This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/authorsrights

Author's personal copy

Original Full Length Article

Thrombin receptor deficiency leads to a high bone mass phenotype bydecreasing the RANKL/OPG ratio

Kukiat Tudpor a,1, Bram C.J. van der Eerden b,1, Prapaporn Jongwattanapisan a, Joris J.T.H. Roelofs c,Johannes P.T.M. van Leeuwen b, René J.M. Bindels a, Joost G.J. Hoenderop a,⁎a Department of Physiology, Radboud Institute for Molecular Life Sciences, Radboud University Medical Center, The Netherlandsb Department of Internal Medicine, Erasmus Medical Center, Rotterdam, The Netherlandsc Department of Pathology, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands

a b s t r a c ta r t i c l e i n f o

Article history:Received 21 August 2014Revised 26 October 2014Accepted 10 November 2014Available online 18 November 2014

Edited by Hong-Hee Kim

Keywords:Thrombin receptorOsteoblastOsteopetrosisOsteoprotegerinReceptor activator of nuclear factor-κB ligand

Thrombin and its receptor (TR) are, respectively, expressed in osteoclasts and osteoblasts. However, theirphysiological roles on bone metabolism have not been fully elucidated. Here we investigated the bonemicroarchitecture by micro-computed tomography (μCT) and demonstrated increased trabecular and corticalbone mass in femurs of TR KO mice compared to WT littermates. Trabecular thickness and connectivity weresignificantly enhanced. The physiological role of TR on both inorganic and organic phases of bone is illustratedby a significant increase in BMD and a decrease in urinary deoxypyridinoline (DPD) crosslink concentration inTR KO mice. Moreover, TR KO cortical bone expanded and had a higher polar moment of inertia (J), implyingstronger bone. Bone histomorphometry illustrated unaltered osteoblast and osteoclast number and surface infemoral metaphyses, indicating that thrombin/TR regulates osteoblasts and osteoclasts at functional levels.Serum analysis showed a decrease in RANKL and an increase in osteoprotegerin (OPG) levels and reflected a re-duced RANKL/OPG ratio in the TR KO group. In vitro experiments using MC3T3 pre-osteoblasts demonstrated aTR-dependent stimulatory effect of thrombin on the RANKL/OPG ratio. This effect was blocked by TR antagonistand p42/p44-ERK inhibitor. In addition, thrombin also intensified p42/p44-ERK expression and phosphorylation.In conclusion, the thrombin/TR systemmaintains normal bone remodeling by activating RANKL and limiting OPGsynthesis by osteoblasts through the p42/44-ERK signaling pathway. Consequently, TR deficiency inhibits osteo-clastogenesis, resulting in a high bone mass phenotype.

© 2014 Elsevier Inc. All rights reserved.

Introduction

Osteoblasts together with osteoclasts are responsible for bone re-modeling. The bone-forming osteoblasts produce an organic collage-nous matrix and mineralize it with an inorganic calcium-phosphatehydroxyapatite [Ca5(PO4)3(OH)] [1]. The bone-resorbing osteoclastsdegrade the mineralized matrix by two means—creating acidic milieuto resorb the inorganic mineralized phase and secreting proteases todegrade organic collagen I-based extracellular matrix (ECM), [1,2].Bone remodeling is regulated by the coupling mechanism of osteoblastand osteoclast activities [3]. On the one hand, osteoclast developmentrequires a stimulation of its surface receptor activator of nuclearfactor-κB (RANK) by RANK ligand (RANKL) secreted from osteoblasts[4]. On the other hand, osteoblasts secrete a decoy receptor osteopro-tegerin (OPG) that specifically binds and neutralizes RANKL [1,4].

Therefore, a decrease in the RANKL/OPG ratio determines dominationof bone resorption over bone formation and vice versa. Disturbancesof RANKL/OPG causes skeletal abnormalities represented by reduced(osteoporosis) and increased (osteopetrosis) bone mass [5].

The serine protease thrombin (coagulation factor II, EC 3.4.21.5),with known function to convert fibrinogen to fibrin in the final step ofthe coagulation cascade, is also synthesized and secreted by osteoclasts[6,7]. Heterozygous thrombin-deficientmice are lethal in the embryonicphase, while surviving neonates died after a few days due to severehemorrhage [8]. Among other proteases, thrombin has been shown toplay a role in bone remodeling including direct cleavage of the boneprotein osteopontin, which is important for anchoring osteoclasts tothemineralizedmatrix [9]. Recently, thrombin has been found to inhibitosteoclast differentiation and down-regulate RANK expression in isolat-ed pre-osteoclast cultures [10]. In addition, thrombin can also exert itsbiological effect via protease-activated receptors (PARs)-1, -3, and -4[11]. PAR-1 (thrombin receptor, TR) is expressed in osteoblasts, butnot in osteoclasts [12]. PAR-3 is not expressed in murine osteoblasts,whereas PAR-4 has been reported only in osteoblast-like calvarial cells[6]. Therefore, TR is responsible for almost, if not all, thrombin signaling

Bone 72 (2015) 14–22

⁎ Corresponding author at: Department of Physiology (286), Radboud UniversityMedical Center, P.O. Box 9101, 6500 HB Nijmegen, The Netherlands.

E-mail address: Joost.Hoenderop@radboudumc.nl (J.G.J. Hoenderop).1 These authors contributed equally to this work.

http://dx.doi.org/10.1016/j.bone.2014.11.0048756-3282/© 2014 Elsevier Inc. All rights reserved.

Contents lists available at ScienceDirect

Bone

j ourna l homepage: www.e lsev ie r .com/ locate /bone

Author's personal copy

inmouse osteoblast cells [13]. TR is a transmembraneGprotein-coupledreceptor (GPCR) of which the extracellular domain can be activatedupon proteolytic cleavage by thrombin [6]. In vitro experiments showedthat activation of TR increased the proliferation rate of osteoblasts andstimulated the activity of alkaline phosphatase (ALP), an osteoblast dif-ferentiationmarker [14]. TRhas been reported to convey its intracellularsignal via the PLC-β, p42/44-ERK, and PKC pathways [15,16]. By usingin vitro calvarial and long bone cultures, thrombin induced bone resorp-tion, while TR knockout mice (TR KO) manifested decreased migrationand proliferation of mesenchymal cells into the bone fracture area ofbone-drilled model [11,17,18]. These data suggest a potentially physio-logical importance of TR. However, the role of the thrombin/TR systemin bone homeostasis has not been investigated. The purpose of thispresent study was, therefore, to investigate the function of thrombin/TR in bone metabolism using an in vivo TR KO mouse model andin vitro osteoblast cell cultures. We hypothesize that the thrombin/TRsystem is necessary for the normal bone remodeling process.

