To investigate the role of osteoprotegerin (OPG) on alcohol (ethanol)-mediated osteoporosis, we measured a variety of bone remodeling parameters in mice that were either on a control diet, an ethanol (5%) diet, or an ethanol (5%) diet plus OPG administration. OPG diminished the ethanol-induced (1) decrease in bone mineral density (BMD) as determined by dual-energy densitometry, (2) decrease in cancellous bone volume and trabecular width and the increase of osteoclast surface as determined by histomorphometry of the femur, (3) increase in urinary deoxypyridinolines (Dpd's) as determined by ELISA, and (4) increase in colony-forming unit-granulocyte macrophage (CFU-GM) formation and osteoclastogenesis as determined by ex vivo bone marrow cell cultures. Additionally, OPG diminished the ethanol-induced decrease of several osteoblastic parameters including osteoblast formation and osteoblast culture calcium retention. These findings were supported by histomorphometric indices in the distal femur. Taken together, these data show that OPG diminishes ethanol-induced bone loss. Furthermore, they suggest that OPG achieves this through its ability to abrogate ethanol-induced promotion of osteoclastogenesis and promote osteoblast proliferation.
ALCOHOLISM IS a widespread problem accompanied by many pathophysiological manifestations. For example, a large proportion of chronic alcoholics have osteopenia.(1–7) There is evidence that alcohol (ethanol) induces bone loss through both inhibition of osteoblast activity(8–10) and induction of osteoclast activity.(9,11–13)
Osteoclasts, the multinucleated giant cells that remove bone, are derived from colony-forming unit granulocyte-macrophage (CFU-GM), which branch from the monocyte-macrophage lineage during the early differentiation process. Macrophage colony-stimulating factor (M-CSF) is a primary factor that induces hematopoietic stem cells to differentiate into CFU-M. Under the stimulation of M-CSF and RANKL, the cells differentiate into prefusion osteoclasts. With the stimulation of M-CSF, RANKL, and interleukin-1 (IL-1) at this step, the prefusion osteoclasts differentiate into mature osteoclasts with osteolytic activity. The differentiation and activation of osteoclasts is regulated by a variety of factors such as hormones, growth factors, and cytokines (reviewed by Teitelbaum(14)). Most osteoclastogenic factors act through osteoblasts and indirectly stimulate osteoclast maturation and activation. These factors are classified into three categories based on different signal transduction pathways: the vitamin D receptor pathway activated by 1,25-dihydroxyvitamin D3 [1,25(OH)2D3]; the protein kinase A pathway activated by parathyroid hormone (PTH), prostaglandin E2 (PGE2), or IL-1; and the glycoprotein 130 (gp130) pathway activated by the IL-6-type cytokines.(15) Recently, a new cytokine system was established to play a critical role in osteoclastogenesis, the RANKL, and osteoprotegerin (OPG) system.
Osteoclastogenesis occurs through the binding of RANKL, which is present on osteoblasts, to RANK, which is found on the osteoclast precursors. OPG is a protein produced by osteoblasts and bone marrow stromal cells. OPG binds RANKL, effectively blocking it from binding to RANK and thus inhibiting osteoclastogenesis (reviewed by Teitelbaum(14)).
Consistent with the result of increased bone resorption by ethanol in vitro(13) and bone histomorphometry findings of an increased bone resorptive area in alcoholics,(9,16) we have previously documented that alcohol induces osteoclast activity and bone resorption in mice.(17) Furthermore, the ethanol-induced osteoclast activity was associated with increased RANKL mRNA expression in bone marrow cells.(17) Therefore, we hypothesized that inhibition of RANKL should diminish ethanol-induced bone loss. Accordingly, we determined, in this study, if the RANKL inhibitory protein OPG inhibits alcohol-mediated bone loss in a murine model.
