Notch Signaling and the Skeleton

Abstract

Notch 1 to 4 receptors are important determinants of cell fate and function, and Notch signaling plays an important role in skeletal development and bone remodeling. After direct interactions with ligands of the Jagged and Delta-like families, a series of cleavages release the Notch intracellular domain (NICD), which translocates to the nucleus where it induces transcription of Notch target genes. Classic gene targets of Notch are hairy and enhancer of split (Hes) and Hes-related with YRPW motif (Hey). In cells of the osteoblastic lineage, Notch activation inhibits cell differentiation and causes cancellous bone osteopenia because of impaired bone formation. In osteocytes, Notch1 has distinct effects that result in an inhibition of bone resorption secondary to an induction of osteoprotegerin and suppression of sclerostin with a consequent enhancement of Wnt signaling. Notch1 inhibits, whereas Notch2 enhances, osteoclastogenesis and bone resorption. Congenital disorders of loss- and gain-of-Notch function present with severe clinical manifestations, often affecting the skeleton. Enhanced Notch signaling is associated with osteosarcoma, and Notch can influence the invasive potential of carcinoma of the breast and prostate. Notch signaling can be controlled by the use of inhibitors of Notch activation, small peptides that interfere with the formation of a transcriptional complex, or antibodies to the extracellular domain of specific Notch receptors or to Notch ligands. In conclusion, Notch plays a critical role in skeletal development and homeostasis, and serious skeletal disorders can be attributed to alterations in Notch signaling.

1. Introduction

   1. Notch receptors

   2. Skeletal development and bone remodeling

2. Notch Signaling

   1. Notch ligands

   2. Notch receptors

   3. Notch intracellular signaling

   4. Notch target genes

3. Genetic Tools to Study Notch Signaling in the Skeleton

   1. Notch misexpression

   2. Conditional Notch misexpression and Cre driver lines

4. Notch Signaling, Skeletal Development, and Homeostasis

   1. Role of Notch in chondrogenesis and skeletal development

   2. Notch regulation of osteoblast differentiation and function

   3. Notch regulation of osteocyte function

   4. Notch regulation of osteoclastogenesis and bone resorption

5. Role of Hes and Hey in Skeletal Homeostasis

6. Notch and Skeletal Diseases

   1. Skeletal congenital diseases associated with loss-of-Notch function

   2. Skeletal congenital diseases associated with gain-of-Notch function

   3. Role of Notch in primary and metastatic bone tumors

   4. Fracture repair and Notch signaling

   5. Notch and osteoarthritis

7. Controlling Notch Signaling

   1. Biochemical inhibitors of Notch signaling

   2. Inhibitors of the Notch transcription complex formation

   3. Antibodies to Notch and Notch ligands

8. Conclusions

I. Introduction

A. Notch receptors

Notch is a ubiquitous signaling pathway that determines cell fate and function. In mice and humans, there are four Notch receptors and five Delta/Serrate/Lag2 (DSL) ligands that are termed Jagged (Jag)1 and Jag2 and Delta-like (Dll)1, Dll3, and Dll4 (1). Notch and DSL ligands are transmembrane proteins that mediate communication between neighboring cells. Notch receptors engaged by cognate ligands are cleaved by the γ-secretase complex; as a result, the Notch intracellular domain (NICD) is released from the cellular membrane (2). The NICD translocates to the nucleus, where it forms a complex with recombination signal-binding protein for Ig of κ region (Rbpjκ) and mastermind-like (Maml) to regulate transcription (3). The human ortholog of the murine Rbpjκ encodes for CBF1, Suppressor of Hairless, Lag1 (CSL). Gene targets of this canonical Notch signaling are hairy and enhancer of split (Hes)1, Hes5, Hes6, and Hes7, and HES-related with YRPW motif (Hey)1, Hey2, and Hey-like (HeyL). NICD can associate with intracellular proteins other than Rbpjκ and MAML in loosely defined interactions termed the noncanonical Notch signaling pathway.

Levels of Notch activity are tightly regulated by multiple mechanisms, such as post-translational modifications and endocytosis, and these can influence the temporal and cellular expression of Notch receptors and ligands (4, 5). The importance of these control mechanisms is exemplified by numerous inherited and somatic conditions caused by genetic defects in the core and accessory components of the Notch signaling pathway. The role of Notch signaling in skeletal development and homeostasis has been established by studies conducted in genetically modified mice. Mutations in Notch are associated with a selected group of genetic skeletal disorders, and dysregulated Notch signaling occurs in osteosarcoma and osteoarthritis. Modeling of these diseases in mice has provided new insight into disease pathogenesis and established the importance of Notch signaling in skeletal development and function.

B. Skeletal development and bone remodeling

Development of the appendicular skeleton and of selected elements of the axial skeleton occurs in hyaline cartilage templates of future bones. These structures arise from the condensation and chondrogenic differentiation of precursor cells of the mesenchymal lineage. As endochondral ossification progresses, chondrocytes residing in the hyaline cartilage develop a hypertrophic phenotype, induce mineralization of the extracellular matrix, and undergo programmed cell death. Apoptotic hypertrophic chondrocytes release angiogenic factors that promote the invasion of the bone collar by blood vessels, allowing bone cell precursors to colonize the calcified cartilage and to complete the ossification process. A region of the hyaline cartilage, located between the primary ossification center of the diaphysis and the secondary ossification centers of the epiphyses, grows at the same pace as the vascular invasion and in humans calcifies after sexual maturity is reached. This structure, called the growth plate, is responsible for the longitudinal growth of long bones, which is secondary to the proliferation and hypertrophic differentiation of growth plate chondrocytes (6).

Mesenchymal cells residing in the adult bone marrow have the potential to differentiate along the osteoblastic, chondrocytic, or adipocytic lineages (7). Mature osteoblasts can differentiate into quiescent bone-lining cells or become embedded in the bone matrix as osteocytes, cells with a dendritic appearance that play a critical role in mechanotransduction and bone remodeling (8). An alternative fate of mature osteoblasts is to die by apoptosis (9). Osteocytes express high levels of dentin matrix protein (Dmp)1 and Sost, the gene encoding for sclerostin, which is a secreted inhibitor of Wnt/β-catenin signaling (10–12).

Osteoclasts are multinucleated cells of the hematopoietic lineage that arise from the fusion of mononuclear bone marrow precursors (13). Osteoclastogenesis requires the presence of macrophage colony-stimulating factor (M-Csf) and of receptor activator of NF-κb-ligand (Rankl), a soluble protein encoded by TNF (ligand) superfamily 11. Rankl activity is opposed by osteoprotegerin, a soluble decoy receptor encoded by TNF receptor superfamily 11b (14). Osteoblasts and osteocytes govern osteoclast differentiation and bone resorption by secreting Rankl and osteoprotegerin, and osteoclasts trigger an osteoblastic response to couple bone formation to bone resorption during bone remodeling (15, 16). This process results in the continuous renewal of skeletal tissue. Bone modeling and remodeling are the processes that allow the skeleton to adapt to environmental challenges and to regulate mineral metabolism. Modeling occurs when bone formation and resorption are temporally independent and is carried out on different bone surfaces; it serves to shape bones during skeletal development and to adapt to mechanical forces (16). The number and coordinated activity of osteoblasts, osteocytes, and osteoclasts determine changes in skeletal remodeling and bone mass.

II. Notch Signaling

A. Notch ligands

The DSL ligands are single-pass transmembrane proteins with a conserved extracellular domain that contains multiple tandem epidermal growth factor (EGF)-like repeats. There are five DSL ligands: Jag1, Jag2, Dll1, Dll3, and Dll4. Members of the Jag and Dll families are “classic” Notch ligands and are recognized to interact with the extracellular domain of Notch. Additional soluble and transmembrane proteins can interact with Notch receptors and have the potential to modify Notch signaling. Delta-like homolog 1, also known as Pref1, is an inhibitor of Notch, and Delta/Notch-like EGF-related receptor, F3, and NB3, also termed Contactin1 and Contactin6, induce Notch signaling but act through Deltex-dependent noncanonical mechanisms (17–20). Microfibril-associated glycoproteins 1 and 2 can either induce or suppress Notch activity, and nephroblastoma overexpressed interacts with the extracellular domain of Notch1 and can regulate Notch-dependent transactivation (21–24).

B. Notch receptors

The four Notch receptors are evolutionary conserved single-pass transmembrane proteins (Figure 1). Multiple EGF-like tandem repeats are a prominent feature of the extracellular region of Notch and the site of interaction with cognate ligands (25). Precursors of mammalian Notch receptors are cleaved by furin-like proprotein convertases in the Golgi compartment. This event leads to the formation of a heterodimer where the N terminus of the transmembrane domain interacts with the C terminus of the extracellular region, forming a heterodimerization domain (HD). The C terminus of the extracellular region contains three Lin12-Notch repeats, and with the HD they form a negative regulatory region (NRR), which prevents receptor activation in the absence of ligands (26). Point mutations in the HD of NOTCH1 that allow ligand-independent activation of the receptor lead to a gain-of-Notch function and are associated with T-cell acute lymphoblastic leukemia (27, 28). Mutations in the HD of NOTCH2 are less common and are rarely associated with splenic marginal zone B-cell lymphomas (29). The transmembrane domain of Notch contains cleavage sites recognized by the γ-secretase complex and are critical for signal activation. The intracellular domain of the four Notch receptors consists of an Rbpjκ-association module (RAM) domain, a nuclear localization sequence, seven ankyrin (ANK) repeats, and two closely associated nuclear localization sequences. The C terminus contains a proline (P)-, glutamic acid (E)-, serine (S)-, threonine (T)-rich motif (PEST) domain, which is the substrate of ubiquitin ligases that target the NICD for proteasomal degradation (30).

Figure 1.

Domains of the four Notch receptors. The upper panel shows the domain and motif organization of a generic human/murine Notch receptor before cleavage at the S1 site by furin-like convertases in the Golgi compartment. The extracellular domain contains a leader peptide (LP) and multiple EGF-like tandem repeats followed by Lin12-Notch repeats (LNR) and the HD. The transmembrane domain (TMD) is located between the extracellular and intracellular domains. The NICD contains a RAM, a nuclear localization sequence (NLS), ANK repeats, and tandem NLS, which are followed by a PEST domain. The lower panel shows the domains and motifs of heterodimeric individual receptors; the NRR is formed by the LNR and HD following cleavage at the S1 site. Notch1 and Notch2 have 36 EGF-like repeats; in green are those required for binding of Notch1 and Notch2 to cognate DSL ligands. Notch1 and Notch2 have a similar NICD, and Notch3 has 34 EGF-like repeats and a shorter NICD than Notch1 and Notch2. Notch4 has 29 EGF-like repeats and an NICD that is shorter than that of other receptors and lacks the tandem NLS located between the ANK repeats and the PEST domain.

