to–body weight ratio and spleen cell number and a decrease in bone marrow cell number compared to

NATURE MEDICINE

VOLUME 13

[

NUMBER 12

[

DECEMBER 2007

the WT bone marrow recipients (Fig. 4c). The reverse transplantation, in which WT bone marrow was transplanted into mutant or WT recipients, rescued the extramedullary hematopoiesis, as evidenced by the comparable spleen weight and cell numbers between the two groups (Fig. 4d). FACS analyses showed that 94–96% of the blood cells were GFP+ and thus donor derived (Fig. 4c,d). Phenotype reversal by bone marrow transplantation further suggests that the osteopetrotic defect originates from the hematopoietic, rather than the mesenchymal, lineage. Finally, to specifically address the predicted effects of gain of PPAR-g function on osteoclasts and bone resorption in vivo, we treated WT or gf/f-Tie2Cre mice with BRL at 10 mg/kg/d for 6 weeks (Supplementary Fig. 6 online). BRL treatment led to a significant increase in both deoxypyridinoline bone resorption marker levels (Supplementary Fig. 6a) and osteoclast numbers (Supplementary Fig. 6d) in WT but not in mutant mice. These results suggest that PPAR-g activation in vivo promotes osteoclast-mediated bone resorption in a receptor-dependent manner. Consistently with previous studies2,10,29, BRL also significantly decreased the abundance of the osteocalcin bone formation marker (Supplementary Fig. 6b), as well as osteoblast numbers (Supplementary Fig. 6e), in both WT and mutant mice. Consequently, BRL-mediated bone loss was partially alleviated in the mutants (15%) compared to WT controls (35%) (Supplementary Fig. 6c). This evidence suggests that, in addition to the established notion that TZDs can inhibit osteoblast differentiation and bone formation, stimulation of osteoclast differentiation and bone resorption may be a previously unrecognized pathway that further contributes to TZD-mediated bone loss (Supplementary Discussion and Supplementary Fig. 7 online). In summary, this study reveals an unexpected role for PPAR-c and its ligand in promoting osteoclast differentiation and bone resorption. Loss of function by targeted PPAR-c deletion impairs osteoclast differentiation and bone resorption, resulting in osteopetrosis and extramedullary hematopoiesis. In contrast, gain of function by ligand activation of PPAR-c accelerates osteoclast differentiation and bone resorption in a receptor-dependent manner. These findings have potential clinical implications, as they suggest that long-term rosiglitazone usage in the treatment of type 2 diabetes and insulin resistance may cause osteoporosis, owing to a combination of decreased bone formation and increased bone resorption. They also suggest that selective PPAR-c modulators may provide a new strategy for the treatment of bone diseases associated with increased osteoclast activity, such as osteoporosis and rheumatoid arthritis. METHODS Mice. We bred Ppargflox/flox (gf/f) mice15 (backcrossed to C57BL/6J for at least four generations) with Tie2Cre transgenic mice14 to generate gf/f and gf/f-Tie2Cre+/ mice, which we further backcrossed to C57BL/6J mice for three more generations. Unless otherwise stated, we performed all experiments with gf/f-Tie2Cre (+Cre) and littermate gf/f (–Cre) control mice, which were generated by breeding male gf/f-Tie2Cre mice with female gf/f mice. The UbGFP transgenic mice were a generous gift (see Acknowledgments). All protocols for mouse experiments were approved by the Institutional Animal Care and Use Committee of the Salk Institute. Hematological analyses. We collected bone marrow and spleen cells from 6–18-week-old male mice unless otherwise indicated. Mouse clonogenic hematopoietic progenitor cells were quantified with MethoCult media (StemCell Technologies). We quantified hematopoietic stem cells as a Lin–cKit+Sca1+ population with the BD LSRI flow cytometer and the following reagents (BD Pharmingen): biotin-labeled antibodies to the lineage markers CD4, CD8, B220, Ter119, CD11b (Mac-1) and Gr-1, streptavidin- allophycocyanin,

1501


LETTERS phycoerythrin-labeled antibody to CD117 (c-Kit), and FITC-labeled antibody to Sca1. Live cells were gated by DAPI exclusion. Complete blood cell count and white blood cell differential analyses were performed using the HEMAVET HV950FS hematology instrument (Drew Scientific).

