Implant wear induces inflammation, but not osteoclastic

Implant Wear Induces Inflammation, but NotOsteoclastic Bone Resorption, in RANK�/� Mice

Weiping Ren,1 Bin Wu,1 Xin Peng,1 Jing Hua,2 Hsiao-Nan Hao,1 Paul H. Wooley1

1Department of Orthopedic Surgery, Wayne State University School of Medicine,University Health Center 7C, 4201 St. Antoine Boulevard, Detroit, Michigan 48201

2Department of Computer Science, Wayne State University, Detroit, Michigan 48201

Received 9 September 2005; accepted 15 February 2006

Published online 15 June 2006 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jor.20190

ABSTRACT: Signaling of RANK (receptor activator of nuclear factor kappa B) through its ligandRANKL appears critical in osteolysis associated with aseptic loosening (AL). The purpose of thisstudywas to investigate the role of RANK in amurine osteolysismodel developed inRANKknockout(RANK�/�) mice. Ultra highmolecular weight polyethylene (UHMWPE) debris was introduced intoestablished air pouches on RANK�/�mice, followed by implantation of calvaria bone from syngeneiclittermates.Wild typeC57BL/6 (RANKþ/þ) mice injectedwith eitherUHMWPEor saline alonewereincluded in this study. Pouch tissues were collected 14 days after UHMWPE inoculation formolecular and histology analysis. Results showed thatUHMWPE stimulation induced strong pouchtissue inflammation in RANK�/� mice, as manifested by inflammatory cellular infiltration,pouch tissue proliferation, and increased gene expression of IL-1b, TNFa, and RANKL. However,theUHMWPE-induced inflammation inRANK�/�micewasnot associatedwith the osteoclastic boneresorption observed inRANKþ/þmice. InRANKþ/þmice subjected toUHMWPEstimulation, a largenumber of TRAPþ cells were found on the implanted bone surface, where active osteoclastic boneresorption was observed. No TRAPþ cells were found in UHMWPE-containing pouch tissuesof RANK�/� mice. Consistent with the lack of osteoclastic activity shown by TRAP staining, nosignificant UHMWPE particle-induced bone resorption was found in RANK�/� mice. A wellpreserved bone collagen content (Van Gieson staining) and normal plateau surface contour[microcomputed tomography (mCT)] of implanted bone was observed in RANK�/� mice subjected toUHMWPE stimulation. In conclusion, this study provides the evidence that UHMWPE particlesinduce strong inflammatory responses, but not associated with osteoclastic bone resorption inRANK�/� mice. This indicates that RANK signaling is essential for UHMWPE particle-inducedosteoclastic bone resorption, but does not participate in UHMWPE particle-induced inflammatoryresponse.� 2006 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res

24:1575–1586, 2006

Keywords: RANK; RANKL; wear debris; osteolysis; osteoclastogenesis

INTRODUCTION

Aseptic loosening (AL) is characterized by thedevelopment of an inflammatory membrane at thebone–implant interface, leading to bone resorp-tion and loss of prosthesis fixation.1 Enhancedosteoclastogenesis (formation of mature osteo-clasts from their precursor stem cells) is a hall-mark of osteoporotic complications, includingthe bone loss associated with AL.2 Two importantregulators of osteoclastogenesis have been re-cently identified.3,4 Receptor activator of nuclearfactor kappa B (RANK, TNFR11B) is a novel type

I transmembrane receptor of the TNF receptorsuperfamily (TNFRSF) that was originally identi-fied using a dendritic cell cDNA library.5 RANK isubiquitously expressed in human tissues, butis particularly apparent on the cell surface ofosteoclasts and their precursors.4 Its ligand,RANKL, also known as ODF/TRANCE/OPGL, isa TNF-related cytokine that is produced bymarrowstromal cells and osteoblasts.6 The availability of asoluble form of recombinant human RANKL,comprising the extracellular domain of the protein(sRANKL), has allowed the in vitro generationof large numbers of bone resorbing osteoclastsfrom hemopoietic precursors in the absence ofstromal/osteoblast cells.7 An emerging conceptis that RANK represents a common mediatorfor inflammatory osteoclastogenesis.8–10 Li et al.8

reported that osteoclastogenesis was not observed

JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2006 1575

Correspondence to: Weiping Ren (Telephone: 313-996-7240;Fax: 313-745-4406; E-mail: [email protected])

� 2006 Orthopaedic Research Society. Published by Wiley Periodicals,Inc.

