Formation of Osteoclast-Like Cells from Peripheral Blood of Periodontitis Patients Occurs Without Supplementation of Macrophage Colony-Stimulating Factor

J Clin Periodontol 2008; 35: 568–575 doi: 10.1111/j.1600-051X.2008.01241.x

Stanley T. S. Tjoa1, Teun J. de Formation of osteoclast-like cells Vries1,2, Ton Schoenmaker1,2, Angele Kelder3, Bruno G. Loos1 and Vincent Everts2 from peripheral blood of 1Department of Periodontology, Academic Centre for Dentistry Amsterdam (ACTA), Universiteit van Amsterdam and Vrije periodontitis patients occurs Universteit, Amsterdam, The Netherlands; 2Department of Oral Cell Biology, Academic Centre for Dentistry Amsterdam (ACTA), without supplementation of Universiteit van Amsterdam and Vrije Universteit, Amsterdam, Research Institute MOVE, The Netherlands; 3Department of Hematology, VUMC, Vrije Universiteit, macrophage colony-stimulating Amsterdam, The Netherlands factor

Tjoa STS, de Vries TJ, Schoenmaker T, Kelder A, Loos BG, Everts V. Formation of osteoclast-like cells from peripheral blood of periodontitis patients occurs without supplementation of macrophage colony-stimulating factor. J Clin Periodontol 2008; 35: 568–575. doi: 10.1111/j.1600-051X.2008.01241.x.

Abstract Aim: To determine whether peripheral blood mononuclear cells (PBMCs) from chronic periodontitis patients differ from PBMCs from matched control patients in their capacity to form osteoclast-like cells. Material and Methods: PBMCs from 10 subjects with severe chronic periodontitis and their matched controls were cultured on plastic or on bone slices without or with macrophage colony-stimulating factor (M-CSF) and receptor activator of nuclear factor-kB ligand (RANKL). The number of tartrate-resistant acid phosphatase-positive (TRACP1) multinucleated cells (MNCs) and bone resorption were assessed. Results: TRACP1 MNCs were formed under all culture conditions, in patient and control cultures. In periodontitis patients, the formation of TRACP1 MNC was similar for all three culture conditions; thus supplementation of the cytokines was not needed to induce MNC formation. In control cultures, however, M-CSF or M-CSF/RANKL resulted in higher numbers compared with cultures without cytokines. Upregulations of osteoclast marker mRNA cathepsin K and carbonic anhydrase II conﬁrmed the osteoclastic character. Bone resorption was only observed when PBMCs were cultured in the presence of M-CSF and RANKL. Key words: osteoclast formation; periodontitis; Conclusion: Our data indicate that PBMCs from periodontitis patients do not need peripheral blood priming by M-CSF to become osteoclast-like cells, suggesting that PBMCs from periodontitis patients are present in the circulation in a different state of activity. Accepted for publication 20 March 2008

Conflict of interest and source of Chronic periodontitis is a multi-factorial primary aetiological factor of periodontal funding statement disease characterized by inflammation of disease is bacterial plaque (Kinane 1999), The authors declare that they have no the periodontium, leading to loss of soft studies have shown that not everyone is conflict of interests. tissue attachment and alveolar bone (Page equally susceptible to periodontal disease The funding for this study was received et al. 1997). If left untreated, this condition (Loe et al. 1986). The host response has from the Royal Netherlands Academy of may ultimately lead to tooth exfoliation. been considered as an important disease- Arts and Sciences (T. J. de V.). Although it is generally accepted that the modifying factor (Page et al. 1997). 568 r 2008 Blackwell Munksgaard No claim to original US government works Osteoclast formation in periodontitis 569

