Macrophage-elicited osteoclastogenesis in response to Brucella abortus infection requires TLR2/MyD88-dependent TNF- production

Transcription

1 Article Macrophage-elicited osteoclastogenesis in response to Brucella abortus infection requires TLR2/MyD88-dependent TNF- production M. Victoria Delpino,* Paula Barrionuevo,*, Gilson Costa Macedo, Sergio Costa Oliveira, Silvia Di Genaro, Romina Scian,* M. Cruz Miraglia, Carlos A. Fossati,* Pablo C. Baldi,* and Guillermo H. Giambartolomei*,,1 *Instituto de Estudios de la Inmunidad Humoral (CONICET), Facultad de Farmacia y Bioquímica, and Laboratorio de Inmunogenética, Hospital de Clínicas José de San Martín, Facultad de Medicina, Universidad de Buenos Aires, Buenos Aires, Argentina; Department of Biochemistry and Immunology, Institute of Biological Sciences, Federal University of Minas Gerais, Belo Horizonte-Minas Gerais, Brazil; and Instituto de Investigaciones Biológicas-San Luis (CONICET), Facultad de Química, Bioquímica y Farmacia, Universidad Nacional de San Luis, San Luis, Argentina RECEIVED APRIL 5, 2011; REVISED SEPTEMBER 29, 2011; ACCEPTED OCTOBER 14, DOI: /jlb ABSTRACT Osteoarticular complications are common in human brucellosis, but the pathogenic mechanisms involved are largely unknown. In this manuscript, we described an immune mechanism for inflammatory bone loss in response to infection by Brucella abortus. We established a requirement for MyD88 and TLR2 in TNF- -elicited osteoclastogenesis in response to B. abortus infection. CS from macrophages infected with B. abortus induced BMM to undergo osteoclastogenesis. Although B. abortus-infected macrophages actively secreted IL- 1, IL-6, and TNF-, osteoclastogenesis depended on TNF-, as CS from B. abortus-infected macrophages failed to induce osteoclastogenesis in BMM from TNFRp55 / mice. CS from B. abortus-stimulated MyD88 / and TLR2 / macrophages failed to express TNF-, and these CS induced no osteoclast formation compared with that of the WT or TLR4 / macrophages. Omp19, a B. abortus lipoprotein model, recapitulated the cytokine production and subsequent osteoclastogenesis induced by the whole bacterium. All phenomena were corroborated using human monocytes, indicating that this mechanism could play a role in human osteoarticular brucellosis. Our results indicate that B. abortus, through its lipoproteins, may be involved in bone resorption through the pathological induction of osteoclastogenesis. J. Leukoc. Biol. 91: ; Abbreviations: / knockout, BMM bone marrow-derived monocyte(s), Ckb cytoplasmic creatine kinase, CS culture supernatant(s), HKBA heat-killed Brucella abortus, IDEHU Instituto de Estudios de la Inmunidad Humoral (CONICET), KO knockout, L-Omp19 lipidated outer membrane protein 19, MOI multiplicities of infection, Omp19 outer membrane protein 19, Pam 3 Cys synthetic lipohexapeptide tripalmitoyl-s-glyceryl-cys-ser-lys4-oh, TRAP tartrate-resistant acid phosphatase, TSA tryptose soy agar, U-Omp19 unlipidated outer membrane protein 19 Introduction Human brucellosis is a protean disease with a diversity of clinical signs and symptoms [1, 2]. It can affect virtually any organ or system, causing focal forms that account for 30% of the reported cases [3]. Osteoarticular involvement is the most common presentation of localized disease, which may affect peripheral joints, sacroiliac joints, or the spine [4 8]. As occurs with other joint infections, persistence of the disease in the infected joint may result in tissue damage. Joint damage is often associated with the presence of Brucella organisms in these joints, as evidence by the recovery of bacterium from the synovial fluid of patients [9, 10], demonstrating that the skeletal symptoms are caused by a true infection of the joint. Loss of bone is a serious complication of localized bacterial infection of bones or the adjacent tissue. When present within or near the bone, bacterial infections are associated with an inflammatory process that results in localized osteolysis; and osteoarticular brucellosis proved not to be the exception [11 15]. Under normal physiological conditions, bone is resorbed periodically by osteoclasts, whereas new bone is formed by osteoblasts [16], which regulate osteoclastic bone resorption, involving differentiation of new osteoclasts and activation of mature osteoclasts [16]. The differentiation of new osteoclasts is dependent on the balance between the expression of RANKL and that of its decoy receptor, osteoprotegerin, in osteoblasts [16, 17]. During chronic inflammatory bone diseases, cellular recruitment contributes to bone loss. RANKL and proinflammatory cytokines, such as TNF-, IL-1, and IL-6, have been shown to be important for disease progression and bone 1. Correspondence; IDEHU, Facultad de Farmacia y Bioquímica, Universidad de Buenos Aires, Junín 956 4to. Piso, 1113 Buenos Aires, Argentina. ggiambart@ffyb.uba.ar /12/ Society for Leukocyte Biology Volume 91, February 2012 Journal of Leukocyte Biology 285

2 loss [18 22]. T cells and B cells have been reported to contribute to the acceleration of bone resorption through the production of cytokines and RANKL [23 25]. Macrophages in inflamed tissues cannot only differentiate to osteoclasts [26], but they also appear to accelerate bone resorption through the production of proinflammatory cytokines [25]. Despite the diversity of signs and symptoms of brucellosis, the inflammatory trait of the disease, present in the acute and chronic phases of human brucellosis [1], together with the detection of bacterium in the inflamed tissues, suggests that Brucella stimulates a robust inflammatory response at sites of localization. In fact, we and others [27 35] have demonstrated that Brucella abortus infection elicits in a variety of cell types a proinflammatory cytokine response. The most likely molecular basis of the inflammatory effectiveness of B. abortus has become apparent recently. Although B. abortus possesses LPS, the generic endotoxin of Gram-negative organisms, it may derive its inflammatory capacity from lipoproteins. Differently from the LPS from the Enterobacteriaceae, Brucella LPS has thus far been found to be virtually devoid of proinflammatory activity [36]. Moreover, we have shown recently that the production of proinflammatory cytokines by different cell types of the innate immunity is induced by Brucella lipoproteins rather than LPS [27 31]. Bacterial lipoproteins are powerful inflammatory molecules that are capable, for example, of inducing inflammatory cytokines, such as IL-6, IL-1, and TNF-. The genome of B. abortus contains no less than 80 genes encoding putative lipoproteins [37]. As it is the lipoprotein lipid moiety shared by all bacterial lipoproteins that endows this type of molecule with inflammatory properties [38], the inflammatory potential of Brucella must be considerable. In a recent work, we have demonstrated that Brucella spp. can infect and survive within human osteoblasts and that this infection elicits the secretion of proinflammatory cytokines and chemokines that might be involved in the osteoarticular