Bone is a complex connective tissue composed of a calcified extracellular matrix in which different cell types are embedded. The complexity of bone is particularly evident in the cells present in this tissue and the multifaceted interactions between them [1]. Bone cells belong to two different families: the osteoblastic and the osteoclastic families (see Figure 1) [2,3].

There are several cell types within the osteoblastic lineage, including osteoblasts, osteocytes and lining cells. All of them derive from mesenchymal precursors that differentiate into osteoblasts, which eventually evolve into osteocytes or lining cells [4]. Cells of the osteoblastic lineage modulate the proliferation and differentiation of cells belonging to the osteoclastic lineage, mainly by the RANKL-OPG-RANK signaling pathway [5,6,7]. Osteocytes derive from some osteoblasts that become embedded and surrounded by bone matrix [8]. These cells are emerging as the responders to mechanical stimuli that regulate bone formation and resorption, as well as key regulators of bone metabolism. Osteocytes modulate bone turnover through the modulation of the Wnt pathway and other pathways that influence the activity of both osteoblasts and osteoclasts [9,10,11].

Osteoclasts derive from hematopoietic precursors. These cells are formed by the fusion of cells of the monocyte-macrophage lineage and are responsible for bone resorption. As mentioned before, osteoclastogenesis is influenced by osteoblasts and osteocytes, which produce several factors critical for the differentiation of osteoclast precursors, such as the receptor activator of nuclear factor kappa-B ligand (RANKL) and the macrophage colony-stimulating factor (MCS-F) [7]. RANKL interacts with RANK (Receptor activator of nuclear factor kappa-B), present in the membrane of osteoclastic precursors, promoting the activation of the Nuclear factor Kappa B, which induces cell fusion and differentiation to originate multinucleated mature osteoclasts [12,13]. It is important to note that osteoblastic cells also express osteoprotegerin (OPG), which acts as a soluble decoy receptor for RANKL, thus impairing the RANKL-RANK interaction [14]. Mature osteoclasts locate on specific surfaces of bone, where they break down and resorb bone matrix, replacing old bone with new bone [12].

Bone is under constant turnover throughout life in order to maintain its properties. It is believed that this process, known as bone remodeling, occurs, in part, randomly. However, sometimes, osteocytes can mark the site where a remodeling cycle must be started in order to repair microcracks, which represent the so called “targeted remodeling” [15,16]. Bone remodeling is carried out by the cyclic and coupled activity of osteoclasts and osteoblasts. The process starts when osteoclast precursors are recruited to the bone surface, where they differentiate and remove a small volume of bone. Then osteoclasts undergo apoptosis and osteoblasts arrive to the region to fill up the defect with new bone [17]. Bone mass homeostasis depends on the balance between bone formation and bone resorption. Any disequilibrium between bone formation and bone resorption leads to changes in bone mass. This is the case of osteoporosis and osteoarthritis, two common skeletal diseases that tend to show changes of bone mass in opposite directions [18].

The maintenance of bone mass requires proper cell differentiation and cell activity. Osteoblast and osteoclast differentiation processes are highly organized and driven by deep changes in the gene expression patterns that in turn result in cells with different shapes and functions [3,19]. Since bone remodeling requires the sequential action of osteoclasts and osteoblasts at a given region, the differentiation of both cell types is controlled in a time and site-specific manner, and influenced not only by intrinsic factors, but also by some systemic and environmental factors. Emerging lines of evidence suggest that epigenetic mechanisms play an important role in establishing cellular identities and controlling gene expression. Herein, we summarize the current knowledge about the role of epigenetics, and specifically DNA methylation marks, in bone homeostasis and pathogenesis.

