ARTHRITIS & RHEUMATISMVol. 52, No. 10, October 2005, pp 3192–3201DOI 10.1002/art.21343© 2005, American College of Rheumatology

Analysis of the Kinetics of Osteoclastogenesis in Arthritic Rats

Georg Schett,1 Marina Stolina,2 Brad Bolon,3 Scot Middleton,2 Matt Adlam,2 Heather Brown,2

Li Zhu,2 Ulrich Feige,2 and Debra J. Zack2

Objective. To analyze the kinetics of osteoclasto-genesis in 2 models of chronic immune-mediated arthri-tis and 1 model of acute arthritis.

Methods. Adjuvant-induced arthritis (AIA) andcollagen-induced arthritis (CIA) in Lewis rats were usedas models of chronic arthritis. Acute arthritis wasinduced in Lewis rats by injecting carrageenan into thehind paw. Osteoclasts were identified by cathepsin Kimmunohistochemistry at various time points after theonset of arthritis. The location, size, and nucleation ofosteoclasts were also analyzed.

Results. In both AIA and CIA, multinucleated andcathepsin K–positive osteoclasts first were observed onthe day of disease onset. Initially, osteoclasts werelocalized at the periosteum next to the synovial mem-brane and in subchondral bone channels. The number,size, and nucleation of osteoclasts rapidly increased,leading to severe bone loss within days after diseaseonset. In addition, numerous mononucleated cathepsinK–positive osteoclast precursor cells emerged in thesynovial membrane. All osteoclasts (cathepsinK–positive, multinucleated, attached to bone) and oste-oclast precursors (cathepsin K–positive, mononucleatedor multinucleated, within synovial tissue) were alsopositive for a macrophage-specific marker. Upon induc-tion of acute arthritis with carrageenan, osteoclastsformed transiently in subchondral bone, but regressed 7days after disease onset.

Conclusion. Functional osteoclasts are generatedat the earliest stage of arthritis, and new precursors arecontinuously formed in the synovial membrane to re-plenish the osteoclast pool. These data indicate thatantiresorptive therapies may provide the most effectivebone protection, when treatment is started soon afterthe onset of arthritis.

Bone erosion is a typical sign of rheumatoidarthritis (RA) and is still considered the best surrogatemarker of joint destruction (1). The appearance of boneerosions thus indicates the ability of synovial inflamma-tion to cause articular damage, which increases thelikelihood of a poor functional outcome (2,3). Althoughskeletal damage increases with disease duration, boneerosion is no longer regarded as an exclusive feature oflate-stage disease. Bone damage usually starts early inthe course of disease, and can be identified even withcomparatively insensitive detection tools such as conven-tional radiography. Approximately one-half of RA pa-tients have visible erosions after only 6 months (4). Thus,subclinical bone damage develops very early in thedisease process, possibly even from the start.

Osteoclast precursors and mature osteoclastshave been detected in the synovial membranes of ani-mals with various forms of experimental arthritis as wellas in humans with RA, whereas normal synovial tissuedoes not harbor osteoclasts (5–9). Inflamed synoviumhas a particular capacity to invade bone (10). Currentthinking holds that the several cell populations residingin the inflamed synovial membrane provide signals thatstimulate osteoclast formation and facilitate bone re-sorption. Synovial fibroblast-like cells and activated Tcells produce RANKL, which is a potent stimulator ofosteoclastogenesis (5,11–13). RANKL can activate cellsof the monocyte/macrophage lineage that exist in theinflamed synovium; these cells constitute a large pool ofosteoclast precursors. The colocalization of a potentosteoclast-inducing cytokine and its target population of

1Georg Schett, MD: Amgen, Inc., Thousand Oaks, California,and Medical University of Vienna, Vienna, Austria; 2Marina Stolina,PhD, Scot Middleton, MS, Matt Adlam, PhD, Heather Brown, BS, LiZhu, BS, Ulrich Feige, PhD (current address: ESBATech, Zurich-Schlieren, Switzerland), Debra J. Zack, MD, PhD: Amgen, Inc.,Thousand Oaks, California; 3Brad Bolon, DVM, PhD: Amgen, Inc.,Thousand Oaks, California, and GEMpath, Inc., Cedar City, Utah.

Dr. Schett was a visiting scientist at Amgen. Drs. Stolina,Bolon, Adlam, Feige, and Zack, and Mr. Middleton, Mr. Zhu, and MsBrown own stock in Amgen.

Address correspondence and reprint requests to Debra J.Zack, MD, PhD, Amgen, Inc., One Amgen Center Drive, B-38-2-B,Thousand Oaks, CA 91320. E-mail: dzack@amgen.com.

Submitted for publication February 3, 2005; accepted inrevised form June 30, 2005.

3192

osteoclast precursors provides the basis for rapid oste-oclastogenesis in arthritis.

Although the link between bone erosion andsynovial osteoclast formation has not been directlyproven in RA, compelling evidence from animal modelsof experimental arthritis suggest such a link. Agents thatinhibit the action of osteoclasts, such as bisphosphonatesor the RANKL-binding soluble receptor osteoprote-gerin, can inhibit or even arrest bone erosion (5–7,14–19). Gene therapy with molecules that inhibit drivers ofosteoclastogenesis effectively inhibit bone erosion (20).Bone erosion does not develop when arthritis is inducedin mice that have been genetically modified to bedeficient in osteoclasts (21,22). Thus, detection of oste-oclasts in synovium can be deemed a suitable tool withwhich to explore the early pathogenesis of bone destruc-tion in experimental arthritis.

