SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Objective

To investigate why bisphosphonates are less effective at preventing focal bone loss in rheumatoid arthritis (RA) patients than in those with generalized osteoporosis, and the mechanisms involved.

Methods

The response of osteoclasts to alendronate (ALN) in tumor necrosis factor–transgenic (TNF-Tg) mice that develop erosive arthritis and in wild-type littermates was studied. TNF-Tg and wild-type mice were given ALN, and the osteoclast numbers in the inflamed joints and in the long bones were compared. The expression levels of Bcl-xL in the osteoclasts of TNF-Tg and wild-type mice were examined by immunostaining. The effect of overexpression of Bcl-xL and Ets-2 proteins on ALN-induced osteoclast apoptosis was determined using an in vitro osteoclast survival assay and retrovirus transfer approach.

Results

ALN reduced osteoclast numbers in the metaphyses by 97%, but by only 46% in the adjacent inflamed joints. Bcl-xL expression was markedly higher in osteoclasts in the joints than in those in the metaphyses of TNF-Tg mice. Bcl-xL or Ets-2 overexpression protected osteoclasts from ALN-induced apoptosis, and TNF stimulated Bcl-xL and Ets-2 expression in osteoclasts. Overexpression of Ets-2 increased Bcl-xL messenger RNA in osteoclasts, while a dominant-negative form of the Ets-2 blocked the protective effect of Bcl-xL or TNF on ALN-induced apoptosis.

Conclusion

The reduced efficacy of bisphosphonates to stop bone erosion in the inflamed joints of RA patients may result from local high levels of TNF up-regulating Ets-2 expression in osteoclasts, which in turn stimulates Bcl-xL expression in them and reduces their susceptibility to bisphosphonate-induced apoptosis.

Bisphosphonates are used widely to treat bone disorders associated with increased osteoclast activity. As antiresorptive agents, they are very effective in the treatment of postmenopausal osteoporosis. However, several clinical trials have shown that bisphosphonates fail to prevent local bone loss in patients with inflammatory arthritis (1, 2), but the reasons for this poor efficacy have yet to be elucidated. Bisphosphonates reduce bone resorption by two main mechanisms: inhibition of bone resorption and induction of apoptosis of osteoclasts. At low dosages, they suppress osteoclast function, which is associated with increased osteoclast numbers on bone surfaces, perhaps as compensation for decreased bone resorption (3). High dosages of bisphosphonates reduce osteoclast numbers at bone resorption sites by promoting their apoptosis (4, 5). Thus, regulation of osteoclast apoptosis directly affects the ability of osteoclasts to resorb bone.

Two distinct signaling pathways control cell apoptosis: one is plasma membrane receptor dependent and the other is triggered by intracellular stress. Bisphosphonate-induced apoptosis is triggered by intracellular stress as a consequence of disrupted cholesterol biosynthesis (6, 7). In osteoclasts, this pathway is regulated by the Bcl-2 family of proteins and involves mitochondrial release of cytochrome c, which leads to the activation of caspases 9 and 3 (8). Bcl-2 family members consist of pro- and antiapoptotic proteins, such as Bcl-2, Bcl-xL, Bax, and Bid. The relative ratio and activation states of these molecules determine the fate of cells (9, 10). Osteoclasts express higher levels of Bcl-xL protein than Bcl-2, and thus Bcl-xL is believed to be an important regulator of apoptosis in these cells. Macrophage colony-stimulating factor (M-CSF), a key survival factor for cells in the myeloid lineage, stimulates Bcl-xL expression (11, 12). Osteoclasts generated from bone marrow cells of transgenic mice that overexpress Bcl-xL under the control of the tartrate-resistant acid phosphatase (TRAP) promoter are more resistant to serum withdrawal–induced cell death (13, 14), demonstrating that Bcl-xL is involved in the control of osteoclast apoptosis.

The expression of Bcl-xL is regulated by the transcription factors NF-κB, activator protein 1 (AP-1), and the Ets family members, mainly Ets-1, Ets-2, and PU.1 (15–17). Interestingly, gene knockout studies have demonstrated that NF-κB, c-Fos, and PU.1 are all essential for osteoclastogenesis during development (18, 19). Both NF-κB and c-Fos are critical for the differentiation of CD11b+/c-Fms+/TRAP-osteoclast precursors to TRAP+ osteoclasts, while PU.1 works at a much earlier stage and is essential for the commitment of multipotent cells to become myeloid progenitors (20).