Presently, epigenetics is defined as, “The study of stable genetic modifications that results in changes in gene expression without a corresponding alteration in DNA sequence [20]”. Importantly, epigenetic marks integrate intrinsic and environmental stimuli and confer both lineage commitment and phenotypic plasticity [21]. Thus, epigenetic marks could be considered as a link between genotype, environment, phenotype and disease. Epigenetic mechanisms comprise DNA methylation, post-translational histone modifications and non-coding RNAs [22,23]. Although the mechanisms of epigenetic inheritance during cell division are well established, heritable epigenetic patterns from parents to offspring are starting to be revealed [24,25].

In eukaryotes DNA methylation consists in the covalent addition of methyl groups to cytosines that precede guanines (CpG) [26]. In vertebrate genomes, CpG sites are predominantly methylated [27]. However, the globally methylated pattern is disrupted in regions known as CpG islands. These areas are present in approximately 70% of gene promoters. On average, CpG islands are 1 kb long, have an elevated C + G content and are frequently demethylated [28]. DNA methylation variations do not occur exclusively at CpG islands. Recently, it has been also shown that methylation occurs at a short distance from the CpG islands (at “CpG island shores”), rather than in the islands themselves [29]. The term CpG island shore refers to regions of lower CpG density that lie in the vicinity (~2 kb) of CpG islands. The addition of methyl groups to cytosines is catalyzed by DNA methyltransferases (DNMTs) [30]. DNMT1 was the first methyltransferase identified in mammals [31]. DNMT1 maintains DNA methylation during cell division by reading and copying the pattern in the hemimethylated strand [31]. Other members of the family are DNMT3A, DNMT3B, responsible for de novo methylation [32], and DNMT2, which has been recently shown to methylate tRNAs [33]. DNA methylation is considered an efficient repressor of transcriptional activity. Until recently, it was commonly thought that methyl groups directly prevent the binding of essential transcription factors to their targets. Although this is true for a specific set of transcription factors, it is not a general phenomenon. In fact, the binding sites for many transcription factors do not have CpGs. Emerging evidences support that the presence of methyl groups models the surrounding chromatin, inducing a DNA conformation less accessible to the transcription machinery. The mechanisms underlying this packing change are not fully understood yet, but many studies have concentrated on nucleosome structure, methyl-binding proteins, such as MECP2, MBD2 and MBD3, and interactions with chromatin remodeling enzymes [34,35,36]. Whatever the mechanism, on the basis of its potential to silence promoters, DNA methylation is supposed to play an important role in cell commitment and cell-specific gene expression.

How methylation is specifically targeted to a subset of promoters is generating intense debate. Apparently, methylation patterns are established early in the embryo [32,37,38]. Several studies support the idea that the ultimate methylation profile is determined by the underlying DNA sequence. In this sense, it has been shown that local DNA sequence is one of the main determinants for targeting DNA methylation to a specific locus. Thus, sequence variation between individuals might contribute to differential methylation patterns [38,39]. In fact, recent findings suggest that allele-specific methylation (ASM) is a common feature across the genome [40,41]. Notably, most of the ASM is strongly associated with SNPs genotypes [42,43]. Subsequent changes in the methylation pattern, generally of a tissue-specific nature, occur following implantation (i.e., repression of pluripotent genes) [44,45]. It has been proposed that tissue-specific changes occur through mechanisms apparently recruiting molecules needed for de novo methylation and demethylation (see Figure 2). Whereas demethylation may occur by active or repairing mechanisms [46,47], de novo methylation may be mediated by polycomb complexes [48].

An impaired mesenchymal differentiation negatively affects bone mass. Several studies suggest that the osteogenic capacity of mesenchymal cells decrease with aging [49]. However, the contribution of this event to the decline of bone mass associated with aging is not clear yet. Osteogenic differentiation of mesenchymal cells towards osteoprogenitor and osteoblastic cells is regulated by several mechanisms, including DNA methylation. Kang et al. demonstrated that promoter methylation changes during mesenchymal cell differentiation [50]. Likewise, it has been proposed that active demethylation of gene promoters (i.e., osteocalcin, osterix, or Runx2), via GADD45-dependent mechanisms, is involved in the ostegenic differentiation of mesenchymal cells [51]. In fact, as reported by Locklin et al., osteogenic differentiation may be modulated by demethylating agents [52]. DNA methylation marks are not only important in osteoblastogenesis, but also afterwards in the osteoblast to osteocyte transition. Our group demonstrated that CpG methylation at regulatory regions controls the expression of the alkaline phosphatase and sclerostin genes. We observed that the hypermethylation of ALPL and SOST promoters was inversely correlated with gene expression in osteoblastic cells. Furthermore, we showed that the presence of methyl groups at the proximal promoter of SOST markedly decreased the transcriptional activity of this sequence, presumably by impairing the binding of essential transcription factors to the core promoter. In addition, we demonstrated that the methylation of those promoters changes during osteoblast differentiation towards osteocytes and controls gene expression in a cell-specific manner [53,54]. DNA methylation at ALPL promoter increased progressively during osteoblast differentiation, silencing ALPL expression in osteocytes. DNA methylation represses SOST expression in osteoblasts, whereas the physiological demethylation of its promoter favored the expression of this gene in osteocytes (see Figure 2). Consistent with this observation, SOST promoter remains methylated in other cell types that do not express sclerostin [53]. The results of other investigators also support that the expression of a number of genes important for osteogenic differentiation and osteoblast/osteocyte activity (including podoplanin, osteopontin, Brachury transcription factor, estrogen receptor, aromatase, collagen cross-linking enzyme lysyl oxidase or the homeobox protein Dlx-5) is regulated by DNA methylation [55,56,57,58,59,60,61,62].

On the other hand, DNA methylation contributes to regulate self-renewal and differentiation capacity of hematopoietic cells, the early precursors of osteoclasts [63]. The subsequent differentiation of osteoclast precursors towards mature osteoclasts is tightly regulated by soluble factors secreted by osteoblasts and osteocytes, such as RANKL, OPG and MCS-F [7,64]. As first reported by Kitazawa et al., and recently confirmed by our group in human samples, RANKL and OPG expression is regulated by DNA methylation in osteoblastic cells [65,66]. CpG methylation at the regulatory regions of RANKL and OPG genes is associated with low transcript levels. In turn, the demethylation of their promoters mediated by 5-azadeoxycitidine, a demethylating agent, induces the expression of both genes. Based upon these evidences, epigenetic mechanisms appear to be important for osteoclast differentiation (reviewed by Yasui et al. [67]). However, it is worth emphasizing that besides DNA methylation, other epigenetic marks, such as microRNAs or chromatin modifiers, are also involved in determining the differentiation and activity of bone cells, recently reviewed by Delgado-Calle et al., Kato S et al. and Earl CS et al. [68,69,70].

Common skeletal disorders, such as osteoporosis and osteoarthritis, are the result of a complex interplay of genetic and acquired factors. However, despite tremendous efforts, including several GWAS, the genetic factors so far identified explain less than 10% of the genetic risk. This suggests that mechanisms not related to DNA sequence may be involved in the development of these diseases. This could be the case of DNA methylation and its mediators. DNA methylation marks are heritable, at least through cell divisions, control gene response to environment, change with aging and underlie cell commitment and the spatiotemporal control of gene expression. DNA methylation plays an important role in the differentiation of cells of the osteoblastic and osteoclastic lineages. Therefore, it is tempting to speculate that the aberrant phenotypes observed in bone diseases might be the consequence of a combination of intrinsic and environmental factors, including gene sequence variations and epigenetic signatures (see Figure 3). However, it is important to note that DNA methylation is only one of the mechanisms underlying gene expression. Thus, the integration of knowledge from both epigenenomics and genomics, together with other “omics” (i.e., transcriptomics, proteomics) will be essential for the full understanding of the underlying mechanisms that govern the initiation and progression of bone diseases. Although still a long way to go, further studies in bone epigenetics may open a new door for drug development combining genetic and epigenetic strategies.