A key issue for a full understanding of the onsetand kinetics of bone damage in human RA is determin-ing how long it takes for functional osteoclasts to ariseand expand once synovial inflammation has begun. Toinvestigate this issue, we studied the kinetics of synovialosteoclastogenesis in Lewis rat models of chronicimmune-mediated arthritis (adjuvant-induced arthritis[AIA] and collagen-induced arthritis [CIA]) and acuteinflammation (carrageenan-induced paw swelling).

MATERIALS AND METHODS

Animals and induction of arthritis. Young adult Lewisrats (44 males and 44 females) weighing 80–100 gm werepurchased from Charles River (Wilmington, MA). Animalswere acclimatized for 1 week under normal environmentalconditions and fed a pelleted rodent chow (#8640; HarlanTeklad, Madison, WI) with tap water ad libitum.

AIA was induced in male rats (n � 24) by a singleintradermal injection (into the base of the tail) of 0.5 mg ofheat-killed Mycobacterium tuberculosis H37Ra (Difco, Detroit,MI) suspended in paraffin oil. CIA was induced in female rats(n � 24) by multiple intradermal injections (into the skin of theback) with a total of 1 mg of porcine type II collagen(Chondrex, Redmond, WA) emulsified 1:1 in Freund’s incom-plete adjuvant (Difco). Groups of rats with AIA or CIA werekilled at disease onset (day 0), or 1, 5, 10, or 20 days afterdisease onset. Arthritis was not induced in control animals(n � 5 animals per model).

Acute arthritis was induced in male rats (n � 8) bysubcutaneous injection (into the left hind paw) of 100 �l of 2%type IV carrageenan (Sigma, St. Louis, MO) solubilized insaline. Untreated animals (n � 4) were used as controls. Ratswere killed 1, 4, or 7 days after injection of carrageenan.

This study was conducted in accordance with federalanimal care guidelines and was preapproved by the Institu-tional Animal Care and Use Committee of Amgen.

Assessment of paw swelling. Swelling of hind paws wasassessed daily from disease onset to day 20 after disease onset.In rats with AIA, paw swelling was measured by water pleth-ysmography, as previously described (23). In rats with CIA,swelling was quantified by measuring the ankle diameter of thehind paws using calipers (Fowler Sylvac Ultra-Cal Mark III;Sylvac, Crissier, Switzerland). Paw swelling was not evaluatedfor carrageenan-inoculated animals because of the acute na-ture of this disease model.

Conventional histologic analysis and detection of os-teoclasts. At necropsy, the right hind paw of rats with AIA andCIA and the left hind paw of animals with carrageenan-induced acute arthritis were removed at the fur line, justproximal to the tibiotarsal (hock) joint, fixed in zinc formalinfor 2 days, and then decalcified with a 1:4 mixture of 8N formicacid and 1N sodium formate. Paws were then divided longitu-dinally along the median axis, processed into paraffin, and cutserially at 4 �m. One section was used for conventionalhistopathologic assessment and was stained with hematoxylinand eosin. The other section was used for osteoclast assess-ment and was subjected to an indirect immunoperoxidaseprocedure to detect cathepsin K, an osteoclast-specific pro-tease.

Immunohistochemistry was performed with an auto-mated tissue stainer (model Mark 5; DPC, Flanders, NJ)according to a standard method (24). Briefly, sections werepretreated with 0.1% trypsin in 1% CaCl2 (Sigma) for 15minutes, blocked with CAS Block (Zymed, San Francisco, CA)for 10 minutes, and incubated with a proprietary rabbit poly-clonal anti–cathepsin K antibody (1 �g/ml; Amgen, ThousandOaks, CA) for 60 minutes. The primary antibody was localizedusing sequential 30-minute incubations with biotin-conjugatedgoat anti-rabbit polyclonal secondary antibody (1:200 dilution;Vector, Burlingame, CA), peroxidase-blocking solution (Dako,Carpinteria, CA) for 25 minutes, and avidin–biotin complexand peroxidase reagents (ABC Elite Kit, Vector). The reactionwas visualized using diaminobenzidine (DAB� SubstrateChromogen system; Dako) for 3 minutes.

Double-labeling technique for osteoclasts and macro-phage markers. Cells expressing macrophage markers werelocalized in sections stained to reveal osteoclasts by incubationwith monoclonal mouse anti-rat CD68 antibody (clone ED-1;Serotec, Oxford, UK). Anti–cathepsin K reactivity on oste-oclasts was detected as described above, except that alkalinephosphatase–conjugated goat anti-rabbit antiserum (Vector)was used as the secondary antibody and an alkaline phospha-tase ABC kit (Vector) was used to demonstrate the reactionproduct. Anti-CD68 reactivity was detected using a biotinyl-ated horse anti-mouse antiserum (Vector), followed by con-struction of an immunoperoxidase bridge as described above.

