Osteoblasts and adipocytes are derived from a common bone marrow progenitor (Prockop, 1997; Vaananen, 2005; Valtieri and Sorrentino, 2008). The primary function of osteoblasts is well defined. Osteoblasts generate bone matrix to support bone growth, turnover, and remodeling. Additionally, osteoblasts have been reported to participate in hematopoietic stem cell mobilization from their bone marrow niche and to support hematopoiesis (Nagayoshi et al., 2006; Mayack and Wagers, 2008). In contrast, the physiological significance of adipocytes residing within bone marrow is less well defined.
Most fat depots act as dynamic energy reservoirs to maintain circulating triglyceride and free fatty acid levels (Caserta et al., 2001). During severe caloric restriction, fat catabolism becomes an important source of energy and peripheral fat depots decline in mass. In contrast, severe caloric restriction is associated with increased bone marrow adiposity (Batista et al., 1988; Bohm, 2000; Cui et al., 2006; Duque, 2008; Bredella et al., 2009). Bone marrow adipocytes sequester and esterify fatty acids during starvation, a finding which has fostered the belief that the bone marrow fat depot does not function as an important energy reservoir for peripheral tissues (Tavassoli, 1974; Bathija et al., 1979; Tran et al., 1981; Gimble et al., 1996; Abella et al., 2002). However, bone marrow adipocytes have been implicated in regulating cell differentiation through their generation and release of adipokines (Kilroy et al., 2007).
c-Kit (CD117), a type 3 receptor tyrosine kinase, is a member of the platelet-derived growth factor (PDGF) family of cytokine receptors (Qiu et al., 1988). Kit receptor-deficient KitW/W-ν mice have been reported to have hypertriglyceridemia and hypercholesterolemia (Hatanaka et al., 1986). Chylomicrons, very low density lipoprotein, and intermediate density lipoprotein are higher in the KitW/W-ν mice compared to WT mice. Additionally, KitW/W-ν mice have reduced lipoprotein lipase activity (Hatanaka et al., 1986; Langner et al., 1989). These observations, made prior to recognition that the white spotting locus (W) codes for c-kit, suggest that c-kit signaling plays an important but largely underappreciated role in fat metabolism.
A reciprocal relationship between bone marrow adiposity and bone mass is often observed (Akune et al., 2004). Membrane-associated kit ligand-deficient KitSl/Sl-d mice having defective c-kit signaling are osteopenic. These mice have reduced bone formation and impaired osteoblast differentiation (Lotinun et al., 2005). We therefore evaluated whether reduced c-kit signaling results in a reciprocal increase in bone marrow adiposity in mice and rats.
MATERIALS AND METHODS
Animals were maintained under standard conditions with a 12-hr light, 12-hr dark cycle in accordance with the NIH Guide for the Care and Use of Laboratory Animals. Food (standard chow) and water were provided ad libitum. The experimental protocols were approved by the Institutional Animal Care and Use Committee.
Studies in Mice
These studies were conducted to evaluate the effect of kit receptor deficiency on bone marrow adiposity. We have previously shown adipocytes to be present in bone marrow of 12-week-old mice (Menagh et al., 2010). Therefore, 4-week-old male and female WBB6F1/J-KitW/KitW-ν (KitW/W-ν) and wild-type (WT) WBB6F1/J littermates (N = 8 mice per group) were purchased from the Jackson Laboratory (Bar Harbor, ME) and maintained until 12 weeks of age. For tissue collection, the mice were anesthetized with ketamine (100 mg/kg) and xylazine (2.5 mg/kg) or with 2–3% isoflurane delivered in oxygen. Body composition was determined via dual-energy X-ray absorptiometry (DXA), and then death was induced by cardiac excision. Tibiae and lumbar vertebrae were removed and stored in 70% ethanol for bone histomorphometry. In a subset of female mice (N = 3 per group), bone marrow was removed from right femora for adipocyte culture.
