E X PE R IM ENTA L C ELL R E S EA RC H

3 21 ( 2 0 14 ) 16 7 – 17 7

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/yexcr

Research Article

Tumor necrosis factor stimulates osteoclastogenesis from human bone marrow cells under hypoxic conditions Takayuki Nomuraa,b, Mineyoshi Aoyamab,n, Yuko Waguri-Nagayac, Yoh Gotoa,b, Mieko Suzukib, Ken Miyazawaa, Kiyofumi Asaib, Shigemi Gotoa a

Department of Orthodontics, School of Dentistry, Aichi-Gakuin University, Chikusa-ku, Nagoya 464-8651, Japan Department of Molecular Neurobiology, Nagoya City University Graduate School of Medical Sciences, Mizuho-ku, Nagoya 467-8601, Japan c Department of Orthopaedic Surgery, Nagoya City University Graduate School of Medical Sciences, Mizuho-ku, Nagoya 467-8601, Japan b

article information

abstract

Article Chronology:

Bone homeostasis is maintained by the balance between osteoblastic bone formation and osteoclastic

Received 18 July 2013

bone resorption. In this study, we used human bone marrow cells (BMCs) to investigate the role of

Received in revised form

hypoxic exposure on human osteoclast (OC) formation in the presence of tumor necrosis factor (TNF).

6 November 2013

Exposing the BMCs to 3%, 5%, or 10% O2 in the presence of receptor activator of NF-κB ligand (RANKL)

Accepted 27 November 2013

and macrophage colony-stimulating factor (M-CSF) generated tartrate-resistant acid phosphatase

Available online 17 December 2013

(TRAP)-positive multinuclear cells, consistent with OCs. The addition of TNF under hypoxic conditions

Keywords:

generated significantly greater numbers of mature OCs with more nuclei than OCs generated under

Human osteoclast

normoxic conditions. Longer initial hypoxic exposure increased the number of OC precursor cells and

Human bone marrow cells (BMCs)

facilitated the differentiation of OC precursor cells into multinucleated OCs. Quantitative RT-PCR

Hypoxia

analysis revealed that RANKL and TNFR1 were expressed at higher levels in non-OC cells from BMCs

Tumor necrosis factor (TNF)

under hypoxic conditions than under normoxic conditions. Furthermore, to confirm the involvement of TNF-induced signaling, we examined the effects of blocking antibodies against TNFR1 and TNFR2 on OC formation under hypoxic conditions. The TNFR1 antibody was observed to significantly suppress OC formation. These results suggest that hypoxic exposure plays an important role in TNFinduced osteoclastogenesis from human BMCs. & 2013 Elsevier Inc. All rights reserved.

Introduction Bone homeostasis is maintained by the balance between osteoblastic bone formation and osteoclastic bone resorption [1–3]. Osteoclasts (OCs) are mature bone-resorbing multinucleated cells

n

Corresponding author. Fax: þ81 52 842 3316. E-mail address: [email protected] (M. Aoyama).

0014-4827/$ - see front matter & 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.yexcr.2013.11.020

that differentiate from monocyte/macrophage lineage cells under the tight regulation of osteoblast cells in the local bone environment [1,4–6]. Multinucleated OCs are characterized by the presence of tartrate-resistant acid phosphatase (TRAP) activity, expression of vitronectin receptors (VNR), and pit-forming activity

168

E XP ER I ME NTAL C E LL RE S E ARCH

on bone. They are formed in co-cultures of mouse osteoblasts and hematopoietic cells in the presence of osteotrophic factors such as 1–25-dihydroxyvitamin D3 [1,25(OH)2D3], parathyroid hormone (PTH), and interleukin (IL)-11 [6–8]. Osteoblasts express receptor activator of NF-κB (RANK) ligand (RANKL) and macrophage colonystimulating factor (M-CSF), which are essential cytokines for OC differentiation [9,10]. M-CSF is constitutively expressed by osteoblasts, whereas RANKL expression is up-regulated by osteotrophic factors. OC precursors express RANK and the M-CSF receptor (CSF1R), and differentiate into mature OCs in the presence of both cytokines [1,4,6]. Thus, osteoblast-lineage cells provide a suitable microenvironment for osteoclastogenesis [11]. Tumor necrosis factor (TNF) is a pro-inflammatory cytokine that stimulates bone resorption and is expressed in abundance at sites of inflammation and in the bone microenvironment [12]. A variety of cell types express TNF protein including monocytes/macrophages, OC precursors, and mature OCs themselves [13,14]. Merkel et al. demonstrated that TNF mediates the orthopedic osteolysis induced by implant-derived particles [15]. TNF also plays important roles in the bone loss associated with osteoporosis and periodontitis [16,17]. Several biological responses are induced by TNF acting via the TNFR1 and TNFR2 cell-surface receptors, which are also termed TNFR p55 and TNFR p75, respectively [18–20]. Both TNFR1 and TNFR2 initiate intracellular signals that stimulate the proteolytic breakdown of IκB, which is a cytoplasmic inhibitor of NF-κB protein [21,22]. The activated NF-κB then translocates from the cytoplasm to the nucleus, where it induces the transcription of several TNF-responsive genes. In addition, the binding of TNF to TNFR1 triggers programmed cell death in many cells. This process depends upon the presence of the “death domain” located in the cytoplasmic region of TNFR1, but not in TNFR2 [23,24]. Murine TNF binds to both murine TNFR1 and TNFR2 with high affinity, whereas human TNF binds to murine TNFR1 with higher affinity than to murine TNFR2 [18,19]. Hypoxia is a feature of skeletal conditions including rheumatoid arthritis [25], pathological fracture [26], primary bones tumors [27], and cancer metastases to bone [28], periodontitis and orthodontic tooth movement [29,30]. Measurement of oxygen partial pressure (pO2) in bone marrow aspirates from normal volunteer donors yield a mean value of 6.6% [31]. The pO2 is often considerably lower in environments such as inflamed tissues, infected tissues, tumors, wounds, fracture sites or sites of orthodontic tooth movement [32], Hypoxia stimulates both the formation and activation of OCs from feline [33], murine [28,34], and human [35,36] monocytes. In addition, we previously demonstrated that hypoxic stress enhances OC differentiation from murine bone marrow cells (BMCs) [37]. In the present study, we investigated human OC formation from human BMCs using a hypoxic culture system to confirm the effect of hypoxia on OC formation we observed using mouse BMCs. In addition, we examined the role of the pro-inflammatory cytokine TNF in human OC differentiation under relative hypoxic conditions. We also explored the contribution of hypoxia to TNFinduced osteoclastogenesis using human BMCs.

