Leptin Inhibits Osteoclast Generation†
The authors have no conflict of interest
Abstract
Originally, leptin was described as a product of adipocytes that acts on the hypothalamus to regulate appetite. However, subsequently, it has been shown that leptin receptors are distributed widely and that leptin has diverse functions, including promotion of hemopoietic and osteoblastic differentiation. It has been recognized for some time that both serum leptin and bone mass are correlated positively to body fat mass and, recently, we have shown a direct positive relationship between serum leptin and bone mass in nonobese women. We now report that leptin inhibits osteoclast generation in cultures of human peripheral blood mononuclear cells (PBMCs) and murine spleen cells incubated on bone in the presence of human macrophage colony-stimulating factor (hM-CSF) and human soluble receptor activator of NF-κB ligand (sRANKL). The half-maximal concentration inhibitory of leptin was approximately 20 nM in the presence of sRANKL at 40 ng/ml but decreased to approximately 2 nM when sRANKL was used at 5 ng/ml. The majority of the inhibitory effect occurred in the first week of the 3-week cultures. Inhibition did not occur when the PBMC cultures were washed vigorously to remove nonadherent cells or when purified CD14+ monocytes were used to generate osteoclasts, indicating an indirect or permissive effect via CD14− PBMC. Leptin increased osteoprotegerin (OPG) messenger RNA (mRNA) and protein expression in PBMC but not in CD14+ cells, suggesting that the inhibitory effect may be mediated by the RANKL/RANK/OPG system. Leptin may act locally to increase bone mass and may contribute to linkage of bone formation and resorption.
INTRODUCTION
There is accumulating evidence that leptin is more than an energy-regulating hormone because it also is involved in reproduction,1 angiogenesis,2 hemopoiesis and immune function,3-5 sympathetic and endocrine function,6 and brain development.7 Several lines of evidence suggest that leptin also has a role in bone metabolism. Clinical studies have shown that obesity protects against osteoporosis,8, 9 that both bone mass10, 11 and serum leptin12, 13 are related positively to body fat mass. Recently, we have established that in nonobese women participating in the Geelong Osteoporosis Study there is a positive relationship between serum leptin and bone mass that is independent of age, body weight, and body fat mass.14 Studies in mice with mutations resulting in leptin deficiency (ob/ob) or lack of functional leptin receptor (db/db) have produced differing results. Earlier studies found that systemic leptin treatment of leptin-deficient mice stimulates bone formation15 and bone growth16 and that the db/db mouse is osteopenic.17, 18 In contrast, the recent study of Ducy and colleagues19 found that both mutants have high bone mass because of increased bone formation and that intracerebroventricular (i.c.v.) infusion of leptin caused bone loss in leptin-deficient and wild-type mice. In human fetuses that are at 18-35 weeks gestation, a negative correlation between serum leptin and a serum bone resorption marker was shown, suggesting that leptin might inhibit bone resorption and increase bone mass.20 Disordered parathyroid hormone-calcium function with and without osteopenia has been described in leptin-deficient humans.6 Recently, Thomas and colleagues have shown that in vitro treatment of human bone marrow stromal cells with leptin promotes their differentiation to bone-forming osteoblasts while inhibiting their differentiation to adipocytes.21
Multinucleate bone-resorbing osteoclasts form in the bone marrow by the differentiation and fusion of macrophage colony-stimulating factor (M-CSF)-dependent mononuclear hemopoietic precursors in cell-cell contact with mesenchymal cells of the stromal/osteoblast lineage (reviewed by Chambers,22 Suda et al.,23 and Roodman24). The recent identification of receptor activator of NF-κB ligand (RANKL), a new type II membrane-bound protein of the tumor necrosis factor (TNF) ligand superfamily that mediates osteoclast differentiation (reviewed by Suda et al.25), has allowed the in vitro generation of osteoclasts in the absence of stromal cells/osteoblasts,26-28 thus allowing examination of direct effects of regulators of osteoclastogenesis on hemopoietic precursors. RANKL acts on osteoclast precursors via a membrane receptor RANK29, 30 but also can bind to a soluble protein osteoprotegerin (OPG).28, 31, 32 By acting as a “decoy” receptor, OPG is a potent inhibitor of osteoclastogenesis.