Osteoclasts are the cells ultimately significant increase in osteoclast formation Table 1. Characteristics of controls and responsible for alveolar bone resorp- and bone resorption activity compared patients tion. They are multinucleated giant with non-osteoporotic postmenopausal Pair C/P Age Gender Ethnicity Smoking cells (Roodman 1996) originating from and healthy controls, respectively. the monocyte/macrophage lineages Spontaneous osteoclastogenesis, for 1 C 41 M Caucasian 1 (Udagawa et al. 1990, Fujikawa et al. example osteoclast formation without P 40 M Caucasian 1 1996). Human osteoclast precursor cells addition of cytokines such as M-CSF 2 C 27 M North African À have been found in the mononuclear and RANKL, was higher in osteoporotic P 26 M North African À fraction of peripheral blood (PBMCs), women and arthritis patients (Ritchlin 3 C 29 M African black 1 P 29 M African black 1 and in vitro these cells differentiate into et al. 2003, D’Amelio et al. 2005). 4 C 45 F African black À osteoclasts (Faust et al. 1999, Shalhoub Similarly, increased spontaneous osteo- P 44 F African black À et al. 1999). clastogenesis was found in cancer 5 C 35 M Caucasian 1 Key regulators of osteoclast differen- patients with bone metastases compared P 35 M Caucasian 1 tiation are macrophage colony-stimulating with healthy controls or cancer patients 6 C 32 F North African À factor (M-CSF), receptor activator of without bone metastases (Roato et al. P 31 F North African À nuclear factor-kB (RANK) ligand 2005). This increase in osteoclast 7 C 48 M Caucasian 1 (RANKL) and osteoprotegerin (OPG). formation and osteoclast activity was P 49 M Caucasian 1 8 C 40 F Indian À In a mouse model, it was shown that also suggested to occur in PBMCs P 41 F Indian À the growth factor M-CSF is indispensa- from periodontitis patients compared 9 C 40 M Indian 1 ble to the proliferation of osteoclast with non-periodontitis patients (Brunetti P 43 M Indian 1 progenitors and their differentiation et al. 2005). However, it is not clear from 10 C 37 F African black À into mature osteoclasts (Tanaka et al. the latter study whether the increased P 37 F African black À 1993). However, M-CSF alone could osteoclast formation was due to the high- C, control; P, patient; M, male; F, female; 1, not induce osteoclasts to form resorption er average age of the patients (mean smoker; À , no smoker. pits on dentine slices (Jimi et al. 1999). 42.8 Æ 15.2 years) compared with the Differentiation and activation of osteo- controls (mean 30.5 Æ 5.2 years), clasts is triggered by RANKL, which is because it is known that osteoclast ACTA who received restorative dental a member of the tumor necrosis factor formation from human PBMCs may treatments and regular dental check-ups, (TNF)-related cytokines, and this cyto- increase with age (Cheleuitte et al. or were subjects from private dental kine has a direct involvement in the 1998, Jevon et al. 2002). clinics. A control subject was selected differentiation, activation and survival In the current study, we further in- on basis of the following criteria: max- of osteoclasts and their precursors vestigated whether PBMCs from perio- imum of one missing tooth per quadrant (Lacey et al. 1998). Binding of RANKL dontitis patients have an increased (third molars excluded), a radiographic to its receptor RANK stimulates osteo- capacity to form osteoclasts by analys- distance of o3 mm between the cemen- clast precursors to differentiate into ing the dependency of M-CSF and toenamel junction (CEJ) and the alveo- osteoclasts and activates mature osteo- RANKL. Osteoclastogenesis was eval- lar bone crest on all teeth on recent clasts to resorb bone. OPG is a decoy uated in PBMC cultures from perio- (not older than 1 year) bite-wings radio- receptor for RANKL, it competes with dontitis patients and controls who were graphs. The control subjects were matched RANK for binding of RANKL, thereby matched for age, sex, ethnicity and and paired with the patient group for age, acting as an inhibitor of osteoclast smoking. gender, ethnicity and smoking habits maturation and activity (Simonet et al. (Table 1), because these parameters are 1997, Lacey et al. 1998). Although the established risk determinants for perio- importance of the RANK–RANKL–OPG Material and Methods dontitis (Borrell & Papapanou 2005) and system in osteoclast biology was mainly could possibly influence osteoclasto- Subjects explored in mice, it is believed that the genesis (Jevon et al. 2002). regulatory mechanisms of human osteo- The patient group consisted of 10 un- An extensive medical history of each clast differentiation are similar (Suda treated patients diagnosed with general- participant was taken by a written ques- et al. 1999). ized severe chronic periodontitis. These tionnaire and by an interview. Each Various studies were undertaken to patients were referred to the Department subject was free from systemic diseases understand the cellular mechanisms of Periodontology at the Academic Cen- and had no clinical symptoms of bacter- underlying bone breakdown in people tre for Dentistry Amsterdam (ACTA) ial, viral or parasitic infections at the with osteopenic diseases, such as auto- for diagnosis and/or treatment of perio- time of the study, except for perio- immune arthritis (Ritchlin et al. 2003), dontitis. Generalized severe chronic dontitis in the patient group. None of post-menopausal osteoporosis (D’Amelio periodontitis patients were defined as the subjects in the study had taken any et al. 2005) and bone tumours (Roato patients who never had periodontal ther- form of medication that could affect et al. 2005). From these studies, it apy or at least not in the last 10 years their periodontal status, such as anti- appeared that differences in osteoclast and who presented with alveolar bone biotics, anti-inflammatory agents and recruitment exist at the level of periph- loss X50% of the root length on at least immune suppressants during the preced- eral blood. Peripheral blood mono- seven teeth on a set of full mouth dental ing 6 months. Women older than 46 years nuclear cells (PBMCs) from patients radiographs (Bizzarro et al. 2007). were excluded from the study to prevent with postmenopausal osteoporosis Patients were carefully matched with skewing of the data due to the possible (D’Amelio et al. 2005) and psoriatic healthy controls. Control subjects were onset of menopause-related osteoporo- arthritis (Ritchlin et al. 2003) exhibit a recruited among dental patients at sis. All subjects received both verbal r 2008 Blackwell Munksgaard