in RANK gene knockout (RANK�/�) micefollowing IL-1 treatment, illustrated by the com-plete absence of TRAPþ and cathepsin Kþ osteo-clasts in calvaria. In addition, the failure ofRANK�/� mice to mount a significant osteoclasticresponse in the presence of experimental inflam-matory arthritis further supports the concept thatRANK plays a gate-keeping role in controllingosteoclastogenesis,10 and mice lacking RANKLhave defects in osteoclastogenesis that leads tosevere osteopetrosis.10 However, there is contro-versy as to whether TNFa can stimulate myeloidosteoclastogenesis by a RANK/RANKL indepen-dent mechanism.11,12 Sabokbar et al.13 foundthat the addition of TNFa to macrophages isola-ted from human loosening periprosthetic tissuesinduced the formation of TRAPþ multinucleatedosteoclasts capable of extensive lacunar resorp-tion. TNFa-induced osteoclastogenesis cannot beblocked by the treatment with either osteoprote-gerin (OPG) or RANK:Fc IgG (which inhibitRANKL-induced osteoclastogenesis), suggestingthis process is distinct from the RANK/RANKLsignaling pathway. These data are interestingbecause one favored hypothesis concerning thepathogenesis of AL is that wear particles gener-ated from the prosthesis stimulate phagocytosis bymacrophages with the subsequent release of pro-inflammatory cytokines.14–16 Ultra high molecu-lar weight polyethylene (UHMWPE) has beenwidely used in total joint prosthesis, andUHMWPE components removed at revision sur-gery usually show evidence of wear.17,18 Using ourestablished mouse osteolysis model, we havedemonstrated that UHMWPE particles provokesignificant inflammation, osteoclastogenesis, andbone resorption.19,20 The purpose of this study wasto investigate the role of RANK in UHMWPEparticle-induced inflammatory osteolysis by theuse of RANK knockout (RANK�/�) mice.

MATERIALS AND METHODS

Ultra High Molecular WeightPolyethylene (UHMWPE) Particles

UHMWPE particles (the generous gift of Dr. JohnCuckler from University of Alabama, Birmingham, AL)were used to stimulate inflammatory reactions. Thesize and distribution of UHMWPE particles was eval-uated using a Coulter particle counter and by scann-ing electron microscopy, as described elsewhere.21

The UHMWPE particles had a mean particle diameterof 2.6 mm, with a range from <0.6 mm to 21 mm. Theparticles were washed in 70% ethanol solution to re-move endotoxin and heat sterilized, and the absence of

endotoxin was confirmed using the Limulus assay(Endosafe; Charles Rivers, Charlestown, SC). Particleswere suspended in sterile PBS and stored at 48C untiluse.

Mouse Osteolysis Model

Institutional approval was obtained for all animalprocedures. The use of the bone implanted inflammatoryair pouch as a model of debris-induced osteolysis hasbeen described previously.19 Air pouches were estab-lished by subcutaneous injection of sterile air on theback of C57BL/6 mice of either RANKþ/þ or RANK�/�

phenotype (the generous gift of Amgen Inc., Seattle,WA). Each experimental group consisted of 10 mice.Six days later, mice were anesthetized by the intra-peritoneal injection of pentobarbital (50 mg/kg; FisherScientific, Pittsburgh, PA). A 0.5-cm incision overlyingthe air pouch was made and a section of calvaria bonefrom a genetically identical donor mouse (approximately0.4� 0.25 cm) was inserted into the pouch. The pouchlayers and the skin incision were then closed using 4-0Prolene sutures. On the following day, pouches wereinjected with 0.5 ml saline containing 1% UHMWPEparticles. Pouches injected with saline alone were usedas negative controls in RANKþ/þ mice. Mice weresacrificed in a carbon dioxide chamber 2 weeks afterbone implantation. The pouch membranes containingimplanted bone were harvested for histological andmolecular analyses.

Histological Assessment

Pouch tissues were fixed in 10% formalin, decalcified in10% EDTA, and paraffin embedded. Paraffin sections(6 mm) were stained with hematoxylin & eosin (H&E)and the microscopic digital images photographed andanalyzed as described previously.19,22 Modified naphtholAS-D chloroacetate esterase staining was used to detectmonocytes/macrophages. Osteoclast formation wasdetermined by tartrate-resistant acid phosphatase(TRAP, EC3.1.3.2) staining in paraffin tissue sections(6 mm) using a commercial kit (Sigma, St. Louis, MO), asdescribed previously.23 The presence of dark purplestaining granules in the cytoplasm was considered asthe specific criterion for TRAPþ cells. Positive TRAPlocalization was quantified by pixel area count andreported as a percentage of the of total implanted bonearea in the porch tissue. Bone collagen depletion wasdetermined by Van Gieson stain and quantified byimage analysis19,24 using the ImagePro software pack-age (MediaCybernetics, Silver Spring, MD, USA).