Materials and methods Reagents The following reagents were used: recombinant human (rh) MCSF, TNF, neutralizing mouse anti-human TNF receptor 1 (TNFR1;

3 21 ( 2 0 14 ) 16 7 – 17 7

Mab225) monoclonal antibody and anti-human TNF receptor 2 (TNFR2; Mab226) monoclonal antibody (R&D Systems, Minneapolis, MN, USA); and rh soluble RANKL (Sigma, St Louis, MO, USA).

Human OC culture BMCs were obtained from approximately 40 patients who underwent total hip arthroplasty for osteoarthritis and the study was approved by the Ethic Committee of Nagoya City University. A specimen of 15 mL of bone marrow was aspirated from each patient during rasping of the femoral canal prior to implant placement. All of the subjects gave their written informed consent before participating in the study. BMCs were isolated by density gradient centrifugation with Lymphosepar (Ficoll-Conray solution, Immuno-Biological Laboratories, Fujioka, Japan) at 1800 rpm for 30 min at room temperature. Nucleated cells were collected from the interface between the plasma and the Lymphosepar solution. The cells were washed with phosphate-buffered saline (PBS), and resuspended in Stem Pro 34 medium, which uses serum replacement without fetal bovine serum (FBS) (Invitrogen Corp., NY, USA).

In vitro assay of TRAP-positive OC formation In vitro osteoclastogenesis was assayed by seeding cells in three replicate wells of a 24-well plate at a density of 2  106 cells per well and incubating for 21 days in Stem Pro 34 containing RANKL (100 ng/ mL) and M-CSF (50 ng/mL). Cultures were treated with TNF (10 ng/ mL) at the indicated period of time in the presence of RANKL and M-CSF from day 0. Cells incubated under hypoxic conditions were cultured in a controlled atmosphere culture chamber (Bellco Glass, Vineland, NJ, USA). This airtight apparatus had a humidified atmosphere with inflow and outflow control valves. The apparatus was flushed for 15 min prior to incubating the cells with gas mixtures containing 5% CO2 and 10%, 5% or 1% O2 and N2 as the balance of the atmosphere. The apparatus chamber was sealed to maintain the gas composition and was incubated at 37 1C. The medium was changed every 3 d by replacing 0.5 mL of used culture medium with an equal quantity of fresh medium containing the appropriate cytokines. After culturing for 21 d, cells were fixed and stained for TRAP as described previously [37,38]. The number of multinucleated TRAP-positive cells was counted using an Olympus inverted microscope at 40  magnification. Cells with three to 10 nuclei were counted as TRAP-positive small cells and cells with more than 10 nuclei were counted as TRAPpositive large cells.

Cell imaging analysis OC differentiation was assessed using TRAP activity staining. BMCs were seeded in replicates of three wells in 24 well plates at a density of 2  106 cells per well and were cultured under normoxic or hypoxic (5% O2) conditions for various periods of hypoxia. After 21 d in culture the cells were fixed and stained for TRAP activity. The nuclei were stained with Hoechst 33258 (Wako Pure Chemical Industries, Osaka, Japan) to obtain a total nuclear count. Highcontent cellular images were acquired at 20  magnification with an IN Cell Analyzer 2000 (GE Healthcare, Buckinghamshire, UK). Small or large TRAP-positive cells were defined as described above. Otherwise, cells with a cell area less than 2000 μm2 were counted as TRAP-positive small area cells and cells with a cell area more than 2000 μm2 were counted as TRAP-positive large area cells.

E XP E RI ME N TAL CE L L R ES E ARC H

Purification of OCs and non-OCs OCs were purified by placing the crude OC preparation on plastic dishes and treating the cells with pronase-EDTA, as previously reported [37,38]. Briefly, non-adherent cells including stromal cells were harvested before pronase treatment. The crude OC preparation was transferred to 24-well dishes and the cells were washed with PBS, and then treated with 1 mL of PBS containing 0.001% pronase (Calbiochem, La Jolla, CA, USA) and 0.02% EDTA for 10 min to remove osteoblast cells. More than 90% of the adherent cells on the dish were TRAP-positive mononuclear and multinucleated cells, and were classified as OCs. The cells detached by the pronase procedure were harvested as non-adherent cells and included osteoblast cells. Both floating cells and non-adherent cells were collected as non-OCs.

Isolation of RNA and quantitative reverse transcription PCR (Q-RT-PCR) Total RNA was isolated using TRIzol™ reagent (Invitrogen) according to manufacturer0 s protocol. Reverse transcription was carried out from total RNA samples using random primers and Ready-To-Go You-Prime First Strand Beads (GE Healthcare Bio-Science Corp., Piscataway, NJ). The resulting cDNAs were subjected to quantitative-reverse transcription PCR (Q-RT-PCR) for selected genes using the 7700 Sequence Detection System (Applied Biosystems, Foster City, CA). The primer pairs used for amplification are shown in Table 1. Real-time PCR was performed using SYBR Green Master Mix Reagents (Applied Biosystems). Amplification was performed by activation of AmpliTaq Gold polymerase at 95 1C for 10 min, followed by amplification for 40 cycles at 95 1C for 15 s and 60 1C for 1 min. The amount of RNA was calculated using relative standard curves for each mRNA of interest and β-actin (ACTB). Normalization

169

3 21 ( 2 0 14 ) 16 7 – 17 7

to ACTB was conducted to account for variability in the quality and total concentration of RNA, as well as in the efficiency of the RT reaction.

Inhibition of TNFR1 and TNFR2 BMCs (2  106) were seeded to 24 well plates in the absence of TNF, RANKL, and M-CSF and were incubated overnight. The BMCs were then pre-incubated for 30 min with 3 μg/mL anti-TNFR1 monoclonal antibody or anti-TNFR2 monoclonal antibody before the addition of TNF (10 ng/mL), RANKL (100 ng/mL), and M-CSF (50 ng/mL). The cells were cultured for 21 d and the medium was changed every 3 d by replacing 0.5 mL of used culture medium with fresh medium. After 21 d, cells were fixed and stained for TRAP activity. The number of multinucleated TRAP-positive cells was counted using an Olympus inverted microscope at 40  magnification.