MATERIALS AND METHODS
Eight-week-old male C57/Bl6 mice (Charles River Laboratories, Wilmington, MA, USA) were housed individually in plastic cages under standard laboratory animal conditions. Mice were fed liquid ethanol or control diets as we have previously described.(17) Briefly, a National Research Council (NRC) liquid diet(18) that contains 100% of the vitamin and mineral requirements set by the NRC was used. The ethanol-treated group diet contained 26% of energy as ethanol, 31.5% as carbohydrates, 12.5% as protein, and 30% as fat (diet no. 710279; Dyets, Bethlehem, PA, USA) as previously described,(19) resulting in a dietary ethanol concentration of 5% (vol/vol) and blood ethanol concentrations of 9.9 mmol/liter and 7.3 mmol/liter at 5 h and 20 h postfeeding, respectively.(18) An isocaloric level of maltose was substituted for ethanol in the nonethanol control diet. Diets were prepared fresh daily. To minimize differences due to food intake, mice from both the control and the ethanol + OPG groups were matched with a mouse from the ethanol group. To accomplish this, food intake for each ethanol-fed mouse was measured and then its respective match-fed control diet-fed mouse and ethanol-fed + OPG mouse received the same volume of food the following day. Using this technique, there was no difference in body weight between groups at the end of the study (data not shown).
As the time animals were randomized to be fed with control or ethanol diet, ethanol-fed mice were randomized to receive either injections (through intraperitoneal injection) of vehicle (1% bovine serum albumin [BSA] in 1× PBS; n = 10) or recombinant mouse OPG/Fc chimera (R & D Systems, Inc., Minneapolis, MN, USA) at 1 mg/kg of body weight (n = 10) twice a week and continued for 4 months. At the end of 4 months, animals were killed by CO2 asphyxiation. The University of Michigan Animal Care and Use Committee approved the animal protocols.
Bone mineral density (BMD) was measured using DXA on an Eclipse Peripheral Dexa Scanner (Norland Medical Systems, Ft. Atkinson, WI, USA) using research software. To measure whole body BMD, mice were anesthetized with ketamine pre- and postethanol feeding and placed in sternal recumbency on the scanner. The mice were scanned at 10 mm/s with a resolution of 0.5 mm × 0.5 mm. BMD was determined in a window that excluded the calvarium and tail. To measure femoral BMD, the right femur was excised from soft tissue and placed on the scanner in lateral position. The femur was scanned at 5 mm/s with a resolution of 0.1 mm × 0.1 mm. BMD was determined in a window that encompassed the entire femur. Short-term BMD precision (%CV) was ∼3% for both of these techniques.
Quantification of deoxypyridinolines
At the time of killing, urine was collected and frozen at −80°C until assayed. Urine creatinine was measured using a creatinine kit (Quidel, Mountain View, CA, USA) as recommended by the manufacturer. Deoxypyridinolines (Dpd's) were measured using the Pyrilinks-D kit (Quidel) as recommended by the manufacturer. A standard curve was created using a four-parameter fit generated by MetraFIT software (Quidel). Dpd was corrected for variations in urine concentration by dividing the Dpd value (nmol/liter) by the creatinine value (mmol/liter) for each sample.
Quantification of CFU-fibroblast and CFU-osteoblast and culture calcium retention
CFU-fibroblast (CFU-F) and CFU-osteoblast (CFU-OB) were measured as previously reported,(20) with minor modification. Briefly, mouse femora were flushed with 3 ml of phenol red-free α-minimum essential medium (α-MEM) to obtain marrow cells. The cells were rinsed twice and the total cell numbers per femur was determined and resuspended to obtain a single cell suspension of 1 × 107 cells/ml. For determination of CFU-F, 1.5 × 106 (150 μl) cells were seeded into 4 ml of complete media (α-MEM, 15% FBS, 50 μg/ml of ascorbic acid, and 10 mM ofβ-glycerophosphate) + 10 nM of dexamethasone/well of 6-well plates. Assays were performed in duplicate for each animal. One-half of the media was removed and replaced on day 5 with α-MEM, containing 30% FBS, 100 μg/ml of ascorbic acid, and 20 mM of β-glycerophosphate to result in a final concentration of the original complete media. Dexamethasone was not added after the initial culture period. On day 10 of culture, cells were fixed, stained for ALP, and counterstained with hematoxylin. ALP+ colonies containing ≥20 cells were enumerated. To determine CFU-OB, cultures were established in duplicate and maintained as for CFU-F except 2.5 × 106 cells (250 μl of 1 × 107 cells/ml) were plated per well. Media was replaced every 5 days as described previously. Cells were cultured for 25 day at which time they were fixed in 18% formaldehyde, 50% ethanol, and 32% deionized water for 1 minute and then stained using Von Kossa's method. Stained colonies were enumerated. To measure calcium content in cultures, cells were prepared and grown as for CFU-OB. Calcium content was measured using standard biochemistry (Calcium Kit; Fisher Scientific Co., Pittsburgh, PA, USA).