As osteoclasts differentiate from hemopoietic precursors of the monocyte/macrophage lineage22-24 and leptin activates cells of this lineage,33 we hypothesized that leptin might regulate monocyte differentiation to osteoclasts. Therefore, we assessed the effects of leptin on osteoclastogenesis in a model using human and murine precursors cultured on bone slices and on the regulation of OPG and RANK in these cells.
MATERIALS AND METHODS
Materials
Recombinant murine leptin (lot numbers 08876, 09776, and 10876; purity >95%; endotoxin level <0.1 ng/μg [1 EU/μg] of leptin) and soluble recombinant human RANKL (sRANKL) were purchased from Peprotech (Rocky Hill, NJ, USA). Recombinant mouse and human leptin were generously provided by Dr A.F. Parlow of The National Hormone and Pituitary Program (NHPP). Recombinant human M-CSF (hM-CSF) was generously provided by the Genetics Institute (Cambridge, MA, USA).
Cell preparation
Whole blood was obtained from healthy donors under a protocol approved by Barwon Health Research and Ethics Committee. Peripheral blood mononuclear cells (PBMCs) were isolated from buffy coats (Duo-Add blood bags; Baxter, Deerfield, IL, USA) using Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden). Before further purification, granulocyte contamination was reduced to <1% by a second Ficoll-Paque separation step. Cell purification was achieved using CD antibody-coated magnetic cell sorting (MACS) MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) and a VS+ selection column was mounted on a magnetic separator using standard techniques.34 Purity was assessed using flow cytometry (FACSCalibre; Becton Dickinson, Bedford, MA, USA).
Six-week-old male mice were killed by cervical dislocation. The spleens were removed aseptically, placed in modified minimal essential medium (MEM), and passed through a 100-μm mesh to produce a cell suspension. The cells were pelleted by centrifugation at 400g for 3 minutes and resuspended at 107 cells/ml in MEM containing 10% fetal bovine serum (FBS).
Osteoclast generation
PBMCs (106/well) were settled on to 4 × 4 × 0.5 mm slices of devitalized bovine cortical bone in 96-well culture plates for 2 h when the media and nonadherent cells were aspirated, without washing (unwashed PBMCs). In other experiments, the bone slices were removed from the wells, vigorously rinsed in medium, and then placed in the wells of a new culture plate with fresh medium (washed PBMCs). CD14+ cells (105/well) were settled onto bone slices and were not aspirated or washed at 2 h. Mouse spleen cells (106/well) were settled onto bone slices according to the protocol used for unwashed PBMCs. The cells were maintained in 200 μl of MEM with 10% heat-inactivated FBS in the presence of hM-CSF (25 ng/ml) and sRANKL (40 ng/ml) with or without leptin (3-100 nM). Unless otherwise stated, cell culture experiments described in this report were conducted in media containing 25 ng/ml of hM-CSF and 40 ng/ml of sRANKL (M-CSF/RANKL). Media and additives were replaced twice weekly. At 21 days, the cultures were fixed in 1% formalin and reacted for tartrate-resistant acid phosphatase (TRAP) activity. The formation of osteoclasts was quantified by counting the number of TRAP+ multinuclear cells. In selected experiments, calcitonin receptor (CTR) expression was determined using125I-salmon calcitonin (125I-sCT) autoradiography as previously described.35 Then, bone slices were cleaned to remove the cells and the percentage of the bone surface that was resorbed was quantified with scanning electron microscopy (SEM) at 100-150× magnification.
Gene expression
PBMCs or CD14+ was incubated in plastic tissue culture flasks for various times at 37°C in 5% CO2 in MEM containing M-CSF/RANKL in the presence or absence of leptin. Then, they were lysed in RNAzol B solution and total RNA was extracted according to the manufacturer's instructions. The method used for routine reverse-transcription polymerase chain reaction (RT-PCR) and the sense/antisense primers for OPG and RANK have been described.36 Real-time (TaqMan) PCR was performed using an ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's protocol. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe and primers were from Applied Biosystems. The OPG and RANK probes and primers were designed using Primer Express software (Applied Biosystems) and are detailed in Table 1.
