PDF hosted at the Radboud Repository of the Radboud University Nijmegen

The following full text is a publisher's version.

For additional information about this publication click this link. http://hdl.handle.net/2066/32614

Please be advised that this information was generated on 2021-10-03 and may be subject to change. Identification of Novel Regulators Involved in Early Phase Osteoblast Differentiation

Diana Sylvia de Jong

Identification of Novel Regulators Involved in Early Phase Osteoblast Differentiation

een wetenschappelijke proeve op het gebied van de Natuurwetenschappen, Wiskunde en Informatica

Proefschrift

ter verkrijging van de graad van doctor aan de Radboud Universiteit Nijmegen op gezag van de Rector Magnificus prof. dr. C. W. P. M. Blom, volgens besluit van het College van Decanen

in het openbaar te verdedigen op woensdag 26 oktober 2005, des namiddags om 1.30 uur precies

door

Diana Sylvia de Jong

Geboren op 27 januari 1976 te Sneek Promotores: prof. dr. W. Olijve prof. dr. E. J. J. van Zoelen

Manuscriptcommissie: prof. dr. G. J. M. Martens dr. P. M. van der Kraan

Reproduction: PrintPartners Ipskamp, Enschede ISBN10: 9090196943 ISBN13: 9789090196947

The research presented in this thesis was performed at the department of Applied , Faculty of Sciences at the Radboud University of Nijmegen. The studies were financially supported by the Netherlands Institute for Earth and Life Sciences (NWO-ALW) grant 809.67.023. Financial support for the reproduction was provided by Organon Nederland BV and the Radboud University of Nijmegen.

NFO f c r g a n o n

N etherlands Organisation fo r Scientific Research ^ Onmogelijk is alleen dat wat niet uitgeprobeerd is

Impossible is only what has not been tried yet Contents

page Chapter 1 Introduction in the Control of BMP Signaling on Osteoblast Differentiation 7 1.1 Bone 8 1.1.1 Embryonic development 8 1.1.2 Osteoblast differentiation, transcriptional regulation and marker genes 9 1.1.2.1 Stages of osteoblast differentiation 9 1.1.2.2 Transcriptional regulation and marker genes of Self Renewal 9 1.1.2.3 Transcriptional regulation of Lineage Commitment, Prolife­ ration and Maturation 9 1.1.2.4 Marker genes of Lineage Commitment, Proliferation and Maturation 11 1.1.2.5 Transcriptional regulation and marker genes of Mineralisation 12 1.1.3 Hormonal regulation o f osteoblast differentiation 13 1.1.4 Polypeptide growth factors in bone 16

1.2 Bone Morphogenetic Proteins 17 1.2.1 BMPs in mesenchymal stem cell differentiation and osteoclastformation 17 1.2 .2 BMP receptors and signaling 18

1.3 Regulators of BMP signaling 22 1.3.1 Extracellular regulation o f BMP activity 22 1.3 .2 Regulation o f BMPs at the membrane level 24 1.3.3 Intracellular regulation o f BMP activity 2 5 1.3.3.1 Inhibitory Smads 25 1.3.3.2 Proteins binding signaling Smads 25 1.3.3.3 Degradation o f signaling Smads 29

1.4 Other signaling routes in osteoblast differentiation and interactions 30 1.4.1 MAP-kinases 30 1.4.2 Wnt-signaling 33 1.4.3 The Notch pathway 3 5

1.5 Clinical skeletal disorders associated with osteoblast differentiation 37

1.6 Aim of the thesis 39

References 41

- 4 - page Chapter 2 Microarray analysis of Bone Morphogenetic Protein, Transforming Growth Factor P, and Activin Early Response Genes During Osteoblastic Cell Differentiation 57

Chapter 3 Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation 71

Chapter 4 Regulation of Notch signaling genes during BMP2- induced differentiation of osteoblast precursor cells 91

Chapter 5 The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation 101

Chapter 6 General Discussion 121 6.1 Regulators of BMP-2-induced osteoblast differentiation 122 6.2 BMP-induced gene expression profiles in diagnosis, drug target discovery and therapy 124 6.2.1 Diagnosing bone diseases 124 6.2.2 Drug target discovery 125 6.2.3 Tissue engineering therapies 128 6.2.3.1 Stem cells in bone tissue engineering 129

Chapter 7 Full color figures 133

Chapter 8 Summary ra Samenvatting 145 Summmary 146 Samenvatting 148

Dankwoord 151

List of Publications 153

Curriculum Vitae 155

- 5 - - 6 - 1

Introduction in the Control of BMP Signaling on Osteoblast Differentiation

Submitted to the International Journal of

Diana S. de Jong1, Wiebe Olijve1’2 and Everardus J.J. van Zoelen3

1) Dept. of Applied Biology, Radboud University Nijmegen, Nijmegen, the Netherlands 2) Organon Research Centre USA, Cambridge, USA 3) Dept. of Cell Biology, Radboud University Nijmegen, Nijmegen, the Netherlands Chapter 1

1.1 Bone

1.1.1 Embryonic development

The formation of the skeleton and the bones of the skull is one of the last events in the entire process of embryonic development. In mice, the development of the major organ systems is being initiated around embryonic day 8.5 (E8.5), whereas the initiation of bone development occurs only around E12. At E14.5 ossification is initiated and at E16.5 bone formation takes place in almost all major ossification centers of the skeleton and the skull. However, full ossification is only achieved approximately two weeks after birth (1). Bone development can occur through two mechanisms of ossification: intramembranous ossification and endochondral ossification (Figure 1). In the process of intramembranous ossification, bone growth begins with the condensation of mesenchymal cells that expand and form a membranous structure. Subsequently, these mesenchymal cells differentiate directly into osteoblasts in the primary centers of ossification. Calcification by the mature osteoblasts is not coordinated but occurs in irregularly distributed patches. These mature osteoblasts form the endosteum, while at the periphery mesenchymal cells continue to differentiate, following the same steps, to form the periosteum (Figure 1A). During endochondral ossification, the condensed embryonic mesenchymal cells first differentiate into chondroblasts, which secrete the cartilaginous matrix (cartilage model). Subsequently, the chondroblasts proliferate, differentiate into chondrocytes and eventually become hypertrophic and undergo programmed cell death. Osteoclasts then resorb the cartilage matrix and adjacent mesenchymal stem cells differentiate to osteoblasts, which form new bone in the primary center of ossification. Subsequent ossification occurs in the same fashion in secondary centers of ossification (Figure 1B) (2).

A Intramembranous Ossification Primary centers of ossification

Mesenchymal Membranous condensation structure

B Endochondral Ossification . Secondary center Chondrocyte Epiphysis \ of ossification hypertrophy " Growth plate-- -Cancellous bone

Cartilage model ------Bone marrow

1' Diaphysis -Endosteum & I ~ Periosteum Mesenchymal condensation Metaphysis* Cortical bone

Figure 1. Bone Development. Schematic diagrams of the initial phases of (A) Itramembranous or (B) Endochondral ossification. See text for details. (Adopted from Baron (2003), see ref. 2)

- 8 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

The majority of the long bones are derived from mesenchymal cells of the somites and are formed through endochondral ossification, while the majority of the bones in the skull are derived from mesenchymal cells from the neural crest and are formed by the process of intramembranous ossification. However, no lineage relationship exists between the origin of the mesenchyme and the type of bone that it forms (1).

1.1.2 Osteoblast differentiation, transcriptional regulation and marker genes

1.1.2.1 Stages o f osteoblast differentiation Mesenchymal stem cells (MSCs) have the potential to differentiate into various cell types including osteoblasts, chondroblasts, myoblasts, adipocytes and fibroblasts, depending on the signaling cascades activated during the initial phase of differentiation (for review see ref. 3). In the process of osteoblast differentiation, several stages can be distinguished (Figure 2): self renewal, in which the MSCs maintain an undifferentiated ; lineage commitment, in which the MSCs receive the primary signal to differentiate to osteoprogenitor cells; proliferation, in which a subset of the osteoprogenitor cells differentiate to pre-osteoblasts; matrix maturation, in which the pre­ osteoblasts differentiate to mature osteoblasts that secrete the extracellular matrix; mineralization, in which hydroxyapatite crystals are deposited in the extracellular matrix and osteoblasts differentiate either into osteocytes, that are completely embedded in the mineralized matrix and support the bone structure, or in lining cells that form the protective bone surface; and finally apoptosis, in which the osteocytes undergo programmed cell death. The osteoclast, which is responsible for bone resorption, is derived from cells of the mononuclear/phagocytic lineage (4) and will not further be discussed here.

1.1.2.2 Transcriptional regulation and marker genes o f Self Renewal Factors maintaining the undifferentiated state of MSCs include LIF (leukemia inhibitory factor) (for review see ref. 5), Sca-1/Ly6a (stem cell antigen 1/lymphocyte antigen 6) (6), and interleukin 18 (IL-18) (7). Moreover, transcriptional regulators, including members of the helix-loop- helix (HLH) family such as Twist and the Id (inhibitor of differentiation) proteins, have been proposed as mediators in maintenance and expansion of the mesenchymal progenitor population (8­ 10) (Figure 2). In general, factors promoting proliferation of the MSC will be inhibitory for the final differentiation stages.

1.1.2.3 Transcriptional regulation o f Lineage Commitment, Proliferation and Maturation Many extracellular factors, including the bone morphogenetic proteins (BMPs), are known to induce differentiation of MSCs to osteoprogenitor and osteoblast cells, and their action will be discussed elsewhere in this chapter. Important transcriptional regulators of the differentiation of osteoprogenitor cells to mature osteoblasts include Bapx1/Nkx3.2 (bagpipe protein 1), Msx-1/Hox 7.1 and Msx-2/Hox 8.1 (muscle-segment homeobox proteins 1 and 2), the homeobox protein Hoxa2, Dlx-5 and Dlx-6 (distal- less like homeobox proteins 5 and 6), core-binding factor a1 (Cbfa1/Runx2/PEPB2aA/AML3) and osterix (Osx) (Figure 2). Bapx1 is one of the early markers for somite development and is expressed in all cartilaginous condensations during embryonic development (11). Moreover, Bapx1 has been shown to be required for bone development since Bapx1 knockout mice have deformed cranial bones and fail to initiate chondroblast differentiation of MSCs, resulting in severe malformations of the skeleton. The failure of Bapx1 knock-out mice to initiate chondroblast differentiation is reflected by the finding that expression of the Sox-9, which is the most important transcription

- 9 - Chapter 1

Cx43 Alp2 Ocn C o lla Dmp1 Dlx5 Bsp Opn Bax Opn Phex C b fa l C o lla Dmp1 C b fa l Ocn C b fa l, Osx Proteoglycans Sost

LINEAGE SELF RENEVWL PROLIFERATION MINERALIZATION COMMITMENT r~ nr ~ i r ~

PTH/PTHrP, DIFFERENTIATION FACTORS Glucocorticoids RA, BMPs, IHH, PDGFs, Vitamin D TGF-ß Leptin IGFs, FGFs, Leptin LIF BMPs Dlx5, Msx1, Msx2, Dlx5, Cbfal, FG Fs TGFß C b fal Dlx5, C b fa t M sx2 ® ------► ► W g ) 0sx -)► Cell death