Statistical analysis The data are expressed as mean7S.E.M. Statistical analysis was performed by Student0 s t-test or one-way factorial analysis of variance combined with Bonferroni correction for all comparison pairs. Differences with P valueso0.05 were considered significant.

Results Effect of TNF on osteoclastogenesis from BMCs under hypoxic conditions The effect of hypoxic conditions on the number of TRAP-positive multinuclear OC cells formed in these cultures was examined

Table 1 – Primers used for RT-PCR analysis cDNA

Sequence (50 –30 )

Human DCSTAMP (NM_001257317.1) Human NFATC1 (NM_006162.3) Human HIF1A (NM_001530.3) Human VEGF (NM_001025366.1) Human IGF2 (NM_000612.4) Human SDF1 (NM_000609.6) Human RANKL (NM_033012.3) Human RANK (NM_003839.2) Human TNFα (NM_000594.2) Human TNFR1 (NM_001065.3) Human TNFR2 (NM_001066.2) Human ACTB (NM_001101.3) (fwd, forward primer; rev, reverse primer).

ACAGCCCTTGGGAAGTGCAT (fwd) GCAAACTCCCAAATGCTGGATAA (rev) GAAGACCGTGTCCACCACCA (fwd) CGAAGTTCAATGTCGGAGTTTCTG (rev) CAGCCGCTGGAGACACAATC (fwd) TTTCAGCGGTGGGTAATGGA (rev) GAGCCTTGCCTTGCTGCTCTAC (fwd) CACCAGGGTCTCGATTGGATG (rev) GGAACCCACATTGGCCTGA (fwd) CCGGCGAGGCAGAATATAACAC (rev) GAGCCAACGTCAAGCATCTCAA (fwd) TTAGCTTCGGGTCAATGCACAC (rev) AAGATGGCACTCACTGCATTTATAG (fwd) TGATGTGCTGTGATCCAACGA (rev) CCATCATCTTTGGCGTTTGCTA (fwd) GACCAAAGTTTGCCGTGTGTGTA (rev) GTGACAAGCCTGTAGCCCATGTT (fwd) TTATCTCTCAGCTCCACGCCATT (rev) ACAGAACACCGTGTGCACCT (fwd) GCACAACTTCGTGCACTCC (rev) ACACCGTGTGTGACTCCTGTGA (fwd) CCGAGTGCAGGCTTGAGTTTC (rev) TGGCACCCAGCACAATGAA (fwd) CTAAGTCATAGTCCGCCTAGAAGCA (rev)

Amplified region (bp) 15–187 1950–2093 1609–1776 1061–1208 4681–4784 214–321 552–731 6917–7021

423–534 786–888 304–425 1043–1228

170

E XP ER I ME NTAL C E LL RE S E ARCH

3 21 ( 2 0 14 ) 16 7 – 17 7

Fig. 1 – Effects of TNF on RANKL-induced OC formation in human BMCs under hypoxic conditions. (A) Human BMCs were cultured with TNF in the presence of RANKL and M-CSF for 21 d, whereupon they were fixed and stained for TRAP activity as a marker of osteoclastogenesis (red/purple staining). OC formation after culture for 21 d under normoxic (20% O2) or hypoxic (10%, 5%, 3%, or 1% O2) conditions is shown. (B) TRAP-positive multinucleated cells (MNCs) containing three or more nuclei were counted as human OCs derived from seven individual donors. The results are expressed as the relative ratio to the cell number of OCs in the presence of TNF under normoxic conditions in each individual. Each bar represents the mean7S.E.M. (n¼7). n Po0.05, compared between the same fraction of cells cultured under normoxic conditions and under hypoxic conditions. # Po0.05, compared with cells cultured in the absence of TNF under the same oxygen concentration conditions.

(Fig. 1). In the presence of RANKL and M-CSF, cultured human BMCs exposed to 3%, 5%, or 10% O2 differentiated into TRAPpositive multinuclear OC cells (Fig. 1A), and only a few, small TRAP-positive cells were observed under normoxic conditions. The addition of TNF under hypoxic conditions resulted in significantly larger numbers of mature OCs with larger cytoplasm and more nuclei than were generated under normoxic conditions. The largest increases in OC number were observed at 3%, 5% and 10% O2, whereas OC formation was completely suppressed at 1% O2. These observations were verified using BMCs from seven individual donors. The BMCs were cultured in M-CSF and RANKL for 21 d in the presence or absence of TNF, whereupon the multinucleated OCs formed, including both small and large OCs, were counted in the same experiments (Fig. 1B). The maximal number of both small OCs and large OCs as well as the total number OCs was produced in the presence of TNF at 3%, 5% and 10% O2. In the absence of TNF, hypoxic treatment did not promote OC formation significantly.

OCs into multinucleated cells is an important process in OC functional differentiation [39,40]. Thus, we investigated the effect of hypoxic exposure on the fusion of premature OCs (preOCs) into multinucleated OCs by counting large OCs with more than 10 nuclei. In the presence of TNF, longer continuous hypoxic exposure resulted in the formation of more large OCs than shorter hypoxic exposure. Also, large OCs were rarely observed with hypoxic exposure from later time points (days 7–21 or days 14–21). In the absence of TNF, OC formation was rarely observed under either normoxic or hypoxic conditions (Fig. 2). To verify these observations, BMCs from five individual donors were incubated in the presence or absence of TNF with M-CSF and RANKL for 21 days, whereupon the number of large OCs and the total number of multinucleated OCs were counted in the same experiments (Fig. 3). Early and longer hypoxic exposure resulted in increased numbers of preOCs and facilitated the differentiation of preOCs into multinucleated OCs. TNF treatment induced OC formation synergistically with hypoxic exposure.