Quantification of osteoclast formation in bone marrow cultures
Osteoclast formation was determined as previously reported,(20) with minor modifications. Briefly, marrow cells were collected as described previously. To determine osteoclast number, the cells were plated at 1.5 × 106 cells/2 cm2 well on a 13-mm2 round tissue culture coverslips (4-6 replicate cultures/mouse). The cultures were maintained for 9 days in the presence of 10−8 M of 1,25 (OH)2D3 with replacement of half the medium every 3 days. To confirm the identification of osteoclasts, several replicates were evaluated for TRAP (Acid Phosphatase Kit; Sigma Chemical Co., St. Louis, MO, USA) and calcitonin receptors (CTR) by a previously described method.(21) Briefly, at the end of the culture period, cells were incubated with125I-salmon calcitonin (0.02 μCi) in the absence or presence of excess unlabeled salmon calcitonin (10−7 M) at room temperature for 2 h. Cells then were washed twice with PBS and fixed with 2.5% glutaraldehyde in PBS. The coverslips were dipped in LM-1 photographic emulsion (Eastman Kodak, Rochester, NY, USA) and maintained at 4°C for 14 days. Cells were counterstained with methyl green and then scored as mononuclear, 2-5 nuclei, or more than nuclei. Results are reported as number of multinucleated TRAP+ cells and CTR+ cells per/coverslip.
Assessment of individual CFU-F colonies for ability to support osteoclast formation
CFU-F osteoclastogenic activity was determined as previously reported(20) with minor modifications. Briefly, 100 μl of 2 × 105 bone marrow cells/ml in α-MEM was placed into wells of 96-well plates containing 100 μl of α-MEM, 30% FBS, 100 μg/ml OF ascorbic acid, 20 mM OF β-glycerophosphate in each well of 96-well plates. Cells were incubated for 5 days, at which time spleen cells were isolated from C57/BL6 mice and resuspended to 1 × 106/ml in α-MEM with 20% FBS and 20 nM of 1,25(OH)2D3. Then, 100 μl of media was removed from each well and 100 μl of the spleen cell suspension was added. On day 9, one-half of the media was removed and replaced with α-MEM with 20% FBS and 20 nM of 1,25(OH)2D3. On day 13, the cells were stained for TRAP and each well was examined microscopically for the presence of CFU-F (>20 cells) and osteoclastic cells (TRAP+).
Quantification of CFU-GM
To determine CFU-GM formation, 1.25 × 105 of the bone marrow cells (i.e., 250 μl of 5 × 105 cells/ml) were seeded into 2.5 ml of a methylcellulose-based media containing CFU-GM promoting growth factors (MethoCult HCC-3534; Stem Cell Technologies, Vancouver, BC, Canada) in a 15-ml sterile test tube. The cells were vortexed and bubbles were allowed to settle. Then, 1.1 ml of cell/media mixture was placed into a 35-mm cell culture dish using a 16G 1.5-in needle. Assays were performed in duplicate. The duplicate 35-mm dishes with cells were placed into a 150-mm plate along with a covered 35-mm dish filled with 3 ml of sterile water. The cells were incubated at 37°C 5% CO2 for 7 days, at which time CFU-GMs (defined as >50 cells) were counted.