STEM CELL MESENCHYMAL OSTEOPROGENITOR PRE-OSTEOBLAST MATURE OSTEOCYTE STEM CELL OSTEOBLAST

Sca1/Lya6, IL-18 LINING CELL Twist Id 1-3 w

IHH, PTHrP, RA, BM Ps B apxl, Sox9, Dlx5, C b fa l w

PPARy2

MyoD ► MYOBLASTS

------w FIBROBLASTS Figure 2. Stages of osteoblast differentiation. Schematic diagram of the differentiation pathway of mesenchymal stem cells towards osteoblast cells, including factors that regulate differentiation and markers for the various stages of differentiation. See text for details. (Adopted from Lian J.B. et al (2003), see ref. 4) factor for chondrogenesis, and the expression of markers for chondrocyte differentiation such as the type II collagen (Col2a1) gene is down-regulated (12, 13). Msx-1 and Msx-2 have been shown to regulate bone development, since both Msx-1 and Msx- 2 knock-out mice display severe defects in craniofacial bone development (14, 15). Moreover, Msx-2 knock-out mice also have defects in endochondral bone formation, which is caused by defective proliferation and subsequent differentiation of progenitor cells (15). In addition, over-expression of Msx-2 in MSCs induces the differentiation of these cells to osteoblastic cells, as shown by the up- regulation of osteoblast markers and up-regulated matrix mineralization (16). Msx-1 and Msx-2 are highly expressed in MSCs and osteoprogenitor cells but much less in more differentiated cells, since Msx-1 expression must be down-regulated for progression of osteoblast differentiation in vivo (17­ 19). The homeobox protein Hoxa-2 has been described not so much as an initiator of osteoblast differentiation but more as a regulator of induced osteoblast differentiation. Hoxa-2 has been shown to negatively regulate intramembranous bone formation by inhibiting the expression of Sox-9 and Cbfa1, which is a key-regulator of osteogenesis (20). Dlx-5 and Dlx-6 were shown to be expressed in the developing limb bud as well as in all skeletal structures after the first cartilage formation (21). Moreover, Dlx-5 was shown not only to induce chondroblast differentiation (22), but also osteoblast differentiation (23, 24), and it is directly involved in the expression regulation of Cbfa1 and multiple osteoblast marker genes (25-27).

- 10 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

Cbfal is a transcription factor of the Runx family of transcriptional regulators (28, 29) and its regulation and function in bone have been studied extensively over the last decade. The gene for Cbfal encodes three isoforms that are regulated by two promoters. The type I isoform is regulated by one promoter, whereas the type II and III isoforms are under the control of another promoter and are formed by different translational start sites (30). In addition, the type I isoform is constitutively expressed in non-osseous tissue and in osteoprogenitor cells, and its expression does not change during osteoblast differentiation. In contrast, the type II and III isoform transcripts increase specifically during differentiation of primary osteoblasts, and in response to BMP-2 treatment of osteoprogenitors (31, 32). Furthermore, regulation of the activity of the Cbfal protein through post- translational modifications (33) as well as through protein-protein interactions (34, 35) has been reported (see below). Cbfal is a key regulator of bone development since Cbfa1 knock-out mice in which the function of all isoforms is disrupted, exhibit impaired chondroblast differentiation and complete absence of skeletal ossification due to a maturational arrest of osteoblasts (36). Moreover, most of the well-established bone markers are known target genes of at least one of the Cbfal isoforms (28, 30, 36-39). Until recently, it was assumed that the impaired chondroblast differentiation and complete absence of skeletal ossification seen in Cbfa1 knock-out mice was caused primarily by the disruption of the function of the Cbfal type II isoform because of its bone-specific expression pattern. However, Xiao et al. (39) have shown that selective Cbfal type II-deficient mice have limited impairment of osteoblast maturation and form almost complete mineralized skeleton, which is characterized by severe abnormalities in endochondral bone formation and almost normal intramembranous bone formation. Specifically, Cbfal type II-null mice have almost normal chondroblast differentiation and maturation, but the subsequent osteoblast differentiation and maturation is severely impaired (39). Therefore, it is likely that Cbfal type I, which is constitutively expressed in osteoprogenitor cells, plays an important role in the early stages of mesenchymal cell development, whereas the “bone- specific” Cbfal type II is necessary for complete skeletogenesis and osteoblast differentiation, especially in the later stages of endochondral ossification (39). In addition to Cbfal, the transcription factor Osx is specifically expressed in all developing bone. More specifically, Osx is expressed similarly to Cbfa1 in differentiating chondrocytes, but not in hypertrophic chondrocytes, and is expressed in all cell types of the osteoblastic lineage (28, 40). Like Cbfal, Osx is a key-regulator of bone development, since Osx knock-out mice display normal patterning of endochondral skeletal elements, but ossification of these elements is severely delayed. In addition, a lack of mineralized bone was observed in the skeletal elements formed by intramembranous ossification due to complete inhibition of osteoblast differentiation, and embryos died shortly after birth due to respiratory failure (40). Furthermore, expression of Cbfa1 was normal in Osx-null cells, while expression of several osteoblast marker genes was severely reduced or completely lacking. However, these cells expressed Sox9, Col2a1 and other chondroblastic markers, indicating that the cells that could not differentiate into osteoblasts acquired a chondroblastic phenotype (40). It has been assumed that Osx expression is regulated by Cbfal, since Osx is not expressed in Cbfa1-null mice while Cbfa1 is normally expressed in Osx-null mice (39, 40). However, reports on the mechanism of Osx regulation are scarce and the results contradictory, due to the existence of cell-type specific isoforms that arise by differential promoter use (27, 4l).

1.1.2.4 Marker genes of Lineage Commitment, Proliferation and Maturation Due to their importance in regulating the differentiation of osteoprogenitor cells to mature osteoblasts, the transcription factors Dlx5, Cbfal and Osx are considered to be marker genes for Lineage Commitment, Proliferation and Maturation. Furthermore, during osteoblast differentiation

- l l - Chapter 1

several genes are expressed that are not involved in the regulation of differentiation, but encode proteins required for the function of the osteoblast (Figure 2). These marker genes for Proliferation and Maturation include Col1a (type I collagen), Alp2 (alkaline phosphatase type liver/bone/kidney), osteocalcin (Ocn), osteopontin (Opn), bone sialoprotein (Bsp), dentin matrix protein 1 (Dmp1) and various proteoglycans such as decorin and lumican. Col1a forms the major constituent of the organic component of the extracellular matrix (ECM). Synthesis of the ECM by osteoblasts contributes to cessation of cell growth leading to a “matrix maturation” state when induced expression of Alp2 and specialized bone proteins renders the ECM competent for mineral deposition. The subsequent up-regulation of calcium- and phosphate- binding proteins such as Ocn, Opn, Bsp and Dmp1 may function to regulate the ordered deposition of mineral, the amount of the hydroxyapatite crystals, and the size of the crystals (4, 42). In addition, expression regulation of the genes for the small leucine-rich proteoglycans decorin and lumican reflects the differentiation status. Decorin negatively regulates the capacity of the cells to mineralize the matrix and its expression is down-regulated with progression of differentiation (43, 44). In contrast, lumican is up-regulated during osteoblast differentiation and maturation (45). Furthermore, bone matrix synthesis supports both cell-cell and cell-matrix interactions that are mediated by transmembrane proteins and several classes of adhesion proteins. Of the integrins, which couple the extracellular matrix to cytoskeletal elements, the P1 integrins play a key role in early bone development and osteoblast differentiation by binding various collagens, which results in the induction of signals that lead to expression of genes characteristic for the osteoblast phenotype (46). The mature osteoblast also expresses genes that do not function to regulate osteoblast differentiation or function, but form the major determinants for osteoclast differentiation. These include RANKL ( activator o f nuclear factor kB ), osteoprotegerin (Opg) and CSF-1 (colony-stimulating factor 1) (for review see ref. 47). Binding of RANKL to its receptor Rank induces the differentiation of mononuclear/phagocytic lineage cells into osteoblasts, thus linking osteoblastogenesis to osteoclastogenesis (47). Opg is a secreted receptor of the tumor necrosis factor receptor family, and by competing with Rank for ligand binding, it acts as a decoy receptor for binding of RANKL, inhibiting osteoclast differentiation (47).

1.1.2.5 Transcriptional regulation and marker genes o f Mineralization Maturation of osteoblasts into osteocytes, as well as osteocyte activity and survival, is regulated by the formation of gap junctions (48, 49). These gap junctions are formed mostly by the connexin 43 (Cx43) protein and provide metabolic and electric coupling of different osteocytes, as well as of osteocytes with osteoblasts and lining cells. Disruption of the formation of gap-junctions by Cx43-deficiency results in dysfunctional osteoblasts which leads to retarded ossification in the entire skeleton (50). Osteocytes express genes that are also expressed in osteoblasts, but generally at a higher level. These include Cx43, the initiator of mineralization Dmp1 (51), positive regulators of mineralization such as Phex (phosphate-regulating gene with homologies to endopeptidases on the X-chromosome) (for review see ref. 52), and inhibitors of mineralization such as Of45/Mepe/osteoregulin (osteoblast/osteocyte specific factor 45) (53) and Ocn (54). Specific osteocyte markers include the E11 antigen gene, which is up-regulated during fracture healing and may therefore be involved in proliferation, migration, adhesion or differentiation (55); the gene for the actin filament fimbrin, which is expressed in the so-called osteocyte processes (thin cellular extensions) and maintains integrity of these processes (56); and Sost, which encodes a secreted glycoprotein that inhibits osteoblast differentiation and mineralization of the ECM (57, 58).

- 12 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

It should be noted here that particularly (mature) osteoblasts are not all identical in their expressed gene repertoire (59, 60), and it has been suggested that extensive diversities between various osteoblasts might contribute to the heterogeneity in trabecular micro-architecture (61), site- specific differences in disease manifestation, and regional variations in the availability of osteoblasts to respond to therapeutic agents (62).

1.1.3 Hormonal Regulation o f osteoblast differentiation

Hormonal factors that are known to contribute to the induction of differentiation of MSCs to osteoprogenitor and osteoblast cells include PTH/PTHrP (parathyroid hormone/parathyroid hormone- related protein), Indian hedgehog (IHH), Vitamin D, retinoic acid (RA), sex steroids, glucocorticoids and leptin (Figure 2). PTH is synthesized and secreted almost exclusively by the parathyroid gland and regulates calcium homeostasis as an endocrine factor. Its access to the skeleton is via the circulation, and although PTH knock-out mice are viable, they show retarded bone development caused by poorly developed capillary invasion leading to reduced cartilage matrix mineralization and diminished numbers of osteoblast progenitor cells. In addition, apoptosis of cells of the osteoblast lineage is enhanced in these mice. It has been speculated that the anabolic effect of intermittent PTH administration in adults is due to enhanced recruitment, commitment, and/or proliferation of osteoprogenitors, as well as inhibition of apoptosis of cells of the osteoblast lineage, as observed in fetal bone development (63). Furthermore, it has been shown that the anabolic effects of PTH require Cbfa1-dependent signaling (64), and that the expression of several Cbfa1 target genes is regulated by PTH (for review see ref. 65). In contrast to PTH, PTHrP is expressed in a variety of cells including proliferative and hypertrophic chondrocytes where it functions as a paracrine/autocrine factor. PTHrP is essential for normal endochondral bone formation since PTHrP knock-out mice die shortly after birth, and exhibit severe skeletal defects due to diminished chondrocyte proliferation, associated with premature chondrocyte maturation and accelerated bone formation (66). PTHrP expression is regulated by IHH, and PTHrP and IHH are speculated to control a negative feedback loop that regulates the rate and synchrony of chondrocyte differentiation (67). Mice lacking IHH expression show a similar but distinct phenotype from PTHrP knock-out mice, characterized by reduced chondrocyte proliferation and maturation as well as a failure of osteoblast development in endochondral bones, while in addition they die shortly after birth due to respiratory failure (68). IHH mediates its effects on endochondral bone growth by both PTHrP-dependent and -independent pathways and therefore the phenotype of IHH knock-out mice is more severe than that of PTHrP knock-out mice (69). Vitamin D is a steroid made in the skin by the action of sunlight. Vitamin D itself is biologically inert and must undergo two successive hydroxylations in the liver and kidney, respectively, to become the biologically active 1,25 dihydroxy vitamin D3 {1,25(OH)2D3}. This biologically active form of vitamin D is the major regulator of calcium homeostasis in the body and protects the organism from calcium deficiency via effects on the intestine, kidney, parathyroid gland, and bone. The renal conversion of 1,25(OH)2D3 is tightly regulated by serum calcium levels through PTH and inorganic phosphate in a feedback loop (70). Biologically active vitamin D exerts its effect via a complex of the (VDR) and a retinoic acid X receptor (RXR) on osteoblasts. Mice lacking VDR show impaired bone formation after weaning and die within 15 weeks after birth (71). In general, 1,25(OH)2D3 is thought to inhibit osteoblast progenitor proliferation and to induce osteoblast differentiation in vivo. However, depending on the differentiation status of the cells, different effects of 1,25(OH)2D3 have been observed (for review see ref. 72). On the other hand,