Although Notch1 and Notch2 are structurally similar, their functions are not redundant. Notch1 null mice die during development secondary to vascular malformations, and a hypomorphic Notch2 allele causes vascular and renal defects that lead to perinatal death (31, 32). Functional differences between the two receptors are possibly related to their temporal and cellular expression and to variations in the affinity of their extracellular domain for cognate ligands because the NICD of Notch1 and Notch2 appear to be functionally equivalent (33, 34). The structure of Notch3 diverges modestly from that of Notch1 and Notch2. Notch3 null mice are viable and fertile, although they exhibit vascular abnormalities; and NOTCH3 mutations in humans cause cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) syndrome (35–37). Notch4 is dispensable for embryonic development, but some of its functions overlap with those of Notch1 (38).

Expression of individual Notch receptors and ligands is confined to selected cell types, ensuring selective receptor activation by specific ligands and providing an initial level of signal regulation (39). More complex mechanisms to control the activity of Notch receptors and the specificity and binding affinity for cognate DSL ligands are provided by the glycosylation of the EGF-like repeats in the Notch extracellular domain. Glycosylation is carried out by the sequential actions of Pofut1 and three fringe proteins, namely lunatic fringe (Lfng), manic fringe, and radical fringe (4, 40–42). The levels of Notch receptors at the cell surface are governed by endocytosis and by the activity of ubiquitin ligases, including Cbl, which targets Notch for degradation in the lysosome, Itch, and its human homolog AIF4 (43, 44). An alternate destiny for ubiquitinated Notch receptors is the recycling endosome, which allows their re-exposure at the cell surface.

C. Notch intracellular signaling

The maturation and activation of Notch receptors require four proteolytic events that occur at cleavage sites located in the HD domain and are sequentially numbered S1 to S4 (30) (Figure 2). The S1 site is cleaved in the Golgi compartment by furin-like proprotein convertases, an event necessary for the formation of a functional Notch heterodimer (45). DSL ligands bound to the Notch extracellular domain are internalized by the ligand-expressing cell; this process, defined as trans-endocytosis, is necessary for ligand-dependent activation of the Notch receptors (30, 46, 47). Trans-endocytosis exposes the S2 site and allows its cleavage by the TNF α conversion enzyme, which is a disintegrin and metalloprotease domain metalloprotease (48). The S2 cleavage generates an intermediate peptide, and this is cleaved at the S3 and S4 site by a γ-secretase complex, which contains the transmembrane proteases Presenilin1 and Presenilin2 (49). Cleavage of the S3 and S4 sites results in the release of the NICD, which translocates to the nucleus to regulate transcription (2, 50). In the absence of NICD, Rbpjκ associates with corepressor proteins to suppress transcription. After nuclear translocation, NICD forms a ternary complex with Rbpjκ and Maml and recruits coactivators of transcription that displace the transcriptional corepressors (51). As a consequence, the Notch target genes Hes1, Hes5, Hes6, and Hes7 and Hey1, Hey2, and HeyL are induced (52–54) (Figure 3).

Figure 2.

Activation of Notch receptors. DSL ligands expressed by a signal-sending cell (top) engage Notch receptors on a signal-receiving cell (bottom). The DSL ligand bound to the extracellular domain of Notch is internalized by the ligand-expressing cell, a process termed trans-endocytosis, exposing the S2 cleavage site of Notch to the activity of the metalloprotease disintegrin and metalloprotease domain (ADAM) 10. This proteolytic event creates a peptide consisting of the transmembrane and intracellular regions of Notch and exposes the intramembranous S3 and S4 sites, which are recognized by the γ-secretase complex. Cleavage at the S3 and S4 sites by γ-secretase activity releases the Notch intracellular domain, which translocates to the nucleus to regulate transcription.

Figure 3.

Transcriptional events regulated by Notch receptors. A, In the absence of the NICD, Rbpjκ associates with transcriptional repressors and histone deacetylase complexes (HDAC). These interactions occur at the promoter region of Notch transcriptional targets, such as members of the hairy enhancer of split (Hes) and HES-related with YRPW motif (Hey) gene families. B, After nuclear translocation, NICD associates with Rbpjκ and Maml to form a ternary NICD/Rbpjκ/Maml complex. These events lead to the recruitment of transcriptional activators, displacement of the transcriptional repressors, and subsequent expression of Hes and Hey. C, Maml associates with CDK8, which phosphorylates (P) the PEST domain of the NICD. D, Maml recruits the E3 ubiquitin-ligase F-box and WD repeat domain-containing (Fbxw) 7, which ubiquitinates (U) the NICD and leads to its proteasomal degradation. As a result, the active transcriptional complex is disassembled, allowing reinstitution of transcriptional suppression after association of Rbpjκ with the transcriptional repressors and the HDAC (A).

Rbpjκ consists of three evolutionary conserved regions; namely, N-terminal, β-trefoil, and C-terminal domains (55). The N-terminal and β-trefoil domains bind to DNA, and the β-trefoil domain serves as the site of interaction with the RAM domain of the NICD. The C-terminal domain of Rbpjκ interacts with the ANK domain of the NICD and with the N terminus of Maml, and it is necessary for the assembly of the ternary NICD/Rbpjκ/Maml complex (56). Similarly to the null mutations of Notch1 and Notch2, the homozygous global down-regulation of Rbpjκ is lethal during early embryonic development due to vascular abnormalities (57). Although deletion of Rbpjκ precludes the ability of NICD to induce a transcriptional response via the NICD/Rbpjκ/Maml ternary complex, Rbpjκ functions independent from Notch signaling are possible and are found during pancreatic development (58–60). For this reason, studies where the Rbpjκ deletion is carried out to prevent Notch activity are not always representative of an inhibition of Notch canonical signaling and should be interpreted with caution because other signals may be affected.

Maml proteins contribute to the formation of an active transcriptional complex and determine the duration of the NICD-mediated transcriptional event by interacting with cyclin (Ccn)-dependent kinases (CDKs) (61). Cdk8 phosphorylates the PEST domain of the NICD, causing the disassembly of the ternary complex and leading to the ubiquitination of the NICD by E3 ubiquitin-ligase F-box and WD repeat domain-containing 7 (51, 62).

D. Notch target genes

Hes and Hey are evolutionary conserved basic helix-loop-helix (bHLH) transcription factors. There are seven Hes proteins, termed Hes1 through Hes7, and with the exception of Hes2 and Hes3, these transcription factors are targets of Notch canonical signaling, (63–65). Hes1, Hes3, and Hes5 are required to preserve the undifferentiated state of precursor cells during development and adult life (66, 67). Hes7 is critical for proper somite segmentation, and Hes6 is a suppressor of Hes1 (68, 69). There are three Hey transcription factors, Hey1, Hey2, and HeyL, and, like Notch, they are required for vascular development (70–74).

Hes and Hey proteins share a high degree of structural homology, and the bHLH domain allows homodimerization or heterodimerization of Hes with Hey or with other transcriptional regulators that harbor bHLH domains. This determines the specificity of their binding to DNA, adding a layer of complexity to the biological impact of Notch signaling (70). Structural differences between Hes and Hey proteins are limited. Hes proteins are characterized by the presence of a conserved proline in the bHLH domain, whereas a glycine is found in the corresponding site in Hey proteins. An additional difference lies in the tetrapeptide motif found in the C terminus, where Hes proteins exhibit a WRPW sequence and Hey proteins exhibit the YRPW motif (54). For the most part, Hes and Hey proteins act as transcriptional repressors, a function mediated by the bHLH domain. In addition, the WRPW motif of Hes1 can interact with intracellular proteins to form a transcriptional repressor complex (75).

III. Genetic Tools to Study Notch Signaling in the Skeleton

A. Notch misexpression

1. Notch loss of function

Global inactivation of Notch1 and Notch2 results in embryonic lethality so that these constitutive loss-of-function models do not allow for the phenotypic characterization of adult mice (Table 1) (31, 32). Notch3 and Notch4 null mice are viable and fertile, but little is known about the function of Notch3 and Notch4 in the skeleton (37, 38). Because of the early lethality of Notch1 and Notch2 null mice, mouse models of conditional Notch1 and Notch2 inactivation have been used to establish the skeletal functions of these Notch receptors (76, 77). In these models, selected exon sequences are flanked by loxP sites, so that Cre recombination results in gene inactivation (76, 77). To block activation of all Notch receptors, investigators have used the global and conditional inactivation of Presenilin1 and Presenilin2 to prevent cleavage of Notch receptors and the release of the NICD and consequent Notch signal activation (78, 79).

Table 1.

Mouse Models for the Study of Notch Signaling

Mouse Strain	Genetic Modification	Consequence on Notch Signaling	Ref.
Loss of function
Notch1 null	Insertion of a neomycin cassette into the exon coding for the EGF repeat 32	Absent Notch1	31
Notch2 null	Insertion of a neomycin cassette into the exon coding for the EGF repeat 14 and 0.3 kb upstream intronic sequence	Dysfunctional Notch2	32
Notch3 null	Replacement of the sequences coding for EGF repeats 8–12 with a neomycin cassette	Absent Notch3	37
Notch4 null	Replacement of sequences coding for amino acids 1249–1434 in the extracellular domain with a neomycin cassette	Nonfunctional Notch4	38
Notch1 conditional	Flanking of the putative promoter and the exon encoding the signaling peptide with loxP sequences	Absent Notch1 after Cre recombination	76
Notch2 conditional	Flanking of exon 3 with loxP sites	Truncated and nonfunctional Notch2 after Cre recombination	77
Presenilin1 conditional	Flanking of exon 2 and exon 3, inclusive of the start of the coding sequence, with loxP sites	Reduction of Presenilin1 after Cre recombination	78
Presenilin2 null	Replacement of exon 5 and neighboring intronic sequences with a hygromycin cassette	Absent Presenilin2	79
Rbpjκ conditional	Flanking of sequences coding for DNA binding domain with loxP sites	Nonfunctional Rbpjκ	80
Dominant negative Maml	Insertion of a loxP-flanked neomycin cassette and sequences coding for amino acids 13–74 in the Rosa26 locus	NICD/Rbpjκ interaction precluded after Cre recombination	81
Gain of function
RosaNotch	Insertion of a loxP-flanked neomycin STOP cassette and the sequence coding for the Notch1 NICD in the Rosa26 locus	Constitutive expression of Notch1 NICD after Cre recombination	83
RosaNotch2	Insertion of a loxP-flanked neomycin STOP cassette and the sequence coding for the Notch2 NICD in the Rosa26 locus	Constitutive expression of Notch2 NICD after Cre recombination	85
Notch reporters
Transgenic Notch reporter (TNR)	Transgene containing 4 Rbpjκ -binding sites and the basal simian virus 40 promoter governing EGFP expression	EGFP induced by formation of NICD/Rbpjκ/Maml complex	88
Notch activity sensor (NAS)	Transgene containing 12 Rbpjκ -binding sites and the minimal β-globin promoter directing nlacZ expression	nlacZ induced by formation of NICD/Rbpjκ/Maml complex	89
Hes1-EmGFPSAT	Replacement of all Hes1 exons with the EmGFP coding sequence	Phenocopies endogenous Hes1 expression	90

To inactivate Rbpjκ and determine whether Notch canonical signaling is responsible for a Notch effect in a target cell, Notch activation is studied in the context of the Rbpjκ inactivation. Rbpjκ conditional mice were created by placing loxP sites flanking sequences coding for the DNA binding domain (80). These mice also can be used to determine the function of Rbpjκ under basal and, presumably, non-Notch-activating conditions. An alternative approach to block canonical Notch signaling is the use of a dominant negative Maml, containing sequences necessary to bind to NICD and Rbpjκ, but lacking sequences necessary to recruit coactivators of transcription (81).

