Insulin Signaling in Bone Marrow Adipocytes

Bone marrow adipose tissue (BMAT) comprises approximately 8% of total fat mass and thus representing a significant fat depot in adult humans with a role in bone homeostasis and whole body energy metabolism [1]. BMAT is more predominant in the appendicular than in axial skeleton [2]. The fatty acid composition of BMAT varies considerably based on its anatomical location, but tibia BMAT was found to have a fatty acid profile that resembles white and classical brown adipose tissues [3]. In the post-natal organism, BMAT originates from progenitor cells that are distinct from peripheral adipose tissues. BMAT is thought to be derived from bone marrow stromal (skeletal, mesenchymal) stem cells (BMSC) present within the bone marrow stroma and that are capable for differentiation, in addition to adipocytes, into osteoblasts and chondrocytes [1]. Currently, there is no consensus regarding the phenotype of BMAT progenitors but a number of markers have been proposed to identify adipocyte progenitors within the bone marrow, including osterix [4, 5], Prx1 and Nestin1 (reviewed in [6]), leptin receptor [7], Rankl [8], Znf423 [9••], and Hoxa11 [10].

BMAT plays a role in lipid storage, skeletal remodeling, and hematopoietic regulation but the mechanisms mediating and integrating these diverse functions have not yet been fully delineated [3]. In addition, BMAT is recognized as an endocrine organ producing local and systemic factors including adiponectin, dipeptidyl peptidase 4 (DPP4), legumain (LGMN), secreted frizzled-related protein 1 (sFRP-1), and delta-like 1 (DLK1) (also known as preadipocyte factor 1 (Pref1)) [9••, 11••, 12‐14]. BMAT undergoes pathologic changes during aging and in a number of diseases [1, 2, 4]; e.g., it expands in anorexia nervosa, states of estrogen deficiency, glucocorticoid excess, and growth hormone deficiency [2, 11••].

Another evidence about the role of BMAT to participate in regulating whole body energy metabolism is its ability to respond to insulin [15], to activate Sirt1, a key cellular energy sensor, and to induce a thermogenic gene program [16]. BMAT may contribute to systemic glucose and fatty acid clearance [3]. In addition, BMAT responds to insulin-sensitizing anti-diabetic medications such as thiazolidinedione (TZD) drugs, PPARγ agonists [4, 15, 17].

In the current review, we will discuss the role of insulin signaling in bone marrow adipocyte formation, metabolic functions, and its contribution to cellular senescence (Fig. 1). We will also summarize common factors involved in the regulation of BMSC differentiation fate into BMAT with a special reference to obesity and type 2 diabetes (T2D).

Insulin exerts anabolic effect on the bone metabolism and it has a critical role in the regulation of skeletal development and bone integrity [48]. Insulin signaling represents a key metabolic pathway important for the bioenergetic demand of bone cells [49‐53]. Our group has recently examined the effects of obesity on BMAT and its role in regulating skeletal energy demands. We have observed a unique metabolic phenotype of BMAT in obese mice and obese humans that exhibit enhanced insulin signaling, which was opposite to what we observed in peripheral adipose tissue [18••, 22••]. In more details, we identified a pro-adipogenic potential of BMSC in mice fed HFD and in obese humans driven by increased insulin signaling associated with enhanced oxidative phosphorylation, leading to BMAT expansion. We have also observed enrichment in a unique BMSC population with high expression of insulin receptor (IR+) in obese subjects [22••]. Interestingly, our study suggests that in murine and human obesity, bone marrow microenvironment does not exhibit insulin resistance phenotype. While this seems at variance with previously published data that employed animal models, most of the studies have employed insulin receptor (INSR)-deficient mice as a model for insulin resistance [49, 50, 51••]. This discrepancy may be explained by the biological differences between INSR-deficient mice, which is a suitable model for insulin deficiency taking place in long-standing T2D and obese subjects who exhibited insulin resistance phenotype but no manifest diabetes. Further, murine studies did not examine the presence of intrinsic changes in BMSC, as the observed insulin resistance may have been related to microenvironmental factors. On the other hand, the study by Wei et al. [51••] corroborates our findings as it demonstrates that mice with enhanced insulin signaling in bone are protected from the severe systemic insulin resistance phenotype. Our study demonstrates that maintenance of insulin responsiveness in BMSC of obese subjects is “a protective mechanism” allowing fat storage in bone marrow, when peripheral tissues manifest impaired insulin signaling (Table 1 summarizes the major findings of these studies and illustrating similarities and differences between studies). We think that tissue-specific responses to insulin are highly relevant for the bone field with respect to understanding the pathophysiology of obesity and T2D-associated bone disease.

Stem cells existing in states of commitment and differentiation have specific bioenergetic needs that determine their functions [54, 55]. Bone marrow consists of heterogeneous population of BMSC with different lineage commitments [56, 57]. Our group has recently examined whether metabolic programming of BMSC is upstream of lineage commitment [22] (Tencerova et al., 2019 Bone Research, accepted https://​doi.​org/​10.​1038/​s41413-019-0076-5).