Semiquantitative scoring of lesions and histomorpho-metric analysis. Synovial inflammation and bone erosion wereassessed in hematoxylin and eosin–stained sections using semi-quantitative scoring systems, as previously described(17,23,24). Inflammation in AIA and CIA was scored on ascale of 0–4, where 0 � normal, 1 � few inflammatory cells inthe perisynovial tissue, 2 � mild inflammation (few small focalaggregates; modest build-up in the perisynovial tissue), 3 �moderate inflammation (many small aggregates; extensive inthe perisynovial tissue), and 4 � marked inflammation (largeaggregates; extensive in the perisynovial tissue).

KINETICS OF OSTEOCLASTOGENESIS IN ARTHRITIC RATS 3193

Bone erosion in AIA was scored on a scale of 0–5,where 0 � normal, 1 � minimal erosion (few sites in the tarsalbones), 2 � mild erosion (modest number of sites in the tarsalbones), 3 � moderate erosion (many sites in the tarsal bones),4 � marked erosion (partial destruction of the tibia andextensive destruction of the tarsal bones), and 5 � extensiveerosion (fragmentation of the tarsal bones and full-thicknesscortical penetration of the tibia).

Bone erosion in CIA was scored on a scale of 0–5,where 0 � normal, 1 � minimal erosion (1–2 small shallowsites), 2 � mild erosion (1–4 sites of medium size and depth),3 � moderate erosion (�5 sites partially extending through thecortical bone), 4 � marked erosion (multiple foci partly orcompletely extending through the cortical bone), and 5 �extensive erosion (cortical penetration at �25% of the bonelength). Analysis included the tibiotarsal articulation and allintertarsal joints.

In addition, the number of cathepsin K–labeled oste-oclasts (multinucleated cells attached to bone) and precursors(mononucleated and multinucleated cells within synovial in-flammatory tissue) were analyzed quantitatively by histomor-phometry of the navicular bone and the adjacent talonavicularjoint. The number, size, and nucleation of osteoclasts and theirprecursors were assessed in the various compartments of thenavicular bone and talonavicular joint (Figure 1), as follows:area 1 � synovial inflammatory tissue (osteoclast precursors),area 2 � periosteum, area 3 � subchondral bone, area 4 �trabecular bone, and area 5 � osteophytes. The same areaswere measured in each of the various groups. Previous studieshave validated the navicular bone as a sensitive indicator of theextent of arthritic changes in experimental arthritis in rats (24).All parameters were analyzed using commercial image analysissoftware (OsteoMeasure version 2.2; Osteometrics, Atlanta,GA), as previously described (18).

Statistical analysis. All results are expressed as themean � SEM. Groups were compared with the nonparametricKruskal-Wallis test, using GraphPad Prism version 4 software(GraphPad Software, San Diego, CA). P values less than 0.05were considered significant.

RESULTS

Occurrence of bone loss in parallel with a rapidincrease in joint inflammation in AIA and CIA. We firstaddressed the course of joint swelling, as well as his-topathologic evidence of synovial inflammation andbone erosion, in rats with AIA and CIA. After inductionof AIA, paw swelling started 9 days after immunization(i.e., disease onset [day 0]), rapidly increased during thefollowing 5 days, and reached a plateau by 10 days afterdisease onset (Figure 2A). Paw swelling in CIA started10 days after immunization (i.e., disease onset [day 0]),peaked after 3 days, and remained stable up to 20 daysafter disease onset (Figure 2B).

Synovial inflammation in AIA was evident, butminimal, from 1 day before disease onset to 1 day afteronset, was marked at 5 and 10 days after onset, and then

regressed (Figure 2C). In contrast, inflammation in CIAwas not apparent until the day of disease onset, at whichtime it was mild, but already substantial. As with AIA,inflammation in CIA peaked at 5 and 10 days after onsetand then regressed (Figure 2D).

Bone erosions were detectable at the onset ofclinical disease in AIA and CIA, although the lesionswere shallow and confined to small areas of the tarsalbones (Figures 2E and F). Erosions developed moreslowly than did synovial inflammation, becoming sub-stantial only after 10 days and peaking at day 20 afterdisease onset in both models. The most pronouncedincrease in erosions was observed between day 5 and day10 in rats with AIA and between day 1 and day 5 in ratswith CIA (Figures 2E and F).

Figure 1. Compartments of the navicular bone in which osteoclastswere assessed in Lewis rats. Shown are histologic sections of the hindpaw of A, a healthy rat (control), C, a rat with adjuvant-inducedarthritis (AIA), and D, a rat with collagen-induced arthritis (CIA), aswell as B, a schematic representation of the specific areas scored.Sections were stained for cathepsin K (brown) and counterstained withhematoxylin and eosin. Images show the navicular bone, the talona-vicular joint, and the head of the talus, as labeled in A. Osteoclasts(brown dots) were assessed in the following compartments: area 1 �synovial inflammatory tissue, area 2 � periosteum, between thesynovial inflammatory tissue and the cortical bone, area 3 � subchon-dral bone, including the Haversian channels, area 4 � trabecular bonebetween the 2 layers of subchondral cortical bone, and area 5 �osteophytes at both sides of navicular bone. Green arrows indicate themajor directions of osteoclast-mediated bone destruction in AIA andCIA. The color-coded drawing of the navicular bone in B shows theboundaries of the areas of measurement as numbered above, exceptfor area 1 (the synovial inflammatory tissue), which is not shown.(Original magnification � 50.)