The right femur was dissected from soft tissue, fixed in 10% phosphate-buffered formalin for 24 h, decalcified in 14% EDTA, and embedded in a lateral position in paraffin. 5 μM thick longitudinal sections from the middle of the femur were stained with hematoxylin and eosin or trichrome stain. Histomorphometry was performed using Bioquant Nova Imaging Software and digitizer tablet (Bioquant, Nashville, TN, USA). Histomorphometric parameters of the secondary spongiosa in the distal femoral metaphysis were measured in standard rectangular area (1.5 mm2 at least 0.2 mm proximal to the distal femoral growth plate, to exclude primary spongiosa.(22) Parameters were measured at a magnification of 100×. The terminology used was that recommended by the Histomorphometry Nomenclature Committee of the American Society for Bone and Mineral Research.(23)
Repeated measures analysis of covariance (ANCOVA) was used to test for significant differences in mean BMD among groups, using weight as a covariate. Factorial ANOVA was used to test for significant differences in parameters other than BMD. Fisher's least significant difference was used for post hoc analysis. Statistical significance was determined as p ≤ 0.05.
Previously, we have documented that ethanol liquid diet intake induces blood ethanol levels of 10 mmol/liter in ethanol-fed mice, (17) which represents the blood levels ob-tained in chronic alcoholics. (24,25) Furthermore, we have reported that exposure to ethanol at these levels results in loss of BMD and increased osteoclastogenesis. (17) However, a direct connection between osteoclastogenesis and ethanol-induced bone loss has not been established previously. To determine if OPG, an inhibitor of osteoclastogenesis, could prevent the ethanol-induced decline in BMD, mice were either fed either a control diet, a 5% ethanol diet, or a 5% ethanol diet plus administration of OPG for 4 months, followed by measurement of excised right distal femoral BMD using DXA. Distal femur, as opposed to whole femur, was evaluated because it is composed primarily of trabec-ular bone, and changes of BMD are more readily detectable at this site than at the whole femur (J. Zhang, unpublished observation, 2000). Ethanol induced an 17.9% loss of distal femoral bone mineral content (BMC); whereas OPG administration cut this approximately in half to 8.9% (Fig. 1). To determine if the OPG-mediated prevention of ethanol-induced BMD loss was associated with diminished bone resorption, urinary Dpd level, an indicator of bone resorption, was quantified. In agreement with the BMD data, ethanol induced a mild increase of Dpd levels, which was diminished by OPG administration (Fig. 2). These data establish that OPG administration partially protects mice from ethanol-induced bone loss. Furthermore, the Dpd data suggest that OPG mediates this effect, in part, through decreasing bone resorption.
Ethanol-induced decline of BMD is associated with both stimulation of osteoclastogenesis and increased RANKL expression in the marrow.(17) Thus, it follows that ethanol may induce osteoclastogenesis through its ability to induce RANKL. To evaluate this hypothesis, we determined the effect of OPG administration on ethanol-induced formation of both bone marrow-derived CFU-GM, which are the osteoclast precursors, and mature osteoclasts. OPG had no effect on CFU-GM formation, which was increased ∼28.9% and 18.7% in ethanol-fed mice receiving either vehicle or OPG, respectively, compared with control-fed animals (Fig. 3A). In contrast, ethanol induced an ∼32.7% increase of osteoclast-like cells that was completely abrogated by OPG (Fig. 3B). In fact, OPG reduced the number of osteoclast-like cells by ∼56.8% below those of the control animals. To ensure that changes of CFU-GM and osteoclast-like cells were not caused by general changes in bone marrow cell number, we determined the number of cells obtained per femur. There was an average of 5.3 ± 1.51 × 107 cells/femur and no differences were detected among the groups (data not shown). These data, taken together with our previous demonstration that ethanol induces RANKL expression in the bone marrow,(17) suggest that ethanol induces CFU-GM formation independent of RANKL, but the formation of ethanol-induced osteoclasts is dependent on RANKL expression.
RANKL is expressed on osteoblasts and bone marrow stromal fibroblasts.(26,27) Therefore, it follows that ethanol may up-regulate CFU-F's ability to induce osteoclastogenesis through increased RANKL expression. To test this hypothesis, we determined ethanol-mediated up-regulation of CFU-F-induced osteoclastogenesis and OPG's ability to inhibit this activity. Ethanol induced an ∼450% increase in osteoclasts associated with CFU-F (Fig. 4). OPG diminished this by 350%. These data show that OPG can inhibit CFU-F-mediated osteoclastogenesis.