Immunohistological Staining forTNFa, IL-1b, CD68, and RANKL

Paraffin tissue sections were deparaffinized, and rinsedsections were immersed into antigen retrieval buffer(BioGenex, San Ramon, CA) with microwave heating

1576 REN ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2006 DOI 10.1002/jor

(power 60% for 2 min, followed by power 30% for 5 min),followed by cooling to room temperature. Sectionswere blocked with 1.5% normal goat serum for 1 h,and incubated with goat anti-mouse TNFa, IL-1b, CD68,and RANKL antibodies (2 mg/ml, Santa Cruz Biotech-nology, Santa Cruz, CA) at 48C overnight. After rinsing,sections were incubated for 30 min with biotin-conju-gated secondary antibody (Vector Laboratories, Burlin-game, CA). Avidin biotin enzyme reagent was thenapplied onto sections for 30 min. Color was formed byadding 3.30-diaminobezidine tetrahydrochloride (DAB),with counterstaining using Mayer’s hematoxylin. Innegative control sections, the primary antibody wasomitted or an irrelevant antibody applied at the sameconcentration as the primary antibody. Digital images ofrepresentative fields of view were captured and anal-yzed using the ImagePro software package.

Enzyme Linked Immunosorbent Assay (ELISA)

ELISA assays were performed to examine the proteinexpression of IL-1b and TNFa in the homogenates ofpouch membrane using commercial kits (R&D, Minnea-polis, MN) following manufacturer’s instructions.

Gene Level of RANKL and Cathepsin K (CPK)

Total RNA from pouch homogenates was reversetranscribed to cDNA as described elsewhere.22 Realtime quantitative PCR was carried out according to themanufacturer’s instruction (Perkin Elmer-Applied Bio-systems, Foster City, CA). Primers used in this studyhave been described elsewhere.20 To standardize thetarget gene level with respect to variability in RNA andcDNA concentrations, the housekeeping gene GADPHwas used as an internal control. To determine therelative level of gene expression, the comparative CT

(threshold cycle) method with arithmetic formulas wasused. Subtracting the CT of the housekeeping gene fromthe CT of target gene yields the DCT in each group(control and experimental groups), which was enteredinto the equation 2�DCT and calculated for the exponen-tial amplification of PCR. The gene activity in controlgroup (PBS) was arbitrarily assigned to 1 to serve asreference. The expression of the target gene fromexperimental groups therefore represents the fold-difference expression relative to the reference gene.

Microcomputed Tomography(MicroCT) Image Acquisition

Mice were scanned 2 weeks after bone implantation(right before termination of the experiment) using anRS-9 In Vivo MicroCT Scanner (General ElectricSystems, London, Ontario). The scan was performed ata resolution of 0.093 mm (cubic voxel length) using themanufacturer’s defined settings. After data acquisition,two-dimensional projection data was reconstructed into

three-dimensional volumes, which were then reorientedin the appropriate dimensions to place the data sets inconsistent orientation for image analysis.

Statistical Analysis

Statistical analysis among groups was performed usingthe ANOVA test, with the Schafer formula for post hocmultiple comparisons, using the SPSS software package(version 7.5; SPSS Inc., Chicago, IL). Data wereexpressed as mean� standard error of the mean. Ap-value of less than 0.05 was considered as significantdifference.

RESULTS

UHMWPE Particle-Induced Tissue Inflammationin RANKþ/þ and RANK�/� Mice

UHMWPE particle induced a similar level ofpouch tissue inflammation in RANK�/� micewhen compared with RANKþ/þ mice. As seen inFigure 1A, gross pathology verified that pouchesinjected with UHMWPE particle developed apronounced inflammatory response in bothRANKþ/þ and RANK�/� mice, as compared withcontrol pouches that received saline injection.H&E staining (Fig. 1B) revealed that UHMWPEparticle stimulation significantly increased boththe membrane thickness and the number ofinfiltrating cells in pouches, as compared to salinecontrols. Quantitative image analysis of histologi-cal sections revealed that both cellular infiltration(7,878� 1,531 mm2) and membrane thickness(0.132� 0.04 mm) in RANK�/� mice were signifi-cantly increased, as compared with saline controlmice (4,789� 1,065 mm2 and 0.089� 0.03 mm,respectively; p< 0.05), but showed no differencecompared with RANKþ/þ mice. There was nodifference with respect to any UHMWPE parti-cle-induced inflammation parameters betweenRANK�/� mice and RANKþ/þ mice (Fig. 2).

