High-Density Lipoprotein (HDL) metabolism and bone mass

Abstract

It is well appreciated that high density lipoprotein (HDL) and bone physiology and pathology are tightly linked. Studies, primarily in mouse models have shown that dysfunctional and/or disturbed HDL can affect bone mass through many different ways. Specifically, reduced HDL levels have been associated with the development of an inflammatory microenvironment that affects the differentiation and function of osteoblasts. In addition, perturbation in HDL metabolic pathways favor adipoblastic and restrain osteoblastic differentiation through, among others, the modification of specific bone-related chemokines and signaling cascades. Increased bone marrow adiposity also deteriorates bone osteoblastic function and thus bone synthesis, leading to reduced bone mass. In this review, we present the current knowledge and the future directions with regards to the HDL-bone mass connection. Unraveling the molecular phenomena that underline this connection will promote the deeper understanding of the pathophysiology of bone-related pathologies, such as osteoporosis or bone metastasis, and pave the way towards the development of novel more effective therapies against these conditions.

Introduction

Recent advances in the field of lipoproteins highlight a multifunctional role of high density lipoprotein (HDL) in health and disease (Constantinou et al. 2015). HDL has been for decades an intriguing lipoprotein which attracted the attention of biomedical community, mainly because of its important role in atheroprotection (Constantinou et al. 2015). Indeed, the inverse correlation between HDL cholesterol (HDL-C) levels and the risk for developing coronary heart disease (Gofman et al. 1954, Havel et al. 1955, Miller et al. 1975, Tsompanidi et al. 2010, Kypreos et al. 2013, Karavia et al. 2014), suggested that high HDL-C levels in plasma are protective against the development of atherosclerosis. As a result, the majority of studies in the literature focused on the understanding of HDL-C levels in human pathology, a simplified approach to HDL that dates back to the early days when little was known about HDL structure and function. However, more recent data from experimental mice and clinical trials indicated that HDL particle functionality, as determined by its apolipoprotein (apo) and lipid content, is far more important in atheroprotection than HDL-C levels alone (Tsompanidi et al. 2010, Karavia et al. 2014, Constantinou et al. 2015). Even though HDL is usually referred to as the “good cholesterol”, certainly it is far more than just a “cholesterol”. HDL is rather a macromolecular assembly of proteins and lipids synthesized in the circulation as a result of a concerted action of apolipoproteins, lipid transporters and plasma enzymes. The main lipid cargo of mature HDL particles is esterified cholesterol. However, other lipids (phospholipids, sphingolipids, ceramides etc) are also part of HDL lipidome (Filou et al. 2016).

Though HDL was originally brought under the spotlight due its importance in protection from atherosclerosis, more recent findings propose a multifunctional role of HDL in numerous other biological processes including, inflammation, oxidative stress, nitric oxide production, and regulation of plasma glucose homeostasis (Karavia et al. 2014). In addition to atherosclerosis and coronary heart disease, recent data indicate that HDL may play pivotal role in the pathology and treatment of other diseases including morbid obesity, nonalcoholic fatty liver disease, type 2 diabetes mellitus, obstructive pulmonary disease, and numerous diseases of the central nervous system (Constantinou et al. 2015). Recently, we identified a novel role of HDL in the pathogenesis of degenerative and metabolic bone diseases using experimental mouse models (Triantaphyllidou et al. 2013, Blair et al. 2016), suggesting that low and dysfunctional HDL may contribute to an increased prevalence of these diseases by influencing molecular processes associated with bone synthesis and catabolism. Here, we review all new information pertinent to the effects of HDL and its major apolipoproteins on bone metabolism and function.

Principles of HDL biogenesis and metabolism

Recent data indicate that HDL is a mixture of lipoprotein particles with densities in the range of 1.063 to 1.21 g/ml and depending on their lipid composition these particles may assume a discoidal or spherical geometry. Mature spherical HDL particles contain approximately 45-55 % apoproteins, 26-32% phospholipids, 15-20% esterified cholesterol, 3-5% free cholesterol, and approximately 5% triglycerides (Tsompanidi et al. 2010). The main protein component of HDL is apolipoprotein A1 (APOA1) which plays a key role in the biogenesis and functions of HDL (Zannis et al. 2004). However, studies in mice showed that, other apolipoproteins such as apolipoprotein E (APOE) (Kypreos & Zannis 2007) and apolipoprotein CIII (APOCIII) (Kypreos 2008) are also capable of promoting the de novo biogenesis of HDL in a pathway similar to the one for the formation of APOA-I containing HDL. Notably, APOA1-containing HDL appear to be structurally and functionally distinct from APOE-containing HDL particles (Filou et al. 2016).

