Cardiotrophin-1 Is an Osteoclast-Derived Stimulus of Bone Formation Required for Normal Bone Remodeling

Cardiotrophin-1 Is an Osteoclast-Derived Stimulus of Bone FormationRequired for Normal Bone Remodeling

Emma C Walker,1 Narelle E McGregor,1 Ingrid J Poulton,1 Sueli Pompolo,1 Elizabeth H Allan,1 Julian MW Quinn,1,2

Matthew T Gillespie,1,2 T John Martin,1,2 and Natalie A Sims1,2

ABSTRACT: Cardiotrophin (CT-1) signals through gp130 and the LIF receptor (LIFR) and plays a major rolein cardiac, neurological, and liver biology. We report here that CT-1 is also expressed within bone in osteo-clasts and that CT-1 is capable of increasing osteoblast activity and mineralization both in vitro and in vivo.Furthermore, CT-1 stimulated CAAT/enhancer-binding protein-� (C/EBP�) expression and runt-related tran-scription factor 2 (runx2) activation. In neonate CT-1−/− mice, we detected low bone mass associated withreduced osteoblasts and many large osteoclasts, but increased cartilage remnants within the bone, suggestingimpaired resorption. Cultured bone marrow (BM) from CT-1−/− mice generated many oversized osteoclastsand mineralized poorly compared with wildtype BM. As the CT-1−/− mice aged, the reduced osteoblast surface(ObS/BS) was no longer detected, but impaired bone resorption continued resulting in an osteopetroticphenotype in adult bone. CT-1 may now be classed as an essential osteoclast-derived stimulus of both boneformation and resorption.J Bone Miner Res 2008;23:2025–2032. Published online on July 28, 2008; doi: 10.1359/JBMR.080706

Key words: osteoblasts, osteoclast, gp130, cytokines, coupling, bone histomorphometry

INTRODUCTION

MANY CYTOKINES SIGNAL through the gp130 co-receptor, including oncostatin M (OsM), leukemia in-

hibitory factor (LIF), interleukin (IL)-6, and IL-11. Thesecytokines stimulate differentiation of osteoblasts (bone-forming cells)(1,2) and osteoclasts (bone-resorbing cells) byupregulating RANKL production by the osteoblast.(3–5)

Osteoblastic control of osteoclast differentiation is a promi-nent feature of bone biology and, although osteoclasts alsomodify osteoblast activity the factors involved, termed“coupling factors,” are less well defined.(6,7)

LIFR is a gp130 associating co-receptor used, not only byLIF, but also by cardiotrophin-1 (CT-1), CNTF, CLC/CLF,and cardiotrophin-2 (CT-2). In humans, an inactivating mu-tation in LIFR leads to a lethal skeletal dysplasia.(8) LIFRdeletion in mice is similarly neonatal lethal with osteopenia(very little trabecular bone) and many oversized osteo-clasts.(9) gp130 deletion in mice leads to neonatal lethality,dwarfism, and osteopenia associated with many oversizedosteoclasts and few osteoblasts.(10) Consistent with this,specific deletion of SHP2/Ras/MAPK signaling down-stream of gp130 also results in osteopenia in adult mice.(11)

In contrast, mice lacking IL-11R� have high bone mass inadulthood characterized by low bone turnover(12) and theresponse of IL-6–deficient mice to osteoclastic stimuli suchas ovariectomy(13) and experimental arthritis(14) is im-paired.

Whereas its receptor subunits are clearly critical, the roleof CT-1 in bone metabolism is unclear. CT-1 administrationstimulates osteoclast formation from bone marrow (BM)co-cultured with osteoblasts.(5) CT-1 mRNA is expressedby rat calvarial osteoblasts,(15) but whereas CT-1 has beenshown in a range of tissues, including cartilage,(16) bone hasnot been examined.

To study whether CT-1 plays a role in normal bone bi-ology, we assessed expression of CT-1 in the skeleton, stud-ied the effects of recombinant CT-1 on osteoblasts, andadipocyte differentiation and analyzed bones from CT-1−/−

mice.(17)

MATERIALS AND METHODS

Animal, skeletal, and serum analyses

CT-1−/− mice were obtained from Dr Michael Sendtner(University of Würzburg)(17) and backcrossed for 10 gen-erations onto C57Bl/6. Double fluorochrome labeling wasperformed as described,(18) specimens were fixed in 4%paraformaldehyde and embedded in methylmethacry-late,(18,19) and 5-�m sections were stained with toluidineblue(18) or Xylenol orange.(20) Histomorphometry was car-ried out in the secondary spongiosa of the proximal tibia(Osteomeasure; Osteometrics).(18) Femoral and tibiallength and femoral diameter were determined from contactX-rays with ImageJ 1.36b.(21) Femoral BMD, cortical cir-cumference, and thickness were measured by pQCT(Stratec X-CT Research SA+; Stratec Medizintechnik).(19)The authors state that they have no conflicts of interest.

1St Vincent’s Institute, Fitzroy, Victoria, Australia; 2Department of Medicine, St Vincent’s Hospital, The University of Melbourne,Fitzroy, Victoria, Australia.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 23, Number 12, 2008Published online on July 28, 2008; doi: 10.1359/JBMR.080706© 2008 American Society for Bone and Mineral Research

2025

JO803127 2025 2032 December

Statistically significant effects were determined by one- ortwo-way ANOVA followed by Fisher’s posthoc test usingStatView 4.0. All data are presented as means ± SE. p <0.05 was considered statistically significant.