While it appears that these transcription factors do not play a significant role in osteoclast survival (5), the role of Ets-2 in osteoclasts has not been well studied. Ets-2−/− mice die before day 8.5 of gestation due to defective trophoblast function (21). After birth, Ets-2 is highly expressed in late-stage myeloid cells, and its expression is rapidly up-regulated upon induction of macrophage differentiation from progenitors. Ets-2 also mediates M-CSF–dependent macrophage survival through regulation of Bcl-xL expression (11). Recently, increased expression of Ets protein has been observed in blood and synovial samples from arthritic animals and patients, suggesting that cytokines, such as tumor necrosis factor (TNF), which are increased in blood and synovial fluid of patients with inflammatory arthritis, may stimulate the expression of Ets family members (22–24). Thus, it is possible that in erosive inflammatory arthritis, high local levels of TNFα stimulate expression of Ets members in osteoclasts and their precursors, triggering Bcl-xL survival signals that render osteoclasts more resistant to bisphosphonate-induced cell death.

The purpose of this study was to test the hypothesis that the inflammatory microenvironment alters the apoptosis machinery in osteoclasts through the TNF/Ets-2/Bcl-xL pathway, leading to decreased susceptibility to bisphosphonate-induced injury. We used alendronate (ALN), a widely administered bisphosphonate, and TNF-transgenic mice (TNF-Tg) as an animal model of chronic inflammatory, erosive arthritis. ALN-induced apoptosis and expression levels of Bcl-xL were compared between osteoclasts on the eroded joint surfaces of TNF-Tg mice and those in the adjacent metaphyses. The effect of Bcl-xL and Ets-2 overexpression on ALN-induced apoptosis was assessed using a retroviral transfer approach and by proinflammatory cytokine treatment of osteoclast cultures in vitro. The relationship between TNF, Bcl-xL, and Ets-2 was determined, and our findings indicate that under inflammatory conditions, TNF stimulates osteoclasts to up-regulate Ets-2 expression, which leads to increased Bcl-xL expression and osteoclast resistance to ALN-induced apoptosis.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Reagents and animals.

Recombinant murine TNFα, M-CSF, and RANKL were purchased from R&D Systems (Minneapolis, MN). ALN (2,[3-pyridinyl-]1-hydroxyethylidene-1, 1-bisphosphonate) was a gift from Dr. Tomi Sawyer (ARIAD Pharmaceuticals, Cambridge, MA). TNF-Tg mice (3647 TNF-Tg line) in a CBA × C57BL/6 background (25) were obtained from Dr. G. Kollias (Vari, Greece) and they were genotyped by polymerase chain reaction (PCR) of tail samples. The Institutional Animal Care and Use Committee of the University of Rochester approved all studies.

Retroviral vectors.

The coding regions of respective genes were amplified by PCR from complementary DNA (cDNA) and cloned into the pMX-puro retroviral vector (26). Each 5′ primer contains a Kozak sequence following the start codon; c-Fos and NF-κB p65 cDNA were of murine origin, and Bcl-xL, Ets-2, and ΔEts-2 were of human origin. A 1.1-kilobase sequence corresponding to the DNA binding domain of Ets-2 and functioning as a dominant-negative mutant is present in ΔEts-2, as described previously (27). These retrovirus vectors were transfected into the Plat-E retroviral packaging cell line (28), and viral supernatant was collected 48 hours later.

Osteoclastogenesis and viral infection.

To generate osteoclast precursors, freshly isolated wild-type (WT) bone marrow cells were cultured for 3 days with M-CSF conditioned medium in α-modified essential medium (Gibco BRL, Grand Island, NY) supplemented with 10% fetal calf serum (Hyclone Laboratories, Logan, UT). Adherent cells were 90% CD11b+ and c-Fms+ by fluorescence-activated cell sorting analysis, and were used as osteoclast precursors.

For retroviral infection, the osteoclast precursors were infected with viral supernatants in the presence of M-CSF (25 ng/ml) and polybrene (2 μg/ml). On day 2 of infection, puromycin (2 μg/ml) was added into the culture to select gene-integrated cells. The cells were then cultured with M-CSF (10 ng/ml) and RANKL (5 ng/ml) for 5 or 6 days at 37°C in an atmosphere of 5% CO2/95% air, and were fed every 2 days by replacing half of the spent medium with fresh medium and M-CSF and RANKL. The cells were fixed and stained for TRAP activity to identify osteoclasts. Mature osteoclasts were considered to be TRAP+ cells containing ≥3 nuclei. For osteoclast survival, mature osteoclasts were treated with ALN (20–50 μM) with or without TNFα (20 ng/ml) for 24 hours. The number of TRAP+ viable and apoptotic osteoclasts was counted, as we have described previously (29).

Preparation and histomorphometry of bone sections.

After mice were killed, the limbs were removed, fixed in 10% buffered formalin, decalcified in 10% EDTA, and embedded in paraffin. Sections (5 μm thick) were then stained for TRAP activity and counterstained with hematoxylin and eosin. Histomorphometric analysis was performed in sections from TNF-Tg mice treated with ALN or phosphate buffered saline (PBS), using Osteomeasure image analysis software (Osteometrics, Atlanta, GA) as previously described (30). Osteoclast numbers were expressed per tibia in longitudinal sections and per mm of eroded knee joint surface.