Studies in cell cultures, as well as studies in experimental mouse models, showed that biogenesis of classical APOA1-containing HDL involves the lipid transporters ATP-binding cassette A1 (ABCA1) and G1 (ABCG1) and the plasma enzyme Lecithin:Cholesterol Acyl Transferase (LCAT) (Soutar et al. 1975, Chroni et al. 2003; Fitzgerald et al. 2004). In the early steps of HDL formation, lipid-free APOA1 which is secreted at about 70% by the liver (Timmins et al. 2005) and 30% by the intestine (Brunham et al. 2006), interacts functionally with a dimeric form of the lipid transporter ABCA1 to acquire phospholipid and cholesterol (Nagata et al. 2013), thus forming a minimally lipidated APOA1. Through a series of intermediate steps that involve ABCG1, minimally lipidated APOA1 gradually forms discoidal HDL particles, which are then converted into spherical particles by plasma enzyme lecithin:cholesterol acyltransferase (LCAT) (Soutar et al. 1975). APOA1 on both discoidal and spherical HDL particles then is capable of interacting with scavenger receptor class B type I (SRBI) present on the surface of cells (Van Eck et al. 2005, Krieger 2001, Liu et al. 2002) to deliver cholesterol esters to the cell.

Studies in mice support that the interactions of APOA1 on HDL with cell surface SRBI are important for the atheroprotective functions of HDL (Trigatti et al. 2003). Additional steps in the metabolism of HDL involve its further processing by plasma enzyme cholesteryl-ester transfer protein an enzyme that mediates the exchange of HDL cholesteryl-esters with VLDL (very low density lipoprotein) triglycerides. This step contributes to the reduction of HDL particle size, making it a more suitable substrate for SRBI, and the eventual catabolism of HDL cholesteryl esters by the LDL receptor through receptor mediated uptake of cholesteryl ester-rich LDL. Additional processing of HDL in circulation involves hydrolysis of its phospholipids and triglycerides by various lipases (lipoprotein lipase, hepatic lipase, and endothelial lipase), and the transfer of phospholipids from VLDL/LDL to HDL by phospholipid transfer protein (Zannis et al. 2004).

Proteomic analyses revealed a diversity in HDL proteome that depending on the method of HDL isolation, includes more than 85 different proteins (Karlsson et al. 2005, Karlsson et al. 2005, Heller et al. 2005, Rezaee et al. 2006, Vaisar et al. 2007; Shah et al. 2013), in addition to APOA1, APOE and APOCIII. Moreover, these studies showed that the plasma abundance of HDL-associated proteins is insufficient to permit one copy of protein for each HDL particle, suggesting that different proteins may be associated with different HDL particles that are differentially distributed across the HDL density spectrum (Tsompanidi et al. 2010). More recent data indicate that HDL proteome dictates its lipidome and subsequently HDL particle functionality (Filou et al. 2016), suggesting that the understanding of HDL proteome and the factors affecting it, are crucial steps in successfully improving HDL functionality.

Biogenesis and function of bone cells

Osteoblast, the bone forming cell, and the “osteoblast family”

Bone is a form of specialized connective tissue that performs several essential functions including mechanical support for soft tissues, protection of internal organs, calcium homeostasis, and support of haematopoiesis. The maintenance of bone mass in mammals depends upon the fine balance between bone synthesis and resorption, two vital tasks accomplished by osteoblasts and osteoclasts, respectively. Continuous bidirectional cross-talk between osteoblasts-osteoclasts is coordinated in time and space through a process that is fundamental for bone homeostasis and is called “bone remodeling” (Figure 1). This process is under the strict control of autocrine, paracrine and endocrine signals and is essential for the adaptation of bone to the mechanical loading and for the repair after damage (Sims & Gooi 2008). Most importantly, disturbances in the coupling between bone synthesis and/or bone degradation, is the leading cause for numerous bone-related pathologies, including osteoporosis.

Figure 1.

Diagram depicting the growth factors, cytokines and receptors that are involved in the “coupling” between osteoblasts and osteoclasts and the regulation of osteoclast maturation and bone resorption. These phenomena that take place in the bone marrow, are described in detail in the text. It should be noted that mature osteoclasts should adhere tightly to bone surfaces (mainly through ανβ3 intern adhesions) in order to accomplish bone resorption. Moreover, note that the activated osteoblasts are large, cuboidal cells, with hefty nucleus; on the contrary, inactive bone lining cells are spindle shaped, with small elongated nucleus. Under specific stimuli, bone lining cells become metabolically active osteoblasts, regaining their bone-producing capacity. Abbreviations: BLC: bone lining cells, CSF1: colony stimulation factor-1, HSC: hematopoietic stem cells, IL: interleukin, MMP: matrix metalloproteases, OCL: osteoclasts, OBL: osteoblasts, OCT: osteocyte, OPG: osteoprotegerin, RANK: receptor activator for nuclear factor k B, RANKL: RANK-Ligand, SC: stromal cells.

In the present review we refer to bone cells in the context of the air-breathing vertebrates. In this context, bone is a complex tissue with advanced features not found in the skeleton of vertebrate phylogenetic precursors, such as fish. Bone usually replaces solid and avascular mesenchymal tissues namely mineralized cartilage or fibrocartilage. Indeed, during bone formation in vitro, cartilage genes are invariably induced at early stages, suggesting that cartilage differentiation is required for osteoblasts formation (Blair et al. in press). This process is more profound in conditions where endochondral ossification is involved, such as normal long bone growth or fracture healing.