3194 SCHETT ET AL

Initiation of osteoclastogenesis at the onset ofAIA and CIA. Next, we performed a quantitative analysisof osteoclasts and their precursors in various subcom-partments of the talonavicular joint, the most highlyinvolved site of bone destruction in rats with AIA (21).In both AIA and CIA, numerous osteoclast precursorsas well as osteoclasts were generated immediately after

the onset of disease, and these cells were found in varioussubcompartments of the joint and in the neighboring bone(Figure 3). In both AIA and CIA, osteoclasts firstappeared at the so-called “junction zone” between thesynovial membrane and periosteal lining, as well as in bonechannels (also known as Haversian channels) in subchon-dral bone, beneath the articular cartilage (Figures 4A–F).

Figure 2. Clinical and histologic course of adjuvant-induced arthritis (AIA) and collagen-induced arthritis (CIA). A, C, and E, AIA was induced in male Lewis rats and B, D, and F,CIA was induced in female Lewis rats (n �15 animals per model). For each model, thecontrol cohort consisted of 5 untreated animals. Hind paw swelling (A and B) was measured1 day before disease onset (day –1), the day of disease onset (day 0), daily until day 5 afterdisease onset, and then on days 7, 10, 15, and 20 after disease onset. Histopathologic indicesof synovial inflammation (C and D) and bone erosion (E and F) were determined in the righthind paw using semiquantitative scoring systems (graded 0–4 for inflammation and 0–5 forbone erosion; see Materials and Methods for details). These analyses were performed 1 daybefore disease onset (day –1), on the day of disease onset (day 0), and on days 1, 5, 10, and20 after disease onset. Values are the mean and SEM.

KINETICS OF OSTEOCLASTOGENESIS IN ARTHRITIC RATS 3195

Whereas osteoclasts located in the periosteumdisappeared with progression of disease, their numberscontinuously increased within the (Haversian) channelsof subchondral bone (Figure 3). This latter expansionresulted directly in the extensive resorption of subchon-

Figure 3. Number and localization of osteoclasts and their precursorsin adjuvant-induced arthritis (AIA) and collagen-induced arthritis(CIA). A, AIA was induced in male Lewis rats and B, CIA was inducedin female Lewis rats (n � 15 animals per model). Osteoclasts and theirprecursors were identified using an indirect immunohistochemicaltechnique to detect the osteoclast-specific protease cathepsin K.Analyses were performed on the day of disease onset (day 0) as well ason days 1, 5, 10, and 20 after disease onset at the following sites in thetarsonavicular bone: synovial membrane, periosteal lining, subchon-dral bone, trabecular bone, and osteophytes. Values are the mean andSEM.

Figure 4. Sites of early osteoclast formation and bone resorption inadjuvant-induced arthritis (AIA). Sections of the talonavicular jointfrom male Lewis rats with AIA were stained with hematoxylin andeosin (A, C, E, G, and I), and corresponding serial sections werestained for the osteoclast marker cathepsin K (brown) (B, D, F, H, andJ). On the day of disease onset (day 0), osteoclast formation (arrow)can be seen in a subchondral bone channel next to the joint margin (Aand B) and at the periosteal lining (C and D). On day 5 after diseaseonset, the subchondral bone channel is filled with numerous oste-oclasts (arrow) (E and F), and the surfaces of subchondral bone (G andH) and trabecular bone (I and J) are undergoing extensive resorptionby osteoclasts (arrow). (Original magnification � 40 in A, B, E, F, G,and H; � 100 in C and D.)

3196 SCHETT ET AL

Figure 5. Osteoclast precursors as mononuclear cathepsin K–positive and macrophage marker–positive cells in synovial tissue from rats withcollagen-induced arthritis (CIA). Sections of hind paws from female Lewis rats with CIA were labeled for the osteoclast-specific protease cathepsinK (brown) (A and B) or were double-labeled for cathepsin K (blue) and the macrophage-specific marker ED-1 (brown) (C–I). A, On day 3 afterdisease onset, the inflamed synovial membrane shows numerous mononucleated cathepsin K–positive osteoclast precursors in the lining layer (blackarrow), the connective tissue beneath the lining (arrowhead), and the joint effusion (red arrow). B, Higher-magnification view of A, showingosteoclast precursors (arrow). C–I, Double labeling for ED-1 and cathepsin K reveals double-labeled osteoclast precursors (black) (blackarrowhead) and single-labeled macrophages (brown) (black arrow) in various tissue compartments. On day 0 (disease onset), there are manymacrophages, but no osteoclast precursors, in the synovial membrane of normal rats (C) or in rats with CIA (D). On day 1 (E) and day 3 (F) afterdisease onset, many osteoclast precursors are mingled with macrophages in the synovial lining layer. On day 0, osteoclasts can be seen together withmacrophages within bone channels near the junction zone (G) and the subchondral region (H). On day 5, macrophages, osteoclast precursors,unpolarized preosteoclasts (red arrow), and polarized osteoclasts (red arrowhead) can be seen inside a subchondral bone erosion (I). (Originalmagnification � 100 in A; � 200 in G–I; � 400 in B–F.) J, Quantification of the proportion of osteoclasts or osteoclast precursors as a percentageof the total macrophage marker–positive cells in the synovial lining layer, synovial sublining, and subchondral bone channels. Values are the meanand SEM.