|
|
Western analysis
PBMC (108 cells) in MEM/20% FBS were settled onto plastic flasks for 2 h when the medium was replaced. The cells were incubated for 24 h in MEM/10% FBS containing M-CSF/RANKL with or without leptin (100 nM) and then lysed by the addition of 1 ml of lysis buffer (50 mM of HEPES, 10 mM of Tris-HCl, pH 7.2, 1% Triton X-100, 10% glycerol, 100 mM of NaF, and 5 mM of EDTA). OPG protein was immunoprecipitated using anti-human OPG antibody (R&D Systems, Minneapolis, MN, USA) and samples were probed for OPG using goat anti-sheep immunoglobulin G-horseradish peroxidase (IgG-HRP; Sigma, St. Louis, MO, USA). OPG bands were detected by enhanced chemiluminescence (ECL) using Hyperfilm (Amersham Pharmacia Biotech, Little Chalfont, Buckinghamshire, UK).
OPG protein secretion
PBMCs were incubated in plastic tissue culture flasks at 37°C in 5% CO2 in MEM containing M-CSF, 25 ng/ml, in the presence or absence of leptin, 10−7 M. At 1, 2, 4, 8, 12, and 24 h, aliquots of media were removed for measurement of OPG protein by ELISA. An ELISA for human OPG was developed in a 96-well plate format using the following reagents from R & D Systems: capture, anti-human OPG monoclonal antibody: detection, anti-human OPG biotin affinity-purified polyclonal antibody; protein, recombinant human OPG-Fc chimera; detection, streptavidin-HRP detection kit. The sensitivity was 32 pg/ml and the intra- and interassay CVs were 0.05 and 0.06, respectively.
RESULTS
Osteoclast generation—unwashed PBMCs
Unwashed PBMC incubated on bone slices (106/slice) for 21 days with hM-CSF (25 ng/ml) and sRANKL (40 ng/ml) generated substantial numbers of TRAP+ multinucleate giant cells together with a variable number of TRAP+ and TRAP− mononuclear cells (Fig. 1). Extensive lacunar resorption was evident when the bone surface was examined with SEM (Fig. 1D). Because these multinucleate cells also possessed abundant CTRs35 (Fig. 1B), we consider that they are genuine osteoclasts. No osteoclasts or resorption pits were apparent when the cultures did not include sRANKL (Fig. 1A). When leptin was added to the cultures for the entire 21 days, we found that it was a potent concentration-dependent inhibitor of both osteoclast generation and bone resorption (Figs. 1C and 2A). In eight independent experiments using 100 nM of murine leptin, osteoclast generation was reduced by 83 ± 7% and resorption was reduced by 87 ± 4% (mean ± SEM). In control and leptin-treated cultures, the ratio of resorption/osteoclast number was not different, indicating that leptin inhibited the generation of osteoclasts not their bone-resorbing activity, once formed. Treatment with leptin for the first week only of the 3-week culture produced approximately two-thirds of the inhibitory effect and treatment for the first 2 weeks was as effective as 3 weeks of treatment (Fig. 2B). In other experiments, when leptin treatment was confined to the second week or the third week, little effect on osteoclast generation and bone resorption was observed (results not shown).
Leptin inhibition of osteoclast generation and bone resorption in human PBMC cultures. PBMCs were cultured on bone slices for 21 days with various treatments as indicated. Cells were incubated with125I-sCT (with and without excess unlabeled sCT) and then were fixed, reacted for TRAP, and processed for CTR autoradiography. After quantification of TRAP+ and CTR+ cells (mean ± SEM/slice; n = 3), the bone slices were stripped and processed for SEM: (A) hM-CSF alone, TRAP+/CTR− mononuclear cells only (1345 ± 270); (B) hM-CSF and sRANKL, TRAP+/CTR+ multinucleate giant cells (osteoclasts; 367 ± 50), TRAP+/CTR+ mononuclear cells (16 ± 4), and TRAP+/CTR− mononuclear cells (55 ± 9) present; (C) hM-CSF, sRANKL, and leptin, TRAP+/CTR− mononuclear cells (1380 ± 360) and TRAP+/CTR+ multinucleated cells (9 ± 3/slice; indicated with arrows); (D) hM-CSF and sRANKL, SEM showing multiple resorption lacunae on the bone surface. In some experiments, bone slices treated with leptin showed moderate numbers of osteoclasts. No resorption was seen in the absence of sRANKL although variable limited resorption was seen in the presence of leptin (panels A-C, bar = 50 μm; panel D, bar = 100 μm).
