For decades, investigators from multiple disciplines have pondered the complex relationship governing the distinction between bone and fat within the marrow microenvironment. Pioneering work led by Friedenstein, Owen, and others defined what is now referred to as the bone marrow mesenchymal stromal/stem cell (BMSC) as the common progenitor for both marrow adipocytes and osteoblasts.1, 2 The relative proportion of these populations changes as a function of advancing age and disease states, such as osteoporosis and aplastic anemia. In their seminal 1992 work, Beresford, Owen, and their colleagues at Oxford University introduced data supporting the hypothesis that an inverse relationship exists between BMSC adipogenesis and osteogenesis,3 namely, that factors promoting BMSC adipogenic differentiation did so at the expense of osteogenesis and vice versa.3 This study has set the paradigm that has since been the subject of multiple in vitro and in vivo analyses.3 Lineage determination by BMSCs has been compared with a seesaw or teeter-totter governed by the relative strength of the adipogenic versus osteogenic factors present within the microenvironment.4
Almost simultaneous with the publication of Beresford and colleagues,3 the gene encoding leptin, the first modern adipokine, was mapped and cloned in the ob/ob murine strain.5, 6 These mice, identified in the 1950 s, were remarkable for their display of hyperphagia and obesity.7 The discovery of leptin revolutionized our narrow view of adipose tissue as simply a passive energy-storage reservoir. Since then, thousands of articles on leptin alone have documented that adipose tissue is a dynamic endocrine organ contributing to systemic feedback loops integrating multiple peripheral organs with the central nervous system (CNS) as well as adipose tissue metabolism itself.8 Leptin's actions are mediated in part through multiple isoforms of its surface receptor which interact with the Janus Kinase (JAK) and STAT transcription factors. Both short and long forms of the leptin receptor exist on undifferentiated BMSCs, on hematopoietic cells, within the hypothalamus of the CNS, on adipocytes, on osteoblasts, and elsewhere. While the ob/ob murine strain displays an inactivating mutation of the leptin gene, the db/db murine strain and the rat fa/fa counterpart reflect an inactivating mutation in the long (but not short) isoform of the leptin receptor.9, 10 While rodents with all three mutations exhibit hyperphagia and obesity, only the ob/ob strain can be treated effectively by parabiosis to a wild-type control mouse with normal circulating levels of leptin or by treatment with recombinant leptin.8, 11, 12
A number of studies had reported the impact of leptin or leptin receptor mutations on bone biology in rodent models.13–17 Furthermore, leptin expression was detected in bone marrow adipocytes in vivo and found to promote osteogenesis in BMSCs in vitro, implicating leptin as an autocrine and/or paracrine factor locally.18, 19 However, Ducy and colleagues' publication in 2000 has remained a landmark study in the field.20 Using histologic analyses of femoral and vertebral bones, Ducy and colleagues reported that ob/ob mice exhibited increased bone mass relative to wild-type controls.20 Following chronic intracerebroventricular (ICV) leptin administration, bone mass and strength were reduced in ob/ob and wild-type C57BL6 mice based on X-ray analysis, three-point bending, and histology.20 Leptin actions were linked to the CNS via feedback loops mediated by the sympathetic nervous system (SNS) and the β-adrenergic receptor pathway.20 This set the stage for subsequent publications documenting leptin as a bone catabolic agent.21, 22 This conclusion conforms to the hypothesis that adipose and bone tissue are inversely related because an abundance of adipose tissue and the secretion of leptin downregulate the formation of bone.