dral cortical bone in both AIA and CIA (Figures 4G andH). Furthermore, in AIA, the trabecular bone in themarrow cavity faced a massive attack by osteoclasts 5days after disease onset (Figures 4I and J). Interestingly,this compartment was almost unaffected in CIA. Oste-oclast precursors were also found within the synovialmembrane itself (Figure 3). Their numbers increasedafter the formation of an inflammatory synovial infil-trate and peaked on day 10 and day 5 after disease onsetin AIA and CIA, respectively. Finally, at later stages ofAIA and CIA, even the osteophytes (raised nodules ofnew bone formed to stabilize the degenerating joint)also contained osteoclasts (Figure 3).

Increased size and nuclear number of osteoclastsafter the onset of AIA and CIA. To further characterizeosteoclast production in arthritis, we measured theaverage size and nucleation of osteoclasts attached toeroding bone surfaces. As early as 1 day after diseaseonset, mature osteoclasts (mean � SEM size 1,711 �162 �m2 in AIA and 1,346 � 167 �m2 in CIA) weremore than twice as large as osteoclast precursor cells.Osteoclast size peaked on day 5 (2,801 � 764 �m2) andday 10 (3,080 � 361 �m2) after the onset of arthritis inAIA and CIA, respectively. Similarly, the number ofnuclei in activated osteoclasts lining the bony surfacesincreased with the progression of arthritis, reaching amaximum of 5 nuclei per cell in both models on day 5following disease onset.

Monocyte/macrophage lineage–derived cathep-sin K–positive mononucleated cells in the synovium andmultinucleated cells attached to eroding bone. To testwhether the cathepsin K–positive osteoclasts and oste-oclast precursors originated from the monocyte/macrophage lineage, we performed a double-labelingprocedure using antibodies directed against the oste-oclast marker cathepsin K and the macrophage markerED-1. A proportion of mononucleated cells in thesynovial membrane, both in the synovial lining layer andin the synovial sublining, as well as in the joint effusion,expressed cathepsin K (Figures 5A and B). All cathepsinK–positive mononucleated and multinucleated cells inthe synovial membrane (osteoclast precursors), as wellas all multinucleated cells attached to bone (osteoclasts),were colabeled with the macrophage marker (Figures5A–D). This included many cathepsin K–positive mono-nucleated cells that had been released from the inflamedsynovium into the joint cavity (Figure 5A). CathepsinK–positive cells that did not express the macrophagemarker were not found. In contrast, only a small pro-portion of the cells positive for the macrophage markerwere also positive for cathepsin K (Figure 5I). This

indicates that osteoclast differentiation occurs only in asmall proportion of cells of the monocyte/macrophagelineage and suggests that these cells undergo differenti-ation due to signals such as RANKL expressed from theneighboring synovial membrane.

Macrophages appeared in the periarticular softtissues 1 day before the onset of clinical disease, whereasosteoclast precursors appeared later (see Figure 3). Theproportion of osteoclast precursors within the monocyte/macrophage compartment significantly increased duringthe first days of AIA (data not shown) and CIA (Figure5E), suggesting an increased differentiation of monocyte/macrophages into osteoclasts. In contrast, double-labeled osteoclasts surrounded by macrophages werefound within cortical and subchondral bone channels onthe day of disease onset (Figures 5G and H). At laterstages of disease, bone erosions exhibited the full set ofosteoclast differentiation steps, with macrophages,mononucleated cathepsin K–positive osteoclast precur-sors, multinucleated cathepsin K–positive osteoclast pre-cursors resembling preosteoclasts in the vicinity of bone,and mature polarized osteoclasts in direct contact withthe bone surface (Figure 5I). The proportion of oste-oclast precursors within the monocyte/macrophage com-partment significantly increased in the synovial lininglayer and the synovial sublining of rats with AIA (datanot shown) and CIA (Figure 5J) during the first day ofarthritis and reached �25% on day 3 after disease onset.In contrast, differentiation of osteoclasts in bone chan-nels reached 25% far earlier, on the day of disease onset,and almost every macrophage marker–positive cell haddifferentiated into an osteoclast on day 1 after diseaseonset.

Osteoclastogenesis triggered by acute inflamma-tion. To study whether acute inflammation in generaltriggers osteoclastogenesis and to study the kinetics ofosteoclastogenesis in more detail, we induced hind pawswelling in rats by the subcutaneous injection of carra-geenan. Although this model does not lead to thegeneration of a hyperplastic synovial membrane, a mas-sive increase in subchondral osteoclasts was observed inmultiple joints (both tibiotarsal and intertarsal) as earlyas 4 days after inoculation with carrageenan (Figures6A–F). After day 4, a rapid decline in osteoclast num-bers was observed, suggesting that increased osteoclas-togenesis in this model is transient. However, this burstof osteoclastogenesis was sufficient to induce a wideningof the subchondral bone channels, as detected by histo-morphometric analysis of the subchondral bone of theankle and talonavicular joint, suggesting that osteoclastswere functionally active (Figures 6C and D).