Immunohistochemistry.

Deparaffinized sections were quenched with 3% hydrogen peroxide and treated for 30 minutes for antigen retrieval. Sections were then stained with a rabbit anti–Bcl-xL antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and a biotinylated goat anti-rabbit secondary antibody (Dako, Carpinteria, CA). The biotin was detected using standard avidin–biotin–peroxidase technology (Dako) with diaminobenzidine (Sigma, St. Louis, MO) as the chromogen. An adjacent section was used for TRAP staining, as described previously (5).

In vivo apoptosis analysis.

Four-month-old TNF-Tg and WT mice were given intraperitoneal injections of ALN (10 mg/kg) daily for 4 days. The mice were killed by CO2 overdose and cervical dislocation 24 hours after the last ALN injection, and the fore limbs were fixed and processed for histologic analysis. Decalcified sections were prepared and stained for TRAP activity, and morphologically viable and apoptotic osteoclasts were counted in the metaphyseal area and arthritic knee joints at the bone–pannus junction. Apoptotic osteoclasts were identified in TRAP-stained bone sections using standard morphologic criteria: cell shrinkage, nuclear condensation and fragmentation, cytoplasmic condensation and intense cytoplasmic TRAP staining, as described previously (13). It has been previously confirmed, using acridine orange staining and the TUNEL assay (4, 13), that TRAP+ osteoclasts with these appearances in culture plates and in tissue sections are apoptotic.

Real-time PCR.

Extraction of RNA from osteoclast precursors or infected cells was performed using the RNeasy kit and the QiaShredder (Qiagen, Valencia, CA). As described previously (5), cDNA synthesis was performed. Quantitative PCR amplification was performed with gene-specific primers using a Rotor-Gene 2000 real-time amplification operator (Corbett Research, Mortlake, Australia). The primer sequences included the following: c-Fos forward 5′-TGACAGTTGGACCCAAGACA-3′, reverse 5′-CTCCATTCAGTCACCCCAGT-3′; NF-κB p65 forward 5′-GGGTCAGTGTGACCGAAGAT-3′, reverse 5′-GGAAGTCAGAAGTGGGTGGA-3′; Bcl-xL forward 5′-CGGAGAGCGTTCAGTGATCT-3′, reverse 5′-TGCAATCCGACTCCCAATA-3′; Bcl-2 forward 5′-GTATGATAACCGGGAGATCGTGATGAAG-3′, reverse 5′-GGACACATGGTGGCACAGG-3′; Bax forward 5′-TGCAGAGGATGATTGCTGAC-3′, reverse 5′-GATCAGCTCGGGCACTTTAG-3′; Bid forward 5′-TCACAGACCTGCTGGTGTTC-3′, reverse 5′-GTCTGGCAATGTTGTGGATG-3′; Ets-1 forward 5′-TCCAGACAGACACCTTGCAG-3′, reverse 5′-GGTGAGGCGGTCACAACTAT-3′; Ets-2 forward 5′-AATGCAGGCACCAAACTACC-3′, reverse 5′-GTCCTGGCTGATGGAACAGT-3′; PU.1 forward 5′-GGCAGCAAGAAAAAGATTCG-3′, reverse 5′-TTTCTTCACCTCGCCTGTCT-3′; and actin forward 5′-AGATGTGGATCAGCAAGCAG-3′, reverse 5′-GCGCAAGTTAGGTTTTGTCA-3′. The relative standard curve method was used to calculate the amplification difference for each primer set (31). The standard curve was made from 4 points corresponding to 10-fold cDNA dilution series for each gene. For each sample, the relative amount was calculated from the respective standard curves.

Western blot analysis.

Infected cells were lysed with radioimmunoprecipitation assay buffer (50 mM HEPES [pH 7], 1% Triton X-100, 1 mM EDTA, 0.1% sodium dodecyl sulfate, 1% sodium deoxycholate, 150 mM NaCl with protease inhibitors, and sodium orthovanadate). The lysates (20 μg of protein) were immunoblotted with rabbit anti–Bcl-xL (Santa Cruz Biotechnology) or mouse antiactin antibody (Sigma).

Caspase 3 assay.

Infected cells were treated with ALN (20 μM) for 24 hours and then lysed in Triton X-100 lysis buffer (20 mM HEPES [pH 7.4], 1% Triton X-100, 2 mM EDTA, and protease inhibitors). The caspase 3 activity was measured using Ac-DEVD-AMC fluorogenic substrate (N-acetyl-Asp-Glu-Val-Asp-7-amino-4-methylcoumarin) in a colorimetric reaction. Fluorescence was read at an excitation wavelength of 360 nm and an emission wavelength of 460 nm. An AMC standard curve (starting with 10 μM and making 2-fold serial dilutions) was included.