If only it was this simple! A growing body of evidence complicates the story by suggesting that leptin plays an anabolic role in bone formation. Independent studies reported that ob/ob mice display reduced femoral bone mineral content and density and trabecular bone volume relative to wild-type controls.23 Furthermore, leptin administration subcutaneously increased bone mineral content while reducing marrow adipocyte numbers in ob/ob mice.24 Likewise, in rat models, leptin administration reduced bone loss owing to ovariectomy and reduced marrow adipogenesis resulting from tail suspension.25, 26
The current issue of the Journal of Bone and Mineral Research brings new evidence to this continuing dilemma.9, 10 First, using the ob/ob model, Bartell and colleagues directly compare the impact of ICV and subcutaneous leptin administration on bone.9 As expected, the mice showed reduced food intake, body weight, and serum glucose and insulin levels following 12 days of leptin administration.9 Furthermore, the mice showed increased mineral apposition rates, bone mineral content, and bone mineral density and decreased marrow adipocyte numbers based on histologic and dual-energy X-ray absorptiometric (DXA) analyses.9 This was accompanied by increased BMSC expression of osteogenic mRNAs and reduced expression of mRNAs associated with adipogenesis and osteoclastogenesis.9 At the serum level, leptin administration correlates with significant increases in circulating insulin-like growth factor 1 (IGF-1), osteocalcin, osteoprotegerin (OPG), and receptor activator of NF-κB ligand (RANKL) levels.9 Second, Williams and colleagues use state-of-the-art micro–computer tomographic (µCT) methods to perform a comprehensive comparison of bone parameters in tibias (appendicular skeleton) and lumbar vertebrae (axial skeleton) between wild-type and db/db mice.10 In the tibias, cortical and trabecular bone volume and thickness, as well as trabecular number, were decreased significantly in db/db mice.10 In the vertebrae, cortical and trabecular bone thicknesses were decreased significantly in db/db mice, and while other parameters were decreased, they did not achieve statistical significance.10 These changes in bone measures were paralleled by a 40% decrease in circulating osteocalcin levels in db/db mice as well as significantly reduced bone strength and stiffness based on three-point bending and nanoindentation tests.10 Together, both publications lend further support to the conclusion that leptin serves an anabolic rather than catabolic role with respect to osteoblast and chondrocyte function.9, 10 Williams and colleagues10 suggest that anorectic actions of leptin and reduced caloric intake might explain the apparent catabolic action of ICV leptin reported by Ducy and colleagues.20
What else might account for these discrepancies in the literature regarding leptin's osteogenic actions? In their Discussions, both Bartell and colleagues and Williams and colleagues spotlight some of the technical and theoretical issues that might be responsible (these and other are summarized in Table 1).9, 10 It is unlikely that the background strain of the murine lines is responsible because the major studies all have used a C57BL6 background originating from Jackson Laboratories. Nevertheless, within-strain variations have been reported for other body phenotypes such as weight gain,27 and the bone parameters of the ob/ob and db/db mice may prove to be another example. However, this is essentially a conclusion of last resort. There are other parameters to consider. One example is leptin dosage. While Bartell and colleagues9 administered 0.38 and 1.56 µg of leptin per day intracerebroventricularly, Ducy and colleagues20 administered a concentration of 8 ng/h intracerebroventricularly (or 0.192 µg/d). Likewise, as highlighted by Williams and colleagues,10 histomorphometric and µCT analyses may yield different outcomes and conclusions owing to the unique advantages and technical limitations each method presents. While each of these elements may be contributory, none can explain fully the current state of conflicting evidence.
|Dynamics||Experimental model: In vitro versus in vivo|
|Scope of gene mutation: Tissue-specific versus whole-body (global) mutation|
|Site of action: Central versus peripheral|
|Route of administration: intracerebroventricular versus subcutaneous|
|Bone site (gross): Appendicular versus axial|
|Bone site (micro): Cortical versus trabecular|
|Age: Young versus old|
|Variables||Dose level of leptin|
|Delivery mechanism and frequency|
Do human clinical studies shed any light on these questions? Independent investigations in Estonia, Japan, Turkey, and the United States have correlated circulating leptin and related adipokine levels with bone mineral density in pre- and postmenopausal women.28–34 Studies of healthy postmenopausal women in Estonia (n = 88) and Japan (n = 139) positively associated leptin levels with increased bone mineral density,28, 32 but when controlled for hormonal influences or insulin resistance, the leptin association with bone mineralization was no longer significant.28 In contrast, a Turkish study evaluating premenopausal (n = 24) and postmenopausal women with (n = 36) and without (n = 30) osteoporosis failed to document a significant correlation between leptin levels and bone mineral density.33 Consistent with this is a recent US-based prospective study (n = 3075) that did not find any correlation of leptin levels with the risk of fracture (n = 406) in either elderly men or women over a mean of 6.5 years.31 Thus the jury is still out on the impact of circulating leptin levels on human bone biology.
What are the next directions? To date, studies have used global mutations of the ob and db genes alone or in the context of adrenergic or other receptor mutations. To dissect the role of leptin and its receptor discretely at the peripheral (BMSC) and central (hypothalamic) levels, more selectively restricted gene modifications will be required. Cell- and tissue-specific deletion of the ob and db genes using Cre/lox technologies in transgenic mice may address this impasse. Both µCT and histomorphometry should be used in parallel to assess bone parameters in such studies in both genders. Another variable to consider is circadian timing. Endogenous circulating leptin levels exhibit robust circadian oscillation,35 and the circadian transcriptional/translational regulatory system has peripheral (BMSC) and central (hypothalamic/suprachiasmic nucleus) components known to regulate both adipogenesis and osteogenesis.36, 37 Thus the timing of leptin administration itself may be a modulator in this complex endocrine system.