Immunohistochemistry

Legs from wildtype neonates were fixed, decalcified, andembedded in paraffin.(22) Serial sections were used forTRACP stain(23) and immunohistochemistry(22) with thefollowing modifications. Endogenous peroxidase was inhib-ited with 1.7% hydrogen peroxide (Merck) in methanol for30 min and permeabilized with 1% sodium borohydride/PBS (Sigma), TNT (0.1 M Tris-HCL, pH 7.5, plus 0.15 MNaCl and 0.05% Tween 20) and 15 min in 0.1% trypsin(JRH, Lenexa, KS, USA) before 30 min in 1% BSA(Sigma), 0.05% Tween 20 (BDH, Poole, UK), and 5% nor-mal rabbit serum in PBS. Sections were incubated over-night with 5 �g/ml anti-mouse CT-1 (R&D Systems, Min-neapolis, MN, USA), followed by 6 �g/ml biotinylated goatanti-rat IgG (Vector Laboratories, Burlingame, CA, USA)in blocking solution for 45 min and 1.6 mg/ml Streptavidin-HRP (DakoCytomation, Glostrup, DK) in PBS for 45 min.CT-1+ cells were visualized with 0.5 mg/ml diaminobenzi-dine (Sigma, St Louis, MO, USA) in PBS plus 0.006% hy-drogen peroxide (Merck).

Primary cell culture

Osteoclastogenic potential of BM was determined bystimulating BM cell preparations from long bones of maleand female mice (105 cells/10-mm well) with RANKL andmacrophage-colony-stimulating factor (M-CSF; R&D Sys-tems).(23) TRACP+ multinucleated cells (MNCs) werecounted at day 7. For resorption assays, BM (105 cells/6-mmwell) was plated on dentine slices and stimulated withRANKL and M-CSF. After 14 days, dentine was strippedwith 0.25 M NH4OH. Resorption pits were identified bystaining with xylene-free black ink, and residual ink wasremoved by wiping against absorbent paper.(24) The area ofdentine surface resorbed was measured (Osteomeasure;Osteometrics).(25) Osteoblast potential was assessed in BMpreparations from long bones of adult mice,(26) plated at 8× 106 cells/cm2 in 16-mm plates. Seven days after initialculturing, cells were differentiated with �-MEM (JRH Bio-sciences) plus 15% heat-inactivated FBS (HIFBS) andascorbate (50 �g/ml; Sigma), for mineralization assays, �-glycerophosphate was used (10 mM; Sigma); for adipogen-esis, cells were grown in �-MEM plus 15% HIFBS with 5mM rosiglitazone (Sigma). Cell extracts were assessed forALP activity.(27) Mineralized nodules were assessed byAlizarin red staining solubilized overnight in 10% cetylpyri-dinium chloride (Sigma), and absorbance was measured at562 nm. Adipogenesis was assessed by oil red O stainingsolubilized in isopropanol and measured at 515 nm.

Kusa 4b10 assays

The Kusa 4b10 cell line is maintained in �-MEM with10% FBS (JRH Biosciences). Cells were seeded at 1.2 × 104

cells/well in 12-well plates (Greiner Bio-one) in �-MEM +10% FBS at 37°C with 5% CO2 overnight. The following

day (day 0), the medium was replaced with osteoblast dif-ferentiating (above) or adipogenic medium (�-MEM +15% HIFBS), insulin (6.6 × 10−8 M; Novo Nordisk), 3-iso-butyl-1-methylxanthine (2.5 × 10−10 M; Sigma), dexameth-asone (10−8 M; Sigma), and murine CT-1 (R&D Systems)as described. ALP activity, mineralization, and adipogen-esis were assessed as above. Short-term effects of CT-1 ongene expression were assessed in nondifferentiated cells byaddition of the protein from day 0; for differentiated osteo-blasts, Kusa 4b10 cells were cultured in osteoblast differen-tiating media and CT-1 added at day 17 of differentiation.To determine whether CT-1 stimulates C/EBP� throughtranscription or translation, osteoblastic Kusa 4b10 cellswere pretreated with actinomycin D (10 �g/ml, 2 h) or cy-cloheximide (5 �g/ml, 4 h).

Real-time RT-PCR

RNA was isolated using trizol (Invitrogen) and concen-tration determined by spectrophotometer (NanodropND1000). cDNA was synthesized (random primers, 10 mMdNTP, 5× first strand buffer, 0.1 M DTT, RNaseOUT, Su-perscriptTM III RT 200U/�l) using 5 �g material as fol-lows: 5 min at 65°C; 5 min at 4°C; 60 min at 50°C; 15 min at70°C; � at 4°C (Bio-Rad iCycler). Amplification was car-ried out with AmpliTaq Gold (Perkin-Elmer), SYBRGreen (Molecular Probes), and specific oligonucleotideprimers (Sigma-Genosys). RT-PCR conditions used were 1cycle for 10 min at 95°C; 40 cycles for 30 s at 95°C, 1 min at60°C for 30 s at 72°C; 1 cycle for 1 min at 95°C, 30 s at 55°C,and 30 s at 95°C (Stratagene Mx3000P). HPRT1, the house-keeping gene, was not regulated by CT-1.

6xOSE2 reporter assay

UMR106.01 cells were seeded into a 96-well plate at 1 ×104 cells/well in �-MEM + 10% FBS. After 6 h, cells weretransfected with 0.1 �g/well of 6xOSE2 pDNA with Fugeneat 3:1 to pDNA. Transfection mix was incubated at roomtemperature for 30 min before adding to the cells. Cellswere incubated overnight at 37°C with 5% CO2 beforeCT-1 or PTH(1-34) treatment. At harvest, a luciferase assaywas carried out (Promega) with Passive Lysis Buffer, Pro-tocol D, for cell preparation and Protocol C for plate read-ing (Polarstar).

In vivo CT-1 administration

Recombinant murine CT-1 (R&D Systems) was admin-istered to 7-wk-old male C57/Bl6 mice by calvarial injec-tion.(28) Briefly, mice were administered daily injections (25�l) of 0.2 �g mCT-1 or saline over the right hemicalvariafor 5 consecutive days. Calcein (20 mg/kg) was injectedintraperitoneally on days 1 and 14. mLIF was used as apositive control and showed a more mild effect than CT-1.Parietal bones were collected 24 h after the last injection,fixed, and analyzed by histomorphometry.(29) All animalstudies were approved by St Vincent’s Health (Melbourne)Animal Ethics Committee.

