Osteogenic differentiation of mesenchymal stem cells in multiple myeloma: Identification of potential therapeutic targets

Role of the transcription factor Runx2 

Under physiological conditions, the osteogenic differentiation of mesenchymal cells is tightly regulated either by system hormones, such as parathyroid hormone (PTH), estrogens, and glucocorticoids or by local growth factors, including the bone morphogenetic protein (BMP) family, transforming growth factor–β, and fibroblast growth factor 2 [5]. Moreover, these factors activate specific intracellular signal pathways that modify the expression and activity of several transcription factors in mesenchymal and osteoprogenitor cells, which result in osteoblastic differentiation 5, 6.

In the last several years, most of these transcription factors have been identified. Runt-related transcription factor 2 (Runx2), also named Cbfa1 or AML3, is the major transcription factor regulating osteoblast commitment and osteogenic differentiation of mesenchymal cells 6, 7, 8. Studies in mice lacking Runx2 indicated that the expression of Runx2 is critical for mesenchymal cell differentiation toward the osteoblast lineage. Runx2-deficient mice (Runx2–/–) completely lack osteoblasts and bone formation, demonstrating a pivotal role for this factor in osteoblastogenesis 6, 7, 8. However, Runx2 overexpression also impairs bone formation in mice [9], indicating that, depending on the stage of osteoblast differentiation, Runx2 could have different effects on the bone formation process. Runx2-overexpressing mice also show enhanced bone resorption, possibly through increased expression of the osteoclast stimulating factor receptor activator of nuclear factor κB ligand by osteoprogenitor cells [10]. Human osteoblast differentiation is primarily associated with increased Runx2 activity without changes in messenger RNA or protein expression 11, 12. Activation of Runx2 in human bone marrow (BM) mesenchymal cells induces expression of the osteoblast-related markers collagen I, alkaline phosphatase, and osteocalcin during the early stages of the osteoblast maturation [11].

Both expression and activity of Runx2 are tightly regulated by other transcription factors as well as protein–DNA or protein–protein interactions. Runx2 itself is regulated by phosphorylation by the extracellular signal-regulated kinase/mitogen-activated protein kinase pathway [13]. Hey-1, Hoxa2, Stat1, and Sox9 interact with Runx2 and inhibit its expression and/or transcriptional activity and thus are negative regulators of osteoblast differentiation. Runx2 activity is also positively regulated by transcription factors such as TAZ, Hoxa10, or BAPX-1 5, 6, 8. In addition, multiple signaling pathways converge to interact with Runx2 to regulate osteoblast differentiation, including binding with activator protein 1 and activating transcription factor 4 that with Runx2 regulate osteocalcin and Osterix (Osx) 5, 6, 8. Because Osx is a zinc finger transcription factor that is important for osteoblast differentiation. Osx-deficient mice lack osteoblast formation. Osx acts as a downstream of Runx2 5, 6. In mouse systems, osteogenic factors stimulate osteogenesis through regulation of these transcription factors. BMP-2 promotes Runx2 and Osx expression in murine osteoprogenitor and osteoblastic cells, and transforming growth factor–β and fibroblast growth factor 2 may enhance osteoblast differentiation by increasing Runx2 expression and activity 5, 6.