Leptin inhibition of osteoclast generation and bone resorption in human PBMC cultures. (A) Concentration-responsive effect of leptin inhibition on sRANKL-induced osteoclastogenesis. PBMCs (106 cells/well) were cultured for 21 days on bone slices with hM-CSF (25 ng/ml) and sRANKL (40 ng/ml) in the presence of increasing concentrations of leptin. Representative data from one experiment (n = 7 bone slices per treatment). Groups are significantly different if the annotations do not contain the same letter, p < 0.05; analysis of variance (ANOVA); Fishers one-way multiple comparison. (B) Effect of limited treatment with leptin on osteoclastogenesis. PBMCs were cultured with hM-CSF (25 ng/ml) and sRANKL (40 ng/ml) in the absence or presence of 100 nM of leptin for 21 days (21), the first 7 days, (7), or the first 14 days (14). Pooled data from three independent experiments using different blood donors; n = 21 bone slices per treatment; p < 0.0001. Groups are significantly different if the annotations do not contain the same letter; p < 0.0001; analysis of variance (ANOVA); Fisher's one-way multiple comparison.
Because sRANKL at 40 ng/ml is a maximally effective concentration in this osteoclastogenesis model (data not shown), we wondered whether the sensitivity to leptin inhibition might be blunted under these conditions. Therefore, we examined the concentration response of the inhibitory effect of leptin in the presence of a submaximal concentration of sRANKL of 5 ng/ml. Under these conditions, sensitivity to leptin inhibition was increased substantially (Fig. 3). The half-maximal inhibitory concentration for leptin decreased from approximately 20 nM in the presence of 40 ng/ml of sRANKL (Fig. 2A) to approximately 2 nM in the presence of 5 ng/ml of sRANKL (Fig. 3) and the lowest leptin concentration showing significant (p < 0.05) inhibition of osteoclast formation decreased from 30 to 3 nM. As shown in the right panel of Fig. 3, resorption is reduced and is a less reliable parameter when a submaximal sRANKL concentration is used.
Inhibitory effect of leptin in the presence of submaximal sRANKL concentration. PBMCs (106 cells/well) were cultured for 21 days on bone slices with hM-CSF (40 ng/ml) and sRANKL (5 ng/ml) in the presence of increasing concentrations of leptin. Representative data from one experiment (n = 8 bone slices per treatment). Groups are significantly different if the annotations do not contain the same letter; p < 0.05; analysis of variance (ANOVA); Fisher's one-way multiple comparison.
To confirm that the inhibitory effect of leptin was specific and not caused by a contaminating factor, we tested other batches of human leptin obtained from Peprotech and also human and mouse leptin obtained from the NHPP. All leptin preparations tested produced essentially similar responses in the presence of 40 ng/ml of sRANKL (data not shown). The NHPP human leptin was used in the experiment shown in Fig. 3 that examined the response in the presence of 5 ng/ml of sRANKL.
Osteoclast generation—washed PBMC and CD14+ cells
Although adherent PBMCs are predominantly monocytes, our unwashed PBMC cultures were invariably contaminated by other nonadherent or weakly adherent cells, particularly lymphocytes and platelets. Because the human osteoclast precursor in PBMC has been shown to be the CD14+ monocyte,34, 37 we wondered whether leptin acted directly on these precursors or indirectly via a CD14− cell type. To address this question, osteoclastogenesis assays were established using either washed PBMCs or 95-97% pure CD14+ PBMCs. Osteoclast generation and bone resorption equivalent to that obtained from 106/well unwashed PBMCs was obtained using 106/well washed PBMCs or 105/well CD14+ cells, but in both cases leptin had no effect on either osteoclast number or bone resorption (Fig. 4).
Leptin does not inhibit osteoclast generation in cultures of washed PBMCs or purified CD14+ cells. PBMCs were prepared as described in the Materials and Methods section and then CD14+ cells were isolated using MACS. Cells were cultured on bone slices with hM-CSF and sRANKL in the absence or presence of leptin (100 nM). Columns represent mean ± SEM of data from eight experiments for PBMC cultures and four experiments for the CD14+ cultures using different blood donors. (A) Comparison of unwashed and washed PBMCs. (B) Comparison of unwashed PBMCs and purified CD14+. Groups are significantly different if the annotations do not contain the same letter; p < 0.001; analysis of variance (ANOVA); Fisher's one-way multiple comparison.