RESULTS

To determine which cells in bone express CT-1 protein,immunohistochemistry was carried out on neonatal skeletal

WALKER ET AL.2026

samples. In vertebrae, rib, and femora, multinucleated os-teoclasts showed very strong cytoplasmic immunostainingfor CT-1 (Figs. 1A and B, negative control shown in Fig.1C). In contrast, no staining was detected in osteoblasts orcells within the bone marrow. Specific staining for CT-1 wasverified in wildtype osteoclasts cultured from BM treatedwith RANKL and M-CSF for 7 days (Fig. 1D).

Because gp130 and LIFR are expressed in osteoblasts,(30)

we determined whether CT-1 influences osteoblast differ-entiation using the murine stromal Kusa 4b10 cell line,which expresses LIFR and differentiates into osteoblasts oradipocytes under appropriate culture conditions.(27) CT-1treatment of these cells dose-dependently elevated ALPactivity (Fig. 2A) and mineralization (Figs. 2B and 2C),indicating a novel effect of CT-1 on bone formation. Treat-ment with 5 ng/ml IL-11 or LIF, both of which stimulateALP activity in other osteoblast-like cell cultures,(30,31) in-creased mineralization and ALP activity in this cell line toa similar extent (Figs. 2A and 2B).

Because osteoblasts and adipocytes share stromal cellprecursors, we also treated Kusa 4b10 cells grown in adipo-genic medium with CT-1. CT-1 dose-dependently reducedadipogenesis (Fig. 2D), and the few adipocytes formed inthe presence of CT-1 contained smaller lipid droplets thanadipocytes in untreated cultures (Fig. 2E). IL-11, which re-duces adipocyte formation in 3T3-L1 cells,(32) also reducedadipocyte formation, but LIF had a very mild effect (Fig.2D).

Because CT-1 increased osteoblast activity in vitro, wesought to determine whether CT-1 could enhance bone for-mation in vivo. CT-1 administration over the calvariae ofyoung mice(28) increased calvarial thickness (Fig. 2F) byincreasing the mineral appositional rate (MAR) but did notincrease the bone surface over which bone formation oc-curred (MS/BS) (Figs. 2G–2I).

To determine the mechanism by which CT-1 stimulatesbone formation, we studied the activation of intracellularsignaling pathways downstream of gp130. We noted an in-crease in both STAT3 and ERK phosphorylation in Kusa4b10 cells treated with CT-1 by Western blot (data notshown), confirming a direct effect of CT-1 on these cells.We then analyzed early changes in mRNA expression

(within 48 h) in response to CT-1 in undifferentiated Kusa4b10 cells. CT-1 treatment did not significantly alter mRNAlevels for a range of osteoblast markers or key bone regu-latory proteins including runx2, osteocalcin, osterix, ALP,sclerostin, c-fos, osteoprotegerin (OPG), RANKL, IGF-1,gp130, connexin 43, or N-cadherin (data not shown).mRNA levels for adipocyte markers adiponectin, peroxi-some proliferator activated receptor � (PPAR�), and

FIG. 1. Osteoclasts stain positive for CT-1 protein. (A) Osteo-clasts (arrows) show strong immunohistochemical staining forCT-1 in newborn mouse bone. Scale bar � 50 �m. (B) TRACPstain, specific for osteoclasts and their precursors, on a serial sec-tion confirms the identity of CT-1+ cells. (C) Negative control (noCT-1 antibody). (D) Osteoclast and mononuclear precursors frommouse BM macrophages (BMMs) treated with RANKL and M-CSF for 7 days, stained for CT-1 protein. Note positive cytoplas-mic staining in the multinucleated osteoclast and mononuclearprecursors. Scale bar � 50 �m.

FIG. 2. CT-1 stimulates mineralization and inhibits adipogenesisin vitro and stimulates bone formation in vivo. Kusa 4b10 cellswere cultured in the presence or absence of CT-1 at doses from0.625 to 10 ng/ml. (A) CT-1 stimulated alkaline phosphatase ac-tivity (ALP); this was significant for 10 and 5 ng/ml at day 7 andfor all doses at days 10 and 14. Effects of 5 ng/ml recombinantmurine IL-11 (white diamond) and LIF (black triangle) are shownon the same axis to the right of day 14. (B) Alizarin red stainingwas higher (p < 0.001) at all doses of CT-1 compared with un-treated controls at all time points. Effects of murine IL-11 and LIFare shown at day 21. (C) Representative Alizarin red stainingafter 19 days of CT-1 treatment. (D) CT-1 treatment of Kusa 4b10cells in adipogenic media reduced adipogenesis, detected by oilred O staining; this was significant (p < 0.001) at all doses and alltime points assessed. Effects of murine IL-11 and LIF are shownat day 17. (E) Representative oil red O staining of untreated andCT-1–treated wells. Data are mean of three independent experi-ments each carried out in triplicate wells. (F–I) In vivo treatmentwith CT-1 for 5 days over calvariae of wildtype mice increasedcalvarial thickness (F), bone formation rate (G), and mineral ap-positional rate (MAR) (H), but not mineralizing surface (I). n �6 mice per group; ap < 0.05; bp < 0.01; cp < 0.001 vs. PBS-treatedcontrols.

CARDIOTROPHIN-1 IN BONE 2027

Fig 1 live 4/C

C/EBP� were also not significantly altered (data notshown). However, C/EBP� mRNA expression by undiffer-entiated Kusa 4b10 cells was induced 10-fold 1 h after CT-1treatment (Fig. 3A); levels returned to normal at 6 h andthen increased again at 24 and 48 h. The rapid increase inC/EBP� expression was confirmed in Kusa 4b10 cells cul-tured in osteoblast-differentiation media for 17 days (Fig.3B) and in rat osteosarcoma UMR 106-01 cells (Fig. 3C);thus, the response is not cell line or species specific. Cyclo-heximide treatment resulted in a superinduction of C/EBP�mRNA that was not altered in the presence of CT-1 at 30min or 1 h. In support of CT-1 modifying C/EBP� transcrip-tion, the CT-1–enhanced C/EBP� mRNA levels at 1 and 4h was blocked by actinomycin D, and no enhancement overbaseline levels was observed with cycloheximide pretreat-ment (Fig. 3D). Because C/EBP� stimulates runx2 phos-phorylation and activates osteocalcin transcription,(33,34) wedetermined whether CT-1 activates transcription from a6xOSE2 reporter construct.(35) We used PTH as a positivecontrol, detecting a maximal 8-fold induction at 4 h. CT-1induced a 2-fold increase in transcription, significant from 2h onward (Fig. 3E), confirming runx2 activation by CT-1.

