Adipocytes are widely distributed throughout the human body, and their physiological functions differ depending on location. Bone marrow adipose tissue (BMAT) is a unique adipocyte depot located within the medullary cavity of bones. While the presence of BMAT has been known since the late 19^th century, the physiological importance of BMAT remains incompletely understood. Humans are born with very little BMAT, but it gradually expands with age, and by adulthood, BMAT constitutes about 50-70% of total marrow volume (1). Two types of BMAT have previously been described (2). Constitutive BMAT (cBMAT) is located within the distal region of long bones and in caudal vertebrae, and as suggested by the name, remains constitutively present despite a wide variety of physiological interventions. Alternatively, regulated BMAT (rBMAT) is in proximal tibia and distal femur. rBMAT is typically seen as single cells or in small clusters of adipocytes interspersed with hematopoietic cellularity. rBMAT volume changes under a variety of physiological and pathological conditions. For example, cold exposure, acute myeloid leukemia, exercise, and lactation lead to decreased rBMAT volume (3). Alternatively, expansion of rBMAT is observed with type 1 and type 2 diabetes, obesity, growth hormone deficiencies, impaired hematopoiesis, osteoporosis, as well as caloric restriction (CR) (3). BMAT also functions to influence bone and hematopoiesis by serving as an endocrine organ, contributing to circulating concentrations of adiponectin, stem-cell factor, leptin, and receptor activator for NF-κB (RANK) ligand (4–6).

Bone is a dynamic organ that responds to local and systemic stimuli to regulate resorption and formation of bone. Most human and mouse data suggest an inverse relationship between bone marrow adiposity and bone mineral density (7); however, this is not always the case. In C57Bl/6J mice, long-term high fat diet leads to increased BMAT with variable loss to trabecular and cortical bone mass (8, 9). Following bariatric surgery in mice, there is a dramatic loss of BMAT, in addition to a loss trabecular and cortical bone mass (10). Recent work demonstrates that BMAT has a negative effect on bone mass, although, under conditions of energy deficit, BMAT lipolysis helps to maintain bone mass (11). BMAds arise from skeletal stem cells, which also give rise to osteoblasts. Fate determination is controlled by several key transcription factors including runt-related transcription factor 2 (Runx2) and osterix (Osx), which promote osteoblastogenesis (12, 13) and proliferator-activated receptor gamma (PPARγ) and CCAAT/enhancer-binding protein alpha (C/EBPα), which promote adipogenesis (14). The bone marrow microenvironment is complex. In addition to impacting bone health, BMAds have also been shown to influence hematopoiesis. Several studies have suggested that BMAds are a negative regulator of hematopoiesis (15, 16). However, other studies suggest that BMAds support the function of hematopoietic cells (11, 17, 18). Thus, interactions between BMAT and other cell types within the marrow are complex and remain poorly understood.

CR has been shown to increase lifespan and improve overall metabolic health (19). Unlike white adipose tissue (WAT), BMAT paradoxically increases during CR (20), suggesting that BMAT and WAT are metabolically and functionally distinct. The physiological purpose for increased BMAT with CR is unknown; however, a recent study from Li et al. demonstrates that lipolysis of BMAds helps to fuel the bone and myelopoiesis during times of CR (11). M olecular signals that lead to these changes in BMAT volume remain poorly understood. Previous hypotheses have included potential roles for leptin, estradiol, fibroblast growth factor 21, ghrelin, and cortisol (21–25). Previously, Cawthorn et al. (26) showed that CR in female mice leads to BMAT increase, without changes in circulating leptin concentrations. Using mice and rabbits, these studies also showed that BMAT expansion was observed only when circulating corticosterone concentrations were elevated, implying a potential role for glucocorticoids in BMAT expansion following CR.

Glucocorticoids (GC) are corticosteroids that are essential for vertebrate biology. Due to their anti-inflammatory effects, they are widely used to treat a variety of inflammatory diseases. Glucocorticoids are mainly synthesized by the cortex of the adrenal gland. Production of GCs is regulated by the hypothalamic-pituitary-adrenal axis. Mechanisms of action by GCs are also tightly regulated by enzymatic conversion between active and inactive forms. Two enzymes regulate the conversion between active and inactive GCs. 11β-hydroxysteroid dehydrogenase 1 (11β-HSD1) catalyzes the conversion of inactive cortisone (11-dehydrocorticosterone in mice) to active cortisol (corticosterone in mice), while 11β-HSD2 performs the opposite conversion (27, 28). GCs bind to the glucocorticoid receptor (GR, gene name Nr3c1), a member of the nuclear receptor family. This receptor functions to regulate GC-responsive genes. Furthermore, activation of GRs is regulated by its subcellular distribution. When unbound, GR resides in the cytosol as a monomer, stabilized by several heat shock proteins. Once bound by a GC, a conformation change occurs, leading to exposure of two nuclear localization signals, at which point GR is transported to the nucleus. Once nuclear, GR directly binds DNA through GC-response elements. The effects of GC on adipose tissue are complex. Under most circumstances, GC stimulates fatty acid uptake and lipolysis in adipocytes (29, 30). Chronic GC treatment leads to metabolic impacts such as insulin resistance, dyslipidemia, and obesity (31). GC are present in almost all adipocyte differentiation protocols (32). In particular, the use of dexamethasone, a synthetic GR ligand, is common. Several groups have used adipocyte GR-knockout models with mixed results (31). Most studies demonstrate that GR is not required for development or maintenance of adipose depots (33–35). However, some studies suggest it may play an important role during high-fat diet feeding (36). Importantly, none of these studies have investigated the impact of adipocyte GR on bone biology.