Materials and methods

Animal experiments and collection of samples

TR KO mice and WT littermates (75% C57BI/6 × 25% 129/Sv back-ground, 10–12 weeks old) purchased, respectively, from The JacksonLaboratory (Bar Harbor, ME, USA) and Charles River (Maastricht, TheNetherlands) were genotyped. All experiments were approved by theInstitutional Animal Care and Use Committee of the University ofAmsterdam. Four male mice of WT and the other four male TR KOmice were placed in metabolic cages. Twenty four-hour food andwater consumption; and fecal and urine excretions were collected forweight, volume, electrolyte, and DPD determinations. The mice wereanesthetized under 2% (v/v) isoflurane. Blood was collected for electro-lyte, RANKL, and OPG determinations before termination. Bones wereremoved, rinsed in iced-cold PBS and kept in 75% (v/v) ethanol at 4 °Cfor further μCT analyses. Kidney and duodenal tissues were removed,rinsed in iced-cold PBS and kept in −80 °C and/or PLP for further RNAand protein extraction, and cryosection, respectively as described previ-ously [19].

Serum, urine, and fecal analyses

Serum and urine Ca2+ concentrations were determined by a colori-metric assay as described previously [20]. Fecal samples were desiccat-ed at 50 °C overnight and stored at−80 °C until analysis. Each 24-h fecalsample was dissolved in 20 ml nitric acid (1.5 M) at room temperaturefor 12 h. The solution was adjusted to a total volume of 20 ml withdeionized water, and Ca2+ was determined with the same methodsfor serum and urine Ca2+ analyses. A flame spectrophotometerFCM6343 (Eppendorf) was used to measure serum and urine Na+ con-centrations and urineK+ concentrations. Urine osmolalitywas obtainedusing a Halbmikro-Osmometer K-7400 (Knauer).

μCT measurements

Femurs from male WT and TR KO mice (n = 4) were scanned at aresolution of 9 μm, using a SkyScan 1172 system (SkyScan). Accordingto guidelines recently published [21], the following settings wereused: X-ray power and tube current were 40 kV and 0.25 mA, respec-tively. Beamhardening (20%)was reduced using a 1mmaluminium fil-ter, ring-artifacts were reduced (set at 5), exposure time was 5.9 s andan average of three pictures was taken at each angle (0.9°) to generatefinal images. Using different software packages from SkyScan (NRecon,CtAn and Dataviewer), bone microarchitectural parameters wereassessed in trabecular and cortical bones of all mice. The trabecularbone parameters: trabecular tissue volume, bone volume, volume frac-tion (BV/TV), thickness, number and patterning factor (connectivity of

trabeculae) were determined in the distal metaphysis of the femur(scan area of 1.3 mm from distal growth plate towards femoral center).In the mid-diaphysis (scan area of 0.43 mm in the femoral center), cor-tical volume, thickness, J (proxy for bone strength) and perimeter wereanalyzed. For image processing, trabecular bone was manually selectedand cortical bonewas automatically selected. In addition, trabecular andcortical BMD was measured on the basis of calibration scanning, usingtwo phantoms with known density (0.25 mg/cm2 and 0.75 mg/cm2;SkyScan) under identical conditions as for the femurs (method notefrom SkyScan provided on http://www.skyscan.be). All terms used inμCT experiments comply with the standard nomenclature, symbols,and units previously described [21].

Bone histomorphometry

After excision, femurs were routinely embedded in methyl-metacrylate as described before [22]. Sections of 6 μm were subjectedto tartrate-resistant acid phosphatase (TRAP) staining. Sections weredeacrylated, hydrated and rinsed in 0.2 M sodium acetate/50 mMtartaric acid for 5 min. Naphthol AS-MX (0.5 mg/ml) and 1.1 mg/mlFast red TR salt (both from Sigma) were added and incubated for120 min at 37 °C. Counterstaining was performed with hematoxylinfor 5 s and after air-drying; the sections were embedded in Permount(Thermo Fisher Scientific). For osteoblast and osteocytemeasurements,sections were stained with a Goldner staining as described before [23].Images were taken from the TRAP and Goldner stainings with a NikonEclipse E400 system (Nikon). Measurements were performed, usingthe software package Bioquant (Version 7.20; Bioquant Image AnalysisCorporation).

Preparation of cultured pre-osteoblastic cells

MC3T3pre-osteoblastic cellswerepurchased fromATCC andgrown inα-minimal essential medium (α-MEM) (Gibco) supplemented with 10%(v/v) heat-inactivated fetal bovine serum (Gibco), 50 μg/ml penicillin/streptomycin (Gibco), 50 μg/ml vitamin C (Sigma), and 10 mM β-glycerophosphate (Sigma) in 5% (v/v) CO2 at 37 °C. The medium wasrefreshed twice a week. The cells were seeded at 2.5 × 104 cells/cm2

into 6-well plates and left in the incubator for 24 h prior to exposureto different treatments for 16 h.

qPCR expression analyses

To evaluate mRNA expression of RANKL and OPG, total RNA was ex-tracted from thrombin-treated (16 h) MC3T3 cells using TriZol TotalRNA Isolation Reagent (Life Technologies BRL) according to themanufacturer's protocol. The obtained RNA was subjected to DNasetreatment (Promega) to prevent genomic DNA contamination. Thereaf-ter, 2 μg of RNA was reverse transcribed by Moloney-Murine LeukemiaVirus-Reverse Transcriptase (Invitrogen). The cDNA was used to deter-mine mRNA expression levels by SYBR green PCR both of the targetgenes of interest and of the housekeeping gene Gapdh as an endoge-nous control. Primers targeting the genes of interest were designedusing the computer program Primer3 (v.0.4.0) and are listed in Table 1.

Table 1Mus musculus oligonucleotide sequences used in RT-PCR.

Gene Forward primer Reverse primer

Tnfsf11 CATGTGCCACTGAGAACCTTGAA CAGGTCCCAGCGCAATGTAACTnfrsf11b CAATGGCTGGCTTGGTTTCATAG CTGAACCAGACATGACAGCTGGAGapdh TAACATCAAATGGGGTGAGG GGTTCACACCCATCACAAAC

Tnfsf11, receptor activator of nuclear factor-κB ligand gene; Tnfrsf11b, osteoprotegeringene; Gapdh, glyceraldehyde-3-phosphate dehydrogenase gene.

15K. Tudpor et al. / Bone 72 (2015) 14–22

Author's personal copy

Western blot analysis

MC3T3 cell lysates were added to sample buffer and incubated for30 min at 37 °C. Homogenates of mouse duodenal and kidney tissues;and cells were prepared and analyzed as previously described [24,25].Protein concentrations were determined by a BCA protein assay kit(Bio-Rad) and volumes of sampleswere normalized accordingly. Subse-quently, 30-μg sampleswere loaded and separated on 8% SDS-PAGE gelsfor p42/44-ERK. Protein fractions were transferred to polyvinylidenedifluoride–nitrocellulose membranes (Immobilon-P, Millipore). Pro-teins on the membranes were detected using rabbit monoclonal anti-phospho-p42/44 ERK and anti-p42/44 ERK antibodies (Cell SignalingTechnology). Mouse polyclonal anti-β-actin Ab was from Santa Cruz.

Analytical procedures

RANKL was measured with the DuoSet ELISA development kit (R&DSystems, Inc.). OPGwasmeasuredwith anOPG ELISA kit (Abnova). DPDcrosslinks were measured with the MicroVue total DPD EIA kit (QuidelCorporation). All protocols were in accordance with the manufacturer'sinstructions. 96-well plateswere read at the specifiedwavelengths, on aBenchmark Plus™ microplate spectrophotometer (Bio-Rad).