We next examined newborn CT-1−/− mice to determinethe importance of CT-1 in skeletal development. CT-1−/−

neonates were of normal size but histology of femora andtibias showed a dramatic reduction in trabecular bone (Figs.4A and 4B). In wildtype mice, trabecular bone extendedwell into the metaphysis, but in CT-1−/− mice, trabecularbone was only observed close to the growth plates. Histo-

morphometry showed a significant reduction in trabecularbone volume (BV/TV) and trabecular number (TbN), butnot trabecular thickness, and markedly less osteoblasts thanin wildtype (Figs. 4C–4F). Closer inspection showed manylarge osteoclasts on the bone surfaces and groups of chon-drocytes within the trabecular bone (Figs. 4A and 4B). In-deed, the number of osteoclasts per unit bone surface wasdoubled in CT-1−/− neonates (Fig. 4G), yet destruction ofthe growth plate cartilage was impaired, indicated by anincrease in cartilage as a proportion of trabecular bone (Fig.4H). Furthermore, the size of individual CT-1−/− osteoclastswas doubled compared with wildtype osteoclasts (osteo-clast area per cell [�m/cell]: wildtype, 226 ± 11; CT-1−/−, 438

FIG. 3. CT-1 stimulates transcription of C/EBP� and activity ofa runx2 reporter assay. (A–C) Fold change in C/EBP�:�PR� ex-pression (real-time RT-PCR) was upregulated by 50 ng/ml CT-1(filled squares) compared with controls (empty diamonds and dot-ted lines) in (A) undifferentiated Kusa 4b10 cells, (B) osteoblastsderived from Kusa 4b10 cells, and (C) UMR 106-01 osteoblast-like cells. (D) Actinomycin D (A) but not cycloheximide (C)blocked the elevation in C/EBP� mRNA expression induced byCT-1. (E) CT-1 (50 ng/ml) rapidly activated a 6xOSE2 (runx2)reporter construct in UMR 106-01 osteoblast-like cells; hPTH(1-34) 10 nM was used as a positive control. Data are means fromthree independent primary cultures, each with triplicate wells.ap < 0.05; bp < 0.01, cp < 0.001 vs. untreated controls.

FIG. 4. Osteopenia, reduced osteoblasts, and increased osteo-clasts in newborn CT-1−/− mice. Representative von Kossa–stained sections of newborn (A) wildtype and (B) CT-1−/− tibiasshowing reduced trabecular bone and chondrocyte stacks (ar-rows) in CT-1−/− trabeculae. Histomorphometry showed reduced(C) trabecular bone volume (BV/TV), (D) trabecular number(Tb.N), and (E) normal trabecular thickness (Tb.Th), as well as(F) reduced osteoblast surface (ObS/BS), (G) increased osteoclastsurface (OcS/BS), and (H) increased cartilage volume within tra-becular bone (CtgV/BV) in CT-1−/− (KO) tibias compared withwildtype (WT). n � 4–5 per group, ap < 0.05, bp < 0.01 vs. WT.

WALKER ET AL.2028

± 10; p < 0.05), but the number of nuclei per osteoclast wasnot elevated (nuclei number per osteoclast: wildtype, 2.3 ±0.1; CT-1−/−, 2.6 ± 0.3).

To determine whether increased osteoclast number andsize in CT-1−/− mice was cell lineage autonomous, BM wascultured from CT-1−/− and wildtype mice with RANKL andM-CSF. More osteoclasts were generated from CT-1−/−

marrow compared with wildtype (Fig. 5A). This may relateto an effect early in osteoclast differentiation, because thenumber of TRACP+ mononuclear cells was also increasedin the absence of CT-1 (TRACP+ MNC/cm2: wildtype, 646± 88; CT-1−/−, 1154 ± 85; p < 0.01). CT-1−/− BM macrophageproliferation was not altered compared with wildtype mar-row (data not shown). As observed in vivo, osteoclasts de-rived from CT-1−/− BM in vitro were larger than those fromwildtype BM (Figs. 5B and 5C) and, whereas slightlyhigher, the number of osteoclast nuclei per cell was notsignificantly altered (Fig. 5D). Despite impaired resorptionin vivo, the ability of osteoclasts derived from CT-1−/− BMto resorb bone in vitro was not impaired (data not shown),nor was actin ring formation disrupted (data not shown).Consistent with an increase in the number of osteoclasts,DC-STAMP mRNA levels, indicative of osteoclast fusion,were elevated in osteoclasts derived from CT-1−/− BM com-pared with wildtype (Fig. 5E).

To study whether the low osteoblast number seen in CT-1−/− neonates also related to a change within the BM mi-croenvironment, we cultured BM under osteoblast-differ-

entiating and mineralizing conditions. This showed im-paired ALP activity (Fig. 5F) and mineralization (Fig. 5G)from CT-1−/− BM compared with BM from wildtype litter-mates. CT-1−/− and wildtype BM generated adipocytes tothe same level in response to rosiglitazone (oil red O [�M]:wildtype � 37.1 ± 1.7; CT-1−/− � 36.3 ± 0.7).