Kit ligand-deficient mice.
These studies were conducted to evaluate the effect of membrane-associated kit ligand deficiency on bone marrow adiposity. The skeletal phenotype of these mice has been described in detail (Lotinun et al., 2005). Four-week-old male WBB6F1/J-KitSl/KitSl-d (KitSl/Sl-d) mice and WT littermates (N = 8 mice per group) were purchased from the Jackson Laboratory (Bar Harbor, ME) and maintained until 12 weeks of age. In a follow-up study, 5-week-old male mice (N = 7–9 mice per group) were divided into three treatment groups: (1) WT, (2) KitSl/Sl-d + carrier (buffered saline), and (3) KitSl/Sl-d + secreted kit ligand (mouse recombinant stem cell factor, Sigma, St. Louis, MO). Kit ligand (100 μg/kg/day, week 1; 150 μg/kg/day, week 2; and 200 μg/kg/day, week 3) was delivered in saline by daily subcutaneous administration for 3 weeks. For tissue collection, all mice were anesthetized with ketamine (50 mg/kg) and xylazine HCl (2.5 mg/kg), and death was induced by cardiac excision. Tibiae were removed and stored in 70% ethanol for bone histomorphometry.
Study in Rats
This study was conducted to assess the effects of kit receptor blockade with gleevec on bone marrow adiposity in rats. Thirteen-month-old male Sprague Dawley rats were obtained from Harlan Sprague Dawley (Madison, WI). We used rats in this study because they have much higher levels of bone marrow adiposity than mice (Menagh et al., 2010). Rats were randomized by weight into two treatment groups (N = 6 rats per group): (1) vehicle or (2) gleevec. Gleevec (50 mg/kg/day; Novartis, Basel, Switzerland) or vehicle was administered intraperitoneally for 7 days. For tissue collection, rats were anesthetized with ketamine (50 mg/kg) and xylazine HCl (5 mg/kg), and death was induced by exsanguination followed by cardiac excision. Tibiae were removed and stored in 70% ethanol for bone histomorphometry.
Total body, femoral, and vertebral bone mineral content (BMC, g) in selected studies was determined prior to necropsy using a PIXImus small animal densitometer (Lunar Corp., Madison, WI).
The histological methods used here have been described in detail (Iwaniec et al., 2008). In brief, the proximal tibial metaphysis and lumbar vertebrae 3–4 were dehydrated in graded increases of ethanol and xylene and embedded undecalcified in methyl methacrylate. Sections (4-μm thick) were cut with vertical bed microtomes (Leica/Jung 2065 and 2165) and affixed to slides precoated with a 1% gelatin solution. Sections were stained according to the von Kossa method with a tetrachrome counterstain, alcian blue, or toluidine blue and used for determining cellular endpoints. Data were collected under visible light using the OsteoMeasure System (OsteoMetrics, Atlanta, GA). For histomorphometric data collection, the sample area within a vertebral body began 0.3 mm away from the cranial and caudal growth plates and included secondary spongiosa only. The tibial sample site was located 0.5 mm distal to the growth plate and also included secondary spongiosa only. Tissue area, adipocyte number and area, and mast cell number were measured. Mast cells were measured because they have been implicated in regulation of adipocyte and osteoblast differentiation (Liu et al., 2009) and KitW/W-ν and KitSl/Sl-d mice are mast cell deficient. The data are expressed as adipocyte density (number/mm2) and size (μm2) and mast cell density (number/mm2). Adipocytes were identified morphologically as large circular- or oval-shaped cells bordered by a prominent cell membrane and lacking cytoplasmic staining because of alcohol extraction of intracellular lipids during processing. This method was previously validated by fat extraction and analysis (Menagh et al., 2010). Mast cells were identified based on the presence of intensely metachromatic cytoplasmic granules. Mast cell identification was confirmed using electron microscopy (Turner et al., 2010).
Adipocyte cell culture.