manifestations of brucellosis. Such response was amplified further by subsequent interactions between osteoblasts and monocytes in the face of B. abortus infection [39]. To establish a link between B. abortus infection on one hand and loss of bone on the other, we tested the hypothesis that the infection might create a microenvironment that would promote the generation of osteoclasts, the only cells known so far to be able to degrade bone. For that purpose, we examined the bacterial determinants and the immune mechanisms involved in B. abortus-infected, macrophageinduced osteoclastogenesis. Using murine macrophages, we detailed a mechanism by which TLR2/MyD88-mediated signaling to B. abortus stimulates a potent proinflammatory response that drives BMM to undergo osteoclastogenesis via TNF- induction. Furthermore, we demonstrated, using Omp19 as a lipoprotein stimulant model, that B. abortus lipoproteins are the TLR2 ligands that induce such effect. The phenomenon was corroborated using human monocytes, indicating that this mechanism could play a role in human brucellosis. MATERIALS AND METHODS Ethics statement Human PBMCs were isolated from healthy blood donors in accordance with the guidelines of the Ethical Committee of the IDEHU Institute (Buenos Aires, Argentina). A written consent was obtained from all blood donors. All animal procedures were performed according to the rules and standards for the use of laboratory animals of NIH. Animal experiments were approved by the Ethical Committee of the IDEHU. Animals Six- to 8-week-old female MyD88, TLR2, and TLR4 KO mice (originally provided by Shizuo Akira, Osaka University, Osaka, Japan), TNFRp55 KO mice [40], and WT mice (all of them on the C57BL/6 background) were provided by Federal University of Minas Gerais (Belo Horizonte, Brazil) or University of La Plata (Argentina). Animals were housed in groups of five, under controlled temperature (22 C 2 C) and artificial light under a 12-h cycle period. Mice were kept under specific pathogen-free conditions in positive-pressure cabinets and provided with sterile food and water ad libitum. Bacteria B. abortus S2308 was grown overnight in 10 ml tryptic soy broth (Merck, Buenos Aires, Argentina) with constant agitation at 37 C. Bacteria were harvested by centrifugation for 15 min at 6000 g at 4 C and washed twice in 10 ml PBS. The numbers of bacteria in stationary-phase cultures were determined by comparing the ODs at 600 nm with a standard curve [27]. When indicated, Brucella organisms were washed five times for 10 min each in sterile PBS, heat-killed at 70 C for 20 min (HKBA), aliquoted, and stored at 70 C until they were used. Absence of B. abortus viability after heat killing was verified by the absence of bacterial growth on TSA. Lipoproteins and LPS B. abortus L-Omp19 and U-Omp19 were obtained as described [27]. L-Omp19 is processed correctly and fully acylated with palmitoyl esters in Brucella spp. and when expressed in Escherichia coli [27, 41], indicating that the pathway for lipoprotein maturation is functionally shared between both bacteria. Both recombinant proteins contained 0.25 endotoxin U/ g protein, as assessed by Limulus amoebocyte lysates (Associates of Cape Cod, East Falmouth, MA, USA). Protein concentration was determined by the BCA method (Pierce, Rockford, IL, USA) using BSA as standard. B. abortus S2308 LPS and E. coli O111k58H2 LPS were provided by Ignacio Moriyon (University of Navarra, Pamplona, Spain). Pam 3 Cys was purchased from Boehringer Mannheim (Indianapolis, IN, USA). Cells and media All experiments were performed at 37 C in 5% CO 2 atmosphere in -minimum essential medium, supplemented with 2 mm l-glutamine, 10% heatinactivated FBS (Gibco-BRL, Life Technologies, Grand Island, NY, USA), 100 U penicillin/ml, and 100 g streptomycin/ml (complete medium). Thioglycolate-elicited peritoneal macrophages were isolated as described previously [27] from MyD88, TLR2, or TLR4 KO mice and C57BL/6 WT mice. PBMCs were obtained by Ficoll-Hypaque (GE Healthcare Bio-Sciences, Uppsala, Sweden) gradient centrifugation from human blood collected from healthy adult individuals. All volunteers gave informed consent prior to participation in the study. Monocytes were obtained after centrifugation of PBMC on a Percoll (GE Healthcare Bio-Sciences) gradient and were resuspended in complete medium. The viability of cells was 95% in all of the experiments, as measured by a trypan blue exclusion test. Infection Murine peritoneal macrophages and human monocytes were cultured in 24-well plates at a density of cells/well in complete medium with- 286 Journal of Leukocyte Biology Volume 91, February

3 Delpino et al. Brucella abortus induces osteoclastogenesis out the addition of antibiotics. Cells were infected with B. abortus S2308 at different MOI for 2 h in medium containing no antibiotics. Cells were washed extensively to remove uninternalized bacteria, and infection was maintained for an additional 24 h in the presence of antibiotics (100 g/ml gentamicin and 50 g/ml streptomycin) to kill remaining extracellular bacteria. To monitor Brucella intracellular replication, infected cells were washed and lysed at several intervals postinfection with 0.1% (vol/vol) Triton X-100. The number of intracellular viable bacteria (in CFU/well) was determined by plating serial dilutions onto TSA plates. The number of bacteria internalized after 24 h of infection into peritoneal macrophages was as follows: MOI 25 ( bacteria), MOI 50 (15, bacteria), MOI 100 (27, bacteria). B. abortus-stimulated macrophage/monocyte assays Peritoneal macrophages from C57BL/6 WT mice; MyD88, TLR2, or TLR4 KO mice; or human monocytes were infected with B. abortus at various MOI or stimulated for 24 h with different concentrations of HKBA, L-OMP19, U-OMP19 (1000 ng/ml), B. abortus LPS (1000 ng/ml), E. coli LPS (10 ng/ml), or Pam 3 Cys (50 ng/ml). CS were harvested, sterilized by filtration through a m nitrocellulose filter, and stored at 70 C until used for ELISA or for osteoclast formation assays. Osteoclast formation assay BMM were induced to undergo osteoclastogenesis as described previously [42]. Briefly, BM cells from C57BL/6 or TNFRp55 KO mice were cultured in complete medium containing 5 ng/ml murine rm-csf (R&D Systems, Minneapolis, MN, USA) for 12 h in 24-well plates. Nonadherent cells were harvested and cultured with 30 ng/ml M-CSF in 24-well plates for an additional 24 h. Nonadherent cells were washed out, and adherent cells were collected and used as BMM ( cells/0.5 ml/well), which were seeded onto glass coverslips in 24-well plates for 6 days and cultured in complete medium containing 30 ng/ml M-CSF and 0.2 ml CS from murine peritoneal macrophages infected with B. abortus or stimulated with Brucella antigens. As positive controls of osteoclast formation, BMM cultures received 50 ng/ml murine TNF- or human RANKL. On Day 3, the culture media and all reagents were replaced. To identify osteoclasts, cells were fixed in 4% paraformaldehyde and stained for TRAP (Sigma-Aldrich, St. Louis, MO, USA). TRAP-positive, multinucleated (more than three nuclei) cells were defined as osteoclasts, and the number was determined by microscopic counts. We also determined the ability of direct B. abortus infection to stimulate osteoclast formation from BMM. For these experiments, mouse BMM ( cells/0.5 ml/well) were cultured for 6 days in complete medium containing 30 ng/ml murine M-CSF and B. abortus (MOI 100) or 0.2 ml culture medium from B. abortus, as described above. To obtain human osteoclasts from human monocytes, the same procedure was followed using recombinant cytokines and growth factors from human origin. Assessment of vitronectin receptor expression Vitronectin receptor (CD51) expression was determined by fluorescent microscopy using a PE-labeled anti-mouse CD51 (BioLegend, San Diego, USA). CD51-positive, multinucleated (more than three nuclei) cells were defined as osteoclasts. RT-PCR analysis Total RNA was extracted with Trizol (Invitrogen Carlsbad, CA, USA), and cdna was obtained by reverse transcription, according to the manufacturer s instructions. Primer sequences for RANK, cathepsin K, and braintype Ckb have been described elsewhere [43 45]. An optimal number of PCR cycles was determined initially by using a variable number of cycles to identify a linear range of amplification for each cytokine transcript. Thirty cycles (RANK), 35 cycles (Ckb), or 20 cycles (cathepsin K and -actin) of denaturation (94 C), annealing (58 C), and elongation (72 C) were performed in a MJ Mini thermocycler (Bio-Rad, Hercules, CA, USA). Positive and negative controls were included in each assay to confirm that only cdna PCR products were detected and that none of the reagents was contaminated with cdna or PCR products. Cytokines and RANKL assays Concentrations of IL-1, IL-6, TNF- (BD PharMingen, San Diego, CA, USA), RANKL (R&D Systems), and TGF- (ebioscience, San Diego, CA, USA) in CS were determined by using commercially available ELISA kits. Pit formation assay BMM from C57BL/6 or TNFRp55 KO mice or human monocytes ( cells/0.25 ml/well) were plated on dentine disks (BD BioCoat Osteologic (BD Biosciences, San Diego, CA, USA) in 96-well culture dishes and cultured in complete medium containing 100 l CS from B. abortus-, L-Omp19-, or U-Omp19-stimulated human monocytes or peritoneal macrophages from WT mice in the presence of M-CSF (30 ng/ml) for 6 days. Media and all reagents were replaced every day to avoid acidification of medium. After culture with cells, dentine discs were washed with 1 M NH 4 OH to remove adherent cells. After rinsing with water, dentine discs were visualized by light microscopy to determine and enumerate resorption lacunae. Evaluation of osteoclast formation in an in vivo model Six- to 8-week-old C57BL/6 or TLR2 KO mice were anesthetized with ketamine chlorhydrate (150 mg/kg) and xylazine (15 mg/kg) and then injected intra-articularly in the knee joint with 50 l HKBA ( bacteria), L-Omp19 (500 ng), U-Omp19 (500 ng), or vehicle (PBS). E. coli LPS (500 ng) and Pam 3 Cys (50 ng) were used as a positive controls. Mice were killed at 5 days postadministration, and whole knee joints were removed. Knee joints were fixed for 4 days using 4% paraformaldehyde, followed by decalcification in 10% EDTA in 1 mm Tris HCl (ph 7.4) for up to 2 weeks at 4 C. Decalcified specimens were processed for paraffin embedding [46]. Histological sections of the proximal tibiae (7 m) were stained for TRAP as described. TRAP-positive, multinucleated (more than three nuclei) cells were defined as osteoclasts. Statistical analysis Statistical analysis was performed with one-way ANOVA, followed by Post Hoc Tukey test using GraphPad Prism 4.0 sofware. Data are represented as mean sd. RESULTS CS from B. abortus-stimulated murine macrophages induce BMM-derived osteoclastogenesis Osteoclasts play an important role in bone resorption and originate from the fusion of precursors belonging to the monocyte/macrophage lineage in the BM [47, 48]. This process may involve soluble mediators from inflammatory cells, such as macrophages, in conjunction with M-CSF [42]. To determine if factors produced by B. abortus-infected macrophages could induce osteoclast formation from BMM, these cells were stimulated with M-CSF in conjunction with CS from peritoneal macrophages infected with B. abortus, and ex vivo osteoclastogenesis was determined by the generation of multinucleated, vitronectin receptor- and TRAP-expressing cells. TNF- and RANKL were used as a positive control. The formation of osteoclast-like cells was induced by CS from B. abortus-infected macrophages but not by those of uninfected macrophages. The magnitude of the osteoclastogenesis induced by CS was Volume 91, February 2012 Journal of Leukocyte Biology 287

4 related directly to the MOI used to infect macrophages (Fig. 1). Confirming ex vivo osteoclastogenesis, the up-regulation of the expression of three genes involved in osteoclast differentiation, such as RANK, cathepsin K, and brain-type Ckb, was corroborated further by RT-PCR in BMM cultures stimulated with CS from Brucella-infected macrophages (Fig. 1D). As expected, RANKL and TNF- induced TRAP- and vitronectin receptor-expressing cells, as well as up-regulation of the three genes involved in osteoclast differentiation (Fig. 1). The direct effect of bacteria on osteoclastogenesis was also examined. Culture of BMM with B. abortus alone was unable to induce osteoclastogenesis (data not shown), indicating that the bacterium itself does not induce osteoclastogenesis. Taken together, these results indicate that CS from B. abortus-infected peritoneal macrophages can promote functional osteoclast formation from BMM. B. abortus-infected murine macrophages produce the proinflammatory cytokines TNF-, IL-6, and IL-1 but not RANKL or TGF- Differentiation of BMM into osteoclasts has been demonstrated previously with RANKL and TNF- in conjunction with M-CSF [49, 50]. Yet, although the ability of B. abortus to induce the secretion of proinflammatory cytokines has been reported previously in macrophages and in a variety of other cell types [27 33, 51], their role in Brucella-induced osteoclastogenesis has not been determined. Thus, the production of TNF-, IL-1, IL-6, and soluble RANKL in CS from peritoneal macrophages in response to B. abortus infection was examined. Infection resulted in a significant (P 0.05) secretion of IL-6, IL-1, and TNF- in a MOI-dependent manner (Fig. 2A). On the contrary, RANKL pro- Figure 1. CS from B. abortus-stimulated murine macrophages induced BMM-derived osteoclastogenesis. BMM cells were stimulated with CS from peritoneal macrophages, infected or not (Control) with B. abortus for 24 h in conjunction with M-CSF. TNF- and RANKL were used as positive controls. After 5 days, osteoclastogenesis was determined by the generation of multinucleated TRAP (A and C)- and vitronectin (B)-expressing cells. Representative digital images were taken by light microscopy (A) or fluorescent microscopy (B), and TRAP-positive, multinucleated cells were identified and counted (C). Bars express the mean sem of duplicates. Data shown are from a representative experiment of five performed. **P 0.01; ***P versus control. Up-regulation of the expression of three genes involved in osteoclast differentiation RANK, cathepsin K, and brain-type Ckb was corroborated by RT-PCR (D). 288 Journal of Leukocyte Biology Volume 91, February

