In this issue of the Journal of Bone and Mineral Research, an article by Turner and colleagues1 addresses the important topic of the relationship between the adipocyte-derived hormone leptin, bone, and energy homeostasis. They report that the leptin-deficient and obese ob/ob mice and the leptin receptor–deficient db/db mice display an osteopetrotic-like skeletal phenotype at 15 weeks of age with high vertebral trabecular bone volume (BV/TV) despite a markedly decreased bone formation rate (BFR), as a result of decreased bone resorption. They also show that these phenotypes (low BFR and low resorption) can be corrected both by subcutaneous injections of leptin and by leptin hypothalamic gene therapy. Because this suggested that leptin can act both at the periphery and in the central nervous system to affect bone, they then performed marrow transplantation experiments in lethally irradiated mice. They report that transplantation of leptin receptor–deficient db/db bone marrow into lethally irradiated wild-type mice drastically reduces bone formation to levels indistinguishable from the db/db mice. Furthermore, mice engrafted with db/db marrow did not show overt alterations in energy homeostasis, suggesting that energy homeostasis is not affected by the changes in bone. They conclude that leptin acts on bone primarily through peripheral pathways and that its effects are to increase bone formation, but even more bone resorption. These results are in strong opposition with prevailing theories and call into question what leptin does in bone; ie, increasing or decreasing bone formation, and how it does it; ie, centrally through a hypothalamic relay or directly at the periphery.
These findings raise several puzzling questions. First, the ob/ob and db/db skeletal phenotypes reported here are radically different from a previous report showing a 45% to 70% increase in BFR in ob/ob and db/db mice, respectively, compared to wild-type littermates.2 Second, the ability of peripheral leptin to affect bone, albeit supported by in vitro data published by others3 is in sharp contradiction with some in-depth studies from other laboratories.4, 5 Third, peripheral leptin corrects the bone phenotype, although no significant alterations in energy homeostasis are noted, suggesting no direct link between bone and energy regulation in these leptin-deficient or leptin receptor–deficient mouse models.
The original concept and the prevailing school of thought concerning the effects of the fat-derived hormone leptin on bone rely mainly on a series of pioneering and robust in vivo studies conducted by the Ducy and Karsenty groups (initially at the Department of Molecular and Human Genetics, Baylor College of Medicine, Houston, TX, USA; then at the Department of Pathology and Cell Biology [Ducy] and the Department of Genetics and Development [Karsenty], College of Physicians and Surgeons, Columbia University, New York, NY, USA). In 2000, these investigators presented their first in vivo evidence to the fact that bone gain in the ob/ob or db/db mice occurred despite gonadal failure, resulting from the absence of leptin signaling.2 Both ob/ob and db/db mice exhibited high trabecular bone mass in the proximal tibiae and vertebrae, whereas intracerebroventricular (ICV) infusion of leptin into ovariectomized mice corrected this phenotype and resulted in a decreased BV/TV in both mutant and control groups. Interestingly, direct in vitro leptin treatment of primary ob/ob osteoblasts had no effect on collagen synthesis or mineralization, whereas db/db osteoblasts were indistinguishable from wild-type in culture, suggesting that leptin acted on bone solely via hypothalamic pathways. Moreover, the central action of leptin was proposed to target osteoblasts and not osteoclasts, because these authors found normal osteoclast function in the absence of leptin signaling. Subsequent work using β2-adrenergic receptor Adrb2-null mice or pharmacological treatment with β-blockers further supported the concept of a central inhibitory role of leptin on the skeleton, and proposed a model whereby activation of leptin receptors in the brain activates the sympathetic nervous system, which in turn suppresses osteoblast activity and increases osteoclast numbers via osteoblast-derived receptor activator of NF-κB ligand (RANKL).4 Furthermore, this group performed transplantation of wild-type bone marrow into irradiated Adrb2-null mice, and this normalized bone formation parameters; conversely, transplantation of Adrb2-null marrow into irradiated wild-type mice increased bone formation significantly, confirming that signaling from the sympathetic nervous system through Adrb2 in osteoblast is required for leptin's antiosteogenic function.6
Despite the strong evidence produced by these laboratories, other groups were less successful in reproducing the negative effects of leptin on bone, obtaining either partially consistent or opposing results. Steppan and colleagues7 reported that peripheral administration of leptin increases bone density and bone mineral content in ob/ob mice, suggesting not only that peripheral leptin can affect bone (although leptin can cross the blood-brain barrier and could have acted centrally) but also that the effect of leptin on bone is to increase bone density, opposite to the model suggested by Karsenty et al. Similarly, Burguera and colleagues8 showed counteracting effects of leptin on bone loss induced by estrogen deficiency, again suggesting an overall stimulatory effect of leptin on bone formation. Bridging the two opposing theories, Hamrick and colleagues9 reported that, consistent with observations from the Karsenty group, ob/ob mice exhibit a high bone mass in the lumbar vertebrae, but a contrastingly low trabecular and cortical bone mass in the femur, proposing a concurrence of both axial catabolic and appendicular anabolic effects of leptin on the skeleton. To add complexity, several in vitro studies pointed toward a direct positive effect of leptin on osteoblast differentiation and inhibition of osteoclastogenesis, again putting into question the absolute dependency on central and neuronal regulation as well as the nature of the effects of leptin on bone cells.3