To investigate the role of GC in BMAT expansion, we utilized a previously described BMAd-specific Cre mouse model to knock out GRs in BMAds (11). Female BMAd-Nr3c1^-/- mice had a small but significant increase in trabecular bone volume fraction (Tb. BV/TV), trabecular thickness, and distal tibial bone volume (BV). Male BMAd-Nr3c1^-/- mice did not show changes in bone parameters. Loss of GR in BMAds did not alter the response of young or adult male mice to CR. However, BMAd-Nr3c1^-/- mice had decreased circulating white blood cells in the light cycle without changes to hematopoietic progenitor populations, suggesting that GRs in BMAds may play a regulatory role in hematopoiesis.

BMAds are a unique group of adipocytes residing within the bone marrow microenvironment. rBMAT volume expands under a variety of physiological and pathological conditions including CR (20) and anorexia nervosa (4, 45, 46). The molecular mechanisms during CR leading to increased BMAT are unknown. Previous studies using mouse models have shown that BMAT expansion occurs when circulating GC are elevated (26). Both excess synthetic and endogenous GCs are known to cause bone loss (47). In general, an inverse relationship exists between BMAT volume and bone mass, including in patients with anorexia nervosa (48–50). Together these results suggest that excess GC during CR may lead to increased BMAT and subsequent bone loss (51, 52). Indeed, work from Pierce et al. have shown that GR-deficiency in Osx-expressing cells led to loss of cortical and trabecular bone mass in mice fed ad libitum or CR (53). However, previous studies using adipocyte-specific deletion of GR have not investigated changes to bone mass (31, 33–35). Indeed, in these studies, the direct effects of GR in BMAd would not be distinguishable from potential indirect effects caused by the deletion of GR in other adipose depots.

In this study, our results suggest that GRs in BMAds are not required for the expansion of BMAT with CR. One explanation is that GR plays an important role in the early differentiation of BMAds, and the BMAd-Cre model used in these studies targets only mature BMAds (11). Another possibility is that the loss of GR in BMAds is compensated by other GC-responsive proteins such as the mineralocorticoid receptor (MR), which can also directly bind to GC. However, evidence for robust compensation by MR has not been observed in adipocyte-specific GR-KO models (33, 54). In the current studies, we also did not observe significant changes to MR expression levels in BMAd-Nr3c1^-/- mice (data not shown). Several groups have targeted GC activity through knock-out of the enzymes involved in their activation, such as 11β-HSD1, a major regulator of the tissue-specific effects of GC (55, 56). Previous work has shown that global deletion of 11β-HSD1 produces a favorable metabolic state (57–59); however, the impact of 11β-HSD1-deficiency in BMAds has not been evaluated. With the recent invention of genetic tools like the BMAd-Cre mouse (11), future experiments can investigate this hypothesis that 11β-HSD1 is important for BMAT responses to GC. It is possible that GR in BMAds is important with other stressors that lead to BMAT expansion such as aging or obesity (60). Alternatively, it may be that elevated GC and BMAd-GR do not contribute to BMAT expansion following CR.

Whereas our results did not suggest a critical role for GR in BMAd in expansion of BMAT during CR, we did observe a small but significant increase in trabecular bone volume fraction and thickness as well as distal cortical bone volume in the tibiae of adult female BMAd-Nr3c1^-/- mice at baseline. Sex-specific differences in bone parameters are well documented (61, 62), and 6 weeks of 30% CR did not lead to loss of tibial bone mass in adult male mice. Indeed, the clinical data surrounding bone loss from CR is variable. One study showed that six months of CR in young adults does not lead to significant bone loss (63). However, similar studies in adults show a significant reduction in bone following CR (64, 65). Amongst these conflicting findings, one consensus is that patients with anorexia nervosa are at an increased fracture risk due to bone loss (66–69). Of note, anorexia nervosa is a condition of more severe caloric deficiency than that achieved in most CR studies. In mouse and rabbit models, results have also been mixed. One common observation is that CR results in stunted bone growth and low bone mass in some regions of the skeleton of young mice (21) and rabbits (26). Indeed, our studies support this finding, which is why we then chose to investigate CR in adult mice (34+ weeks). In adult animals, most studies suggest that CR leads to bone loss (21, 70). However, results have been variable both in terms of the severity of bone loss and the location of bone loss within the skeleton (21, 70). One study showed that bone loss occurs with 6 weeks of CR, even when performed in combination with exercise (71), which typically positively supports bone health (72). Interestingly, some studies suggest that long-term CR may protect against or delay age-related bone loss (73, 74). Several factors could influence the impact of CR on bone mass including sex, age, length of CR, strain of mice, and region of the skeleton being investigated. While our CR diet includes micronutrient supplementation, in other dietary restriction studies (21), the availability of calcium and other essential minerals is of concern.

Due to its location within the marrow niche, as well as its endocrine functions, many studies have aimed to investigate the roles of BMAd on hematopoiesis. Several studies suggest an inverse relationship between BMAT and hematopoiesis (10, 15, 16, 18). However, others suggest a supportive role for BMAds in hematopoiesis (11, 18, 75). In our study, loss of GR from BMAds led to a significant decrease in circulating white blood cells. Female BMAd-Nr3c1^-/- mice showed a decrease in lymphocytes, while male mice showed a significant reduction in monocytes. These changes were only significant when the blood draw was performed during the day (ZT5). Confirming previous studies showing that hematopoiesis is circadian (43), nearly all blood cell populations were reduced when the blood draw was performed at night (ZT17). We did not observe changes in circulating red blood cells in BMAd-Nr3c1^-/- mice. While GR has previously been shown to mediate stress erythropoiesis (41), we hypothesize that the lack of changes in circulating red blood cells in BMAd-Nr3c1^-/- mice is because this process mostly occurs in the spleen (76). In WAT, adipocyte GR suppresses the immune system, maintaining immune homeostasis (77, 78). While circulating white blood cells were decreased during the day in BMAd-Nr3c1^-/- mice, we did not observe differences in the abundance of bone marrow progenitor cell populations in the tibiae of BMAd-Nr3c1^-/- mice. We observed a subtle but significant change in ST-HSCs in the femur of BMAd-Nr3c1^-/- mice, but the frequency of CMP or CLP populations was not different. BMAds locally interact with hematopoietic cells and contribute to whole-body metabolism through the secretion of adipokines (79, 80). Lipodystrophic A-ZIP/F1 mice, which lack BMAT, have delayed hematopoietic regeneration in the long bones following irradiation (5), a process that involves the secretion of stem cell factor (SCF) from BMAds (5, 80, 81). Loss of SCF from BMAT reduces the bone marrow cellularity, hematopoietic stem and progenitor cells, common myeloid progenitors, megakaryocyte-erythrocyte progenitors, and granulocyte-monocyte progenitors (81). Supporting this hypothesis, there is a depletion of BMAT accompanied by a decrease in bone marrow erythroid cells and anemia following bariatric surgery (10). Our data suggests that loss of GR from BMAd lowers circulating white blood cells without altering bone marrow hematopoietic progenitor cell populations.

Our experiments demonstrate that loss of GR in BMAd of female mice led to a small but significant increase in trabecular bone volume fraction and trabecular thickness. Loss of GR in BMAd of male mice showed no changes to bone parameters. CR in young mice resulted in a decrease in tibial and femur lengths as sufficient caloric intake is important for growth in developing bones. CR studies in both young and adult male mice demonstrated that GR in BMAd is not required for BMAT expansion following CR. This study has several limitations. One caveat to the BMAd- Cre mouse model is that the insertion of a flipped Cre gene in the 3’ untranslated region of adiponectin leads to a small, but significant, decrease in circulating adiponectin concentrations (11). To confirm that excess GC do not work directly on BMAds to promote BMAT expansion during CR, future mouse models should target other aspects of the GC pathway, such as deletion of MR or 11β-HSD1. It is also highly likely that GC induce BMAT expansion by promoting the differentiation of new adipocytes, and the BMAd-Cre model used herein targets mature adipocytes. In conclusion, our data suggest that BMAd-GR is not required for BMAT expansion following CR.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was approved by Institutional Animal Care & Use Committee University of Michigan. The study was conducted in accordance with the local legislation and institutional requirements.

RS: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Writing – original draft, Writing – review & editing. JV: Data curation, Writing – review & editing. MA: Data curation, Writing – review & editing. ZL: Data curation, Writing – review & editing. KL: Data curation, Writing – review & editing. AM-H: Data curation, Writing – review & editing. KH: Data curation, Writing – review & editing. JM: Data curation, Writing – review & editing. RU: Data curation, Writing – review & editing. HH: Data curation, Writing – review & editing. IH: Data curation, Writing – review & editing. HM: Data curation, Writing – review & editing. OM: Writing – original draft, Writing – review & editing.