- 13 - Chapter 1

1,25(OH)2D3 does not have a direct active role in the mineralization process; it maintains normal levels of blood calcium and phosphate levels to ensure normal mineralization to occur. In addition, 1,25(OH)2D3 stimulates the osteoblasts to produce RANKL, thereby stimulating osteoclast proliferation and maturation which results in balanced bone remodeling (70). Retinoic acid (RA) is the most active metabolite of vitamin A, and multiple isomers of RA exist of which all-trans RA (T-RA) and 9-cis RA (9C-RA) are the most prominent. RA levels are tightly regulated by the activity of retinaldehyde dehydrogenases (RALDH/ALDHla) and the cytochrome P450RA (P450RA/CYP26) enzymes, which synthesize and degrade RA respectively (for review see ref. 73). This tight regulation of RA levels is necessary, since elevated or subnormal RA levels cause congenital malformations including severe skeletal defects (74, 75). Two families of nuclear receptors have been identified that relate retinoic acid signaling: the RA receptors (RARa, -P and y) and the retinoid X receptors (RXRa, -P and y), which function predominantly as RAR/RXR heterodimeric transcription factors during embryonic development, although RXR homodimers have also been described to exists (76, 77). In the RAR/RXR heterodimer, RXR functions as an orphan receptor, whereas in RXR homodimers it is specifically activated by 9C-RA (78). However, it is currently unclear under which physiological conditions the RXRs act as 9C-RA-dependent receptors. Furthermore, RARs in the RAR/RXR heterodimer complex are activated by both T-RA and 9C-RA (78). In addition to the nuclear receptors, cytoplasmic retinol binding proteins (CRBPs) and cytoplasmic RA binding proteins (CRABPs) bind to retinol and RA respectively, which have roles in controlling RA availability in the nucleus (for review see ref. 79). The RA receptors and binding proteins are expressed in distinct and sometimes overlapping spatiotemporal patterns during embryogenesis, and disruption of their function leads to several skeletal defects, predominantly abnormalities associated with endochondral bone formation (79). In particular, RA has been shown to regulate both chondroblast differentiation and chondrocyte hypertrophy. At the early stages of chondroblast differentiation, RA inhibits the expression and activity of Sox9 and subsequently reduces the expression of chondroblast marker genes, thus influencing the timing of condensation and chondroblast differentiation. In addition to this inhibitory effect on early stage chondroblast differentiation, RA positively regulates the maturation of chondrocytes by inducing the expression of Cbfa1 and its target genes, while simultaneously inhibiting the expression of Sox9 (79). In addition to these effects on chondrogenesis, RA has also been described to enhance osteoblast differentiation, in part because RA induces the expression of BMPs, but also because RA and BMPs cooperate to induce differentiation of mesenchymal cells to osteoblasts (80-82). It has been hypothesized that this action of RA is mediated by the Ets transcription factors, which are differentially expressed throughout osteoblast differentiation. Furthermore, these factors are able to interact with Cbfa1 and show expression regulation by RA (83). Sex steroids form a group of steroid hormones that consist of estrogens, androgens and progesterone. Particularly estrogen has been shown to play an important role in maintaining bone mass. It is believed, however, that the major effect of estrogen is not a stimulation of the bone forming osteoblasts, but primarily an inhibition of the proliferation and differentiation of the bone resorbing osteoclasts. In postmenopausal women, estrogen deficiency causes an increase in bone remodeling activity during which bone resorption is no longer coupled to bone formation, resulting in a net loss of bone (osteoporosis) (84). Recently it has been shown that also in men estrogen is very important for inhibiting bone resorption. Androgens such as testosterone are known to enhance bone size in both men and women (for review see ref. 85). Androgens decrease bone resorption by acting directly on osteoclast function, and can regulate the production of bone-resorbing factors produced by mature osteoblasts (86).

- 14 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

Progesterone levels also decrease at menopause, but direct evidence for an effect of progesterone in postmenopausal bone loss is still lacking. However, it has been shown that progesterone can promote bone formation in vivo both independently and in combination with estrogen. In addition, progesterone may have a similar effect on bone resorption as estrogens. Finally, it has been shown that progesterone consistently blocks glucocorticoid responses by binding to glucocorticoid receptors, but it has also been suggested that progesterone may act directly through the in osteoblast cells (for review see ref. 87). Glucocorticoids such as dexamethasone have been reported to promote the differentiation of MSCs to osteoblasts in vitro. However, in vivo it has been shown that glucocorticoids impair the differentiation of MSCs toward cells of the osteoblast lineage and prevent the terminal differentiation of osteoblastic cells. In addition, glucocorticoids stimulate apoptosis of mature osteoblasts. These two effects combined result in a decrease in the number of bone forming cells. Glucocorticoids also inhibit the function of the remaining mature osteoblasts. Furthermore, osteoclast proliferation and maturation is stimulated by glucocorticoids, resulting in increased bone resorption. This stimulation of bone resorption is most likely responsible for the initial bone loss in metabolic bone diseases observed after glucocorticoid exposure. The additional inhibition of bone formation by glucocorticoids will cause a decrease in bone remodeling and a continued increased risk of fractures (for review see ref. 88). In addition to these direct effects of glucocorticoids on osteoblasts and osteoclasts, glucocorticoids also indirectly inhibit bone formation by decreasing the expression of the growth factor IGF-I, which is a positive regulator of osteoblast function. This is an important mechanism that contributes to the decreased bone formation observed after glucocorticoid exposure(89). Leptin is a signaling peptide that acts on the hypothalamus, thereby stimulating the sympathetic nervous system, which results in an attenuation of appetite and an enhancement of lipolysis in adipocytes, with concomitant heat production. Leptin is mainly produced in adipose tissue, but it is also expressed in and secreted by other tissues such as muscle, gastric epithelium, breast epithelium and bone (for review see ref. 90). Leptin exerts its effects via a specific receptor (OB-R1), which is expressed in cells of the hypothalamus, cerebral cortex, amygdala, thalamus, cerebellum and anterior pituitary as well as in tissues such as adipose, liver, kidney, heart, muscle and bone. Leptin has been shown to be a regulator of bone formation, since mice deficient in leptin production or signal transduction are not only obese and hypogonadic, but also display a markedly elevated bone strength and mass, indicating that leptin negatively regulates bone formation. In contrast, however, leptin has also been shown to enhance proliferation, differentiation and mineralization of primary osteoblasts, and to prolong the life span of primary osteoblasts by inhibiting apoptosis. Furthermore, leptin has been shown to inhibit differentiation of osteoclasts, indicating that leptin positively regulates bone formation and remodeling (90). Recently, a model has been proposed in which the effect of leptin on bone formation was proposed to be dependent on the concentration and mode of exposure, determined by the source from which leptin is reaching the bone cells. This model postulates that the central effect of leptin via the hypothalamus could act as a brake on the anabolic actions that hormones, local factors and mechano- sensing have on bone mass, while locally produced leptin promotes bone formation, thus ensuring a proper balance of bone mass in response to the nutritional status (90).

- 15 - Chapter 1

1.1.4 Polypeptide growth factors in bone

Bone is a rich source of polypeptide growth factors that have important actions in the regulation of bone formation and bone resorption. These include insulin-like growth factors (IGFs), fibroblast growth factors (FGFs), platelet-derived growth factor (PDGF), transforming growth factor P (TGF-P) and the bone morphogenetic proteins (BMPs) (Figure 2). Two insulin-like growth factors, IGF-I and IGF-II, have been identified, which are present in the systemic circulation and are synthesized by multiple tissues, including bone, where they act as local regulators of cell metabolism (for review see ref. 91). In the circulation, IGFs exist in a complex with IGF binding proteins (IGFBPs) and a non-IGF binding component termed acid-labile subunit (ALS) (for reviews see refs. 91, 92). IGFs have a general anabolic effect on bone remodeling, since mice lacking IGF-I exhibit decreased bone formation. IGFs modestly promote proliferation of cells of the osteoblastic lineage, and enhance the function of mature osteoblasts by increasing bone matrix synthesis. In addition, IGFs prevent apoptosis of osteoblasts resulting in a larger population of mature osteoblastic cells. Moreover, IGFs increase collagen synthesis by osteoblasts and inhibit collagen degradation by down-regulation of collagenases (93-96). PTH is the major inducer of IGF synthesis in osteoblasts, explaining the anabolic effect of PTH in bone (97). On the other hand, glucocorticoids inhibit IGF expression as mentioned above. The six IGFBPs are also produced by bone cells, and they can prolong the half life of IGFs in the circulation, either neutralize or enhance their biological activity, and mediate transport of IGFs to their target cells (92). FGFs comprise a family of 22 structurally related proteins, which can be subdivided into 6 subfamilies, all with similar functions but expressed in specific spatial and developmental patterns. Four distinct FGF receptor tyrosine kinases (FGFRs) bind and are activated by these various FGF family members. FGFs are considered important for neovascularization, wound healing and bone formation, and disturbances in FGF signaling result in various skeletal deformations (for review see ref. 98). For example, a point mutation in the transmembrane domain of FGFR3 forms the molecular basis for Achondrodysplasia, the most common genetic form of dwarfism in humans. During bone development, the four FGFRs have distinct, mostly non-overlapping expression patterns, but in general they are expressed in mesenchymal condensations, proliferating chondrocytes, hypertrophic chondrocytes, differentiating osteoblasts and mature osteoblasts. There is a strong overlap between the expression patterns of the various FGFs and their receptors (98). FGFs control the balance between undifferentiated and differentiated osteogenic cells by increasing the proliferation of immature cells and chondrocytes and by promoting the differentiation and apoptosis of more mature osteoblasts. In addition to the effects on bone formation, FGFs also directly regulate bone resorption by increasing osteoclast proliferation and activation of mature osteoclasts (98). Moreover, indirect actions of FGFs on bone development involve the regulation of IHH/PTHrP, which results in an inhibition of bone growth (99). PDGF is a dimer of the products of four genes, PDGF-A, -B, -C and -D, and experimental evidence has been presented for -AA, -BB, -CC or -DD homodimers as well as for an -AB heterodimer. In addition, there are two PDGF receptors, a and p. PDGFs and their receptors are expressed during embryogenesis in distinct patterns particularly in the mesoderm, and disturbances in PDGF signaling during embryogenesis result in severe growth deformations, generalized hemorrhages and embryonic death (for reviews see refs. 100, 101). PDGFs are implicated to play a role in inflammation, wound healing, angiogenesis, apoptosis and bone formation. In bone, PDGF-AA, -AB and -BB are expressed in the osteoblast lineage cells and regulate the proliferation of pre-osteoblastic cells, which results in an increase in the number of osteoblastic cells, and subsequently bone collagen synthesis. However, these PDGFs do not have a direct