Osteoclast generation—mouse spleen
To confirm that the inhibitory effect on osteoclastogenesis is not species specific, we also tested leptin in assays using mouse spleen cells as osteoclast precursors. After 10 days incubation with hM-CSF (25 ng/ml) and sRANKL (40 ng/ml), mouse spleen cells generated numbers of osteoclasts similar to those found using unwashed human PBMCs as precursors. Leptin inhibited mouse osteoclast generation (Fig. 5), although the magnitude of the inhibition was substantially less than that seen routinely in the human assay (approximately 35% vs. 83%). Statistically significant inhibition of osteoclastogenesis did not occur below 100 nM of leptin. However, the inhibitory effect of leptin on resorption was more sensitive (significant at 1 nM, p < 0.05) and of greater magnitude (approximately 60%).
Leptin inhibition of osteoclast generation and bone resorption in mouse spleen cultures. Mouse spleen cells were cultured on bone slices for 10 days in the presence of hM-CSF (25 ng/ml) and sRANKL (40 ng/ml) in the absence or presence of leptin (1, 10, or 100 nM). Columns represent mean ± SEM of data from four experiments (48 bone slices). Groups are significantly different if the annotations do not contain the same letter; p < 0.001; analysis of variance (ANOVA); Fisher's one-way multiple comparison.
Regulation of OPG and RANK—unwashed PBMC
The osteoclastogenesis experiments established that the inhibitory effect of leptin is dependent on the presence of CD14− PBMC but not dependent on the presence of stromal/osteoblasts. In addition, because sRANKL and hM-CSF had been added to the cultures, the mechanism is unlikely to involve modulation of the expression of these factors. Considering the potent inhibitory effect of OPG on osteoclast generation,32 we next examined the effect of leptin on expression of OPG messenger RNA (mRNA) in PBMCs and found that it was markedly increased. Leptin induced a concentration-responsive increase in OPG mRNA in PBMCs and a reciprocal decrease in RANK mRNA (Figs. 6A and 6B; Table 2). Western analysis of cell lysates from cultures of PBMCs treated with leptin showed increased OPG protein expression (Fig. 6C).
|
|
Effect of leptin on OPG and RANK mRNA expression in human PBMCs. Unwashed PBMCs cultured in tissue culture flasks were treated with vehicle or leptin (2, 20, or 200 nM) for 24 h. (A) RT-PCR analysis of OPG mRNA and RANK mRNA expression. (B) RANK and OPG mRNA expression are presented as ratios compared with GAPDH mRNA expression. (C) Western analysis of OPG protein expression in unwashed PBMCs treated with vehicle (control) or leptin (100 nM) for 24 h. Giant cell tumor (GCT) of bone was used as a positive control. Results are representative of two experiments.
To determine the time course of OPG expression in response to leptin, PBMCs were treated with leptin (100 nM) or vehicle for various times when the cells and media were harvested for OPG mRNA and protein measurement, respectively (Fig. 7). The OPG mRNA treated/control ratio (measured by real-time PCR, normalized for β-actin mRNA expression) rose by 4 h and peaked at 12 h at a level approximately 10 times above baseline. OPG protein in the media (measured by ELISA) increased markedly by 12 h in leptin-treated cultures. There also was an increase in OPG protein secretion in control cultures that did not begin until 24 h.
Short time course of OPG mRNA and protein expression in human PBMC after leptin treatment. Unwashed PBMCs were incubated in 24-well plates with media containing hM-CSF (25 ng/ml). They were treated with vehicle (control) or leptin (100 nM) for the times indicated when the cells were harvested for mRNA and the media for protein. OPG and β-actin mRNA were quantified by real-time PCR and OPG protein by ELISA. (A) OPG mRNA normalized for β-actin mRNA expression treated/control ratios. (B) OPG protein concentration in media (pg/ml).