5 Delpino et al. Brucella abortus induces osteoclastogenesis Figure 2. Murine macrophages produce proinflammatory cytokines in response to B. abortus. Peritoneal macrophages were infected with B. abortus at different MOI (25 100) or stimulated with HKBA ( bacteria/ml; A) and E. coli (Ec) LPS (10 ng/ml), B. abortus (Ba) LPS (1000 ng/ml), L-Omp19 (10 ng/ml, 100 ng/ml, or 1000 ng/ml), U-Omp19 (1000 ng/ml), or Pam 3 Cys (50 ng/ml) or nonstimulated (B). TNF-, IL-6, and IL-1 were determined in CS by ELISA. Bars express the mean sem of duplicates. Data shown are from a representative experiment of five performed. *P 0.05; **P 0.01; ***P versus control. duction was not detected when macrophages were infected with B. abortus (data not shown). To test whether viable bacteria were necessary to induce a proinflammatory response upon macrophage infection, the ability of HKBA to induce the secretion of TNF-, IL-1, IL-6, and soluble RANKL was also examined. The production of TNF-, IL-1, and IL-6 was markedly enhanced in CS from murine macrophages that were stimulated with HKBA when compared with the unstimulated cells. Cytokine production was a function of the amount of bacteria present in the culture. A significant (P 0.05) cytokine secretion was detected in cultures containing between and bacteria/ml (Fig. 2A), a similar bacterial concentration to that able to elicit the secretion of mediators with live bacteria. HKBA was unable to induce the secretion of RANKL (data not shown). As it has been reported that TGF- can stimulate osteoclastogenesis [25], TGF- levels were also measured in CS. TGF- levels from B. abortus-infected cells or stimulated with HKBA were not significantly different (P 0.05) from those of unstimulated cells (data not shown). These results indicate that macrophages infected with B. abortus secrete proinflammatory cytokines but do not produce soluble RANKL or TGF-. They also suggest that this inflammatory response could be induced by a structural component of B. abortus. CS from B. abortus L-Omp19-stimulated murine macrophages induce BMM-derived osteoclastogenesis Experiments were then conducted to evaluate the contribution of B. abortus LPS to the production of inflammatory mediators and the osteoclastogenesis elicited by CS from B. abortus-infected macrophages. For this, BMM were stimulated with CS from peritoneal macrophages treated with B. abortus LPS for 24 h in conjunction with M-CSF, and ex vivo osteoclastogenesis was determined by the generation of TRAP-expressing cells and by the expression of RANK, cathepsin K, and Ckb genes. TNF-, IL-1, and IL-6 concentrations in macrophage CS were determined by ELISA. TNF- or CS from E. coli LPS-stimulated macrophages were used as positive controls. Although CS from HKBA-treated macrophages induced a significant (P 0.05) osteoclast formation from BMM, CS from macrophages stimulated with highly purified B. abortus LPS (at concentrations comparable with the ones estimated to be present in the concentration of bacteria used [52]) were unable to elicit osteoclastogenesis (Fig. 3). Likewise, B. abortus LPS was un- Volume 91, February 2012 Journal of Leukocyte Biology 289

6 able to stimulate the secretion of inflammatory mediators from macrophages (Fig. 2B). As expected, TNF- or CS from E. coli LPS-stimulated macrophages induced osteoclast formation (Fig. 3). As B. abortus LPS is not involved in the secretion of inflammatory mediators from macrophages that led to osteoclastogenesis and taking into account that we have demonstrated that B. abortus lipoproteins are able to induce the secretion of cytokines by monocytes/macrophages [27, 29, 30], we hypothesized that Brucella lipoproteins could be the structural component that endows macrophage CS with the capacity to induce osteoclastogenesis. To test this hypothesis, we used L-Omp19 as a Brucella lipoprotein model. L-Omp19-treated macrophages induced significant (P 0.01) secretion of TNF-, IL-1, and IL-6 but not RANKL in a dosedependent manner (Fig. 2B, and data not shown). Concomitantly, CS from L-Omp19-treated macrophages induced a significant (P 0.05) osteoclast formation from BMM, as determined by the generation of TRAP-expressing cells and by the expression of RANK, cathepsin K, and Ckb genes (Fig. 3). Cy- tokine secretion and osteoclast formation induced by Omp19 were dependent on the lipid moiety of the molecule, as UOmp19 was unable to induce these phenomena. The requirement for lipidation was supported further by the fact that Pam3Cys, a lipohexapeptide with an irrelevant peptide sequence, also induced the production of all mediators and the subsequent osteoclast formation (Figs. 2B and 3). These results indicate that the inflammatory response elicited by B. abortus, which causes BMM to undergo osteoclastogenesis, is induced mainly by Brucella lipoproteins. MyD88 and TLR2 determine cytokine production by murine macrophages and concomitant BMM osteoclastogenesis in response to B. abortus infection or L-Omp19 stimulation We have reported that TLR2 signaling is crucial for inflammatory responses induced by B. abortus and L-Omp19 [27 29]. Thus, to assess the requirements for the TLR signaling pathways in proinflammatory cytokine production and subsequent osteoclastogenesis induction in response to B. abortus infection Figure 3. CS from L-Omp19-stimulated murine macrophages induced BMM-derived osteoclastogenesis. BMM cells were stimulated with CS from peritoneal macrophages infected with B. abortus (MOI 100) or stimulated with HKBA (1 106 and bacteria/ml) and E. coli LPS (10 ng/ml), B. abortus LPS (1000 ng/ml), L-Omp19 (10 ng/ml, 100 ng/ml, or 1000 ng/ml), U-Omp19 (1000 ng/ml), or Pam3Cys (50 ng/ml) or nonstimulated for 24 h in conjunction with M-CSF. TNF- was used as a positive control. After 5 days, osteoclastogenesis was determined by the generation of multinucleated, TRAP-expressing cells (A and B). Representative digital images were taken by light microscopy (A); TRAP-positive, multinucleated cells were identified and counted bars; and symbols express the mean sem of duplicates (B). Bars express the mean sem of duplicates. Data shown are from a representative experiment of five performed. *P 0.05; **P 0.01 versus control. Up-regulation of the expression of three genes involved in osteoclast differentiation RANK, cathepsin K, and brain-type Ckb was corroborated by RT-PCR (C). 290 Journal of Leukocyte Biology Volume 91, February