Wnt signaling 

Several studies have demonstrated that Wnt signaling plays a critical role in the regulation of osteoblast formation 14, 15, 16, 17, 18. Canonical Wnt signaling pathway is activated by Wnt 1/3a that triggers the phosphorylation of GSK3/Axin complex, leading to the stabilization and nuclear translocation of the active dephosphorylated (dephospho) β-catenin, which in turn activates the lymphoid enhancer factor-1/T-cell factor transcription system 14, 15, 16, 17, 18. In murine systems activation of canonical Wnt signaling in osteoblast progenitors induces osteogenic differentiation 17, 18. In particular, BMP-2 and other osteogenic molecules induce osteoblastic differentiation of murine mesenchymal stem cells by stimulating Wnt signaling through modulation of Wnt stimulators and/or inhibitors [18]. Several molecules negatively regulate canonical Wnt signaling by inducing phosphorylation and subsequent degradation of β-catenin inhibiting osteoblast differentiation in murine osteoprogenitor cells. Dickkopfs (DKKs), including DKK-1 [19], the secreted frizzled related proteins (sFRPs), such as sFRP-1, -2, -3, -4, and Wnt inhibitory factor-1 are the major soluble Wnt inhibitors present in murine osteoblasts, which block early osteoblast formation and induce the death of immature cells 20, 21. The mechanism by which these inhibitors block the canonical Wnt signal pathway is different. sFRP family are able to bind Wnt soluble agonists or the frizzled (FZ) family of the Wnt receptor, whereas DKKs prevent the activation of the Wnt signaling pathway by binding to LRP5/6 and Kremen abolishing Wnt/LRP6 signaling 20, 21, 22.

In vivo models support the role of canonical Wnt signaling in the regulation of bone formation 23, 24, 25. Inactivating mutations of the LRP5 Wnt coreceptor cause osteoporosis [23], indicating that Wnt-mediated signaling via LRP5 affects bone accrual during growth and is important for the establishment of peak bone mass. Constitutive activating LRP5 mutations impair the action of normal antagonists of the Wnt pathway such as DKK-1 and increase Wnt signaling, which result in high bone density 24, 25. In addition a targeted destruction of LRP5 in mice induce a low bone mass phenotype due to a decrease of osteoblast proliferation and function independently to Runx2 as demonstrated by a normal Runx2 expression in Lrp5–/– mice [26]. On the other hand, it has been reported that canonical Wnt signaling promotes osteogenesis by directly stimulating Runx2 gene expression [27]. In sFRP-1 null mice, which exhibits activated Wnt signaling and high bone mass phenotype, there is a significant increase of both T-cell factor and Runx2 expression [27]. Similarly, overexpression of Wnt10b in mice protect from bone loss of estrogen deficiency and shifts mesenchymal stem cells toward osteoblastic lineage by induction of Runx2 [28]. All these results demonstrate that canonical Wnt signaling plays an important role in bone formation both independently and dependent to Runx2 expression.

Despite the consistent findings between human genetic studies and mouse studies, which indicate that activation of canonical Wnt signaling stimulate bone formation, recent in vitro data obtained with human mesenchymal cells indicate that canonical Wnt activation by Wnt3a in human BM mesenchymal cells suppresses osteogenic differentiation rather than stimulates osteoblastogenesis as observed in murine osteoprogenitor cells 29, 30, 31, 32. These results suggest that Wnt signaling is required to maintain human mesenchymal cells in an undifferentiated state. We can suppose that the effect of canonical Wnt signaling on osteogenesis in human mesenchymal may depend to the level of Wnt activity given that hyperactivation of Wnt signaling by overexpressing LRP5 [24] can enhance osteogenesis, whereas exogenous levels of Wnt3a inhibit osteoblast differentiation 29, 30. Alternatively, the effect of canonical Wnt signaling may depend on the stage of differentiation of the cells. Accordingly, it has been reported that canonical Wnt signaling antagonizes the terminal steps of osteogenic differentiation as demonstrated by the evidence that mice lacking the Wnt inhibitor DKK-2 are osteopenic [33]. Consistently an increased expression of Wnt antagonists during late osteoblast differentiation has been shown 34, 35. Finally, we could suppose a different behavior between murine and human osteoprogenitor cells.