We investigated whether UHMWPE particlescan induce macrophage infiltration in pouch tis-sues in RANK�/� mice. As shown in Figure 3B,nonspecific esterase (NSE) stained cells weresignificantly increased in UHMWPE stimulatedpouch tissues from both RANK�/� and RANKþ/þ

mice. These NSEþ cells were particularly concen-trated in regions of tissue with marked particleaccumulation. The local macrophage accumulationwas confirmed by immunohistochemical stainingusing anti-CD68, a specific cell surface marker foractivated macrophages (Fig. 3A). These dataindicated that the UHMWPE particle-induced

IMPLANT WEAR INDUCES INFLAMMATION IN RANK�/� MICE 1577

DOI 10.1002/jor JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2006

recruitment of macrophages was not impaired inRANK�/� mice.

UHMWPE Particle-Stimulated IL-1b and TNFaProduction in Pouch Tissues from RANK�/� Mice

UHMWPE particle-stimulated tissue inflamma-tion was invariably accompanied by local expres-sion of proinflammatory cytokines, such as IL-1band TNFa. As seen in Figure 4A, immunostainingrevealed that increased staining of both IL-1b andTNFa occurred in UHMWPE stimulated pouchesfrom RANK�/� mice, compared with saline con-trols. Staining for these cytokines in UHMWPEstimulated pouches was predominantly locatedwithin the cytoplasm of inflammatory cell aggre-gates. There was no significant difference withrespect to the staining intensity of either IL-1b orTNFa between RANK�/� and RANKþ/þ mice. Theprotein levels of IL-1b and TNFa was alsomeasured in the supernatants of pouch homoge-nates using the ELISA assay (Fig. 4B), and asimilar protein expression profile was observed.These data verify that, similar to RANKþ/þ mice, asignificant increase of proinflammatory cytokineproduction can be induced by UHMWPE particlestimulation in RANK�/� mice.

UHMWPE Particle Did Not InduceOsteoclastic Bone Resorption in RANK�/� Mice

Immunohistochemical staining was used to inves-tigate RANKL protein expression in RANK�/�

mice. As depicted in Figure 3, intense RANKLstaining was observed in UHMWPE stimulatedpouches from both RANK�/� and RANKþ/þ mice,compared with levels seen in saline controls.RANKL staining was predominantly seen at themembrane interface with implanted bone, and inthe cytoplasm of inflammatory cells that wereforming debris associated aggregates. Similarto the immunohistochemical findings, real timeRT-PCR assay (Fig. 5) showed that UHMWPEparticles significantly increased RANKL geneexpression in pouch tissues of both RANKþ/þ mice(5.2-fold) and RANK�/� mice (3.9-fold), as com-pared with saline control (p< 0.05). There is nostatistical difference of RANKL gene level betweenRANK�/� mice and RANKþ/þ mice.

Although UHMWPE particle increased RANKLgene production in RANK�/� mice, no UHMWPEparticle-stimulated osteoclastogenesis and boneresorption changes were observed in RANK�/�

mice. As shown in Figure 5B, UHMWPE particlestimulated CPK gene expression in RANKþ/þ mice

Figure 1. Representative pouch tissues histology of both macroscopic (A) andmicroscopic (B) appearance. All the main micrographs (B) are tissue sections stainedwith hematoxylin and eosin (H&E). (Original magnification, �100.) B, implanted bone;M, pouch membrane. Note the significant inflammatory cellular infiltration andproliferated pouch membranes in UHMWPE containing pouches of both RANKþ/þ andRANK�/� mice.

1578 REN ET AL.


(4.1-fold). Interestingly, CPK gene expression wassignificantly decreased in RANK�/� mice, andrepresents only 25% of total CPK level, as com-pared with saline controls, or 10% of correspondingRANKþ/þ mice (p< 0.05). Constitutive expressionof a housekeeping gene, GAPDH, was essentiallyconstant with a relative standard error of �3% forall groups in the presence or absence of UHMWPEparticle stimulation (data not shown).