As CT-1−/− mice aged, the neonatal osteopenia graduallyreversed. In males, BV/TV was higher in CT-1−/− mice com-pared with wildtype mice at 4, 10, and 26 wk of age (Figs.6A and 6D). TbTh was increased in males at 10 and 26 wkof age (Fig. 6C). BV/TV and TbN were normalized in fe-male CT-1−/− mice at 4 wk, and at 10 and 26 wk of age,BV/TV and TbN were greater in CT-1−/− females comparedwith wildtypes (Figs. 6A and 6B). At 26 wk of age, TbThwas also significantly increased in CT-1−/− females com-pared with wildtypes (Fig. 6C). This was not restricted tothe tibias. Femoral Tb.BMD was elevated in male and fe-male CT-1−/− mice compared with wildtype mice (Fig. 6E).In contrast, cortical BMD was lower in male and femaleCT-1−/− mice compared with wildtype controls (Fig. 6F).Cortical size was not altered; femoral cortical thickness,periosteal circumference, endosteal circumference, andfemoral width were not altered by lack of CT-1 in male orfemale mice at any age studied (data not shown). Because

FIG. 5. Increased osteoclast generation and impaired mineral-ization is reproduced ex vivo using BM of CT-1−/− mice. (A–E)Osteoclasts were generated from CT-1−/− and wildtype BM stimu-lated with RANKL and M-CSF. Osteoclast number (A), size (Band C), nuclei per osteoclast (D), and DC-STAMP expression (E)were increased in osteoclasts formed from CT-1−/− marrow com-pared with wildtype. Mean values and individual observations areshown in B and D. The increase in nuclear number was not sta-tistically significant. (F and G) Primary osteoblast cultures de-rived from male and female CT-1−/− BM showed (F) reduced ALPactivity and reduced mineralization (G) compared with wildtypes.Images in G are representative wells for each experimental group.Data are from three independent experiments. ap < 0.05; bp < 0.01;cp < 0.001 vs. wildtype controls.

FIG. 6. High BV/TV and BMD in older CT-1−/− mice. (A) In-creased BV/TV in male (M) CT-1−/− tibias (filled bars) at 4, 10,and 26 wk of age and in female (F) CT-1−/− tibias at 10 and 26 wkof age compared with wildtypes (empty bars). (B and C) Trabec-ular number (Tb.N) and trabecular thickness (Tb.Th) in CT-1−/−

mice compared with wildtype. (D) von Kossa–stained represen-tative tibias from 10-wk-old male wildtype and CT-1−/− mice;white box indicates region used for histomorphometric analysis.(E) Femoral trabecular BMD (Tb.BMD) measured by pQCT waselevated in 10-wk-old male and female CT-1−/− mice comparedwith wildtype controls. (F) In contrast, cortical BMD (Ct.BMD)was reduced. n � 7–10 animals per group. ap < 0.05; bp < 0.01;cp < 0.001 vs. wildtype age-matched controls.

CARDIOTROPHIN-1 IN BONE 2029

bone length was dramatically reduced in gp130 signalingmutant mice(11) and in LIFR and gp130-null neonates,(9,10)

we measured it in CT-1−/−, but femoral length was only veryslightly reduced (by ∼5%) and only in 26-wk-old CT-1−/−

females (p < 0.05), so CT-1 is clearly not the maingp130:LIFR-signaling cytokine regulating bone size.

The elevated osteoclast surface and size observed in CT-1−/− neonates was retained in adult CT-1−/− mice (Fig. 7A).At all ages and in both sexes, the osteoclasts in vivo werelarger and had a greater area per cell than wildtype osteo-clasts (Fig. 7B). The average nuclei number per osteoclastwas not significantly changed (Fig. 7C). CT-1−/− osteoclastsseem to have impaired function in vivo throughout life,because BV/TV is higher than wildtype and cartilage rem-nants are retained in the trabecular bone(11,12) in 4-wk-oldfemale and 10-wk-old male CT-1−/− mice (Fig. 7D).

As well as impaired bone resorption, CT-1−/− neonateshad very few osteoblasts, but this was not detected in theadults. ObS/BS was not altered in 4-, 10-, or 26-wk-old maleor female CT-1−/− mice (Fig. 7E). The only indications of

reduced bone formation in older mice were reduced osteoidvolume in 10-wk-old female CT-1−/− mice (Fig. 7F) and lowMAR and mineralizing surface in 10-wk-old male CT-1−/−

mice (MS/BS [%]: wildtype � 24.7 ± 2.7; CT-1−/− � 16.4 ±1.5, p < 0.05; MAR [�m/d]: wildtype � 1.83 ± 0.09; CT-1−/−

� 1.21 ± 0.08, p < 0.01). These defects were not detected inthe 10-wk-old female mice, nor in males or females at anyother time point.

DISCUSSION

CT-1 is an important cytokine in cardiac biology, liverpathogenesis, and motoneuron function, and we have nowshown a critical role in bone biology. CT-1 is expressed inosteoclasts, stimulates bone formation in vitro and in vivo,and is essential for normal bone resorption and neonatalbone formation. Taken together with the osteoblastic ex-pression of LIFR and the known communication betweenthe cell types, this suggests CT-1 may be a coupling factorproduced by the resorbing osteoclast that promotes boneformation.

The structure of trabecular and cortical bone is con-stantly changing because of repeated cycles of bone remod-eling. In these continual cycles where osteoclasts resorbbone that is replaced by osteoblastic bone formation, thebalance between the cell activities must be equal so that anadequate skeletal structure is maintained. The process bywhich the cell activities are matched is known as coupling.A number of suggestions have been made as to how osteo-clasts can control the function of osteoblasts, including therelease of factors from the resorbed bone matrix or releaseof locally acting factors, known as “coupling factors,” thatcan stimulate bone formation.(6,7) CT-1 provides an appeal-ing candidate for this role, because it is a stimulus of boneformation that is detected in osteoclasts, although it shouldbe noted that CT-1 has also been detected in human se-rum,(36) suggesting an endocrine role is also possible.

As well as indirectly stimulating osteoclast differentiationthrough osteoblasts,(5) we show here that CT-1, like LIFand IL-11,(28,31) stimulates bone formation in vivo and invitro. The CT-1–induced increase in C/EBP� indicates amechanism for this effect. Whereas C/EBP� stimulates adi-pogenesis by upregulating C/EBP� expression,(37) it alsoacts synergistically with runx2 to activate osteocalcin tran-scription through a C/EBP enhancer element.(33,34) CT-1also stimulates runx2 activation at OSE2 sites, althoughosteocalcin mRNA expression was not increased within 48h of CT-1 administration. However, it is very likely thatC/EBP� effects on bone formation are not limited to en-hancing osteocalcin expression because the promoter ofother genes that regulate bone formation includingSmad6(38) and Nell-1(39) also contain OSE2 sites. C/EBP�also modifies expression of other factors that increase boneformation including 25-hydroxyvitamin-D3-24-hydroxy-lase(40) and IGF-1.(41) The importance of C/EBPs in bonebiology is indicated by low bone mass in mice transgenic forC/EBP,(42) which heterodimerizes with C/EBP� andC/EBP� to inhibit osteocalcin transcription.