In addition to its ability to inhibit RANKL-induced osteoclastogenesis, OPG can bind TNF-related apoptosis-inducing ligand (TRAIL), a TNF-like molecule that can induce apoptosis. Thus, it is plausible that OPG diminishes loss of BMD through inhibition of osteoblast apoptosis as has been established for PTH.(28) To test this possibility, we measured the effect of ethanol ingestion and OPG administration on CFU-F formation, osteoblast colony formation, and osteoblast function. Both ethanol and OPG alone or combined had no effect on CFU-F formation (Fig. 5A). However, ethanol reduced osteoblast colony formation by 31.8%; OPG completely blocked the ethanol-induced decline of osteoblast function (Fig. 5B). In parallel to the decline in osteoblast colony formation, ethanol ingestion was associated with decreased calcium retention in culture and OPG blocked this decline (Fig. 5C). These data show that OPG administration has a protective effect against an ethanol-induced decline of osteoblast production and function.
To determine if these in vitro observations were reflected in vivo, bone histomorphometry was performed on the femur (contralateral to the femur used for obtaining bone marrow cells). Ethanol induced loss of cancellous bone area, which OPG prevented (Table 1). Additionally, ethanol increased the number of osteoclasts per millimeter of bone and the overall osteoclast perimeter and decreased the number of osteoblasts per millimeter of bone and the overall osteoblast perimeter (Table 1). OPG administration prevented the ethanol-induced changes (Table 1). Finally, ethanol ingestion induced a decrease of trabecular width, which OPG abrogated (Table 1). These data are consistent with the in vitro data and show that OPG diminishes ethanol-induced loss of bone at the cellular level in vivo.
Table Table 1.. Static Histomorphometry of Distal Femoral Cancellous Bone in Mice Fed a Control Diet, an Ethanol Diet, or an Ethanol Diet With OPG Administration for 4 Months
This study provides strong evidence that OPG prevents an ethanol-mediated decline of BMD through inhibition of osteoclast formation. These data are consistent with previous reports that ethanol induces osteoclastogenesis and inhibits osteoblast colony formation(17) and OPG inhibits osteoclastogenesis.(29) In addition to the novel finding that OPG prevented ethanol-induced osteoclastogenesis, results from this study suggest that OPG diminishes ethanol-mediated inhibition of osteoblast formation.
OPG, a secreted soluble decoy receptor for RANKL, mediates its bone-sparing effects by binding to RANKL, which inhibits RANKL-induced osteoclastogenesis.(29) Administration of recombinant OPG has been shown to increase bone density in mice an rats,(29,30) diminish ovariectomy-mediated loss of bone volume in the proximal tibial metaphysis of mice,(29) prevent tail suspension-induced osteopenia in mice,(31) and rescue OPG-deficient mice from osteoporosis.(32) Adenoviral delivery of OPG was shown to ameliorate bone resorption in a mouse ovariectomy model of osteoporosis.(33) A single subcutaneous dose of OPG was shown to reduce bone turnover as determined by several biochemical markers such as urinary N-telopeptide and Dpd in a randomized, double-blind, placebo-controlled clinical study that was conducted in postmenopausal women.(34) Furthermore, OPG administration was shown to decrease osteoclast number and osteoclast perimeter in the proximal tibial metaphysis of hypercalcemic C-26 adenocarcinoma-bearing mice.(35) OPG administration also was shown to inhibit osteoclast activity in cancer cell-induced osteoclastogenesis.(36–40) In agreement with these findings, this report shows that OPG prevented ethanol-mediated osteoporosis and that this was associated with decreased osteoclastogenesis. The absence of a treatment group that received OPG alone renders it difficult to interpret the magnitude of the impact that ethanol had on OPG's ability to modulate bone remodeling. However, the data clearly show that OPG diminished ethanol-mediated bone loss.