7 Delpino et al. Brucella abortus induces osteoclastogenesis or L-Omp19 stimulation, we infected macrophages from C57BL/6 WT mice or from MyD88 KO, TLR2 KO, and TLR4 KO mice with B. abortus or treated them with L-Omp19 and assessed the production of each cytokine in CS. Macrophages from WT mice secreted IL-6 and TNF- when infected with B. abortus or treated with L-Omp19. In contrast, B. abortus infection or stimulation with different concentrations of L-Omp19 failed to elicit secretion of both cytokines in MyD88 KO mice (Fig. 4). This indicates that IL-6 and TNF- induction by B. abortus or its lipoproteins is dependent on MyD88 and therefore, likely to use a TLR. TNF- and IL-6 production by TLR2 KO mouse macrophages infected with B. abortus or stimulated with L-Omp-19 was reduced significantly (P 0.05) compared with that of WT mice. On the contrary, macrophages from TLR4 KO mice produced similar amounts of TNF- and IL-6 compared to those from wild type mice. As expected, macrophages from TLR2 KO and TLR4 KO mice did not produce proinflammatory cytokines in response to their cognate ligands (Pam 3 Cys and E. coli LPS, respectively), whereas macrophages obtained from WT mice responded to stimulation with both antigens (Fig. 4). To verify the requirement for TLR in eliciting osteoclastogenesis, CS from B. abortus-infected or L-Omp19-stimulated C57BL/6, MyD88 KO, TLR2 KO, and TLR4 KO mice were used to stimulate BMM in the presence of M-CSF. The formation of osteoclast-like cells was only induced by CS from WT or TLR4 KO macrophages, either stimulated with L- Omp-19 or infected with B. abortus. CS from MyD88 KO- or TLR2 KO-stimulated macrophages did not induce osteoclastogenesis (Fig. 5). U-Omp19 was unable to induce cytokine secretion or osteoclastogenesis in macrophages of any mouse strain. Collectively, these results indicate that TLR2 signaling is involved in mediating the cytokine responses to B. abortus and its lipoproteins and that these responses determine concomitant osteoclastogenesis. Osteoclast formation, induced by murine macrophages, infected with B. abortus, or stimulated by L-Omp19, is mediated by TNF- As mentioned, a key cytokine that can be involved in the differentiation of BMM into osteoclasts is TNF- [49, 50]. To assess the role of TNF- in the osteoclastogenesis elicited by B. abortus or L-Omp19, BMM from TNFRp55 KO mice were stimulated with CS from WT macrophages, infected with B. abortus, or treated with L-Omp19, and osteoclastogenesis was evaluated by the generation of TRAP-expressing cells. BMM from C57BL/6 WT mice were used as control. CS from B. abortusinfected or L-Omp19-stimulated WT macrophages were unable to induce osteoclastogenesis of BMM obtained from TNFRp55 KO mice. In contrast, all CS were able to induce osteoclasto- Figure 4. MyD88 and TLR2 determine cytokine production by murine macrophages in response to B. abortus infection or L-Omp19 stimulation. Peritoneal macrophages from WT, MyD88, TLR2, and TLR4 KO were infected with B. abortus at MOI 100 or stimulated with L-Omp19 (10 ng/ml, 100 ng/ml, or 1000 ng/ml), U-Omp19 (1000 ng/ml), Pam 3 Cys (50 ng/ml), or E. coli LPS (10 ng/ml) or nonstimulated. TNF- (A) and IL-6 (B) were determined in CS by ELISA. Bars express the mean sem of duplicates. Data shown are from a representative experiment of five performed. *P 0.05; **P 0.01; ***P versus control. Volume 91, February 2012 Journal of Leukocyte Biology 291

8 Figure 5. MyD88 and TLR2 determine BMM osteoclastogenesis in response to B. abortus infection or L-Omp19 stimulation. BMM cells from WT mice were stimulated with CS from peritoneal macrophages from WT, MyD88, TLR2, and TLR4 KO mice that were infected with B. abortus at MOI 100 or were stimulated with L-Omp19 (10 ng/ml, 100 ng/ml, or 1000 ng/ml), U-Omp19 (1000 ng/ml), Pam3Cys (50 ng/ml), or E. coli LPS (10 ng/ml) or nonstimulated. After 5 days, osteoclastogenesis was determined by the generation of multinucleated, TRAP-expressing cells (A and B). Representative digital images were taken by light microscopy (A) and TRAP-positive, multinucleated cells were identified and counted (B). Bars express the mean sem of duplicates. Data shown are from a representative experiment of five performed. *P 0.05; ***P versus control. genesis of BMM obtained from C57BL/6 mice. CS from U-Omp19-treated macrophages were unable to induce osteoclastogenesis of BMM from both mouse strains (Fig. 6). As expected, TNF- did not induce osteoclastogenesis of 292 Journal of Leukocyte Biology Volume 91, February 2012 TNFRp55 KO BMM, but it elicited osteoclastogenesis of WT BMM. Conversely, RANKL induced osteoclastogenesis of BMM from both strains (Fig. 6). Thus, TNF-, acting through TNFR1, determines osteoclastogenesis.

9 Delpino et al. Brucella abortus induces osteoclastogenesis Figure 6. Osteoclast formation, induced by murine macrophages, infected with B. abortus, or stimulated by L-Omp19, is mediated by TNF-. BMM cells from TNFRp55 / were stimulated with CS from peritoneal macrophages from WT mice that were infected with B. abortus at MOI 100 or stimulated with L-Omp19 (10 ng/ml, 100 ng/ml, or 1000 ng/ml), U-Omp19 (1000 ng/ml), or Pam 3 Cys (50 ng/ml) or nonstimulated. After 5 days, osteoclastogenesis was determined by the generation of multinucleated, TRAP-expressing cells (A and B). Representative digital images were taken by light microscopy (A), and TRAP-positive, multinucleated cells were identified and counted (B). Bars express the mean sem of duplicates. Data shown are from a representative experiment of five performed. **P 0.01; ***P versus control. Human monocytes elicited osteoclastogenesis in response to B. abortus stimulation through TNF- secretion As osteoarticular brucellosis is undoubtedly the most common localization of active human brucellosis, the next experiments were designed to determine whether the effect of B. abortus and L-Omp19 on osteoclastogenesis could be extended to human monocytes, as it has been demonstrated that peripheral human monocytes are able to differentiate to osteoclast-like cells [53, 54]. For this, purified human monocytes were stimulated in the presence of M-CSF with CS collected from monocytes, infected with B. abortus, or treated with HKBA, L-Omp19, or U-Omp19, and osteoclast formation was evaluated by TRAP staining. CS from B. abortus-infected, HKBA- or L-Omp19-treated, but not U-Omp19-stimulated, monocytes were able to induce the differentiation of human monocytes into osteoclasts (Fig. 7A). We also asked if osteoclast formation was mediated by TNF-. Experiments were conducted as described above but in the presence of an anti-tnf- antibody. The TNF- blocking antibody completely abrogated osteoclastogenesis induced by CS from human monocytes, infected by B. abortus, or treated with HKBA or L-Omp19 (Fig. 7B), whereas an isotype control had no effect (Fig. 7C). These results indicate that monocyte-secreted TNF- induced by B. abortus or its lipoproteins could be involved in the osteoarticular focalization of human brucellosis. L-Omp19 and B. abortus induce functional osteoclast-like cells Our premise is that B. abortus infection might create a microenvironment that would promote the generation of osteoclasts, leading to bone loss. Thus, we assessed the functional activity of Brucella-induced, osteoclast-like cells by their ability to resorb dentine. Human monocytes or mouse BMM treated with CS from B. abortus-infected or L-Omp19-stimulated human monocytes or mouse peritoneal macrophages were able to induce a significant (P 0.05) dentine resorption in a dose-dependent manner. On the contrary, CS from U-Omp19-stimulated or from unstimulated cells did not (Fig. 8). In experiments performed with human cells, the resorption of dentine was blocked by anti-tnf antibody but not by the isotype control (Fig. 8C and D). Additionally, macrophages from TNFRp55 KO mice were unable to induce dentine resorption (Fig. 8A and B). Taken together, these results indicate that CS from Brucella-infected mouse or human monocytic cells can promote functional osteoclast formation from BMM or human monocytes. B. abortus and L-Omp19 induce osteoclast in the tibiae of mice in a TLR2-dependent manner To finally determine the in vivo relevance of our hypothesis, HKBA, L-Omp19, and U-Omp19 were injected in the knee joint of C57BL/6 WT or TLR2 KO mice. E. coli LPS and Volume 91, February 2012 Journal of Leukocyte Biology 293