Importance of timing of hypoxic treatment for OC formation in the presence of TNF

Differential gene expression with hypoxia and TNF treatment

The effect of hypoxic exposure on OC differentiation was examined by exposing BMCs to hypoxic conditions (O2 5%) at different time points during OC differentiation in vitro. Hypoxic exposure during the early culture time points (days 1–7, days 1–14, or days 1–21) enhanced differentiation of BMCs into OCs, but from later time points (days 7–21 or days 14–21) it failed to enhance OCs differentiation. It should be noted that the fusion of mononuclear

To understand the molecular mechanisms responsible for the enhancement of OC differentiation with hypoxic exposure, we examined changes in gene expression in the cultured cells. Total RNA was prepared from the cultures treated with TNF under normoxic or hypoxic conditions (5% O2) for the entire 21 d of culture with M-CSF and RANKL. Q-RT-PCR was performed on total RNA from separated OC and non-OC cell populations to determine

E XP E RI ME N TAL CE L L R ES E ARC H

3 21 ( 2 0 14 ) 16 7 – 17 7

171

Fig. 2 – The effect of changes in the time course of hypoxia on OC differentiation. Human BMCs were cultured under normoxic or hypoxic conditions with M-CSF and RANKL for the following times: the first 7 d, 14 d or the entire 21 d or the last 7 d or 14 d in the absence or presence of TNF. At the end of the culture period, the cells were stained for TRAP activity and the nuclei were stained with Hoechst 33258. The numbers of TRAP-positive multinucleated cells per well were counted using an IN Cell Analyzer. The first column from the left: plan of the time course of human BMCs cultured for 21 d with M-CSF and RANKL in the presence or absence of TNF. Hypoxic exposure was added for various time periods as indicated (solid). The second and fourth columns from the left: the representative morphological appearance of human BMCs in the presence or absence of TNF. The third and fifth columns from the left: Nuclear counts and cell areas of individual TRAP-positive MNCs in the absence or the presence of TNF were measured by the IN Cell Analyzer.

whether OC cells or other cells, including osteoblasts, expressed hypoxia-induced genes. Initially, to confirm that hypoxic treatment enhanced hypoxia marker genes, we examined the gene expression of hypoxia inducible factor 1a (HIF1a) and vascular endothelial growth factor (VEGF). We also examined the gene expression of stromal cellderived factor 1 (SDF1) and insulin-like growth factor 2 (IGF2), which were increased in non-OCs under hypoxic conditions in our previous study using mouse cells [37]. HIF1a was expressed predominantly in non-OCs, and was not upregulated in OCs or non-OCs under hypoxic conditions. VEGF was also expressed predominantly in non-OCs, and was upregulated in these cells with hypoxic treatment. SDF1 expression was markedly upregulated under hypoxic conditions in non-OCs, although the same

treatment did not enhance IGF2 expression in non-OCs or OCs (Fig. 4A). We also examined the gene expressions of cell fusion marker, dendritic cell-specific transmembrane protein (DC-STAMP), and differentiation marker, nuclear factor of activated T-cells, cytoplasmic, calcineurin-dependent 1 (NFATc1). DC-STAMP was expressed predominantly in OCs, and upregulated slightly in these cells with hypoxic treatment. In addition, NFATc1 expression increased significantly in OCs under hypoxic conditions (Fig. 4B). RANKL was expressed predominantly in non-OCs whereas RANK was expressed predominantly in OCs (Fig. 5A). These data confirmed the effective separation of the OC and non-OC cell populations. Interestingly, hypoxic treatment enhanced RANKL expression in non-OCs (Fig. 5A). Hypoxic treatment also enhanced

172

E XP ER I ME NTAL C E LL RE S E ARCH

3 21 ( 2 0 14 ) 16 7 – 17 7

Fig. 3 – Early and longer hypoxic exposure increased OCs and facilitated the differentiation of multinucleated OCs. Human BMCs from five individual donors were treated in normoxic or hypoxic conditions with M-CSF and RANKL for the following different periods: the entire 21 d, the first 7 d or 14 d, or the last 7 d or 14 d in the presence of TNF. At the end of the culture period, the cells were stained for TRAP activity and the number of TRAP-positive MNCs was counted. The results are representative of BMCs from five individuals. Results are expressed as the relative ratio to the cell number of OCs in the presence of TNF under normoxic conditions for each individual. Each bar represents the mean7S.E.M. (n¼ 5). n Po0.05, compared with culture cells under normoxic conditions for the entire 21 d. (A) Hypoxic exposure was added for various time periods as indicated (solid). (B) Numbers of total TRAP-positive MNCs were counted and are shown. (C) The number of TRAP-positive MNCs with small nuclear number (small NN cells) or with large nuclear number (large NN cells) was counted, and are shown respectively in the top panels. The numbers of TRAP-positive MNCs with smaller area (small area cells) or with larger area (large area cells) were counted, and are shown respectively in the bottom panels.

TNFR1 expression in non-OCs, whereas TNFR2 expression did not change significantly in OCs or non-OCs. TNF expression did not change significantly with hypoxic exposure (Fig. 5B).

TNFR1 and TNFR2 signaling for OC formation under hypoxic conditions To examine the involvement of TNF signaling in regulating osteoclastogenesis, we treated human BMCs in the presence of RANKL and M-CSF with blocking antibodies against TNFR1 and TNFR2 and measured TRAP-positive cell formation under hypoxic conditions (Fig. 6). The antibody against TNFR1 strongly inhibited TRAP-positive cell formation induced by TNF. To assess the possibility of crosstalk between TNFR1 and TNFR2 signaling, we added blocking antibodies against TNFR1 and TNFR2 together. Unexpectedly, the blocking effects by both antibodies were weaker than the effect of the antibody against TNFR1 alone. These results suggest that both TNFR1- and TNFR2-mediated signals are important for TNF-induced OC formation under hypoxic conditions.

Discussion Many studies of OC biology have utilized primary murine progenitor cells for the generation of OCs because mouse models offer significant advantages in terms of experimental manipulation. Although studies employing murine cells have yielded much valuable information, the use of human cells is likely to translate more readily into clinical benefits [41]. Previous studies in bone cell biology have shown that there are differential responses and effects between humans and mice. Prostaglandin E2 (PGE2) strongly inhibits RANKL-induced OC formation in human peripheral blood mononuclear cell cultures [42], whereas PGE2 enhances the RANKL and M-CSF-induced differentiation of mouse bone marrow-derived monocyte/macrophages into OCs [43–45]. The effect of PGE2 on in vitro OC formation in human culture systems is the opposite of mouse culture systems [42]. Human CD14þ cells pretreated with PGE2 produce an inhibitory factor(s) for OC formation not only in human CD14þ cells, but also in mouse OC precursor cells [42]. Thus, while it is important to investigate human osteoclastogenesis using human cells, recognition of the differences between species used in these studies is critical for