No claim to original US government works 570 Tjoa et al. and written information about the pur- leucocyte marker CD45 (PerCP- fore, the number of TRACP1 MNCs pose of the study. The Medical Ethical labelled) was used to discriminate was determined after 21 days of cultur- Committee of the Academic Medical between white blood cells and unlysed ing. The cultured cells were washed Center of the University of Amsterdam red cells or debris. Lymphoid markers with PBS and fixed in 4% PBS-buffered approved the study and all subjects (CD19 PE, CD3 FITC or CD3 APC) formaldehyde for 5 min. and stained for signed an informed consent. were used to discriminate between B-cells TRACP activity using the leucocyte and T-cells. CD8 (FITC-labelled) and acid phosphatase kit (Sigma). The CD4 (PE-labelled) were included to nuclei were visualized by incubating PBMC isolation identify the cytotoxic/suppressor T-cells the cell cultures with diamidino-2- Venous blood was collected by veni- and T-helper/inducer cells, respectively. phenylindole-dihydrochloride (DAPI) puncture in lithium heparin-containing CD14 (APC-labelled) was included to in PBS. Overall, adhered cells after 21 blood collecting tubes (Vacutainer, identify monocytes. Data acquisition days seemed viable, because apoptotic Becton Dickinson, Plymouth, UK) was performed using a FACScalibur cells (shrunken nuclei) were scarcely from both the patient and his corre- equipped with an argon and red diode seen after DAPI staining. sponding control. Blood samples of the laser and analysis was performed using To assess the number of TRACP1 patient and his matched control were Cellquest software (Becton Dickinson). MNCs, five micrographs were taken taken on the same day and time and Cells negative for CD19, CD3 and from pre-determined positions per well were processed within 3 h. Two milli- CD14 were classified as ‘other, e.g. with a digital camera (Leica, Wetzlar, litres of blood was kept separate for granulocytes and NK cells’. Based on Germany) mounted on an inverted light determination of the leucocyte composi- the forward–sideward scatter of cells, an microscope (Leica) equipped with a tion. PBMCs were isolated from whole estimate was made of the total number fluorescence lamp and appropriate filters blood with Ficoll-Paque density gradient. of granulocytes present. These cells to visualize blue stained nuclei, using Heparinized whole blood was diluted in a are lost after Ficoll density gradients 20 Â magnification and quantified manu- 1:1 concentration with Hank’s Balanced and consequently not included in the ally using Image-Pro plus software Salt Solution (HBSS; Gibco BRL, Pais- cell cultures. (Media Cybernetics, Silver Spring, ML, ley, Scotland, UK). Five parts of diluted USA). The centre of the well was first blood were layered on top of three parts found with a low magnification, then Osteoclastogenesis assays lymphoprep (Axisshield Po CAS, Oslo, with an objective of 20 Â ,thismiddle Norway) and centrifuged without brake PBMCs were cultured in 96-well plates micrograph was taken, next the adjacent at 1000 Â g for 30 min. at room tempera- at a density of 106 cells per well on fields to the left, right and to the top and ture. The interface was carefully col- plastic and on 650-mm-thick bovine bottom were photographed. Only cells lected and washed two times with cortical long bone slices. The formation with three or more nuclei were consid- a-minimum essential medium (a-MEM; of multinucleated cells (MNCs) was ered as MNC. The number of nuclei may Gibco BRL)15% foetal calf serum assessed after 3 weeks of culturing on reflect the maturity of osteoclasts (Piper (FCS; HyClone, Logan, UT, USA)11% plastic, and bone resorption was evalu- et al. 1992). Based on the number of antibiotics [PSF: 100 U/ml penicillin, ated after 4 weeks. The PBMCs were nuclei per cell, each MNC was assigned 100 mg/ml streptomycin and 250 ng/ml cultured under three conditions: (1) to one of the three groups: 3–5 nuclei or amphotericin B (antibiotic antimyotic a-MEM (Gibco BRL) supplemented 6–10 nuclei or 410 nuclei. The data solution, Sigma, St. Louis, MO, USA)]. with 5% FCS, (2) a-MEM1FCS11% were expressed as mean numbers of PSF and M-CSF (R&D Systems, Min- TRACP1 MNCs per mm2. 1 Flow cytometry neapolis, MN, USA), (3) a-MEM 5% FCS11% PSF and M-CSF and RANKL Two millilitres of blood was analysed RNA analysis and real-time quantitative (PreProtech, Rocky Hill, NJ, USA). PCR by flow cytometry to differentiate M-CSF was used at a concentration of leucocytes. Four-colour flow cytometry 25 ng/ml and RANKL was used at a RNA from cells before culture and from was performed at the Hematology concentration of 40 ng/ml, concentra- cells cultured for 3 weeks was isolated Laboratory of the Vrije Universiteit tions previously found to be optimal in using the RNeasy Mini Kit (Qiagen, Medical Center (VUMC). The immuno- these assays (Olivier et al. 2008). Hilden, Germany) according to the phenotypic labelling was performed Although a broad range of M-CSF can manufacturer’s instructions. The RNA as follows: the erythrocytes were lysed be used in osteoclastogenesis assays, a concentration was measured with the for 10 min. using Pharmlyse (Becton concentration of 25 ng/ml M-CSF RiboGreen Kit (Molecular Probes, Dickinson, Franklin Lakes, NJ, USA) proved to be optimal (Hodge et al. Eugene, OR, USA). One hundred nano- and washed once with phosphate- 2007). Cells were cultured in duplicate grams of RNA was used in the reverse buffered saline (PBS) containing 0.1% and the culture media was changed transcriptase reaction which was per- human serum albumin. The cells were twice a week. The cultures were incu- formed according to the MBI Fermentas incubated for 15 min. at room tempera- bated in a humidified 5% CO2 atmo- cDNA Synthesis Kit (Vilnius, Lithua- ture with either fluorescein isothiocya- sphere at 371C. nia), using both Oligo(dT)18 and D(N)6 nate (FITC), phycoerythrin (PE), primers. peridinyl chlorophyllin (PerCP) or allo- Real-time PCR primers were de- phycocyanin (APC) conjugated mono- Tartrate-resistant acid phosphatase signed using the Primer Express soft- staining and cell count clonal antibodies (Becton Dickinson) ware, version 2.0 (Applied Biosystems, and washed once with PBS containing TRACP is widely used as a marker Foster City, CA, USA) and have been 0.1% human serum albumin. The pan- enzyme of osteoclast-like cells. There- used in a previous study (de Vries et al. r 2008 Blackwell Munksgaard No claim to original US government works Osteoclast formation in periodontitis 571