Statistical analyses. All statistical analyses were performed with Student’s t-test and are represented as means ± s.d. unless otherwise stated. * indicates P o 0.05; ** indicates P o 0.01, *** indicates P o 0.005; n.s. indicates not significant (P 4 0.05).

Bone analyses. For mCT in vivo imaging, we killed and immediately scanned littermate mice (3–4 months old) at 27-mm resolution with the eXplore Locus scanner (GE Healthcare). BVF and BMD were determined with the eXplore MicroView 2.0 software. We performed histomorphometric analyses with the Osteomeasure Analysis System (Osteometrics). We injected calcein (20 mg/kg) 2 and 7 d before bone collection. We performed X-ray analysis with a Faxitron cabinet X-ray (Hewlett Packard). For TRAP and alkaline phosphatase staining, we fixed the femurs in 2% paraformaldehyde for 24 hr, decalcified them with 10% EDTA (pH 7.5) for 8 d, emerged them in 30% sucrose in PBS for 24 hr, embedded them in optimal cutting temperature compound (Tissue-Tek) and sectioned them at 10 mm. TRAP+ or alkaline phosphatase+ cells were stained with the Leukocyte Acid Phosphatase or Alkaline Phosphatase staining kits, respectively (Sigma). We measured urine deoxypyridinoline cross-links with the Metra DPD EIA kit (Quidel) and normalized them by creatinine level (Thermo Electron). Serum alkaline phosphatase was measured with the QuantiChrom Alkaline Phosphatase Assay kit (BioAssay Systems). Serum osteocalcin was measured with the mouse osteocalcin EIA kit (BTI).

Note: Supplementary information is available on the Nature Medicine website.

In vitro differentiation. Osteoclasts were differentiated from splenocytes or bone marrow cells as described21. Briefly, we differentiated cells with 100 ng/ml of macrophage colony-stimulating factor (MCSF) in a-MEM containing 10% FBS for 3 d, then with 100 ng/ml of MCSF and 100 ng/ml of RANKL (R&D Systems) for 3 d (or 30 min for western blot analyses). We identified mature osteoclasts as multinucleated (43 nuclei), TRAP+ cells. We differentiated osteoblasts as described2. Briefly, bone marrow cells were differentiated in a-MEM containing 10% FBS and freshly added 50 mg/ml ascorbic acid and 10mM a-glycerophosphate for 14–21 d with a medium change twice a week. Gene expression analyses. RNA was reverse transcribed into cDNA and analyzed with SYBR Green–based quantitative RT-PCR in triplicate or quadruplicate. We used antibodies to the following proteins for western blotting: PPAR-g and IkBa (Santa Cruz), b-actin (Sigma), phospho–c-Jun (on Ser63), phospho–c-Jun (on Ser73) and c-Jun (Cell Signaling). Promoter analyses. For transient transfection, we transfected a luciferase reporter into the mouse macrophage cell line RAW264.7 cells along with expression plasmids for b-gal, PPAR-g1 and RXRa using FuGENE 6 (Roche). On the next day, we treated the cells with 1 mM BRL or DMSO vehicle control overnight. We normalized luciferase activity by b-galactosidase activity. GL3B vector was a negative control; PPRE-luc was a positive control. We performed electrophoretic mobility shift and ChIP assays as described30. We performed the ChIP assays with bone marrow–derived monocytes and macrophages differentiated in MCSF for 3 d and two PPAR-g–specific antibodies (monoclonal E8 and polyclonal H100, Santa Cruz). ChIP output was measured by quantitative PCR and normalized by 10% input. All assays were repeated at least three times. Retroviral gene transduction. The mouse c-fos cDNA was amplified and cloned into pMIGR retroviral vector (pMIGRF) that contains a GFP gene downstream of an internal ribosomal entry site. We achieved retrovirus packaging by transfecting pMIGR and the packaging plasmid pEcopak into 293 cells, and we collected viral supernatant 48 hr later. We performed viral inoculation as described25. Briefly, we suspended bone marrow cells in the viral supernatants containing polybrene (8 mg/ml) and incubated them for 4 h at 37 1C. We then removed the viral supernatants and differentiated the cells into osteoclasts for 6 d. Bone marrow transplantation. We labeled donor mice with GFP by crossing them with Ub-GFP mice to generate gf/f–Ub-GFP or gf/f-Tie2Cre–Ub-GFP mice. Recipient mice (12–14 weeks old) were irradiated at 9 Gy, reconstituted with 2 106 bone marrow cells on the same day retro-orbitally, and analyzed 12–20 weeks after transplantation. We determined reconstitution efficiency in blood cells by FACS of the GFP marker.