Primary bone marrow stromal cells were cultured as previously described (Menagh et al., 2010). Briefly, the cells were cultured in α-MEM supplemented with 10% FBS and antibiotics at an initial density of 1 × 106 cells per milliliter in 10-cm2 dishes. After 10 days in culture, the cells were split and seeded at 5 × 106 cells per milliliter in six-well plates prior to induction of adipocyte differentiation. A differentiation-inducing media consisting of the base α-MEM and 10% FBS media supplemented with 0.5 mM isobutylmethylxanthine, 1 μM dexamethasone, and 0.5 μM insulin was added for 7 days to allow mature adipocytes to develop. Adipocytes were identified after staining with oil red-O. The number of adipocytes per field was determined by counting adipocytes at 10× in five randomly selected fields per well in six wells per treatment.
Differences among treatment groups were determined using a t-test for comparisons between two treatment groups or one-way ANOVA (SPSS 17.0, SPSS, Chicago, IL) for multiple comparisons. When significant treatment effects were detected with one-way ANOVA, a Bonferroni post hoc test was used to evaluate pairwise comparisons. If ANOVA assumptions of homogeneity of variance were not met, a Kruskal–Wallis test followed by a Tamhane post hoc test was applied. Differences were considered significant at P < 0.05. All data are reported as mean ± SE.
Studies in Mice
The effect of kit receptor deficiency on body composition in 12-week-old female mice is shown in Fig. 1. Significant differences in body weight or percent body fat were not detected between WT and kit receptor-deficient KitW/W-ν mice. However, total body, femur, and lumbar vertebra BMC was significantly lower in the KitW/W-ν mice compared to WT mice.
The effects of kit receptor deficiency, gender, and anatomical location on bone marrow adiposity in 12-week-old mice are shown in Fig. 2. Representative photomicrographs of bone marrow from WT and KitW/W-ν mice are depicted in Fig. 2A,B. The effects of kit receptor deficiency and gender on bone marrow adiposity in the proximal tibia are shown in Fig. 2C,D. Bone marrow adipocyte density was greater in female than male WT mice, but adipocyte size was greater in the male than female WT mice. Adipocytes were not detected in bone marrow of either female or male kit receptor-deficient KitW/W-ν mice. The effects of kit receptor deficiency and anatomical location on bone marrow adiposity in a separate cohort of female mice are shown in Fig. 2E,F. Adipocyte density was greater and adipocyte size tended to be greater (P < 0.07) in the proximal tibia than in the lumbar vertebra, respectively. Adipocytes were not detected in either tibia or vertebra in the KitW/W-ν mice.
The effect of disruption of c-kit signaling on in vitro bone marrow adipocyte differentiation is shown in Fig. 3. Oil red-O-positive cells were common in cultures derived from bone marrow of 12-week-old WT female mice, but were much less numerous in cultures from kit receptor-deficient KitW/W-ν mice.
The effects of membrane-associated kit ligand on bone marrow adiposity in mice are shown in Fig. 4. Adipocytes were present in the proximal tibia in WT mice. Adipocytes were not detected in the proximal tibia of membrane-associated kit ligand-deficient KitSl/Sl-d mice or KitSl/Sl-d mice treated with secreted kit ligand.
We evaluated skin as a positive control for the actions of administered secreted kit ligand. Representative micrographs of skin from WT and mutant mice are shown in Fig. 5. Subdermal adipocytes were detected in all genotypes. As expected, mast cells were absent from the subdermal fat depot of c-kit-signaling-deficient mice. However, mast cells were common in the subdermal fat depot of WT mice and KitSl/Sl-d mice treated with secreted kit ligand.
Studies in Rats
The effect of the receptor tyrosine kinase antagonist gleevec on bone marrow adiposity and mast cell density in proximal tibia of 13-month-old male rats is shown in Fig. 6. Adipocytes and mast cells were common in bone marrow of control rats, and mast cells were often localized immediately adjacent to adipocytes. Bone marrow adiposity was 47% lower in gleevec-treated rats compared to control rats. The difference in adiposity was due to fewer adipocytes in the gleevec-treated animals; significant differences in adipocyte size were not detected with treatment. Mast cell number was 74% lower in gleevec-treated rats compared to control rats.