2006). To avoid amplification of genomic cating the bone slices in 10% ammo- Table 2. Teeth present and description of DNA, each amplicon spanned at least one nium hydroxide for 30 min. The bone severity of periodontal destruction in patients intron. The primers used were as follows: slices were then thoroughly washed and controls cathepsin K: forward primer CCA with demineralized water and subse- Patients Controls TATgTgggACAggAAgAgAgTT, reverse quently incubated in a 10% saturated primer TgCATCAATggCCACAgAgA; alum (KAl(SO4)2 Á 12H2O) solution for Total # teeth present 29.0 Æ 2.1 28.8 Æ 2.4 Carbonic anhydrase II: forward primer 10 min. Finally, the bone slices were # teeth with 26.0 Æ 3.7 0 Æ 0 TggACTggCCgTTCTAggTATT, reverse washed again with demineralized water X30% bone loss primer TCTTgCCCTTTgTTTTAATg and the resorption pits were stained with # teeth with 13.9 Æ 5.5 0 Æ 0 X50% bone loss gAA; b2-microglobulin: forward primer Coomassie brilliant blue. Five micro- AAgATTCAggTTTACTCACgTC, re- graphs from pre-determined positions Values are reported as mean Æ standard devia- verse primer TgATgCTgCTTACATg in the centre of the bone slices were tion. Percentage bone loss was based on mea- TCTCg. taken with a digital camera mounted on surements on the periapical radiographs. The external standard curve used in a light microscope using 20 Â magni- the PCR reactions is cDNA from the fication and the percentage of resorbed quantitative PCR human reference total bone area was quantified using Image- Table 3. Blood cell composition of controls RNA (Stratagene, La Jolla, CA, USA), Pro plus software (Media Cybernetics). and patients which is composed of total RNA Controls Patients obtained from 10 human cell lines. (n 5 10) (n 5 10) Real-time PCR was performed on the Statistical analysis ABI PRISM 7000 (Applied Biosys- Statistics were performed using Graph- Whole blood tems). The reactions were performed Pad InStat version 3.00 (GraphPad Soft- T-cells 21.6 Æ 7.3 19.8 Æ 6.6 with 5 ng cDNA in a total volume of ware, San Diego, CA, USA). A paired B-cells 5.2 Æ 4.9 3.8 Æ 1.7 25 ml containing SYBR Green PCR Monocytes 7.4 Æ 4.0 7.1 Æ 2.6 Student’s t-test (two tailed) was per- Other (e.g. 62.4 Æ 10.5 65.4 Æ 15.1 Master Mix, consisting of SYBR Green formed for the comparisons between granulocytes I Dye, AmpliTaq Gold DNA polymer- patients and controls regarding number and NK cells) ase, dNTPs with dUTP instead of dTTP, of teeth present, blood cell counts, num- In culture passive reference and buffer (Applied ber of TRACP1 MNCs on tissue culture T-cells 51.1 Æ 11.2 49.3 Æ 9.6 Biosystems) and 300 nM of each primer. plates or bovine cortical bone slices, as B-cells 11.7 Æ 9.4 9.8 Æ 4.8 After an initial activation step of the well as comparisons between culturing Monocytes 19.3 Æ 12.8 19.1 Æ 9.8 AmpliTaq Gold DNA polymerase for on tissue culture plates or bovine corti- Other (e.g. NK 9.9 Æ 4.9 11.5 Æ 11.0 cells) 10 min. at 941C, 40 cycles of a two-step cal bone slices. PCR were run consisting of a denatura- Comparisons between the formation Values reported are mean percentage Æ SD. tion step at 951C for 30 s and annealing of TRACP1 MNCs of control and Percentage of each peripheral blood cell type and extension step at 601C for 1 min. patient groups under three different was determined by flow cytometry. For cells in Subsequently, the PCR products were culture conditions on the tissue culture culture, cell percentages were corrected for the subjected to melting curve analysis to plates were performed by means of percentage of granulocytes, because this cell type is lost during the density gradient separa- test if any unspecific PCR products were repeated-measures ANOVA and, if appro- generated. The PCR reactions of the priate, followed by Tukey’s multiple tion. No significant differences were found in blood cell composition between the control and different amplicons had equal efficien- comparison post-test. We constructed a cies. b2-microglobulin was used as the patient group (paired Student’s t-test, two- general linear model (ANCOVA) to ana- tailed) in both whole blood as well as in culture. the housekeeping gene. Expression lyse the influence of smoking, gender, of this gene was not affected by the sex and ethnicity on the formation of experimental conditions. Samples multinucleated cells. Within the patient/ X30% bone loss and 13.9 (SD Æ 5.5) were normalized for the expression control comparison, we have used the teeth with X50% bone loss, attesting of b2-microglobulin by calculating DCt values of multinuclearity as a fixed factor, to the fact that substantial perio- (Ctb2-microglobulin À Ctgene of interest), and and age, gender, ethnicity and smoking dontal destruction was present in these expression of the different genes is as co-variables. patients. Per definition, no bone loss was expressed as 2 À (DCt). Expression data The level of bone resorption was found in controls (Table 2). at 21 days of culture were expressed as analysed by the Mann–Whitney (U) test. fold increase from t 5 0. Leucocyte composition