- 16 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

stimulatory effect on the function of differentiated osteoblasts but instead inhibit the rate at which bone mineralization occurs. In addition, PDGFs enhance bone resorption by increasing osteoclast proliferation and by inducing the expression of specific collagenases and other matrix metalloproteases. Finally, PDGFs indirectly regulate bone formation by inhibiting the expression of IGF-I and -II and by stimulating the expression of interleukin 6 (IL-6), which enhances bone resorption by increasing osteoclast formation and recruitment (for review see ref. 102). Three isoforms of TGF-P (TGF-P1, -P2, and -P3) have been described with very similar biological activities. TGF-P is the founding member of a family of growth factors which also include BMPs, activins, nodal and MIS/AMH (Mullerian inhibiting substance /anti-Mullerian hormone) (for review see ref. 103). Members of the TGF-P family are produced by a large variety of cell types, and are known to regulate cell migration, adhesion, proliferation, differentiation and apoptosis (for reviews see refs. 104, 105). Moreover, TGF-P family members have major roles during embryogenesis and skeleton formation, since mice deficient in genes belonging to this family or their signal transduction pathways often show early embryonic lethality, various skeletal disorders and holoprosencephaly, a forebrain malformation (106). The most important members of this family for bone formation are the BMPs, and their function, signal transduction and regulation will be discussed in detail below. In bone, TGF-P stimulates the replication of pre-osteoblasts, bone collagen synthesis, and it inhibits bone resorption by inducing osteoclast apoptosis. On the other hand, TGF-P opposes the effects of BMPs on osteoblastic cell differentiation and inhibits late-stage osteoblast differentiation (for review see ref. 107).

1.2 Bone Morphogenetic Proteins

1.2.1 BMPs in mesenchymal stem cell differentiation and osteoclast formation

As mentioned above, BMPs are members of the TGF-P family of growth factors. Over 20 BMP family members have currently been identified, which can be classified in multiple subfamilies of highly related proteins. BMP-2, BMP-4 and the Drosophila homologue decapentaplegic (Dpp) constitute one such subfamily. A second subfamily is formed by BMP-5, BMP-6/Vgr1 (BMP- 6/vegetal-related growth factor 1), BMP-7/OP1 (BMP-7/osteogenic protein 1) and BMP-8/OP-2, while a third is formed by GDF-5 (growth differentiation factor 5), GDF-6/BMP-13 and GDF- 7/BMP-12. Like all members of the TGF-P family, BMPs are synthesized as large precursors, which are processed and proteolytically cleaved to yield mature active protein homodimers (108). Proteolytic cleavage of BMPs is mediated by the pro-protein convertases furin and PC6 (109). All members of the TGF-P family are cysteine knot proteins. BMPs are distinguished from other members of the family by having, in general, seven conserved cysteines in the mature protein, rather than nine. BMPs are secreted as biologically active molecules, and not in a latent form such as TGF-Ps where the pro­ region remains non-covalently complexed with the mature region and therefore require extracellular proteolytic cleavage to obtain a biologically active protein (108). Although the BMPs were first identified by their ability to induce ectopic bone formation in vivo (110), these proteins are now known to be involved during embryonic development in mesoderm formation and patterning, left- right asymmetry, neural patterning, somite patterning, skeletal development, limb patterning and in

- 17 - Chapter 1

the development of nearly all organs including tooth, lung, kidney, gut, skin and gonads (for review see ref. 108). The primary function of BMPs in bone formation is to block the differentiation of MSCs to myoblasts and to induce their differentiation toward chondroblasts and osteoblasts (111-113). In order to inhibit MyoD-mediated myoblast differentiation, BMPs induce the expression of Msx-1 and the inhibitor of differentiation genes (Id1-3), which in turn inhibit the expression of the genes for MyoD and (114-118). To promote chondrocyte proliferation and differentiation during endochondral bone formation, BMPs induce or sustain the expression of the homeobox genes Bapx1/Nkx3.2, Dlx5, Sox9, Msx-1 and Msx-2 in parallel with or downstream of Hedgehog proteins (119-124). Furthermore, BMPs enhance the expression of Cbfa1 and Osx in chondrocytes (125, 126), and interactions between BMP signaling molecules (Smads) and the transcription factor Cbfa1 are necessary for proper chondrogenic differentiation (112). BMPs stimulate osteoblast differentiation by repressing Bapx1 expression which results in a repression of chondroblast differentiation (127). In addition, they induce the expression of Dlx5 (23). Dlx5 then up-regulates the osteoblast specific transcription factors Cbfa1 and Osx in a direct manner (26, 27). Cbfa1 and Osx subsequently regulate the expression of osteoblast marker genes such as Alp2, Ocn, Col1a, Opn, Bsp, Dmp1 and Sost (28, 30, 37, 38, 40, 51, 128). Interaction of Smads with Cbfa1 is required for the targeting of Smads to the nucleus and subnuclear sites of active transcription, and thus for optimal osteoblast differentiation (129, 130). Mutations in Cbfa1 that inhibit its interaction with Smads result in severe defects in bone formation, due to inhibition of osteoblast differentiation (131). In addition, BMPs induce apoptosis of terminally differentiated osteoblasts by the action of Msx-2, a process that occurs independently of BMP-induced osteoblast differentiation, (132-134) (Figure 2). BMPs are synthesized by many cells including osteoblasts, and play an autocrine role in chondroblast/osteoblast differentiation and function (135). Moreover, BMP expression is regulated by estrogens, retinoic acid and BMPs themselves. Promoters of various BMP genes also contain Cbfa1 responsive elements (136-140), which suggests the existence of a positive feedback loop for regulating BMP expression in which Cbfa1 plays a central role. BMPs also have both an indirect and a direct role in osteoclastogenesis. BMPs indirectly stimulate osteoclast differentiation by stimulating osteoblast differentiation, since inducers of osteoclastogenesis such as RANKL and CSF-1 are expressed by mature osteoblasts (47). Moreover, it has been shown that Cbfa1 is needed for optimal RANKL expression (141, 142). BMPs directly stimulate osteoclast differentiation by sensitizing the cells to the effects of RANKL on cell genesis and survival (143). To temper these effects, BMPs also induce the expression of osteoprotegerin (see above) and inhibit expression of collagenase (144, 145).

1.2.2 BMP receptors and signaling

Like the other members of the TGF-P family, BMPs exist mainly as homodimers and exert their effects through activation of specific heterotetrameric combinations of so-called type I and type II transmembrane serine/threonine kinase receptors: BMPs exert their effect through combinations of BMP type II receptor (BMPR-II), ActR-II or ActRIIB and the type I receptors Alk-3, Alk-6 or Alk-2. In contrast, activins signal through combinations of the activin type II or IIB receptor (Act-RII/Act-RIIB) and the type I activin receptor-like kinase 4 (Alk-4), while TGF-Ps elicit their effects through combinations of the TGF-P type II receptor (TGF-PR-II) and the type I receptors Alk-

- 18 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

5 or Alk-1 (103) (Table 1). Upon ligand binding, the constitutive kinase of the type II receptor phosphorylates the GS-domain in the juxtamembrane region of the type I receptor which subsequently becomes activated (103). Unlike TGF-Ps and activins, which first bind to the type II receptor after which a type I receptor becomes recruited to the complex and is subsequently activated (103), BMPs first bind to the type I receptor, and it has been hypothesized that two distinct modes of receptor binding mediate BMP signaling (Figure 3). It has been proposed that BMP type I and type II receptors are present on the cell surface as pre-existing heteromeric and homomeric complexes, including complexes of both type I and type II (BMPR-II/Alk-3, BMPR-II/Alk-6, Act-RII/Alk-3, Act-RII/Alk-6, Act-RIIB/Alk-3 or Act-RIIB/Alk-6) and complexes of either type II (BMPR-II/BMPR-II, Act-RII/Act-RII, Act- RIIB/Act-RIIB, BMPR-II/Act-RII, BMPR-II/Act-RIIB or Act-RII/Act-RIIB) or type I receptor combinations (Alk-3/Alk-3, Alk-6/Alk-6, Alk3/Alk6) (146). Signal transduction through the receptors can be achieved either by binding to two preformed type I/II dimeric complexes, subsequently inducing a conformational change that activates this complex, or by binding to high affinity type I dimeric receptor complexes, subsequently recruiting unliganded BMPR-II homodimers into the complex (146). Furthermore, each of the different receptor oligomers may allow association of different cytoplasmatic proteins (Figure 3, indicated as X, Y and Z) that may participate in the regulation of distinct pathways. It has been proposed that binding of BMP to the pre-existing type I/II heterodimers initiates the Smad signal transduction pathway, while binding of BMP to ligand-induced type I/II heterodimers initiates Smad-independent signal transduction pathways such as the MAP- kinase (p38, ERK and JNK) pathways (147). Two such cytoplasmic proteins that participate in the regulation of BMP-induced signaling pathways are Bram1 (BMP receptor associated molecule 1) and Xiap (X-chromosome-linked inhibitor of apoptosis protein), which specifically interact with type I BMP receptors in the cytoplasm. Both Bram1 and Xiap recruit Tab1 (Tak1 binding protein 1), an activator protein of Tak1 (TGF-P activated kinase 1; a MAP kinase kinase kinase), to the receptor complex, indicating that Bram1 and Xiap are involved in the activation of the Tab/Tak/p38 pathway by BMPs (148, 149). In addition to the differential regulation mediated by pre-formed receptor complexes, the binding affinity of the various BMPs for type I receptors differs, and the level of receptor expression varies in different cell types. Moreover, the function of the type I receptors also varies according to cell type. This suggests that in addition to the mode of receptor oligomerisation, the specificity of the type I receptor can determine which signaling pathway is used as well (147).

Table 1. Receptor specificity of TGF-ft family members Ligand Type II receptor Type I receptor Activin ActR-II/ActR-IIB Alk-4 (ActR-IB)

TGF-ß TGF-ßR-II Alk-5 (TGF-ßR-I) Alk-1

BMP BMPR-II Alk-3 (BMPR-IA) ActR-II/ActR-IIB Alk-6 (BMPR-IB) Alk-2 (ActR-I) ActR-II, activin type II receptor; ActR-IIB, activin type IIB receptor; TGF-PR-II, TGF-P type II receptor; BMPR-II, BMP type II receptor; Alk 1-6, activin receptor-like kinases 1-6; ActR-I, activin type I receptor, ActR-IB, activin type IB receptor; TGF-PR-I, TGF-P type I receptor; BMPR-IA, BMP type IA receptor; BMPR-IB, BMP type IB receptor.