Although many other cytokines signal throughgp130:LIFR, they do not compensate for the lack of CT-1 in

FIG. 7. Increased osteoclasts, but less resorption in CT-1−/−

mice, and normal osteoblast numbers. (A and B) Osteoclast sur-face (OcS/BS) and osteoclast area per osteoclast (OcAr) wereelevated in male (M) and female (F) CT-1−/− mice (filled bars)compared with wildtype (empty bars). (C) Nuclear number perosteoclast (NNc/Oc) was not altered in CT-1−/− compared withwildtype. (D) Cartilage volume as a percentage of bone volume(CtgV/BV) was elevated in CT-1−/− mice compared with wildtype.(E) Neither male nor female CT-1−/− mice showed any alterationin tibial osteoblast surface (ObS/BS) at any time point from 4 wkonward compared with wildtype animals (empty bars). (F) In fe-male mice only, osteoid volume (OV/BV) was reduced at 10 wk.ap < 0.05; bp < 0.01; cp < 0.001 vs. wildtype controls. n � 7–10animals per group, except for 26-wk-old wildtype females, whereonly one sample had BV/TV > 0.5 allowing measurement of cellsurface parameters.

WALKER ET AL.2030

the knockout mice. The striking parallel between the CT-1−/− neonate bone phenotype and that of mice renderednull for either receptor subunit(9,10) also suggests the tra-becular phenotype of gp130 and LIF-R knockouts may oc-cur specifically because of the lack of CT-1 signaling. How-ever, the dwarfism and early lethality caused by the absenceof gp130 and LIFR was not detected in CT-1−/−, indicatingthat another family member or members are critical fornormal bone length and for the aspects of the gp130 andLIFR phenotypes that cause early death.

The reversal of the trabecular bone phenotype as CT-1−/−

mice age is striking. In normal neonate mice, trabecularbone forms very rapidly; growth plate cartilage is resorbedby osteoclasts and replaced with bone by osteoblasts. Thelow BV/TV in CT-1−/− neonates indicates that, because car-tilage resorption is impaired, the effect of reduced boneformation must be dominant at this stage of trabecular de-velopment. As the mice age, despite reduced potential ofCT-1−/− progenitor cells in the BM to differentiate into os-teoblasts, ObS/BS in adult CT-1−/− bone is normal, and anychange in bone formation in vivo is mild, sex specific, andtransient. The low level of bone resorption is retained, how-ever, and as a result, the trabecular phenotype reversesleading to a mild osteopetrosis.

The predominance of the osteoclast defect on bone struc-ture in adult mice also relates to the way trabecular bone isremodeled with age. Repeated remodeling cycles replacethe many thin trabeculae generated from the growth platewith a structure containing fewer thickened trabeculae (i.e.,Tb.N reduces and Tb.Th increases). In CT-1−/− mice, thintrabeculae are not removed because of the low level ofresorption, and Tb.N remains high. Any reduction in boneformation does not seem to contribute to the trabecularphenotype in adult CT-1−/− mice, because the osteoblast-dependent increase in Tb.Th. associated with age still oc-curs in the absence of CT-1.

The many osteoclasts detected in CT-1−/− mice contrastswith CT-1 stimulation of osteoclast formation in the pres-ence of osteoblasts.(5) Because high osteoclast numbers inthe absence of CT-1 were maintained ex vivo, the increasedpotential for osteoclast formation must relate to a change inthe BM microenvironment. The contrast in resorptive ac-tivity of the osteoclasts in vivo and ex vivo suggests thedefect in resorption is not cell autonomous, and osteoclastactivity is modified through a paracrine or endocrinemechanism in the absence of CT-1. Because LIFR is ex-pressed by BM cells, including macrophages and osteo-blasts, but not by osteoclasts,(30) any local effect of CT-1 onosteoclast function must be mediated by other cells in theBM microenvironment, such as the osteoblast. This wouldinclude the possibility that CT-1–deficient osteoblastsmanufacture a bone matrix with altered composition.

CT-1 is an important cytokine for many biological sys-tems; we now show it is also important for the skeleton. Itis expressed in differentiated osteoclasts, is essential fornormal bone resorption, and is capable of stimulating boneformation in vitro and in vivo. CT-1 may be one of thecoupling factors signaling from the active osteoclast to theosteoblast to promote bone formation and is necessary fornormal bone formation and bone resorption.

ACKNOWLEDGMENTS

The authors thank Dr John Wark and Susan Kantor atDepartment of Medicine, Royal Melbourne Hospital, foruse of pQCT, Dr Bettina Holtmann and Prof MichaelSendtner, University of Wurzburg, for CT-1−/− mice, andProf Gerard Karsenty, Columbia University, for the6xOSE2 construct. We also thank staff at the BioresourcesCentre, St Vincent’s Health for excellent animal care. Thework was supported by NHMRC (Australia) ProgramGrant 345401 to NAS, MTG, and TJM. NAS is supportedby a NHMRC (Australia) Senior Research Fellowship.

REFERENCES

1. Bellido T, Borba VZ, Roberson P, Manolagas SC 1997 Acti-vation of the Janus kinase/STAT (signal transducer and acti-vator of transcription) signal transduction pathway by interleu-kin-6-type cytokines promotes osteoblast differentiation.Endocrinology 138:3666–3676.

2. Jay PR, Centrella M, Lorenzo J, Bruce AG, Horowitz MC 1996Oncostatin-M: A new bone active cytokine that activates os-teoblasts and inhibits bone resorption. Endocrinology137:1151–1158.