2. Notch gain of function

Although transgenic mouse models have been used to deliver the Notch1 NICD to the skeletal environment, the constitutive gain-of-Notch1 function frequently results in a pronounced phenotype and lethality at an early age (82). A more commonly used mouse model to examine the effects of Notch activation in the skeleton is the Rosa^Notch conditional mouse (83, 84). In these mice, the Rosa26 locus is targeted with a DNA construct encoding for the Notch1 NICD, preceded by a STOP cassette flanked by loxP sites and cloned downstream of Rosa26 regulatory elements. A similar model was created for the conditional activation of Notch2 (85). Expression of the Notch1 or Notch2 NICD from the targeted Rosa26 locus occurs after the excision of the STOP cassette by Cre recombination of loxP sequences (86, 87). To ensure that Notch activation has occurred, it is important to demonstrate up-regulation of canonical Notch target genes, such as Hes1, Hey1, Hey2, and HeyL, or up-regulation of Notch signaling using reporter assays or mouse reporter lines. This can be done in vitro by transfecting cells from experimental mice with reporter constructs or in vivo by crossing experimental mice with lines expressing enhanced green fluorescent protein (EGFP) or LacZ downstream Rbpjκ binding motifs (88, 89). Hes1-EmGFP^SAT, where Hes1 regulatory elements drive the expression of emerald green fluorescent protein (EmGFP), can be used as surrogates to demonstrate activation of canonical Notch signaling (90).

B. Conditional Notch misexpression and Cre driver lines

To recombine loxP flanked DNA sequences, various transgenic and knock-in lines expressing the Cre recombinase in cells of the osteoblastic, osteoclastic, and chondrocytic lineages are available (91). These lines can either activate Notch in gain-of-function models, such as the Rosa^Notch mouse, or inactivate various components of Notch signaling, where DNA sequences of specific genes are flanked by loxP sites. To study the consequences of the Notch activation or inactivation in undifferentiated and differentiated cells of the osteoblastic lineage, conditional mice are bred with mice expressing Cre under the control of Sp7 (encoding for osterix), the Bglap (encoding for osteocalcin) or the 2.3-kb fragment of the collagen type I α1 (Col1a1) promoter, respectively (92–94). To study the misexpression of Notch in osteocytes, mice are crossed with transgenics delivering Cre under the control of the Dmp1 or the Sost promoter (95, 96). To activate or inactivate Notch in undifferentiated and differentiated osteoclasts, conditional mice are crossed with mice carrying a knocked-in Cre in the endogenous M lysozyme locus of myeloid cells or in the CathepsinK locus (97–99). An alternative to express Cre in undifferentiated cells of the osteoclast lineage is the use of Cd11b-Cre transgenics, but Cre expression is not osteoclast-specific and is exclusive to male mice (100).

To induce Cre-mediated recombination in mesenchymal cellular condensations in the limb bud, the paired-related homeobox 1 (Prx1) enhancer is available (101). To induce Cre-mediated recombination in chondrocytes, mouse models where the collagen type II α1 (Col2a1), aggrecan, or sex-determining region Y-box (Sox) 9 regulatory elements drive Cre expression are used (102–106). The aggrecan and Sox9 models express a recombinase/estrogen receptor fusion protein (CreER^T2), and the Cre recombinase is induced after the administration of tamoxifen (105, 106).

IV. Notch Signaling, Skeletal Development, and Homeostasis

A. Role of Notch in chondrogenesis and skeletal development

The initial demonstration of Notch1 expression in the proliferative zone of the growth plate suggested a role of Notch signaling in chondrocyte differentiation. Studies conducted in ATDC5 cells, an immortalized teratocarcinoma cell line that can develop a chondrocyte-like phenotype in vitro, demonstrated that the Notch ligand Dll1 suppressed chondrogenic differentiation (107). Accordingly, overexpression of the Notch1 and Notch2 NICD in ATDC5 cells inhibited the expression of Sox9, a transcription factor that is critical for chondrocyte differentiation (108–110). Although Notch is considered a transcriptional activator, recruitment of the NICD/Rbpjκ complex to Rbpjκ consensus sequences in the Sox9 promoter could cause a direct inhibitory effect on Sox9 transcription (110). Alternatively, the suppression of Sox9 transcription by Notch may be secondary to the induction of Hes and Hey, which are known transcriptional repressors (70). Consistent with a suppressive role of Notch in chondrogenesis, exposure of limb micromass cultures to a γ-secretase inhibitor, to prevent activation of Notch, enhanced chondrogenic differentiation (111). Selected effects of Notch in chondrocytes are exerted through the transcriptional repressors Hes1 and Hey1, which bind to N-boxes adjacent to a Sox9 consensus sequence in the Col2a1 enhancer to suppress type II collagen transcription (112). Notch inhibits chondrogenesis in vivo, and Dll1 arrests chondrocyte differentiation in chicken embryos, preventing the formation of the hypertrophic zone in the growth plate (113).

Deletion of Presenilin1, Notch1, and Notch2 in cells of the developing limb bud was carried out to determine the function of Notch in endochondral bone formation. The conditional inactivation of Presenilin1 in a global Presenilin2 null genetic composition or the dual Notch1 and Notch2 inactivation in the limb bud using a Prx1-Cre enhancer led to the accumulation of hypertrophic chondrocytes in the growth plate and severe skeletal malformations (114). The conditional dual Notch1 and Notch2 or the single Rbpjκ deletion in the limb bud caused an early increase in cancellous bone volume, possibly secondary to an elongation of the growth plate, and facilitated the commitment of mesenchymal cells to the osteoblastic lineage (114, 115). The study also reported that inactivation of Notch2, but not of Notch1, in Prx1-expressing cells recapitulated the phenotype, indicating that Notch2 carries out the inhibitory functions of the Notch signaling pathway during endochondral bone formation (114, 115). In accordance with these reports, overexpression of the Notch1 NICD in the limb bud impairs the formation of hyaline cartilage bone templates in the appendicular skeleton, and this effect was precluded by the deletion of Rbpjκ (116, 117).

Further investigations to determine the function of Notch in chondrocytes examined the consequences of Notch misexpression in Col2a1-expressing cells and demonstrated that Notch impairs chondrocyte proliferation and differentiation (118). Most of the effects required activation of canonical signaling and were reversed by the down-regulation of Rbpjκ (110, 117). Under basal conditions, Rbpjκ also suppressed chondrocyte differentiation, and its inactivation resulted in elongation of the hypertrophic chondrocyte zone of the growth plate in the appendicular skeleton (118). The function of Notch target genes in endochondral bone formation was investigated after the conditional inactivation of Hes1 in chondrocytes in the context, or not, of a global Hes5 null background (119, 120). The study suggested that Hes1 is dispensable for endochondral bone formation, but this might have been due to genetic compensation from other Hes paralogs (65, 121, 122). Indeed, the conditional inactivation of Hes1 in cells expressing the Prx1 enhancer in a dual Hes3 and Hes5 global null background increased femoral length, demonstrating an inhibitory function in cartilage and suggesting that Hes may be accountable for some of the effects of Notch in this tissue (123).

B. Notch regulation of osteoblast differentiation and function

Initial studies addressing the function of Notch in the specification of the osteoblastic fate were carried out in murine cell lines (124–129). Stable Notch1 activation in C2C12 myoblastic cells, ST-2 stromal cells, and osteoblastic MC-3T3 cells suppressed osteoblastic differentiation (124–126). In ST-2 cells, Notch suppressed Wnt/β-catenin signaling, but not bone morphogenetic protein (BMP) signaling (126). The inhibitory effect of Notch on Wnt signaling is also observed in bone marrow stromal cells from transgenics expressing the Notch1 NICD under the direction of the Col1al promoter. In Notch overexpressing cells, cytosolic β-catenin levels and the stimulation of alkaline phosphatase activity by Wnt3 are suppressed by Notch (82). In addition, Notch opposes the transactivating potential of a stable form of β-catenin, suggesting that Notch opposes Wnt signaling by mechanisms upstream and downstream of β-catenin signaling. The downstream effects appear to be mediated by Hes1, forming a complex with Groucho and T-cell specific factor and thus preventing T-cell factor interactions with β-catenin, which are necessary to induce transcription of Wnt-dependent genes. In contrast to the results obtained in stable cell lines expressing Notch1 NICD, the transient activation of Notch enhanced BMP activity and osteoblastogenesis (117, 127, 128). This work demonstrated that the duration of the cellular exposure to Notch signaling can determine the outcome of the osteoblast differentiation program. However, in vivo models have confirmed that Notch inhibits osteoblastic cell differentiation.

Notch enhances the replication of cells of the osteoblastic lineage, an effect that correlates with increased CcnD1 and CcnE expression and suppressed osteoblast differentiation (130). This may be secondary to the inhibition of Wnt/β-catenin signaling or of the function of Runx2, which is required for osteoblast maturation. The NICD can interact directly with Runx2 and inhibit the transactivation of Bglap (encoding for osteocalcin) (130). In addition, Hes and Hey proteins inhibit the function of Runx2, contributing to the suppression of osteoblastogenesis by Notch (114). A common mesenchymal cell progenitor can give rise to osteoblasts and adipocytes, and mesenchymal cells enter each lineage in a mutually exclusive manner (131). Therefore, signals that determine the fate of osteoblasts often influence adipogenesis. In ST-2 stromal cells, Notch enhances the adipogenetic effect of cortisol, but down-regulation of Notch experiments has concluded that Notch is both required and dispensable for adipocyte specification (125, 132, 133). The reason for these controversial results is not apparent. Hes1 favors the initial phases of adipogenesis by suppressing Pref1 but inhibits terminal adipocyte differentiation by repressing the transcription of genes required for adipocyte maturation (134, 135).