3198 SCHETT ET AL

Osteoclasts were predominantly localized in theHaversian channels beneath the cartilage (Figures 6Cand E). In normal rats, these channels contained bloodvessels suspended in a mesh of fine connective tissues,

but no osteoclasts (Figures 6D and F). Functionalactivity of osteoclasts induced by carrageenan injectionwas evident from a widening of the subchondral chan-nels, as detected by histomorphometric analysis of thesubchondral bone of the ankle and talonavicular joint(data not shown).

DISCUSSION

In this study, we examined the kinetics of oste-oclastogenesis in 3 animal models of arthritis: 2 chronicimmune-mediated conditions and 1 acute inflammatorysystem. Our data showed that osteoclasts appear at, orimmediately after, the onset of clinical arthritis, increaserapidly in size and nucleation (indices of a functionallyactive state), and populate virtually all bony compart-ments of the inflamed joint. Initially, osteoclasts appearat the junction between the periosteum, synovial inser-tion, and cartilage as well as in (Haversian) channelswithin subchondral bone. Soon after, myriad osteoclastprecursors develop in the synovial membrane, concom-itant with a massive increase in osteoclasts along thesurfaces of subchondral and trabecular bone, whichresults in rapid resorption of bone. Osteoclastogenesis inchronic conditions (AIA and CIA) appears to be acontinuous process, whereas osteoclast expansion istransient when the inflammatory process is acute andself-limiting (carrageenan-induced arthritis) (25).

One major finding of this study confirmed theearly formation of osteoclasts in arthritic joints. Giventhe fact that osteoclasts are primarily responsible forbone damage, this observation suggests that the ele-ments needed to drive bone destruction are formed veryearly in the course of arthritis progression. Interestingly,bone erosion does not occur, even if arthritis is highlyaggressive and longstanding, in animals in which oste-oclasts have been targeted by genetic deletion of essen-tial genes needed for osteoclast development or byadministration of effective drugs that interfere withosteoclastogenesis (5–7,14–22). Since osteoclasts arealso a common feature in the joints of patients with RA,their contribution to the genesis of irreversible bonedamage in human disease is quite likely (8,9,26,27).

Although osteoclastogenesis cannot be directlyinvestigated in vivo at the cellular level in humans withRA, various imaging techniques have confirmed theearly occurrence of bone erosion. Even comparativelyinsensitive conventional radiography shows evidence ofbone dissolution (presumably mediated by osteoclasts)in 40% of patients with RA of 6 months’ duration (4),while more advanced techniques, such as magnetic res-

Figure 6. Carrageenan-induced paw swelling and increased osteoclastformation in Lewis rats. Acute arthritis was induced in male Lewis rats(n � 8) by injection of carrageenan (day 0), and injected hind pawswere analyzed histopathologically on days 1, 4, and 7 after injection.Uninjected animals (n � 4) served as controls. Osteoclasts in A, thetibiotarsal (ankle) joint and B, the talonavicular joint were quantifiedby using cathepsin K immunolabeling. C and D, Histomorphometry ofthe subchondral bone regions of C, the tibiotarsal joint and D, thetalonavicular joint, showing the relative amount of subchondral bonereplaced by widened bone channels (percentage of bone-free surfaceof the total subchondral surface). Values are the mean and SEM. * �P � 0.05 versus control. E, In rats with carrageenan-induced arthritisevaluated on day 4, osteoclasts (brown) were localized in the subchon-dral bone channels. F, In normal control rats evaluated on the sameday, bone channels did not contain osteoclasts. G and H, Higher-magnification views of E and F, respectively. Arrows indicate subchon-dral bone channels; arrowheads indicate osteoclasts. (Original magni-fication � 40 in E and F; � 200 in G and H.)

KINETICS OF OSTEOCLASTOGENESIS IN ARTHRITIC RATS 3199

onance or ultrasound imaging, may ultimately serve todetect bone erosion at even earlier stages of disease(28,29). Importantly, the observation that bone erosionis frequent in RA and occurs early in the disease implies2 major facts. First, osteoclasts must begin forming at, orsoon after, the onset of disease in order for there to besufficient time to resorb the quantities of bone necessaryto yield erosive lesions that can be visualized by conven-tional radiography. Second, osteoclast expansion duringthis early phase of arthritis progression must rapidlyevolve a sufficient number of osteoclasts to allow thedestruction of a significant amount of bone surface.

The 2 animal models of chronic immune-mediated arthritis examined in this study, AIA and CIA,effectively demonstrated that both these criteria arefulfilled in aggressive joint-eroding conditions. Oste-oclasts are generated in substantial numbers right fromthe onset of arthritis, and the huge numbers ofmonocyte/macrophages that infiltrate the inflamed syno-vium represent a pool of precursors from which torecruit new osteoclasts.