3. Udagawa N, Takahashi N, Katagiri T, Tamura T, Wada S,Findlay DM, Martin TJ, Hirota H, Taga T, Kishimoto T, SudaT 1995 Interleukin (IL)-6 induction of osteoclast differentia-tion depends on IL-6 receptors expressed on osteoblastic cellsbut not on osteoclast progenitors. J Exp Med 182:1461–1468.

4. Horwood NJ, Elliott J, Martin TJ, Gillespie MT 1998 Osteo-tropic agents regulate the expression of osteoclast differentia-tion factor and osteoprotegerin in osteoblastic stromal cells.Endocrinology 139:4743–4746.

5. Richards CD, Langdon C, Deschamps P, Pennica D, Shaugh-nessy SG 2000 Stimulation of osteoclast differentiation in vitroby mouse oncostatin M, leukaemia inhibitory factor, cardiotro-phin-1 and interleukin 6: Synergy with dexamethasone. Cyto-kine 12:613–621.

6. Martin TJ, Sims NA 2005 Osteoclast-derived activity in thecoupling of bone formation to resorption. Trends Mol Med11:76–81.

7. Sims NA, Gooi JH In Press Bone Remodeling: Multiple cel-lular interactions required for coupling of bone formation andresorption. Semin Cell Dev Biol (in press).

8. Dagoneau N, Scheffer D, Huber C, Al-Gazali LI, Di Rocco M,Godard A, Martinovic J, Raas-Rothschild A, Sigaudy S, UngerS, Nicole S, Fontaine B, Taupin JL, Moreau JF, Superti-FurgaA, Le Merrer M, Bonaventure J, Munnich A, Legeai-Mallet L,Cormier-Daire V 2004 Null leukemia inhibitory factor receptor(LIFR) mutations in Stuve-Wiedemann/Schwartz-Jampel type2 syndrome. Am J Hum Genet 74:298–305.

9. Ware CB, Horowitz MC, Renshaw BR, Hunt JS, Liggitt D,Koblar SA, Gliniak BC, McKenna HJ, Papayannopoulou T,Thoma B, Cheng L, Donovan PJ, Peschon JJ, Bartlett PF,Willis CR, Wright BD, Carpenter MK, Davison BL, GearingDP 1995 Targeted disruption of the low-affinity leukemia in-hibitory factor receptor gene causes placental, skeletal, neuraland metabolic defects and results in perinatal death. Develop-ment 121:1283–1299.

10. Shin HI, Divieti P, Sims NA, Kobayashi T, Miao D, KaraplisAC, Baron R, Bringhurst R, Kronenberg HM 2004 Gp130-mediated signaling is necessary for normal osteoblastic func-tion in vivo and in vitro. Endocrinology 145:1376–1385.

11. Sims NA, Jenkins BJ, Quinn JM, Nakamura A, Glatt M,Gillespie MT, Ernst M, Martin TJ 2004 Glycoprotein 130 regu-lates bone turnover and bone size by distinct downstream sig-naling pathways. J Clin Invest 113:379–389.

12. Sims NA, Jenkins BJ, Nakamura A, Quinn JM, Li R, Gillespie

CARDIOTROPHIN-1 IN BONE 2031

MT, Ernst M, Robb L, Martin TJ 2005 Interleukin-11 ReceptorSignaling Is Required for Normal Bone Remodeling. J BoneMiner Res 20:1093–1102.

13. Poli V, Balena R, Fattori E, Markatos A, Yamamoto M,Tanaka H, Ciliberto G, Rodan GA, Costantini F 1994 Inter-leukin-6 deficient mice are protected from bone loss caused byestrogen depletion. EMBO J 13:1189–1196.

14. Wong PK, Quinn JM, Sims NA, van Nieuwenhuijze A, Camp-bell IK, Wicks IP 2006 Interleukin-6 modulates production ofT lymphocyte-derived cytokines in antigen-induced arthritisand drives inflammation-induced osteoclastogenesis. ArthritisRheum 54:158–168.

15. Liu F, Aubin JE, Malaval L 2002 Expression of leukemia in-hibitory factor (LIF)/interleukin-6 family cytokines and recep-tors during in vitro osteogenesis: Differential regulation bydexamethasone and LIF. Bone 31:212–219.

16. Sheng Z, Pennica D, Wood WI, Chien KR 1996 Cardiotro-phin-1 displays early expression in the murine heart tube andpromotes cardiac myocyte survival. Development 122:419–428.

17. Oppenheim RW, Wiese S, Prevette D, Armanini M, Wang S,Houenou LJ, Holtmann B, Gotz R, Pennica D, Sendtner M2001 Cardiotrophin-1, a muscle-derived cytokine, is requiredfor the survival of subpopulations of developing motoneurons.J Neurosci 21:1283–1291.

18. Sims NA, Clement-Lacroix P, Da Ponte F, Bouali Y, Binart N,Moriggl R, Goffin V, Coschigano K, Gaillard-Kelly M, Kop-chick J, Baron R, Kelly PA 2000 Bone homeostasis in growthhormone receptor-null mice is restored by IGF-I but indepen-dent of Stat5. J Clin Invest 106:1095–1103.

19. Sims NA, Brennan K, Spaliviero J, Handelsman DJ, Seibel MJ2006 Perinatal testosterone surge is required for normal adultbone size but not for normal bone remodeling. Am J PhysiolEndocrinol Metab 290:E456–E462.

20. Horn D, Rivas P, McCluskey B, Mundy GR, Gutierrez G 2002A new staining technique for undecalcified bone sections thatenhances visualization of fluorochromes. J Bone Miner Res17:S416.

21. Sims NA, Dupont S, Krust A, Clement-Lacroix P, Minet D,Resche-Rigon M, Gaillard-Kelly M, Baron R 2002 Deletion ofestrogen receptors reveals a regulatory role for estrogen recep-tors-beta in bone remodeling in females but not in males. Bone30:18–25.

22. Sims NA, White CP, Sunn KL, Thomas GP, Drummond ML,Morrison NA, Eisman JA, Gardiner EM 1997 Human andmurine osteocalcin gene expression: Conserved tissue re-stricted expression and divergent responses to 1,25-dihydroxy-vitamin D3 in vivo. Mol Endocrinol 11:1695–1708.