45Ca2+ transport measurement in the intestinal Caco-2 cell line

For radioactive 45Ca2+ transport experiments, Caco-2 cells wereseeded onto 0.33 cm2 polycarbonate Transwell® inserts (CorningCostar) for 21 days in DMEM (Bio Whittaker) supplemented with 10%(v/v) FCS (PAA), 2 mM L-glutamine, and 10 μl/ml non-essential aminoacids at 37 °C in a humidity controlled incubator with 5% (v/v) CO2

before use. The day prior to the experiments, transepithelial electricalresistance (TEER) was measured using an epithelial volt-ohmmeter(EVOM,World Precision Instruments). On the experimental day, culturemedium was removed and replaced with physiological salt solution(PSS: 140 mM NaCl, 2 mM KCl, 1 mM K2HPO4, 1 mM MgCl2, 1 mMCaCl2, 5 mM D-glucose, 5 mM L-alanine, 5 μM indomethacin, 10 mMHEPES/Tris, pH 7.4). The apical compartment contained 100 μl andbasolateral compartment contained 600 μl. After 30 min of incubation,PSS was replaced with the same volumes of the pre-warmed identicalsolution with and without 10-nM thrombin (T4648, Sigma); apical me-dium contains 3 μCi/ml 45Ca2+. Ten microliters of basolateral mediumwas collected at time points 10, 30, and 60 min and analyzed for radio-activity in a Perkin Elmer liquid scintillation counter. Unidirectional fluxfrom the apical side to the basolateral side (JA → B) was calculated withEqs. (1) and (2) as previously described [26].

JA→B ¼ RA→B

SA � Að1Þ

SA ¼ CA

CTð2Þ

where RA → B was the rate of tracer appearance in the basolateral side(cpm/min); SA was the specific activity in the apical side (cpm/nmol);A was the surface area of the Transwell; CA was a mean of radioactivityof 45Ca2+ in the apical side (cpm); and CT was the total Ca2+ in the api-cal side (nmol).

Compounds

Thrombin, U0126, and U73122 were purchased from Sigma.Thrombin receptor antagonist (SCH79797) was from Axon Medchem.Chelerythrine was from Research Biochemicals International.

Statistical analysis

If not specified otherwise, the data are expressed as mean ± SEM.Multiple sets of data were compared by ANOVA. The significant differ-ences between the means of the two groups were analyzed by pairedStudent's t-test, while multiple comparisons between groups were per-formed by Tukey's post hoc multi analysis. The level of statistical signif-icance is P b 0.05. All data were analyzed by GraphPad Prism (version4.0c for Mac OS X; GraphPad Software).

Results

TR-deficient mice have increased bone mass

Gross observation revealed that TR KO mice have an increasedbody weight compared to WT littermates (27.2 ± 1.9 vs 30.1 ± 0.3 g,P b 0.05). TR KO mice consumed similar amounts of food and waterand excreted the same extent of feces compared to WT mice (3.0 ±1.1 vs 3.6 ± 1.8 g, 4.8 ± 1.5 vs 4.8 ± 0.5 ml, and 2.3 ± 0.6 vs 2.8 ±0.7 g, respectively, P N 0.05). Subsequently, we analyzed the bone tissueof TR KO mice. First, we compared the bone microarchitecturalstructure of the femoral metaphysis and diaphysis between WT andTR KO mice. Data in Τable 2 showed that, in the WT and TR KO groupsrespectively, trabecular bone volume (BV), femoral head volume(Fh.V), and trabecular number (Tb.N) were significantly increased(0.47 ± 0.01 vs 0.81 ± 0.06 mm3, 4.15 ± 0.09 vs 5.06 ± 0.18 mm3,and 3.13 ± 0.04 vs 3.98 ± 0.19 mm−1, respectively, P b 0.05) whereastrabecular separation (Tb.Sp) was decreased (145 ± 0.3 vs 129 ±4.2 μm, P b 0.05) (Table 2). Diaphyseal endocortical volume (Ec.V)was also increased (0.46 ± 0.02 vs 0.54 ± 0.01 mm3). The mice lackingTR had a small, albeit significant, increase in femur length (15.79± 0.08vs 16.26 ± 0.03 mm, P b 0.05).

TR deficiency increased cortical bone size, strength proxy, and bonemineraldensity

At the level of the cortical bone, asmeasured in the distalmetaphysisof the femur, endocortical and cortical volumes were significantlyincreased (3.13 ± 0.07 vs 3.87 ± 0.17 mm3 and 1.01 ± 0.02 vs1.20 ± 0.02 mm3 respectively, P b 0.05, Figs. 1A–C), while diaphysealthickness was reduced (142 ± 1.9 vs 123 ± 2.2 μm, P b 0.05, Fig. 1D).Also a significant increase in diaphyseal bone circumference (corticalperimeter, 10.40 ± 0.02 vs 12.85 ± 0.23 mm, P b 0.05, Fig. 1E) was ob-served. The bone expansion led to a higher J (0.42 ± 0.01 vs 0.60 ±

Table 2Bonemicroarchitectural data and femur length fromwild-type (WT) and thrombin recep-tor-deficient (TR KO) mice.

Parameter WT TR KO

MetaphysisBV (mm3) 0.47 ± 0.01 0.81 ± 0.06a

Fh.V (mm3) 4.15 ± 0.09 5.06 ± 0.18b

Tb.N (mm−1) 3.13 ± 0.04 3.98 ± 0.19b

Tb.Sp (μm) 145 ± 0.3 129 ± 4.2a

DiaphysisEc.V (mm3) 0.46 ± 0.02 0.54 ± 0.01b

Whole femurFemur length (mm) 15.79 ± 0.08 16.26 ± 0.03a

Femurs from wild-type (WT) and thrombin receptor-deficient (TR KO) mice (n = 4)were analyzed in detail by μCT. Parameters indicated are femoral head volume (Fh.V),bone volume, metaphyseal bone volume (BV), trabecular separation (Tb.Sp), trabecularnumber (Tb.N), diaphyseal endocortical volume (Ec.V) and femoral length. Data are pre-sented as mean ± SEM.

a P b 0.05 versus wild-type.b P b 0.01 versus wild-type.

16 K. Tudpor et al. / Bone 72 (2015) 14–22

Author's personal copy

0.01mm4, P b 0.05, Fig. 1F). BMDwas also higher in TR KOmice (1.21±0.004 vs 1.25 ± 0.01 g/cm3, P b 0.05, Fig. 1G).

TR deficiency increased trabecular bone thickness and connectivity

Detailed structural analysis of the femurs revealed that trabecu-lar bone volume fraction in the distal metaphyseal bone compartmentof TR KO mice was elevated (15.14 ± 0.15 vs 20.94 ± 0.01%, P b 0.05,Figs. 2A and B), which is explained by increased trabecular thickness(48.37 ± 0.12 vs 52.53 ± 0.41 μm, P b 0.05) and trabecular number(3.13 ± 0.04 vs 3.98 ± 0.19 mm−1, P b 0.05) (Fig. 2C and Table 2, re-spectively). This led to reduced trabecular separation (145 ± 0.3 vs129 ± 4.2 μm, P b 0.01) and more trabecular connectivity (indicatedby a decrease in trabecular patterning factor, 23.27 ± 0.75 vs 15.65 ±0.76 mm−1, P b 0.01) as shown in Table 2 and Fig. 2D, respectively.