The present studies investigated the effects of disrupted c-kit signaling on bone marrow adiposity. Kit receptor-deficient KitW/W-ν mice had no bone marrow adipocytes. Additionally, they had greatly reduced numbers of adipocyte precursors. Bone marrow adipocytes were likewise not detected in either membrane-associated kit ligand-deficient KitSl/Sl-d mice or in KitSl/Sl-d mice treated with secreted kit ligand. Furthermore, short-duration treatment of rats with the kit receptor antagonist gleevec resulted in lower bone marrow adipocyte density.
Expression of kit receptor is controlled by tissue-specific enhancers and silencers, and, as a consequence, kit receptor is limited in its cellular distribution (Berrozpe et al., 1999). Cells expressing kit receptor within bone marrow of WT mice include hematopoietic and mesenchymal stem cells, osteoclasts, mast cells, and megakaryocytes (Gattei et al., 1996; Thalmeier et al., 1996; Huss and Moosmann, 2002; Grimbaldeston et al., 2005; Horny et al., 2007; Conrad et al., 2008). It is well recognized that hematopoietic lineage cells, including osteoclasts and mast cells, interact with mesenchymal lineage cells through c-kit signaling (Ikuta et al., 1991; Otsuka et al., 1994; Gattei et al., 1996; Marsicano et al., 1997; Koma et al., 2005; Shiozawa et al., 2008). As a consequence, c-kit signaling may play a role in mesenchymal cell differentiation. In this regard, recent studies suggest that c-kit mediates parathyroid hormone-induced osteoblast recruitment and differentiation (Turner et al., 2010).
KitW/W-ν mice possess a point mutation in the tyrosine kinase domain that inactivates the receptor in one allele and a mutation in the second allele that results in lack of receptor at the cell surface (Copeland et al., 1990; Zsebo et al., 1990). As a consequence, activation of c-kit signaling by kit ligand is greatly reduced in kit receptor-expressing cells in bone marrow of mutant mice. The ligand for the kit receptor is encoded by the Steel (Sl) locus and is referred to variously as kit ligand, steel factor, stem cell factor, or mast cell growth factor (Zsebo et al., 1990). Several variants of kit ligand are expressed in a cell-specific manner by mesenchymal stem cells, adipocyte precursors, fibroblasts, and osteoblasts (Gattei et al., 1996; Huss and Moosmann, 2002; Koma et al., 2005). Kit ligand is synthesized as a relatively stable transmembrane protein, but alternative splicing can result in a protein that undergoes rapid proteolysis to produce a secreted form (Ashman, 1999). Membrane-associated kit ligand-deficient KitSl/Sl-d mice have a complete loss of function mutation (Sl) on one allele caused by 4-kb deletion in genomic DNA and a partial loss of function mutation (Sl-d) on the second allele. The transmembrane and cytoplasmic domains of membrane-associated kit ligand are absent, and a protein similar to the secreted kit ligand is produced, which binds to and activates kit receptor (Zsebo et al., 1990; Ashman, 1999).
Our failure to detect bone marrow adipocytes in KitSl/Sl-d mice suggests that membrane-associated kit ligand is essential for adipocyte differentiation in bone marrow. This finding implies that autocrine or cell to cell interaction between cells expressing kit receptor and cells expressing membrane-associated kit ligand is required for c-kit signaling to promote bone marrow adipogenesis. This conclusion is supported by our observation that administration of secreted kit ligand, while inducing mast cell differentiation in skin, does not result in appearance of bone marrow adipocytes.