Osteoblasts are metabolically active cells, and as such they possess large nucleus containing 2-4 nucleoli, very rich rough endoplasmatic reticulum and Golgi apparatus. The name “osteoblast” has its origins from the Greek words “osteo”-“óστó” that means bone and “blast”-“βλασταίνω” that means “been born”. These cells are responsible for the production of the extracellular matrix that is composed mainly of collagen type 1, with smaller amounts of specialized proteins including osteocalcin (Blair et al. 2007). Osteoblasts are under the tight control of paracrine, autocrine and endocrine signals. In fact, virtually all common signal transduction pathways converge on osteoblasts regulating their homeostasis (de Gorter DJJ & ten Dijke P 2013).

New osteoblasts differentiate from mesenchymal stem cells in a skillfully coordinated fashion at the bone surfaces. Key transcription factors for osteoblast differentiation are osterix and RUNX2; the former is often used to regulate cre recombinase for osteoblast-specific gene expression (Blair et al. 2008). The molecular pathways that are responsible for osteoblast differentiation, proliferation and ultimately apoptosis are complex (de Gorter DJJ & ten Dijke P 2013). Among the most thoroughly studied and potent regulators of osteoblast differentiation are the mitogen activated kinases (MAPK), the Wnt/β-catenin, the bone morphogenic proteins (BMP) and transforming growth factor-β (TGFβ), the Hedgehog, the Notch, the insulin-like growth factor 1(IGF1), the fibroblast growth factor (FGF) and the Calcium signal transduction pathways that ultimately converge on RUNX2, which is the master regulator of osteoblast differentiation (Blair et al. in press). The osteoblasts are connected through tight junctions and gap junctions consisted mainly of connexin-43. The tight junctions exclude low molecular matrix materials, while bone is alive. Notably, however, this is not the case for dead bone, which is immediately permeable (Eberhardt et al. 2001). The most common bone formation defects, usually include mutations in type I collagen (osteogenesis imperfecta) (Forlino & Marini 2016). There are many causes of low- or high- bone mass that are secondary to defects in specific factors, including serum lipids and lipoprotein receptors, as discussed elsewhere in this review.

Except from the bone-forming cells, the “osteoblast family” also includes the bone lining cells and the osteocytes. Bone lining cells are bone surface residing cells that are spindle shaped and metabolically and functionally quiescent. These cells separate bone matrix from the tissue extracellular fluid, creating a type of a tight “epithelium”. The separation of matrix and extracellular space is an essential feature of bone, which is often not appreciated. Under the influence of specific signals, bone lining cells can regain their bone forming capacity and produce bone. As osteoblastic function progresses and bone matrix expands, surface osteoblasts are eventually incorporated into the matrix that they have produced being transformed to osteocytes, the third member of the “osteoblast family” (Bonewald 2011). Osteocytes are permanent component of living bone and the most abundant cells in human skeleton. Indeed, in the fully mature skeleton, osteocytes compose the 90-95% of the total number of bone cells, which is enormous given that osteoblasts constitute only the 4-6% (Bonewald 2011). Since these cells are entrapped in mineralized bone matrix they harbor a delicate network of cell membrane processes, in order to communicate with each other and with the bone surface cells. These processes contain connexin-43 and run in small channels, the canaliculi, which permeate the extracellular matrix (Blair et al. in press). For years it was believed that osteocytes are terminally differentiated cells with small impact on skeletal metabolism. Nevertheless, it is now clear that the osteocyte network is surprisingly active, serving as an endocrine unit with key-roles in bone remodeling and homeostasis (Bonewald 2011).

One of the most important functions of skeleton is to receive and respond to mechanical stimuli, primarily related to gravity and movement. Osteocytes are the most mechanosensitive cells of bone. Indeed, their large number and the perfectly developed cell-cell and cell-surface communications, render them capable of receiving mechanical cues, which, in turn are converted to biochemical and ultimately molecular response, through a finely-tuned but extremely complex process called mechanotransduction (Thompson et al. 2012, Papachristou et al. 2009). Another worth-mentioning function of osteocytes is the production of sclerostin (Delgado-Calle et al. 2016), a protein that inhibits further growth of bone. Other important signals produced by osteocytes and osteoblasts are mediated by fibroblast growth factor 23 (FGF23), a specialized fibroblast growth factor family protein, which responds to 1,25-dihydroxyvitamin D3 and reduces phosphate reabsorption in the kidney (Huang et al. 2013).