Histochemical TRAP staining was used toaddress whether UHMWPE particle can induceosteoclast formation inRANK�/�mice. As shown in

Figure 6A, a discrete focus of TRAP staining wasobserved at the interface between implanted boneand pouch membrane in pouches with salineinjection alone. The bone morphology remainedintact, and no resorption lacunae were observed.Intensified TRAP staining was found in pouchesfollowing UHMWPE stimulation, and extendedinto the adjoining tissues in RANKþ/þ mice. Asdepicted in Figure 6B, a significant (p< 0.05)increase in TRAPþ cells (6.5-fold) was observed inUHMWPE-containing pouches of RANKþ/þ mice,compared to saline controls. Regions where TRAPþ

Figure 2. Quantitative histological assessment of UHMWPE-stimulated total cellcounts (A) and pouch membrane thickness (B). Minimum of three separate sections perspecimen were evaluated in a blinded fashion using the ImagePro software. Dataexpressed asmean�SD for total 30mice (10mice per group). p< 0.05; ND, no statisticaldifference.



cells localized were often pitted, suggesting activeosteoclastic bone resorption. However, no TRAPstaining was observed in any specimens fromRANK�/� mice, suggesting that osteoclastogenesisis completely blocked in the absence of a functionalRANK gene.

Van Gieson stain was performed to quantify thebone collagen contents of implanted bone in pouchtissues, and representative images of van Giesonstained sections are shown in Figure 7A.UHMWPE particle stimulation dramaticallyincreased the loss of bone collagen content at thebone surface in close contactwith the inflammatorypouch membranes, in comparison to the bonecollagen changes in sections from control (saline-stimulated) pouches. Quantitative image analysisindicated that the bone collagen content ofimplanted bone was well preserved in RANK�/�

mice (Fig. 7B, p< 0.05).

As shown in Figure 8, the mCT imaging analysisof implanted bones showed a significant alterationof plateau surface contour in pouches withUHMWPE particle stimulation in RANKþ/þ mice,as compared to control (saline-stimulated) pouches,suggesting the progressive implanted bone degra-dation. However, UHMWPE particle stimulationdid not induce any visible alteration of plateausurface contour in pouches from RANK�/� mice.

DISCUSSION

RANK�/� mice develop severe osteopetrosis char-acterized by the accumulation of newly synthe-sized bone and defective bone resorption.8,9

Hematopoesis in these mice is unaffected, withthe exception that osteoclasts are absent through-out the skeleton.8,9 Our study revealed thatUHMWPE stimulation induced strong pouch

Figure 3. Macrophage stainingand immunohistological staining ofCD68andRANKLin pouch membrane. The number of infiltrating macrophages in pouch membrane wasevaluated by both immunostaining of CD 68, a macrophage cell marker (A) andnonspecific esterase (NSE) histological stain (B). Immunohistological staining of RANKLin pouch membrane was shown in (C). B, implanted bone; M, pouch membrane. Arrowsindicate positive staining. (Original magnification, �200).

1580 REN ET AL.


tissue inflammation in RANK�/� mice, as mani-fested by inflammatory cellular infiltration, pouchtissue proliferation, and increased gene expressionof IL-1b, TNFa, and RANKL. However, theUHMWPE-induced inflammation in RANK�/�

mice was not associated with the osteoclastic bone

resorption observed in RANKþ/þ mice, as evi-denced by the lack of TRAPþ cells, remarkablelower CPK gene production, and well preservedbone morphology and bone collagen contents ofimplanted bone in pouch tissues. Our data indi-cates that RANK signaling is essential for

Figure 4. Expression of IL-1b and TNFa in pouch membranes. (A) Immunohisto-chemical staining of IL-1b and TNFa in pouch membranes. B, implanted bone; M, pouchmembrane. Arrowheads indicate positive staining. (Original magnification, �200.) (B)The protein level of IL-1b and TNFa in the supernatants of the pouch membranehomogenates was measured by ELISA. Data expressed as mean�SD for total 30 mice(10 mice per group). *p< 0.05.



UHMWPE particle-stimulated inflammatory os-teoclastogenesis, and may play a critical role in theAL development.