Statistical analysis.

All results are given as the mean and SEM. Comparisons were made by analysis of variance and Student's t-test for unpaired data. P values less than 0.05 were considered significant.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Increased resistance to ALN-mediated apoptosis and higher Bcl-xL expression in osteoclasts in the inflamed joints than in the metaphyses of TNF-Tg mice.

In the absence of any inflammatory signals, osteoclasts are typically found in greatest numbers in growing mice in the metaphyseal region of long bones, where they are involved in the modeling and remodeling of the bone. In the joints of TNF-Tg mice, osteoclasts are also found at sites of inflammation-mediated focal erosions. To examine whether osteoclasts in these focal erosions respond differently to bisphosphonate-induced cell death compared with those involved in normal bone modeling and remodeling, we treated 4-month-old TNF-Tg mice with established arthritis and WT littermates with ALN (10 mg/kg/day for 4 days), a protocol that we have used previously to induce osteoclast apoptosis in vivo (5). Consistent with previous observations, ALN killed most of the osteoclasts in the tibial metaphyses of WT and TNF-Tg mice (Figure 1). However, the reduction of the number of osteoclasts in the arthritic knee joints of ALN-injected TNF-Tg mice was significantly less than that in the metaphyses (percent reduction of osteoclasts compared with PBS treatment: 97% in metaphyses versus 46% in knee joints) (Figures 1A–D).

thumbnail image

Figure 1. Effect of alendronate (ALN) on osteoclast numbers in the metaphysis and in the adjacent inflamed joint surfaces of tumor necrosis factor–transgenic (TNF-Tg) mice. Four-month-old TNF-Tg mice with established arthritis and their wild-type (WT) littermates (n = 5 in each group) were treated with phosphate buffered saline (PBS) or ALN (10 mg/kg/day) for 4 days. After 4 days, the animals were killed, sections of their long bones were prepared and stained for tartrate-resistant acid phosphatase (TRAP) activity, and histomorphometric analysis was performed. Shown are TRAP-stained sections of the metaphysis (green arrows) and knee joint (black arrows) of a PBS-treated (A) or ALN-treated (B) TNF-Tg mouse and of a PBS-treated (C) or ALN-treated (D) WT mouse (original magnification × 2). High-power micrographs (original magnification × 20) show the potent effects of ALN on metaphyseal osteoclasts (E) (97% apoptosis) versus its limited effects on osteoclasts at the erosion front (F) (46% apoptosis) in TNF-Tg mice. Note the typical appearance of TRAP-stained osteoclast apoptotic bodies (yellow arrows) compared with those of viable polarized osteoclasts (red arrows). The number of viable TRAP+ osteoclasts (Ocls) was counted in the metaphysis and eroded joint surfaces of PBS- and ALN-treated TNF-Tg mice (G). Values are the mean and SEM of 5 mice per group, and are representative of 2 independent experiments. ∗ = P < 0.05 versus metaphysis in ALN-treated mice.

Download figure to PowerPoint

Since ALN reduces osteoclast numbers by inducing their apoptosis, we compared the number of apoptotic osteoclasts in TRAP-stained sections of the arthritic joints with those of the tibial metaphyses in ALN-treated TNF-Tg mice. The apoptotic osteoclasts were identified using standard morphologic criteria: cell shrinkage, nuclear condensation and fragmentation, cytoplasmic condensation and intense cytoplasmic TRAP staining, as described previously (5). We observed many more apoptotic osteoclasts in the metaphyseal areas versus the arthritic joints (Figures 1E and F), suggesting that osteoclasts in the inflamed sites of the TNF-Tg mice have enhanced survival potential.

To determine if the ALN-resistant osteoclasts express higher levels of cell survival proteins, we examined Bcl-xL expression by immunohistochemistry. Bcl-xL was highly expressed in TRAP+ osteoclasts in the arthritic joints of TNF-Tg mice (Figure 2). In contrast, expression of Bcl-xL in osteoclasts in the metaphyses of these mice was low or undetectable, while high expression was seen in nearby osteoblasts within the same sections. Similarly, we observed strong Bcl-xL staining in TRAP+ osteoclasts in specimens taken from inflamed joints of rheumatoid arthritis (RA) patients, and these cells also expressed TNF (results not shown). Osteoblasts in the same sections were Bcl-xL and TNF positive, but TRAP negative. These observations demonstrate that osteoclasts in the inflamed joints of TNF-Tg mice and RA patients express elevated levels of Bcl-xL protein.

thumbnail image

Figure 2. Bcl-xL expression in osteoclasts on eroded joint surfaces of tumor necrosis factor–transgenic (TNF-Tg) mice. Long bone sections from 4-month-old TNF-Tg mice were immunostained for Bcl-xL (A and B), and adjacent sections were stained for tartrate-resistant acid phosphatase (TRAP) activity (C and D). High expression of Bcl-xL is seen in osteoclasts (black arrows) at the eroded joint surface (A and C), but not in osteoclasts at the base of the growth plate (B and D), where high expression is seen in osteoblasts (yellow arrows). Results are representative of 5 pairs of TNF-Tg and wild-type mice. (Original magnification × 20.)