A reciprocal relationship between bone mass and bone marrow adiposity has often been noted in rodents and humans (Akune et al., 2004; Pei and Tontonoz, 2004; Morita et al., 2006). A deficiency in PPARγ, a key mediator of adipocyte differentiation, reduces marrow fat and enhances osteogenesis in mice (Akune et al., 2004). This finding has been interpreted as evidence for a cause and effect relationship where increased adipocyte differentiation precludes the differentiation of osteoblasts. However, other studies suggest that changes in osteoblast differentiation need not be directly coupled to changes in adipocyte differentiation (Menagh et al., 2010).
The present findings, in which bone mass was reduced in kit receptor-deficient KitW/W-ν mice in spite of complete absence of marrow adipocytes, support this alternative view. Membrane-associated kit ligand-deficient KitSl/Sl-d mice, also shown in the present study to be deficient in bone marrow adipocytes, are also osteopenic (Lotinun et al., 2005). In contrast to the greatly reduced adipocyte precursor pool in bone marrow of c-kit- receptor-deficient mice, the osteoblast precursor cell population was relatively normal. Instead, a defect in differentiation of the osteoblast precursor population to osteoblasts was likely responsible for the osteoblast deficit (Lotinun et al., 2005). Thus, the deficiencies in bone marrow adipocytes and osteoblasts in c-Kit-signaling mice do not appear tightly coupled to one another. However, the precise mechanisms of action of c-kit signaling in adipocyte and osteoblast differentiation will require further elucidation. Binding of kit ligand to kit receptor leads to activation of multiple pathways, including phosphatidyl-inositol-3 (PI3)-kinase, phospholipase C-gamma, Src kinase, Janus kinase/signal transducers and activators of transcription (STAT), and mitogen-activated protein (MAP) kinase pathways (Reber et al., 2006). A critical role for the PI3K pathway has been demonstrated in adipogenesis of human mesenchymal stem cells (Yu et al., 2008). There is also evidence for MAP kinases acting as positive and negative mediators of adipocyte differentiation (Aubert et al., 1999; Aouadi et al., 2006). Thus, activation of a key kit-receptor signaling pathway may be required for adipocyte differentiation from mesenchymal stem cells in bone marrow.
Short-duration (1 week) treatment of rats with the kit receptor antagonist gleevec reduced bone marrow adiposity. In addition to kit receptor, gleevec has the potential to antagonize PDGF receptor-α, colony-stimulating factor-1 receptor, and the tyrosine kinase domain of the Abelson protooncogene (abl) (Dewar et al., 2005; Lotinun et al., 2005). Nevertheless, the marked similarity of gleevec-treated rats to c-kit-signaling-deficient mice implicate kit receptor blockade as the cause for the observed reduction in adipocytes and mast cells. The rapid reduction in adipocyte density after treatment with gleevec further suggests that maintenance of mature adipocytes in bone marrow requires c-kit signaling.
Mast cells may be relevant to these studies because they have been implicated in regulation of adipogenesis (Liu et al., 2009). Pronounced species differences in the distribution of mast cells among tissues have been reported. Although mast cells are common in bone marrow of rats and humans, they were not detected in bone marrow of mice (Lowry et al., 2008), an observation confirmed in the present studies. Therefore, the absence of bone marrow adipocytes in c-kit-signaling-deficient mice is unlikely directly due to concurrent mast cell deficiency. This conclusion is further supported by our observation that administration of secreted kit ligand induced dermal mast cells in membrane-associated kit ligand-deficient mice but failed to induce bone marrow adipocytes.
In conclusion, the present studies demonstrate that c-kit signaling through membrane-associated kit ligand is required for adipocyte differentiation in bone marrow of sexually mature mice and suggest that inhibition of kit signaling is the likely mechanism for the rapid reduction in the number of mature adipocytes observed in bone marrow of gleevec-treated rats. Taken together, the results implicate that c-kit signaling plays an important but previously unsuspected role in regulating bone marrow adiposity.
The authors thank Ms. Dawn Olson for performing DXA analysis.