Regulation of OPG and RANK—CD14+ versus CD14− cultures
Untreated purified CD14+ cultures did not exhibit any detectable OPG mRNA expression and, in contrast to PBMCs, treatment with leptin did not result in increased expression (Table 2; Fig. 8). Furthermore, in these cells, OPG protein levels were below the detection limit of the ELISA assay (<32 pg/ml) with and without treatment with leptin for 24 h. However, untreated CD14− cells expressed OPG mRNA and treatment with leptin (100 nM) for 12 h increased this 7.4-fold. In two experiments with CD14− cells treated with leptin (100 nM) for 24 h, OPG levels in the media increased from <32 to 70 pg/ml and from 100 to 145 pg/ml, respectively. RANK mRNA was readily detectable in untreated CD14+ cultures and treatment with leptin for 24 h resulted in 3- to 6-fold reduction in expression (Table 2; Fig. 8). We did not measure RANK protein.
OPG and RANK mRNA expression in leptin-treated CD14+ monocytes. CD14+ cells were purified from PBMCs using MACS. Cells were incubated in media containing hM-CSF (25 ng/ml) and treated with vehicle or leptin (100 nM) for 24 h. OPG and RANK mRNA expression were assessed by RT-PCR.
Expression of leptin receptor
To identify targets for leptin in PBMCs, we examined mRNA expression for the long form of the leptin receptor38 [Ob-R(L)] in various fractions sorted according to CD expression. Ob-R(L) mRNA was detectable in all fractions examined including CD14+, CD4+, CD8+, nonadherent CD4−/CD8−, and granulocytes (results not shown).
DISCUSSION
We have shown that leptin inhibits the in vitro differentiation of cells present in the PBMC fraction of human blood to osteoclasts and that the inhibitory effect is dependent on the presence of CD14− PBMCs. We also have shown that leptin inhibits osteoclast differentiation from mouse spleen cells. Thus, we have identified leptin as a potential local inhibitor of bone resorption in vivo. Thomas et al.21 have shown that leptin also has a local effect to increase bone formation. This effect, together with simultaneous inhibition of bone resorption, as implied by our results, could lead to substantial increases in bone mass and bone strength. Indeed, recent preliminary data of Cornish et al.39 showed that systemic administration of leptin to mature male mice for 4 weeks increased bone strength by >20%. On the other hand, the results of Ducy and colleagues19 suggest a central role of leptin to decrease bone mass. Thus, there is now an increasing body of evidence that leptin can regulate bone metabolism in multiple ways, by both locally and centrally mediated mechanisms.
Previous reports have documented effects of leptin on a variety of hemopoietic cells, including multilinear myeloid and lymphoid progenitors and it has been shown to induce proinflammatory responses in mature hemopoietic cells.5 Among the identified targets for leptin is the candidate osteoclast precursor, the peripheral blood monocyte, which is activated by leptin.33 We have found that leptin acts directly on CD14+ monocytes to down-regulate expression of RANK, the receptor for RANKL. However, this alone is not sufficient to inhibit osteoclast differentiation, which is dependent on the presence of nonadherent or weakly adherent CD14− PBMCs, most probably lymphocytes. Both CD4+ and CD8+ T lymphocytes express leptin receptor and leptin enhances their activation and increases their synthesis of interleukin-2 and interferon-γ.40, 41 We have confirmed mRNA expression of the long form of leptin receptor in CD4+ and CD8+ T lymphocytes and also found expression in nonadherent CD4−/CD8−, which would mostly be B lymphocytes. Therefore, there are a number of potential target cells in the CD14− PBMC population that might be mediating the effect of leptin on osteoclast generation. When the osteoclastogenesis assays were treated with leptin for periods of <3 weeks, we found that leptin was a much more effective inhibitor when added in the earlier part of the culture, particularly the first week. This finding suggests that leptin inhibits an earlier, rather than later, phase of osteoclast differentiation. Alternatively, this temporal effect may be caused by progressive depletion of the mediating CD14− PBMCs with time because of media changes and cell death.
OPG was investigated as a candidate mediator of the leptin effect because of its known potent inhibitory effect on osteoclastogenesis.31, 32 We found that OPG expression was stimulated by leptin in PBMCs but not in purified CD14+ cells, suggesting that OPG production by one or more cell types in the CD14− PBMC population may be responsible for the inhibition of osteoclast generation seen. Leptin increased OPG mRNA expression by 4 h and protein by 12 h, suggesting a direct effect on a target cell, rather than indirect effect mediated by another factor(s) or the result of leptin-induced phenotypic change. There are no previous reports of leptin modulating OPG expression, although it has been shown to be up-regulated in B cells and dendritic cells by stimulation of CD40,42 a related member of the TNF receptor superfamily. Although we do not have any direct evidence that leptin inhibition of osteoclast generation is mediated by OPG, we believe that this could be obtained by using hemopoietic cells obtained from OPG-deficient (“knockout”) animals or by the use of an anti-OPG blocking antibody. We have attempted the latter experiments but find that the anti-OPG antibodies used directly inhibit our osteoclast generation assay.