Stem cells get energy supply from either glycolysis or oxidative phosphorylation (OxPhos) and they differ in the choice of energy substrate depending on whether they are in the growth or differentiation stage. Pluripotent embryonic stem cells prefer anabolic glycolysis, which is also the preferred metabolic process of rapidly proliferating cells [58, 59]. Hematopoietic progenitor cells exhibit differentiation dependent use of glycolysis or OxPhos [60, 61]. Osteoblast lineage cells employ both oxidative and glycolytic metabolic pathways in undifferentiated state, but during osteoblast differentiation, glycolysis is the preferred energy source [62]. On the other hand, a pre-adipocytic embryonic murine cell line 3T3-L1 employs OxPhos during adipocyte differentiation [62]. Thus, commitment to either osteoblasts or adipocytes is associated with a characteristic bioenergetic profile that is maintained during differentiation. We have recently observed that human BMSC and committed murine BMSC progenitors exhibit similar bioenergetic phenotype [22••] (Tencerova et al., Bone Research, accepted https://​doi.​org/​10.​1038/​s41413-019-0076-5). Employing high-throughput technologies including RNAseq and metabolomics, we found that committed murine adipocyte progenitors (named BMSC^adipo) and osteoblast progenitors (named BMSC^osteo) [57] exhibit a distinct metabolic program that is dependent on insulin signaling with higher oxidative phosphorylation in BMSC^adipo compared with preferable glycolysis in BMSC^osteo. Our findings demonstrate that the BMSC exhibit cellular responses to exogenous metabolic signals, which regulate their differentiation fate and the expansion of osteoblast versus adipocyte progenitor populations dependent on the prevalent metabolic environment. Future studies are needed to investigate the contribution of bioenergetic properties of BMSC to whole body energy homeostasis.

Metabolic pathways play a critical role in aging [63]. Insulin signaling belongs to nutrient sensing pathways that are important not only for energy metabolism but also for regulation of cellular senescence in different cell types [64].

Insulin signaling has been shown to regulate mitochondrial function [65, 66], which produces the most of cellular energy in form of ATP but this process is also accompanied with formation of intermediates such as reactive oxygen species (ROS) that play a role as a second messenger. Recent study has demonstrated that ROS increases insulin sensitivity via oxidative modification of the insulin receptor (autophosphorylation) or inactivation of protein tyrosine phosphatases, including PTEN and PTP1B [67]. Thus, this oxidative challenge suggests activation of cellular adaptation via modulation of insulin action and antioxidant system [68‐70]. In physiological condition, ROS is also required for adipocyte differentiation, which is controlled by insulin signaling [71]. On the other hand, chronic exposure to ROS in obesity and diabetes is associated with insulin resistance [72] as ROS is also a trigger for cell damage, inflammation, and senescence. Insulin signaling regulates mitochondrial function and biogenesis by inhibiting FOXO1, which tunes redox signaling by maintaining NAD+/NADH ratio for activation of SIRT1/PGC1α important for normal mitochondrial function [73, 74]. SIRT1 deacetylase activity is essential for prolongation of lifespan and delay of cellular senescence [75, 76]. Thus, the ratio between mitochondrial ROS production and NAD+/NADH redox complex needs to be maintained at the levels, to which the cellular metabolic capacity can adapt. This regulation can be tissue-specific. Insulin signaling in connection to mitochondrial function has been intensively investigated in liver, muscle, or adipose tissue [65]. However, it has not been studied in this context in BMAT and BMSC. We have recently reported that enhanced insulin signaling in BMSC of obese subjects leads to increased OXPHOS activity accompanied with higher ROS production [22••]. We suggested that the presence of this hypermetabolic status of BMSC leads to accelerated senescence phenotype and consequently impairment of stem cell functions and increased risk of bone fragility in obesity and diabetes (depicted in Fig. 1).

Obesity and T2D are increasingly recognized as risk factor for bone fractures [77‐81]. New therapeutic strategies are needed based on understanding the biological and molecular mechanisms of BMAT formation. Thus, unraveling the relationship between BMAT and bone metabolism at the molecular level is relevant.

Based on our understanding of insulin signaling and BMAT formation, it is possible that inhibiting insulin signaling within BMSC may serve as a protective mechanism against expansion of BMAT. This notion is supported by experience of using TZD in treatment of T2D. These drugs increase insulin sensitivity but led to increased BMAT formation and increased risk for fractures [77, 80, 81].

Other approach is metabolic slowing induced by caloric restriction or nutrient supplementation that can prevent accumulation of unused intermediates from metabolic processes and support mitochondrial functions in metabolically active tissues [82‐84]. On the other hand, calorie restriction in animals is associated with increased BMAT formation, but its effects on humans remain to be determined.

Given the interactions among medications and lifestyle modifications/interventions, the relative effect on BMAT metabolic phenotype and insulin signaling within bone microenvironment needs studying to identify specific approaches for prevention and treatment of metabolic bone diseases associated with obesity and T2D.