E XP E RI ME N TAL CE L L R ES E ARC H

3 21 ( 2 0 14 ) 16 7 – 17 7

173

Fig. 4 – Q-RT-PCR analysis of the expression of selected mRNAs involved in hypoxic treatment and OC differentiation. Human bone marrow cultures were incubated for 21 d in osteoclastogenic culture medium then OCs were separated from the non-adherent cells (non-OCs) using the pronase procedure. Total RNA was prepared from the cell populations and reverse transcribed into cDNA prior to Q-RT-PCR analysis. The expression of β-actin served as a normalization control. The results were expressed as the expression relative to the expression in non-OCs in the presence of TNF under normoxic conditions for each individual. Each bar represents the mean7S.E.M. (n ¼7). nPo0.05, compared between cells from the same preparation cultured under normoxic conditions and under hypoxic conditions. # Po0.05, compared between non-OC and OC cells under the same culture conditions.

understanding the function of cytokines [46]. We previously demonstrated that hypoxic stress enhances OC differentiation from murine BMCs [37], and now have confirmed that human OCs could be generated from human BMCs under the same hypoxic conditions, although the culture medium differed between the studies. Human BMCs were cultured in serum-free Stem Pro 34 medium using serum replacement, while mouse BMCs were cultured in α-modification minimum essential medium (α-MEM) containing 10% FBS. The serum-free culture conditions were useful for detecting and quantifying the maturation-promoting effects of hypoxia in the presence of TNF when human BMCs were cultured, while at the same time prohibiting significant proliferation of fibroblasts. Moreover, we previously reported elevated gene expressions of both IGF2 and SDF1 in mouse non-OCs under hypoxic conditions [39]. However, in our present study, SDF1 gene expression was increased, but IGF2 was decreased in human nonOCs under hypoxic conditions, possibly representing the discrepancy in signaling between human and mouse cells. TNF directly stimulates the differentiation of OC precursors through a pathway that is not inhibited by osteoprotegerin (OPG) and does not require the presence of stromal support cells or the addition of exogenous RANKL [47,48]. TNF also has many effects on stromal cells, and the presence of stromal cells is required for the stimulation of OC activity by TNF [49]. Previous reports by Ragab et al. and Lam et al. have suggested that TNF primarily stimulates OC differentiation through direct actions on OC precursors [50,51]. This mechanism is distinctly different from that utilized by most other bone resorptive agents that act through the RANKL pathway It is likely that RANKL acts synergistically with TNF to stimulate OC differentiation both in vivo [52] and in vitro [51,53]. Hypoxia is a common feature in neoplastic and inflammatory conditions and is a poor prognostic indicator [36]. It has been

suggested that there is a delicate balance between hypoxiainduced OC activity and hypoxia-induced OC apoptosis, which mediates pathological bone resorption [36]. The difference between chronic and intermittent hypoxia is also evident from studies on osteoclast differentiation under those conditions. Chronic exposure to 2% O2 completely prevents OC formation, whereas cells cultured under a hypoxia-reoxygenation schedule produce more OCs than a normoxic culture [36]. In our hypoxic culture system, cells were exposed to complete reoxygenation when the culture medium was replaced every 3 d, which may enhance OC differentiation. Hypoxia-reoxygenation occurs during injury, ischemia and reperfusion, but it is distinct from hypoxia and produces different responses. Intermittent hypoxia specifically increases cancer cell migration [54] and spontaneous metastasis formation in vivo [55] to levels above that observed with continuous hypoxia. Intermittent hypoxia also influences transcription factor activation. The transcription factor NF-κB is not activated by hypoxia, but it is rapidly and strongly activated upon re-exposure to normal oxygen pressure [56]. Reoxygenation also stimulates HIF-1 signaling. A pool of HIF-1-regulated transcripts remains untranslated in stress granules during hypoxia that are depolymerized upon reoxygenation, allowing translation of the sequestered transcripts [57]. Furthermore, the largest increases in OC number were observed at 3%, 5%, and 10% O2, while 1% O2 completely suppressed OC formation in our culture system. Measurement of pO2 in bone marrow aspirates from normal volunteer donors yielded a mean value of 6.6% [31], but pO2 is often considerably lower under pathological conditions [32]. Although 3%, 5%, and 10% O2 treatments reflect hypoxic treatments compared with the 20% O2 used for normal culture conditions in vitro, it is still controversial whether these lower O2 concentrations suitably reflect physiologically or pathologically relevant hypoxia.

174

E XP ER I ME NTAL C E LL RE S E ARCH

3 21 ( 2 0 14 ) 16 7 – 17 7

Fig. 5 – Q-RT-PCR analysis of the expression of selected mRNAs involved in RANKL and TNF signaling. Human bone marrow cultures were incubated for 21 d in osteoclastogenic culture medium and OCs were separated from the non-adherent cells (non-OCs) using the pronase procedure. Total RNA was prepared from the cell populations and reverse transcribed into cDNA prior to Q-RT-PCR analysis. The expression of β-actin served as a normalization control. The results were expressed as the relative ratio to the expression in non-OCs in the presence of TNF under normoxic conditions in each individual. Each bar represents the mean7S.E.M. (n¼ 7). nPo0.05, compared between cells from the same preparation cultured under normoxic conditions and under hypoxic conditions. # Po0.05, compared between non-OC and OC cells under the same culture conditions. To understand the molecular mechanisms of enhanced OC differentiation with TNF treatment under hypoxia, we performed Q-RT-PCR using total RNA prepared from BMCs treated with TNF under normoxic or hypoxic conditions. Furthermore, to determine whether OCs or other cells, including osteoblasts, expressed hypoxia-induced genes, we separated OCs from non-OC using pronase treatment as previously reported [37,38]. Our data indicate that hypoxia promotes TNFR1 expression non-OCs. To clarify the separation of OC and nonOC in each cell preparation, we compared the expression of RANK, which is expressed in OC, with that of RANKL, which is expressed in non-OC. RANKL was only expressed in the non-OC cell preparation, which indicates that the OC cell population was purified without contamination from non-OC cells. On the other hand, RANK was expressed in both OC and non-OC, which indicated little contamination of OC within non-OC separation. Interestingly, RANKL expression increased in non-OC under hypoxic conditions. This suggests that additional cell-cell interactions occur between pre-OCs and non-OCs via membrane-bound RANKL on non-OCs including osteoblast cells,

and that such an interaction stimulates OC formation via signaling from RANK. In addition, increased exogenous soluble RANKL will also stimulate OC formation. Moreover, TNFR1 was upregulated in non-OC after hypoxic treatment, which suggests that TNF signal transduction is enhanced. Antibody-mediated blockage of TNF signaling inhibited OC formation under hypoxic conditions (Fig. 6). Interestingly, the blocking effects by both antibodies were weaker than the effect of the antibody against TNFR1 alone. These data might suggest that these two receptors not only function independently, but also that they could influence each other via crosstalk between the various signaling pathways initiated by TNFR1 and TNFR2 stimulation [58]. Such crosstalk might indeed contribute to antagonistic effects of the two receptors on other0 s signaling pathways. Specifically, hypoxia treatment might predominantly induce TNFR1 signaling, while suppressing TNFR2 signaling, and together, these effects enhance human osteoclastogenesis. Further studies are thus needed to clarify the crosstalk between TNFR1 and TNFR2 signaling under hypoxic conditions.