Together with canonical Wnt signaling, a noncanonical Wnt pathway independent to the activation of β-catenin has been extensively demonstrated (reviewed in 14, 16, 36, 37). Noncanonical Wnt signals are transduced through FZD receptor and Ror2 coreceptor to several cascades involving disheveled or Ca++ dependent pathways. Rho family small GTPase (RhoA, Rac) and JNK are downstream effectors of disheveled. Nemo-like kinase and the nuclear factor of activated T cells are Ca++ effectors of noncanonical pathways 36, 37. Wnt-4, -5, and-11 proteins have been identified as specific activator of noncanonical Wnt pathway 14, 16, 36, 37. Recent evidences suggest that noncanonical Wnt pathway activation by Wnt5a blunted the inhibitory effect of Wnt3a on osteogenic differentiation of human mesenchymal cells and stimulate osteoblast differentiation 38, 39. Similarly, noncanonical Wnt-4 signaling enhances bone regeneration of mesenchymal stem cells in vivo [38]. The pro-osteogenic effect of Wnt5a could be mediated by activation and homodimerization of the Ror2 receptor in mesenchymal cells, whereas Wnt3a has not effect on Ror2 activation and homodimerization 39, 40. It has been consistently shown that Ror2 overexpression in human mesenchymal stem cells induces expression of the osteogenic transcription factors Osx and Runx2 and induces osteogenic differentiation 40, 41, 42. Finally, it has been recently demonstrated that noncanonical Wnt pathway through Nemo-like kinase represses peroxisome proliferation activated receptor–γ transactivation and induces Runx2 expression promoting osteoblastogenesis in BM mesenchymal stem cells suggesting a potential relationship between noncanonical pathway and Runx2 activity [43].

All these evidences suggest that the noncanonical Wnt signaling pathway is involved in mesenchymal cells osteogenic differentiation.

Effect ofMM cells onRunx2 activity in osteoprogenitor cells 

Impairment of osteoblast and bone formation in MM is mainly due to the block of the osteogenic differentiation process of mesenchymal cells induced by MM cells. Coculture of human MM cells with osteoprogenitor cells inhibited osteoblast differentiation in long-term BM cultures. MM cells reduced early osteoblast precursors, fibroblast colony-forming units (CFU-F), and the more differentiated osteoblast precursor, the colony-forming osteoblast units (CFU-OB), as well as expression of osteoblastic differentiation markers, alkaline phosphatase, osteocalcin, and collagen I [44]. These in vitro observations have been confirmed in vivo in MM patients [45], although this is controversial 46, 47.

The inhibitory effect of MM cells on osteoblast differentiation appears to be mediated by the capacity of MM cells to inhibit Runx2 activity in human mesenchymal stem and osteoprogenitor cells [44]. It has been consistently reported that the number of mesenchymal and osteoblastic Runx2-positive cells, were lower in patients with osteolytic lesions as compared to patients without skeletal involvement [44]. Suppression of Runx2 activity by MM cells is mediated, at least in part, by the cell-to-cell contact between MM and osteoprogenitor cells. This cell-to-cell contact involves interactions between very late antigen 4 (VLA-4) on MM cells and vascular cell adhesion molecule 1 (VCAM-1) on osteoblast progenitors, as demonstrated by the capacity of a neutralizing anti–VLA-4 antibody to reduce the inhibitory effects of MM cells on Runx2/Cbfa1 activity [44]. The role of cell-to-cell contact via VLA-4/VCAM-1 interaction in the development of bone lesions by osteoclast activation and osteoblast inhibition in MM has been recently demonstrated using in vivo mouse models [48]. When the human MM cell line, JJN3, which strongly expresses VLA-4, is implanted in irradiated severe combined immunodeficient (SCID) mice, the mice developed lytic lesions and marked osteoblastopenia with a significant reduction of bone formation [4]. In addition to VLA-4/VCAM-1, other adhesion molecules appear to be involved in the inhibition of osteoblastogenesis by human MM cells. For example, neural cell adhesion molecule to neural cell adhesion molecule interactions between MM cells and stromal/osteoblastic cells decrease bone matrix production by osteoblastic cells, and may contribute to development of bone lesions in MM patients [49]. Soluble factors contribute to the inhibitory effects of MM cells on osteoblast differentiation by mesenchymal stem cells and Runx2/Cbfa1 activity [44]. Interleukin-7 (IL-7) decreases Runx2/Cbfa1 promoter activity and the expression of osteoblast markers in osteoblastic cells [43]. Moreover, IL-7 can inhibit bone formation in vivo in mice [50], as well as CFU-F and CFU-OB formation in human BM cultures and reduces Runx2 activity in human osteoprogenitor cells [44]. The potential involvement of IL-7 in MM has been supported by the demonstration of higher IL-7 plasma levels in MM patients compared to normal subjects [51] and by the capacity of blocking antibodies to IL-7 to partially blunt the inhibitory effects of MM cells on osteoblast differentiation [44]. These studies suggest that MM cells block Runx2 activity and osteoblast differentiation either by cell-to-cell contact or by secreting IL-7, which leads to a reduction in the number of more differentiated osteoblastic cells. Other soluble factors may also be involved in this process. The hepatocyte growth factor (HGF) is produced by MM cells and its high levels in BM of MM correlated with those of alkaline phosphatase [52]. HGF inhibits in vitro osteoblastogenesis induced by BMP-2 and BMP-induced expression of alkaline phosphatase in both human and murine mesenchymal cells. Moreover, HGF treatment reduced the expression of transcription factors Runx2 and Osx as well as Smad [52].