- 19 - Chapter 1

Figure 3. Signal transduction via BMP receptors and regulation of BMP-induced signaling. Signal transduction via BMP receptors can be mediated 1) by binding of BMP to two preformed type I/II dimeric complexes, subsequently inducing a conformational change that activates this complex, or 2) by binding of BMP to high affinity type I homo- /heterodimeric receptor complexes, subsequently recruiting unliganded BMPR-II homodimers into the complex. Furthermore, each of the different receptor oligomers may allow association of different cytoplasmatic proteins (indicated as X, Y and Z) which may participate in the regulation of distinct pathways. See text for details. Regulation of BMP-induced signaling is achieved by I) extracellular agonists and antagonists, II) pseudoreceptors, III) inhibitory Smads, IV) Smad binding proteins and IV) proteasomal degradation. See text for details. See page 134 for full color figure. (Adopted from Canalis et al. (2003), see ref. 209)

Upon receptor activation, members of the TGF-P family signal via Smads (Figure 3). Three classes of Smads exist: 1) receptor regulated Smads (R-Smads) that can be activated by BMPs (Smad- 1, -5 and -8) or TGF-P/activins (Smad-2 and -3); 2) a common mediator Smad (Co-Smad), Smad-4; and 3) inhibitory Smads (I-Smads), Smad-6 and -7 (for reviews see refs. 103, 150). Smad proteins contain three domains, an amino-terminal MAD homology domain MH1, a divergent proline-rich

- 20 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

linker domain, and a carboxy-terminal MH2 domain. The MH1 domain is conserved between R- and Co-Smads, binds to several cytoplasmic partners and, for some R-Smads, has been shown to bind DNA and to interact with various nuclear proteins. The MH2 domain, which is highly conserved among all Smads, binds to cytoplasmic and nuclear proteins and is required for Smad oligomerization (for reviews see refs. 103, 150, 151). After activation of the type I receptor, R-Smads in the cytoplasm are recruited to the receptor complexes and are activated by phosphorylation of two serine residues in the carboxy-terminal SS(V/M)S motif (152-154). Activated R-Smads then form heteromeric complexes with Smad-4 and become efficiently translocated to the nucleus (for review see ref. 155). Smad-4 can bind DNA on GTCT containing sequences, also termed Smad binding elements (SBEs), and artificial multimeric SBE containing promoters are efficient enhancers of BMP and TGF- P signaling (156, 157). Furthermore, BMP R-Smads have been reported to also bind GCAT motifs, GCCG-rich containing sequences, GGCG elements and CAGC elements in different target genes (158-162). Moreover, multimerization of SBE, GGCG and CAGC motifs has been shown to enhance sensitivity for BMP-induced signaling (161). However, the affinity of Smads for DNA binding to the above sequences is relatively weak (163), unless additional co-activators or co-repressors are present to efficiently regulate the expression of target genes (164). Smad-mediated activation of transcription has been shown to be dependent on the general transcription co-activators CBP/p300 (CREB binding protein/p300) (Figure 3). Smad-1, -2, -3 and -4 bind CBP/p300 through their MH2 domain, and phosphorylation of the R-Smads enhances Smad- dependent CBP/p300-mediated signaling (165-167). Moreover, CBP/p300 have intrinsic histone acetyltransferase activity (168), which suggests that they can potentiate Smad-dependent gene transcription by loosening the chromatin structure. In addition, it has been shown that activated R- Smads can bind the co-activator P/CAF (p300/CBP-associated factor), thereby enhancing transcriptional activity of the Smads. P/CAF, like CBP/p300, has intrinsic histone acetyltransferase activity and can bind to CBP/p300, thereby potentiating its effect on Smad-dependent transcription activation, possibly by increasing the affinity of R-Smads for their DNA binding sequences (169). Furthermore, it has been shown that the expression of several BMP response genes in their basal state is inhibited by the binding of Hoxc-8 to their promoter regions (144, 170). Hoxc-8 is capable of binding histone deacetylases (HDACs) rendering the chromosome transcriptionally inactive (171). Binding of activated R-Smad complexes to Hoxc-8 relieves the binding of Hoxc-8 to DNA and HDACs, while the subsequent recruitment of CBP/p300 by Smad-1 might be involved in the transcriptional activation of these BMP response genes (144, 170). BMPs also elicit negative gene expression regulation, and recently it has been shown that Bapx1/Nkx3.2 is a BMP-dependent transcriptional repressor. Bapx1 associates with Smad-1 and -4 upon receptor activation of the Smads, and subsequently this complex binds DNA on Bapx1/Nkx3.2 response elements. Upon DNA binding, Bapx1 recruits HDAC1 in a Smad-4-dependent manner. It has been postulated that this association of Smad-1/4, Bapx1 and HDAC1 then represses the transcription of genes that inhibit chondrogenic differentiation, resulting in the induction of chondroblast differentiation (172). Recently, it has been shown that endocytosis can play a role in signal transduction of multiple cytokine families (for review see ref. 173). Endocytosis of ligand/receptor complexes can either be mediated by clathrin-coated pits or by caveolae. Clathrin is the major protein in the formation of clathrin-coated pits, while caveolin is the major protein in the formation of caveolae. Clathrin- mediated endocytosis of TGF-P/receptor complexes is very important for signal transduction. SARA (Smad anchor for receptor activation) is a FYVE domain containing protein, which has been shown to bind to Smad-2 and -3 as well as the TGF-P receptor complex, thereby recruiting Smad-2/-3 to the receptor complex upon ligand binding (174). This recruitment of Smad-2/-3 to the receptor complex

- 21 - Chapter 1

has been shown to take place at the plasma membrane, but SARA-mediated localization of the TGF- P/receptor/SARA/Smad-2 complex in the endosome is necessary for TGF-P receptor complexes to phosphorylate Smad-2 (175). The FYVE domain of SARA has been shown to be sufficient for early endosomal localization and activation of TGF-P receptor complexes, but this process can be enhanced by cooperation with Hsg/Hrs (hepatic growth factor-regulated tyrosine kinase substrate), another FYVE domain containing protein (176, 177). SARA is specific for TGF-P signaling and, at the moment, no homologue of SARA has been identified for BMP signaling. In Drosophila, Dpp forms long-range concentration gradients in the developing wing extending over 40 cell diameters, which is required to specify cell fates along the anterior-posterior axis of the wing (173). This gradient is not formed by free diffusion, but rather employs clathrin- mediated endocytosis according to the so-called planar transcytosis model. Upon receptor binding, Dpp is internalized into early endosomes through dynamin-dependent endocytosis. Subsequently, a small percentage of these endosomes is degraded in the lysosome whereas the rest form so-called argosomes, membrane vesicles resembling exosomes, which are released on exocytic fusion with the plasma membrane. Subsequent rounds of endocytosis and exocytosis by the target cells establish spreading of Dpp over several cell diameters to form a gradient. The rate of lysosomal degradation in combination with the rate of endosomal trafficking thereby determine the maximal range of Dpp signaling (173). In bone, caveolae and its structural protein caveolin have been identified in both chondrocytes and osteoblasts. In chondrocytes caveolin-1, -2 and -3 are expressed, whereas in osteoblasts only caveolin-1 and -2 are expressed (178, 179). The precise role of caveolae in the regulation of osteoblast differentiation is still largely unknown. However, in C2C12 mesenchymal progenitor cells it has been shown that inhibition of caveolin-mediated endocytosis of BMP/receptor complexes enhances BMP-induced osteoblast differentiation, indicating that this type of endocytosis negatively regulates BMP signaling (180).

1.3 Regulators of BMP signaling

To ensure that the BMPs are active at the right place and the right moment, tight regulation of their activities is needed. This regulation is achieved at three levels: extracellularly, at the membrane level, and intracellularly.

1.3.1 Extracellular regulation o f BMP activity

Regulation of BMP activity at the extracellular level usually occurs through binding of BMPs by extracellular proteins that block the molecular interfaces for binding of BMP to its type I and type II receptors. Such extracellular antagonists include noggin (181, 182), chordin (183-185), ventroptin (186) and DAN family members (57, 187, 188). In addition, BMP antagonists such as follistatin can exert their function by binding to both BMPs and their receptors, rendering an inert complex incapable of transmitting signals (189). Finally, extracellular regulation can be achieved by agonistic proteins, such as twisted gastrulation (190), that also bind BMPs and thereby enhance binding to their receptors (Figure 3).

- 22 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

Noggin was first identified as a secreted protein capable of inducing dorsal development in Xenopus embryos (191). Noggin is a cysteine knot protein that binds with different affinities to BMP- 2, -4, -5, -6, and -7, GDF-5 and GDF-6 but not to other members of the TGF-P family (181, 192-194). Recently, it has been reported that noggin can associate with the cell surface through binding of heparan sulfate proteoglycans, where it remains functional for the binding and inhibition of BMPs (195). Noggin is released from the cell surface through the activity of the endosulfatase Qsulf1, which results in a restoration of BMP responsiveness in cells. However, binding of noggin to heparan sulfate proteoglycans is not essential for binding and antagonism of BMPs, but rather functions to regulate the in vivo diffusion of noggin and therefore the formation of gradients of BMP activity such as observed in early embryonic development (195, 196). In bone, BMPs can induce the expression of noggin in both chondroblastic and osteoblastic cells (197, 198), and noggin expression inhibits the differentiation of osteoblast lineage cells resulting in reduced bone formation and osteoporosis (198­ 201). Chordin was first identified as a secreted cysteine-rich protein that acts as a dorsalizing factor in Xenopus development (202). Chordin binds specifically to BMPs and not to other members of the TGF-P family (202). The biological activity of chordin resides in its cysteine-rich domains (CRs), especially in CR1 and CR3 (203, 204). In mice, three proteins with high homology to Xenopus chordin have been identified: mouse chordin (Chrd), chordin-like (Chl) and chordin-like 2 (Chl-2) (183-185). Chl is able to bind and inhibit BMP signaling in mice, but can also bind TGF-Ps. In comparison, Chl-2 interacts directly with BMPs but not with TGF-Ps. Moreover, the spatio-temporal expression pattern of Chl is distinct from that of Chrd, since expression of both Chl and Chl-2 is restricted to mesenchyme-derived cell types such as the chondrocytes of the developing skeleton during embryogenesis, whereas Chrd is predominantly expressed during early embryogenesis in the anterior primitive streak and subsequently in the node and axial mesoderm. In addition, Chrd has been shown to be essential for inner and outer ear development as well as for pharyngeal and cardiovascular organization (183-185, 205). Furthermore, it has been proposed that Chl-2 may negatively regulate the generation and maturation of chondrocytes and consequently endochondral ossification (185). The activity of Chrd is proposed to be regulated by BMP-1, a metalloprotease first identified in bone extracts capable of inducing ectopic bone formation (110) and structurally unrelated to BMP growth factors. BMP-1 is the shorter, alternatively spliced, product of mTld (mammalian tolloid) that shows high similarity to Drosophila Tolloid (206) and Xenopus Xolloid, which can specifically cleave chordin and chordin/BMP complexes in order to release biologically active BMP (207). In mice, four BMP-1/tolloid-like proteases are known: BMP-1, mTld, Tll-1 (tolloid-like 1) and Tll-2 (tolloid-like 2), of which only BMP-1 and Tll-1 are capable of cleaving chordin. Moreover, differences in enzymatic activity and expression domains of these proteases suggest that BMP-1 is the major chordin antagonist in early mammalian embryogenesis and in skeletogenesis (208). Ventroptin is a BMP antagonist that is mainly expressed in the chicken ventral retina, where it plays a role in retinal patterning (186). Ventroptin has three CR domains, which are homologous with those of chordin. Moreover, ventroptin binds BMPs and has BMP-neutralizing activity similar to that of chordin and noggin (186). Although ventroptin is also expressed at other sites than the ventral retina, its role in skeletal development is unknown. DAN (differential screening-selected gene aberrative in neuroblastoma) is a tumor suppressor gene expressed in cranial, facial and limb mesenchyme. Other proteins belonging to the DAN family of secreted glycoproteins include gremlin, cerberus, dante, caronte, PRDC (protein related to DAN and cerberus) and sclerostin (for review see ref. 209). In bone, gremlin is co-expressed with DAN in