The structure model index estimating whether a bone structure isplate-like (valueof 0) or rod-like (value of 3) revealed that TRdeficiencyleads to amore plate-likemicroarchitecture (1.99±0.04 vs 1.70±0.03,P b 0.01, Fig. 2E).

Histomorphometrical parameters of WT and TR KO mice

In order to investigate bone turnover, histological analyses of femo-ral metaphyseal bone sections from WT and TR KO mice were per-formed with Goldner's trichrome (Figs. 3A and B) and TRAP (Figs. 3Cand D) staining for osteoblasts and osteoclasts, respectively. Osteoblastnumber and surface were not significantly different between the WTand TR KO groups (0.61 ± 0.05 vs 0.91 ± 0.25 no./mm and 1.78 ±0.17 vs 2.96 ± 0.85%, P N 0.05, Figs. 3E and F, respectively). Similarly,Figs. 3G andH also showed no changes in osteoclast number and surface

Fig. 1. Cortical bonemicroarchitecture and bonemineral density of wild-type (WT) and TR-deficient (KO)mice. A, the cross-sectional geometry ofmouse femoral metaphyseswas deter-minedwith μCT. Endocortical and cortical volumesweremeasured at metaphyseal areas (B and C). Cortical thickness, cortical perimeter (mid-shaft circumference), and polar moment ofinertia (J) were obtained from diaphyseal analyses (D, E, and F). Femoral BMD is depicted (G). N = 4 per group, *P b 0.05 in comparison to WT.

Fig. 2. Trabecular bone structural analysis of wild-type (WT) and TR-deficient (KO) mice femurs by μCT. Representative 3D-construction specimens show trabecular bone structure ofWT (upper) and TR KO (lower) mice (A). Arrows indicate majorities of rod-like and plate-like trabecular bones in WT and TR KO mice, respectively. Basic set of variables includesbone volume fraction (BV/TV), trabecular thickness, trabecular patterning factor (1/trabecular connectivity) and structure model index (plate-like, value of 0 and rod-like, value of3) are depicted (B–E). TV, tissue volume; BV, bone volume. N = 4 per group, *P b 0.05 in comparison to WT.

17K. Tudpor et al. / Bone 72 (2015) 14–22

Author's personal copy

when corrected for bone surface (1.69 ± 0.11 vs 1.49 ± 0.21 no./mmand 15.94 ± 1.40 vs 15.12 ± 2.77%, P N 0.05).

TR deficiency decreased urinary bone resorption marker protein, but didnot affect serum and urinary Ca2+ contents

Because bone proteases are involved in ECM degradation [5]and bone mass was increased, urinary DPD crosslinks (degradationproducts of collagen I); and serum and urinary Ca2+ were measured.Urinary DPD concentration in TR KO mice was reduced compared toWT (141 ± 27 vs 79 ± 3 nM in WT and TR KO, respectively, P b 0.05,Fig. 4), indicating that TR is required for proper bone resorption.Electrolyte analyses revealed that urinary Ca2+ excretion was un-changed, but polyuria (0.8 ± 0.2 vs 2.1 ± 0.3 ml, P b 0.05) and alkalineurine (pH 6.7 ± 0.1 vs 7.1 ± 0.1, P b 0.05) were observed (Table 3).Serum Ca2+ concentration was not affected (2.23 ± 0.03 vs 2.16 ±0.15, P N 0.05). Additionally, urinary Na+ excretion was enhancedwhereas serum Na+ and K+ values appeared normal (see Table 3).

TR deficiency decreased serum RANKL/OPG ratio

Based on the fact that TR is expressed in osteoblasts and thefinding of an abnormal increase in bone mass, we hypothesized that

TR deficiency might alter osteoblast function to secrete RANKL andOPG. Fig. 5A shows that the serum RANKL level was significantlydecreased from 1419 ± 576 pg/ml to 355 ± 34 pg/ml in WT and TRKO mice, respectively, P b 0.05, while the OPG level was increased(992±29pg/ml vs 1189±57 pg/ml inWT and TRKOmice, respective-ly, P b 0.05) (Fig. 5B). As a consequence, the RANKL/OPG ratio was sig-nificantly reduced (1.00 ± 0.35 vs 0.22 ± 0.03, P b 0.05, Fig. 5C).

Thrombin upregulated RANKL/OPG ratio in osteoblast-culture medium viaTR-dependent and p42/44-ERK-mediated pathway

In order to elucidate the mechanism of thrombin action, TR-expressing MC3T3 pre-osteoblastic cells [13] were cultured withthrombin-containing medium (0.1–10 nM) for 16 h. The RANKL/OPGratio in the cultured medium was significantly upregulated after incu-bation with 10-nM thrombin (1.00 ± 0.06 vs 1.47 ± 0.13 in controland 10-nM thrombin-treated respectively, P b 0.05, Fig. 6A). RANKLand OPG mRNA expression was analyzed in the cells treated with amaximal dose of thrombin (10 nM). Interestingly, only RANKL mRNAwas increased whereas the OPG mRNA and RANKL/OPG mRNA ratiowere unchanged by thrombin incubation (Figs. 6B–D). We further in-vestigated if thrombin action was dependent on TR and if so, which sig-naling pathway was responsible for this stimulatory effect. Fig. 6Eshows that pre-incubation (30 min prior to thrombin) with TR antago-nist, 20 μMSCH79797, prevents anupregulation of theRANKL/OPG ratioby thrombin (1.00 ± 0.14, 1.47 ± 0.27, and 0.84 ± 0.11 in control, 10-nM thrombin-treated, and 10-nM thrombin-treated + SCH79797, re-spectively). Moreover, the inhibitors for p42/44-ERK (25 μM U0126)and PLC-β (10 μM U73122) blocked the effect of thrombin, whereasPKC inhibitors (10 μM Chelerythrine) failed to reverse the effect ofthrombin (1.87 ± 0.31, P b 0.05 compared to control, Fig. 6E). Further-more, immunoblotting analysis showed that both phosphorylated andtotal p42/44-ERK expression is stimulated by thrombin (100 ± 23 vs170 ± 21 and 100 ± 14 vs 177 ± 2, respectively, P b 0.05, Fig. 6F).

Fig. 3. Bone histomorphometry in wild-type (WT) and TR-deficient (KO) mice. Goldner(A–B) and TRAP (C–D) stainings were performed on femoral bone sections from WT(A and C) and TR KO (B and D) mice and images were taken of the metaphyses. Fromthese stainings, osteoblast number (E) and surface (F) as well as osteoclast number(G) and osteoclast surface (H) per bone surface area were quantified (n = 4 per group).

Fig. 4. Deoxypyridinoline (DPD) crosslink concentration in urine samples of wild-type(WT) and TR-deficient (KO) mice. n = 4 per group. *P b 0.05 in comparison to WT.

Table 3Blood and urine parameters of wild-type (WT) and thrombin receptor-deficient (TR KO)mice.

Parameter WT TR KO

SerumCa2+ (mM) 2.23 ± 0.03 2.16 ± 0.15Na+ (mM) 136.1 ± 0.7 136.8 ± 1.6K+ (mM) 7.1 ± 0.1 7.4 ± 0.3

UrineCa2+ excretion (μmol/24 h) 5.5 ± 1.2 9.5 ± 1.1Na+ excretion (μmol/24 h) 99 ± 21 199 ± 44*K+ excretion (μmol/24 h) 194 ± 48 416 ± 91Osmolality (mOsm) 2430 ± 478 1653 ± 222Volume (ml/24 h) 0.8 ± 0.2 2.1 ± 0.3*pH 6.7 ± 0.1 7.1 ± 0.1*

Data are presented as mean ± SEM. *P b 0.05 versus WT. Wild-type (WT) and thrombinreceptor deficient (TR KO) mice (n = 4).

18 K. Tudpor et al. / Bone 72 (2015) 14–22

Author's personal copy

Increased BMD in TR-deficient mice was not due to intestinal Ca2+

hyperabsorption

To investigate the origin of the increment of BMD in TR KO mice,fecal Ca2+ content was measured in both mouse strains, but no sig-nificant difference was observed between the groups (105 ± 25 vs145 ± 56 μmol/g/day, Fig. 7A). Furthermore, 10-nM thrombin had noeffect on transepithelial Ca2+ transport in Caco-2 cells that endogenous-ly express TR, TRPV6, and CaBP-D9K [27,28] (Fig. 7B). Moreover, a 2-hincubationwith 10-nM thrombin did not affect PD and TEER, suggestingthat thrombin had also no effect on the paracellular transport (Fig. 7C).

Discussion

Serine proteases and their receptors have been characterized asimportant players in bone metabolism as exemplified by tissue- andurokinase-type plasminogen activators and plasminogen/plasmin [29–31]. The present study provides evidence that the thrombin/TRsystem is required for a normal bone resorption processmainly by stim-ulating RANKL secretion from osteoblasts, thereby indirectly triggeringosteoclastogenesis. Furthermore, the system seems to be crucial forproper degradation of both inorganic and organic bone compartments.

These conclusions are drawn from the following observations in TRKO mice and osteoblast cells. First, TR KO mice manifested incrementsin both cortical and trabecular bone mass, increased BMD, and changesin bone microarchitecture. Second, urine analyses showed a reductionof DPD crosslink concentration, an organic-phase bone resorptionmarker. Third, TR deficiency decreased the serum RANKL/OPG ratio.Fourth, in vitro experiments revealed that thrombin upregulated theRANKL/OPG protein ratio in osteoblast-culturing medium via the TRand p42/44-ERK-mediated pathway. Finally, thrombin had no direct ef-fect on intestinal Ca2+ transport.

The physical appearance of the TR KO mice was normal, but theirbody weights were higher compared to WT in spite of normal foodandwater intakes. These observations can be explained by the increasesin bone size and mass as well as possibly adipocyte hypertrophy sincethrombin was reported to control the size of adipocytes [10]. Indeed,femur length was increased in TR KO mice, as were various parametersof cortical and trabecular bone mass. In addition, a transformation ofrod-like into plate-like microarchitecture indicated that TR-deficientmice gained bone [32]. As a consequence of themicrostructural changesin cortical and trabecular bones, a higher J ensued. This finding confirmsthe osteoporotic phenotype since the combined measurement of BMDand J has been used as important indicators for bone strength and stiff-ness [33]. It is unlikely that the observed increase in bone mass andstrength in TR KOmice was due to the enhanced bone-forming activityof osteoblasts, since bone histomorphometry showed that osteoblastnumber and surface (Ob.N/BS and Ob.S/BS, respectively) were not al-tered. Moreover, osteoclast number (Oc.N/BS) was also not changed.These findings are in linewith the study by Song and colleagueswho re-ported that Ob.S/BS and Oc.N/BSwere not different betweenWT and TRKO mice [11]. Therefore, the observed increase in bone mass likely re-sulted from a decrease in the cytokine-producing activity of osteoblastsas signified by decreases in the serum RANKL/OPG ratio and urinaryDPD concentration in the present study. However, how the change inRANKL did not affect osteoclast histology is complex and needs to beclarified in further studies.

RANKL and OPG can be considered as central players harmonizingbone formation and resorption [34]. RANKL-deficient mice completely

Fig. 5. SerumRANKL and OPG levels inwild-type (WT) and TR-deficient (KO)mice (A andB, respectively). C, Relative RANKL/OPG ratio was calculated from A and B. n = 4 pergroup, *P b 0.05 in comparison to WT.

Fig. 6. Stimulation of RANKL/OPG synthesis ratio by thrombin in osteoblasts depends on TR and p42/44-ERK signaling pathway. RANKL/OPG ratiowas obtained frommedia of MC3T3 cellstreated with thrombin with indicated concentrations for 16 h (A). Cells treated with 10 nM thrombin were analyzed for RANKL and OPG mRNA expression (B–D). E, RANKL/OPG ratioswere calculated upon treatments of MC3T3 cells with thrombin with different inhibitors. −, non-thrombin-treated. F, Immunoblotting analysis of phosphorylated and total p42/44-ERK expression (upper andmiddle left panels, respectively). β-actinwas used as internal control (lower left panel). Relative p42/44-ERK expressionwas calculated from the blot densities(right panel). n = 6 per group, *P b 0.05 in comparison to non-thrombin-treated condition.

19K. Tudpor et al. / Bone 72 (2015) 14–22

Author's personal copy

lack osteoclasts and have severe osteopetrosis whereas OPG deficiencyessentially causes the opposite traits i.e., osteoporosis with severe corti-cal and trabecular bone thinning and porosity [35,36]. Various factorscan positively or negatively affect RANKL and OPG synthesis. For exam-ple, 1,25(OH)2D3, PTH, interleukin-11 (IL-11), and prostaglandin E2(PGE2) stimulate RANKL synthesis whereas estrogen and transforminggrowth factor-β (TGF-β) are inhibitors of RANKL production [5,35,37].RANKL and proteases are bone regulators that work in close inter-relationship. For instance, a cysteine protease cathepsin K localized inlysosome in the ruffled border of osteoclasts has been identified as amajor osteoclastic protease and plays a principal role in degradingthe ECM [5]. RANKL, upon its binding to RANK, has been reported tostimulate cathepsin K mRNA and protein expression [38]. Besides,ADAM (a disintegrin and metalloproteinase domain), a member ofmetalloproteinases, also indirectly facilitates RANKL-induced osteoclas-togenesis by shedding a RANKL enhancer TNF-α [5]. Importantly, MMPand ADAM family members have been shown to cleave RANKL in vitro,increasing the RANKL concentration in the culture medium [23,39]. Inthe present study; however, the possibility that thrombin cleavesRANKL was excluded by the observation that an increase in theRANKL/OPG ratio after thrombin treatment was blocked by TR antago-nist. Altogether, the thrombin/TR signaling cascade seems essential forRANKL-dependent osteoclastogenesis.

Osteoclast cells are derived from hematopoietic stem cells (HSCs)that reside in the endosteal region i.e., the active bone modeling andremodeling interface region between bone and bone marrow [40].Recently, Mansour and colleagues demonstrated that the proper osteo-clast function is imperative forHSC niches and osteoblast differentiation[40]. Even though TR is not expressed in osteoclasts, it was detected inHSC and found to be important for cell repopulation [12,41]. In addition,Aronovich and co-workers reported that TR ablation reduced bonemassin adult female, but not newborn mice [41]. Taken together with ourfinding of a high bone mass phenotype in TR-KO adult male mice, therole of TR may be dependent on gender and developmental stage. Ofnote, a functional androgen response element is located in the promoterregion of TR, which is ubiquitously expressed in bone-derived cancercells [42,43]. In fact, our preliminary study revealed that TR deficiencyhad no effect on both trabecular and cortical μCT parameters in femalemice (data not shown). The roles of sex hormones on TR-associatedbone remodeling still need to be elucidated in future studies. Interest-ingly, thrombin/TR was found to be involved with osteoblast differenti-ation, which is comprised of three overlapping phases: collagenousphase (synthesis of collagen type 1, a major component of ECM), non-collagenous phase (osteopontin and ALP expression), and mineralizingphase (osteocalcin expression) [14,44]. In vitro experiment showed thatthrombin inhibited ALP activity in primary osteoblast-like cells throughTR, indicating the negative effect of thrombin on the non-collagenousphase [14]. Accordingly, the results in the present study suggestedthat lower concentration of urinary DPD crosslinks and increased BMDmight be due to the enhanced collagenous and mineralizing phasesin the absence of TR. In other words, it could be extrapolated that

thrombin/TR negatively influences all steps of bone formation by osteo-blasts. On the other hand, RANKL synthesis and OPG synthesis were in-dependent of the osteoblast differentiation stages, but rather relied onthe presence of an intact TR/p42/44-ERK signaling system [44].

The serine proteases can exert their physiological effects via varioussignaling pathways including p42/44-ERK, p38-MAPK, PKC, and Rac1[15,16,31,45,46]. Upon stimulation by thrombin; however, TR conveysits signal via either PKC through Gq11/Gi2 or p42/44-ERK through G12/G13 [16,47–50]. Based on our finding that thrombin stimulated RANKLexpression through p42/44 ERK signaling, it was thus presumed inthe present study that TR mediated its effect via G12/G13. An activationof p42/44-ERK signaling by other serine proteases demonstrated dif-ferential effects on osteoblast activity. Urokinase, through urokinasereceptor/p42/44-ERK, inhibited bone-forming osteoblast activitywhereas plasmin induced OPG expression and preserved normal BMD[30,31]. Of note, thrombin has been reported to stimulate IL-6 synthesisin osteoblasts through the p42/44-ERK pathway [15]. IL-6 was a potentstimulator of RANKL production and surface membrane expression [29,51]. Therefore, in the present study, thrombin/TR might also indirectlyraise the RANKL/OPG ratio in osteoblasts through IL-6.

Many proteases including thrombin are synthesized as inactive pre-cursors. Prothrombin mainly produced in parenchymal cells of the liveris converted to active thrombin by coagulation factor Xa (FXa) [52].Recently, Karlstrom and colleagues reported that prothrombin wasalso expressed in osteoclasts as well as in the metaphyseal trabecularbone and to a lesser extent, in cortical bone [7]. The highest expressionof prothrombin in the subepiphysis, the bone-growing area, might ex-plain our observation of longer femurs in TR KO mice in such a waythat TR served as bone growth suppressor [7]. This research group alsodemonstrated interplay between RANKL and FXa in osteoclasts that, ina co-culture of the monocyte/macrophage cells, RANKL induced theprecursor differentiation into mature osteoclast phenotypes and theprothrombinase activity of FXa [53]. This finding suggested a positivefeedback of thrombin synthesis from osteoclasts, which in turn exertedits stimulatory effect on osteoblasts as observed in the present study. Inaddition to the effect on osteoclast maturation presented in the currentstudy, thrombin has been found to recruit non-adherent osteoclast pre-cursors and inhibit osteoclast differentiation, but these effects were TR-independent [10,11]. Moreover, thrombin could not directly inhibit os-teoclast differentiation unless bonemarrow cells were primedwith PTHand 1,25(OH)2D3 [10]. Therefore, we postulate here that thrombin acti-vates osteoclastogenesis by majorly inducing RANKL synthesis andinhibiting OPG production from osteoblasts in a TR-dependent manner.

Apart frombone, thrombin and TR are also expressed in the intestine[27,54]. Therefore, an excessive BMD in TR-deficient bone might betaken up from Ca2+ hyperabsorption in the gastrointestinal tract.However, intestinal Ca2+ absorption signified by fecal Ca2+ excretionappeared normal in TR KO mice. Moreover, in vitro data indicates thatthrombin had no effect on Ca2+ transport as well as electrical parame-ters across Caco-2 cells, suggesting that both transcellular Ca2+ absorp-tion and paracellular Ca2+ absorption were unaffected. These data are

Fig. 7. Roles of thrombin and TR on intestinal Ca2+ transport. Fecal Ca2+ excretion represents a whole non-absorbed Ca2+ in gastrointestinal tract (A). Time-response in vitrotransepithelial Ca2+ transport in Caco-2 cells endogenously expressing TR, TRPV6, and CaBP-D9K upon treatment of 10 nM thrombin (B). Epithelial electrical parameters of Caco-2 cellswere determined after 2-h incubation with and without 10 nM thrombin (C), n = 4 per group. Transepithelial potential difference (PD) and transepithelial electrical resistance (TEER)were directly measured.

20 K. Tudpor et al. / Bone 72 (2015) 14–22

Author's personal copy

in linewith a previousfinding that an activation of PAR-2, but not PAR-1(TR) changed intestinal paracellular permeability and ion transport[55]. Further study is needed to define the source of Ca2+ depositedin TR-deficient bone.

In conclusion, the thrombin/TR system regulates bone remodelingby stimulating RANKL synthesis in osteoblasts via a p42/44-ERK sig-nalingpathway. Consequently, RANKL/RANKbinding induces osteoclastmaturation and concomitant bone resorption. In TR KOmice, the lack ofTR causes a decrease in RANKL and partially increases OPG secretionfrom osteoblasts, thereby retarding proper osteoclastogenesis, whichresults in a high bone mass phenotype (see Fig. 8). We suggest thatTR is a kind of ‘gate keeper’ for RANKL production during osteoblastdifferentiation.

Acknowledgments

This work was supported by the Netherlands Organisation forScientific Research (grant number NWO-ALW 819.02.012, ZonMw9120.8026). J.G.J.H. is supported by a NWO VICI grant (016.130.668).We thank Titia E. Woudenberg-Vrenken, Femke van Zeeland, andMarijke Schreuders-Koedam for their excellent technical assistance.

Authors' roles: KT and BCJE performed the research, analyzed thedata, and wrote the paper. PJ performed the research. JJTHR reviewedthe manuscript. JPTML, RJMB, and JGJH wrote and reviewed themanuscript.

References

[1] Li B, Ling Chau JF, Wang X, Leong WF. Bisphosphonates, specific inhibitors of osteo-clast function and a class of drugs for osteoporosis therapy. J Cell Biochem 2011;112:1229–42.

[2] Gowen M, Lazner F, Dodds R, Kapadia R, Feild J, Tavaria M, et al. Cathepsin K knock-outmice develop osteopetrosis due to a deficit inmatrix degradation but not demin-eralization. J Bone Miner Res 1999;14:1654–63.

[3] Baron R, Hesse E. Update on bone anabolics in osteoporosis treatment: rationale,current status, and perspectives. J Clin Endocrinol Metabol 2012;97:311–25.

[4] Boyce BF, Xing L. The RANKL/RANK/OPG pathway. Curr Osteoporos Rep 2007;5:98–104.

[5] Georges S, Ruiz Velasco C, Trichet V, Fortun Y, Heymann D, Padrines M. Proteasesand bone remodelling. Cytokine Growth Factor Rev 2009;20:29–41.

[6] Pagel CN, Sivagurunathan S, Loh LH, Tudor EM, Pike RN, Mackie EJ. Functional re-sponses of bone cells to thrombin. Biol Chem 2006;387:1037–41.

[7] Karlstrom E, Norgard M, Hultenby K, Somogyi-Ganss E, Sugars R, Andersson G, et al.Localization and expression of prothrombin in rodent osteoclasts and long bones.Calcif Tissue Int 2011;88:179–88.

[8] Sun WY, Witte DP, Degen JL, Colbert MC, Burkart MC, Holmback K, et al. Prothrom-bin deficiency results in embryonic and neonatal lethality inmice. Proc Natl Acad SciU S A 1998;95:7597–602.

[9] Senger DR, Perruzzi CA, Papadopoulos-Sergiou A, Van de Water L. Adhesive proper-ties of osteopontin: regulation by a naturally occurring thrombin-cleavage in closeproximity to the GRGDS cell-binding domain. Mol Biol Cell 1994;5:565–74.

[10] Sivagurunathan S, Pagel CN, Loh LH,Wijeyewickrema LC, Pike RN, Mackie EJ. Throm-bin inhibits osteoclast differentiation through a non-proteolytic mechanism. J MolEndocrinol 2013;50:347–59.

[11] Song SJ, Pagel CN, Campbell TM, Pike RN, Mackie EJ. The role of protease-activatedreceptor-1 in bone healing. Am J Pathol 2005;166:857–68.

[12] Karlstrom E, Norgard M, Hultenby K, Somogyi-Ganss E, Sugars R, Andersson G,Wendel M. Localization and expression of prothrombin in rodent osteoclasts andlong bones. Calcif Tissue Int 2011;88:179–88.

[13] Bluteau G, Pilet P, Bourges X, BilbanM, Spaethe R, Daculsi G, et al. The modulation ofgene expression in osteoblasts by thrombin coated on biphasic calcium phosphateceramic. Biomaterials 2006;27:2934–43.

[14] Abraham LA, MacKie EJ. Modulation of osteoblast-like cell behavior by activation ofprotease-activated receptor-1. J Bone Miner Res 1999;14:1320–9.

[15] Tokuda H, Kato K, Natsume H, Kondo A, Kuroyanagi G, Matsushima-Nishiwaki R,et al. Involvement of AMP-activated protein kinase in thrombin-stimulated interleu-kin 6 synthesis in osteoblasts. J Mol Endocrinol 2012;49:47–55.

[16] Kaufmann R, Schafberg H, Rudroff C, Nowak G. Thrombin receptor activation resultsin calcium signaling and protein kinase C-dependent stimulation of DNA synthesisin HEp-2g laryngeal carcinoma cells. Cancer 1997;80:2068–74.

[17] Gustafson GT, Lerner U. Thrombin, a stimulator of bone resorption. Biosci Rep 1983;3:255–61.

[18] Hoffmann O, Klaushofer K, Koller K, Peterlik M, Mavreas T, Stern P. Indomethacin in-hibits thrombin-, but not thyroxin-stimulated resorption of fetal rat limb bones.Prostaglandins 1986;31:601–8.

[19] Renkema KY, Velic A, Dijkman HB, Verkaart S, van der Kemp AW, NowikM, et al. Thecalcium-sensing receptor promotes urinary acidification to prevent nephrolithiasis. JAm Soc Nephrol 2009;20:1705–13.

[20] Hoenderop JG, Muller D, Van Der Kemp AW, Hartog A, Suzuki M, Ishibashi K, et al.Calcitriol controls the epithelial calcium channel in kidney. J Am Soc Nephrol2001;12:1342–9.

[21] Bouxsein ML, Boyd SK, Christiansen BA, Guldberg RE, Jepsen KJ, Muller R. Guidelinesfor assessment of bone microstructure in rodents using micro-computed tomogra-phy. J Bone Miner Res 2010;25:1468–86.

[22] van der Eerden BC, Hoenderop JG, de Vries TJ, Schoenmaker T, Buurman CJ,Uitterlinden AG, et al. The epithelial Ca2+ channel TRPV5 is essential for proper os-teoclastic bone resorption. Proc Natl Acad Sci U S A 2005;102:17507–12.

[23] Gruber HE. Adaptations of Goldner's Masson trichrome stain for the study ofundecalcified plastic embedded bone. Biotech Histochem 1992;67:30–4.

[24] de Groot T, Lee K, Langeslag M, Xi Q, Jalink K, Bindels RJ, et al. Parathyroid hormoneactivates TRPV5 via PKA-dependent phosphorylation. J Am Soc Nephrol 2009;20:1693–704.

[25] Van Baal J, Yu A, Hartog A, Fransen JA, Willems PH, Lytton J, et al. Localization andregulation by vitamin D of calcium transport proteins in rabbit cortical collectingsystem. Am J Physiol 1996;271:F985–93.

[26] Charoenphandhu N, Tudpor K, Pulsook N, Krishnamra N. Chronic metabolicacidosis stimulated transcellular and solvent drag-induced calcium transportin the duodenum of female rats. Am J Physiol Gastrointest Liver Physiol 2006;291:G446–55.

[27] Vetrano S, Ploplis VA, Sala E, Sandoval-Cooper M, Donahue DL, Correale C, et al. Un-expected role of anticoagulant protein C in controlling epithelial barrier integrityand intestinal inflammation. Proc Natl Acad Sci U S A 2011;108:19830–5.

[28] O'Brien CA, Gubrij I, Lin SC, Saylors RL, Manolagas SC. STAT3 activation in stromal/osteoblastic cells is required for induction of the receptor activator of NF-kappaBligand and stimulation of osteoclastogenesis by gp130-utilizing cytokines orinterleukin-1 but not 1,25-dihydroxyvitamin D3 or parathyroid hormone. J BiolChem 1999;274:19301–8.

Fig. 8. Cartoon postulating themolecular mechanism of thrombin/TR action on bonemetabolism. In wild-type (WT) mice, TR is expressed on osteoblasts, of which precursors secrete re-ceptor activator of nuclear factor-κB ligand (RANKL), an osteoclast-activating factor (green line); and by osteoblasts themselves osteoprotegerin (OPG), an anti-maturation factor of os-teoclasts (red line). In WT mice, RANKL/OPG secretion ratio was in balance. In TR-deficient mice, the lack of TR causes a reduction in RANKL and an upregulation of OPG secretion fromosteoblasts and impairs osteoclastogenesis and bone resorptive capacity. This eventually results in an increase in bone mass. (For interpretation of the references to color in this figurelegend, the reader is referred to the web version of this article.)