TAZ, a Runx2/Cbfa1 transcriptional coactivator, has been recently shown to modulate the osteogenic potential of the human mesenchymal stem cells 8, 53. TAZ expression is lower in MM patients as compared to healthy donors [54]. The repressed osteogenesis and TAZ expression were both partially restored by neutralization of tumor necrosis factor–α [54].

The inhibition of Runx2 in BM osteoprogenitor cells by MM cells through the previously mentioned mechanisms suggest that Runx2 could be a target to counterbalance the inhibition of osteogenic differentiation in MM. Modulation of Runx2 activity has been shown with anabolic agents, such as PTH amino-terminal peptide 1–34 55, 56. Moreover, in vivo mouse models have shown that Runx2 gene transfer enhances osteogenic activity of BM mesenchymal cells [57]. Clearly, further studies will be necessary to evaluate the feasibility of this approach in MM.

Role of IL-3 in the inhibition of osteoblasts formation by MM cells 

Interleukin-3 (IL-3) has been reported as a potential osteoblast inhibitor in MM patients [58]. In both murine and human system, IL-3 indirectly inhibited osteoblast formation in a dose-dependent manner, without affecting cell growth at concentrations comparable to those seen in BM plasma from patients with MM. IL-3 levels in BM plasma from patients with MM were increased in approximately 70% of patients compared to normal controls or patients with monoclonal gammopathy of undetermined significance (MGUS) patients [59]. IL-3 is also produced by T lymphocytes in the MM bone microenvironment [60]. Importantly, BM plasma from patients with MM with high levels of IL-3 inhibited osteoblast formation in human cultures, and this inhibition was partially reversed by addition of a neutralizing antibody to human IL-3 [58]. The inhibitory effect of IL-3 was increased in the presence of tumor necrosis factor–α, a cytokine induced in the MM marrow microenvironment. The effect of IL-3 is indirect and mediated by CD45+/CD11b+ monocyte/macrophages in both human and mouse primary culture systems. IL-3 increased the number of CD45+ hematopoietic cells in stromal cell cultures, and depletion of the CD45+ cells abolished the inhibitory effects of IL-3 on osteoblasts. Importantly, reconstituted CD45+ depleted cultures with CD45+ cells, restored the capacity of IL-3 to inhibit osteoblast differentiation. On the other hand, IL-3 had no direct effect on osteoblast progenitors or on Runx2 expression and activity in both murine and human BM cells [58].