Osteoclasts, the bone degrading cells

Osteoclasts are multinucleated cells derived from monocyte-family precursors (Blair et al. 2008), in sharp contrast to osteoblast that are of mesenchymal origin. Osteoclast precursors are mononucleated cells that during the process of differentiation and under the influence of specific factors that will be mentioned briefly later in this review, fuse forming the mature multinucleated cells, capable of bone resorption (Figure 1). The name “osteoclast” originates from the Greek words “osteo”-“óστó” that means bone and “clast” that means cut/destroy. As their name defines, the main function of these cells is the degradation of bone extracellular matrix. In order to accomplish this function osteoclasts adhere tightly to bone surfaces through specialized integrin (mainly ανβ3) adhesions. Acid-secreting H+-ATPase, which resides at their apical surfaces, facilitates bone mineral resorption by adding acid to it, solublilizing phosphate and calcium (Blair et al. 1989). The acid secretion is supported by chloride-proton exchange and chloride channels (Palagano et al. 2015, Schlesinger et al. 1997). Osteoclasts secrete acid proteinases, including cathepsins B and K and matrx metalloproteases (MMP) -9 and -13 that degrade collagen. Removal of bone components is achieved mainly by vacuolar translocation (Blair et al. 2008). As with other macrophage family cells, osteoclasts are dependent upon the macrophage colony stimulation factor-1 (CSF1), via the tyrosine kinase receptor c-fms, for survival and differentiation (Pixley & Stanley 2004, Blair et al. 2005). CSF1 is produced by mesenchymal cells including osteoblasts and stromal cells and is responsible for the differentiation proliferation and survival, of the committed cells and their progression to committed pre-osteoclasts (Figure 1). Osteoclast precursors also express the receptor RANK (receptor activator for nuclear factor k B) that regulates the maturation of mono- and multi- nucleated pre-osteoclats through its binding with RANK ligand (RANKL). RANKL is a type II homotrimeric transmembrane protein produced by osteoblast and stromal cell and is also secreted by activated T cells (Martin & Sims 2015). In addition to RANKL, osteoblasts and stromal cells also produce and secrete osteoprotegerin (OPG) a soluble RANK decoy receptor, which prevents RANK from acting at distant sites as a scavenger, and pharmacologically is a useful inhibitor of osteoclast formation (Boyle et al. 2003, Martin & Sims 2015). Increased RANKL/OPG ratio promotes osteoclast formation and thus bone resorption (Figure 1).

Primary osteoclast defects are rare, usually including defects in acid secretion, acid proteinases, and cellular adhesion (Blair et al. 2004).

There is a tremendous amount of additional information on osteoclast differentiation, too much for a general review. Molecules including pituitary hormones, sex steroids, glucocorticoids, specific cytokines and growth factors are implicated in the regulation of bone formation and degradation (Blair et al. in press); nevertheless, our appreciation of the overall physiological control of bone homeostasis is still vague.

Obesity, bone marrow fat, HDL and bone metabolism

The past few years there is gradually increasing volume of data documenting a strong connection between fat and bone metabolism (Papachristou & Blair 2016). Indeed, it has been demonstrated that disturbances in lipid metabolic pathways differentially affect bone cells resulting in the development skeletal pathologies. In this vein, a large number of epidemiological and animal studies have attempted to explore the association between obesity and bone mass. Notably, however, the results are contrasting and it is not clear whether obesity has a positive or negative effect on bone mass. As a matter of fact, several, mainly older studies, suggested that obesity in terms of increased body weight has a protective role against bone loss and osteoporosis development (Reid 2008, Reid et al. 1994). This notion was mainly supported by the fact that increased body weight results in augmented mechanical stimulation on bone, which promotes osteoblast activation and bone formation, while, on the other hand it inhibits osteoclastogenesis and bone resorption (Papachristou et al. 2009). From a metabolical standpoint, obesity is associated with the secretion of the pancreatic hormones insulin, resistin, amylin, as well as with increased estrogen levels, molecules that serve as critical regulators of bone metabolism (Barsh et al. 2002). In addition, visceral adipocytes can secrete adipokines, and in particular leptin, adiponektin, and resistin that also variably affect bone cells (Barsh et al. 2002). However, more recent studies have shown that the incidence of osteoporosis and bone fractures is increased in obese individuals compared to individuals within normal weight (Cao 2011). In this line, a more recent study on healthy premenopausal women showed that central adiposity is strongly associated reduced bone quality, stiffness and bone formation rate (Cohen et al. 2013). In addition, epidemiological studies in humans propose that fat mass determines the quality of bone in a manner that is independent of body weight, in postmenopausal women (Reid et al. 1994). One possible explanation for this is that the local phenomena that take place within bone marrow have strong effect on bone cells function, and thus on bone mass, that possibly overcomes the impact of bone-acting circulating hormones. Indeed, conditions that are characterized by increased visceral adiposity are associated with increased bone marrow fat and reduced bone mass (Ng & Duque 2010) (Figure 2). The increased bone marrow adiposity that is observed in these conditions is primarily attributed to the fact that osteoblasts and bone marrow adipocytes that have common progenitor, the mesenchymal stem cell, represent the two sides of the same coin. Therefore, increased bone marrow adiposity is accompanied by decreased number of osteoblasts and vise versa (Reid 2008).

Figure 2.

This diagram shows the molecular mechanisms that underline the effect of impaired and/or dysfunctional HDL on bone mass. The red arrows represent the “positive” effects, the blue arrows represent the “negative” effects and the green arrows are indicative of “no” or “undefined” effect. Abbreviations: ANXA2: Annexin-2, BM-ADC: bone marrow adipocytes, CLCX12: CXC chemokine ligand 12, Coll: collagen, IL: interleukin, MCP: macrophages, MSC: mesenchymal stem cells, OC: osteocalcin, OCL: osteoclast, OBL: osteoblast,ON: osteonectin,OPN: osteopontin, RANK: receptor activator for nuclear factor k B, RANKL: RANK Ligand, TNFα: tumor necrosis factor alpha.