Studies in genetically modified mice have enabled the manipulation of Notch signaling in cells of the osteoblastic lineage at defined stages of differentiation. Notch1 induction under the control of the Sp7 or Bglap promoter, expressed in osteoblast precursors and mature osteoblasts, respectively, caused profound osteopenia secondary to impaired bone formation (136). In agreement with these findings, the dual conditional Notch1 and Notch2 inactivation in Sp7-expressing cells increased osteoblast number, bone formation, and cancellous bone volume (137). It is of interest that mice harboring the activation of Notch1 in Sp7- or Bglap-expressing cells failed to form a compact cortical bone. The appearance of the cortical shell resembled that of embryonic cortical bone, suggesting that a compact cortex failed to develop (138). The phenotype of mice where Notch1 is activated in Sp7-expressing cells evolved with time to resemble the phenotype of the Notch1 activation in mature osteoblasts and osteocytes. This is expected because cell lineage-tracing studies have shown that a terminal fate of cells expressing Sp7 is the osteocyte (139, 140). At 3 months of age, Notch activation in Sp7-expressing cells increased bone volume and osteoblast number, but bone formation was disrupted (136). The inhibitory role of Notch in osteoblast differentiation/function was observed in transgenic mice expressing the Notch1 NICD under the control of a 3.6-kb fragment of the Col1a1 promoter (82). These mice exhibited profound osteopenia, and the mechanism appeared to involve an inhibition of Wnt/β-catenin signaling by Notch (82, 126). It is of interest that Notch was found to interact with nuclear factor of activated T cells (Nfat), suggesting that the effects of Notch in cells of the osteoblast lineage are modulated by interactions with other signaling pathways (141, 142). However, it remains to be determined how these interactions regulate osteoblast cell fate and function.

The actions of Notch1 evolve in mature osteoblasts, and induction of the Notch1 NICD in cells expressing a 2.3-kb fragment of the Col1a1 promoter caused an osteopetrotic phenotype due to a suppression of bone resorption. However, osteoblasts were dysfunctional and deposited immature woven bone, confirming the inhibitory role of Notch in osteoblast differentiation and function (130, 136). The phenotype involves canonical Notch signaling because the conditional deletion of Rbpjκ in the context of the Notch1 activation reverses the effects described (143). It is of interest that under basal conditions, Rbpjκ is dispensable, and the conditional inactivation of Rbpjκ in mature osteoblasts or osteocytes is without phenotypic consequences (143, 144). The findings confirm an inhibitory role of Notch in osteoblastogenesis, although the consequences depend on the timing of Notch induction during cell differentiation. Notch activation in the early phases of the osteoblast differentiation program prevents the maturation of cells capable to synthesize a mineralized matrix, whereas Notch induction in mature cells precludes further differentiation and causes an accumulation of dysfunctional osteoblasts (82, 114, 130, 136). These events are possibly mediated by suppression of Runx2 transcription, decrease of Wnt signaling, and increase in cellular proliferation secondary to the up-regulation of CcnD1 and CcnE1 expression (82, 130).

C. Notch regulation of osteocyte function

To determine the function of Notch signaling in osteocytes, our laboratory characterized the skeletal consequences of the Notch1 NICD overexpression and of the Notch1, Notch2, and Rbpjκ inactivation in cells expressing the 9.6-kb fragment of the Dmp1 promoter (95). Dmp1 is preferentially, although not exclusively, expressed by osteocytes. Activation of Notch1 in osteocytes caused an osteopetrotic phenotype by decreasing cancellous bone resorption and formation, so that bone remodeling was suppressed (136, 145). The phenotype was compartment-specific because cortical bone formation was increased, but cortical bone was porous, with the appearance of trabecular bone, possibly because it failed to reach maturity. The dual inactivation of Notch1 and Notch2 in osteocytes, like in osteoblasts, caused a modest increase in cancellous bone volume and a decrease in bone resorption (145). Although under basal conditions Rbpjκ does not play a role in osteoblast or osteocyte function, the phenotypic changes observed after the activation of Notch1 in osteocytes were reversed in the context of the inactivation of Rbpjκ (144). These findings indicate that canonical signals are responsible for the phenotype, as they are for the actions of Notch in osteoblasts.

The mechanism of Notch1 action in osteocytes was secondary to an induction of osteoprotegerin and a suppression of the Wnt antagonists sclerostin and Dkk1, leading to an increase in Wnt/β-catenin signaling (144, 145). This induction of Wnt signaling occurs by indirect mechanisms and is not observed in osteoblasts because these cells express low levels of sclerostin and Notch does not have the opportunity to down-regulate Sost in a cell where Sost is hardly expressed (12). These findings do not contradict the direct inhibitory effect of Notch on Wnt signaling in stromal cells and osteoblasts that is likely responsible for the impaired osteoblast maturation (82, 126). The skeletal phenotype of Notch1 and β-catenin activation in osteocytes has a degree of overlap, and Wnt signaling is known not only to enhance osteoblastogenesis but also to repress osteoclastogenesis indirectly by inducing osteoprotegerin and directly by acting on osteoclast precursors (11).

The interactions of Notch and Wnt in osteocytes are complex, and Notch receptors are induced in mice expressing a constitutively active β-catenin mutant under the control of the Dmp1 promoter (146). This would create a positive-feedback loop where Wnt enhances Notch signaling and where Notch, by down-regulating Sost, increases levels of Wnt activity in osteocytes. It also provides a mechanism of Notch induction in osteocytes because activation of Notch signaling requires cell-to-cell contact and this is not bound to occur in osteocytes, cells that communicate via a canalicular network. It is also possible that mechanical forces induce Notch signaling in osteocytes either directly or indirectly through the activation of Wnt. Mechanical loading enhances Wnt signaling in osteocytes, and our laboratory has found that fluid shear stress activates Notch signaling in osteocytic cell lines (147) (our unpublished observations).

D. Notch regulation of osteoclastogenesis and bone resorption

Notch signaling regulates osteoclast differentiation and function by direct and indirect mechanisms, but the effects of Notch1 and Notch2 on osteoclastogenesis and bone resorption are distinct and opposite. Whereas Notch1 inhibits, Notch2 enhances osteoclastogenesis by direct and indirect mechanisms. Inactivation of Notch1 in osteoblasts increases bone resorption and causes bone loss by suppressing osteoprotegerin expression (114, 130, 148). Accordingly, activation of Notch1 in mature osteoblasts and osteocytes induces osteoprotegerin and suppresses bone resorption (130, 136, 145). In contrast to these effects, activation of Notch2 in osteoblasts induces Rankl and enhances osteoclastogenesis (149).

Notch1 has direct effects on osteoclast precursors. In vitro studies in bone marrow mononuclear precursors revealed that activation of Notch receptors, after exposure to the Notch ligands Dll1 or Jag1 or expression of the Notch1 NICD, suppressed osteoclastogenesis (148, 150). Accordingly, the deletion of Notch1, Notch2, and Notch3 in myeloid cells enhanced osteoclastogenesis, but not bone resorption, and Notch null mice are sensitized to the stimulatory effects of Rankl on bone resorption (148). Similar results were obtained after the inactivation of Rbpjκ in myeloid cell cultures, indicating that the inhibitory effects of Notch in osteoclast differentiation in vitro are mediated by canonical mechanisms (151, 152). However, under selected experimental conditions in vitro, nonspecific activation of Notch signaling can induce osteoclastogenesis, whereas suppression of Notch inhibits osteoclast differentiation (149, 151, 153–155). This may reflect the fact that signaling from all Notch receptors was altered nonspecifically.

In contrast to the inhibitory effects of Notch1 on osteoclastogenesis in vitro, work in human and murine osteoclast precursor cells revealed that Notch2 induces osteoclastogenesis. The mechanism involves interactions of the Notch2 NICD with nuclear factor (Nf)-κB, that lead to the transcription of Nfatc1, a gene critical for osteoclastogenesis (149, 153). Additional in vitro reports indicated that inhibition of the ubiquitin E3 ligase ligand of numb-protein X 2 precluded osteoclast differentiation by suppressing levels of the Notch2 NICD (156). The number of osteoclast precursors as well as their potential to differentiate to mature osteoclasts is increased in mice harboring a global gain-of-function knock-in mutation in Notch2 (149). The observations demonstrate that in contrast to Notch1, activation of Notch2 induces osteoclastogenesis and enhances bone resorption.

V. Role of Hes and Hey in Skeletal Homeostasis

In vitro studies demonstrated that Hes1 and Hey1 suppress the transcriptional activity of Runx2 and the expression of gene markers characteristic of the osteoblastic phenotype (157–161). In accordance with these observations, transgenic overexpression of Hes1 in immature osteoblasts causes osteopenia secondary to impaired osteoblastic function, and the conditional deletion of Hes1 in osteoblasts, in the context of the global dual Hes3 and Hes5 inactivation, increases bone formation and cancellous bone volume (123). The ubiquitous overexpression of Hey1 in transgenic mice causes osteopenia and a modest reduction in mineral apposition rate (162). Accordingly, female mice heterozygous for a Hey1 null allele and homozygous for a HeyL null allele display increased cancellous bone volume associated with more osteoblasts (115). However, phenotypes obtained from global gene misexpression could be the result of developmental or systemic nonskeletal events. Studies on Hey2 overexpression or conditional deletion in osteoblastic cells demonstrated that Hey2 decreases cancellous bone volume (163). The observations indicate that Hey1, Hey2, and HeyL have similar inhibitory functions in cells of the osteoblastic lineage and are consistent with the effects of Notch1 activation in osteoblasts, suggesting that Hes and Hey may contribute to the skeletal actions of Notch signaling in bone.

VI. Notch and Skeletal Diseases

A. Skeletal congenital diseases associated with loss-of-Notch function (Table 2)

Table 2.