Download figure to PowerPoint

Bcl-xL overexpression prolongs survival of osteoclasts and protects them from ALN-induced apoptosis in vitro.

To test if Bcl-xL overexpression affects osteoclast survival, we cloned the human bcl-xL cDNA into the pMX-puro retrovirus vector and infected WT osteoclast precursors with retroviral supernatants containing Bcl-xL or green fluorescent protein (GFP) virus. Two days after viral infection and puromycin selection, cells were treated with M-CSF and RANKL for an additional 5 days to induce osteoclast formation. Under these conditions, 90% of cells are infected, as indicated by GFP positivity. After the formation of multinucleated osteoclasts was observed by inverted microscopy, the medium in the cultures was replaced with fresh medium without M-CSF and RANKL. Our preliminary experiments showed that under these culture conditions, fully formed mature osteoclasts began to die by 8 hours. Thus, between 8 and 32 hours after M-CSF and RANKL were withdrawn, the cells were fixed and the numbers of viable and apoptotic osteoclasts were determined according to previously described morphologic criteria (4, 11).

Bcl-xL overexpression significantly reduced the percentage of apoptotic osteoclasts compared with GFP virus controls and also prolonged mature osteoclast lifespan (Figure 3). Eighty percent of Bcl-xL–infected cells survived for at least 20 hours after M-CSF and RANKL withdrawal, while almost all control cells were dead after 20 hours. To confirm Bcl-xL overexpression, total RNA and protein were obtained from Bcl-xL– or GFP-infected osteoclasts, and real-time PCR and Western blot analyses were performed. Compared with GFP-infected cells, Bcl-xL–infected cells expressed much higher levels of bcl-xL messenger RNA (mRNA) and protein (Figure 3C), indicating successful overexpression of the bcl-xL viral transgene.

thumbnail image

Figure 3. Effect of Bcl-xL overexpression on osteoclast survival. Wild-type bone marrow cells were cultured with macrophage colony-stimulating factor (M-CSF) for 3 days to enrich osteoclast precursor cells. The cells were then infected with green fluorescent protein (GFP) or Bcl-xL retroviral supernatants and treated with M-CSF and RANKL to generate mature osteoclasts. After microscopic confirmation of the presence of multinucleated mature osteoclasts, the culture medium was replaced with medium without M-CSF and RANKL for various times before tartrate-resistant acid phosphatase (TRAP) staining. A, The percent of apoptotic osteoclasts (Ocls) in these cultures is presented as the mean and SEM from 4 culture wells. ∗ = P < 0.05 versus GFP-infected group. Similar results were obtained in 2 additional independent experiments. B, Representative photomicrographs of TRAP-stained cultures taken at 8 hours after M-CSF and RANKL withdrawal (original magnification × 4). Note the outline of osteoclasts that have undergone apoptosis in the GFP-infected cultures (arrows) compared with the live osteoclasts in the Bcl-xL–infected cultures. C, Overexpression of the Bcl-xL transgene was confirmed by real-time polymerase chain reaction and Western blotting. Quantification of these results is shown as the mean and SEM of 3 samples. Color figure can be viewed in the online issue, which is available at http://arthritisrheum.org.

Download figure to PowerPoint

To examine the effect of Bcl-xL overexpression on ALN-induced osteoclast apoptosis, Bcl-xL– or GFP-infected osteoclasts were treated with various concentrations of ALN (Figure 4A). Numerous Bcl-xL–infected osteoclasts (37%) survived after treatment with ALN (20 μM), compared with GFP-infected cells (7%), indicating that Bcl-xL-infected osteoclasts were significantly more resistant to ALN-induced apoptosis. Since TNF is highly expressed by inflammatory cells, its effect on ALN-induced death was examined using a similar protocol. We confirmed the previous report (32) that TNF treatment significantly attenuated the apoptosis-inducing effect of ALN (Figure 4B).

thumbnail image

Figure 4. Effect of Bcl-xL overexpression and tumor necrosis factor (TNF) treatment on alendronate (ALN)–induced osteoclast (Ocls) apoptosis. Bcl-xL– or green fluorescent protein (GFP)–infected mature osteoclasts were generated as described in Figure 3. The cells were treated with ALN (0, 20, or 50 μM) for 24 hours (A), or mature osteoclasts were treated with ALN with or without TNF (20 ng/ml) for 24 hours (B). Cells were then fixed and stained for tartrate-resistant acid phosphatase activity. The percent of viable osteoclasts (upper panels) and apoptotic osteoclasts (lower panels) is presented as the mean and SEM from 4 culture wells. ∗ = P < 0.05 versus the GFP-infected group or versus phosphate buffered saline (PBS)–treated cells, with ALN at the same concentration. Similar results were obtained in 3 additional independent experiments.