In their study, Turner and colleagues1 derived an alternative model whereby leptin enhances bone turnover; ie, both bone formation and bone resorption, and this primarily via peripheral pathways. Thus, suppression of leptin signaling in ob/ob and db/db mice leads to low BFR and osteocalcin levels and even lower serum cross-linked C-telopeptide (CTX), despite increased osteoclast surface, leading to high bone volume. As a first step, the authors conducted a histomorphometric analysis of obese ob/ob and db/db mice in comparison to lean wild-type littermates. Consistent with Ducy and colleagues,2 vertebral bone volume was increased by approximately twofold. Interestingly, however, and in contrast to the earlier report from Ducy and colleagues,2 Turner and colleagues1 found a lower trabecular BFR in the lumbar vertebrae and femurs. Furthermore, again in divergence with Ducy and colleagues,2 Turner and colleagues1 report a suppressed number of osteoblasts, whereas Ducy and colleagues2 showed no changes in number of osteoblasts. Like Ducy and colleagues,2 however, they found higher osteoclast numbers in femurs, but report a decrease in CTX and a novel and critically important finding: a marked increase in cartilage remnants in the primary spongiosa of ob/ob and db/db mice. These are characteristic features of osteopetrosis and, together with the markedly decreased serum CTX values, solidly establish the fact that bone resorption is markedly decreased in absence of leptin signaling. This finding explains the increased BV despite significant decreases in osteoblast numbers and BFR. Taken together, these findings point toward reduced bone formation as well as even further reduced bone resorption; ie, an overall suppression of bone turnover with a positive balance, in the leptin-deficient animals.
The contrasting histomorphometric data in these articles are difficult to explain. Of note, however, both groups agree on an increase in vertebral BV/TV in ob/ob mice, although Ducy and colleagues2 also reported higher bone mass in the femoral bone, whereas Turner and colleagues1 showed no change. But the two groups differ markedly in their interpretation on how this high BV came about. Ducy and colleagues,2 who studied 1-month-old ob/ob mice that had been fed a low-fat diet (before their weights exceeded those of wild-type controls), as well as 3- and 6-month-old obese animals, attribute this to their finding of an increased bone formation at all three time points. This led these authors to conclude that leptin has a negative effect on osteoblasts. In contrast, Turner and colleagues,1 who studied only the already obese group at 4 months of age, find that bone formation is decreased but bone resorption is decreased even more, leading to high BV/TV and to the opposite conclusion that leptin exerts a positive effect on bone formation (and bone resorption). It is possible that differences in the severity and duration of hyperinsulinemia in this strain of mice may generate such diverse findings because insulin signaling is known to regulate bone remodeling through both osteoblast and osteoclast function.10–13 At present, however, no clear explanation for such divergent histomorphometric results can be provided. More careful and detailed studies of bone remodeling and metabolism in ob/ob and db/db mice are therefore required.
The second major point of divergence between the findings of these two groups of investigators is where the effect of leptin is being exerted: in brain, in bone, or in both. In agreement with Steppan and colleagues,7 Turner and colleagues1 report that subcutaneous administration of leptin for a period of 3 weeks effectively reversed most of the bone formation parameters, inducing a higher BFR and osteoblast-lined bone surface. Because subcutaneously administered leptin can cross the blood brain barrier, these changes could still be attributed to the effects of leptin in the hypothalamus. To circumvent this problem, Turner and colleagues1 designed a series of experiments to distinguish central (indirect) from peripheral (direct) actions of leptin on bone formation. To address the centrally mediated effects, vector-carrying adenoassociated viruses expressing leptin or green fluorescent protein (GFP) control were overexpressed by stereotaxy in the third ventricle of the ob/ob mice hypothalamus, where leptin receptors are expressed abundantly. This induced a reversal of the high vertebral BV and an increase in osteoblast-lined bone perimeter, establishing that indeed at least some of the effects of leptin on bone formation are exerted centrally. Unfortunately, no information was provided on BFR or bone resorption in these experimental animals. Consequently, it is quite difficult to reconcile the increase in osteoblasts with a decrease in BV, unless BFR and/or bone resorption were very low, which would contradict the proposed hypothesis that leptin favors bone formation, although it could increase osteoblast numbers. Alternatively, it could indicate that central leptin inhibits and peripheral leptin favors bone formation, a possibility not discussed in the article. Using the same approach, this group had previously reported normalization of high trabecular bone mass abnormalities in the distal femur and vertebrae with leptin gene therapy,14 showing successful reduction of vertebral bone mass in either ob/ob or wild-type mice following ICV injection of leptin, consistent with Ducy and colleagues.2 It should, however, be noted that an analogous study by Bartell and colleagues15 using a shorter-duration (12 days instead of 28 days) and higher leptin dose (1.5 µg/d versus 0.19 µg/d) resulted in an increased bone mineral density and mineral apposition rate, confirming that leptin can act on bone via the hypothalamus, but showing that it increases (not decreases) bone formation, thereby adding to the confusion about what are indeed the effects of central leptin on bone. These somewhat contradictory results certainly raise concern about the net effect of central leptin on bone formation and bone resorption and whether the central effects are similar or opposite to the effects of peripheral leptin.