- 23 - Chapter 1

cranial, facial and limb mesenchyme, and gremlin has been shown to regulate chondrogenesis (210, 211). In contrast to other members of the DAN family, sclerostin, the product of the gene Sost, does not function as a classical BMP antagonist; unlike other BMP inhibitors that block BMP-induced osteoblast differentiation and matrix mineralization, sclerostin has no effect on osteoblast differentiation but specifically inhibits calcium deposition in the matrix, probably in concert with an as yet unidentified late phase BMP-induced gene (58). Mice over-expressing Sost exhibit low bone mass and decreased bone strength (57), and in humans homozygous loss-of-function mutations in Sost lead to the disease named sclerosteosis, which is characterized by increased bone density and progressive skeletal outgrowth in skull, mandible, ribs, clavicles and all long bones (212, 213). The related heterozygote humans are clinically normal, and can only be identified by minor changes on skull radiographs (214). Furthermore, it has recently been shown that patients with Van Buchem disease, which is clinically similar to sclerosteosis but less severe, show a homozygous 52 kb deletion downstream of SOST, and it has been hypothesized that this deletion results in down-regulation of the SOST gene product by a cis regulatory action or position effect (215-217). This suggests that the severity of sclerosteosis/Van Buchem disease is dependent on the amount of functional sclerostin present, such that a reduction to 50% of the normal amount of sclerostin has no clinical effects, while further reductions of the normal level of sclerostin result in Van Buchem disease (> 50% reduction) or sclerosteosis (100% reduction). Follistatin was first identified as an activin-binding protein that inhibits activin function (218, 219), but it can also bind and inhibit various BMPs (189, 220). In contrast to other BMP binding proteins, follistatin is noncompetitive with the BMP receptors for ligand binding, but forms a trimeric complex with BMP and its receptor, rendering the BMPs incapable of transducing a signal through their receptors (189). Activin can induce endochondral bone formation in vivo as well as osteoblastogenesis in vitro, and this activity can be inhibited by follistatin (221, 222). Mice deficient in the follistatin gene are retarded in growth, have skeletal defects of the hard palate and the ribs, and die within hours of birth due to respiratory failure (223). Twisted gastrulation (Tsg) is a secreted cysteine-rich protein that was first identified in Xenopus as a BMP-binding protein that enhances BMP signaling (190). However, recent reports have shown that Tsg can also function as a BMP antagonist in Drosophila, and Xenopus by binding to both chordin and BMPs, resulting in an enhanced function of chordin (224-226). In mice, Tsg has been reported to act as a BMP agonist in skeletogenesis but as a BMP antagonist in T-cell development. This was shown by the observation that Tsg-deficient mice exhibit severe growth retardation with delayed endochondral ossification as well as an atrophic thymus in which BMP signaling is enhanced (227).

1.3.2 Regulation o f BMPs at the membrane level

At the membrane level, pseudoreceptors can regulate the signaling capacity of various BMPs (Figure 3). Bambi/Nma (BMP and activin membrane-bound inhibitor/non-metastatic gene A) is a transmembrane protein that is related to the type I receptors of the TGF-P family, showing highest homology to BMP type I receptors, but it lacks an intracellular kinase domain (228). Homologues of this pseudoreceptor have been identified in zebrafish, Xenopus, humans and mice (228-232). In mice, Bambi is expressed in multiple tissues including bone, where it is expressed by osteoblasts (232). During embryonic development, Bambi is co-expressed with BMP-2 and BMP-4 and its expression is regulated by these BMPs (228, 229, 231). Bambi stably associates with all TGF-P family type II

- 24 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

receptors, can bind TGF-P family ligands and inhibits BMP, activin as well as TGF-P signaling. It is believed that Bambi acts as a dominant-negative receptor whose effect is mediated by the shortened intracellular domain, which resembles the homodimerisation interface and prevents the formation of active receptor complexes (228, 229).

1.3.3 Intracellular regulation o f BMP activity

BMPs transduce their signal via specific R-Smads and the Co-Smad Smad-4. Intracellular modulation of Smad activity occurs by three mechanisms: inhibitory Smads, proteins binding signaling Smads, and degradation of signaling Smads (Figure 3).

1.3.3.1 Inhibitory Smads The inhibitory Smad-6 and Smad-7 lack a C-terminal phosphorylation motif, and in addition to an entire MH2 domain they contain only short segments with MH1 domain homology (233, 234). Both Smad-6 and -7 can be induced by various members of the TGF-P family, but in general Smad-6 is most potently induced by BMPs while Smad-7 is most potently induced by TGF-P (234, 235). Over-expression experiments have shown that Smad-6 and -7 can bind to activated type I receptors of both TGF-P and BMPs, and interfere with the phosphorylation of R-Smads by competing with the R-Smads for receptor binding. Because of the inhibition of phosphorylation of the R-Smads, the subsequent heterodimerization and nuclear localization are also impaired (233, 234, 236). However, at low expression levels Smad-6 specifically blocks signaling by the BMP/Smad-1 pathway. This inhibition does not occur through interfering with receptor-mediated phosphorylation of Smad-1, but by specifically competing with Smad-4 for binding to receptor-activated Smad-1. Smad-6 binding to receptor-activated Smad-1 yields a transcriptionally inert complex (237). Furthermore, it has been reported that Smad-6 can also function as a transcriptional co-repressor by binding to the homeobox protein Hoxc8 and the transcription co-repressor C-terminal binding protein (CtBP), thereby potentiating the repressor activity of both proteins (238, 239). In addition, both Smad-6 and Smad-7 were found to interact with HDACs, thereby inhibiting gene transcription directly (171). Conversely, Smad-6 activity can be regulated by Amsh (associated molecule with SH3 domain of STAM). Upon BMP stimulation, Amsh interacts with Smad-6 in the cytoplasm thereby repressing the inhibitory function of Smad-6 (240).

1.3.3.2 Proteins binding to signaling Smads Co-activators or repressors of Smad signaling can exert their effect through binding with Smads. This binding usually occurs via the MH2 domain of the Smads, but it can also occur via a combination of the MH2 and MH1 domains, via the MH1 domain alone, or via the linker region of the Smads. Proteins binding Smads through the MH2 domain include Oaz (Olf-1/EBF-associated ), Msg1 (melanocyte specific gene 1), the Ski family, Skip (Ski binding protein), Tob (transducer of ErbB2), Tgif (TG-interacting factor), Smad interacting protein-1 (Sip-1/Zeb-2/Zfhx1b), 8EF1, Piasy (protein inhibitor of activated signal transducer and activator of transcription y), MAN1, Sane (Smad-1 antagonistic effector) and menin, while proteins binding both MH1 and MH2 include Snip-1 (Smad nuclear interacting protein 1) and Znf8 (zinc finger protein 8). The multi-zinc finger protein Oaz specifically associates with BMP-activated R-Smads, and enhances Smad-mediated transcriptional activation of target genes in a cell-specific manner. Oaz can interact with Smad binding elements and it has been postulated that Oaz functions by enhancing the

- 25 - Chapter 1

binding of the Smads to DNA (241-243). In addition, Oaz can associate with poly(ADP-ribose) polymerase 1, an association that is essential for Oaz-mediated BMP-dependent gene regulation (243). Msg1 is a non-DNA-binding transcriptional activator that associates with Smad-4 in a TGF- P/BMP signaling-dependent manner. In addition, Msg1 also binds to CBP/p300, and it has been postulated that Msg1 functions to stabilize the Smad/CBP/p300 transcriptional activator complex thereby enhancing Smad-dependent gene transcription (244, 245). Furthermore, Msg1 is expressed in multiple tissues of mesodermal origin during development, including the limb bud, which indicates a possible role for Msg1 in skeletal development (246). Members of the Ski family include Ski itself and Sno (ski-related novel gene) (247). Ski was first identified as the avian homologue of the viral transforming protein of the Sloan-Kettering retroviruses (v-Ski) (248). Subsequently, it was shown that Ski can induce oncogenic transformation and myogenesis in avian cells (249). Ski is uniformly expressed throughout the cell, but in association with other proteins it has DNA binding properties (250). Furthermore, Ski and Sno can form a complex via their C-terminal regions (251, 252). Ski represses BMP, TGF-P and activin signaling as well as activation of target genes in mammalian cells (253-256). Ski can bind Smad-1 to -5, but not the inhibitory Smad-6 and -7, and an interaction of Ski with Smad-1 to -5 is required for its repression of BMP/TGF-P/activin signaling (254, 256, 257). Furthermore, interaction of Ski with Smad-1, -2, -3 and -5 requires receptor activation, while the interaction with Smad-4 is constitutive (254, 256, 257). Two mechanisms are known by which Ski can inhibit signaling by TGF-P family members. First, Ski can bind to Smad-4 in the cytoplasm, where it induces a constitutive binding of Smad-2/-3 with Smad-4. This binding is Smad-4-dependent, but independent of Smad phosphorylation, and inhibits ligand-dependent phosphorylation of Smad-2/-3 yielding inactive Smad complexes (258). Second, nuclear Smad/Ski/Sno complexes are able to bind to DNA at SBEs (254, 257). By recruitment of different co-repressors such as Hipk2 (homeodomain-interacting protein kinase 2) and N-CoR/SMRT, which are involved in the histone deacetylase macromolecular complex, nuclear Ski and Sno can repress signaling by TGF-P family members (259-261). Loss of Smad-4 binding by Ski completely abrogates the ability of Ski to inhibit BMP signaling, but it is still able to inhibit TGF-P signaling. This suggests that inhibition of BMP signaling occurs mainly through the first mechanism, while inhibition of TGF-P signaling results from a combination of both inhibition mechanisms (262). Ski is involved in the of craniofacial structures as was shown by the defects in patterning of vertebral and cranial skeletal structures in Ski knock-out mice (263). Skip (Ski interacting protein) was first identified as a nuclear protein able to bind to Ski (264) and was subsequently shown to act as a transcriptional co-activator of EBNA2 (Epstein-Barr virus nuclear antigen 2) (265). In addition, Skip augments TGF-P dependent transcription by binding to Smad-2 and -3, but this binding is not sufficient for complete activation of Smad-dependent transcription. Moreover, Skip competitively interacts with Ski/Sno for Smad-2/-3 binding, thereby relieving the Ski-mediated transcriptional repression of TGF-P (266). It is not known whether Skip also functions in BMP signaling, but it is likely that it is also able to bind Smad-1 and -5 in order to compete with Ski/Sno for transcriptional regulation of BMP target genes. Tob is a member of the PC3/BTG/Tob family of genes, which are involved in cell replication and differentiation (267, 268). Tob preferably binds the inhibitory Smad-6 and -7, thereby enhancing the interaction of Smad-6 with activated type I BMP receptors, which results in the inhibition of BMP signaling (269). In addition, Tob can associate with Smad-1, -4, -5 and -8, resulting in recruitment of these Smads to nuclear bodies and repression of BMP-induced Smad-dependent transcription in osteoblasts (270). Tob expression is induced during osteoblast differentiation, which is indicative for