Download figure to PowerPoint

TNFα-increased Bcl-xL expression in WT osteoclast precursors.

To examine if TNF promotes osteoclast survival by up-regulating Bcl-xL expression, WT osteoclast precursors were treated with TNF for various times and bcl-xL mRNA expression was assessed by real-time PCR. TNF (20 ng/ml) significantly increased bcl-xL expression in these cells by 2 hours, and its effect peaked at 8 hours (Figure 5A). Furthermore, Bcl-xL protein levels were also increased, as shown by Western blot analysis (Figure 5B). The influence of TNF on other Bcl-2 family proteins, including Bcl-2, Bax, and Bid, was examined using the same RNA samples. TNF slightly reduced Bcl-2 and Bax mRNA by 1-fold and did not affect Bid expression significantly (Figure 5A). These data suggest that high local concentrations of TNF may increase Bcl-xL expression in osteoclasts and thus enhance their survival.

thumbnail image

Figure 5. Effect of tumor necrosis factor (TNF) on Bcl-xL expression in osteoclast precursors. Wild-type marrow cells were cultured with macrophage colony-stimulating factor for 3 days to induce osteoclast precursor formation and then treated with TNF. The expression levels of bcl-xL, bcl-2, bax, bid, NF-κB p65, c-fos, ets-1, ets-2, and pu.1 mRNA were examined by real-time polymerase chain reaction at various times and are presented in arbitrary units standardized to a β-actin control as described in Materials and Methods. A, Data are the mean and SEM from 3 samples. Similar results were obtained in 2 additional independent experiments. B, Protein levels of Bcl-xL were examined by Western blotting 24 hours after TNF treatment and compared with levels of actin as a loading control. Results are representative of 2 independent experiments. PBS = phosphate buffered saline.

Download figure to PowerPoint

Ets-2 mediation of TNFα-induced Bcl-xL expression in osteoclast precursors.

The bcl-x promoter contains functional binding sites for NF-κB, AP-1, and Ets member transcription factors (11). In osteoclasts, TNF is a strong activator of NF-κB and c-Fos (33–36), but it is not known if it regulates Ets expression. To determine which Ets factors mediate TNF-induced Bcl-xL expression, we treated osteoclast precursors with TNF and examined the expression levels of c-fos, NF-κB, ets-1, ets-2, and PU.1 by real-time PCR. As expected, TNF increased NF-κB and c-fos expression. Among the 3 Ets family members, only ets-2 mRNA levels increased significantly in response to TNF (Figure 5A). Of note, the expression levels of ets-1 were very low in these cells, compared with ets-2 and PU.1.

To determine if overexpression of NF-κB, c-Fos, or Ets-2 could enhance osteoclast survival in the presence of ALN, WT osteoclast precursors were infected with retroviral supernatants containing NF-κB p65, c-Fos, or Ets-2 retrovirus and cultured with M-CSF and RANKL to form osteoclasts. The mature osteoclasts were then challenged with ALN for 24 hours. Under these conditions, Ets-2–infected cells, but not NF-κB– or c-Fos–infected cells, were more resistant to ALN-induced apoptosis (Figure 6A). The protective effect of Ets-2 was confirmed by measuring caspase 3 activity, an indicator of cell apoptosis. ALN doubled caspase 3 activity, which was abolished in Bcl-xL– or Ets-2–overexpressing cells. To ensure that the absence of increased survival in the NF-κB and c-Fos over-expressing osteoclasts was not due to low expression levels, we examined transgene expression by real-time PCR. All 3 genes, including NF-κB p65, c-Fos, and Ets-2, were overexpressed at comparable levels (results not shown).

thumbnail image

Figure 6. Effect of overexpression of Ets-2 or a dominant-negative mutant of Ets-2 (ΔEts-2) on tumor necrosis factor (TNF)–mediated osteoclast survival and Bcl-xL expression. A, Mature osteoclasts (Ocls) were infected with Ets-2, NF-κB p65, or c-Fos retroviral supernatants, treated with alendronate (ALN) (20 μM) for 24 hours, and then stained for tartrate-resistant acid phosphatase (TRAP) activity to determine the number of viable and apoptotic osteoclasts in the cultures. Data are the mean and SEM from 4 culture wells. ∗ = P < 0.05 versus the green fluorescent protein (GFP)–infected group. Similar results were obtained in 2 additional independent experiments. B, At the indicated times, bcl-xL mRNA levels were determined in Ets-2–overexpressing cells by real-time polymerase chain reaction (PCR). Values are the mean and SEM from 3 samples. ∗ = P < 0.05 versus GFP-infected cells at the indicated times. Data are representative of 2 independent experiments. C, The ΔEts-2– or GFP virus–infected mature osteoclasts were treated with ALN (20 μM) in the presence of TNF (20 ng/ml) or phosphate buffered saline (PBS) for 24 hours. The percent of apoptotic osteoclasts was determined in TRAP-stained plates, and values are presented as the mean and SEM from 4 culture wells. Bcl-xL mRNA expression was examined by real-time PCR, and values are presented as the mean and SEM from 3 samples. ∗ = P < 0.05 versus the TNF-treated, GFP-infected group. Data are representative of 2 independent experiments.