E XP E RI ME N TAL CE L L R ES E ARC H

3 21 ( 2 0 14 ) 16 7 – 17 7

175

Fig. 6 – The effects of anti-human TNFR1 and anti-human TNFR2 blocking antibodies on OC differentiation. (A) Human BMCs were preincubated for 30 min with 3 μg/mL anti-TNFR1 or anti-TNFR2 monoclonal antibodies before the addition of TNF, RANKL and M-CSF. The cells were cultured for 21 d, whereupon they were fixed and stained for TRAP activity. The morphological appearance of human bone marrow cultures treated for OC formation for 21 d under hypoxic (5% O2) conditions is shown. (B) TRAP-positive MNCs containing three or more nuclei were counted as human OCs. The results are expressed as the relative ratio to the cell number of OCs in the presence of TNF under hypoxic conditions for each individual. Each bar represents the mean7S.E.M. (n¼ 5). n Po0.05, compared with culture cells under hypoxic conditions in the presence of TNF without anti-TNFR1 or anti-TNFR2 monoclonal antibodies.

In conclusion, hypoxic treatment stimulated TNF-induced OC formation from human BMCs. The expression of RANKL and TNFR1 were both greater in non-OC under hypoxic conditions than under normoxic conditions. Longer initial hypoxic exposure in our hypoxic-reoxygenation culture system enhanced OC formation in the presence of TNF. The blockage of TNF signaling inhibited OC formation under hypoxic conditions. These results suggest that hypoxic exposure plays an important role in TNF-induced osteoclastogenesis from human BMCs.

Conflict of interest All authors state that they have no conflicts of interest.

Acknowledgments This study was supported in part by a Grant-in Aid for Scientific Research (C) from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

references [1] S.L. Teitelbaum, F.P. Ross, Genetic regulation of osteoclast development and function, Nat. Rev. Genet. 4 (2003) 638–649. [2] W.C. Horne, A. Sanjay, A. Bruzzaniti, R. Baron, The role(s) of Src kinase and Cbl proteins in the regulation of osteoclast differentiation and function, Immunol. Rev 208 (2005) 106–125. [3] G. Karsenty, H.M. Kronenberg, C. Settembre, Genetic control of bone formation, Annu. Rev. Cell Dev. Biol. 25 (2009) 629–648. [4] R. Baron, L. Neff, P. Tran Van, J.R. Nefussi, A. Vignery, Kinetic and cytochemical identification of osteoclast precursors and their differentiation into multinucleated osteoclasts, Am. J. Pathol. 122 (1986) 363–378. [5] T.J. Chambers, Regulation of the differentiation and function of osteoclasts, J. Pathol. 192 (2000) 4–13. [6] T. Suda, N. Takahashi, N. Udagawa, E. Jimi, M.T. Gillespie, T.J. Martin, Modulation of osteoclast differentiation and function by the new members of the tumor necrosis factor receptor and ligand families, Endocr. Rev. 20 (1999) 345–357. [7] N. Takahashi, T. Akatsu, N. Udagawa, T. Sasaki, A. Yamaguchi, J.M. Moseley, T.J. Martin, T. Suda, Osteoblastic cells are involved in osteoclast formation, Endocrinology 123 (1988) 2600–2602. [8] W.J. Boyle, W.S. Simonet, D.L. Lacey, Osteoclast differentiation and activation, Nature 423 (2003) 337–342. [9] H. Yasuda, N. Shima, N. Nakagawa, K. Yamaguchi, M. Kinosaki, S. Mochizuki, A. Tomoyasu, K. Yano, M. Goto, A. Murakami, E. Suda, T. Morinaga, K. Higashio, N. Udagawa, N. Takahashi, T. Suda,

176

[10]

[11]

[12]

[13] [14] [15]

[16]

[17]

[18]

[19] [20]

[21] [22]

[23]

[24]

[25]

[26]

[27]

[28]