Several lines of evidence exist that show that in addition to decreased bone production, ethanol-mediated bone loss is associated with increased bone resorption in chronic alcoholics. For example, chronic alcoholics show increased urinary excretion of hydroxyproline(41) and histomorphometric evidence of increased resorptive surface and osteoclast number measurements.(9,16) Thus, the ability to inhibit ethanol-mediated bone resorption in alcoholics, through compounds such as OPG and bisphosphonates, should inhibit progression of osteopenia in alcoholics.
The observation that OPG diminished ethanol-induced osteoclastogenesis but not CFU-GM formation is consistent with the current knowledge of osteoclast formation. Specifically, pluripotent stem cells become committed to CFU-GM through cytokines such as IL-1 and IL-6. RANKL then acts on the CFU-GM to induce osteoclast formation. Thus, OPG, which inhibits RANKL, should have no impact on CFU-GM formation as opposed to inhibiting osteoclast formation. These results are consistent with the previous report that OPG administration to mice decreased osteoclastogenesis in bone marrow cultures derived from mice administered various cytokines and hormones [IL-1β, TNF-α, PTH, PTH-related peptide (PTHrP), and 1,25(OH)2D3] that are known to increase bone resorption.(42)
We cannot readily account for the observation that OPG administration was associated with decreased ethanol-induced CFU-F-mediated osteoclastogenesis. Specifically, in this assay, the CFU-F are from the mice being treated, while the osteoclast precursors are derived from suspension of spleen from untreated mice. Thus, although it is probable that ethanol ingestion induced RANKL expression in the CFU-F in vivo, which in turn could induce osteoclastogenesis, the mechanism through which OPG administration diminished their ability to induce osteoclastogenesis once the CFU-F are grown in vitro is not clear. OPG should, in theory, not impact the development of the CFU-F, because OPG mediates its effect on RANKL protein. One possibility is that OPG forms a negative feedback mechanism on the RANKL, thus resulting in down-regulation of RANKL production from the CFU-F, thus the CFU-F from the OPG-treated group would express less RANKL, resulting in less production of osteoclasts from the splenic suspensions. Further experiments are required to clarify this activity.
The ability of OPG to prevent an ethanol-mediated decrease of osteoblast colony formation was an unexpected finding. This finding was supported further by the histomorphometric observation that OPG administration abrogated ethanol's ability to decrease the osteoblast perimeter. Taken together, these observations suggest that OPG promotes bone production through a mechanism other than a direct effect on osteoclastogenesis. This postulation is in line with a previous report that describes that OPG increased the mineral phase of bone production in a murine tail suspension model.(31) The mechanism through which ethanol inhibits osteoblasts currently is not well defined but may involve induction of apoptosis. OPG's ability to bind TRAIL(43) may allow it to inhibit TRAIL-mediated apoptosis, although the affinity for TRAIL at 37°C is low.(44) Thus, OPG may mediate its effect through inhibition of TRAIL-mediated apoptosis or a similar mechanism to inhibit ethanol-mediated apoptosis of osteoblasts.
The observation that ethanol reduced CFU-OB (von Kossa positive) but not CFU-F (ALP+) cells suggests that there is a differential effect of ethanol based on the maturity of the cell. Specifically, CFU-Fs are less differentiated cells than CFU-OBs and may have been less sensitive to the effects of ethanol. Differential effects based on the state of differentiation have been observed previously in osteoblastic cells. For example, condition media from prostate cancer cells inhibited bone nodule formation and ALP activity more rapidly in a well-differentiated osteosarcoma line than in primary fetal rat calvaria cells, which contain many undifferentiated osteoprogenitor cells.(45)
In conclusion, OPG inhibited ethanol-induced bone loss in mice. The OPG effect on bone was associated with decreasing ethanol-induced osteoclastogenesis and preventing an ethanol-induced decrease of osteoblast formation. These results suggest that in addition to decreased bone production, increased bone resorption contributes to ethanol-induced bone loss. Furthermore, these results suggest that OPG may effectively inhibit bone loss in chronic alcoholics.
This work was supported by the National Institutes of Health (NIH) grants R01 AG15904 (to E.T.K.) and T32 RR07008 (to E.T.K.) and the Program in Comparative Integrative Genetics and Biology, University of Michigan.