Download figure to PowerPoint

Bcl-xL has been shown to be a direct target gene of the Ets-2 transcription factor in myeloid cells (11). To determine if bcl-xL is a target gene of Ets-2 in osteoclasts, levels of bcl-xL mRNA were examined in Ets-2–infected cells. Ets-2 overexpression increased bcl-xL expression in a time-dependent manner (Figure 6B). Increased ets-2 mRNA levels in the Ets-2–infected cells, but not in the Bcl-xL– or GFP-infected cells, confirmed overexpression of the Ets-2 retrovirus (results not shown). Finally, to functionally demonstrate that Ets-2 mediates TNF-induced Bcl-xL expression and osteoclast survival, loss-of-function experiments were performed using a dominant-negative form of the Ets-2 transgene (ΔEts-2), which was used previously to further accelerate M-CSF withdrawal–induced macrophage apoptosis (11). Overexpression of ΔEts-2 blocked the protective effect of TNF on ALN-induced osteoclast death (Figure 6C). Accordingly, TNF-induced Bcl-xL expression was also abolished in these cells. Taken together, these findings indicate that increased Ets-2 expression by TNF up-regulates expression of the cell survival protein Bcl-xL, which makes osteoclasts more resistant to ALN-induced apoptosis.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The current study was designed to determine why bisphosphonates are less effective at preventing focal bone loss in patients with inflammatory bone diseases than in patients with osteoporosis. To this end, we examined the response of osteoclasts to ALN in TNF-Tg mice that have established erosive arthritis. We found that ALN diminishes osteoclast numbers in the metaphyses by 97% in both TNF-Tg and WT mice, but by only 46% at sites of focal joint erosion. Expression of the antiapoptotic protein, Bcl-xL, is significantly increased in osteoclasts in the diseased joints of TNF-Tg mice (Figure 2) and patients with RA (results not shown), suggesting that the local environment favors osteoclast survival. In support of this hypothesis, we demonstrated that 1) TNF stimulates Bcl-xL and Ets-2 expression; 2) Bcl-xL overexpression reduces ALN-induced apoptosis; and 3) Ets-2 mediates TNF-induced Bcl-xL expression. Thus, we propose a model to explain the reduced efficacy of bisphosphonate in the treatment of chronic inflammatory bone diseases (1, 2), in which TNF up-regulates Ets-2 expression in osteoclast precursors, which stimulates transcription of the bcl-xL gene and makes osteoclasts more resistant to apoptotic signals.

Our model emphasizes the importance of the microenvironment in determining the response of osteoclasts to certain therapies. Increased osteoclast survival capacity in inflamed joints indicates that a higher concentration or more potent drug may be required locally in arthritic joints. To support this, we show that TNF treatment and Bcl-xL overexpression produce protective effects in osteoclasts only when 20 μM or lower concentrations of ALN are used, but not at high concentrations (Figure 4). Since the local ALN concentration in the inflamed joints and metaphyses of ALN-treated TNF-Tg mice is not known, we cannot exclude the possibility that the concentration of ALN in the joints is lower than that in the growth plate of long bones, given the high affinity of bisphosphonate for bone (36). However, the fact that metaphyseal and joint osteoclasts are separated by only hundreds of microns, and inflamed sites tend to have increased blood flow and thus exposure to potentially higher bisphosphonate concentrations, makes this highly unlikely. Increased Bcl-xL staining in osteoclasts in the inflamed joints and resistance to ALN-induced death of Bcl-xL-infected osteoclasts indicate that inflammation-induced changes within osteoclasts themselves contribute at least in part to the failure of bisphosphonate to prevent local bone loss in inflammatory bone diseases, suggesting a fundamental difference between osteoclasts generated by homeostatic versus inflammatory signals in our model.