10 multinucleated cells in the tibiae of TLR2 KO mice (Fig. 9). These results indicate that the presence of B. abortus and L- Omp19 within the bone tissue can promote an inflammatory response that leads to the induction of multinucleated, TRAPpositive osteoclast cells and that this phenomenon is mediated by TLR2. DISCUSSION Figure 7. Human monocytes elicited osteoclastogenesis in response to B. abortus stimulation through TNF- secretion. Peripheral human monocytes were stimulated with CS from peripheral human monocytes infected with B. abortus at different MOI (MOI ) or stimulated with HKBA ( and bacteria/ml; A), E. coli LPS (10 ng/ ml), B. abortus (LPS 1000 ng/ml), L-Omp19 (10 ng/ml, 100 ng/ml, or 1000 ng/ml), U-Omp19 (1000 ng/ml), Pam 3 Cys (50 ng/ml), or RANKL (50 ng/ml) or nonstimulated for 24 h in conjunction with M-CSF. TNF- was used as a positive control. After 5 days, osteoclastogenesis was determined by the generation of multinucleated, TRAPexpressing cells (A), in the presence of anti-tnf- antibody (B) or its isotype control (C). TRAP-positive, multinucleated cells were identified and counted. Bars express the mean sem of duplicates. Data shown are from a representative experiment of five performed. **P 0.01; ***P versus control. Pam 3 Cys were used as positive controls; PBS as a negative control. Five days later, animals were killed, whole knee joints were removed, and histological sections of the proximal tibiae were stained for TRAP. An extensive and widespread osteoclastogenesis, defined as TRAP-positive, multinucleated cells, was observed in the tibiae of all WT animals injected with HKBA, L-Omp19, Pam 3 Cys, or E. coli LPS, whereas in those inoculated with U-Omp19 or PBS, osteoclastogenesis was not observed. Conversely, only E. coli LPS was able to induce TRAP-positive, Osteoarticular brucellosis is the most common localization of active brucellosis, and bone loss has been consistently reported in its three most frequent forms of osteoarticular involvement (sacroiliitis, spondylitis, and peripheral arthritis) [11 15]. Although the clinical and imaging aspects of osteoarticular brucellosis have been widely described [13, 14], the pathogenic mechanisms of bone loss caused by Brucella have not been investigated at the molecular and cellular levels. In this manuscript, we describe an immune mechanism for inflammatory bone loss that may take place in response to infection by B. abortus. Although several reports have documented the roles that T and B cells play in inflammatory bone loss [55 58], it is becoming apparent that macrophages are also important immune cells involved in this phenomenon [42]. Our results demonstrate the contribution of the macrophage in response to infection by B. abortus and the resulting induction of osteoclastogenesis. Upon infection with B. abortus, macrophages released inflammatory mediators that were able to induce the formation of osteoclasts from BM undifferentiated cells. Osteoclast development was corroborated phenotypically by the formation of TRAP- and vitronectin receptor-positive, multinucleated cells, which up-regulated the expression of three genes involved in osteoclast differentiation, such as RANK, cathepsin K, and Ckb, and functionally, by the ability of these cells to induce dentine resorption. Moreover, the presence of B. abortus or L-Omp19 in tibiae of mice induced the formation of a TRAP-positive, multinucleated osteoclast. Induction of osteoclastogenesis was not a phenomenon unique to mouse cells. CS from B. abortus-infected human monocytes also induced osteoclastogenesis in monocytes of healthy human donors. Thus, the parallelism between mice and humans underscores the usefulness of the mouse model developed in this investigation for future studies of immune mechanisms that may be implicated in the pathogenesis of the osteoarticular forms of human brucellosis. In chronic inflammatory bone diseases, such as rheumatoid arthritis, the proinflammatory cytokines TNF-, IL-1, and IL-6, as well as RANKL and TGF-, have been shown to be important for disease progression and osteoclastogenesis [18 22, 25, 42]. TNF- has been reported to stimulate osteoclastogenesis by a RANKL-independent mechanism [49, 50]. Infection of murine macrophages by B. abortus elicited the secretion of TNF-, IL-1, and IL-6 but not RANKL. As B. abortus-mediated osteoclastogenesis was induced by CS containing these cytokines, it appears that as it happens in the inflammatory bone loss in response to infection by Porphyromonas gingivalis [42], RANKL would not be involved in the loss of bone induced by B. abortus. Kim et al. [59] reported previously that TGF- plays a role in TNF- -induced osteoclastogenesis. As TGF- is pres- 294 Journal of Leukocyte Biology Volume 91, February

11 Delpino et al. Brucella abortus induces osteoclastogenesis Figure 8. L-Omp19- and B. abortus-induced functional osteoclast-like cells. Functional activity of B. abortus- and L-Omp19-induced osteoclasts-like cells was determined by their ability to resorb dentine. BMM from WT and TNFRp55 / (A and B) or peripheral human monocytes (C and D) were cultured on dentine discs under the same conditions as described above. After 5 days, cells were removed, and dentine resorption was determined by light microscopy (A and C), and the number of resorption pits was counted (B and D). Bars express the mean sem of duplicates. Data shown are from a representative experiment of five performed. *P 0.05; **P 0.01; ***P versus control. ent in FBS, we cannot definitively rule out a role for TGF- in B. abortus-induced osteoclastogenesis. Cytokine production by murine macrophages and concomitant osteoclastogenesis in response to B. abortus were not dependent on bacterial viability, as both phenomena were also induced by exposure to HKBA, suggesting that they were elicited by a structural bacterial component. We established that the structural element responsible for such response was not B. abortus LPS. On the contrary, B. abortus lipoproteins seem to be determinant. L-Omp19, a prototypical B. abortus lipoprotein [27], mimicked the phenotypic and functional changes induced by B. abortus, which lead to osteoclast activation. U-Omp19 was unable to induce any of these changes, confirming that the lipid moiety is required for the production of proinflammatory mediators, which induce osteoclast activation, as it has been reported with different cells types [27 30]. Although it is difficult to determine the actual Omp19 concentration expressed by the Brucella organism in vivo, in this Volume 91, February 2012 Journal of Leukocyte Biology 295