Role of Wnt inhibitors and Wnt signaling in MM-induced osteoblast suppression 

The potential involvement of inhibitors of Wnt signaling in the suppression of osteoblast formation and function in MM has been investigated. Primary CD138+MM cells overexpress the Wnt inhibitors DKK-1 as compared to plasma cells from MGUS patients and normal plasma cells [61]. Further, using gene expression profiling, a tight correlation between DKK-1 expression by MM cells and the occurrence of focal lytic bone lesions in MM patients has been reported. High DKK-1 levels in BM and peripheral sera in MM patients correlated with the presence of bone lesions [62]. Interestingly, patients with advanced disease, as well as human MM cell lines did not express DKK-1 61, 62, suggesting that DKK-1 may mediate bone destruction in the early phases of disease. MM cells also produce other Wnt inhibitors, including sFRP-3/FRZB. FRZB is highly expressed by CD138+MM cells from patients as compared to MGUS patients, and BM plasma levels are higher in MM patients with bone lesions as compared to those without skeletal involvement [62]. sFRP-2 has been also reported to be produced by some human MM cell lines and by patients with advanced MM bone disease, and can inhibit osteoblast differentiation [63].

The mechanism by which DKK-1 and the other Wnt inhibitors produced by MM cells is related to bone destruction is not completely understood. Neutralizing anti–DKK-1 antibody can block the inhibitory effect of BM plasma of MM patients on BMP-2–induced alkaline phosphatase expression and osteoblast formation by a murine mesenchymal cell line [61], but failed to block the inhibitory effects of MM cells on human BM osteoblast formation [44]. In addition, only high concentrations of DKK-1 are able to inhibit CFU-F and CFU-OB formation and to block β-catenin signaling in human BM osteoprogenitor cells [44]. Recently, we reported that MM cells failed to block canonical Wnt signaling in human BM osteoblast progenitors but inhibited this pathway in murine systems suggesting that blockade of the canonical Wnt pathway does not occur in MM patients [62].

Other mechanisms could be involved in DDK-1–mediated bone destruction in MM. For example, a link between cell adhesion and the Wnt pathway was recently reported. Wnt inhibitors, such as DKK-1, are triggered by cell contact and modulate adhesion of leukemia cells to osteoblasts [64]. Possibly, DKK-1 production by MM cells could be involved in the adhesion of stromal and MM cells, which is critical for osteoclast activation and MM-induced Runx2 inhibition and osteoblast suppression. Further, crosstalk between MM cells and the microenvironment can stimulate both DKK-1 and IL-6 production in human BM cultures [65]. The capacity of DKK-1 to regulate expression of the osteoclast inhibitory factor, OPG, has been also reported in murine osteoblasts [66].

Studies in the SCID-human mice model of MM have shown that anti–DKK-1 increases bone mineral density and the number of osteocalcin positive osteoblasts compared to control mice [67]. Interestingly, a reduction of the number of osteoclastic cells was also observed, suggesting that Wnt signaling could be involved in the regulation of bone resorption [67]. Consistently, it has been reported that activation of the Wnt3a signaling pathway in the bone microenvironment is able to block the development of bone lesions and the growth of MM cells in murine MM models 68, 69. These observations are in line with those showing that canonical Wnt pathway stimulates bone formation in mice. Interestingly, both an anti–DKK-1 antibody and Wnt canonical activation by Wnt3a or by litium 68, 69 are able to blunt the inhibitory effect of MM cells on osteoblast formation independently of the production of DKK-1 by MM cells. These results suggest that the bone microenvironment rather than MM cells is the target of anti–DKK-1 antibody therapy.

