INFORMATIVE NOTE

An opinion on the regulation of bone marrow adipose tissue by dietary fatty acids

S. Lopez^a, A. Lemus-Conejo^b, M.A. Rosillo^b, F.J.G. Muriana^b and R. Abia^b,*

^aDepartment of Cell Biology, School of Biology, University of Seville, Seville, Spain.

^bLaboratory of Cellular and Molecular Nutrition, Instituto de la Grasa, CSIC, Seville, Spain.

*Corresponding author: abia@ig.csic.es

In previous studies, we have identified atherosclerosis, thrombosis, and inflammatory risk factors associated to the ingestion of dietary fats during the postprandial period in humans (Lopez et al., 2018, 2011, 2010, 2008; Ortega-Gomez et al., 2017, 2012; Montserrat-de la Paz et al., 2016; Naranjo et al., 2016; Bermudez et al., 2014, 2011; Varela et al., 2011; Lopez-Miranda et al., 2010; Pacheco et al., 2008, 2006; Abia et al., 2003, 2001, 1999). Our findings included some of the putative mechanisms by which exogenous fatty acids modulate cellular and molecular targets in the aetiology and pathogenesis of cardiovascular disease (Varela et al., 2015, 2014, 2013; Lopez et al., 2013, 2007; Bermudez et al., 2012, 2008; Bellido et al., 2004; Pacheco et al., 2003, 2002, 2001). Recently, our research team has also discovered that neutrophils and monocytes/macrophages undergo phenotype/functional dynamic switch in response to the microenvironment signals during the period that follows the ingestion of dietary fats (manuscripts in preparation). We found that circulating neutrophil number was transiently increased after the ingestion of saturated (milk cream) but not of monounsaturated (refined olive oil) dietary fats. In addition, while total number of circulating monocytes was not postprandially affected, those monocytes expressing CD14 and CD16 at high density were characteristic after the ingestion of saturated and monounsaturated dietary fats, respectively. These observations are the first evidence of a potential crosstalk between exogenous fatty acids and bone marrow (BM) function, and suggest that dietary fats, in a fatty acid-dependent manner, could trigger myeloid cell mobilization in BM.

Hematopoietic stem cells (HSCs) exist in the BM and are in charge to generate all the cells necessary to refill the blood and immune systems. It is hierarchical and firmly orchestrated by a proper environment of cells and factors for controlling the fate decision of HSCs to preserve a steady platelet, erythrocyte, and leukocyte supply (Eaves, 2015; Boulais and Frenette, 2015). HSCs are rare (1 in 10000 BM cells), relatively quiescent (<5% are in cell cycle), and constitute the apex of a differentiation cascade of hematopoietic progenitor cells with gradually reduced regeneration competence while augmented ability of conversion into multiple blood cell lineages. The existence of subpopulations with different phenotypes among the progenitor cells involved in the reconstitution of haematopoiesis and able to produce the entire range of haematopoietic progeny cells has already been documented for a long time (Morrison and Scadden, 2014). According to the latest model of haematopoiesis, the multipotency of HSCs is conserved by the presence of long-term HSCs (LT-HSCs) that differentiate to yield short-term HSCs (ST-HSCs) in BM and no other population of HSCs is devoted to sustaining a certain lineage, but the potential lineage appears as the HSC grow (Adler et al., 2014). Blood and immune system have the requirement of a proper supply of red cells and mature leukocytes with a finite lifespan throughout life. It is more evident after stress situations in which both host defence and repair mechanisms are mobilized. Furthermore, intravenously injected HSCs during BM transplantation can also promote homing and engraftment of HSCs to BM and regenerate the HSC pool to meet the demand of blood and immunity (Lapidot et al., 2005). The endosteal and perivascular regions in bone are anatomical locations for HSCs within the BM. Several stromal cells, including mesenchymal stem cells (MSCs), secrete messengers [e.g. granulocyte-macrophage colony-stimulating factor (GM-CSF), macrophage colony-stimulating factor (M-CSF), and stromal cell-derived factor 1 (SDF-1)/C-X-C motif chemokine ligand 12 (CXCL12)] that bind to cell-surface receptors on HSCs for modulating their survival and functional role (Morrison and Scadden, 2014; Mendelson and Frenette, 2014). In addition, HSCs are retained in restricted hematopoietic niches via adhesion molecules, including vascular cell adhesion molecule-1 (VCAM-1), cadherins, annexin II, and E-selectin, where they receive and integrate extrinsic regulatory cues through the activation of transcription factors of the ETS family and partners, the main of which are crucial to divert the production to myeloid cells (Pittet et al., 2014).