Osteoclast resorption assay Results Statistical analyses revealed that there Population characteristics were no differences between the control PBMCs were cultured in 96-well plates and the patient group in numbers of at a density of 106 cells per well on The ethnicity, gender, age and smoking T-lymphocytes, B-lymphocytes, NK cells, bovine cortical long bone slices and habits data of the control and the patient monocytes or granulocytes (Table 3). analysed for resorption pits. PBMCs group are presented in Table 1. There were cultured as indicated in the pre- was no statistical difference between the vious paragraph. Cells were cultured for patient and the matched control group Formation of osteoclast-like cells on tissue culture plates 28 days, after which they were lysed regarding the number of teeth present. with demineralized water. Cell remnants Further, the patient group presented on TRACP1 MNCs formed in both the were removed mechanically after soni- average 26.0 (SD Æ 3.7) teeth with patient and the control group under the r 2008 Blackwell Munksgaard No claim to original US government works 572 Tjoa et al.

ent. First, higher numbers of TRACP1 MNCs were present when culturing control PBMCs in the presence of M-CSF (M-CSF; M-CSF1RANKL; Fig. 1b–e). In contrast, in the patient group, addition of M-CSF or M- CSF1RANKL did not increase the for- – M MR mation of TRACP1 MNCs (Fig. 1b–e). Overall Fig. 1 mainly appears to show a that the cells from patients are refractory p<0.01 to M-CSF stimulation. p<0.05 We further investigated the osteoclastic character of the cells cultured for p=0.04 3 weeks under the three conditions by 25 25 analysing expression of the osteoclast- 20 20 related genes cathepsin K and carbonic

2 p<0.05

2 anhydrase II. For all culture conditions, 15 15 these markers were 120–493 Â and 10 10 10–46 Â higher, respectively (Fig. 2), MNC/mm

MNC/mm than in PBMCs just after isolation. 5 5 M-CSF alone caused a signiﬁcant increase 0 0 compared with cultures without cyto- b C P C P C P c C P C P C P kines. Expression levels of cultures where M-CSF was added were comparable 10 p<0.05 10 with expression levels of cultures that p=0.02 had received both M-CSF and RANKL. 8 p=0.01 8

2 No differences in expression were 2 6 6 observed between controls and patients. p=0.03 p=0.01 4 4 MNC/mm MNC /mm MNC 2 2 Formation of osteoclast-like cells on 0 0 bovine cortical bone slices C P C P C P C P C P C P In contrast to the abundant TRACP1 d –MMR e –MMRMNCs generated when cultured on tis- 1 Fig. 1. Formation of tartrate-resistant acid phosphatase-positive (TRACP1) multinucleated sue culture plates, very few TRACP cells (MNCs) on tissue culture plates. (a) Micrographs taken after 21 days of culturing and MNCs were detected when cells were consecutive TRACP staining under the conditions: without cytokines (–), macrophage cultured on bone slices (Fig. 3). On colony-stimulating factor (M-CSF) (M) and M-CSF and receptor activator of nuclear bone, no differences were observed in factor-kB ligand (RANKL) (MR). Small triangle: multinucleated cell containing 3–5 nuclei; the formation of TRACP1 MNCs arrowhead: cell containing 6–10 nuclei; large triangle: multinucleated cell containing 410 between patient and control for any of nuclei. (b) Number of multinucleated cells from control (C) and patient (P) peripheral blood the culture conditions. mononuclear cells under the following culture conditions: without cytokines (–), M-CSF (M) We next compared the formation of and M-CSF1RANKL (MR). (c–e) Results as in (b), subdivided as number of nuclei per cell 1 category, number of TRACP1 MNC containing 3–5 nuclei (c), 6–10 nuclei (d) and 410 TRACP MNCs on tissue culture plas- nuclei (e). Data are represented as mean number of TRACP1 MNCs per mm2 Æ SEM tic and on bone slices (Table 4; derived (for culturing without cytokines and with M-CSF: n 5 9 per group; for culturing with from Figs 1 and 3). Numbers of M-CSF1RANKL: n 5 10 per group). TRACP1 MNCs were significantly higher on tissue culture plates than those on cortical bone slices. three culture conditions (Fig. 1a). The cell and 410 nuclei per cell (Fig. 1e), arrangement of the nuclei in large multi- were formed in the control group under nucleated cells was often in a circular culture conditions with M-CSF (M- Bone resorption configuration (Fig. 1a). In the presence CSF; M-CSF1RANKL). General linear of M-CSF in the control group, more model considering the formation of PBMCs were cultured for 28 days on TRACP1 MNCs with X3 nuclei were multinucleated cells as dependent factor bovine cortical bone slices under the found compared with the patient group. and age, gender, race and smoking as three culture conditions and resorption Culturing without the addition of cyto- co-variables indicated that none of these pits were visualized (Fig. 4). Bone kines or with M-CSF1RANKL did not variables contributed to the observed resorption was found exclusively when differ between the groups (Fig. 1b). No differences between patients and cells were cultured in the presence of M- differences in smaller (3–5 nuclei per controls. CSF and RANKL, albeit in a low num- cell; Fig. 1c) TRACP1 MNCs existed. When comparing the formation of ber of control (3 out of 10) and patient However, significantly larger TRACP1 TRACP1 MNCs under the three culture (3 out of 10) PBMC cultures, not allow- MNCs, with 6–10 nuclei (Fig. 1d) per conditions, two patterns became appar- ing any further statistical analysis. r 2008 Blackwell Munksgaard No claim to original US government works Osteoclast formation in periodontitis 573

Cathepsin K Carbonic anhydrase II Table 4. Excess of osteoclasts formed on plastic versus bone p<0.05 p <0.05 800 60 –MMR p<0.05 Controls 600 45 n nnn nnn Plastic/bone 4.0 22.5 17.0 Patients 400 30 Plastic/bone 8.2nnn 11.1nnn 10.6nnn