Genetic Disorders Associated With Notch Dysregulation and Affecting the Skeleton

Disease	Mutation	Major Manifestations
Loss of function
AOS	NOTCH1, EOGT, DLL4, RBPJκ or CSL, ARHGAP31, DOCK6	Cranial aplasia cutis congenita
		Terminal limb defects
Alagille syndrome	JAG1, NOTCH2	Facial dysmorphism
		Vertebral abnormalities
		Bile duct atresia
		Cardiovascular defects
Spondylocostal dysostoses	DLL3, MESP2, HES7, LNF	Vertebral segmentation defects
		Rib anomalies
Spondylothoracic dysostoses	MESP2	Vertebral segmentation defects
		Rib anomalies
Gain of function
Brachydactyly	CHSY1	Short digits
		Short stature
Hajdu-Cheney syndrome	NOTCH2	Acroosteolysis
		Osteoporosis
		Craniofacial development defects
		Neurological complications
		Polycystic kidneys
LMS	NOTCH3	Meningocele
		Craniofacial abnormalities

1. Adams Oliver syndrome

Adams Oliver syndrome (AOS), a rare congenital disorder characterized by aplasia cutis congenita and terminal transverse limb defects, is frequently associated with mutations of genes encoding various components of the Notch signaling pathway (Figure 4). The genetic basis for AOS is heterogeneous, and the disease is known to be associated with six gene mutations. Two genes— ARHGAP31, encoding for Rho guanosine triphosphatase (GTPase)-activating protein 31; and DOCK6, encoding for dedicator of cytokinesis 6—are not associated with alterations in Notch signaling. Four genes, including NOTCH1; DLL4; CSL, the human ortholog of Rbpjκ; and EOGT, encoding for EGF-domain-specific O-linked N-acetylglucosamine transferase (O-GlcNAc), have a function in the Notch signaling pathway. Heterozygous autosomal dominant mutations in ARHGAP31, CSL, DLL4, and NOTCH1 or homozygous recessive mutations in DOCK6 and EOGT, as well as sporadic gene mutations, are associated with the syndrome. AOS presents with significant phenotypic variations, and the terminal limb defects include oligodactyly, syndactyly, hypoplastic nails, and transverse amputations (164, 165). Congenital cardiac defects, including ventricular septal defects, tetralogy of Fallot, and anomalies of arteries and cardiac valves are uncommon and associated with peripheral vascular abnormalities, such as cutis marmorata telangiectatica congenita and retinal hypovascularization.

Figure 4.

Clinical features of AOS. A, Brachydactyly of the left foot and missing toes on right foot. B, Bald area on the scalp. C, Brachydactyly of toes. D, Brachydactyly of fingers. E, Aplasia cutis congenita. G, Short distal phalanges and symphalangism of index finger. [Reproduced from J. A. Meester et al: Heterozygous loss-of-function mutations in DLL4 cause Adams-Oliver syndrome. Am J Hum Genet. 2015;97(3):475–482 (171), with permission. © American Society of Human Genetics.]

It is of interest that NOTCH1 mutations are associated with a higher prevalence of cardiac abnormalities, cutis aplasia, and transverse limb defects (166, 167). Small vessel abnormalities and vascular obstruction by thrombosis during development may be responsible for the terminal transverse limb defects, and these are consistent with the established role of Notch signaling in vascular development (166–169). A variety of NOTCH1 mutations across the length of the receptor are found in AOS. Most of the gene mutations occur in the EGF-like repeats of the extracellular domain, leading to structural changes and a loss-of-NOTCH1 function; in one family, an 85-kb deletion of the NOTCH1 promoter and first exon was reported (167).

Heterozygous mutations in CSL leading to defects in DNA binding without affecting CSL-NICD interactions were found in two families with AOS. The loss-of-function mutations result in decreased binding of CSL to regulatory regions of Notch target genes (170). Missense and nonsense mutations throughout the coding sequence of DLL4 affecting all known structural domains of the protein are associated with AOS and are predicted to bring about DLL4 loss of function (171). Although DLL4 protein is expressed in arteries and is required for vascular development, only a minority of subjects with DLL4 mutations present with cardiovascular defects (172). Autosomal recessive mutations in EOGT cause impaired O-GlcNAc activity and are associated with AOS (173–175). The O-GlcNAc transferase acts on EGF-like domain-containing proteins and glycosylates Notch1 and Dumpy, which, like Notch, plays a critical role in cell-to-cell interactions, and these might become perturbed in AOS (176).

ARHGAP31 is a member of the Rho GAP family of proteins known to inactivate the Rho GTPases, Cdc42 and Rac1, and heterozygous gain-of-function mutations in ARHGAP31 are associated with AOS (177). The truncation of ARHGAP31 results in augmented GTPase activity, down-regulation of Cdc42, and impaired cell proliferation, migration, and survival (178, 179). ARHGAP31 is expressed during development in limb buds, cranium, and cardiac structures, tissues affected in AOS. DOCK6 protein increases the availability of the active guanosine triphosphate-bound form of Cdc42 and Rac1. Recessive DOCK6 homozygous mutations, lacking DOCK6 functional domains and presumed to be null, have been reported in two individuals afflicted by AOS (173, 180).

2. Alagille syndrome

Alagille syndrome is an autosomal dominant disease that presents with cardiovascular defects including tetralogy of Fallot, abnormalities of the craniofacial skeleton and vertebrae, cholestatic liver disease due to bile duct atresia, and kidney anomalies causing renal failure (181). Failure of the vertebrae to fuse ventrally during development causes a characteristic “butterfly” appearance in radiographic images (182) (Figure 5). Craniofacial developmental abnormalities cause craniosynostosis and characteristic facial features, including broad nasal bridge, deep set eyes, pointed chin, prominent forehead, and triangular facies. Patients affected by Alagille syndrome also have short stature and digit abnormalities and may present with osteoporosis, considered to be secondary to liver failure and malnutrition.

Figure 5.

Alagille syndrome butterfly vertebrae. In A and B, arrows point to butterfly vertebrae identified in two patients with Alagille syndrome. Note more severe clefting of vertebrae in patient on the left. [Reproduced from I. D. Krantz et al: Alagile syndrome. J Med Genet. 1997;34:152–157 (316), with permission. © BMJ Publishing Group Ltd.].

Alagille syndrome is associated with loss-of-function mutations of JAG1, and they include total gene deletions, protein-truncating mutations caused by insertions, deletions, and nonsense mutations (183–186). Missense mutations can lead to impaired Notch signaling and impaired processing of JAG1 with accumulation of the protein (185). These mutations are distributed across the extracellular and intracellular domains of the protein. De novo JAG1 mutations leading to haploinsufficiency cause Alagille syndrome, although some mutations play a dominant negative role and impair Notch signaling (186). Rarely, mutations of NOTCH2, either in isolation or in conjunction with mutations of JAG1, have been reported (187, 188).

Null mutations of Jag1 or inactivation of Notch2 in mice result in embryonic lethality. Jag1 null mice have defective embryonic and yolk sac vasculature and die of hemorrhage in early embryogenesis (189). Notch2 null mice succumb to cardiac and renal developmental defects causing hypoplastic kidneys (32). Interestingly, the combined heterozygous inactivation of Jag1 and a hypomorphic Notch2 allele in mice recapitulate features of Alagille syndrome, suggesting that the inactivation of these genes is responsible for the clinical manifestations (190). Inactivation of Jag1 selectively in cells of the cranial neural crest phenocopies the abnormalities of the craniofacial skeleton that characterize Alagille syndrome (191). In accordance with the role of Notch2 in renal development, patients with Alagille syndrome associated with NOTCH2 null mutations frequently present with hypoplastic kidneys and renal insufficiency (188).

3. Spondylocostal and spondylothoracic dysostosis

Spondylocostal dysostosis and spondylothoracic dysostosis are characterized by vertebral segmentation defects and rib anomalies secondary to defective somitogenesis (192) (Figure 6). Dominant, recessive, and sporadic mutations of genes encoding various components of the Notch signaling pathway are associated with these disorders (193, 194). Mutations of DLL3, leading to the translation of a truncated or misfolded protein of this Notch ligand, are found in 20–25% of affected individuals (195, 196). Interestingly, inactivation of Dll3 in mice recapitulates the manifestations of spondylocostal dysostosis (196). Spondylothoracic dysostosis, which is observed mostly in people of Puerto Rican descent, is associated with a mutant mesoderm posterior bHLH transcription factor (MESP) 2 allele (197). MESP2 is a Notch target gene, which encodes a bHLH transcription factor critical for somitogenesis, and Mesp2 null mice exhibit vertebral developmental defects (198, 199). Accordingly, individuals displaying abnormal segmentation of the thoracic vertebrae, typical of spondylocostal dysostosis, harbor homozygous nonsense mutations of MESP2 (200). Hes7, a Notch target gene, regulates the transcription of Lfng which, by regulating the glycosylation of Notch, changes the affinity of Notch receptors for its ligands (42). Lfng-mediated modifications of Notch enhance its ability to be activated by Dll1 but reduce Notch1 signaling in the presence of Jag1. Inactivation of Lfng or Hes7 in mice leads to abnormal development of the rib cage and vertebral column, and homozygous missense loss-of-function mutations in LFNG and HES7 are associated with spondylocostal dysostosis in humans (68, 201–204).

Figure 6.

Spondylocostal dysostosis due to homozygous mutations in DLL3. A, All vertebrae show abnormal segmentation, and the ribs show irregular points of fusion along their length. However, there is an overall symmetry to the thoracic cage. (Image courtesy of Yanick Crow.) B, Because of the similarity to smooth, eroded pebbles on a beach, the authors of the original publication suggested calling the radiological appearance the “pebble beach” sign. [Reproduced from P. D. Turnpenny, et al: Abnormal vertebral segmentation and the notch signaling pathway in man. Dev Dyn. 2007;236(6):1456–1474 (194), with permission. © John Wiley & Sons, Inc.]

B. Skeletal congenital diseases associated with gain-of-Notch function

1. Brachydactyly

Brachydactylies consist of diverse congenital disorders, often associated with disturbances in the BMP/growth differentiation factor and hedgehog signaling pathways. Brachydactyly is an autosomal recessive disorder characterized by bilateral preaxial brachydactyly or shortening of the digits of the hands and feet (205, 206). Affected individuals also display facial dysmorphism, dental anomalies, sensory hearing loss, and growth, motor, and mental retardation. A null allele of chondroitin sulfate synthase (CHSY)1 was discovered in members of a Jordanian family diagnosed with brachydactyly presenting with symmetric bilateral preaxial brachydactyly, short stature, micrognathia, and learning disabilities (Figure 7) (207). The disease was inherited as autosomal recessive, and a nonsense mutation secondary to a 1-kb deletion causing a frameshift in CHSY1 was reported.

Figure 7.