The reduced bone mass observed in obese individuals is also linked to the fact that obesity elicits low-grade inflammatory response, primarily mediated by the pro-inflammatory cytokines tumor necrosis factor-α (TNFα), interleukine-6 (IL6) and IL1β (Okin et al. 2013) and affects bone formation-resorption equilibrium and hence bone mass. Activated T lymphocytes can directly bind macrophages, triggering the production and secretion of the aforementioned cytokines. Under normal conditions, HDL inhibits the T lymphocyte-macrophage interaction preventing the initiation of the inflammatory response (Burger et al. 2002). It is now well appreciated that inflammation has strong impact on bone remodeling differentially affecting the function of both the osteoblasts and the osteoclasts and thus possesses a central role in the development of bone-related metabolic pathologies, such as osteoporosis (Redlich & Smolen 2012). Osteoblastic differentiation and activation are under the strict control of the Wnt/β-catecinin signaling axis. Briefly, this axis is activated when extracellular Wnt binds to a specific co-receptor complex composed of a member the Frizzled family of receptors (FZD) and the extracellular domain of the LDL receptor- related proteins (LRPs) -5 and -6 (in vertebrates) (Pandur & Kuhl 2001). Among the common substrates of the Wnt/b-catenin cascade, which is referred as the “canonical Wnt pathway”, are RUNX2, COX2, parathyroid hormone (PTH) and Indian hedgehog (IHH), crucial regulators of osteoblastic physiology. In an inflammatory background, TNFα, IL6 and IL1β activate Dickkopf, Sclerostin, and secreted Frizzled-related proteins (sFRP), which impede the formation of the Wnt–LRPs–FZD assembly and hence suppress the anabolic downstream effects of the Wnt/b-catenin pathway on osteoblasts (Clevers 2006). Another mechanism that is responsible for the inflammation-related bone loss is the activation of the IL6-JAK-STAT [(JUN N-terminal kinase) / (signal transducers and activators of transcription)] signal transduction pathway that restrains the activation of the mitogen-activated protein kinase (MAPK) - AP1 signaling cascade, which in turn regulates the expression of the osteoblastogenic factors RUNX2, Osterix, Collagen type 1, Osteocalcin, PTHrP and OPG (Redlich & Smolen 2012; Krum et al. 2010). A similar effect is evoked via the activation of the TNF - SMAD ubiquitylation regulatory factor (SMURF) 1/2 cascade that also deteriorates the expression of osteoblast regulators (Kaneki et al. 2006).

The effect of low-grade inflammation is also pronounced on osteoclasts. More specifically, activated T lymphocytes express RANKL and therefore have the ability to directly bind and activate osteoclast-precursor cells, via the RANK-RANKL axis (Devlin et al. 1998, Ma et al. 2004). In addition, pro-inflammatory molecules secreted by the bone marrow adipocytes can enhance osteoclasts differentiation and activation in a RANK/RANKL - independent fashion (Kobayashi et al. 2000). Inflammation is also associated with the production of the monocyte chemoattractant protein 1 (MCP1), a pro-inflammatory cytokine that is also implicated in the transcription of osteoclast-specific genes, primarily through the JAK-STAT signaling cascade (Redlich & Smolen 2012). Nevertheless, less is know about the direct effect of HDL on osteoclasts physiology, a research field that is under scrutiny (Figure 2).

In an effort to shed light into the effect of diet-induced obesity on bone mass, a recent study on adult mice that were fed high-fat diet for 11 weeks, uncovered that bone responds to increased body weight in a bimodal fashion. Initially, obesity has an anabolic effect on bone due to enhanced mechanical loading and/or secretion of adipokines or growth factors that augment bone synthesis. Remarkably, however, following prolonged insult with high-fat diet animals enter a second, catabolic phase characterized by reduced bone mass, probably related with diet-induced alterations in metabolic profile of these obese mice (Lecka-Czernik et al. 2015).

The effect of the components of the HDL metabolic pathways on cartilage and bone

HDL and cartilage homeostasis

A considerable volume of epidemiological studies has uncovered a strong connection between osteoarthritis (OA) and pathological conditions such as hypertension, dyslipidemia, coronary heart disease and type 2 diabetes. Therefore, it not surprising that several authors suggest that OA should be considered a facet of metabolic syndrome (Velasquez & Katz 2010).

In symphony with the clinical data, molecular studies have shown that perturbations in lipid metabolic pathways are strongly related to the pathobiology of OA. Indeed, a sophisticated comparative proteome analysis on hypertrophied OA chondrocytes revealed that pathways regulating lipid metabolism, prostaglandin synthesis, glutathione metabolism and metabolism through the cytochrome p450, display differences between normal and hypertrophic OA chondrocytes (Tsolis et al. 2015). Moreover, gene expression analyses on human-derived OA chondrocytes uncovered disturbed expression levels of genes that regulate cholesterol influx and efflux and lipid metabolism (Tsolis et al. 2015). It has been also documented that the expression of a set of genes that regulates cholesterol efflux, namely ABCA1, APOA1 and LXRα, is significantly suppressed in chondrocytes derived from OA in comparison to normal cartilage (Tsezou et al. 2010). Notably, following treatment with the LXR agonist TO-901317, the OA chondrocytes displayed augmented ABCA1 and APOA1 expression and cholesterol efflux (Tsezou et al. 2010). This finding is in line with a previous study by Collins-Racie and colleagues having examined 48 nuclear receptors and found that the expression levels of 23 of them had significant differences between normal and OA human cartilage samples (Collins-Racie et al. 2009). Interestingly, the mRNA of the nuclear receptors LXRα and LXRβ, and their heterodimeric partners retinoid X receptor (RXR)α and RXRβ, as well as the LXR target genes ABCG1 and apolipoproteins D and E, were greatly reduced in the OA as compared to the normal cartilage, a finding proposing that the HDL metabolic pathway is deregulated in OA. Their finding that the LXR signaling alterations are restored with the use of the LXR agonist TO-901317 suggests LXR signaling modulation could add to the therapeutic armamentarium against OA (Collins-Racie et al. 2009).