n 200 15 Not significant. nnnp 0.001. Relative expression Relative expression o 0 0 Plastic/bone: ratio of number of osteoclast-like – M MR – M MR – M MR – M MR cells formed on plastic compared with that on a Control Patientb Control Patient bone (n 5 9). –, M, MR: culture conditions without cytokines, Fig. 2. Expression of osteoclast-related genes. Expression of cathepsin K (a) and carbonic with M-CSF and with M-CSF1RANKL. anhydrase II (b) was assessed in peripheral blood mononuclear cell (PBMC) cultures from periodontitis patients and matched controls (n 5 5) who either received no cytokines (–), macrophage colony-stimulating factor (M-CSF) alone (M) or M-CSF and receptor activator of nuclear factor-kB ligand (MR). Data are represented as fold increase compared with with cytokines does not influence osteo- expression by PBMCs at isolation (mean increase Æ SEM). clast formation. Although cultures of controls and patients contained comparable numbers of CD141 cells, properties (e.g. RANKL expression) of B- and T-lymphocytes could differ between controls and patients. Precursors of healthy controls, however, can still be influenced by these cytokines. A similar insensitivity for priming was observed by Roato et al. (2005) who compared – M MR osteoclast formation from PBMCs from a healthy controls and cancer patients with secondary bone loss due to meta- 4 static disease. When PBMCs from cancer patients without bone-associated 3 metastases or from healthy controls were stimulated with increasing concen- 2 trations of RANKL, a dose-dependent increase in osteoclast-like cells was MNC/mm2 1 found, while this dose dependency was not seen when PBMCs of cancer 0 patients with bone breakdown were sti- C P C P C P mulated with increasing concentrations b –MMR of RANKL (Roato et al. 2005). Fig. 3. Formation of tartrate-resistant acid phosphatase-positive (TRACP1) multinucleated Previously, it was found in various cells (MNCs) on bovine cortical bone slices. (a) Micrographs taken after 21 days of culturing studies comparing osteoclast formation under the conditions: without cytokines (–), macrophage colony-stimulating factor (M-CSF) from PBMCs from various bone (M) and M-CSF and receptor activator of nuclear factor-kB ligand (RANKL) (MR). Nuclei destructive (osteopenic) diseases that were visualized with diamidino-2-phenylindole-dihydrochloride. Bar 5 50 mm. (b) Number 1 PBMCs from patients gave rise to of TRACP MNCs of controls (C) and patients (P) under the following culture conditions: 1 without cytokines (–), M-CSF (M) and M-CSF1RANKL (MR). Data are mean number of more TRACP MNCs than controls. TRACP1 MNCs per mm2 Æ SEM (for culturing without cytokines and with M-CSF: n 5 9 This was found, for instance, in post- per group; for culturing with M-CSF1RANKL: n 5 10 per group). No significant differences menopausal osteoporosis (D’Amelio et al. were observed for any of the culture conditions. 2005), bone tumors (Roato et al. 2005) and also in periodontitis (Brunetti et al. 2005). Also, in another study in osteoclast-like cells, whereas in case of which we assessed osteoclast formation Discussion controls, addition of these cytokines was from peripheral blood of chronic liver In the present study, we found that needed to induce a higher number of disease patients, we found that osteo- the formation of TRACP1 MNCs from MNCs. A possible explanation might be clast formation from patients with PBMCs of chronic periodontitis patients that in chronic periodontitis patients, chronic liver disease with osteoporosis may differ from the formation of these priming of the precursors of osteoclasts is higher than that in similar patients cells by blood cells of control subjects: already takes place in the peripheral without osteoporosis (Olivier et al. 2008). PBMCs from periodontitis patients did blood, possibly by CD41 cells present To our surprise, and not in line with the not require supplementation of M-CSF in the circulation (Brunetti et al. 2005); mentioned bone diseases with concomi- or M-CSF/RANKL for the formation of and therefore additional stimulation tant excessive bone degradation, more r 2008 Blackwell Munksgaard No claim to original US government works 574 Tjoa et al.