At the very first stage of clinically apparent jointinflammation, multinucleated osteoclasts were alreadyfound in 2 locations in the affected joint: in the connec-tive tissues at the junction of the synovial membrane andthe periosteum and in the vascular channels spanningthe cortical bone underneath the articular cartilage.These locations are in close contact with vulnerablebone surfaces, thus supplying the basis for rapid andeffective destruction of the joint architecture. Theseearly sites of osteoclast accumulation correspond pre-cisely to the locations at which radiographically evidentbone erosions first develop in RA: localized bone ero-sions at the joint margins, which represent the site ofinsertion of the synovial membrane at the periosteum,and juxtaarticular osteopenia, which reflects the widen-ing of vascular channels in the subchondral bone (30).These channels are used physiologically for the nutritionof bone. However, in the case of arthritis, the channelsseem to represent a weak point in the joint architecture,since they effectively serve to concentrate inflammatorycells that produce osteoclastogenic factors (5) and oste-oclast precursors in a confined space located immedi-ately adjacent to the bony substrate that provides thebasis for osteoclast-mediated joint dissolution.

To allow the relentless resorption of bone thatculminates in joint destruction, however, sufficientquantities of osteoclasts have to be formed on a contin-uous basis. Our data indicate that this process dependson the persistent presence of synovial inflammation. Inthe carrageenan-injection model of acute arthritis, the

induction of temporary inflammation in periarticulartissues led to a rise in osteoclast numbers and promotedearly bone-resorptive changes, such as widening of sub-chondral bone channels. However, the rapid decline inthe severity of this synovial leukocyte infiltrate wasaccompanied by regression of multinucleated oste-oclasts, so extensive bone damage did not emerge. Incontrast, in the 2 models of chronic inflammatory arthri-tis, the inflamed synovial membrane as well as theleukocyte-filled marrow cavity continued to supply largenumbers of osteoclast precursors. The precursor pool inthe synovial membrane may be a particularly importantpool for osteoclast recruitment, since these cells arefound diffusely throughout the synovium and are alsoreleased into the joint cavity. This abundance of oste-oclast precursors in the synovial membrane also enablesthe rapid fusion and increased nucleation of cells. Inboth models, fusion is accompanied by a continuousincrease in size of the cells, which, as typically seen in theCIA model, outlasts the fusion process and peaks uponpolarization and final differentiation of metabolicallyactive osteoclasts.

In summary, our current data indicate that oste-oclasts are first formed at, or soon after, the onset ofclinical arthritis and are replenished continuously duringthe course of disease progression. These facts furtherintimate that early and effective interference with osteo-clastogenesis represents an attractive strategy by whichto protect against skeletal damage in humans with RA,particularly the irreversible joint dissolution that leads tocrippling. Our findings indicate that potent inhibition ofosteoclastogenesis will likely be necessary beginningright after the onset of RA, when osteoclasts are firstformed and long before skeletal damage is visible onstandard radiographs. Such aggressive bone-sparingtherapy might be achieved either as an indirect conse-quence of an agent’s antiinflammatory properties (e.g.,cytotoxic disease-modifying antirheumatic drugs orcytokine-inhibiting biomolecules) or as a direct result of(specific) osteoclast inhibition (e.g., the RANKL-binding receptor osteoprotegerin or bisphosphonates).From a clinical perspective, the outcome with respect tojoint preservation will be indistinguishable. However,interference with osteoclast expansion will be particu-larly important before full control of inflammation isachieved, since active inflammation will perpetuate theproduction of new osteoclasts. Thus, our findings indi-cate that in aiming for full preservation of the skeletalarchitecture of joints during arthritis, the early andeffective blockade of osteoclastogenesis may be re-quired.

3200 SCHETT ET AL

REFERENCES

1. Goldring S. Osteoporosis and rheumatic diseases. In: Favus MJ,editor. Primer on the metabolic bone diseases and disorders ofmineral metabolism. 5th ed. Washington, DC: American Societyfor Bone and Mineral Research; 2003. p. 379–81.

2. Scott DL, Pugner K, Kaarela K, Doyle DV, Woolf A, Holmes J, etal. The links between joint damage and disability in rheumatoidarthritis. Rheumatology (Oxford) 2000;39:122–32.

3. Pincus T. Rheumatoid arthritis: disappointing long-term outcomesdespite successful short-term clinical trials. J Clin Epidemiol1988;41:1037–41.

4. Van der Heijde DM. Joint erosions and patients with earlyrheumatoid arthritis. Br J Rheumatol 1995;34:74–8.

5. Kong YY, Feige U, Sarosi I, Bolon B, Tafuri A, Morony S, et al.Activated T cells regulate bone loss and joint destruction inadjuvant arthritis through osteoprotegerin ligand. Nature 1999;402:304–9.

6. Redlich K, Hayer S, Maier A, Dunstan CR, Tohidast-Akrad M,Lang S, et al. Tumor necrosis factor �–mediated joint destructionis inhibited by targeting osteoclasts with osteoprotegerin. ArthritisRheum 2002;46:785–92.

7. Romas E, Sims NA, Hards DK, Lindsay M, Quinn JW, Ryan PF,et al. Osteoprotegerin reduces osteoclast numbers and preventsbone erosion in collagen-induced arthritis. Am J Pathol 2002;161:1419–27.

8. Bromley M, Woolley DE. Chondroclasts and osteoclasts at sub-chondral sites of erosion in the rheumatoid joint. Arthritis Rheum1984;27:968–75.

9. Gravallese EM, Harada Y, Wang JT, Gorn AH, Thornhill TS,Goldring SR. Identification of cell types responsible for boneresorption in rheumatoid arthritis and juvenile rheumatoid arthri-tis. Am J Pathol 1998;152:943–51.