- 26 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

a negative feedback loop in BMP signaling. Moreover, Tob-deficient mice have a greater bone mass, resulting from increased numbers of osteoblasts which are also more responsive to BMP-2 (270-272). The homeodomain protein Tgif was first identified as a Smad -binding protein, capable of binding Smad-2 and -3. Binding of Tgif to Smad-2 was shown to inhibit Smad-2-mediated transcription because of competition of Tgif and CBP/p300 for Smad-2 (273). Interestingly, upon TGF-P stimulation, Tgif not only binds N-CoR with its associated histone deacetylase but also mSin3, thereby inhibiting Smad signaling in a manner similar to but distinct from Ski/Sno (273-275). Although Tgif can bind weakly to Smad-1 (273), it is not known whether Tgif can also bind Smad-5 or -8 and function as a transcriptional repressor of BMP signaling. However, Tgif has been shown to activate signaling by Dpp, the Drosophila homologue of BMP2/4 (276), indicating that it might also function as a regulator of BMP signaling in mammals. Sip-1 is a two-handed zinc finger homeodomain which is closely related to 8EF1/Zeb- 1/Zfhx1a (277). Like 8EF1, Sip-1 binds DNA at two E-box sequences (CACCT and CACCTG) and inhibits transcription activation of its target genes in most cell types (277-279). Sip-1 and 8EF1 bind to Smad-1, -2, -3 and -5 upon activation of these R-Smads, but Sip-1 does not interact with Smad-4 (277, 280). In contrast to Sip-1, 8EF1 synergizes with Smad proteins in cells of the osteoblast lineage in order to activate transcription and to promote osteoblastic differentiation (280). Moreover, 8EF1 has been shown to be required for skeletal patterning, since mice deficient for the 8EF1 gene show defects in craniofacial, limb, ribs, sternum and vertebral bones (281). Sip-1 inhibits BMP-induced expression of Alp2 in osteoblastic cells and it has been suggested that extinguishing Sip-1 expression is a prerequisite for the induction of Alp2 (280, 282). These opposing effects of 8EF1 and Sip-1 arise from the differential recruitment of transcriptional co-activators (p300 by 8EF1) and co-repressors (CtBP by Sip-1) to the Smads during development (283). Piasy associates with Smad-3, -4, -2 and -1 with different affinities, and can form stable complexes with Smad-3/-4, Smad-2/-4, and Smad-1/-4. In addition, Piasy can associate with HDAC1. The association of Piasy/Smad complexes with HDAC1 inhibits Smad transcriptional activity by selectively repressing a subset of TGF-P/BMP responses (284). In contrast, another member of the Pias family, Pias3, has recently been shown to activate TGF-P signaling by binding both Smad-3 and CBP/p300 in a ligand dependent manner. Therefore, it has been postulated that the differential action of Piasy and Pias3 is based on a similar mechanism as Sip-1/8EF, where recruitment of CBP/p300 leads to the displacement of CtBP from 8EF1, which then binds CBP/p300 and activates transcription (285). Currently, the function of Piasy in bone development is unknown, but during embryonic development, Piasy is expressed in the developing limb bud indicating a possible role in bone formation (286). The inner nuclear membrane protein MAN1 has been shown to repress BMP-4-mediated transcription activation in Xenopus, and to bind Smad-1, -5 and -8 with high affinity, Smad-2 weakly, and Smad-4 not at all (287). Currently, the molecular action of MAN1 in regulating BMP signaling is unknown, but it may compete with transcription factors and co-activators to bind R-Smads, or it may recruit additional co-repressors to the Smads and thereby repress BMP signaling (287). Sane is a novel protein with sequence similarity to nuclear membrane proteins such as MAN1. Sane specifically inhibits BMP-mediated signaling and transcription activation both in Xenopus and in osteoblastic cells (288). Sane interacts with Smad-1 and -5, but not or only very weakly with Smad-2, -3 and -4. Unlike MAN1, Sane specifically binds type I BMP receptors in addition to the BMP-specific R-Smads, which implies that Sane functions in the cytoplasm. In concordance with this, Sane inhibits the phosphorylation and subsequent nuclear localization of BMP-specific R-Smads. The interaction with both a receptor and R-Smads makes Sane resemble SARA, but because of its

- 27 - Chapter 1

antagonistic effect Sane most likely represents a new mode of regulating intracellular BMP signaling (288). Menin is the product of Men1 and mutations in this gene are responsible for multiple endocrine neoplasia type 1. Menin has been shown to be a nuclear protein, which is widely expressed during development and in adult mammalian tissues (289-292). Recently, menin has been shown to bind a variety of proteins including JunD, Smad-1, -3 and -5, but not Smad-2 or -4. Interestingly, menin represses JunD-activated transcription, but enhances Smad-3-mediated transcription (293-295). In mice, menin has been shown to play an important role in development and bone formation, since homozygous Men1 knock-out mice are embryonically lethal (E14.5) exhibiting defects in cranial/facial development (296). Furthermore, it has been shown that menin has a dual role during osteoblast differentiation: in the early stages of osteoblast differentiation, menin mediates specifically the BMP-induced commitment of MSCs to the osteoblastic lineage, while during the late phase of differentiation it serves as an inhibitor of osteoblast maturation (295, 297). To exert this positive effect on the commitment of MSCs to osteoblasts, menin not only interacts with Smad-1/-5 but also with Cbfa1, thereby enhancing the transcriptional activity of Smad- 1/-5 and Cbfa1 on their target genes. In more differentiated cells, menin does not bind Smad-1/-5 or Cbfa1, but instead interacts with Smad-3 and blocks BMP-induced Smad-1/-5 and Cbfa1 transcriptional activity (297). Snip-1 is a nuclear protein that represses the CBP/p300-dependent transcriptional activation of multiple inducers including TGF-P family members (298, 299). Snip-1 specifically binds Smad-4 upon receptor activation by TGF-P family. In addition, Snip-1 interacts with both CBP and p300 in a ligand dependent manner, and can compete with Smad-4 for binding to CBP/p300, thereby inhibiting Smad-dependent promoter activation (298). The nuclear Kruppel-like zinc finger protein Znf8 interacts with Smad-1, -2, -3, -4 and -5 with different affinities, and interaction between Znf8 and Smads is enhanced upon phosphorylation of R- Smads. Interestingly, it has been shown that Znf8 specifically inhibits a subset of BMP responses. Currently, it is not known whether Znf8 functions as a transcriptional repressor, a general inhibitor such as Snip-1, or a co-repressor such as Ski and Tgif. Moreover, as a result of the different affinities of Znf8 for the various Smads, Znf8 might set thresholds for signaling, providing a new mechanism for cells to have different sensitivities for various ligands (300). Although Znf8 is ubiquitously expressed in mice during embryonic development, it is not known whether Znf8 can function in bone to inhibit BMP responses. The transcription factor Yin Yang 1 (YY1) binds the MH1 domain of Smad-1, -2, -3 and -4 with different affinities, and is able to inhibit TGF-P family member signaling by blocking the binding of activated R-Smads to their DNA binding sites. Furthermore, Smad-induced osteogenic differentiation has been shown to be repressed by ectopically expressed YY1, while it was enhanced upon knock-down of YY1 (301). One of the few proteins that are known to bind Smads through the linker region is Smif (Smad-4 interacting factor), which is a ubiquitously expressed transcriptional co-activator that specifically binds to Smad-4 (302). The formation of a Smif/Smad-4 complex occurs in the cytoplasm and is enhanced by TGF-P/BMP treatment. Upon formation, the complex translocates to the nucleus and, in combination with CBP/p300, it activates transcription of target genes. Smif is not able to bind CBP/p300 directly, which implies that the recruitment of this co-activator occurs through Smad-4/R- Smad complexes. Furthermore, loss of Smif expression causes widespread defects in embryonic development implicating that Smif is involved in many, but not all, developmental processes in which TGF-P family members are involved (302).

- 28 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

Recently, Ciz/Nmp4 (Cas interacting protein/nuclear matrix protein 4) has been identified as a Smad-binding protein and therefore capable of regulating signaling of TGF-P family members, but the molecular mechanism of its binding remains to be established. Ciz is a zinc finger protein that inhibits the transcriptional activation of Smad-1 and -5 on SBEs, thereby suppressing BMP-induced expression of Cbfa1, Alp2, Ocn and Col1a in osteoblastic cells (303, 304). In addition, Ciz expression is regulated by BMP2, which is indicative for a feedback loop to regulate BMP-signaling (304).

1.3.3.3 Degradation of signaling Smads Proteasome-mediated degradation of Smads and Smad binding proteins is an important element in the down-regulation of BMP/TGF-P signaling, and it has been shown that both ubiquitin-dependent and -independent pathways are involved in this regulation of BMP/TGF-P signaling. Smurf-1 (Smad ubiquitination regulatory factor 1) and Smurf-2 both are members of the HECT-family of E3 ubiquitin ligases and have been shown to inhibit BMP/TGF-P signaling (305­ 307). Presently, two mechanisms are known by which Smurfs regulate BMP/TGF-P signaling. First, Smurf-1 and Smurf-2 bind to activated R-Smads at their linker region and ubiquitinate the Smads, after which these are degradated by the proteasome. Smurf-1 binds Smad-1 and -5, while Smurf-2 binds with high affinity to Smad-1 and Smad-2, with intermediate affinity to Smad-3, but does not interact with Smad-4. Moreover, Smurf-1 is more potent in inhibiting BMP signaling than Smurf-2 (305-307). Furthermore, Smurf-2 can also interact with Smad-binding proteins such as Sno. Binding of Smurfs to Smad-binding proteins is mediated by activated R-Smads and ubiquitination of Smad-binding proteins by Smurfs targets them to the proteasome for degradation (308). Second, Smurf-1 and Smurf-2 bind the inhibitory Smad-6 and -7 via their linker region, and can associate with type I receptors via these I-Smads. Binding of Smurf-1 or Smurf-2 to type I receptor/I-Smad complexes ubiquitinates both the receptor and the inhibitory Smads, and targets them to the proteasome for degradation (309-311). Recently, it has been shown that Smurf-1 inhibits osteoblast proliferation and differentiation in vivo and in vitro resulting in impaired bone formation (312-314). Inhibition of osteoblast proliferation and differentiation results from the finding that Smurf-1 not only degradates Smad-1 and -5 in osteoblastic cells, but is also able to bind and ubiquitinate the osteoblast specific transcription factor Cbfa1, which is followed by proteasomal targeting. Subsequent degradation of Cbfa1 results in severely reduced expression of Osx, Alp2, Ocn and Col1a, as well as in inhibition of the formation of bone nodules (312-314). Arkadia is a RING finger domain protein with E3 ubiquitin ligase activity, but it shows no homology to other E3 ubiquitin ligases. Arkadia directly interacts with Smad-7 and induces its ubiquitination and subsequent degradation. In contrast to the Smurfs, Arkadia/Smad-7 complexes are not recruited to the type I receptors, and Arkadia does not degrade the receptors. Therefore, the effect of Arkadia is to enhance both BMP and TGF-P signaling (315). Chip (Carboxyl terminus of Hsc70-interacting protein) is a U-box dependent E3 ubiquitin ligase, which has initially been identified as a co-chaperon protein. Chip binds to the MH2 domain of Smad-1, -2, -3 and -4 independent of receptor activation. Upon binding, the Smads become ubiquitinated and targeted to the proteasome for degradation. Chip-mediated degradation of Smad-1/- 4 results in the inhibition of Smad-mediated transcription activation (316). In addition to these ubiquitination-mediated degradation pathways, it has recently been shown that the Smad-binding protein Snip-1 can be degraded by the proteasome upon interaction with a complex that includes Smad-1, the proteasome P subunit HsN3 and the ornithine decarboxylase antizyme (Az), independent of ubiquitination. The formation of this Smad-1/HsN3/Az complex is

- 29 - Chapter 1

enhanced by type I receptor activation, and Snip1 degradation is therefore Smad-1, Smad-4 and Az- dependent (317). Mechanisms as described above suggest a balance of activated Smad complexes and repressor proteins in the cytoplasm and subsequently the nucleus, which is critical for regulation of the signal transduction pathways activated by TGF-P family ligands.