(Kukita et al. 1990). Especially long exposure time of human CD14-positive osteoclast precursors to TGF-b – possibly occurring on bone slices – results in downregulation of RANK, and subsequently osteoclast formation and resorption was inhibited (Karsdal et al. 2003). In conclusion, the current data strongly suggest that peripheral blood of chronic periodontitis patients contains relatively mature precursors of osteo- a b clasts that have the capacity to become osteoclast-like cells, also without stimu- Fig. 4. Bone resorption micrographs taken after 28 days of culturing with macrophage lation with added M-CSF and/or colony-stimulating factor and receptor activator of nuclear factor-kB ligand: (a) no bone RANKL. Possibly, because these pre- resorption, (b) resorption pits are present (arrows) on the bone surface. Scale bar 5 30 mm. cursors are relatively mature, they need less triggering at the inflammatory site to further differentiate into osteoclasts. TRACP1 MNCs were formed by from monocytes or monocyte progenitor control PBMCs than by PBMCs from cells, but both cell types are differently severe periodontitis patients. Note- induced by cytokines. Our RT-PCR worthy, these higher numbers of further reveals that osteoclast markers References TRACP1 MNCs were only observed cathepsin K and carbonic anhydrase II when cultured with M-CSF or M-CSF/ were increased (120–493 and 10– Anderson, J. M. (2000) Multinucleated giant Â cells. Current Opinion in Hematology 7, RANKL, suggesting that osteoclast pre- 46 Â, respectively) at the time point 40–47. cursors of healthy controls are more sus- where multinucleated cells were ana- Bizzarro, S., Van der Velden, U., ten Heggeler, ceptible to M-CSF or M-CSF/RANKL. lysed, indicating that these cell cultures J. M., Leivadaros, E., Hoek, F. J., Bakker, Our results are in conflict with the contained some characteristics typical S. J., Gans, R. O., Gerdes, V. E., ten Cate, H. results of Brunetti et al. (2005), who of osteoclasts. Remarkably, this increase & Loos, B. G. (2007) Periodontitis is observed a higher number of TRACP1 was seen in cell cultures where no characterized by elevated PAI-1 activity. MNCs and bone resorption by PBMCs cytokines were added and in cell Journal of Clinical Periodontology 34, of periodontitis patients compared with cultures containing M-CSF (M-CSF or 574–580. Borrell, L. N. & Papapanou, P. N. (2005) 1 controls. In the latter study, increased M-CSF RANKL). M-CSF alone Analytical epidemiology of periodontitis. osteoclast formation in periodontitis caused an upregulation, while extra Journal of Clinical Periodontology 32 patients was due to increased RANKL supplementation with RANKL did not (Suppl. 6), 132–158. and TNF-a production by peripheral T- further upregulate osteoclast markers. Brunetti, G., Colucci, S., Pignataro, P., lymphocytes. Although these character- Supplementation with RANKL also Coricciati, M., Mori, G., Cirulli, N., Zallone, istics were not investigated in our did not increase the number of multi- A., Grassi, F. R. & Grano, M. (2005) T cells patient and control groups, a possible nucleated cells, but RANKL was support osteoclastogenesis in an in vitro explanation for the discrepancies obligatory for resorption. model derived from human periodontitis between the latter study and the current In accordance with other studies, patients. Journal of Periodontology 76, 1 1675–1680. investigation could be due to different where the number of TRACP MNCs Cheleuitte, D., Mizuno, S. & Glowacki, J. matching procedures. Both age and sex on cortical bone slices was counted and (1998) In vitro secretion of cytokines by influence the osteoclastogenic potential compared with cells cultured on plastic, human bone marrow: effect of age and estro- 1 of peripheral blood cells (Jevon et al. we found a lower number of TRACP gen status. Journal of Clinical Endocrinology 2002). In our study, the patient group MNCs on bone (Faust et al. 1999, de and Metabolism 83, 2043–2051. consisted of untreated subjects with Vries et al. 2006). Moreover, these D’Amelio, P., Grimaldi, A., Pescarmona, G. P., generalized severe chronic periodontitis, osteoclast-like cells contained fewer Tamone, C., Roato, I. & Isaia, G. (2005) and in order to control known confound- nuclei compared with cells cultured on Spontaneous osteoclast formation from per- ing factors, we meticulously matched tissue culture plates. Consistent with ipheral blood mononuclear cells in postmenopausal osteoporosis. FASEB Journal 19, the healthy controls for age, gender, this is the low number of patients and 410–412. ethnicity and smoking. Moreover, con- controls where bone resorption was de Vries, T. J., Schoenmaker, T., Wattanaroon- trol and patient blood samples were observed on bovine cortical bone slices. wong, N., van den Hoonaard, M., Nieuwen- processed simultaneously; thus, patient It might well be that inhibitory factors huijse, A., Beertsen, W. & Everts, V. (2006) and control were matched within each present in the cortical bone could cause Gingival fibroblasts are better at inhibiting experiment. this effect. One of these factors could be osteoclast formation than periodontal Morphologically, the circular orienta- transforming growth factor-b (TGF-b), ligament fibroblasts. Journal of Cellular tion of nuclei within a multinucleated which is found in high levels in bone. Biochemistry 98, 370–382. Faust, J., Lacey, D. L., Hunt, P., Burgess, T. L., cell (Fig. 1a) is normally not seen in Osteoclasts can activate the latent form Scully, S., Van, G., Eli, A., Qian, Y. & osteoclasts and is suggestive that these of TGF-b (Oursler 1994), and the acti- Shalhoub, V. (1999) Osteoclast markers cells are in fact related to Langhans’ vated form of TGF-b could act as a accumulate on cells developing from human giant cells (Anderson 2000). Like osteo- natural inhibitor of osteoclast activity peripheral blood mononuclear precursors. clasts, Langhans’ giant cells are derived by inhibiting osteoclast formation Journal of Cellular Biochemistry 72, 67–80. r 2008 Blackwell Munksgaard No claim to original US government works Osteoclast formation in periodontitis 575