10. Pap T, Aupperle KR, Gay S, Firestein GS, Gay RE. Invasivenessof synovial fibroblasts is regulated by p53 in the SCID mouse invivo model of cartilage invasion. Arthritis Rheum 2001;44:676–81.

11. Gravallese EM, Manning C, Tsay A, Naito A, Pan C, Amento E,et al. Synovial tissue in rheumatoid arthritis is a source ofosteoclast differentiation factor. Arthritis Rheum 2000;43:250–8.

12. Shigeyama Y, Pap T, Kunzler P, Simmen BR, Gay RE, Gay S.Expression of osteoclast differentiation factor in rheumatoid ar-thritis. Arthritis Rheum 2000;43:2523–30.

13. Lubberts E, Oppers-Walgreen B, Pettit AR, van den Bersselaar L,Joosten LA, Goldring SR, et al. Increase in expression of receptoractivator of nuclear factor �B at sites of bone erosion correlateswith progression of inflammation in evolving collagen-inducedarthritis. Arthritis Rheum 2002;46:3055–64.

14. Herrak P, Gortz B, Hayer S, Redlich K, Reiter E, Gasser J, et al.Zoledronic acid protects against local and systemic bone loss intumor necrosis factor–mediated arthritis. Arthritis Rheum 2004;50:2327–37.

15. Sims NA, Green JR, Glatt M, Schlict S, Martin TJ, Gillespie MT,et al. Targeting osteoclasts with zoledronic acid prevents bonedestruction in collagen-induced arthritis. Arthritis Rheum 2004;50:2338–46.

16. Goldring SR, Gravallese EM. Bisphosphonates: environmentalprotection for the joint? [editorial]. Arthritis Rheum 2004;50:2044–7.

17. Campagnuolo G, Bolon B, Feige U. Kinetics of bone protection byrecombinant osteoprotegerin therapy in Lewis rats with adjuvantarthritis. Arthritis Rheum 2002;46:1926–36.

18. Redlich K, Gortz B, Hayer S, Zwerina J, Warmington K, Ko-stenuik P, et al. Repair of local bone erosions and reversal ofsystemic bone loss upon therapy with anti-TNF in combinationwith OPG or PTH in TNF-mediated arthritis. Am J Pathol2004;164:543–5.

19. Zwerina J, Hayer S, Tohidast-Akrad M, Bergmeister H, RedlichK, Feige U, et al. Single and combined inhibition of tumor necrosisfactor, interleukin-1, and RANKL pathways in tumor necrosisfactor–induced arthritis: effects on synovial inflammation, boneerosion, and cartilage destruction. Arthritis Rheum 2004;50:277–90.

20. Lubberts E, Joosten LA, Chabaud M, van den Bersselaar L,Oppers-Walgreen B, Coenen-de Roo CJ, et al. IL-4 gene therapyfor collagen arthritis suppresses synovial IL-17 and osteoprote-gerin ligand and prevents bone erosion. J Clin Invest 2000;105:1697–710.

21. Redlich K, Hayer S, Ricci R, David JP, Tohidast-Akrad M, KolliasG, et al. Osteoclasts are essential for TNF-�-mediated jointdestruction. J Clin Invest 2002;110:1419–27.

22. Pettit AR, Ji H, von Stechow D, Muller R, Goldring SR, Choi Y,et al. TRANCE/RANKL knockout mice are protected from boneerosion in a serum transfer model of arthritis. Am J Pathol2001;159:1689–99.

23. Feige U, Hu YL, Gasser J, Campagnuolo G, Munyakazi L, BolonB. Anti-interleukin-1 and anti-tumor necrosis factor-� synergisti-cally inhibit adjuvant arthritis in Lewis rats. Cell Mol Life Sci2000;57:1457–70.

24. Bolon B, Morony S, Cheng Y, Hu YL, Feige U. Osteoclastnumbers in Lewis rats with adjuvant-induced arthritis: identifica-tion of preferred sites and parameters for rapid quantitativeanalysis. Vet Pathol 2004;41:30–6.

25. Willoughby D. Carrageenan induced paw edema. In: Winyard PG,Willoughby DA, editors. Inflammation protocols. 1st ed. Totowa(NJ): Humana Press; 2003. p. 109–22.

26. Goldring SR. Pathogenesis of bone erosions in rheumatoid arthri-tis. Curr Opin Rheumatol 2002;14:406–10.

27. Schett G, Redlich K, Smolen JS. The role of osteoprotegerin inarthritis. Arthritis Res Ther 2003;5:239–45.

28. McQueen F, Lassere M, Edmonds J, Conaghan P, Peterfy C, BirdP, et al. OMERACT rheumatoid arthritis magnetic resonanceimaging studies: summary of OMERACT 6 MR imaging module.J Rheumatol 2003;30:1387–92.

29. Peterfy CG. New developments in imaging in rheumatoid arthritis.Curr Opin Rheumatol 2003;15:288–95.

30. Marinova-Mutafchieva L, Williams RO, Funa K, Maini RN,Zvaifler NJ. Inflammation is preceded by tumor necrosis factor–dependent infiltration of mesenchymal cells in experimental arthritis.Arthritis Rheum 2002;46:507–13.

KINETICS OF OSTEOCLASTOGENESIS IN ARTHRITIC RATS 3201