Circulating concentrations of leptin in humans are in the range of 0.1-5 nM.12, 13 Although, in assays using a high concentration of sRANKL (40 ng/ml) we did not observe effects of leptin on bone resorption below 10 nM, significant inhibition of osteoclast generation was seen with 3 nM of leptin when the sRANKL concentration was reduced to 5 ng/ml. Furthermore, leptin significantly inhibited bone resorption at 1 nM in mouse spleen osteoclastogenesis assays. Thomas et al.21 found that leptin concentrations of ≥10 nM were required to promote differentiation of stromal cells and concentrations in the range of 0.1-10 nM have been shown to be effective in various hemopoietic in vitro assays.3-5, 33, 41, 43, 44 Leptin sensitivity may be reduced in our assay because of a suboptimal density of the putative CD14− cell type(s) mediating the response. Furthermore, because bone marrow adipocytes secrete leptin apparently at high levels,45 local concentrations of leptin in the bone marrow microenvironment are likely to be significantly higher than systemic levels.
Ducy et al.19 showed that complete absence of leptin signaling, as found in the leptin-deficient (ob/ob) and leptin receptor-deficient (db/db) mouse, is associated with high bone mass. They also showed that i.c.v. infusion of leptin in ob/ob or wild-type animals resulted in reduced bone formation and loss of bone mass via an undefined mechanism. Their results imply the existence of a neuroendocrine feedback loop enabling leptin to regulate negatively bone formation, analogous to its regulation of appetite. This apparent anomaly raises an important question; why do obese humans with high plasma leptin levels have high bone mass? It is possible that local anabolic effects (which would not be operative in the experiments of Ducy et al.19 because leptin was administered i.c.v.) predominate in this situation. Another explanation is that in overweight humans, the neuroendocrine effects of systemic leptin are different from those when leptin is given i.c.v. to mice. The reason that overweight or obese humans remain so, despite elevated plasma leptin levels in most, is not understood although leptin “resistance” or “relative deficiency” have been implicated.46-48 A similar dysregulation of the central effect of leptin on increased bone resorption might occur in overweight and obese humans.
We have shown that leptin inhibits osteoclast generation and bone resorption via a local mechanism(s) confined to the hemopoietic lineage. This new finding, together with the previous knowledge that the different interrelated phenotypes of the bone marrow mesenchyme can secrete or respond to leptin indicates that leptin may be an important regulatory cytokine within the bone marrow microenvironment. Bone is remodeled continually throughout adult life and maintenance of bone mass relies on close “coupling” of the bone resorption and formation cycles.49 Extrinsic or intrinsic perturbations that produce shift of stromal differentiation toward adipocyte phenotype will directly result in reduced bone formation and indirectly result (via adipocyte-derived leptin) in reduced bone resorption. Shifts toward osteoblast phenotype will have opposite effects. Thus, leptin may contribute to the linkage between bone formation and resorption rates. Because leptin produced by bone marrow adipocytes will inhibit further adipocyte differentiation, the potential exists for homeostatic autoregulation. In addition, leptin's dual local role of reducing resorption while increasing osteoblast differentiation21 suggests that it may have an important bone-conserving function. Involvement of leptin in the antiosteoporotic effect of estrogen is suggested by the observations that estrogen increases leptin production50 and that leptin levels fall after menopause, despite increases in body mass index.51 Although we have shown that leptin regulates OPG and RANK expression in PBMCs, a precise molecular mechanism linking this with the antiosteoclastogenesis effect of leptin remains to be defined.
Acknowledgements
We thank the Genetics Institute for providing hM-CSF, Dr. A.F. Parlow of The National Hormone and Pituitary Program for providing mouse and human leptin, and Prof. Peter Choong of St. Vincent's Hospital, Melbourne, Australia for providing giant cell tumors of bone. This work was supported by the National Health and Medical Research Council of Australia (project grants 980746 and 114248).