Fujikawa, Y., Quinn, J. M., Sabokbar, A., Loe, H., Anerud, A., Boysen, H. & Morrison, E. Simonet, W. S., Lacey, D. L., Dunstan, C. R., McGee, J. O. & Athanasou, N. A. (1996) (1986) Natural history of periodontal disease Kelley, M., Chang, M. S., Luthy, R., Nguyen, The human osteoclast precursor circulates in in man. Rapid, moderate and no loss of H. Q., Wooden, S., Bennett, L., Boone, T., the monocyte fraction. Endocrinology 137, attachment in Sri Lankan laborers 14 to 46 Shimamoto, G., DeRose, M., Elliott, R., 4058–4060. years of age. Journal of Clinical Perio- Colombero, A., Tan, H. L., Trail, G., Sulli- Hodge, J. M., Kirkland, M. A. & Nicholson, G. dontology 13, 431–445. van, J., Davy, E., Bucay, N., Renshaw-Gegg, C. (2007) Multiple roles of M-CSF in human Olivier, B., Schoenmaker, T., Mebius, R. E., L., Hughes, T. M., Hill, D., Pattison, W., osteoclastogenesis. Journal of Cellular Bio- Everts, V., Mulder, C. J., Van Nieuwkerk, K. Campbell, P., Sander, S., Van, G., Tarpley, J., chemistry 102, 759–768. M. J., de Vries, T. J. & Van der Merwe, S. W. Derby, P., Lee, R. & Boyle, W. J. (1997) Jevon, M., Sabokbar, A., Fujikawa, Y., Hiraya- (2008) Increased osteoclast formation and Osteoprotegerin: a novel secreted protein ma, T., Neale, S. D., Wass, J. & Athanasou, activity by peripheral blood mononuclear involved in the regulation of bone density. N. A. (2002) Gender- and age-related differ- cells in chronic liver disease patients with Cell 89, 309–319. ences in osteoclast formation from circulating osteopenia. Hepatology 47, 259–267. Suda, T., Takahashi, N., Udagawa, N., Jimi, E., precursors. Journal of Endocrinology 172, Oursler, M. J. (1994) Osteoclast synthesis and Gillespie, M. T. & Martin, T. J. (1999) 673–681. secretion and activation of latent transform- Modulation of osteoclast differentiation and Jimi, E., Nakamura, I., Duong, L. T., Ikebe, T., ing growth factor beta. Journal of Bone and function by the new members of the tumor Takahashi, N., Rodan, G. A. & Suda, T. Mineral Research 9, 443–452. necrosis factor receptor and ligand families. (1999) Interleukin 1 induces multinucleation Page, R. C., Offenbacher, S., Schroeder, H. E., Endocrine Reviews 20, 345–357. and bone-resorbing activity of osteoclasts in Seymour, G. J. & Kornman, K. S. (1997) Tanaka, S., Takahashi, N., Udagawa, N., the absence of osteoblasts/stromal cells. Advances in the pathogenesis of perio- Tamura, T., Akatsu, T., Stanley, E. R., Kur- dontitis: summary of developments, clinical okawa, T. & Suda, T. (1993) Macrophage Experimental Cell Research 247, 84–93. implications and future directions. Perio- colony-stimulating factor is indispensable for Karsdal, M. A., Hjorth, P., Henriksen, K., dontology 2000 14, 216–248. both proliferation and differentiation of Kirkegaard, T., Nielsen, K. L., Lou, H., Piper, K., Boyde, A. & Jones, S. J. (1992) The osteoclast progenitors. Journal of Clinical Delaisse´, J. M. & Foged, N. T. (2003) Trans- relationship between the number of nuclei of Investigation 91, 257–263. forming growth factor-b controls human an osteoclast and its resorptive capability Udagawa, N., Takahashi, N., Akatsu, T., osteoclastogenesis through the p38 MAPK in vitro. Anatomy and Embryology 186, Tanaka, H., Sasaki, T., Nishihara, T., Koga, and regulation of RANK expression. Journal 291–299. T., Martin, T. J. & Suda, T. (1990) Origin of of Biological Chemistry 278, 44975–44987. Ritchlin, C. T., Haas-Smith, S. A., Li, P., Hicks, osteoclasts: mature monocytes and macro- Kinane, D. F. (1999) Periodontitis modiﬁed by D. G. & Schwarz, E. M. (2003) Mechanisms phages are capable of differentiating into systemic factors. Annals of Periodontology 4, of TNF-alpha- and RANKL-mediated osteo- osteoclasts under a suitable microenviron- 54–64. clastogenesis and bone resorption in psoriatic ment prepared by bone marrow-derived stro- Kukita, A., Bonewald, L., Rosen, D., Seyedin, arthritis. Journal of Clinical Investigation mal cells. Proceedings of the National S., Mundy, G. R. & Roodman, G. D. (1990) 111, 821–831. Academy of Sciences of the United States of Osteoinductive factor inhibits formation of Roato, I., Grano, M., Brunetti, G., Colucci, S., America 87, 7260–7264. human osteoclast-like cells. Proceedings Mussa, A., Bertetto, O. & Ferracini, R. of the National Academy of Sciences (2005) Mechanisms of spontaneous osteo- of the United States of America 87, clastogenesis in cancer with bone involve- 3023–3026. ment. FASEB Journal 19, 228–230. Lacey, D. L., Timms, E., Tan, H. L., Kelley, M. Roodman, G. D. (1996) Advances in bone Address: J., Dunstan, C. R., Burgess, T., Elliott, R., biology: the osteoclast. Endocrine Reviews Teun de Vries Colombero, A., Elliott, G., Scully, S., Hsu, 17, 308–332. Experimental Periodontology H., Sullivan, J., Hawkins, N., Davy, E., Shalhoub, V., Faust, J., Boyle, W. J., Dunstan, c.o. Department of Oral Cell Biology Capparelli, C., Eli, A., Qian, Y. X., Kaufman, C. R., Kelley, M., Kaufman, S., Scully, S., ACTA, Universiteit van Amsterdam and Vrije S., Sarosi, I., Shalhoub, V., Senaldi, G., Guo, Van, G. & Lacey, D. L. (1999) Osteoprote- Universiteit J., Delaney, J. & Boyle, W. J. (1998) Osteo- gerin and osteoprotegerin ligand effects on Van der Boechorststraat 7 protegerin ligand is a cytokine that regulates osteoclast formation from human peripheral 1081 BT Amsterdam osteoclast differentiation and activation. Cell blood mononuclear cell precursors. Journal The Netherlands 93, 165–176. of Cellular Biochemistry 72, 251–261. E-mail: [email protected]

Clinical Relevance destructive diseases. Here, we investi- Practical implications: Osteoclast Scientiﬁc rationale for the study: Peri- gated osteoclast formation from per- formation may occur more rapidly pheral blood contains precursor cells ipheral blood of periodontitis patients. from blood of periodontitis patients. of osteoclasts, the bone-degrading cell Principle ﬁndings: Peripheral blood- The precursors of these cells seem responsible for alveolar bone degrada- derived osteoclast formation in already primed in the circulation tion in periodontitis. Osteoclast forma- periodontitis patients occurs without before entry of the site of periodontal tion from precursors in peripheral induction by the cytokines M-CSF bone breakdown. blood is altered in a variety of bone and RANKL.

r 2008 Blackwell Munksgaard No claim to original US government works