Driven by the aforementioned data, and based on previous results of our research team on the role of the lipoprotein transfer system and HDL metabolic pathways in the pathogenesis of diet-induced obesity and non-alcoholic fatty liver disease (Karagiannides et al. 2008, Karavia et al. 2012), we were tempted to further explore the role of HDL in the pathobiology of OA. For this purpose we applied histological, histomorphomentrical and molecular/biochemical methodologies and found that after the consumption of Western-type diet both LCAT and APOA1 deficient mice developed OA, in contrast to the control groups (Triantaphyllidou et al. 2013). This, suggests that alterations in HDL biogenesis and maturation predispose to the development of OA in mice, following chronic insult with Western-type diet. Notably, we also observed that in sharp contrast to the LCAT^-/- mice, the APOA1^-/- were not quite as obese, an observation that raises the challenging possibility that altered HDL metabolism may have a direct destructive effect on articular cartilage that is independent of body weight, and significantly contributes to the development of OA (Triantaphyllidou et al. 2013).

HDL and Bone Homeostasis

Recent data propose that there is an association between serum HDL levels and bone mass (Papachristou & Blair 2016). However, the results that have been generated from epidemiological studies on humans are contradicting. Indeed, even though a considerable volume of these studies have shown that increased HDL is associated with better bone quality and reduced risk of osteoporosis, others support a negative relation between serum HDL levels and bone mass (Li et al. 2015, Jeong et al. 2010, Ackert-Bicknell 2012). It is believed that factors such as genetic background, age, dietary habits and metabolic status are responsible for this inconsistency. Therefore, research groups that work on this field have recruited experimental animal models in an effort to illuminate the molecular mechanisms that link HDL and bone homeostasis, as will be described in the following sections of the present review article.

The role of Scarb1

Several recent studies have investigated the possible connection between HDL and bone metabolism. Towards this direction the group of Moreau has extensively studied the role of the principal HDL receptor, the Scavenger receptor class B, type I (SRBI), in the regulation of bone mass. SRBI is the product of the Scarb1 gene, and binds HDL with high affinity, facilitating the transportation of cholesterol from peripheral organs to the liver (Acton et al. 1996) a process called reverse cholesterol transport.

In an elegant study, this group examined the phenotypic characteristics of femora obtained from Scarb1 deficient mice, in comparison to their wild type counterparts (Martineau et al. 2014, Martineau et al. 2014). Serologic analysis revealed that the Scarb1 knock out mice had significantly elevated both total plasma and HDL-cholesterol. Dynamic and static (μCT-based) histomorphometry uncovered that Scarb1 deficiency was followed by increased bone volume and elevated osteoblast number and function, particularly in the trabecular bone of of 2 month-old female mice femora. Their in vitro approaches on MSC obtained from knock out (KO) and wild type mice were in symphony with their histological and histomorphometrical results, further supporting the osteoblastogenic behavior of the Scarb1 null mice. SRB1 regulates the uptake of cholesterol at the adrenal glands, and thus Scarb1 deficient mice are characterized by reduced glucocorticoid but elevated adrenocorticotropic hormone (ACTH) levels (Martineau et al. 2014, Martineau et al. 2014). Given that high concentrations of ACTH are anabolic for osteoblasts (Isales et al. 2010) this mechanism may possibly explain the high bone mass phenotype observed in the Scarb1 KO mice. Nevertheless, other mechanisms such as intrinsic cellular alterations of the canonical and non-canonical Wnt pathways that are not directly related to plasma HDL levels may explain the increased bone mass of Scarb1 KO mice (Martineau et al. 2014), further adding to the complexity of the relationship between HDL and bone metabolism.