We have focused on Bcl-xL in the current study because its role in osteoclast survival has been repeatedly documented (12, 14, 17, 37). In addition, we found that TNF stimulates expression of Bcl-xL, but not the other Bcl-2 family members we tested (Figure 5). However, other Bcl-2 family proteins that are involved in the regulation of cell fate may also mediate TNF-induced osteoclast survival. For instance, Bim, a BH3-only proapoptotic protein, restricts osteoclast lifespan, and its expression is suppressed by M-CSF (38). TNF stimulates the expression of cellular inhibitor of apoptosis 1, an inhibitor of apoptotic proteins, in osteoclast precursors through the NF-κB pathway (39). It will be important to investigate the role of these proteins in osteoclast survival in inflammatory bone disorders.

Our finding that NF-κB and c-Fos overexpression failed to promote osteoclast survival is not surprising given that osteoclasts can be generated from the splenocytes of both NF-κB p50/p52–double-knockout and c-Fos–knockout mice when the knockout cells are infected with certain proteins, such as nuclear factor of activated T cells 2. Miyazaki et al have demonstrated that inhibition of NF-κB activation in osteoclasts by overexpressing the NF-κB superinhibitor, IκBα, does not induce cell death (40). Thus, the major function of NF-κB and c-Fos in osteoclast biology is to control the differentiation of osteoclast precursor cells.

Of the 30 Ets family members, Ets-1, Ets-2 and PU.1 have been reported to be involved in bone cell functions (17). Ets-1 interacts with CbFA1 to regulate osteogenesis (41); Ets-2 is required for cartilage and intramembranous bone formation (42); and PU.1 is essential for osteoclastogenesis (20). Additionally, all of these Ets family members transactivate the bcl-x promoter (11, 37). However, in the current study, along with our unpublished observations, we found that ets-1 mRNA levels are very low in osteoclast precursors and undetectable in mature osteoclasts. While PU.1 is abundantly expressed in osteoclasts, its expression is not regulated by TNF, M-CSF, or RANKL stimulation. Consistent with our findings, So et al recently demonstrated that PU.1 expression is not induced during osteoclast formation, unlike microphthalmia transcription factor and osteoclast-associated receptor (43). These results indicate that Ets-1 and PU.1 are unlikely to play a significant role in mature osteoclast survival.

The significance of our finding that TNF stimulates Ets-2 expression is the implication that Ets-2 may be involved in the regulation of osteoclast function during inflammation-related bone loss. Indeed, Ets-2 is not critical for survival of osteoclasts or osteoclast precursors during development, because transgenic mice carrying a dominant-negative form of Ets-2 under control of the c-Fms promoter have a normal bone phenotype (44). However, it is possible that in pathologic conditions, high levels of Ets-2 alter the intrinsic apoptotic machinery in osteoclasts, leading to enhanced survival, which is supported by increased ets-2 mRNA in osteoclast precursors isolated from TNF-Tg mice (mean ± SEM ets-2:actin ratio of 4.9 ± 0.48 in cells of TNF-Tg mice versus 1 ± 0.05 in cells of WT mice). The TNF signaling pathway used to activate Ets-2 is not clear. TNF activates Akt, Jun, and ERK signaling pathways in osteoclasts, and all of them have been reported to link to Ets-2 activation in various cell types (45–47). We are currently investigating the specific pathway by which TNF mediates Ets-2 activation in osteoclasts.

In summary, we have provided experimental evidence to explain the decreased efficacy of bisphosphonates in preventing local inflammatory bone loss due to changes in the joint microenvironment that favor osteoclast survival and involve both paracrine and autocrine mechanisms. However, our findings do not eliminate the possibility of therapeutic efficacy of bisphosphonates in inflammatory bone diseases, because ALN treatment did reduce osteoclast numbers in inflamed joints, although to a lesser degree compared with cells inside the long bones (Figures 1 and 2). Therefore, more potent bisphosphonates, such as zoledronate, may be efficacious. Recently, 2 groups have reported that zoledronate effectively prevents the bone erosion that occurs in the inflamed joints of TNF-Tg mice or animals with collagen-induced arthritis (48, 49). Unfortunately, in those studies, the authors focused on investigating zoledronate-induced inhibition of osteoclast function, rather than survival. Nevertheless, the effectiveness of zoledronate in preventing focal bone erosion is yet undetermined in patients with inflammatory arthritis and awaits the results of clinical trials. Meanwhile, it may be worth examining the efficacy of intraarticular injections of bisphosphonates to achieve sufficiently high local concentrations to prevent joint erosion.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We would like to thank Dr. G. Cheng (University of California, Los Angeles) for providing the Bcl-x WT and Bcl-x κB vectors, Dr. R. Sakamuri (University of Pittsburgh, Pittsburgh, PA) for providing the human Bcl-xL cDNA, Dr. K. Matsuo (Keio University, Tokyo, Japan) for the retrovirus system, and Dr. S. Takeshita (National Institute for Longevity Sciences, Azchi, Japan) for the M-CSF–producing cell line. The authors would also like to thank Bianai Fan for technical assistance with the histologic analysis.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES