Transforming growth factor-beta (TGF-β)/bone morphogenetic protein (BMP) plays a fundamental role in the regulation of bone organogenesis through the activation of receptor serine/threonine kinases. Perturbations of TGF-β/BMP activity are almost invariably linked to a wide variety of clinical outcomes, i.e., skeletal, extra skeletal anomalies, autoimmune, cancer, and cardiovascular diseases. Phosphorylation of TGF-β (I/II) or BMP receptors activates intracellular downstream Smads, the transducer of TGF-β/BMP signals. This signaling is modulated by various factors and pathways, including transcription factor Runx2. The signaling network in skeletal development and bone formation is overwhelmingly complex and highly time and space specific. Additive, positive, negative, or synergistic effects are observed when TGF-β/BMP interacts with the pathways of MAPK, Wnt, Hedgehog (Hh), Notch, Akt/mTOR, and miRNA to regulate the effects of BMP-induced signaling in bone dynamics. Accumulating evidence indicates that Runx2 is the key integrator, whereas Hh is a possible modulator, miRNAs are regulators, and β-catenin is a mediator/regulator within the extensive intracellular network. This review focuses on the activation of BMP signaling and interaction with other regulatory components and pathways highlighting the molecular mechanisms regarding TGF-β/BMP function and regulation that could allow understanding the complexity of bone tissue dynamics.
Bone, a specialized form of connective tissue, is the main element of the skeletal system. Its formation is a very complex but finely orchestrated process in which bone morphogenetic protein (BMP) plays the major role in the regulation of osteoblast lineage-specific differentiation and later bone formation.1 More insight into the functions of BMP has been gained from in vivo experiments using transgenic animals to reflect the importance of BMP in osteogenesis. Studies on loss-of-function and gain-of-function mutations result in various bone-related abnormalities during development.2,
The evolutionary significance of the BMP family is highlighted by the conserved nature of the canonical TGF-β/BMP signaling over at least 700 million years that proves the vitality and importance of BMP for vertebrate physiology.10 The highly conserved canonical TGF-β/BMP linear signaling cascade engages the TGF-β/BMP ligands, two cell surface BMP receptors (BMPRs), and signal transducers, Smads.11 Mechanistically, phosphorylated C-terminus receptor-regulated-Smad (R-Smad), specific for the BMP pathway, interacts with various downstream proteins, including Runx2, which in turn can induce bone differentiation factors.12 On the other hand, non-canonical Smad-independent signaling (p38 mitogen-activated protein kinase, MAPK) pathway also implicates the Runx2 gene to control the mesenchymal precursor cell (MPC) differentiation.1 The coordinated activity of Runx2 and BMP-activated Smads is critical for bone formation. Furthermore, a large number of gene products and pathways pointed to promote osteoblastogenesis and bone formation for maintaining the stability of bone microenvironment.13 Bone dynamics maintain delicate interactions between TGF-β/BMP and other pathways, which are tightly regulated spatiotemporally, giving rise to the remarkable complexity, diversity, flexibility, and delicacy of TGF-β/BMP functions.14,15 Several findings explore different modes of cross-talk between BMP signaling and other major signaling pathways, namely Wnt, Hedgehog (Hh), Notch, and MAPK,16,
Interconnected signaling is responsible for final target gene expressions required for osteogenesis. As perturbations in the signaling result in bone diseases, so there is a great potential for clinical applications of TGF-β/BMP molecules for the treatment of bone disorders. However, the outcomes of clinical applications of BMP largely rely on a precise design of cell therapies. So, the structure of BMP and its receptors is of immense interest due to their pathophysiological implications in osseous and non-osseous diseases. Hence, this review emphasizes on the structure of BMP and its receptors, and its role in integrated regulatory mechanisms in signaling. This review also highlights different modes of cross-talk between BMP signaling and the signaling pathways of MAPK, Wnt, Hh, Notch, and FGF, in which Runx2 is a key transcriptional regulator of osteoblast differentiation and bone formation. This knowledge of the integrated signaling pathways helps to enable the rational design of cell therapies to replace damaged or aged bone tissues.
BMP and its receptors
The human genome encodes more than 20 homodimeric or heterodimeric BMP ligands20 which can be divided into four distinct subfamilies, according to their amino acid sequence homology, structures, and functions: (i) BMP2 and 4; (ii) BMP5, 6, 7, 8a, and 8b; (iii) BMP9 and 10; and (iv) BMP12, 13, and 14.6 BMP, soluble local-acting signaling proteins can function in an endocrine, paracrine, and autocrine manner. BMP2, BMP4 through to BMP7, and BMP9 have been shown to induce intra-membranous and endochondral bone formation. Of these, BMP2, 6, and 9 appear to have important roles in the induction of mesenchymal cell differentiation into osteoblasts.21 However, not all members are truly osteogenic. For example, BMP7 is involved in kidney, eye, and limb development; BMP4, 7, and 14 are important for proper reproductive tissue development; BMP2, 3, and 7 contribute to cartilage regeneration; and BMP12 and 13 are required for normal tendon healing.22,23 Recent studies suggest that BMP6 and 9 may be more effective in promoting orthotopic bone formation when compared to BMP2 and 7.24,25Table 1
BMP is synthesized by osteoblasts as 400–500 amino acid peptide precursors. The large precursor BMP consists a hydrophobic secretory leader (signal peptide at the N-terminus) and a pro-peptide sequence (non-conserved domain) joined to the mature region (C-terminus domain).6 The mature domain of each BMP has a region of seven conserved cysteine amino acids at their C-terminal which contains 100–140 amino acid residues, six of which are involved in forming a characteristic structural motif (a cysteine-knot with two finger-like double-stranded sheets).12 In general, the C-terminal active domain, by a subtilisin-like convertase cleavage at a consensus dibasic Arg-X-X-Arg site, is released to generate mature dimers by disulfide bonds and dimerization which is a prerequisite for bone induction.53 The N-terminal region controls the stability of the processed mature protein, and the downstream sequence adjacent to the cleavage site determines the efficiency of cleavage.54 BMP has sites for N and O glycosylation, which increases the stability and half-life of the protein, in addition to determining the specificity of receptor coupling.55 Nevertheless, only BMP2 and 4 possess a potential secondary cleavage site. BMP2 is a multifunctional growth factor and a homodimer of two 114 peptides subunits, and two receptor-binding motifs have been identified in the BMP2 ligand. The wrist epitope, comprising of residues from both BMP2 subunits, has a high-affinity binding site for BMPRIA, while the knuckle epitope consists of one subunit and weakly binds to BMPRII.45 On the other hand, the monomer (BMP7) is stabilized by three disulfide bonds, i.e., Cys-67-Cys-136 and Cys-71-Cys-138 forming a ring through which the third, Cys-38-Cys-104, passes.56 The cysteine knot constitutes the monomer core from which four strands of anti-parallel β-sheets emanate, forming two finger-like projections. Figure 1 summarizes the genomic and proteomic organization of BMP and its receptors.
BMPRs are composed of three parts: a short extracellular domain, a single membrane-spanning domain, and an intracellular domain with the active serine/threonine kinase region.57 Three BMPRI (BMPRIA or ALK3, BMPRIB or ALK6, and ALK2: IA activin receptors) and three BMPRII (BMPRIIB, ACTRIIA, and ACTRIIB) are known to bind BMP. Out of the three BMPRI, ALK3 and ALK6 are structurally highly similar to each other, whereas ALK2 is less similar to the other two.51 One unique feature of BMPRI is the presence of a highly conserved TTSGSGSG motif (GS domain) in their cytoplasmic region that plays a key role in kinase activity.19 On the other hand, BMPRII has a unique, long C-terminal tail with 530 amino acids after the kinase domain.58 BMPRII has constitutively active kinases and BMPRI has inactive and inducible kinases. Recently, DRAGON, a glycosylphosphatidylinositol-anchored protein, was described as a co-receptor that directly interacts with ligands and receptors to facilitate BMP signaling.59 Again, Samad et al. also reported a second co-receptor: repulsive guidance molecule, a homologue of DRAGON, seems to be similarly involved in the BMP signaling pathway. The BMPRs were housed in a specific membrane domain of the caveolin-1 β (CAV1β) and clathrin-coated pits at cell surface.60
Normally, the BMP ligand initiates signaling by binding to and bringing together BMPRI and BMPRII receptors on the cell surface to form a ternary holocomplex.61 From the current understanding, heteromeric complexes of BMPRI and BMPRII are required for signal propagation. However, BMPRII is the primary binding site of the BMP ligand and phosphorylate of the BMPRII. And only then that BMPRIA signals are promoted to downstream substrates. It is believed that BMPRII does not actually bind the ligand but rather stabilizes and/or accelerates ligand binding to BMPRI. The specificity of the intracellular signals is mainly determined by BMPRI. Most typically, BMP2 and 4 bind to BMPRIA and BMPRIB, whereas BMP6 and 7 bind strongly to ALK2 and weakly to BMPRIB. BMPRII is specific for BMP, whereas ActRII and ActRIIB are shared by activins, myostatin, and BMP.62 The C-terminal tail of BMPRII has important functions in the regulation of cytoskeletal protein functions. BMPRII inhibits the ability of LMK1 to phosphorylate cofilin through interaction with its C-terminal tail and this inhibition is alleviated by BMP4.63Figure 1d,e illustrates the interaction of BMP and its receptor and receptor organization.
BMP signaling and implications in pathophysiology of bone formation
Both BMP and BMPRs are important not only for skeletal development and regeneration but also for the homeostasis of normal skeletal muscle mass. Recently it has been identified that the substitution of ″G″ with ″A″ in the activin A receptor type I/ACVR1 (ALK2) gene of juxta-membrane GS domain at position 617 (c.617 G>A) responsible for both familial and sporadic cases of FOP which is a rare autosomal dominant disorder that causes progressive heterotopic ossification (HO) in skeletal muscle tissue.64 Two clinical features define classic FOP: (i) malformation of the great toes and (ii) progressive HO in specific spatial patterns.65 The exact physiology behind FOP is still unclear, but the ectopic bone is thought to originate from MSCs which lie dormant in soft tissues and differentiate into osteogenic cells. Three states are involved in HO, namely a stimulating event, progenitor, or MSCs and an environment that allows osteogenesis.66 In humans, FOP causes metamorphosis of skeletal muscle and soft connective tissue into a second skeleton of heterotopic bone that mimic the patterns of normal embryonic skeletal formation, important differences are the lack of inflammation in embryonic skeletal induction and the relative absence of lymphocytic inflammatory cells in early fracture healing.67 A number of seminal discoveries provided evidence of profound dysregulation of the BMP signaling pathway in cells from patients who had FOP. The classic and invariable FOP phenotype of great toe malformations and progressive heterotopic endochondral ossification suggested that the primary molecular pathology involves the BMP signaling pathway.68 The recurrent mutation in ACVR1 causes an arginine-to-histidine substitution at p.R206H in ALK2 within the GS domain in FOP, making this one of the most highly specific disease-causing and first gain-of-function mutation in the human genome.69 We know that GS domain is the molecular switches of the type 1 receptor. Destabilization/autophosphorylation due to mutation of the GS domain promotes the correct positioning of glycine-Serine-rich loop and α-helix for kinase activation, consistent with an overactive BMP signaling pathway as the underlying cause of the ectopic chondrogenesis, osteogenesis, and joint fusion seen in FOP.68 The overexpression of BMP4 in lymphocytes has been reported in FOP patients, suggesting that BMP4 is one of the downstream target genes of BMP signaling in a positive feedback loop. Recently, additional genetic mutations have been found in ACVR1 in patients with HO and other skeletal abnormalities. The positions of the mutation in the GS domain are: p.L196P, p.P198-F199delinsL, p.R202I, and p.Q207E.70,
Some of the mutations in the kinase domain have been suggested to be exposed at the interface with the GS domain and to be involved in the interaction with FKBP12/FKBP1A. The GS domain of all TGF-β/BMPRI is a critical site for binding and activation of pathway-specific Smad signaling proteins, and is a binding site of FKBP12, an inhibitory protein that prevents leaky activation of BMPRI in the absence of ligand.76 FKBP12 also recruits a Smad7-Smurf (Smad ubiquitylation regulatory factor)1 ubiquitin ligase that functions normally to regulate the abundance of the receptor at the membrane. Both leaky activation of BMP signaling and accumulation of BMP BMPRI at the cell membrane are seen in FOP cells, suggesting possible aberrant association with FKBP12 in FOP.77 The most likely possibility is that FKBP12 interactions with the GS domain become altered, leading to promiscuous ACVR1/ALK2 activity. However, exactly how the R206H mutation in ACVR1/ALK2 specifically perturbs BMP signaling in FOP remains a question but could involve dysregulation of BMPR functions or degradation/over activation of downstream signaling. This is of importance, as ACVR1 is a type 1 receptor for BMP, whose exogenous administration in soft tissue leads to ectopic bone formation.78 Thus, inhibition of mutant ACVR1 signaling in FOP patients represents an attractive therapeutic approach. Allele-specific RNA interference (ASP-RNAi), small chemical inhibitors of the kinase, including LDN-193189, Smad7 or dorsomorphin has been developed to prevent BMP signaling using a mutant ALK in FOP.79 However, there is still much to learn about the physiology and pathology at the interface of BMP and skeletal muscle, as FOP is, until recently, one of medicine's most elusive mysteries.
Smad (sma and mothers against decapentaplegic) protein: downstream effector of BMP signaling
The Smad is a group of intracellular proteins critical for transducing the signals from the cell surface to the nucleus upon binding of BMP ligands. There are three classes of Smad, namely (i) receptor regulated Smad (R-Smad), (ii) common mediator Smad (C-Smad), and (iii) antagonistic/inhibitory Smads (I-Smad).80 So far, Smad consist of three domains: an N-terminal mad-homology 1 (MH1) domain that carries nuclear localization signal and a DNA-binding region; a middle prolines, and serines/threonines-enriched linker domain that interacts with prolyl-isomerases and ubiquitin ligases; and a C-terminal MH2 domain that binds to BMPRI to mediate Smad signaling9,81. Generally, Smad forms a trimer which consists of two R-Smad and one C-Smad and acts as transcription factors that regulate the expression of certain osteoblastic genes.47 Nevertheless, the event of Smad/BMP interaction has been shown either an inhibitory or stimulatory function spatiotemporally through either Smad-dependent or Smad-independent pathways.12Figure 3 represents the proteomic organization of Smads and mechanism of interaction of Smad MH1 domain with DNA82.
Classical TGF-β/BMP signaling and its regulation by Smad
Among various Smad; R-Smad is the immediate downstream key molecule for the heterodimeric complex of two transmembrane serine/threonine kinase receptors to transduce BMP signal.83 Mechanistically, BMP ligand phosphorylate GS domain of BMPRIA and start the cascade for Smad. R-Smad has the structurally important L3 loop in the MH2 domain, which interacts with the type I receptor. R-Smad 1 and 5 become phosphorylated at the C-terminal serine residues of Ser-Ser-Val/Met-Ser sequence (SSXS motif).1 The specific phosphorylation of the R-Smad is controlled by two mechanisms of the receptor kinase complexes. First, a loop structure in the receptor kinase domain, the L4/5 loop, interacts with the L3 loop of an R-Smad protein. Second, R-Smad, such as Smad2 and Smad3, was phosphorylated by interacting with various cytoplasmic-anchored proteins beneath the cell membrane, like SARA (Smad Anchor for Receptor Activation) and cytoplasmic PML (promyelocytic leukemia) protein.84 Thus, upon BMP stimulation, BMPRII initiates the kinase activity of BMPRI leading to the phosphorylation of R-Smad in a complex cooperative process.
The phosphorylated R-Smad then disassociates from BMPRs, and forms a complex with C-Smad 4 which is imported to the nucleus. Then they interact with DNA-binding proteins or directly regulate transcriptional activity, either as monomers or in association with Smad4.6,85 In the nucleus they can regulate transcription of target genes by directly binding via their MH1 domain to specific DNA sequences depends on the formation of Smad complex with other DNA-binding proteins. The R-Smad/C-Smad heterodimer complex interacts with various transcription factors, co-activators and co-repressors to modulate gene expression.82,86 Smad complex binds the promotor regions of several BMP-responsive genes. Recently, the BMP responsive region in the Id1 promoter has been identified with two critical motifs, i.e., Smad-binding elements and GC-rich boxes (TGGCGCC).87,88 The sequence TGGCGCC (Bre7 motif) is located a short distance upstream of one or more Smad-binding elements motifs containing GTCTG. In the nucleus, Smads are also able to participate in histone modifications and chromatin remodeling.89 Runx2 in concert with R-Smad and/or Smad mediates specific signals to regulate osteoblasts formation and activate the expression of other osteogenesis specific target genes. The β-subunit of PEBP2 stabilizes Runx2 by preventing ubiquitin-dependent degradation.90 Upon activation of BMP signaling cascade, Runx2 and Smad physically interact to regulate the transcription of target genes co-operatively, and thereby induce osteoblast differentiation of MPCs. BMP does not directly induce the expression of Runx2 in mesenchymal cells, but it facilitates the expression of (Dlx5) in osteoblasts, and Dlx5 then induces expression of Runx2 in osteoprogenitor cells.91Figure 4 illustrates the classical TGF-β/BMP signaling and its regulation from BMP expression to Smad signaling including nuclear regulations.
Hh: modulator of BMP signaling
A number of findings have indicated that Hh proteins directly act on osteogenic precursor cells and osteoblasts to stimulate osteogenic differentiation.90 For example, Shh upregulates TGF-βII to inhibit hypertrophic chondrocyte differentiation during bone development. Whereas Shh/Gli2 induced BMP2 expression is responsible for osteoblast differentiation.91 Hence, during development, Shh and TGF-β/BMP pathways could directly regulate key components of each other. Both, Ihh and Shh have been shown to modulate BMP expression in cartilage and bone patterning throughout the axial, appendicular and facial skeleton.90 Shh is involved in fracture healing and bone maintenance. In the initial stages of fracture repair, the expression of Shh has been noticed in proliferating callus-forming cells in the periosteum.14 Additionally, the implantation of Shh-transduced cells increases the bone regeneration in a rabbit model of calvaria defects have also been found.92 In developing axial skeleton, sequential Shh and BMP signals are required for specification of a chondrogenic fate in presomitic tissue.93 Shh interact with the fibroblast growth factor 4 (FGF4) and BMP2 to induce the formation of osteoblast precursors and also involved in osteoblast differentiation by stimulating BMP2 promoter activity.19 The effects of Shh are mediated by transcription factor Gli2, a powerful activator of BMP2 gene expression, which is required for osteoblast differentiation.94 The expression of secreted phosphoprotein 1 (osteopontin) mediated via Gli2 and parathyroid hormone 1 receptor expressed via parathyroid hormone-like protein (PTHrP) signaling.91,95 PTHrP stimulates osteogenic cell proliferation. In developing long bones, Ihh stimulates the proliferation of chondrocytes but prevent differentiation at the growth plate in contract with BMP, and both protein acts on potential progenitor cells to promote osteoblast differentiation.40 Experimentally, Smad4 mutant mice showed in lessen of the expression of Ihh/PTHrP signaling in the growth plate.96 ALP activity also is regulated by Ihh and BMP synergistically. Moreover, upregulation of Ihh improve new bone formation at the osteogenic fronts through upregulation of BMP2.
Shh may increase mesenchymal proliferation via promotion of Msx2, and similarities are present between the Shh, Msx2, and BMP expression during neonatal craniofacial suture development.13 And Ihh induces PTCH and BMP2/4 expression, which results in intramembranous bone formation at the osteogenic fronts during intramembranous ossification.95 In sum, Ihh lies upstream of BMP2/4 signaling and may be a part of the regulatory network that controls BMP expression, thereby influencing bone formation. Discrete expression pattern of Shh versus Ihh in the developing skull indicates some possible differences in functions.97 Shh causes chondrocyte differentiation; Ihh induces differentiation of adjacent perichondrial cells into bone-forming osteoblasts.98 Our knowledge about the molecular nature of the BMP/Hh interaction is still limited; however, more sophisticated cross-talk between Hh and BMP is expected. Figure 5c,g summarizes hedgehog (Hh) signaling in osteoblasts proliferation, differentiation, as well as osteoclast formation.
Wnt/β-catenin: mediator of TGF-β/BMP signaling
The central function of Wnt/β-catenin pathway is the regulation of the stability of the β-catenin. In absence of Wnt ligand, β-catenin phosphorylated for proteasome mediated degradation that is controlled by the so-called ″β-catenin destruction complex″ which comprised of axin, adenomatous polyposis coli (APC), glycogen synthase kinase (GSK) 3 and casein kinase 1α).99 But in presence of Wnt, β-catenin accumulates and translocates to nucleus to initiate downstream gene expression. When Wnt ligand binds to its receptor frizzled (Fz) and co-receptor LRP5/6, the signal is transduced to the β-catenin destruction complex through an intracellular dishevelled protein. And thereby it leads to axin downregulation, GSK3 inactivation and β-catenin stabilization.100 More interestingly, inactivation of GSK3-β and stabilization of β-catenin reduces Smad1 ubiquitination.101 In the developing skeletal system, the Wnt/β-catenin signaling regulates the differentiation of progenitor's cells into osteoblast.18 Wnt/β-catenin pathway also help to increase the bone mass through a number of mechanisms including renewal of stem cells, stimulation of preosteoblast replication, induction of osteoblastogenesis, and inhibition of osteoblast and osteocyte apoptosis as well.102 Several Wnts (Wnt4, Wnt5a, Wnt9a, and Wnt16) are expressed in the perichondrium of developing long bones.103 However, the best-defined venue of BMP/Wnt cross-talk is in the nucleus, whereas the Smad/β-catenin/Lef (lymphocyte enhancer factor) complex regulates a group of shared target genes in a synergistic manner.
In Xenopus, BMP2/4 regulate Wnt-8 expression and cooperatively involved in the mesoderm patterning,104 whereas BMP2 downregulates Wnt7a and β-catenin in chicken embryonic mesenchymal cells in a p38-dependent manner, leading to enhanced chondrogenesis. In preosteoblastic cells, BMP2 modulates nuclear β-catenin signaling through stimulation of Wnt, LRP, and Fz receptors expression in mesenchymal cells, suggesting the interaction between Wnt and BMP2 signaling during osteoblastic differentiation.105 HesR1, TCF, LEF and Herp2 are the transcription factors specifically expressed in osteoblast cells by BMP treatment, and are also involved in Wnt signaling.106 Recent studies in differentiating osteoblasts indicate that Wnt pathway induction stabilizes β-catenin and increases TCF/LEF-dependent gene expression in parallel with β-catenin-independent complex formation between TCF4 and Runx2. Activation of either Runx2 or TCF4 co-enhances TCF and Runx2 activity increases TGF-βRI expression.107 Canonical Wnt/β-catenin pathway interacts with Runx2 as a critical mediator of BMP9-mediated osteogenic signaling.24 BMP9 stimulation of MSCs recruits β-catenin and Runx2 to the osteocalcin promoter.1 Wnt3a and BMP9 also synergized to induce ALP activity in MSCs, ectopic bone formation and matrix mineralization while BMP9-induced ALP activity inhibited by Frzb overexpression or β-catenin knockdown.24 The Wnt/β-catenin signaling pathway is activated during fracture repair. In early pluripotent mesenchymal stem cells, Wnt/β-catenin signaling needs to be precisely regulated to facilitate the differentiation of osteoblasts; by contrast, β-catenin is not needed for chondrocyte differentiation.108 Once mesenchymal stem cells are committed to the osteoblast lineage, activation of Wnt/β-catenin signaling enhances bone formation. Figure 5f highlights the cross-talk between BMP signaling and the signaling pathway of Wnt/β-catenin.
MicroRNAs act as a TGF-β/BMP signaling regulator
Activation of several signaling pathways, including TGF-β/BMP, Wnt/β-catenin, as well as transcription factors which are tightly regulated by differential expression of miRNAs.109 Osteoblast differentiation has been found tightly controlled by several regulators including miRNAs,12,110 and miRNAs can regulate expression of genes during differentiation of MSCs towards osteoblastic cells, resulting in bone formation. With regard to the regulation of bone formation, a growing number of miRNAs are found to be expressed in the developing skeletal system of metazoan. Besides the emerging role of miRNAs during embryo skeletogenesis, miRNA-dependent modulation of gene function can alter skeletal phenotypes across individuals and also within same individual over time.111
Recent evidence suggests that miRNAs might have either positive or negative regulatory function in osteoblast differentiation. For example, miR-218 facilitates differentiation of osteoblast into the final stage of forming mineralized tissue by inhibiting ERB1 (TOB1) and sclerostin (SOST).112 And miR-2861 promotes BMP2 induced ST2 osteoblast differentiation by repressing histone deacetylase 5 (HDAC5) expressions.113 BMP2 stimulates acetylation of Runx2 protein by p300 resulting inhibition of Smurf1-mediated degradation of Runx2 and increases the Runx2 transactivation activity. HDAC5 proteins can deacetylate Runx2 protein, and thus Runx2 protein is vulnerable to Smurf-mediated degradation. Homeobox A2 (Hoxa2), another repressor of Runx2 is targeted by miR-3960 leading to enhanced osteogenesis in BMP2 induced ST2 stromal cells.114 miR-27 promotes osteoblast differentiation by inhibiting APC gene expression, consequently activates Wnt/β-catenin signaling through accumulation non-phosphorylated β-catenin that translocates from cytoplasm to nucleus and then associates with TCF/LEF to induce target gene expression.19 Kapinas et al.115 and Zhang et al.116 have shown that miR-29a and miR-335-5p promotes osteoblast differentiation by downregulating the inhibitors of canonical Wnt/β-catenin signaling such as dickkopf-related protein 1 (Dkk1) and secreted Fz-related protein (sFRP2). MiR-208 enhances BMP2 induced pre-osteoblast differentiation by targeting the transactivation of osteopontin, Runx2, type I procollagen and PTHrP.117 Another miRNA, namely miR-138, regulates the osteogenic differentiation of human MSCs in vivo.118 Microenvironment determines the role of miRNAs for regulation of osteoblast differentiation. Liu et al.119 reported that miR-17 has a role for stimulation or inhibition of osteoblast differentiation that depends on the microenvironment. Table 2120,
On the other hand, there are several miRNAs acting as negative regulators of osteoblast differentiation and mineralization in mesenchymal progenitor cells and bone marrow stromal cell (BMSCs). Runx2 gene expression is inhibited by miR-204 and miR-211.139 MiR-355 is found to be directly targeting Runx2 mRNA, and overexpression of miR-335 in hMSCs inhibits their proliferation and migration, as well as their osteogenic potential.140 TGF-β/BMP signaling pathways and their components in osteoblasts can be inhibited by miRNAs such as hsa-miR-31, hsa-miR-106a, hsa-miR-148a, and hsa-miR-424 and their target genes were found to be as Runx2, Cbfb, BMPs. mir125b and miR26a inhibits osteoblast differentiation by reducing the cell proliferation rate and interference on Smad1 translation, respectively. Yin et al.141 have shown that miR-155 targets the BMP signaling cascade, including Smad1, Smad5, CEBPB, and Runx2. The expression of miR-141 and miR-200a thought to downregulate BMP2-induced pre-osteoblast differentiation through the translational repression of Dlx. Recently, it has been reported that Smads participate in miRNA biogenesis.142 However, these physiological regulation brought out by miRNAs could be important for balancing bone formation and bone resorption processes.
MAPKs and PI3K/AKT pathway and TGF-β/BMP signaling
MAPKs are serine/threonine/tyrosine-specific protein kinases belonging to CMGC (CDK/MAPK/GSK3/CLK) kinase group. Multiple extracellular stimuli can kick off the serial phosphorylation from MAP kinase kinase kinase (MAPKKK) to MAP Kinase.143 However, the linker region of Smad protein is the platform for integrating MAPK/RTK signal with the TGF-β/BMP pathway. The serine rich linker region is loosely organized and structurally flexible, making it accessible for kinases. This region is also rich in threonine and proline residues, favoring phosphorylation by proline directed kinases such as MAPKs.144 Multiple Ser/Thr residues in Smad1 linker can be sequentially phosphorylated by Erk and GSK3-β, creating a docking site for the Smad1/5 specific E3 ubiquitin ligase causes not only ubiquitination and degradation of the Smad but also occludes their interaction with the nuclear pore complex, thereby preventing Smad nuclear translocation and attenuation of the BMP signal.145 But, Wnt/β-catenin signaling, which is known to inactivate GSK3-β, reduces Smad1 ubiquitination and stabilizes the protein.101 More recently, three residues in the linker region of Smad3 (Thr178, Ser203, and Ser207) were identified as Erk1/2 phosphorylation sites both in vitro and in vivo. Erk-mediated phosphorylation of these sites inhibits Smad3 transcriptional activity but does not prevent Smad3 from entering the nucleus146 suggesting the existence of an unknown mechanism for Smad3 inhibition by the linker phosphorylation. Figure 5a shows the role of MAPKs/TAK1 signaling in MSCs differentiation and bone formation.
Studies in Xenopus laevis have indicated that BMPR activation leads to MAPK activation via X-linked inhibitor of apoptosis, TGF-β-activated kinase and Tak-binding protein.147 BMP7 has been shown to induce p38 MAPK through integrin-linked kinases.148 Furthermore, the ligands BMP2 and BMP4 can activate p38 and ERK MAPK, but not JNK. In early mouse embryos, limb-bud outgrowth is promoted by Shh and FGF, whereas termination of this growth requires BMP-mediated inhibition of FGF. On the other hand, defects in cartilage development thought partly because of an elevation in FGF signaling that suppresses chondrogenesis. As a result, BMP function can be suppressed by several signals that activate RTK/MAPK, including EGF, FGF and insulin like growth factor (IGF).149 In addition to R-Smad, MAPKs also phosphorylate and regulate the C-Smad, Smad4, and the inhibitory Smad implicated in the transcriptional regulation of Smad7, therefore indirectly regulating TGF-β signaling.
Turning the focus to how the synergy between the TGF-β and HER2/Ras/MAPK pathways often leads to the secretion of growth factors and cytokines, including TGF-β itself, which in turn promote epithelial-to-mesenchymal transition (EMT). MEK/Erk has been reported to positively regulate Smad3 gene transcription in epithelial cells. Besides MAPK, it has been reported that BMP2 can activate the protein kinase C and PI3K pathways.150 PI3K-activated, plasma membrane-anchored Akt can physically sequester Smad3 and block its nuclear translocation in a kinase-independent manner, without affecting the C-terminal phosphorylation (activation) of Smad3. However, this mechanism may be cell type-specific and dependent on the stoichiometry between the Akt and Smad3 proteins, as Akt can also facilitate Smad3 function, and PI3K/Akt has been shown to be required for the nuclear accumulation of BMP-activated Smad1.150 A third way for PI3K/Akt to debilitate Smad3 is through the inactivation of certain nuclear factors that are necessary for Smad3 function. The regulation of MAPKs by BMP stimulation151 represents an important mechanism for non-Smad TGF-β signaling but the exact mechanism of activation is not so clear.
Notch signaling pathway and TGF-β/BMP signaling
In the nucleus, Notch ICD (intracellular domain) binds specific transcription factors and recruits transcriptional co-activators to induce the expression of bHLH family of genes.152 Notch pathway is active in the early stages of osteoblast differentiation, also by acting on Runx2-dependent osteogenic gene expression.153 Notch could also regulate osteoclastogenesis, through the up-regulation of RANKL and OPG genes, suggesting that the functional cross-talk between osteoblasts and osteoclasts might be also mediated by Notch signaling.152 Notch signaling is also involved in skeletal patterning and somitogenesis.147 Notch1 null mouse embryos exhibits significantly delayed and disorganized somitogenesis.154 Finally, Notch signaling is believed to act also in chondrogenic differentiation, although its exact role and its temporal effects during chondro/osteoblastogenesis are still unclear. Figure 5d summarizes Notch signaling in MSC differentiation. More lately, a sign of the existence of molecular crosstalk between endothelial and osteoblastic cells, angiogenesis in the skeletal system and osteogenesis identified to be coupled, in which Notch signaling is involved. Ramasamy team found that endothelial-cell-specific and inducible genetic disruption of Notch signaling in mice not only impaired bone vessel morphology and growth, but also led to reduced osteogenesis, shortening of long bones, chondrocyte defects, loss of trabeculae and decreased bone mass.155
Other modulators of TGF-β/BMP signaling
There are a number of some other modulators of BMP signaling namely Glucocorticoids (GCs), retinoic acid, laminin, and nitric oxide (NO), nuclear factor kappa B (NF-κB) and hypoxia-inducible factor (HIF). GCs mediate their action on osteoblasts through BMP.156 Retinoic acid is a possible modulator of BMP expression.157 Cell adhesion molecules such as laminin, neural cell adhesion molecules (N-CAM), and integrin are known to interact with BMP and have been localized at the areas of initial mesenchymal condensation158,159. TGF-β activities are also interconnected with mTOR and NO signaling. Lamouille et al.160 reported that mTORC2 is a novel and crucial mediator in TGF-β signaling. Upon TGF-β activation, Smad controls the expression and activities of Snai1 transcription factor involved in EMT, which represses transcription of the E-cadherin gene.161 TGF-β-induced Akt (S473) phosphorylation is dependent on mTORC2 activity.160 Akt blocks phospho-activation of Smad3 by an Akt kinase-dependent mechanism through mTOR and also blocks TGF-β signals downstream of Smad3 activation, but through a mechanism that does not require the kinase activity of Akt or mTOR.162 Attenuated NO and small interfering RNA based silencing of secreted modular calcium-binding protein 1 (SMOC-1) decreased TGF-β, reduced Smad binding to DNA, and decreased mRNA expression of genes regulated by TGF-β.163 In particular, NO reduced the metabolic life of ectopically expressed Smad2 and enhanced its ubiquitination. The endothelial NO/cGMP/PKG pathway interferes with TGF-β/Smad2 signaling by directing the proteasomal degradation of activated Smad.164 Ubiquitously expressed NF-κB transcription factors and HIF also recently found to be involved in TGF-β/BMP signaling. HIF-2α regulates collagen X, MMP-13 and VEGF expression.165 In addition, further possible transcription targets of HIF-2α relates to endochondral ossification have been identified, namely Runx2 and Ihh proteins. Concerning activated NF-κB that influences the amount and remodeling of ECM proteins, and shows indirect positive effects on downstream regulators including HIF-2α, β-catenin and Runx2.166 Concerning bone, NF-κB proteins induce the synthesis of BMP2. In normal conditions RANK/RANKL interaction is regulated by the OPG decoy receptor that by blocking RANKL promotes bone formation.167 Contrary, the interaction between the IL-1/TNF induced RANKL and RANK surface receptor activates NF-κB turn out PGE2 and thereby bone resorption.168
Sometimes BMP cannot generate enough clinical response to bone regeneration treatment. A possible reason might be that inflammatory cytokines inhibit the bone formation and osteoblast differentiation induced by BMP. For example, one inflammatory cytokine, tumor necrosis factor (TNF) a, inhibits osteoblast differentiation in multiple models, including fetal calvaria, bone marrow stromal cells. Together, there seem to be antagonistic effects between BMP/Smad and NF-κB signaling systems. Although the involvement of NF-κB in osteoclastogenesis has been well investigated, little is currently known about the possible involvement of NF-κB in osteoblast differentiation, especially the cross-talk between BMP/Smad and NF-κB signaling in osteoblast differentiation.169 NF-κB activation inhibits osteoblast differentiation at least in part by interfering with Smad signaling. NF-κB is involved in the inhibitory effect of TNFa on BMP2-induced osteoblast differentiation. Eliseev et al. reported that Inhibition of NF-κB represses BMP/Smad signaling and BMP2-induced differentiation through Smad7 leads to the induction of osteoblast differentiation.170 Inhibition of NF-κB could suppress the antiosteoblastogenic activity of TNFa and relieve the suppression of BMP2-induced Runx2 expression mediated by TNFa.171 Hence, TNFa on BMP signaling in osteoblast differentiation is closely linked to the NF-κB pathway. NF-κB activation inhibits osteoblast differentiation at least in part by interfering with Smad signaling. An intracellular balance of signal intensities between NF-κB and BMP/Smad is therefore crucial for osteoblast differentiation.
TGF-β/BMP signaling are finely tuned by a complex network of signaling molecules and biophysical factors, shedding light on some extremely relevant modulators namely FGF, HIF, GC, NO, PI3K, Akt, NF-κB, and mTOR. On a final note it could be remarked that given the considerable variation in the use of receptors, abundance of transcriptional co-regulators and links to other major pathways TGF-β/BMP signaling creates more of a signaling network than a straight linear regulatory path.
Bone microenvironment and bone formation
MSCs derived osteoblasts, osteoclasts, mineralized bone matrix, and some sort of fixed osteocytes within bone are the major components of bone microenvironment.40 Formation of new bone and the regulation of osteoclastogenesis are two main functions of the osteoblasts.172 Osteogenesis comprises a sequential cascade with three critical phases: migration and mitosis of mesenchymal cells, differentiation of mesenchymal cells into chondroblasts, cartilage formation and, finally, substitution of cartilage by bone. During osteogenesis some pluripotent mesenchymal cell lines, bone marrow cells, and osteoblast precursors are the target of BMP. The pluripotent stem cells undergo successive stages of differentiation with a decreasing proliferation potential, giving rise to committed pre-osteoblasts. Subsequently, pre-osteoblasts differentiate into mature osteoblasts that deposit the necessary components to form bone matrix, followed by mineralization.173 Eventually, mature, mineralizing osteoblasts become embedded in the newly secreted bone matrix and undergo terminal differentiation to form osteocytes. Figure 6 shows the evolution of osteoblasts and osteoclasts in bone formation.
Osteoclasts are large multinucleated terminally differentiated cells with a unique ability for bone resorption. They are derived from hematopoietic stem cells. The cells undergo proliferation in response to M-CSF. The precursor cells flaunt receptor and activator of nuclear factor κB (RANK) on the surface, while the ligand RANKL is expressed by the bone marrow stromal cells and osteoblasts.174 Binding of the ligand to the receptor commits the precursor cells to differentiate into osteoclast lineage. The same interaction is also critical for osteoclast formation and can also promote osteoclast activity, since RANK is also present on the surface of terminally differentiated osteoclasts. Osteoprotegerin (OPG) is a soluble decoy receptor and a competitor of RANKL in its binding with RANK and thus can inhibit osteoclastogenesis.175 Therefore, the balance of RANKL and OPG is critical for osteoclast formation and activity. Cathepsin K and MMPs catalyze the degradation of bone matrix, dissolution of the bone mineral, and resorption of the bone.176 A recent study on OVX mice treated with exogenous PDGF-BB or inhibition of cathepsin K showed that, osteogenesis during bone modeling and remodeling is coupled with angiogenesis.177
In general active BMP dimers act as a chemo attractant for undifferentiated mesenchymal stem cells, causing them to move into the area of bone defect or injury and over expression of activated BMPRs ultimately increased production of bone matrix proteins.178 Receptor bounded BMP is displayed on the stem cells' surface. The Smad complex translocates into the nucleus and interacts with several transcription factors such as Runx2/Cbfα1, Osx, Dlx5, and Msx2. The expression of osteoblast genes such as alkaline phosphatase, osteopontin (Opn), and bone sialoprotein are also found to be important for osteogenesis. For instance, activated Smad relieves repression of the osteopontin gene by binding to the repressor Hoxc8 and removing it from the Opn promoter. Hoxc8, a repressor in the BMP pathway binds to the consensus site of the Opn gene and silences its transcription. The Smad1/4 complex, in response to BMP, interacts with its DNA-binding domain and dislodges it from the OPN promotor element, thereby initiating gene transcription and inducing osteoblast differentiation.179 Recently in mouse model it is shown that notch signaling pathway enhances BMP2 responsiveness of Msx2 gene to induce osteogenic differentiation.180 In addition to others, menin is required for BMP-induced osteoblast differentiation.181 Nonetheless, BMP2 recruit's mesenchymal cells surrounding the initial cartilage condensations into chondrogenic fate, BMP6 accelerates hypertrophic and terminal chondrocyte differentiation and mineral accretion while GDF5 is required for joint formation.179 In multipotential mesenchymal cells isolated from human bone marrow, BMP2 and 9 promote chondrogenic differentiation, possibly through activation of Sox-9, a chondrogenic-related transcription factor.182 It is frequently assumed that BMP-activated Smads directly regulate the expression of Runx2 and members of the BMP family of growth/differentiation factors. But the interactions between Runx2 and BMP-specific Smads (Smad1 and 5) suggest their intrinsic cooperation in mediating BMP functions. Although the mechanism is not fully clear. It has been proposed that BMPs can induce both undifferentiated stem cells and more differentiated multipotent cells along chondrogenic or osteogenic pathways. Figure 5 summarizes the up to date integration of different signaling pathways for lineage specific differentiation for MSCs in an exquisitely coordinated manner. Evidently behind its outward rigidity, bone is a highly dynamic organ where homeostasis is tightly controlled and largely dependent upon cellular communication between osteoclasts and osteoblasts. This tight coupling between bone resorption and bone formation is essential for the correct function and maintenance of the skeletal system, repairing microscopic skeletal damage, and replacing aged bone.
To date, a large number of regulatory factors and targets have been identified for the TGF-β/BMP signaling pathway, leading to a complex interactive network that investigators are still trying to unravel. High throughput screening, modeling, structural biology, and bioinformatics will greatly assist the elucidation of the complex mechanisms regarding TGF-β/BMP function and regulation to close the gaps. Knowing how these pathways affect the transformation of osteoblasts to osteocytes in different developmental stages help to exploit the therapy of bone related diseases. However, BMP was named for the ability to form bone, and the name implies that they are capable of inducing the osteogenic differentiation of non-bone cells. This is true for most of the best-studied BMPs, but not all. Furthermore, the name does not acknowledge the breadth of BMP actions, and it is still not clear what controls and how a specific BMP will be osteoinductive. And the number of players in osteogenesis is much wider than that described in this review, and much remains to be understood about the complexity of BMP signaling.
We are grateful to Prof. Pritinder Kaur (Epithelial Stem Cell Biology Laboratory, Peter MacCallum Cancer Centre, Australia) for her helpful suggestion and assistance. We also gratefully thank S.M. Abdul Awal, Signal Transduction Group, Cambridge University, UK, for fruitful discussions.