Chronic metabolic acidosis stimulates net calcium efflux from bone due to increased osteoclastic bone resorption and decreased osteoblastic collagen synthesis. Previously, we determined that incubation of neonatal mouse calvariae in medium simulating physiological metabolic acidosis leads to a significant, cyclooxygenase-dependent, increase in RNA for bone cell receptor activator of NF-κB ligand (RANKL) compared with incubation in neutral pH medium. In this study, we tested the hypothesis that the acid-mediated increase in RANKL expression is a primary mechanism for the stimulated osteoclastic resorption. Acid medium increased the medium concentration of sRANKL without altering the concentration of the decoy receptor osteoprotegerin (OPG). Inhibition of the RANKL pathway with concentrations of OPG up to 25 ng/ml, far greater than physiological, did not significantly decrease the robust acid-induced Ca efflux from bone nor did incubation of the calvariae with a different inhibitor, RANK/Fc (up to 50 ng/ml). Thus acid-induced net Ca efflux appears to involve mechanisms in addition to the RANK/RANKL pathway. Osteoblasts also produce TNF-α, another cytokine that stimulates the maturation and activity of osteoclasts. Incubation of calvariae in acid medium caused a significant increase in TNF-α levels. Incubation of calvariae with anti-TNF (up to 250 ng/ml) did not significantly decrease acid-induced net Ca efflux. However, the combination of RANK/Fc plus anti-TNF caused a significant but subtotal reduction in acid-induced Ca efflux, whereas the combination of RANK/Fc plus an isotype-matched control for the anti-TNF had no effect on Ca release. Thus simultaneous inhibition of RANKL and TNF-α is necessary to reduce acid-induced, cell-mediated net Ca efflux from bone; however, additional osteoblast-produced factors must also be involved in acid-induced, cell-mediated bone resorption.
- chronic metabolic acidosis
- bone resorption
- osteoblastic collagen synthesis
- tumor necrosis factor-α
metabolic acidosis increases urine calcium excretion (20, 44, 45) without a concomitant increase in intestinal Ca absorption (1, 37, 46) resulting in a net loss of Ca (46). The skeleton, as the primary reservoir of body Ca (61), has been implicated as the source of this additional urinary Ca (5, 12, 15, 30, 45). In vivo mild, physiologically relevant, metabolic acidosis depletes bone Ca stores (12, 45). In vitro acute (up to 24 h) metabolic acidosis, modeled as a primary reduction in medium bicarbonate concentration, stimulates Ca efflux primarily through physicochemical mineral dissolution, while cell-mediated mechanisms predominate at later time periods (>24 h) (40). Using neonatal mouse calvaria, we showed that acidosis-induced, cell-mediated Ca efflux from bone is mediated by inhibition of osteoblastic collagen synthesis and stimulation of osteoclastic resorption as measured by release of β-glucuronidase, an enzyme whose secretion correlates with osteoclast-mediated bone resorption (9, 11, 13, 14, 17, 18, 42). This acid-induced, cell-mediated increase in net Ca flux is mediated, at least in large part, by an increase in osteoblastic PGE2 release (18, 41, 55).
Acute metabolic acidosis inhibits RNA accumulation for the immediate early response gene egr-1 (26) while chronic metabolic acidosis decreases RNA abundance for the bone matrix genes, matrix gla protein and osteopontin (23, 24). Recently, we showed that metabolic acidosis leads to an increase in expression of RNA for the osteoclastogenic factor receptor activator of NF-κB ligand (RANKL) (25). The interaction between RANKL, produced by osteoblasts, and its receptor RANK, on the surface of osteoclast precursors, activates the DNA binding protein NF-κB to initiate a differentiation cascade, which leads to functional multinuclear osteoclasts capable of resorbing bone (3, 19, 33, 34). In addition to promoting the maturation of preosteoclasts, RANKL also increases the bone-resorbing activity of mature osteoclasts (7, 27, 43) and decreases osteoclastic apoptosis (57). The osteoclastogenic activity of RANKL can be potentiated by other cytokines such as TNF-α, also produced by osteoblasts (4, 36, 60).
In the current study we tested the hypothesis that inhibition of RANKL and/or TNF-α activity would markedly suppress cell-mediated, acid-induced bone resorption. We found that inhibition of either RANKL activity or TNF-α activity alone was not sufficient to inhibit the acid-mediated bone resorption; however, simultaneous inhibition of both pathways led to a significant decrease in acid-induced Ca efflux.
MATERIALS AND METHODS
Recombinant human osteoprotegerin (OPG) was obtained from PeproTech (Rocky Hill, NJ). RANK/Fc chimera was obtained from Sigma (St. Louis, MO). Rat anti-mouse TNF-α monoclonal antibody (6281–2626) and purified rat IgG1,κ isotype control immunoglobulin were obtained from BD PharMingen (San Diego, CA).
Organ culture of bone.
Exactly 2.8 ml of DMEM containing 15% heat-inactivated horse serum, heparin (10 USP U/ml), and penicillin (100 U/ml) were preincubated at a partial pressure of carbon dioxide (Pco2) of 40 mmHg, at 37°C for at least 3 h in 35-mm dishes (40). Calvariae were dissected from 4- to 6-day-old mice and immediately before adding two bones per dish, all inhibitors were added and then 1 ml of medium was removed to determine preincubation pH, Pco2, and calcium concentration. In the 48-h experiments, at the conclusion of the first 24-h incubation period, medium was removed, analyzed for pH, Pco2, Ca concentration and in some experiments the level of TNF-α, and replaced with similar, fresh preincubated medium. At the conclusion of the second 24-h incubation period, medium was removed and analyzed for pH, Pco2, Ca concentration and in some experiments levels of sRANKL (soluble RANKL) and OPG. In those experiments that extended to 51 h, the medium was analyzed and replaced at both 24 and 48 h with similar fresh preincubated medium and analyzed at the conclusion of the experiment. Ca was measured by an ion-selective electrode (model 10, Nova Biomedical, Waltham, MA). Net Ca flux was calculated as Vm([Ca]f − [Ca]i), where Vm is the medium volume (1.8 ml) and [Ca]f and [Ca]i are the final and initial medium Ca concentrations, respectively. A positive flux indicates efflux of Ca from bone and a negative flux indicates Ca influx into bone. Medium pH and Pco2 were determined with a blood-gas analyzer (ABL5, Radiometer, Copenhagen, Denmark) and the concentration of medium bicarbonate ([HCO3−]) was calculated from pH and Pco2 as described previously (8, 16).
In each experiment, some calvariae were incubated in medium at neutral pH (Ctl, pH ∼7.4) and others at a physiologically acidic pH (Met, pH ∼7.1) produced by a primary reduction in [HCO3−], as a model of metabolic acidosis. Pco2 was maintained at ∼40 mmHg throughout all experiments. To closely replicate physiological conditions, only the HCO3−/CO2 buffer system was used; no other buffers were added to the medium (10). Initial medium pH was reduced to ∼7.4 in Ctl and ∼7.1 in Met by the addition of HCl (maximum added volume = 11 μl of 2.4 M HCl per ml of medium). The initial medium pH, Pco2, and [HCO3−] are reported as means for each of the Ctl and each of the Met groups in the figure legends. There were no differences between the initial medium pH, Pco2, and [HCO3−] in any of the Ctl groups and there were no differences between the initial medium pH, Pco2, and [HCO3−] in any of the Met groups.
sRANKL, OPG, and TNF-α ELISA.
After incubation, medium sRANKL, OPG, and TNF-α concentrations were determined with ELISA kits (R&D Systems, Minneapolis, MN) following manufacturer's directions. Medium was analyzed for sRANKL and TNF-α without dilution and for OPG after 1:8 dilution in the provided diluent. For TNF-α assays, samples containing less TNF-α than the specified lower limit of assay sensitivity (5.1 pg/ml) were set to 0.
All values were expressed as means ± SE. Tests of significance were calculated using the ANOVA and regression analysis using conventional programs (BMDP; University of California, Los Angeles, CA) on a personal computer. P < 0.05 was considered significant.
Acidosis effects on levels of sRANKL and OPG.
Previously, we showed that incubation of neonatal mouse calvariae in physiologically acidic medium increases RANKL RNA accumulation (25). To determine whether there was a concomitant increase in sRANKL protein, medium was assayed at the end of the 24- to 48-h period by ELISA. Incubation in acidic medium induced an approximately fivefold increase in sRANKL levels (Fig. 1, left y-axis). Osteoprotegerin (OPG) is a decoy receptor for RANKL; an increase in its activity would limit RANKL stimulation of osteoclastic activity. Previously, we showed that OPG RNA accumulation was not altered by incubation in acid medium (25); consistent with this result, in this study we do not find an increase in the level of OPG protein (Fig. 1, right y-axis).
Inhibition of RANKL activity.
We used OPG (10 ng/ml) (29) to block the activity of RANKL and determine the effect on acid-induced net Ca efflux from bone. Met induced a significant increase in Ca efflux; the addition of a superphysiological amount of OPG (29) did not alter the net Ca efflux in Ctl or the increase in net Ca efflux induced by Met (Fig. 2). To independently determine the role of RANKL in acid-induced net calcium efflux, we inactivated RANKL with RANK/Fc. RANK/Fc is a chimeric molecule consisting of the ligand-binding portion of RANK fused to the Fc portion of an immunoglobulin and has been previously used to inactivate RANKL (35). Met again induced a significant increase in net Ca efflux. The addition of increasing concentrations of RANK/Fc to 50 ng/ml, which is far in excess of endogenously produced sRANKL (35), did not alter this acid-induced Ca efflux (Fig. 3).
TNF-α, produced by osteoblasts, potentiates the osteoclast-stimulating activity of RANKL (4, 36, 60). To determine whether there was an increase in TNF-α levels, we measured its concentration by ELISA after a 24-h incubation of calvariae in neutral or acidified medium. Incubation in Met led to a significant increase in the TNF-α concentration (Fig. 4). We then inhibited the TNF-α activity with a blocking monoclonal antibody (anti-TNF) (64). Again, Met induced a significant increase in net Ca efflux (Fig. 5). The addition of increasing concentrations of anti-TNF to 250 ng/ml, which is far in excess of endogenously produced TNF-α (64), did not alter this acid-induced Ca efflux (Fig. 5).
Simultaneous inhibition of RANKL and TNF-α.
Because inhibition of either RANKL or TNF-α, individually, did not significantly alter acid-induced Ca efflux from bone, we next determined the effects of simultaneous inhibition of these two pathways. During the 24- to 48-h time interval, when cell-mediated, acid-induced bone resorption predominates, Met again increased net Ca efflux (Fig. 6). The addition of both RANK/Fc (50 ng/ml) and anti-TNF (250 ng/ml) significantly inhibited acid-induced Ca efflux while there was no change in Ca efflux in Ctl (Fig. 6, top). At the 48- to 51-h time period, the combination of RANK/Fc and anti-TNF again significantly inhibited acid-induced Ca efflux, with no change in nonstimulated Ca efflux (Fig. 6, bottom).
To control for nonspecific antibody effects on acid-induced net Ca efflux, an isotype-matched IgG was used in place of the specific anti-TNF. While anti-TNF + RANK/Fc again significantly inhibited acid-induced net Ca efflux from bone, the isotype-matched IgG + RANK/Fc did not alter Ca efflux at either the 24- to 48-h or the 48- to 51-h time interval (Fig. 7).
In the current study, we tested the hypothesis that inhibition of RANKL and/or TNF-α activity would suppress acid-induced, cell-mediated bone resorption. We found that inhibition of either RANKL activity or TNF-α activity alone was not sufficient to suppress acid-mediated bone resorption; however, simultaneous inhibition of both pathways led to a significant decrease in acid-induced calcium efflux.
We previously showed that when neonatal mouse calvariae are cultured in medium acidified by a primary decrease in the concentration of bicarbonate, a model of metabolic acidosis, there is a marked increase in net Ca efflux compared with bone cultured in neutral pH medium. Although the Ca efflux is initially due to acid-induced physicochemical mineral dissolution, at later times (>24 h) there is subsequent suppression of osteoblastic activity and stimulation of osteoclastic activity, leading to enhanced net bone resorption (11, 13, 17, 18, 42). To ensure that we measure primarily cell-mediated effects, we study net calcium efflux in the 24- to 48- and/or 48- to 51-h ranges. The mechanism of the acid-induced, cell-mediated resorption involves increased osteoblastic secretion of PGE2, a known stimulator of osteoclastic bone resorption (18, 41, 55).
Incubation of calvariae in acidic medium also induces a cyclooxygenase-dependent increase in expression of RNA for RANKL, without a change in OPG RNA expression (25). RANKL, a TNF-related protein, is expressed on the osteoblastic cell surface and interacts with a receptor, RANK, on the surface of preosteoclasts and osteoclasts (3, 19, 33, 34). RANKL can also be processed into a soluble form, sRANKL, by the metalloproteinase TNF-α converting enzyme, TACE (52, 54). The interaction of RANKL with RANK stimulates differentiation of preosteoclasts into functionally mature, bone-resorbing osteoclasts (43); increases the resorptive activity of mature osteoclasts (7, 27, 43); and decreases apoptosis of osteoclasts (57). Osteoblasts also produce OPG, a decoy receptor for RANKL, which prevents the RANK/RANKL interaction (3, 19, 33, 34). An increase in the ratio of RANKL to OPG leads to an increase in osteoclastic bone resorption. Metabolic acidosis caused an increase in RANKL protein without affecting OPG levels (Fig. 1). PGE2 has been reported to inhibit expression of OPG in human bone marrow stromal cells (6), human periodontal ligament cells (58), mouse osteoblasts (59), and mouse osteoblastic MC3T3-E1 cells (49). We did not observe a decrease in OPG RNA (25) or protein (this study), suggesting that acidosis may alter the responsiveness of osteoblasts to PGE2. In this study, we found that inhibition of the RANK/RANKL interaction with OPG did not inhibit acid-induced Ca efflux from bone. The lack of inhibition could reflect instability of OPG or a failure of OPG to inactivate RANKL before its association with RANK. We confirmed the lack of inhibition using a fusion protein between the ligand-binding domain of RANK and the Fc portion of an immunoglobulin (RANK/Fc) (35). RANK/Fc alone also did not inhibit acid-induced Ca efflux from bone. We believe that we added sufficient OPG and RANK/Fc to totally block the activity of RANKL. We added ∼140-fold excess of OPG (10 ng/ml; Fig. 2) and ∼300-fold RANK/Fc (50 ng/ml; Fig. 3) compared with the amount of sRANKL (∼120 pg/ml; Fig. 1) present after 48 h of incubation in acid medium. The absolute increase in RANKL (soluble and osteoblast bound) was not determined in this study.
In addition to RANKL, osteoblasts express other cytokines, such as TNF-α, IL-1α and -β, and IL-6, that enhance bone resorption (36, 50, 51). While TNF-α is generally considered to act as an adjunct to RANKL in stimulating osteoclastogenesis (36, 50, 51, 63), TNF-α has been shown to support osteoclast formation during inhibition of the RANKL pathway (4, 38, 56, 64). However, the lack of osteoclasts, leading to osteopetrosis, in RANKL −/− (39) and RANK −/− (47) mice suggests that the RANKL-independent TNF-α pathway does not support normal bone remodeling. Mice overexpressing human TNF-α have increased numbers of osteoclast precursors in bone marrow and spleen, suggesting a role for TNF-α in early osteoclast differentiation (48). TNF-α also suppresses osteoblast differentiation, inhibiting production of the differentiation factor RUNX2/cbfa1 (28).
Acidified medium increased levels of TNF-α. The TNF-α levels in the medium of the cultured calvariae ranged from 11 pg/ml (Ctl) to 19 (Met), similar levels to those reported for normal human bone marrow mononuclear cells (∼4 pg/ml) (22). To determine whether inhibition of TNF-α decreased acid-induced resorption, we used a blocking monoclonal antibody. The addition of ∼400-fold excess of anti-TNF (250 ng/ml) did not inhibit acid-mediated Ca efflux (Fig. 5). While inhibition of either RANKL alone (with OPG or RANK/Fc) or TNF-α alone (with anti-TNF) had no effect on acid-induced bone resorption, simultaneous blockade of both RANKL and TNF-α significantly inhibited acid-induced Ca release (Fig. 6). The combination of RANK/Fc and anti-TNF did not alter resorption under neutral conditions [Fig. 6, compare first bar (Ctl) and third bar (Ctl + RANK/Fc + anti-TNF)], indicating a lack of cellular toxicity. A control, isotype-matched monoclonal antibody + RANK/Fcs did not suppress the acid-induced Ca efflux (Fig. 7), further substantiating that blockade of both pathways is necessary to block the acid-induced bone resorption.
This study indicates that acidified medium increases the amount of each of the resorptive factors RANKL and TNF-α and that each alone is sufficient to induce acid-mediated bone resorption. This is consistent with previous findings that while optimal osteoclast activation is obtained with the combination of RANKL and TNF-α, either cytokine alone can support a degree of osteoclast activation (36, 50, 51, 63). Blockade of either RANKL or TNF-α alone did not decrease acid-induced bone resorption, while blockade of both led to a significant decrease in this bone resorption. However, there was still appreciable acid-induced bone resorption after blockade of both RANKL and TNF-α, suggesting that other proresorptive factors must be involved in the acid-induced bone resorption. Such factors could include IL-1α, IL-1β, and IL-6 (36, 50, 51). Osteoblasts also produce the TNF-related apoptosis ligand (2) which can either inhibit osteoclastogenesis (62) or, in combination with OPG, activate osteoclastogenesis (21). That acid-induced bone resorption involves multiple parallel pathways is consistent with the general requirement for redundant pathways in cytokine-mediated systems (31, 32, 53).
This work was supported in part by National Institutes of Health Grants AR-46289 and DK-56788 and a grant from the Renal Research Institute.
We thank S. Smith for expert technical assistance.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society