21K. Tudpor et al. / Bone 72 (2015) 14–22

Author's personal copy

[29] Daci E, Everts V, Torrekens S, Van Herck E, Tigchelaar-GutterrW, Bouillon R, et al. In-creased bone formation in mice lacking plasminogen activators. J Bone Miner Res2003;18:1167–76.

[30] Furlan F, Galbiati C, Jorgensen NR, Jensen JE, Mrak E, Rubinacci A, et al. Urokinaseplasminogen activator receptor affects bone homeostasis by regulating osteoblastand osteoclast function. J Bone Miner Res 2007;22:1387–96.

[31] Kanno Y, Ishisaki A, Kawashita E, Chosa N, Nakajima K, Nishihara T, et al.Plasminogen/plasmin modulates bone metabolism by regulating the osteoblastand osteoclast function. J Biol Chem 2011;286:8952–60.

[32] van Ruijven LJ, Giesen EB, Mulder L, Farella M, van Eijden TM. The effect of bone losson rod-like and plate-like trabeculae in the cancellous bone of the mandibular con-dyle. Bone 2005;36:1078–85.

[33] Popp AW, Windolf M, Senn C, Tami A, Richards RG, Brianza S, et al. Prediction ofbone strength at the distal tibia by HR-pQCT and DXA. Bone 2012;50:296–300.

[34] Boyce BF, Xing L. Biology of RANK, RANKL, and osteoprotegerin. Arthritis Res Ther2007;9(Suppl. 1):S1.

[35] Kong YY, Yoshida H, Sarosi I, Tan HL, Timms E, Capparelli C, et al. OPGL is a key reg-ulator of osteoclastogenesis, lymphocyte development and lymph-node organogen-esis. Nature 1999;397:315–23.

[36] Bucay N, Sarosi I, Dunstan CR, Morony S, Tarpley J, Capparelli C, et al.Osteoprotegerin-deficient mice develop early onset osteoporosis and arterial calcifi-cation. Genes Dev 1998;12:1260–8.

[37] Shevde NK, Plum LA, Clagett-Dame M, Yamamoto H, Pike JW, DeLuca HF. A potentanalog of 1alpha,25-dihydroxyvitamin D3 selectively induces bone formation. ProcNatl Acad Sci U S A 2002;99:13487–91.

[38] Shalhoub V, Faust J, Boyle WJ, Dunstan CR, Kelley M, Kaufman S, et al. Osteoproteg-erin and osteoprotegerin ligand effects on osteoclast formation from human periph-eral blood mononuclear cell precursors. J Cell Biochem 1999;72:251–61.

[39] Lynch CC, Hikosaka A, Acuff HB, Martin MD, Kawai N, Singh RK, et al. MMP-7 pro-motes prostate cancer-induced osteolysis via the solubilization of RANKL. CancerCell 2005;7:485–96.

[40] Mansour A, Abou-Ezzi G, Sitnicka E, Jacobsen SE, Wakkach A, Blin-Wakkach C. Oste-oclasts promote the formation of hematopoietic stem cell niches in the bone mar-row. J Exp Med 2012;209:537–49.

[41] Aronovich A, Nur Y, Shezen E, Rosen C, Zlotnikov Klionsky Y, Milman I, et al. A novelrole for factor VIII and thrombin/PAR1 in regulating hematopoiesis and its interplaywith the bone structure. Blood 2013;122:2562–71.

[42] Salah Z, Maoz M, Cohen I, Pizov G, Pode D, Runge MS, et al. Identification of a novelfunctional androgen response element within hPar1 promoter: implications to pros-tate cancer progression. FASEB J 2005;19:62–72.

[43] Chay CH, Cooper CR, Gendernalik JD, Dhanasekaran SM, Chinnaiyan AM, Rubin MA,et al. A functional thrombin receptor (PAR1) is expressed on bone-derived prostatecancer cell lines. Urology 2002;60:760–5.

[44] Thomas GP, Baker SU, Eisman JA, Gardiner EM. Changing RANKL/OPGmRNA expres-sion in differentiating murine primary osteoblasts. J Endocrinol 2001;170:451–60.

[45] Aguirre-Ghiso JA, Liu D, Mignatti A, Kovalski K, Ossowski L. Urokinase receptor andfibronectin regulate the ERK(MAPK) to p38(MAPK) activity ratios that determinecarcinoma cell proliferation or dormancy in vivo. Mol Biol Cell 2001;12:863–79.

[46] Camerer E, Barker A, Duong DN, Ganesan R, Kataoka H, Cornelissen I, et al. Local pro-tease signaling contributes to neural tube closure in the mouse embryo. Dev Cell2010;18:25–38.

[47] Hu H, Yamashita S, Hua Y, Keep RF, LiuW, Xi G. Thrombin-induced neuronal protec-tion: role of the mitogen activated protein kinase/ribosomal protein S6 kinase path-way. Brain Res 2010;1361:93–101.

[48] Wysoczynski M, Liu R, Kucia M, Drukala J, Ratajczak MZ. Thrombin regulates themetastatic potential of human rhabdomyosarcoma cells: distinct role of PAR1 andPAR3 signaling. Mol Cancer Res 2010;8:677–90.

[49] Ruiz-Loredo AY, Lopez E, Lopez-Colome AM. Thrombin stimulates stress fiber as-sembly in RPE cells by PKC/CPI-17-mediated MLCP inactivation. Exp Eye Res 2012;96:13–23.

[50] Amiable N,Martel-Pelletier J, Lussier B, Kwan Tat S, Pelletier JP, Boileau C. Proteinase-activated receptor-2 gene disruption limits the effect of osteoarthritis on cartilage inmice: a novel target in joint degradation. J Rheumatol 2011;38:911–20.

[51] Yoshitake F, Itoh S, Narita H, Ishihara K, Ebisu S. Interleukin-6 directly inhibits oste-oclast differentiation by suppressing receptor activator of NF-kappaB signaling path-ways. J Biol Chem 2008;283:11535–40.

[52] Barnhart MI. Cellular site for prothrombin synthesis. Am J Physiol 1960;199:360–6.[53] Karlstrom E, Ek-Rylander B, Wendel M, Andersson G. RANKL induces components of

the extrinsic coagulation pathway in osteoclasts. Biochem Biophys Res Commun2010;394:593–9.

[54] Tudpor K, Lainez S, Kwakernaak AJ, Kovalevskaya NV, Verkaart S, van Genesen S,et al. Urinary plasmin inhibits TRPV5 in nephrotic-range proteinuria. J Am SocNephrol 2012;23:1824–34.

[55] Swystun VA, Renaux B,Moreau F,Wen S, Peplowski MA, HollenbergMD, et al. Serineproteases decrease intestinal epithelial ion permeability by activation of protein ki-nase Czeta. Am J Physiol Gastrointest Liver Physiol 2009;297:G60–70.

22 K. Tudpor et al. / Bone 72 (2015) 14–22