Phenotypic characteristics of syndromic recessive brachydactyly caused by a frameshift mutation of CHSY1. C, X-ray radiograph of male proband's right hand when he was 8 years of age, showing partial duplications of proximal phalanges in digits 1, 2, and 3 (inset: hand photograph). D, X-ray radiograph and pictures of male proband's right foot showing severe skeletal anomalies. The big toe exhibits short and duplicated metatarsals and proximal phalanges. The second and fourth proximal phalanges are duplicate as well. [Reproduced from J. Tian et al: Loss of CHSY1, a secreted FRINGE enzyme, causes syndromic brachydactyly in humans via increased NOTCH signaling. Am J Hum Genet. 2010;87(6):768–778 (207), with permission. © American Society of Human Genetics.]

Chondroitin sulfate is a linear polysaccharide modified with sulfated residues at various positions, and its biosynthesis is catalyzed by six glycosyltransferases, including CHSY1 (208, 209). CHSY1 has a carboxyl-terminal type A glycosyl transferase catalytic domain and an N-terminal fringe domain. Down-regulation of Chsy1 results in enhanced Notch signaling, and cultured fibroblasts from individuals affected by brachydactyly exhibit up-regulation of JAG1 and Notch activation (207). Down-regulation of Chsy1 in zebrafish causes impaired pectoral fin and skeletal development and enhanced Notch signaling (207). These observations suggest that excessive activation of Notch signaling is responsible for the disease. However, the Chsy1 inactivation in the mouse exhibits a more complex phenotype. Chsy1 null mice display brachypodism, patterning defects in distal phalanges, chondrodysplasia, and osteopenia. In the mouse model, there is a reduction in chondroitin sulfation and a shift in cell orientation, and the expression of growth differentiation factor 5, a member of the BMP family of proteins, is altered (210). Importantly, murine Chsy1 null cells do not exhibit alterations in the expression of Notch1, Notch2, Jag1, and Jag2 or Notch target genes Hey1, Hey2, and Hes1. These observations suggest that Notch signaling is not modified by the down-regulation of Chsy1 in the mouse. An additional study has identified various CHSY1 loss-of-function mutations in brachydactyly, although abnormalities in Notch signaling were not reported (211).

2. Hajdu-Cheney syndrome

Hajdu-Cheney syndrome (HCS) is a rare, inherited disease characterized by acroosteolysis of the hands and feet and developmental defects of bones, teeth, and joints causing distinctive craniofacial and skull changes, osteoporosis, and short stature (212–218). The disease was first described by Hajdu in 1948 in a 37-year-old accountant who presented with osteoporosis and acroosteolysis and died of severe neurological complications (217).

HCS has autosomal dominant inheritance, although sporadic cases are common. Patients with HCS exhibit prominent skeletal features including facial dysmorphism and craniofacial defects, such as micrognathia, midface flattening, and dental abnormalities (Figure 8). There is high clinical variability as well as evolution of the phenotypical clinical manifestations. Some signs of the disease present as early as in the first 2 years of life, and others become more evident in young children and adolescents (215). Eventually, adult patients develop classic coarse facial features. Acroosteolysis is frequently observed and can present with symptoms of inflammation, including pain and swelling, and lysis of the phalanges causes short and broad digits. Spinal abnormalities include compression fractures, kyphosis, and scoliosis. Long bone deformities, such as serpentine fibula, are also noted (214, 216). Abnormal dental eruptions, decay, and premature loss of teeth are common. Platybasia and basilar invagination can result in severe neurological complications, including hydrocephalus, central respiratory arrest, and sudden death. Some patients present with renal cysts or polycystic kidneys, or with cardiovascular defects, including septal defects and valve abnormalities (219, 220).

Figure 8.

Identification of NOTCH2 mutations in individuals with Hajdu-Cheney syndrome. A, Facial dysmorphy with micrognathia, thick eyebrows, long philtrum, hypertelorism, and low-set and posteriorly rotated ears. Written consent to publish photographs of this individual was obtained by the authors of the original publication. Phalanges radiograph showing acro-osteolysis of all distal phalanges. Skull radiograph showing characteristic findings of HCS, including platybasia, thickened occipital bone, open sutures, and wormian bones. [Reproduced from B. Isidor et al: Truncating mutations in the last exon of NOTCH2 cause a rare skeletal disorder with osteoporosis. Nat Genet. 2011;43(4):306–308 (221). © Nature Publishing Group.]

HCS is associated with nonsense mutations or deletions leading to the creation of a termination codon in exon 34 of NOTCH2 upstream of the PEST domain (221–224). Because the PEST domain contains sequences necessary for the ubiquitination and degradation of NOTCH2 in the proteasome, the mutations are predicted to lead to the formation of a truncated stable NOTCH2 protein and enhanced NOTCH2 signaling. It is of interest that somatic NOTCH2 mutations causing loss of the PEST domain exhibit enhanced Notch activation and have been identified in B-cell lymphoma, specifically in splenic marginal zone lymphoma (29, 225–227). However, there is no apparent increase in the incidence of B-cell lymphoma in HCS.

The focal osteolysis is accompanied by neovascularization, inflammation, and fibrosis (228–230). Tissue from iliac crest biopsies has been examined in a small number of cases of HCS, but variable results have not elucidated the pathogenesis of the bone loss. Mechanisms responsible for the craniofacial and cardiovascular developmental abnormalities probably relate to the effects of Notch on skeletal and cardiac development, and the short stature may be secondary to the inhibitory effects of Notch on chondrogenesis.

In recent work conducted in our laboratory, a Notch2 mutant mouse harboring a truncating mutation in exon 34 upstream of the PEST domain was created (149). The mutation reproduces the one found in HCS, and Notch2 mutant mice exhibit cancellous and cortical bone osteopenia secondary to increased osteoclast number and bone resorption. In vitro studies demonstrated enhanced osteoclastogenesis and confirmed the stimulatory effect of Notch2 on osteoclast differentiation (153). This was attributed to an induction of Nfatc1 in osteoclast precursors by the Notch2 NICD acting in conjunction with NfκB.

There are no controlled trials on the management of the osteoporosis in patients with HCS, and only anecdotal cases treated with either bisphosphonates or teriparatide have been reported. Bisphosphonates (alendronate and pamidronate) alone or in combination with teriparatide have been used, but there is no clear evidence that either therapy is beneficial (231, 232). Because in experimental mouse models of the disease Notch2 enhances osteoclastogenesis, a consideration would be the use of denosumab, in an effort to block RANKL and inhibit osteoclast formation. Enhanced Notch expression and signaling are found in humans with osteosarcoma, and long-term activation of Notch in osteoblasts causes osteosarcoma in experimental mouse models (233). These are potential concerns when considering the use of teriparatide, which is contraindicated in individuals predisposed to osteosarcoma.

NOTCH2 itself could be a future target for the treatment of HCS. Experimental modalities to control Notch signaling, including the use of antibodies to the Notch extracellular domain or its ligands, and the use of cell membrane permeable peptides that interfere with the formation of the Notch transcriptional complex could be considered (234, 235). However, some of these approaches may result in the nonspecific inhibition of all Notch isoforms in skeletal and nonskeletal tissues and may cause significant unwanted events. There are no studies reported in humans exploring approaches to block NOTCH2 signaling in HCS.

3. Lateral meningocele syndrome

Lateral meningocele syndrome (LMS), or Lehman syndrome, is a rare disorder characterized by facial anomalies, hypotonia, and meningocele with related neurological dysfunction (Figure 9) (236). The clinical features of LMS include classic craniofacial features, such as hypertelorism; high-arched eyebrows; ptosis; low-set eyes; cleft, high, and narrow palate; micrognathia; and empty sella. Affected individuals have developmental delay, intellectual disability, hypotonia, decreased muscle mass and syringomyelia, and cardiac valve abnormalities. Skeletal manifestations include short stature, scoliosis, pectus excavatum, wormian bones, thick calvariae, increased density of the base of the skull, and increased bone remodeling (237).

Figure 9.

Cranial and central nervous system features of LMS. A, Thickened cranial vault, mild dilatation of the subarachnoideal spaces, sellar arachnocele, and verticalization of the tentorium cerebellum at magnetic resonance imaging. B, Cranial vault sclerosis. C–H, Magnetic resonance images showing thickening of the neurocranium and increased distance between cortex and meninges (C) and an empty sella filled with cerebrospinal fluid (D); lateral meningocele (arrows) at the thoracic (E and F); and at lumbar metameres (G and H). [Reproduced from M. Castori et al: Late diagnosis of lateral meningocele syndrome in a 55-year-old woman with symptoms of joint instability and chronic musculoskeletal pain. Am J Med Genet A. 2014;164A(2):528–534 (317), with permission. © Wiley-Liss, Inc.]

Exome sequencing demonstrated the presence of point mutations or short deletions in exon 33 of NOTCH3, upstream of the PEST domain (238). The gain-of-function mutations lead to the translation of a truncated and stable protein, devoid of the PEST domain, and with the consequent potential to enhance Notch signaling. This mechanism is analogous to the one reported for HCS, where affected patients harbor mutations in exon 34 of NOTCH2. Indeed, subjects suffering from either syndrome can exhibit thickened skull and base of the skull, wormian bones, midface hypoplasia, hypertelorism, micrognathia, and cardiac abnormalities (239, 240). Although increased bone turnover and decreased density of bones may occur in LMS, acroosteolysis, a hallmark feature of HCS, has not been reported. The inheritance in LMS is not clear, and autosomal dominant inheritance has been suggested (237). However, most documented cases reveal de novo heterozygous truncating mutations in exon 33 of NOTCH3.

NOTCH3 is primarily expressed in smooth muscle cells of small arteries and pericytes of brain capillaries and plays a fundamental role in vascular development. Dominant mutations of the extracellular domain of NOTCH3, encoded by exons 2 to 24, that lead to an abnormal accumulation of the NOTCH3 extracellular domain in vessels are associated with CADASIL (36, 241). There is no obvious increase in Notch canonical signaling in CADASIL, and the mechanisms responsible for the disease remain enigmatic. Notch3 homozygous null mice are viable, demonstrating that Notch3 is not essential for embryonic development, although mice exhibit defects in distal arteries (35–37). The lack of a pronounced phenotype in Notch3 null mice may be related to the restricted tissue expression of the receptor to small arterioles, thymocytes, T cells, and the central nervous system (242).

There is no specific treatment for patients with LMS. γ-Secretase inhibitors prevent Notch activation, but are not specific. They inhibit all Notch isoforms, and their long-term administration is poorly tolerated due to gastrointestinal side effects (243). Antibodies to the NRR of NOTCH3 can prevent Notch activation and signaling, but their effects in vivo in experimental animals or humans are unknown (244).