In 1999,Dougal et al.9 generatedRANK�/�mice,and demonstrated that RANK expression was notrequired for the commitment, differentiation, andfunctional maturation of macrophages and dendri-tic cells from their myeloid precursors. Thesemacrophages are functionally intact and normallydistributed in the tissues of RANK�/� animals.These findings can explain why UHMWPE parti-cles can induce similar tissue inflammation andmacrophage infiltration responses in pouch tissuesof RANK�/� mice, when compared with RANKþ/þ

mice.RANK is the sole osteoclast receptor for RANKL

in vivo.8 Our previous in vivo study19,20 demon-strated that UHMWPE particle stimulation sig-nificantly increased RANKL production, and apositive association was found between RANKLgene expression and pouch tissue inflammation, aswell as the gene level of IL-1b and TNFa (whichtend to be co-produced at sites of local inflamma-tion). Abnormally higher levels of RANKL (mRNAand protein) were predominantly associated withcells containing wear particles, and may signifi-cantly contribute to the UHMWPE particle-

induced osteoclastogenesis.25 The data from thisstudy illustrate that UHMWPE particles inducesimilar pouch tissue inflammation, and increasedgene expression of RANKL, TNFa, and IL-1b inRANK�/� mice. CPK is a protease known to be acellular marker for mature osteoclasts.26 We foundthat CPK gene expression was significantlydecreased in RANK�/� mice, and represents only25%of totalCPK level observed in saline controls.Arecent study27 reported that RANKL stimulatesCPK mRNA expression in a dose- and time-dependent fashion, indicating that activation ofRANK signaling is critical in the regulation of CPKgene expression. Most importantly, the UHMWPEparticle-induced osteoclastic bone resorptiondevel-oped in RANKþ/þ mice, was not observed inRANK�/�mice. These findings support the conceptthat RANK is the sole receptor of RANKL, and isessential for the development of UHMWPE parti-cle-induced osteoclastic bone resorption. Our find-ings are in agreement with a recent report bySchwarz et al.,28 where treatment with antagonistRANK:Fc IgG successfully prevented titanium (Ti)debris-induced inflammatory osteolysis in a mousecalvaria model, yielding results that were statisti-cally equivalent to data obtained with Ti-treatedRANK�/� mice.29

Figure 5. Real time RT-PCR analysis of mRNA gene copies of RANKL and cathepsinK (CPK) in mouse pouch tissues. Pouch membranes were harvested 14 days after boneimplantation. Total RNA prepared from pouch membranes were reverse transcribed tocDNA for RT-PCR analysis. Values are mean�SD for total 30 mice (10 mice per group).*p< 0.05.

1582 REN ET AL.


A major controversy has arisen due to disagree-ment in findings regarding the role of TNFa in theinduction of osteoclastogenesis in the absence ofRANK signaling. Azuma et al.11 were the first toreport that TNFa can directly induce murineosteoclast precursors to differentiate to functionalosteoclasts in the presence of M-CSF. Similar datawere provided by Kobayashi et al.12 In both thesestudies, bone marrow cells (containing osteoclastprecursors and bone marrow stromal cells) werecultured in the presence of M-CSF to expand thepopulation of osteoclast progenitors prior to purify-ing the cells for use in experiments with TNFa.Interpretation of their results was questioned,

because previous studies have shown the impor-tance of RANKL/RANK signaling for osteocl-astogenesis in RANK�/� mice.6,8 Lam et al.30

demonstrated thatwhen pure populations ofmono-nuclear phagocytes were isolated directly frommurine bonemarrowby immunopurification, thesecells failed to differentiate into osteoclasts in thepresence of M-CSF and TNFa even at super-physiological concentrations. These data suggestedthat TNFa can only induce osteoclastogenesis inprecursor cells previously exposed toRANKL.Theyproceeded to demonstrate that low concentrationsof TNFa (500 pg/ml)were able to synergizewith lowconcentrations ofRANKL(1ng/ml) in the induction

Figure 6. TRAPþ cells are absent in RANK�/� mice. (A) The representative TRAPstaining in paraffin tissue sections (original magnification, �200). TRAP was staineddark red and is indicated by arrows. B, implanted bone; M, pouchmembrane. (B) TRAPþ

cells were quantified by Image Analysis software as described inMaterials andMethods.The value represents TRAPþ cells location quantified as percentage of total implantedbone area. Values are mean�SD for total 30 mice (10 mice per group). *p< 0.05.