12 Figure 9. B. abortus and L-Omp19 induce osteoclastogenesis in vivo. The ability of B. abortus and L-Omp19 to induce in vivo osteoclastogenesis was determined by their ability to induce osteoclast-like cells in tibiae of mice. Longitudinal sections of tibia from inoculated mice were removed after 5 days and were stained for TRAP. TRAP-positive, multinucleated cells were observed (arrows) in the tibiae of all animal injected with HKBA, L- Omp19, Pam 3 Cys, or E. coli LPS in WT mice or in E. coli LPS-injected animals in TLR2 KO mice. Data shown are from a representative experiment of two performed. study, we used this protein as a model of the B. abortus lipoprotein. Yet, considering that its bioactivity is conferred by the lipid moiety, which is likely the same in all of the lipoprotein molecules of this organism, and that the Brucella genome contains the genes of at least 80 putative lipoproteins, many of which were shown to be expressed in the outer membrane of the bacterium [41, 60], one can envision that the local concentration of Brucella lipoproteins in confined tissue spaces may be sufficient to exert their biological effects. In this context, we can imagine that any surface-exposed Brucella lipoprotein may be relevant beyond in vitro assays, and not one lipoprotein but rather a combination of them can cause the proinflammatory response that leads to the loss of bone seen in osteoarticular brucellosis. As TLRs sense pathogen-derived molecules and initiate the inflammatory reactions of innate immune cells [61], the participation of TLRs in the macrophage-elicited osteoclastogenesis in response to B. abortus merits discussion. Our results using KO mice indicate that B. abortus- and L-Omp19-induced macrophage production of inflammatory mediators that lead to osteoclastogenesis is dependent on the adaptor molecule MyD88, the immediate downstream signaling initiator of TLR, and agree in this regard with the results obtained by other investigators [62]. We showed, moreover, that both phenomena are not TLR4- but TLR2-mediated and provided as B. abortus LPS is a TLR4 ligand [27] proof of concept that B. abortus lipoproteins would be the TLR2 ligands used by the bacterium to trigger the release of proinflammatory mediators that lead to osteoclastogenesis. The incapacity of B. abortus or L-Omp19 to induce the formation of a TRAP-positive, multinucleated osteoclast in tibiae of TLR2-deficient mice underscores further the relevance of TLR2-mediated, inflammatory responses in the osteoclastogenesis induced by B. abortus. TNF- regulates various cellular functions, such as proliferation, differentiation, maintenance of a differentiated phenotype, and apoptosis in various types of cells [63, 64]. In the bone microenvironment, TNF- has multiple actions on bone cells [65]. Diseases, such as rheumatoid arthritis, aseptic loosening, and periodontal diseases, have been associated with the accumulation of TNF- and/or other proinflammatory cytokines, which likely mediate local bone destruction by stimulating osteoclast activity [19, 20, 66, 67]. TNF- is a potent inducer of bone resorption by activating mature osteoclasts [68 70] or by stimulating proliferation and differentiation of osteoclasts precursors [49, 71]. TNF- signaling via type I p55 TNFR appeared to be the sole determinant of macrophageelicited osteoclastogenesis as induced by B. abortus and its lipoproteins, as the formation of functional, TRAP-positive, multinucleated cells with the ability to resorb dentine was abolished completely in BMM from TNFRp55 / mice. Although osteoclast progenitors express TNFRp55 and -p75 [42], it has been described that the majority of TNF- biological activities is initiated by TNFRp55 in these cells. Using BMM from TNFR-disrupted mice (p55 /, p75 /, and both), Abu-Amer et al. [72] showed that LPS-induced osteoclastogenesis was mediated by TNF- and was transmitted through the TNFRp55 but not through the -p75. Also, using p55- and p75-blocking antibodies, Azuma et al. [49] demonstrated that the induction of TRAP-positive, multinucleated cells by TNF- was inhibited completely by anti-p55 antibodies. Moreover, our results demonstrated that anti-tnf- antibodies abrogated osteoclast formation and dentine resorption in human cells, further supporting the contention that TNF- determines the osteoclastogenesis induced by B. abortus and underscoring the role of this cytokine in the loss bone reported in human brucellosis. Of note, although TNF- production by B. abortus-infected macrophages seems to be at odds with previous results [73, 74], it has to be considered that these authors results were obtained when macrophages were infected with Brucella species other than B. abortus. On the contrary, other investigators agree with our results, in that B. abortus-infected macrophages secrete TNF- [32, 33]. Finally, the results presented here, together with our previous observations [39, 75], are shedding light on how the interactions of B. abortus with cells involved in bone metabolisms and its intermingling with cells of the innate immunity play a 296 Journal of Leukocyte Biology Volume 91, February

13 Delpino et al. Brucella abortus induces osteoclastogenesis role in the pathogenesis of osteoarticular brucellosis. Upon infection with B. abortus, osteoblasts release MCP-1 [39], which can attract monocytes to the site of infection. In the in vivo situation, attracted and resident-infected monocytes/macrophages can respond to Brucella lipoproteins with the production of proinflammatory cytokines, such as TNF-, inducing osteoclast formation and subsequent bone destruction. TNF- could, in turn, enhance the production of MCP-1 by osteoblasts [39, 76], attracting more monocytes to the site of infection, which under the influence of local TNF-, would develop into osteoclasts in a pathological, vicious circle, dictating the fate of osteoarticular brucellosis. AUTHORSHIP M.V.D., S.C.O., and G.H.G. conceived of and designed the experiments. M.V.D., P.B., G.C.M., S.D., R.S., and P.C.B. performed the experiments. M.V.D., P.B., R.S., M.C.M., and C.A.F. contributed reagents/materials/analysis tools. M.V.D. and G.H.G. wrote the paper. ACKNOWLEDGMENTS This work was supported by grants PICT , PICT , and PICT from Agencia Nacional of Promoción Científica y Tecnológica (ANPCYT; Argentina); PIP from Consejo Nacional de Investigación Científica y Tecnológica (CONICET); UBACYT from Universidad de Buenos Aires; and CNPq/Prosul (#490485/2007-3), CNPq/ANPCy (#490528/ ), and INCT-Vacinas. R.S. is a recipient of a fellowship from ANPCYT. M.V.D., P.B., C.A.F., P.C.B., and G.H.G. are members of the Research Career of CONICET. We thank Horacio Salomón and the staff of Centro Nacional de Referencia del Sida (University of Buenos Aires) for their assistance with Biosafety Level 3 laboratory use. We thank Dr. Ignacio Moriyón (University of Navarra, Pamplona, Spain) for E. coli and B. abortus LPS. REFERENCES 1. Young, E. J. (1995) An overview of human brucellosis. Clin. Infect. Dis. 21, , quiz Pappas, G., Akritidis, N., Bosilkovski, M., Tsianos, E. (2005) Brucellosis. N. Engl. J. Med. 352, Colmenero, J. D., Reguera, J. M., Martos, F., Sanchez-De-Mora, D., Delgado, M., Causse, M., Martin-Farfan, A., Juarez, C. (1996) Complications associated with Brucella melitensis infection: a study of 530 cases. Medicine (Baltimore) 75, Colmenero, J. D., Reguera, J. M., Fernandez-Nebro, A., Cabrera-Franquelo, F. (1991) Osteoarticular complications of brucellosis. Ann. Rheum. Dis. 50, Priest, J. R., Low, D., Wang, C., Bush, T. (2008) Brucellosis and sacroiliitis: a common presentation of an uncommon pathogen. J. Am. Board Fam. Med. 21, Gotuzzo, E., Alarcon, G. S., Bocanegra, T. S., Carrillo, C., Guerra, J. C., Rolando, I., Espinoza, L. R. (1982) Articular involvement in human brucellosis: a retrospective analysis of 304 cases. Semin. Arthritis Rheum. 12, Khateeb, M. I., Araj, G. F., Majeed, S. A., Lulu, A. R. (1990) Brucella arthritis: a study of 96 cases in Kuwait. Ann. Rheum. Dis. 49, Ariza, J., Gudiol, F., Valverde, J., Pallares, R., Fernandez-Viladrich, P., Rufi, G., Espadaler, L., Fernandez-Nogues, F. (1985) Brucellar spondylitis: a detailed analysis based on current findings. Rev. Infect. Dis. 7, Ibero, I., Vela, P., Pascual, E. (1997) Arthritis of shoulder and spinal cord compression due to Brucella disc infection. Br. J. Rheumatol. 36, Kelly, P. J., Martin, W. J., Schirger, A., Weed, L. A. (1960) Brucellosis of the bones and joints. Experience with thirty-six patients. JAMA 174, Aydin, M., Fuat Yapar, A., Savas, L., Reyhan, M., Pourbagher, A., Turunc, T. Y., Ziya Demiroglu, Y., Yologlu, N. A., Aktas, A. (2005) Scintigraphic findings in osteoarticular brucellosis. Nucl. Med. Commun. 26, Colmenero, J. D., Ruiz-Mesa, J. D., Plata, A., Bermudez, P., Martin-Rico, P., Queipo-Ortuno, M. I., Reguera, J. M. (2008) Clinical findings, therapeutic approach, and outcome of brucellar vertebral osteomyelitis. Clin. Infect. Dis. 46, Madkour, M. M. (2001) Osteoarticular brucellosis. Madkour's Brucellosis, 2nd Ed., Springer-Verlag, Berlin, Germany. 14. Madkour, M. M. (2001) Bone and joint imaging. Madkour's Brucellosis, 2nd Ed., Springer-Verlag, Berlin, Germany. 15. Young, E. J. (1989) Clinical manifestations of human brucellosis. Brucellosis: Clinical and Laboratory Aspects (E. J. Young, M. J. Corbel, eds.), CRC, Boca Raton, FL, USA. 16. Suda, T., Kobayashi, K., Jimi, E., Udagawa, N., Takahashi, N. (2001) The molecular basis of osteoclast differentiation and activation. Novartis Found. Symp. 232, Kong, Y. Y., Feige, U., Sarosi, I., Bolon, B., Tafuri, A., Morony, S., Capparelli, C., Li, J., Elliott, R., McCabe, S., Wong, T., Campagnuolo, G., Moran, E., Bogoch, E. R., Van, G., Nguyen, L. T., Ohashi, P. S., Lacey, D. L., Fish, E., Boyle, W. J., Penninger, J. M. (1999) Activated T cells regulate bone loss and joint destruction in adjuvant arthritis through osteoprotegerin ligand. Nature 402, Haynes, D. R. (2004) Bone lysis and inflammation. Inflamm. Res. 53, Merkel, K. D., Erdmann, J. M., McHugh, K. P., Abu-Amer, Y., Ross, F. P., Teitelbaum, S. L. (1999) Tumor necrosis factor- mediates orthopedic implant osteolysis. Am. J. Pathol. 154, Nair, S. P., Meghji, S., Wilson, M., Reddi, K., White, P., Henderson, B. (1996) Bacterially induced bone destruction: mechanisms and misconceptions. Infect. Immun. 64, Wei, S., Kitaura, H., Zhou, P., Ross, F. P., Teitelbaum, S. L. (2005) IL-1 mediates TNF-induced osteoclastogenesis. J. Clin. Invest. 115, Kotake, S., Sato, K., Kim, K. J., Takahashi, N., Udagawa, N., Nakamura, I., Yamaguchi, A., Kishimoto, T., Suda, T., Kashiwazaki, S. (1996) Interleukin-6 and soluble interleukin-6 receptors in the synovial fluids from rheumatoid arthritis patients are responsible for osteoclast-like cell formation. J. Bone Miner. Res. 11, Revell, P. A., Mayston, V., Lalor, P., Mapp, P. (1988) The synovial membrane in osteoarthritis: a histological study including the characterisation of the cellular infiltrate present in inflammatory osteoarthritis using monoclonal antibodies. Ann. Rheum. Dis. 47, Smith, M. D., O'Donnell, J., Highton, J., Palmer, D. G., Rozenbilds, M., Roberts-Thomson, P. J. (1992) Immunohistochemical analysis of synovial membranes from inflammatory and non-inflammatory arthritides: scarcity of CD5 positive B cells and IL2 receptor bearing T cells. Pathology 24, Chu, C. Q., Field, M., Allard, S., Abney, E., Feldmann, M., Maini, R. N. (1992) Detection of cytokines at the cartilage/pannus junction in patients with rheumatoid arthritis: implications for the role of cytokines in cartilage destruction and repair. Br. J. Rheumatol. 31, Adamopoulos, I. E., Sabokbar, A., Wordsworth, B. P., Carr, A., Ferguson, D. J., Athanasou, N. A. (2006) Synovial fluid macrophages are capable of osteoclast formation and resorption. J. Pathol. 208, Giambartolomei, G. H., Zwerdling, A., Cassataro, J., Bruno, L., Fossati, C. A., Philipp, M. T. (2004) Lipoproteins, not lipopolysaccharide, are the key mediators of the proinflammatory response elicited by heat-killed Brucella abortus. J. Immunol. 173, Zwerdling, A., Delpino, M. V., Barrionuevo, P., Cassataro, J., Pasquevich, K. A., Garcia Samartino, C., Fossati, C. A., Giambartolomei, G. H. (2008) Brucella lipoproteins mimic dendritic cell maturation induced by Brucella abortus. Microbes Infect. 10, Barrionuevo, P., Cassataro, J., Delpino, M. V., Zwerdling, A., Pasquevich, K. A., Garcia Samartino, C., Wallach, J. C., Fossati, C. A., Giambartolomei, G. H. (2008) Brucella abortus inhibits major histocompatibility complex class II expression and antigen processing through interleukin-6 secretion via Toll-like receptor 2. Infect. Immun. 76, García Samartino, C., Delpino, M. V., Pott Godoy, C., Di Genaro, M. S., Pasquevich, K. A., Zwerdling, A., Barrionuevo, P., Mathieu, P., Cassataro, J., Pitossi, F., Giambartolomei, G. (2010) Brucella abortus induces the secretion of proinflammatory mediators from glial cells leading to astrocyte apoptosis. Am. J. Pathol. 176, Zwerdling, A., Delpino, M. V., Pasquevich, K. A., Barrionuevo, P., Cassataro, J., Garcia Samartino, C., Giambartolomei, G. H. (2009) Brucella abortus activates human neutrophils. Microbes Infect. 11, Zhan, Y., Kelso, A., Cheers, C. (1993) Cytokine production in the murine response to Brucella infection or immunization with antigenic extracts. Immunology 80, Zhan, Y., Cheers, C. (1995) Differential induction of macrophage-derived cytokines by live and dead intracellular bacteria in vitro. Infect. Immun. 63, Volume 91, February 2012 Journal of Leukocyte Biology 297