C. Role of Notch in primary and metastatic bone tumors

1. Leukemias and Lymphomas

Malignancies arising from hematopoietic cells, such as T-cell acute lymphoblastic leukemia, are often associated with somatic mutations of NOTCH1 and dysregulated Notch signaling, which may cause uncontrolled cell proliferation (245). These mutations occur in the juxtamembrane NRR of NOTCH1 and are associated with constitutive activation of NOTCH1, or they occur in exon 34 upstream of the PEST domain and lead to the translation of a truncated and stable NOTCH1 protein (28, 246, 247). Although hypercalcemia and increased bone resorption may occur in acute T-cell leukemia, radiographic abnormalities of the skeleton are uncommon. These include osteolytic lesions and subperiosteal bone resorption similar to those observed in hyperparathyroidism. The mechanism appears related to the secretion of PTHrP by T cells, and not to enhanced Notch signaling (248, 249).

Somatic nonsense or truncating mutations in exon 34 of NOTCH2 upstream of the PEST domain resulting in a NOTCH2 truncated and stable protein and gain-of-NOTCH2 function are found in 4 to 8% of individuals with diffuse large B-cell lymphomas and in 25% of splenic marginal zone lymphomas (29, 225–227). A single missense mutation in the NRR of NOTCH2, similar to the activating mutations of NOTCH1 reported in T-cell acute lymphoblastic leukemia, was found in a case of splenic marginal zone lymphoma. The mutations are associated with poor clinical outcomes. The occurrence of NOTCH2 mutations in B-cell and splenic marginal zone lymphoma is in accordance with the preferential expression of NOTCH2 in these cells. Importantly, the conditional inactivation of Notch2 in cells of the hematopoietic lineage results in the loss of marginal zone B cells, demonstrating that Notch2 is indispensable for their development (250). Although somatic, the mutations in marginal zone lymphomas are analogous to the germline mutations reported in HCS. However, HCS patients are not known to be predisposed to lymphomas and do not exhibit obvious alterations in B-cell function. Despite the stimulatory effects of Notch2 in osteoclastogenesis, subjects with B-cell and splenic marginal zone lymphomas do not exhibit skeletal manifestations (251). However, chronic lymphoblastic leukemia is associated with osteolytic lesions and hypercalcemia, and precursor B-cell acute lymphoblastic leukemia is associated with bone loss and fractures (252–255). The mechanism of the bone lytic lesions most likely involves the release of cytokines by leukemia cells, and not disrupted Notch signaling.

2. Osteosarcoma

Osteosarcoma is the most common primary bone malignant tumor, but the molecular mechanisms that determine the onset and progression of osteosarcoma are largely unknown. Recent work demonstrated up-regulation of Notch receptors and Notch target genes in osteosarcoma in humans and dogs (256, 257). Because Notch receptors and ligands could become therapeutic targets in osteosarcoma, it is important to determine which receptor and ligand are up-regulated and whether Notch signal activation occurs and is responsible for the progression of the disease. Gene expression analysis in 10 human biopsies obtained from subjects with untreated osteosarcoma and compared to human bone tissue obtained during total hip arthroplasty demonstrated that NOTCH2 and JAG1 were up-regulated in most osteosarcoma samples; NOTCH1 and DLL1 were up-regulated in one patient (257). These results are in contrast with those obtained in osteosarcoma tumor samples from the Pediatric Cooperative Human Tissue Network Biopathology Center. In this study, samples were compared to commercially available control human osteoblast RNAs, and NOTCH1 and JAG1 (but not NOTCH2) mRNA were increased in osteosarcoma (258). Notch target genes, particularly HEY1 and HEY2, were increased in osteosarcoma samples in both studies, verifying activation of Notch canonical signaling. The growth of osteosarcoma cell lines and of xenografts in vivo and CCND1 and CCNE1 and CCNE2 gene expression were suppressed by γ-secretase inhibitors, to prevent Notch activation, and by lentiviruses delivering dominant negative MAML, to prevent Notch-dependent transcription. This suggests that Notch controls the growth of osteosarcoma cells (256). Notch signaling might also play a role in tumor invasiveness because osteosarcoma cell lines with the ability to metastasize have higher levels of expression of NOTCH1, NOTCH2, and DLL1 mRNA and of the Notch target gene HES1 (259).

Recent work revealed that the conditional induction of the Notch1 NICD in mature osteoblasts caused the spontaneous development of osteosarcoma in mice at 5 to 14 months of age. There was complete penetrance in terms of tumor induction because all mice surviving to adulthood developed osteosarcoma (233). The initiation and tumor progression were accelerated by the loss of p53 and required activation of Notch canonical signaling because it was not observed in the context of an Rbpjκ inactivation. These findings reveal an important role for Notch canonical signaling in the initiation and possibly invasive potential of osteosarcoma. It is conceivable that somatic activating mutations of Notch receptors, ligands, or target genes play a role in the onset of the tumor. As a result, components of the Notch signaling pathway may become future therapeutic targets in osteosarcoma.

3. Skeletal metastases in carcinoma of the breast and prostate

Bone metastases are major complications of carcinoma of the breast, and Notch appears to play a role in selected interactions between osteoblasts and metastatic cells. Human bone marrow-derived osteoblasts induce the expression of NOTCH3 and its ligand JAG1, as well as cancer cell colony formation in human carcinoma of the breast cell lines. Systemic inoculation of carcinoma of the breast cell lines, or their direct injection into the bone marrow of athymic mice, induced the formation of osteolytic bone metastases, and down-regulation of NOTCH3 reduced their skeletal metastatic potential (260). This suggests that NOTCH3 plays a role in the invasiveness of breast cancer cells. In a subsequent study, expression of JAG1 in mammary tumor cells was shown to correlate with tumor load and with the ability of tumors to form bone metastases in mice (261). Tumor cells expressing JAG1 activated Notch signaling, which induced IL-6 in osteoblasts, and enhanced osteoclastogenesis and the formation of osteolytic bone metastases. As the lytic lesions formed, they released TGFβ from the bone matrix, and TGFβ in turn up-regulated JAG1, causing further activation of Notch signaling and creating a positive feedback loop favoring the metastatic potential of the tumor. Down-regulation of JAG1 by RNA interference or administration of γ-secretase inhibitors, to prevent Notch activation, decreased the osteolytic potential in this experimental model of carcinoma of the breast.

Carcinoma of the prostate frequently metastasizes to bone, and characteristically these metastases induce osteoblastic woven bone formation and variable degrees of osteoclastic bone resorption (262). A variety of factors and cytokines released by carcinoma of the prostate cells accounts for the osteoblastic response. NOTCH1 and JAG1 are expressed by carcinoma of the prostate, and their expression is associated with the metastatic potential and recurrence of the tumor (263, 264). This would suggest that dysregulation of Notch signaling may be associated with the metastatic potential of the tumor. Indeed, down-regulation of NOTCH1 by RNA interference in human prostate cancer cells decreased their invasiveness (265). This was attributed to a decrease in the expression of matrix metalloprotease (MMP) 9 and of urokinase plasminogen activator receptor. Immunohistochemical analysis of invasive carcinoma of the prostate revealed that the expression of NOTCH1 is associated with the expression of molecular markers of the epithelial mesenchymal transition (EMT), which occurs during cancer progression and tumor invasiveness (266). EMT markers were more intensely expressed in the invasive tumor front (267). These observations would suggest that Notch plays a role in EMT and tumor aggressiveness and in the invasive potential of carcinoma of the prostate. This contention is supported by a study examining low- and high-grade prostate carcinomas with localized disease. Cancer cells were isolated by laser capture microdissection, and expression profiling revealed up-regulated JAG2, NOTCH3, and HES6 transcripts in high-grade tumors (268, 269). HES6, however, acts as an inhibitor of HES1, and it remains to be determined whether the reported changes in expression indicate an induction or a suppression of NOTCH signaling (69).

4. Notch and the EMT

EMT is a process that implies loss of cell polarity and cellular adhesion and the acquisition of a migratory phenotype leading to mesenchymal cell function from the epithelium. EMT takes place during embryonic development as well as during tumor progression and metastases (270). In addition, EMT is a prominent cellular process that occurs during tissue fibrosis. Loss of the epithelial adhesion protein E cadherin is associated with EMT. Notch induces Snail1, an E-cadherin repressor, and as such it plays a role in EMT in cardiac development and oncogenic transformation and tumor invasiveness (271). The suppression of E cadherin transcription may lead to decreased cellular adhesion, enhanced cellular migration, and tumorigenic potential (272). During embryonic development, multiple signaling pathways including those dependent on fibroblast growth factor, BMP, TGFβ, Wnt, and Notch signaling regulate the EMT. TGFβ is a major determinant of EMT, and by inducing Jag1, TGFβ can activate Notch signaling and both pathways can act in conjunction during tumor progression and invasiveness (273, 274).

D. Fracture repair and Notch signaling

Fracture healing is a complex regenerative process that generates connective tissue and new bone after a fracture. The healing of a fracture occurs either by intramembranous bone formation, when the bone is mechanically stable, or by endochondral ossification, when bones are unstable (275). The contribution of cells from multiple lineages is important for fracture healing, and lineage tracing experiments identified cells expressing α smooth muscle actin, a marker of mesenchymal progenitor cells, in the periosteum during the early phases of fracture healing. In this cell population, 2 to 6 days after a fracture, the expression of Notch ligands, receptors, and target genes is suppressed (276). However, other investigators have reported up-regulation of Notch receptors, ligands, and Notch target genes in the fracture callus 5, 10, and 20 days after a fracture of the tibia in the mouse (277). The data, however, are difficult to interpret because determination of transcript levels in fracture calluses was relative to transcript levels obtained from diaphyseal bone. The seemingly contradictory results may represent differential Notch activation in subsets of cell populations involved in fracture healing. Because of the known inhibitory effects of Notch on osteogenesis and chondrogenesis, down-regulation of Notch signaling might be a requirement for the fracture-healing process to occur (278). Indeed, transient and systemic administration of γ-secretase inhibitors, to prevent the activation of Notch signaling, accelerated fracture healing in an open midshaft tibial fracture mouse model (279). It is of interest that in this study only a transient suppression of Notch target genes was demonstrated, but γ-secretase inhibition resulted in an increase in callus volume, cartilage, and bone area. There was a concomitant enhancement of the biomechanical properties of bone in animals treated with the γ-secretase inhibitor. Whereas these results suggest a negative role of Notch in fracture healing, it is important to note that γ-secretase inhibitors are not Notch-specific and suppression of other signals might have contributed to the effects observed. Moreover, work from the same laboratory reported that down-regulation of Rbpjκ in Prx1-expressing cells results in nonunion fractures (280). This would suggest that Notch signaling is necessary for fracture repair, but the mechanisms and signals affected by the down-regulation of Rbpjκ are not clear and may involve a variety of Rbpjκ target genes.