1502

ACKNOWLEDGMENTS We would like to thank the Department of Radiology and Moores Cancer Center at the University of California at San Diego for the bone analyses, M. Yanagisawa (University of Texas Southwestern Medical Center) for providing the Tie2Cre transgenic mice, I. Verma (Salk Institute) for providing the Ub-GFP transgenic mice, S. Hong, J. Jonker, R. Yu and Y. Wan for discussion and critical reading of the manuscript, and L. Ong and S. Ganley for administrative assistance. R.E. is an investigator of the Howard Hughes Medical Institute at the Salk Institute and the March of Dimes Chair in Molecular and Developmental Biology. Y.W. is supported by a postdoctoral fellowship from the American Cancer Society (PF-03-081-01-TBE). This work was supported by the Howard Hughes Medical Institute and the National Institutes of Health (SCOR/HL569898, HL07770 and DK57978). AUTHOR CONTRIBUTIONS Y.W. conceived the project, designed and performed most of the experiments and wrote the manuscript and L.-W.C. performed the experiments by assisting with tissue collection, DNA, RNA and protein isolation, and analyses. R.M.E. supervised the study and edited the manuscript. Published online at http://www.nature.com/naturemedicine Reprints and permissions information is available online at http://npg.nature.com/ reprintsandpermissions