Proteasome inhibitors 

The ubiquitin-proteasome pathway is the major cellular degradative system for several proteins involved in cell proliferation and survival in MM cells. Recently, it has been demonstrated that this pathway may regulate osteoblast differentiation and bone formation in vitro and in vivo in mice [70]. The ubiquitin-proteasome pathway can modulate the expression of BMP-2, which can induce osteoblast differentiation through Wnt signaling and regulates the proteolytic degradation of the osteoblast transcription factor Runx2/Cbfa1 [71]. Different proteasome inhibitors that bind the catalytic β subunits of the 20S proteasome and block its activity are able to stimulate bone formation in neonatal murine calvarial bones [70]. A strong correlation between the capacity of these compounds to inhibit proteasomal activity in osteoblasts and their bone forming activity was also demonstrated [70]. Consistent with these in vitro observations, the administration of the natural proteasome inhibitors PS1 and epoximicin to mice, increased bone volume and bone formation rate >70% after 5 days, indicating a potent stimulatory effect of these drugs on osteoblastic cells [70]. These results are strongly supported by the in vivo observations obtained in MM patients treated with the proteosome inhibitor bortezomib. Increases of total alkaline phosphatase and the isoenzyme, bone specific alkaline phosphatase, have been reported in MM patients that respond to treatment with bortezomib 72, 73, 74. Others have reported a significant increase in bone remodeling markers in all patients treated with bortezomib 75, 76. Recent data indicate that bortezomib at low concentration (2–4nM) induces osteoblast differentiation from human mesenchymal cells in vitro and increases Runx2 activity without stabilization of β-catenin [77]. This results in the expression of osteoblast markers such as collagen I and osteocalcin and in the increase of bone nodule formation by human osteoprogenitor cells [77]. A similar effect was observed using specific proteasome inhibitors, e.g., as MG-132 and MG-262, indicating that the pro-osteogenic effect of bortezomib is due to its capacity to block proteasome activity in osteoblasts and osteoblast progenitors [77]. In line with these observations, bortezomib stimulate bone regeneration in mice in part by the modulation of the bone-specific transcription factor Runx2 and is able to rescue bone loss in a mouse model of osteoporosis [78].

Recently, the capacity of bortezomib at low concentration to stimulate osteoblastogenesis has been clearly confirmed in several osteoprogenitor cell lines both in murine and human system [79], suggesting the potential role of this drug as anabolic agent in MM patients. The authors reported that bortezomib activates β-catenin at high concentration (50-500nM) in human mesenchymal cells; even if the pro-osteogenic effect of Bortezomib occurs at low concentration 77, 78, 79.

PTH 

PTH amino-terminal peptide 1–34 has been shown to stimulate bone formation in mice and rats [56]. Moreover, clinical trials with intermittent PTH administration increases bone mass and reduce the incidence of fracture in patients with postmenopausal or glucocorticoid-induced osteoporosis [80]. Several mechanisms are involved in the anabolic effect of PTH. In vitro and in vivo studies have shown that PTH directly activates survival signaling in osteoblasts, and that delay of osteoblast apoptosis is the major contributor to the increased osteoblast number, at least in mice [56]. This effect requires Runx2-dependent expression of antiapoptotic genes like Bcl-2 55, 81. PTH promote osteoblast differentiation by its ability to promote exit from the cell cycle by decreasing expression of cyclin D and increasing expression of several cyclin-dependent kinase inhibitors 56, 82. PTH it has been shown to activate Wnt signaling in osteoblasts, to inhibit the Wnt antagonist sclerostin in osteocytes and to increase the Notch ligand Jagged1 in osteoblasts 82, 83. Interestingly, increasing evidence indicate that osteoblast expansion by PTH as well as PTH-mediated signal pathway in osteoblasts may stimulate and increase the number of hemopoietic stem cells 84, 85. Possible mechanisms involved in PTH stimulation of the osteoblastic niches include alterations in the expression of hematopoietic cytokines such as the stem cell factor, Notch ligands and insulin growth factor-1/2 [86].

On the basis of all these results, PTH could be used in the future as potential anabolic agent in MM to increase bone formation. Preliminary data obtained in SCID-rabbit MM model indicate that PTH administration, similarly to bortezomib, significantly increased BMD resulting in slower growth of MM cells [87]. In turn, it has recently been shown that the in vivo anabolic effect of bortezomib can be associated to an intermittent elevation of PTH levels in MM patients [88].