E XP ER I ME NTAL C E LL RE S E ARCH

Osteoclast differentiation factor is a ligand for osteoprotegerin/ osteoclast genesis-inhibitory factor and is identical to TRANCE/ RANKL, Proc. Natl. Acad. Sci. USA 95 (1998) 3597–3602. D.L. Lacey, E. Timms, H.L. Tan, M.J. Kelley, C.R. Dunstan, T. Burgess, R. Elliott, A. Colombero, G. Elliott, S. Scully, H. Hsu, J. Sullivan, N. Hawkins, E. Davy, C. Capparelli, A. Eli, Y.X. Qian, S. Kaufman, I. Sarosi, V. Shalhoub, G. Senaldi, J. Guo, J. Delaney, W.J. Boyle, Osteoprotegerin ligand is a cytokine that regulates osteoclast differentiation and activation, Cell 93 (1998) 165–176. K. Maeda, Y. Kobayashi, N. Udagawa, S. Uehara, A. Ishihara, T. Mizoguchi, Y. Kikuchi, I. Takada, S. Kato, S. Kani, M. Nishita, K. Marumo, T.J. Martin, Y. Minami, N. Takahashi, Wnt5a-Ror2 signaling between osteoblast-lineage cells and osteoclast precursors enhances osteoclastogenesis, Nat. Med. 18 (2012) 405–412. C. Kaplan, A. Finnegan, Pro-inflammatory cytokines, RANK-L and bone remodeling in rheumatoid arthritis, Front. Biosci. 8 (2003) d1018–d1029. L. Xing, E.M. Schwarz, B.F. Boyce, Osteoclast precursors, RANKL/ RANK, and immunology, Immunol. Rev. 208 (2005) 19–29. S.L. Teitelbaum, Osteoclasts; culprits in inflammatory osteolysis, Arthritis Res. Ther. 8 (2006) 201–209. M.D. Merkel, J.M. Erdmann, K.P. McHugh, Y. Abu-Amer, F.P. Ross, S.L. Teitelbaum, Tumor necrosis factor α mediates orthopedic implant osteolysis, Am. J. Pathol. 154 (1999) 203–210. R. Pacifici, C. Brown, E. Puscheck, E. Friedrich, E. Slatopolsky, D. Maggio, R. McCracken, L.V. Avioli, Effect of surgical menopause and estrogen replacement on cytokine release from human blood mononuclear cells, Proc. Natl. Acad. Sci. USA 88 (1991) 5134–5138. R. Assuma, T. Oates, D. Cochran, S. Amar, D.T. Graves, TNF IL-1, antagonists inhibit the inflammatory response and bone loss in experimental periodontitis, J. Immunol. 160 (1998) 403–409. M. Lewis, L.A. Tartaglia, A. Lee, G.L. Bennett, G.C. Rice, G.H. Wong, E.Y. Chen, D.V. Goeddel, Cloning and expression of cDNAs for two distinct murine tumor necrosis factor receptors demonstrate one receptor is species specific, Proc. Natl. Acad. Sci. USA 88 (1991) 2830–2834. L.A. Tartaglia, D.V. Goeddel, Two TNF receptors Immunol. Today 13 (1992) 151–153. C.A. Smith, T. Farrah, R.G. Goodwin, The TNF receptor superfamily of cellular and viral proteins: activation, costimulation, and death, Cell 76 (1994) 959–962. A.S. Baldwin, The NF-κB and IκB proteins: new discoveries and insights, Annu. Rev. Immunol. 14 (1996) 649–683. I.M. Verma, J.K. Stevenson, E.M. Schwarz, D. Van Antwerp, S. Miyamoto, Rel/NF-κB/IκB family: intimate tales of association and dissociation, Genes Dev. 9 (1995) 2723–2735. L.A. Tartaglia, T.M. Ayres, G.H. Wong, D.V. Goeddel, A novel domain within the 55 kd TNF receptor signals cell death, Cell 74 (1993) 845–853. L.A. Tartaglia, M. Rothe, Y.F. Hu, D.V. Goeddel, Tumor necrosis factor0 s cytotoxic activity is signaled by the p55 TNF receptor, Cell 73 (1993) 213–216. T. Gaber, R. Dziurla, R. Tripmacher, G.R. Burmester, F. Buttgereit, Hypoxia inducible factor (HIF) in rheumatology: low O2! See what HIF can do!, Ann. Rheum. Dis. 64 (2005) 971–980. Y. Watanabe, Y. Terashima, N. Takenaka, M. Kobayashi, T. Matsushita, Prediction of avascular necrosis of the femoral head by measuring intramedullary oxygen tension after femoral neck fracture, J. Orthop. Trauma 21 (2007) 456–461. J. Dunst, S. Ahrens, M. Paulussen, S. Burdach, H. Jurgens, Prognostic impact of tumor perfusion in MR-imaging studies in Ewing tumors, Strahlenther. Onkol. 177 (2001) 153–159. T. Hiraga, S. Kizaka-Kondoh, K. Hirota, M. Hiraoka, T. Yoneda, Hypoxia and hypoxia-inducible factor-1 expression enhance osteolytic bone metastases of breast cancer, Cancer Res. 67 (2007) 4157–4163.

3 21 ( 2 0 14 ) 16 7 – 17 7

[29] H. Motohira, J. Hayashi, J. Tatsumi, M. Tajima, H. Sakagami, K. Shin, Hypoxia and reoxygenation augment bone-resorbing factor production from human periodontal ligament cells, J. Periodontol. 78 (2007) 1803–1809. [30] H.J. Park, K.H. Baek, H.L. Lee, A. Kwon, H.R. Hwang, A.S. Qadir, K.M. Woo, H.M. Ryoo, J.H. Baek, Hypoxia inducible factor-1α directly induces the expression of receptor activator of nuclear factor-κB ligand in periodontal ligament fibroblasts, Mol. Cells 31 (2011) 573–578. [31] J.S. Harrison, P. Rameshwar, V. Chang, P. Bandari, Oxygen saturation in the bone marrow of healthy volunteers, Blood 99 (2002) 394. [32] J.S. Lewis, J.A. Lee, J.C. Underwood, A.L. Harris, C.E. Lewis, Macrophage responses to hypoxia: relevance to disease mechanisms, J. Leukocyte Biol. 66 (1999) 889–900. [33] M. Muzylak, J.S. Price, M.A. Horton, Hypoxia induces giant osteoclast formation and extensive bone resorption in the cat, Calcif. Tissue Int. 79 (2006) 301–309. [34] T.R. Arnett, D.C. Gibbons, J.C. Utting, I.R. Orriss, A. Hoebertz, M. Rosendaal, S. Meghji, Hypoxia is a major stimulator of osteoclast formation and bone resorption, J Cell Physiol. 196 (2003) 2–8. [35] T.R. Arnett, H. Massey, J.C. Utting, I.R. Orriss, A.M. Flanagan, Hypoxia is a major stimulator of osteoclast formation from human peripheral blood, Calcif. Tissue Int. 72 (2003) 345–346. [36] H.J. Knowles, N.A. Athanasou, Acute hypoxia and osteoclast activity: a balance between enhanced resorption and increased apoptosis, J. Pathol. 218 (2009) 256–264. [37] H. Fukuoka, M. Aoyama, K. Miyazawa, K. Asai, S. Goto, Hypoxic stress enhances osteoclast differentiation via increasing IGF2 production by non-osteoclastic cells, Biochem. Biophys. Res. Commun. 328 (2005) 885–894. [38] K. Nakao, M. Aoyama, H. Fukuoka, M. Fujita, K. Miyazawa, K. Asai, S. Goto, IGF2 modulates the microenvironment for osteoclastogenesis, Biochem. Biophys. Res. Commun. 378 (2009) 462–466. [39] S.H. Lee, J. Rho, D. Jeong, J.Y. Sul, T. Kim, N. Kim, J.S. Kang, T. Miyamoto, T. Suda, S.K. Lee, R.J. Pignolo, B. Koczon-Jaremko, J. Lorenzo, Y. Choi, v-ATPase V0 subunit d2-deficient mice exhibit impaired osteoclast fusion and increased bone formation, Nat. Med. 12 (2006) 1403–1409. [40] M. Yagi, T. Miyamoto, Y. Sawatani, K. Iwamoto, N. Hosogane, N. Fujita, K. Morita, K. Ninomiya, T. Suzuki, K. Miyamoto, Y. Oike, M. Takeya, Y. Toyama, T. Suda, DC-STAMP is essential for cell-cell fusion in osteoclasts and foreign body giant cells, J. Exp. Med. 202 (2005) 345–351. [41] J.J. Cody, A.A. Rivera, J. Liu, J.M. Liu, J.T. Douglas, X. Feng, A simplified method for the generation of human osteoclasts in vitro, Int. J. Biochem. Mol. Biol. 2 (2011) 183–189. [42] I. Take, Y. Kobayashi, Y. Yamamoto, H. Tsuboi, T. Ochi, S. Uematsu, N. Okafuji, S. Kurihara, N. Udagawa, N Takahashi, Prostaglandin E2 strongly inhibits human osteoclast formation, Endocrinology 146 (2005) 5204–5214. [43] Y. Kobayashi, T. Mizoguchi, I. Take, S. Kurihara, N. Udagawa, N. Takahashi, Prostaglandin E2 enhances osteoclastic differentiation of precursor cells through protein kinase A-dependent phosphorylation of TAK1, J. Biol. Chem. 280 (2005) 11395–11403. [44] K. Ono, T. Akatsu, T. Murakami, M. Nishikawa, M. Yamamoto, N. Kugai, K. Motoyoshi, N. Nagata, Important role of EP4, a subtype of prostaglandin (PG) E receptor, in osteoclast-like cell formation from mouse bone marrow cells induced by PGE2, J. Endocrinol. 158 (1998) 1–5. [45] Y. Tintut, F. Parhami, A. Tsingotjidou, S. Tetradis, M. Territo, L.L. Demer, 8-Isoprostaglandin E2 enhances receptor-activated NFκB ligand (RANKL)-dependent osteoclastic potential of marrow hematopoietic precursors via the cAMP pathway, J. Biol. Chem. 277 (2002) 14221–14226.