Cell lineage tracing studies have demonstrated that osteoblast precursors move into the fracture site along invading blood vessels, confirming the importance of vascularization for proper fracture healing (140). Notch signaling plays an important and complex role in vascular development and physiology, and recent work has demonstrated that Notch signaling promotes endothelial cell proliferation and vessel growth in long bones (281–283). However, it is not known whether this angiogenic effect of Notch is manifested during fracture healing. If this were the case, selective inhibition of Notch signaling may or may not be beneficial to the healing process. Interestingly, studies using a dominant negative MAML1 expressed under the control of the msh homeobox 1 promoter to inhibit Notch transcriptional activation demonstrated that Notch signaling is required for successful fracture healing (284).

E. Notch and osteoarthritis

Osteoarthritis is a degenerative joint disease characterized by an aberrant chondrocyte phenotype, cartilage degeneration, inflammation, and fibrosis (285). As osteoarthritis progresses, articular chondrocytes proliferate and acquire aspects of chondrocyte hypertrophy (286, 287). Human osteoarthritic chondrocytes express MMP13, and constitutive activation of Mmp13 in murine articular chondrocytes causes spontaneous joint degeneration, suggesting that this metalloprotease contributes to the cartilage degradation in osteoarthritis (288, 289). Inflammatory molecules, such as IL-1, IL-6, and TNFα are detected in the synovial fluid of osteoarthritic joints, and exposure to these molecules causes joint damage in mice, indicating that inflammation plays a critical role in the progression of the disease (290–292).

Chondrocytes harvested from osteoarthritic cartilage are enriched in cells expressing NOTCH1 in relation to chondrocytes from normal tissue, and NOTCH1 is detected in the clusters of proliferating chondrocytes that characterize the early stages of the disease (293, 294). Levels of NOTCH1, JAG1, and HES5 are higher in osteoarthritic than in normal cartilage, and the NICD of NOTCH1 and NOTCH2 localize in the nucleus of osteoarthritic, but not of normal, chondrocytes, demonstrating NOTCH activation in joints affected by the disease (293, 295).

To recapitulate the conditions of enhanced NOTCH activity observed in human osteoarthritic chondrocytes, the Notch1 NICD was overexpressed in chondrocytes in vitro and in vivo (296–298). Activation of Notch signaling in primary chondrocytes phenocopied the alterations in gene expression observed in osteoarthritic chondrocytes, suggesting that Notch is detrimental to articular chondrocyte function and cartilage integrity (296). Overexpression of the Notch1 NICD in primary chondrocyte cultures and ATDC5 cells induced expression of IL-6, but not of IL-1 or Tnf (297). IL-6 mediated selected effects of Notch activation, suggesting that IL-6 is responsible for the inflammatory processes observed in Notch-induced osteoarthritis (298). Overexpression of Hes1 in ATDC5 cells recapitulated the effects of NICD overexpression, although this implied that Hes1 acted as an activator, and not as a suppressor, of transcription, a commonly accepted property of Hes1 (120, 299). In accordance with the in vitro observations, sustained overexpression of the Notch1 NICD in vivo in postnatal chondrocytes caused joint degeneration in mice. These effects were possibly due to increased levels of IL-6 and to the consequent induction of signal transducers and activators of transcription 3 and MAPK signaling (298). In agreement with these in vivo findings, inactivation of either Rbpjκ or Hes1 in chondrocytes protected against osteoarthritis induced by surgical destabilization of the medial meniscus (120, 293). It is of interest that under selected conditions, Rbpjκ inactivation causes degeneration of the knee, indicating that basal levels of Notch signaling are required to preserve joint structure, or that Rbpjκ has distinct functions from Notch receptors in articular chondrocytes (300). Overall, the observations suggest that activation of NOTCH in articular chondrocytes is detrimental and may contribute to the development and progression of osteoarthritis.

VII. Controlling Notch Signaling

There is considerable interest to control Notch activity in the treatment of disorders associated with dysregulated NOTCH signaling. This is particularly important in selected malignancies, where Notch is believed to play a role in tumorigenesis and tumor invasion. The approach to down-regulate Notch signaling has been diverse and includes the use of biochemical inhibitors of Notch activation, antibodies to Notch receptors or their ligands, and small permeable molecules that prevent the formation of a Notch/Rbpjκ/Maml ternary complex. However, the generalized inhibition of Notch signaling is not without unwanted consequences and has been associated with the development of skin cancer, vascular tumors, and hepatotoxicity (235, 301). These clinical observations are in agreement with experimental mouse models of conditional Notch1 heterozygous inactivation reporting the widespread appearance of vascular tumors causing hemorrhage and lethality (302). Therefore, although the control of Notch activity might be regarded as a promising therapeutic alternative, it needs to be considered in the context of potential serious unwanted events.

A. Biochemical inhibitors of Notch signaling

1. γ-Secretase inhibitors

γ-Secretase inhibitors were developed to reduce amyloid-β protein aggregates in Alzheimer's disease (303). γ-Secretase inhibitors block the cleavage of the Notch receptor induced by Presenilins, and as a result the NICD is not processed and the Notch transcriptional response is precluded (304). γ-Secretase inhibitors have been examined in clinical trials to test their effects on Notch inhibition in cancer. A problem with these compounds has been the lack of specificity and complications associated with their chronic use, such as goblet cell metaplasia-induced gastrointestinal intolerance and skin tumors (305, 306). Another concern is the increased risk of skin cancer observed in a clinical trial testing a γ-secretase inhibitor for its efficacy in the management of Alzheimer's disease (307). About 90 substrates of the γ-secretase complex are known, so that studies using γ-secretase inhibition as a means to suppress Notch signaling must be interpreted with caution (308). Moreover, γ-secretase inhibitors prevent the indiscriminate activation of all Notch receptors, and different Notch receptors carry out different and sometimes opposing functions in the same cell type or organ.

2. Thapsigargicin

Thapsigargicin is an analog of thapsigargin, a natural product inhibitor of the sarco/endoplasmic reticulum Ca²⁺-ATPase. Inhibition of sarco/endoplasmic reticulum Ca²⁺-ATPase precludes the maturation and folding of Notch and its exit from the endoplasmic reticulum, leading to decreased levels of Notch receptor at the cell surface (309, 310). Thapsigargicin is more effective at inhibiting the activity of NOTCH1 mutants causing ligand-independent activation of NOTCH and associated with T-cell acute lymphoblastic leukemia.

B. Inhibitors of the Notch transcription complex formation

To activate the transcription of target genes, Notch forms a complex with Maml and Rbpjκ, and the delivery of dominant negative MAML1 mutant antagonizes Notch signaling (311). Synthetic small cell permeable molecules that target critical protein-protein interactions prevent the assembly of an active Notch transcriptional complex (234). These small peptides penetrate the cell, bind the NICD/Rbpjκ complex, and compete for binding with Maml-suppressing Notch signaling. This results in a nonspecific inhibition of all Notch isoforms, but the long-term efficacy and safety of small molecules that inhibit the assembly of the Notch transcriptional complex is unknown.

C. Antibodies to Notch and Notch ligands

Antibodies directed to either a Notch isoform or a Notch ligand inhibit signal activation and biological effects. To target specific Notch receptors, antibodies to the extracellular juxtamembrane NRR of Notch1, Notch2, and Notch3 have been developed (244, 312). The targeting of this region prevents the initial cleavage and activation of specific Notch isoforms. The antibodies have specific neutralizing activity for individual receptors and do not exhibit cross-reactivity with other Notch paralogs. The long-term safety of these antibodies is not known, but in short-term studies, anti-Notch1 antibodies disrupted retinal vascular development, and the concomitant use of anti-Notch1 and Notch2 antibodies in mice resulted in weight loss and gastrointestinal toxicity.

Antibodies against the extracellular domains of Notch2 and Notch3 block interactions between these two Notch receptors and their ligands Jag1 and Dll4, oppose Notch2/3 activity, and have a favorable antitumor effect (313). Targeting members of the Jagged or Dll family of ligands could prove valuable to block Notch activation in a cell environment where a specific Notch ligand is expressed (314). Neutralizing antibodies for the Notch ligand Dll4 affect endothelial cell proliferation, but their administration in vivo resulted in vascular neoplasms and severe liver pathology (301, 315).

VIII. Conclusions

Notch regulates the differentiation and function of cells of the osteoblastic and osteoclastic lineages and plays a critical role in skeletal development, chondrogenesis, osteoblastogenesis, and osteoclastogenesis. Importantly, the effects of Notch are cell-context dependent. In immature cells of the osteoblastic lineage, Notch suppresses cell differentiation by inhibiting Wnt signaling and by interacting with Runx2 to prevent osteoblast maturation (Figure 10). In mature osteoblasts, Notch represses their differentiated function, whereas in osteocytes, Notch decreases the expression of Sost with a consequent increase in Wnt signaling. The balance of these activities is essential to maintain cellular homeostasis and bone remodeling. Notch1 inhibits osteoclastogenesis, whereas Notch2 by interacting with NfκB induces Nfatc1 and osteoclast differentiation.

Figure 10.

Notch signaling and regulation of bone remodeling. DSL ligands induce cleavage of Notch receptors and the generation of the NICD. Activation of Notch1 in osteoblast precursors suppresses osteoblastogenesis by inhibiting osterix (Osx), runt-related transcription factor (Runx)2 and cytosolic β-catenin, thereby impairing bone formation. The Notch2 NICD in osteoclast precursors associates with Nf-κB) and induces Nf of activated T cells (Nfatc)1 transcription and osteoclastogenesis. Impaired bone formation and increased bone resorption lead to a decrease in bone mass (left). Activation of Notch1 in osteoblastic cells induces osteoprotegerin (Opg), an inhibitor of Rankl, leading to suppressed cancellous bone resorption. Activation of Notch1 in osteocytes suppresses sclerostin and dickkopf (Dkk)1, enhances Wnt signaling, and increases cortical bone formation. Decreased cancellous bone resorption and enhanced cortical bone formation lead to an in increase in bone mass (right).

Gain- and loss-of-function mutations of various components of the Notch signaling pathway result in a variety of congenital disorders with significant skeletal manifestations, reaffirming the role of Notch in skeletal development and function. Alterations in Notch signaling are associated with osteosarcoma and with the skeletal metastatic potential of carcinoma of the breast and the prostate, and Notch plays a role in the development of osteoarthritis and in the early phases of fracture healing. Controlling Notch signaling could prove useful in diseases of Notch gain-of-function and in selected malignancies. However, clinical data are not available, and toxicity of the agents may preclude their use in humans. It will be difficult to restore Notch activity in diseases associated with loss-of-Notch function.