The “enigmatic” role of APOE

ApoE is a 34.2kDa glycoprotein produced by the liver and other peripheral organs. Its main functions are atheroprotection and maintenance of plasma lipid homeostasis, since it mediates the cellular uptake of chylomicron remnants, very low-density lipoprotein and low-density lipoprotein and their clearance from the circulation. Notably, APOE is also implicated in the de novo biosynthesis of HDL (Kypreos & Zannis 2007). In a relatively recent study, investigating the genetic connection between HDL and bone and mineral density (BMD), Ackert-Bicknell reported that APOE, along with PPARγ, ESR1 and IL6, belongs in a set of genes that regulate both HDL and BMD (Ackert-Bicknell 2012). Nevertheless, the role of APOE in bone pathophysiology, still remains quite puzzling. Indeed, back in 2005 Schilling et al., first published the results of a very interesting research study on the effects of APOE deficiency on bone mass. With the use of histological, histomorphometrical, biomechanical and in vitro approaches the authors showed that the ApoE null mice exhibit augmented osteoblastic function and increase rate of bone synthesis (Schilling et al. 2005). Previous in vitro and experimental animal-based studies have documented that APOE deficiency is associated with reduced uptake of vitamin K containing lipoproteins by osteoblasts and that exogenous administration of APOE increases the uptake of chylomicron remnants that contain vitamin K (Niemeier et al. 2005, Newman et al. 2002). Taking into account that Vitamin K is essential for osteocalcin carboxylation the authors correctly surmised that impaired APOE results in enhanced bone mass, paralleling the phenotype of the osteocalcin-depleted mice (Niemeier et al. 2005). Nonetheless, this mechanism does not explain the whole phenotype of APOE deficiency, since studies in mouse models have shown that shortage of osteocalcin is associated with augmented fat mass (Oldknow et al. 2015), in contrast to APOE deficiency that is characterized by reduced body fat content and smaller adipocytes (Huang et al. 2006). To further add to the complexity of the role of APOE in bone biology, a more recent study by the same group showed that the bone mass of APOE null mice is substantially reduced after consumption of diabetogenic high fat diet for 16 weeks. These knock out mice also have an interesting metabolic profile, characterized by normal weight and lowered levels of serum insulin, glucose and leptin (Bartelt et al. 2010). The aforementioned findings reinforce the prevailing notion that body weight per se cannot define the quality of bone; rather it appears that the fine balance between local and systemic metabolic pathways is responsible for the determination of bone mass.

As mentioned previously, osteoblasts and lipoblasts originate from the same precursor, the bone marrow mesenchymal stem cell. Interestingly, APOE deficient mice challenged with high fat display both reduced bone mass and bone marrow adiposity (Bartelt et al. 2010). This observation allows drawing the conclusion that lack of APOE possibly halts bone marrow MSC maturation in early stages and thus affects both lipoblastic and osteoblastic lineage, through unidentified mechanisms. More recently, Dieckerman et al. performed biochemical, histomorphometrical and genetic analyses on transgenic mice and demonstrated that human APOE2 has the most significant effect on bone turnover, in comparison to the other natural human isoforms, namely APOE3 and APOE4. They also presented data introducing the APOE ε2 allele as a latent risk factor for reduced trabecular bone mass and vertebrae fractures in humans (Dieckmann et al. 2013).

The role of APOE in the function of osteoclasts is also poorly understood. The studies of Bartelt et al. and Schilling et al. demonstrated that shortage of APOE does not influence the number or the function of osteoclasts in mice fed both high-fat or chow diet (Bartelt et al. 2010, Schilling et al. 2005). In sharp contrast, a more recent in vitro report on bone marrow-derived macrophages showed that APOE has an inhibitory effect on osteoclasts, which is mediated by c-FOS-, NFATc1- and NF-κB- related signaling cascades (Kim et al. 2013).

In aggregate, the current literature indicates that APOE serves as a potent regulator of mesenchymal stem cells and therefore affects bone mass; nevertheless, its' role on osteoclastic function warrants further investigation.

The role of APOA1

APOA1 is key-molecule in the regulation of HDL biogenesis. Its role in the regulation of bone mass is graphically presented in figure 3. We have recently documented that in addition to its role in atheroprotection, APOA1 is also implicated in the development of non-alcoholic fatty liver disease as well as in the pathogenesis of osteoarthritis in mice (Karavia et al. 2012, Trantaphyllidou et al. 2013). In order to further explore whether HDL possesses a causative role in the regulation of bone mass, we studied the involvement of LCAT and APOA1 in the function of bone cells. For this purpose we used 12-week-old Lcat and ApoA1 KO male mice. Static and dynamic histomorphometrical analyses showed that the bone quality, as well as the rate of new bone formation and the number of tartrate-resistant acid phosphatase (TRAP) positive cells, were similar between the Lcat KO and the wild type animals. Importantly, however, the ApoA1-/- mice had significantly reduced bone mass in comparison their wild type counterparts, implying that it is the impaired synthesis and not the incomplete maturation of HDL that affects bone mass. Bone quality studies including Raman spectroscopy and three-point bending test revealed that these mice had a remarkable reduction of cross-linked collagen, which insinuates a significant deterioration of the biomechanics properties of femora obtained from these mice. Dynamic histomorphometry showed a significant reduction in calcein-labeled and double-labeled surfaces, but no differences in the TRAP positive cells between the two groups. Through a series of in vitro and molecular experiments on mesenchymal stem cells, osteoblasts and osteoclasts isolated and cultured from femora of the two mouse groups we uncovered that the osteoblast-related factors and signaling axes, RUNX2, osterix, COLLa1, RANKL, were significantly impaired in the ApoA1 KO as compared to the wild type mice. On the contrary, genes that regulate osteoclastic differentiation and function namely TRAP (Acp5), cathepsin K (Ctsk) and RANK (Tnfrsf11a), were unaffected (Blair et al. 2016). These interesting findings indicate that the reduced bone mass observed in the ApoA1 KO mice, is attributed primarily to suppressed osteoblastic bone synthesis and not to increased osteoclastic bone degradation.