1. Tolar, J., Teitelbaum, S.L. & Orchard, P.J. Osteopetrosis. N. Engl. J. Med. 351, 2839–2849 (2004). 2. Akune, T. et al. PPAR g insufficiency enhances osteogenesis through osteoblast formation from bone marrow progenitors. J. Clin. Invest. 113, 846–855 (2004). 3. Cock, T.A. et al. Enhanced bone formation in lipodystrophic PPARghyp/hyp mice relocates haematopoiesis to the spleen. EMBO Rep. 5, 1007–1012 (2004). 4. Grey, A. Skeletal consequences of thiazolidinedione therapy. Osteoporos. Int. published online, doi:10.1007/s00198-007-0477-y (28 September 2007). 5. Kahn, S.E. et al. Glycemic durability of rosiglitazone, metformin, or glyburide monotherapy. N. Engl. J. Med. 355, 2427–2443 (2006). 6. Grigoriadis, A.E. et al. c-Fos: a key regulator of osteoclast-macrophage lineage determination and bone remodeling. Science 266, 443–448 (1994). 7. Barak, Y. et al. PPAR g is required for placental, cardiac, and adipose tissue development. Mol. Cell 4, 585–595 (1999). 8. Kubota, N. et al. PPAR g mediates high-fat diet–induced adipocyte hypertrophy and insulin resistance. Mol. Cell 4, 597–609 (1999). 9. Rosen, E.D. et al. PPAR g is required for the differentiation of adipose tissue in vivo and in vitro. Mol. Cell 4, 611–617 (1999). 10. Ali, A.A. et al. Rosiglitazone causes bone loss in mice by suppressing osteoblast differentiation and bone formation. Endocrinology 146, 1226–1235 (2005). 11. Li, M. et al. Surface-specific effects of a PPARg agonist, darglitazone, on bone in mice. Bone 39, 796–806 (2006). 12. Rzonca, S.O., Suva, L.J., Gaddy, D., Montague, D.C. & Lecka-Czernik, B. Bone is a target for the antidiabetic compound rosiglitazone. Endocrinology 145, 401–406 (2004). 13. Sottile, V., Seuwen, K. & Kneissel, M. Enhanced marrow adipogenesis and bone resorption in estrogen-deprived rats treated with the PPARgamma agonist BRL49653 (rosiglitazone). Calcif. Tissue Int. 75, 329–337 (2004). 14. Kisanuki, Y.Y. et al. Tie2-Cre transgenic mice: a new model for endothelial cell-lineage analysis in vivo. Dev. Biol. 230, 230–242 (2001). 15. He, W. et al. Adipose-specific peroxisome proliferator–activated receptor g knockout causes insulin resistance in fat and liver but not in muscle. Proc. Natl. Acad. Sci. USA 100, 15712–15717 (2003). 16. Constien, R. et al. Characterization of a novel EGFP reporter mouse to monitor Cre recombination as demonstrated by a Tie2 Cre mouse line. Genesis 30, 36–44 (2001). 17. Wan, Y. et al. Maternal PPAR g protects nursing neonates by suppressing the production of inflammatory milk. Genes Dev. 21, 1895–1908 (2007). 18. Takakura, N. et al. Critical role of the TIE2 endothelial cell receptor in the development of definitive hematopoiesis. Immunity 9, 677–686 (1998). 19. Cantor, A.B. & Orkin, S.H. Transcriptional regulation of erythropoiesis: an affair involving multiple partners. Oncogene 21, 3368–3376 (2002). 20. Taichman, R.S. Blood and bone: two tissues whose fates are intertwined to create the hematopoietic stem cell niche. Blood 105, 2631–2639 (2005). 21. Kawano, H. et al. Suppressive function of androgen receptor in bone resorption. Proc. Natl. Acad. Sci. USA 100, 9416–9421 (2003).

VOLUME 13

[

NUMBER 12

[

DECEMBER 2007

NATURE MEDICINE

LETTERS 27. Leesnitzer, L.M. et al. Functional consequences of cysteine modification in the ligand binding sites of peroxisome proliferator activated receptors by GW9662. Biochemistry 41, 6640–6650 (2002). 28. Lee, G. et al. T0070907, a selective ligand for peroxisome proliferator-activated receptor g, functions as an antagonist of biochemical and cellular activities. J. Biol. Chem. 277, 19649–19657 (2002). 29. Grey, A. et al. The peroxisome proliferator-activated receptor-g agonist rosiglitazone decreases bone formation and bone mineral density in healthy postmenopausal women: a randomized, controlled trial. J. Clin. Endocrinol. Metab. 92, 1305–1310 (2007). 30. Chui, P.C., Guan, H.P., Lehrke, M. & Lazar, M.A. PPARg regulates adipocyte cholesterol metabolism via oxidized LDL receptor 1. J. Clin. Invest. 115, 2244–2256 (2005).


22. Karsenty, G. & Wagner, E.F. Reaching a genetic and molecular understanding of skeletal development. Dev. Cell 2, 389–406 (2002). 23. Teitelbaum, S.L. Bone resorption by osteoclasts. Science 289, 1504–1508 (2000). 24. Wada, T., Nakashima, T., Hiroshi, N. & Penninger, J.M. RANKL-RANK signaling in osteoclastogenesis and bone disease. Trends Mol. Med. 12, 17–25 (2006). 25. Matsuo, K. et al. Fosl1 is a transcriptional target of c-Fos during osteoclast differentiation. Nat. Genet. 24, 184–187 (2000). 26. Ricote, M., Li, A.C., Willson, T.M., Kelly, C.J. & Glass, C.K. The peroxisome proliferator–activated receptor-g is a negative regulator of macrophage activation. Nature 391, 79–82 (1998).

NATURE MEDICINE

VOLUME 13

[

NUMBER 12

[

DECEMBER 2007

1503