Once neutrophils and monocytes are produced in mature form, they can leave the bone marrow into circulation. The release is tightly controlled by numerous signals such as C-C motif chemokine receptor 2 (CCR2) ligands [e.g. monocyte chemoattractant protein-1 (MCP-1)] and CXCL12. Migration of HPCs is limited, as only a few hundred are normally found outside BM (Massberg et al., 2007). However, proteases [neutrophil elastase (NE), cathepsin G, and metalloproteinases (MMPs)] produced by BM myeloid cells have been noticed to be involved in the cleavage and inactivation of CXCL12 and KIT ligand throughout mobilization of myeloid cells from BM in stress situations (Shen et al., 2010; Klein et al., 2015). Serine protease inhibitor α1-antitrypsin (A1AT) is expressed in BM to prevent the accidental cleavage of niche components by NE and cathepsin G, and extracellular matrix (ECM) disarrangement (Kuiperij et al., 2009). Genes encoding the small Rho guanosine triphosphatases Rac1 and Rac2 that control cell migration further promote the rapid mobilization and entry of HSCs into circulation (Gu et al., 2003). We have preliminary data showing that monounsaturated, when compared to saturated dietary fats, promote: (i) a higher content of A1AT in lipoproteins of intestinal origin, neutrophils, and monocytes (manuscript in preparation); and (ii) the activation of the phosphoinositide 3-kinase (PI3k)-Rac1-Jun N-terminal kinase (JNK) and Rac1-MMP-2 pathways in human perivascular smooth muscle cells (Varela et al., 2014). Therefore, it is tempting to speculate that dietary fats, in a fatty acid-dependent manner, may be involved on migratory capacity and release of HSCs.

In BM, cells of the osteoblast and adipocyte lineages have a common precursor, which comes from MSCs (Tian and Yu, 2015). Tracking the fate of MSCs in aged BM has revealed that close proximity between osteoblasts and adipocytes underscores the reciprocal relation of bone mass and BM adipogenesis. As a result, bone loss caused by osteoporosis is accompanied by a progressive BM adiposity, which suggests that differentiation of MSCs into adipocytes predominates over osteoblasts. To support MSC differentiation into either osteoblasts or adipocytes, it is essential that MSCs be recruited to proper density and confluence. Involved in modulating the lineage commitment, a variety of external stimuli delicately balanced include the release of growth factors [e.g. transforming growth factor-β (TGF-β) family members, bone morphogenic proteins (BMPs), and Wnt proteins] and transcription factors [e.g. peroxisome proliferator-activator gamma 2 (PPARγ2) for adipogenesis and runt-related transcription factor 2 (RUNX2) for osteogenesis]. The canonical Wnt/β-catenin pathway plays a major role in regulating these two differentiation processes (Yuan et al., 2016). Soluble Wnt proteins such as Wnt10b, Wnt1, Wnt6, Wnt7a, and Wnt10a have pro-osteogenic activity and prevent phosphorylation, and thereby degradation of β-catenin. Unphosphorylated β-catenin may then translocate into the nucleus to form transcriptional complexes along with members of the T-cell factor (Tcf)/lymphoid enhancer-binding factor (Lef) nuclear protein family and RUNX2. These processes facilitate the osteogenic differentiation of MSCs, the mineralization by increasing alkaline phosphatase (ALP) activity in pre-osteoblasts, and the expression of the anti-bone resorption osteoprotegerin (OPG). During bone formation, pro-osteogenic Wnt proteins (Wnt10b, Wnt1, Wnt6, Wnt7a, and Wnt10a) coordinate each other and bind to low-density lipoprotein receptor-related protein (LRP) family, most likely LRP5/6. This crosstalk stabilizes β-catenin and allows its translocation, promoting osteoblastogenesis whereas blocking adipogenesis (Cawthorn et al., 2012). However, other Wnt proteins (Wnt4, Wnt5a, and Wnt5b) obstruct these communications and networks, enabling degradation of β-catenin and inducing adipogenesis (Cristancho and Lazar, 2011). Previous studies noticed that MSCs are by-default programmed to differentiate into adipocytes (Lecka-Czernik, 2006; Kirkland et al., 2002). By contrast, when the nuclear envelope intermediate filament lamin A/C is overexpressed, as it occurs in the bone in youth, MSCs enter the osteogenic lineage, preventing MSCs to differentiate into adipocytes (Pajerowski et al., 2007; Bermeo et al., 2015). Disruption of nearly all of these conditions takes place with aging and therefore the balance between MSC osteogenesis and adipogenesis will move towards adipogenic differentiation. In a setting of PPARγ activation, adipogenesis in BM-MSCs is positively regulated. Through mechanisms mediated by PPARγ signalling pathways, Wnt proteins are inhibited, β-catenin is degraded, and lamin A/C expression is suppressed. Interestingly, systemic adipokynes that regulate insulin sensitivity can also influence the fate of MSCs in BM. Insulin impedes the production of OPG in osteoblasts, thus, an increased ratio of OPG to receptor activator of nuclear factor-kB ligand (RANKL) supports the release of active osteocalcin from bone ECM giving rise the release of insulin from pancreas (Chen et al., 2012). We have previously demonstrated that β-cell function and insulin sensitivity are modulated by meals enriched with refined olive oil (source of oleic acid) when compared to meals enriched with butter/milk cream (source of SFAs) (Lopez et al., 2011; Lopez et al., 2008), in agreement with a local action of meal-derived oleic and palmitic acids on pancreatic β-cells (Bermudez et al., 2014). The adipocyte/osteoblast balance is also highly regulated at the level of gene transcription by the transfer of adipogenic microRNAs from adipocytes to osteoblasts, which enhances BM adiposity (Martin et al., 2015).

Exponential accumulation of BM fat begins at birth. This happens more quickly in distal than in proximal bones, so that at 25 years old, BM is approximately 70% fat (Fazeli et al., 2013; Cawthorn et al., 2014). Apart from age, certain conditions such as metabolic diseases (type 1 and 2 diabetes) (Kurra and Siris, 2011), oestrogen withdrawal (Taxel et al., 2008), immobilization (Lau and Guo, 2011), glucocorticoid treatment (Havashi et al., 2009), Cushing’s disease (Geer et al., 2012), and caloric restriction (Devlin et al., 2010) favour fat accumulation within BM instead of bone formation. These risk factors encourage the MSCs to change into its default lineage: adipocytes. BM adipose tissue (BMAT) formation is further induced by several treatment modalities such as radiation, chemotherapy, and thiazolidinediones (Suchacki et al., 2016). Until the last few years, adipocytes were perceived by many as metabolically inactive cells in the BM niche, even as cells only intended to fill trabecular bone holes. It is now clear that the adipocyte within bone is indispensable for the normal function of other neighbouring cells in the marrow microenvironment. BM adipocytes are scattered throughout the hematopoietic tissue, instead of grouped in lobules as in other fat depots (Hardouin et al., 2014). It is relevant that subcutaneous and visceral adipose tissues (white adipose tissue, WAT) have a fatty acid composition different to that found in BM adipose tissue (BMAT), which is richer in saturated fatty acids (SFAs) and poorer in monounsaturated fatty acids (MUFAs) than subcutaneous or visceral WAT (Griffith et al., 2009). Observations from randomized controlled trials and population-based observational studies have shown that a high intake of fatty fish strongly correlates with a reduction in the burden of fragility fracture (Longo and Ward, 2016), suggesting that omega-3 long chain polyunsaturated fatty acids (omega-3 PUFAs), as demonstrated for oleic acid (Gillet et al., 2015; Garcia-Martinez et al., 2014), have potential benefits on BM cells and microenvironment. Under high levels of adipogenesis, BM adipocyte products (free fatty acids and adipokynes) may induce osteoblast apoptosis, even the fat tissue can undergo dramatic and repetitive sequences in the remodelling to allow tissue expansion and removal of dead adipocytes, which is a process mediated by adipose tissue macrophages (ATMs). This scenario causing a vicious circle is known as lipotoxicity (Ng and Duque, 2010). Conversely, when co-cultured with adipocytes having inhibited the fatty acid endogenous biosynthesis, osteoblasts increase survival and improve mineralization (Elbaz et al., 2010). There is now plenty evidence that lipotoxicity by ectopic fat accumulation in BM is common in other organs, including pancreas, muscle, and liver. In pancreas, newly arrived adipocytes (deposition of fat inside and around the pancreas) affect the function and induce the death of β-cells (Long et al., 2014). Interestingly, this fat-induced process may be orchestrated by fatty acids. For example, exposure of human islet cells to palmitic acid [the main SFA in the diet] impairs glucose-stimulated insulin secretion, reduces insulin gene transcription, and induces β-cell apoptosis, whereas oleic acid [the main MUFA in the diet] is cytoprotective for β-cells and even attenuates the pro-apoptotic effects of palmitic acid (Lopez et al., 2010). Similarly, among the adipocyte-secreted factors in BM, palmitic acid has negative effect on osteoblasts in co-culture (Gunaratnam et al., 2014; Wang et al., 2013): (i) lowering mineralization, ALP activity, and expression of osteogenic mRNA markers; (ii) increasing intracellular reactive oxygen species (ROS); and (iii) disrupting the Wnt signalling and BMP2/RUNX2/SMADs pathways. Evidence from proteomic analysis of adipocytes suggests the occurrence of changes during ageing BM from pro-osteogenic, anti-adipogenic, and anti-apoptotic phenotype in young mice to a toxic and pro-adipogenic phenotype in old mice through paracrine mechanisms involving local secretion of adipokynes particularly involved in BM function (Gasparrini et al., 2009). Recent studies have reported that secretion of leptin and adiponectin (the adipokynes whose receptors are expressed by osteoblasts and osteoclasts) is greater from BMAT than from visceral WAT (Cawthorn et al., 2014) and that adiponectin from BM adipocytes has a relevant role in mobilizing HSCs/HPCs into blood to participate in tissue repair and remodelling (Yu et al., 2015).

The link between progressively sedentary lifestyle and cumulative consumption of high-dense foods, exemplified by the “Western style” high-fat diet, has contributed to the global pandemic of obesity and related disorders. This diet-induced adiposity is generally localized in subcutaneous or visceral WAT. There are important metabolic differences between adipocytes in different anatomical compartments; therefore, the location of fat pads at any of these regions is not their only distinctive characteristic. In fact, this adipose heterogeneity is also visualized by their different roles in disease susceptibility. BMAT seems to be a special class of fat depot with unique functions. Despite a similar appearance to subcutaneous and visceral adipocytes, BM adipocytes have divergent physiological, hormonal, and metabolic responses. In the C57BL/6J and C3H/HeJ mouse strains, by using the osmium staining technique for comparative morphology, BMAT may be found in two different groups: constitutive BMAT (cBMAT) and “regulated” BMAT (rBMAT). cBMAT is formed very early after birth in distal region of tibia and in caudal vertebrae, its adipocytes look like those in WAT and are poorly responsive to systemic disturbances. On the other hand, rBMAT is developed later as smaller adipocytes scattered with other BM cells in proximal skeletal sites, being responsive to a range of environmental, metabolic, and genetic cues (Scheller et al., 2014). Strikingly, the analysis of lipids, gene expression patterns, and functions in rBMAT is different from cBMAT (Scheller et al., 2015). It is possible that the origin of BMAT, still unveiled, is dissimilar to that of other fat depots. Thus, the origin of WAT and brown adipose tissue (BAT) is even a matter of on-going debate. After body weight loss induced by caloric restriction or by starvation in an eating disorder such as anorexia nervosa, there is an increase of BMAT. This fat tissue accretion in BM is to some extent unexpected because other fat depots (for example, visceral WAT) are being mobilised to meet energy demands when excessive catabolic activities. It is unclear why this happens. Several hypotheses have been proposed to explain this apparent paradox, including the occupation by BMAT of BM cavity space in response to the eventual loss of trabecular bone, the expansion of BMAT in response to starvation-induced stress as an adaptive strategy for supporting survival of BM cells, and the adiponectin secretion by BMAT to enhance insulin sensitivity and to stimulate appetite (Scheller et al., 2014). Notwithstanding the relevance of BMAT for metabolic and hematopoietic outcomes, it remains unexplored whether dietary fatty acids are instrumental in BM adiposity and in the phenotype of BM adipocytes, which could balance the likely adverse effects of increased adipogenesis on BM organization and function.

ACKNOWLEDGMENTSTOP

This work was supported by the grant AGL2016-80852-R from The Spanish Ministry of Science, Innovation and Universities. MAR acknowledges financial support from the Spanish Research Council (CSIC)/Juan de la Cierva [FJCI-2017-33132].

REFERENCESTOP