23. Quinn JM, Whitty GA, Byrne RJ, Gillespie MT, Hamilton JA2002 The generation of highly enriched osteoclast-lineage cellpopulations. Bone 30:164–170.

24. Hodge JM, Kirkland MA, Aitken CJ, Waugh CM, Myers DE,Lopez CM, Adams BE, Nicholson GC 2004 Osteoclastic po-tential of human CFU-GM: Biphasic effect of GM-CSF. JBone Miner Res 19:190–199.

25. Aoki K, Didomenico E, Sims NA, Mukhopadhyay K, Neff L,Houghton A, Amling M, Levy JB, Horne WC, Baron R 1999The tyrosine phosphatase SHP-1 is a negative regulator of os-teoclastogenesis and osteoclast resorbing activity: Increased re-sorption and osteopenia in me(v)/me(v) mutant mice. Bone25:261–267.

26. Kalajzic I, Kalajzic Z, Kaliterna M, Gronowicz G, Clark SH,Lichtler AC, Rowe D 2002 Use of type I collagen green fluo-rescent protein transgenes to identify subpopulations of cells atdifferent stages of the osteoblast lineage. J Bone Miner Res17:15–25.

27. Allan EH, Ho PW, Umezawa A, Hata J, Makishima F,Gillespie MT, Martin TJ 2003 Differentiation potential of amouse bone marrow stromal cell line. J Cell Biochem 90:158–169.

28. Cornish J, Callon K, King A, Edgar S, Reid IR 1993 The effectof leukemia inhibitory factor on bone in vivo. Endocrinology132:1359–1366.

29. Calvi LM, Sims NA, Hunzelman JL, Knight MC, GiovannettiA, Saxton JM, Kronenberg HM, Baron R, Schipani E 2001Activated parathyroid hormone/parathyroid hormone-relatedprotein receptor in osteoblastic cells differentially affects cor-tical and trabecular bone. J Clin Invest 107:277–286.

30. Allan EH, Hilton DJ, Brown MA, Evely RS, Yumita S, Met-calf D, Gough NM, Ng KW, Nicola NA, Martin TJ 1990 Os-teoblasts display receptors for and responses to leukemia-inhibitory factor. J Cell Physiol 145:110–119.

31. Takeuchi Y, Watanabe S, Ishii G, Takeda S, Nakayama K,Fukumoto S, Kaneta Y, Inoue D, Matsumoto T, Harigaya K,Fujita T 2002 Interleukin-11 as a stimulatory factor for boneformation prevents bone loss with advancing age in mice. JBiol Chem 277:49011–49018.

32. Kawashima I, Ohsumi J, Mita-Honjo K, Shimoda-Takano K,Ishikawa H, Sakakibara S, Miyadai K, Takiguchi Y 1991 Mo-lecular cloning of cDNA encoding adipogenesis inhibitory fac-tor and identity with interleukin-11. FEBS Lett 283:199–202.

33. Gutierrez S, Javed A, Tennant DK, van Rees M, Montecino M,Stein GS, Stein JL, Lian JB 2002 CCAAT/enhancer-bindingproteins (C/EBP) beta and delta activate osteocalcin gene tran-scription and synergize with Runx2 at the C/EBP element toregulate bone-specific expression. J Biol Chem 277:1316–1323.

34. Shin CS, Jeon MJ, Yang JY, Her SJ, Kim D, Kim SW, Kim SY2006 CCAAT/enhancer-binding protein delta activates theRunx2-mediated transcription of mouse osteocalcin II pro-moter. J Mol Endocrinol 36:531–546.

35. Krishnan V, Moore TL, Ma YL, Helvering LM, Frolik CA,Valasek KM, Ducy P, Geiser AG 2003 Parathyroid hormonebone anabolic action requires Cbfa1/Runx2-dependent signal-ing. Mol Endocrinol 17:423–435.

36. Talwar S, Downie PF, Squire IB, Barnett DB, Davies JD, NgLL 1999 An immunoluminometric assay for cardiotrophin-1: Anewly identified cytokine is present in normal human plasmaand is increased in heart failure. Biochem Biophys Res Com-mun 261:567–571.

37. Lane MD, Tang QQ, Jiang MS 1999 Role of the CCAATenhancer binding proteins (C/EBPs) in adipocyte differentia-tion. Biochem Biophys Res Commun 266:677–683.

38. Wang Q, Wei X, Zhu T, Zhang M, Shen R, Xing L, O’KeefeRJ, Chen D 2007 Bone morphogenetic protein 2 activatesSmad6 gene transcription through bone-specific transcriptionfactor Runx2. J Biol Chem 282:10742–10748.

39. Truong T, Zhang X, Pathmanathan D, Soo C, Ting K 2007Craniosynostosis-associated gene nell-1 is regulated by runx2. JBone Miner Res 22:7–18.

40. Dhawan P, Peng X, Sutton AL, MacDonald PN, Croniger CM,Trautwein C, Centrella M, McCarthy TL, Christakos S 2005Functional cooperation between CCAAT/enhancer-bindingproteins and the vitamin D receptor in regulation of 25-hydroxyvitamin D3 24-hydroxylase. Mol Cell Biol 25:472–487.

41. Centrella M, Christakos S, McCarthy TL 2004 Skeletal hor-mones and the C/EBP and Runx transcription factors: Inter-actions that integrate and redefine gene expression. Gene342:13–24.

42. Shirakawa K, Maeda S, Gotoh T, Hayashi M, Shinomiya K,Ehata S, Nishimura R, Mori M, Onozaki K, Hayashi H, Ue-matsu S, Akira S, Ogata E, Miyazono K, Imamura T 2006CCAAT/enhancer-binding protein homologous protein(CHOP) regulates osteoblast differentiation. Mol Cell Biol26:6105–6116.

Address reprint requests to:Natalie Sims, PhD

St Vincent’s Institute9 Princes Street

Fitzroy, Victoria 3065, AustraliaE-mail: [email protected]

Received in original form March 4, 2008; revised form July 4, 2008;accepted July 28, 2008.

WALKER ET AL.2032