E XP E RI ME N TAL CE L L R ES E ARC H

[46] S. Kotake, Y. Nanke, T. Yago, M. Kawamoto, H. Yamanaka, Human osteoclastogenic T cells and human osteoclastology, Arthritis Rheum. 60 (2009) 3158–3163. [47] Y. Azuma, K. Kaji, R. Katogi, S. Takeshita, A. Kudo, A, Tumor necrosis factor-α induces differentiation of and bone resorption by osteoclasts, J. Biol. Chem. 275 (2000) 4858–4864. [48] K. Kobayashi, N. Takahashi, E. Jimi, N. Udagawa, M. Takami, S. Kotake, N. Nakagawa, M. Kinosaki, K. Yamaguchi, N. Shima, H. Yasuda, T. Morinaga, K. Higashio, T.J. Martin, T. Suda, Tumor necrosis factor alpha stimulates osteoclast differentiation by a mechanism independent of the ODF/RANKL-RANK interaction, J. Exp. Med. 191 (2000) 275–286. [49] B.M. Thomson, G.R. Mundy, T.J. Chambers, Tumor necrosis factors alpha and beta induce osteoblastic cells to stimulate osteoclastic bone resorption, J. Immunol. 138 (1987) 775–779. [50] A.A. Ragab, J.L. Nalepka, Y. Bi, E.M. Greenfield, Cytokines synergistically induce osteoclast differentiation: support by immortalized or normal calvarial cells, Am. J. Physiol. Cell. Physiol. 283 (2002) C679–C687. [51] J. Lam, S. Takehita, J.E. Barker, O. Kanagawa, F.P. Ross, S.L. Teitelbaum, TNF-α induces osteoclastogenesis by direct stimulation of macrophages exposed to permissive levels of RANK ligand, J. Clin. Invest. 106 (2000) 1481–1488. [52] J. Li, I. Sarosi, X.Q. Yan, S. Morony, C. Capparelli, H.L. Tan, S. McCabe, R. Elliott, S. Scully, G. Van, S. Kaufman, S.C. Juan, Y. Sun, J. Tarpley, L. Martin, K. Christensen, J. McCabe, P. Kostenuik, H. Hsu, F. Fletcher, C.R. Dunstan, D.L. Lacey, W.J. Boyle,

3 21 ( 2 0 14 ) 16 7 – 17 7

[53]

[54]

[55]

[56]

[57]

[58]

177

RANK is the intrinsic hematopoietic cell surface receptor that controls osteoclastogenesis and regulation of bone mass and calcium metabolism, Proc. Natl. Acad. Sci. USA 97 (2000) 1566–1571. M. Komine, A. Kukita, T. Kukita, Y. Ogata, T. Hotokebuchi, O. Kohashi, Tumor necrosis factor-α cooperates with receptor activator of nuclear factor κB ligand in generation of osteoclasts in stromal cell-depleted rat bone marrow cell culture, Bone 28 (2001) 474–483. L.M. Postovit, D.E. Abbott, S.L. Payne, W.W. Wheaton, N.V. Margaryan, R. Sullivan, M.K. Jansen, K. Csiszar, M.J. Hendrix, D.A. Kirschmann, Hypoxia/reoxygenation: a dynamic regulator of lysyl oxidase-facilitated breast cancer migration, J. Cell Biochem. 103 (2008) 1369–1378. R.A. Cairns, T. Kalliomaki, R.P. Hill, Acute (cyclic) hypoxia enhances spontaneous metastasis of KHT murine tumors, Cancer Res. 61 (2001) 8903–8908. R.A. Rupec, P.A. Baeuerle, The genomic response of tumor cells to hypoxia and reoxygenation. Differential activation of transcription factors AP-1 and NF-kappa B, Euro. J. Biochem./FEBS 234 (1995) 632–640. B.J. Moeller, Y. Cao, C.Y. Li, M.W. Dewhirst, Radiation activates HIF-1 to regulate vascular radiosensitivity in tumors: role of reoxygenation, free radicals, and stress granules, Cancer Cell 5 (2004) 429–441. P.J. Naudé, J.A. den Boer, P.G. Luiten, U.L. Eisel, Tumor necrosis factor receptor cross-talk, FEBS J. 278 (2011) 888–898.