of osteoclastogenesis. Since this concentration ofRANKL was 10-fold below the concentration ofRANKL previously shown to induce osteoclasto-genesis in osteoclast precursors, they concludedthat TNFawas able to act directly on the osteoclastprecursors in the presence of basal, nonstimulatedlevels of RANKL to induce osteoclastogenesis.Evidence for this was also provided by Zhanget al.31 who demonstrated that TNFa stimulationenhanced osteoclastogenesis by RANKL pre-treat-ed murine osteoclast precursors. This effect was

abrogated when they utilized cells from TNFR-1deficient mice. Overall, the exact molecularmechanism of TNFa signaling in osteoclastogen-esis is unclear. Our data suggests that the roles ofTNFa and RANKL in osteoclastogenesis exhi-bit a considerable overlap in the signaling path-ways.31 Data presented here demonstrates thatUHMWPE particle-induced osteoclastogenesis iscontrolled by RANKL/RANK signaling. At thelevel of the osteoclast, the diverse humoral andcytokine signaling regulating bone remodeling

Figure 7. Collagen loss of implanted bones was illustrated by vanGieson staining. (A)Van Gieson stain was performed to evaluate bone collagen content (dark red coloration).Diminished coloration is indicated by arrow (original magnification,�200). (B) Collagencontent of implanted bone was quantified by Image Analysis software as described inMaterials andMethods.Thevalue represents percentage of bone collagen loss.Valuesaremean�SD for total 30 mice (10 mice per group). *p< 0.05.

1584 REN ET AL.


ultimately regulates the activation of RANK onthe osteoclast/osteoclast precursor cell surface.The precise signal transduction pathways thatemanate from this receptor should define the keyintracellular processes that control osteoclasto-genesis.

In summary, our study provides solid evid-ence that RANK/RANKL signaling is essential forUHMWPE particle-induced inflammatory osteo-clastogenesis, and may play a critical role in theAL development. These data provide a biologicalrationale for the RANK-targeted treatmentstrategy; especially at the early stage of weardebris-induced inflammatory responses.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the support ofAmgen Company in providing RANK�/� mice for thisstudy, and the financial support from the Department ofOrthopedic Surgery Fund for Medical Research andEducation.

REFERENCES

1. Goodman SB, Knoblich G, O’Connor M, et al. 1996.Heterogeneity in cellular and cytokine profiles frommultiple samples of tissue surrounding revised hip pros-theses. J Biomed Mater Res 31:421–428.

2. Schwarz EM, Lu AP, Goater JJ, et al. 2000. Tumor necrosisfactor-alpha/nuclear transcription factor-kappaB signal-ing in periprosthetic osteolysis. J Orthop Res 18:472–480.

3. Khosla S. 2001. Minireview: the OPG/RANKL/RANKsystem. Endocrinology 142:5050–5055.

4. Nakagawa N, Kinosaki M, Yamaguchi K, et al. 1998.RANK is the essential signaling receptor for osteoclast

differentiation factor in osteoclastogenesis. Biochem Bio-phys Res Commun 253:395–400.

5. Anderson DM,Maraskovsky E, BillingsleyWL, et al. 1997.A homologue of the TNF receptor and its ligand enhanceT-cell growth and dendritic-cell function. Nature 390:175–179.

6. Roux S, Orcel P. 2000. Bone loss: factors that regulateosteoclast differentiation: an update. Arthritis Res 2:451–456.

7. Lacey DL, Timms E, Tan HL, et al. 1998. Osteoprotegerinligand is a cytokine that regulates osteoclast differentia-tion and activation. Cell 93:165–176.

8. Li J, Sarosi I, Yan XQ, et al. 2000. RANK is theintrinsic hematopoietic cell surface receptor that controlsosteoclastogenesis and regulation of bone mass andcalcium metabolism. Proc Natl Acad Sci USA 97:1566–1571.

9. Dougall WC, GlaccumM, Charrier K, et al. 1999. RANK isessential for osteoclast and lymph node development.Genes Dev 13:2412–2424.

10. Kong YY, Feige U, Sarosi I, et al. 1999. Activated T cellsregulate bone loss and joint destruction in adjuvantarthritis through osteoprotegerin ligand. Nature402:304–309.

11. Azuma Y, Kaji K, Katogi R, et al. 2000. Tumor necrosisfactor-alpha induces differentiation of and bone resorptionby osteoclasts. J Biol Chem 275:4858–4864.

12. Kobayashi K, Takahashi N, Jimi E, et al. 2000. Tumornecrosis factor alpha stimulates osteoclast differentiationby a mechanism independent of the ODF/RANKL-RANKinteraction. J Exp Med 191:275–286.

13. Sabokbar A, Kudo O, Athanasou NA. 2003. Two distinctcellular mechanisms of osteoclast formation and boneresorption in periprosthetic osteolysis. J Orthop Res21:73–80.

14. Gallo J, Kaminek P, Ticha V, et al. 2002. Particle disease. Acomprehensive theory of periprosthetic osteolysis: areview. Biomed Pap Med Fac Univ Palacky OlomoucCzech Repub 146:21–28.

15. Stea S, Visentin M, Granchi D, et al. 2000. Cytokines andosteolysis around total hip prostheses. Cytokine 12:1575–1579.

16. Trindade MC, Song Y, Aspenberg P, et al. 1999. Proin-flammatory mediator release in response to particle

Figure 8. Evaluation of implanted bone resorption by microCT analysis. A represen-tative segmented series of high-resolution microCT images of mouse pouch tissue withimplanted bone. Localized bone resorptionwas noticed inRANKþ/þmicewithUHMWPEparticle injection. Total six mice were scanned (two mice per group).



challenge: studies using the bone harvest chamber. JBiomed Mater Res 48:434–439.

17. Kobayashi A, Bonfield W, Kadoya Y, et al. 1997. The sizeand shape of particulate polyethylene wear debris in totaljoint replacements. Proc Inst Mech Eng [H] 211:11–15.

18. Maloney WJ, Smith RL, Schmalzried TP, et al. 1995.Isolation and characterization of wear particles generatedin patients who have had failure of a hip arthroplastywithout cement. J Bone Joint Surg [Am] 77:1301–1310.

19. Ren WP, Yang S, Wooley PH. 2004. A novel murine modelof orthopaedic wear debris-associated osteolysis. Scand JRheumatol 33:1–10.

20. Ren WP, Yang SY, Fang HW, et al. 2003. Distinct geneexpression of receptor activator of nuclear factor-kB andrank ligand in the inflammatory response to variantmorphologies of UHMWPE particles. Biomaterials 24:4819–4826.

21. Wooley PH, Morren R, Andary J, et al. 2002. Inflammatoryresponses to orthopaedic biomaterials in the murine airpouch. Biomaterials 23:517–526.

22. Yang SY, Ren W, Park Y, et al. 2002. Diverse cellular andapoptotic responses to variant shapes of UHMWPEparticles in a murine model of inflammation. Biomaterials23:3535–3543.

23. Ren WP, Wu B, Mayton L, et al. 2002. Polyethylene andmethyl methacrylate particle-stimulated inflammatorytissue and macrophages up-regulate bone resorption in amurine neonatal bone resorption in a murine neonatalcalvaria in vitro organ system. J Orthop Res 20:1031–1037.

24. Liao F, Schenkel AR, Muller WA. 1999. Transgenic miceexpressing different levels of soluble platelet/endothelial

cell adhesion molecule-IgG display distinct inflammatoryphenotypes. J Immunol 163:5640–5648.

25. Crotti TN, Smith MD, Findlay DM, et al. 2004. Factorsregulating osteoclast formation in human tissues adjacentto peri-implant bone loss: expression of receptor activatorNFkappaB, RANK ligand and osteoprotegerin. Biomater-ials 25:565–573.

26. Saftig P, Hunziker E, Everts V, et al. 2000. Functions ofcathepsin K in bone resorption. Lessons from cathepsin Kdeficient mice. Adv Exp Med Biol 477:293–303.

27. Pang M, Martinez AF, Jacobs J, et al. 2005. RANK ligandand interferon gamma differentially regulate cathepsingene expression in pre-osteoclastic cells. Biochem BiophysRes Commun 328:756–763.

28. Childs L, Paschalis E, Shigeyama Y, et al. 2000. Long-termprotection from wear debris-induced bone resorption andamelioration of established osteolysis by RANK:Fc. Arthri-tis Rheum 43:S206.

29. Childs LM, Paschalis EP, Xing L, et al. 2002. In vivo RANKsignaling blockade using the receptor activator of NF-kappaB:Fc effectively prevents and ameliorates weardebris-induced osteolysis via osteoclast depletion withoutinhibiting osteogenesis. J Bone Miner Res 17:192–199.

30. Lam J, Takeshita S, Barker JE, et al. 2000. TNF-alphainduces osteoclastogenesis by direct stimulation of macro-phages exposed to permissive levels of RANK ligand. J ClinInvest 106:1481–1488.

31. Zhang YH, Heulsmann A, Tondravi MM, et al. 2001.Tumor necrosis factor-alpha (TNF) stimulates RANKL-induced osteoclastogenesis via coupling of TNF type 1receptor and RANK signaling pathways. J Biol Chem 276:563–568.

1586 REN ET AL.


Documents

Implant wear induces inflammation, but not osteoclastic