Figure 3.

Diagram depicting the molecular changes observed in the ApoA-1 knock out mice (B) in comparison to their wild type counterparts (A). All the molecular alterations are described in detail in the text. Note that the increased or decreased expression levels are represented by larger or smaller boxes. For example in the ApoA-1 deficient mice, the expression of CXCL12 is significantly decreased (X0.3), while the expression levels of its' CXCL12 receptor CXCR4 is greatly elevated (X3). Abbreviations: APOA1: Apolipoprotein A-1, CLCX12: CXC chemokine ligand 12, CXCR4: CXC Receptor 4, LBL: lipoblast, MSC: mesenchymal stem cells, OBL: osteoblast, WT: wild type.

It is well-appreciated that bone marrow adiposity possesses an essential role in the regulation of osteoblasts and osteoclasts, and consequently bone mass (Lecka-Czernik and Stechschulte 2014, Scheller et al. 2016). Histological examination of ApoA1 KO mice femora revealed that bone marrow of these animals have significantly elevated number of lipoblasts. In addition to the enhanced adiposity, the mesenchymal stem cells derived from the ApoA1 deficient mice exhibit increased expression levels of the lipoblast master regulators PPARγ and CEBPa (Blair et al. 2016). Taking under consideration our finding that the number of bone marrow total mesenchymal stem cell population of both mouse groups was similar, we hypothesize that APOA1 may perform as a modulator of the mesenchymal stem cells pools within bone marrow, tipping the balance towards osteoblastogenesis (Blair et al. 2016). The mechanisms that underline this phenomenon remain insufficiently elucidated.

One the most potent regulators of mesenchymal stem cell homing and bone synthesis is the CXC chemokine ligand 12 (CLCX12) that binds primarily the receptor CXC Receptor 4 (CXCR4) and is regulated, at least in part, by Annexin-2 (ANXA2) (Jung et al. 2001). We have very recently reported that APOA1 deficiency greatly disturbs the ANXA2-CLCX12-CXCR4 connection, providing a plausible mechanistic explanation for the impaired osteoblastic function observed in the ApoA1 KO mice (Blair et al. 2016). It should also be borne in mind that the CLCX12-CXCR4 axis is also crucial for the regulation of osteoblastic bone marrow niche that participates in an array of processes, including hematopoietic stem cell (HSC) homing and mobilization and bone metastasis (Greenbaum et al. 2013, Kfury & Scadden 2015). Taken together, these reports fuel the attractive hypothesis that APOA1 may have a role in the pathobiology of HSC dormancy and/or motivation and thus it may be involved in the development of hematopoietic neoplasms. Moreover, given that reduced HDL levels are associated to increased metastasis rate (Pan et al. 2012), we advance the intriguing theory that perturbations in APOA1 may be implicated in the pathogenesis of bone metastases through alterations in signaling cascades and molecular axes, such as ANXA2-CLCX12-CXCR4 and RANK-RANKL, which modify the microenvironment of the osteoblastic niche. Definitely, further studies are essential for the substantiation of these hypotheses.

Conclusions and Future Perspectives

As research on HDL progresses, new data in the literature strongly support a causative role of dysfunctional HDL with a number of metabolic disorders, including bone metabolic diseases (Papachristou & Blair 2016, Constantinou et al. 2015). Nonetheless, several critical issues should be addressed in order to unfold the “mysteries” that underline the bone-HDL connection. First, the involvement of additional key-regulators of the lipoprotein metabolism, such as ABCA1, LDLR, SRB1 in bone biology should be extensively assessed. The use of genetically modified mice is the best approach towards the salvation of these issues. Second, the role of bone marrow adiposity must be thoroughly examined. It is known that elevated marrow adiposity variably affects bone cells and is involved in the pathogenies of bone-related pathologies such as osteoporosis and bone metastases. Moreover, novel data suggest that bone marrow microenvironment, and the osteoblastic niche in particular, are greatly influenced by HDL status. Taking into account that osteoblastic niche regulates the fine balance between dormancy and motivation of the HSCs and plausibly participates in the pathogenesis of hematopoietic neoplasias it would be very intriguing to unveil the molecular cobblestone that link HDL metabolism to the development of these diseases. Finally, it should be kept in mind that HDL efficacy does not rely solely on the HDL-C levels, but also on the HDL particle functionality, which is defined by HDL apolipoprotein and lipid content (Filou et al. 2016). For this reason, the in-depth appreciation of the structure-function association of HDL, will pave the way not only towards the thorough understanding of the mechanisms that link HDL and bone mass, but also towards the development of effective pharmaceuticals that will target HDL functionality. It is our expectation that the identification of surrogate markers of HDL functionality may prove invaluable for predicting the risk for developing bone metabolic disorders.

Obviously, many “rivers need to be crossed” in order to achieve a proper understanding of the molecular basis of the complex HDL-bone interactions. However, it is now clear that the functional cross-talk between HDL and bone metabolism determines the beginning of a beautiful friendship that can potentially set the basis for the development of novel HDL-directed pharmaceuticals for the treatment of bone pathologies.