Regardless of the nature of leptin's action in bone, these experiments establish that leptin can affect bone via the hypothalamus but do not exclude the possibility that leptin can also affect bone directly at the periphery. Of note, being produced by adipocytes, leptin is therefore present in bone marrow and does not need to circulate in order to affect bone cells, provided they have the appropriate receptors for this adipokine, as suggested by several studies.16–19 To address this point, Turner and colleagues1 deleted leptin receptors specifically from the bone microenvironment through irradiation and bone marrow transplantation. Remarkably, the absence of leptin receptor following transplantation of db/db bone marrow cells in lethally irradiated wild-type mice resulted in markedly suppressed BFR in the femur and lumbar vertebrae, to values indistinguishable from db/db mice. This firmly establishes that leptin can also act directly within the bone environment and that its effect is to increase bone formation. Here again, however, the data is very limited and no information is provided about the effects of these transplantation experiments on bone resorption or osteoclast numbers.
The last point of divergence between this study and the prevailing concepts is related to the relationship between leptin, bone, and the regulation of energy expenditure. Even though Turner and colleagues1 reevaluate the model of exclusively central regulation of skeletal remodeling by leptin, the functions of leptin in energy metabolism are still suggested to be restricted to the brain, in congruence with current literature. Earlier work by Ferron and colleagues20 and Hinoi and colleagues21 has shown that bone functions as an endocrine organ, capable of regulating total body homeostasis through the release of osteocalcin, which circulates to increase pancreatic insulin secretion and adiponectin-related insulin sensitivity. Centrally-acting leptin was later proposed to affect this metabolic activity of bone-derived osteocalcin via sequential activation of the sympathetic nervous system and upregulation of Esp (embryonic stem-cell protein tyrosine phosphatase), affecting osteocalcin's carboxylation state.22 Keeping in mind the reciprocal regulation of energy homeostasis and the skeleton, it is possible to stretch the logic a step further, proposing that factors affecting bone biology may interfere with total body energy balance. An alternate possibility may be that centrally acting leptin holds a sovereign role in the maintenance of energy homeostasis, and whereas peripheral alterations in leptin signaling control bone remodeling, they are not potent enough to balance energy. Here the authors show that food intake, body weight, abdominal fat pad weight, and blood glucose of wild-type mice engrafted with db/db-derived marrow cells did not resemble obese db/db mice and remained identical to the wild-type metabolic phenotype. Turner and colleagues1 conclude that peripherally acting leptin and the changes it induces in bone do not affect overall energy metabolism, subsiding to the predominantly central pathway. The data presented, however, lacks comprehensive analysis of energy expenditure, including heat and locomotion, and lacks information about bridging molecules between bone and energy such as osteocalcin, insulin, and adiponectin. Although striking, the short duration of these experiments and the narrow scope of metabolic profile examination somewhat limits the derivation of conclusions with regard to the central versus peripheral effect of leptin on energy metabolism. More careful and extensive evaluation is required to fully address this puzzling question.
In summary, and despite the limitations of this study, the article by Turner and colleagues1 raises several important questions. First, is bone formation increased or decreased in leptin-deficient ob/ob or leptin receptor–deficient db/db mice? Second, is bone resorption also markedly decreased in these mice? Third, does central administration of leptin increase or decrease bone formation? Does it affect bone resorption? Fourth, is leptin acting both centrally and peripherally and are its effects on bone formation similar or opposite? Fifth, do alterations in leptin signaling in the bone environment affect energy homeostasis, besides its known central effects? Finally, if the data presented in this article strongly suggest that leptin can affect bone formation through both central and peripheral signaling, it falls short of providing a clear view on the respective roles and effects of these two pathways on bone remodeling and energy homeostasis. This article should nevertheless encourage researchers in our field to revisit these important questions.