1.4 Other signaling routes in osteoblast differentiation and interactions

Other growth factors than BMP and TGF-P also play a role in osteoblast differentiation. These factors exert their effect through various signal transduction pathways. Most of the growth factors activate members of the different MAP-kinase signaling pathways, including p38, SAP/JNK and Erk1/2. Furthermore, the Wnt and Notch signaling pathways have been shown to be important regulators of osteoblast differentiation and therefore these pathways and their interactions with the BMP/Smad-pathway will also be discussed here.

1.4.1 MAP-kinases

For activation of the Erk1/2 MAP-kinase pathway, growth factor receptor complexes activate Ras, upon which Raf1, Mek1/2 and Erk1/2 are activated by stepwise phosphorylation. The Jnk MAP- kinase pathway is also activated through Ras by its ability to activate Rac1, resulting in the subsequent activation of Pak1/2, Map3K4, MKK7 and Jnk1-3. The p38 MAP-kinase pathway can be regulated through both Ras and Tak1. Ras activation leads to the activation of Rac1, which activates p38 through either Mekk1 and Mkk4 or Mlk1 and Map2K6, while activated Tak1 leads to the activation of p38 via MEK3/6 (for reviews see refs. 318-320). In order to regulate gene expression, the Sap/Jnk, Erk1/2 and p38 MAP-kinase pathways employ amongst others the AP1 transcription factor complex, which is composed of dimers of Fos and Jun family members, as well as some members of the CREB/ATF families. Fos and Jun are nuclear effectors of the JNK and Erk1/2 MAP-kinase pathway, while CREB/ATF members are effectors of the p38 MAP-kinase pathway. The Fos proteins (c-Fos, FosB, Fra-1 and Fra-2) can only heterodimerize with members of the Jun family, whereas Jun proteins (c-Jun, JunB and JunD) can both homodimerize or heterodimerize with Fos members in order to form transcriptionally active complexes (for review see ref. 321). In osteoblasts, MAP-kinase/AP1 activity has been shown to be induced by a variety of factors including PTH and vitamin D, but also by TGF-P and BMPs (322-324). BMP activation of the various MAP-kinase/AP1 pathways in osteoblasts has been shown to take place at multiple levels (Figure 4). BMP has been shown to differentially stimulate p38, Erk and JNK activity in osteoblastic cells; secondly, BMP stimulates the activity of several individual AP1 components, including c-Fos, FosB, Fra-1, Fra-2, JunB and ATF2; and thirdly, BMP induces the expression of the AP1 components c-Fos, c-Jun and JunB (117, 325, 326).

- 30 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

Ill Transcription regulation of c-Fos, c-Jun, JunB

Figure 4. Interactions of the MAPKinase and Smad signal transduction pathways. To regulate osteoblast differentiation, multiple points of convergence between the MAPKinase and Smad pathways can be identified: I) BMPs differentially activate p38, Erk1/2 and Jnk, II) BMPs activate individual API components, III) BMPs regulate the expression of API components, IV) p38 MAPKinase stimulates Smad-1 phosphorylation, V) Erk1/2 MAPKinase activates the Cbfal protein, VI) Erk1/2 MAPKinase can phosphorylate R-Smads, and VII) Erk1/2 and Jnk MAPKinase can regulate the activity of Smad binding proteins. See text for details. See page 135 for full color figure

The differential activation of p38, JNK and Erk1/2 MAP-kinase pathways by BMPs may underlie the differential effects of these pathways on BMP-induced osteoblast differentiation. Recently, it has been postulated that BMP-induced activation of the Tab/Tak1/p38 MAP-kinase pathway is possibly mediated by Bram1 and/or Xiap (see above), providing a point of convergence of the p38 MAP-kinase and BMP signal transduction pathways (148, 149, 327, 328). Furthermore, p38 MAP-kinase has been shown to be a positive regulator of osteoblast differentiation, since inhibition of this pathway results in down-regulation of the osteoblast marker genes Cbfa1, Alp2, Col1a, Ocn, Opn and Opg, as well as in down-regulation of matrix mineralization (322, 329-335). Possibly, this results from the convergence of these pathways at the level of Smad-1 phosphorylation, since inhibition of

- 31 - Chapter 1

p38 MAP-kinase has been shown to suppress BMP-induced Smad-1 phosphorylation and its subsequent translocation to the nucleus (Figure 4) (332). It has been shown that JunB is required for optimal expression of Cbfa1 (335). Because JunB is a direct target gene of BMP2 and TGF-P (157), it is possible that expression of JunB is inhibited when Smad-1 phosphorylation and nuclear translocation is blocked by inhibition of p38 MAP-kinase. This will subsequently inhibit Cbfa1 expression and hence osteoblast differentiation. Recent data suggest that in addition to the p38 MAP-kinase pathway, also the JNK pathway positively regulates BMP-induced osteoblast differentiation, since inhibition of c-Jun phosphorylation decreased osteoblast differentiation, as shown by the down-regulation of Ocn expression. However, Alp2 expression was only slightly increased under these conditions and the effect on mineralization capacity of the cells was not determined (336). In contrast to p38, Erk1/2 MAP-kinase has been shown to be mainly a negative regulator of osteoblast differentiation. Although complete inactivation of Erk1/2 activity inhibits osteoblast differentiation (337), this effect is mainly due to the complete inactivation of the Cbfa1 protein upon inhibition of Erk1/2 activity, since Erk1/2-mediated phosphorylation of Cbfa1 has been shown to be essential for Cbfa1 protein activity (Figure 4) (33). Surprisingly, partial inhibition of Erk1/2 MAP- kinase results in residual Cbfa1 activity, and enhancement of BMP-induced osteoblast differentiation, as shown by the up-regulation of Alp2 activity as well as matrix mineralization. Furthermore, expression of constitutively active MEK1, which activates Erk1/2, results in the down-regulation of BMP-induced osteoblast differentiation, as shown by repressed Alp2 activity (338). Another point of convergence of the Erk1/2 MAP-kinase and Smad pathways may be at the level of the R-Smads: for TGF-P it has been shown that Erk1/2 can phosphorylate R-Smads at their linker region, and depending on the cell type, this can result in either a negative or a positive effect on activation of TGF-P signaling (339). In epithelial cells, it has furthermore been shown that Smad-1 can be phosphorylated at its linker region following EGF-induced Erk1/2 activation, which causes nuclear exclusion of Smad-1/-4 complexes and therefore inhibition of BMP-induced transcriptional activity (340). It is not known, however, whether Smad-1 can be phosphorylated at the linker region in osteoblastic cells as a result of BMP-dependent or -independent activation of Erk1/2, nor if this phosphorylation regulates osteoblast differentiation in a positive or negative manner (Figure 4). MAP-kinase and Smad pathways may furthermore be linked through the cytoplasmic and nuclear Smad binding proteins. For instance, Erk1/2 and Jnk have been shown to phosphorylate Tob, an inhibitor of BMP-induced osteoblast differentiation. Activation of Tob inhibits its anti-proliferative function in NIH3T3 cells (341, 342). However, it is not yet known whether phosphorylation of Tob in osteoblasts results in inhibition or enhancement of differentiation (Figure 4). The various members of the AP1 complex are differentially expressed during osteoblast differentiation in vitro. Initially, all Fos and Jun proteins are highly expressed, but during late phase osteoblast differentiation, their levels decline and Fra-2 and JunD become the major proteins of the AP1 complex in fully differentiated osteoblasts (343). Furthermore, it has been shown that several of the individual components of the AP1-complex are important for bone formation during embryonic development, since a number of knock-out and transgenic mice display various bone . In this respect, it has been shown that c-Fos knock-out mice display osteopetrosis (overly dense bones due to osteoclast deficiency), while transgenic mice with ectopic expression of Fra-1 exhibit osteosclerosis (overly dense bones due to increased numbers/activity of osteoblasts). Furthermore, transgenic mice with ectopic expression of c-Fos suffer from osteosarcomas. Finally, mice in which the c-Jun protein is inactivated, suffer from scoliosis, a side-to-side curvation of the spine. Many genes related to the osteoblastic phenotype have been shown to contain AP1 binding sites in their promoters, and therefore to be regulated by this transcription factor complex (for review

- 32 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

see ref. 344). Since many osteoblast marker genes are also regulated by Cbfa1, it is likely that AP1 and Cbfa1 act in concert. Indeed, it has been shown that Cbfa1 and AP1 interact with each other and that this complex regulates collagenase-3 expression during osteoblast differentiation (345, 346). To regulate the activity of the AP1 complex, activated R-Smad/Co-Smad complexes have been shown to interact with individual AP1-components, thereby not only stimulating transcriptional activity of target genes, but also enhancing the up-regulation of AP1 by the Erk1/2 pathway (322, 347).

1.4.2 Wnt-signaling

Wingless/Wnt family members are secreted glycoproteins that bind and activate cell surface receptor complexes consisting of the Frizzled family of G-protein coupled receptors and the low- density lipoprotein receptor-related protein 5 and 6 (LRP-5 and -6) (Figure 5) (348). The family of Wnt proteins can be divided in two subfamilies: the P-catenin-dependent Wnts such as Wnt-1, -2, -3 and -3a, and the P-catenin-independent Wnts such as Wnt-4, -5a, -5b, -6 and -7 (for review see ref. 349). In the P-catenin-dependent pathway, absence of Wnt signaling enables glycogen-synthase kinase 3 (Gsk-3) to phosphorylate P-catenin, thereby inducing the degradation of P-catenin by the ubiquitin-proteasome pathway (Figure 5). Upon activation of Wnt signaling, Gsk3 activity is inhibited by the protein disheveled (Dsh). This prevents P-catenin degradation, resulting in its accumulation, nuclear translocation and association with members of the Lef/Tcf (Lymphoid enhancer binding factor/T-cell-specific factor) transcription factor family, which function in P-catenin mediated gene transcription (349). In contrast, in the P-catenin-independent pathway, the receptor complex consists of Frizzled, or frizzled domain containing proteins, with the proteoglycan Knypek as a co-receptor. Activation of this P-catenin-independent pathway also leads to the activation of Dsh, but at different protein domains than in the P-catenin-dependent pathway. Subsequently, this differential activation of Dsh leads to the activation of the Jnk signal transduction pathway, as well as signal transduction pathways in which Ca2+ flux and small G-proteins are essential (349) (Figure 5). Both the Wnt/P-catenin and Wnt/Ca2+ pathway have been shown to be important for bone formation, such that the P-catenin-dependent pathway has been shown to enhance osteoblast differentiation and possibly also chondroblast differentiation, whereas the Wnt/Ca2+ pathway enhances chondroblast differentiation and possibly inhibits osteoblast differentiation (349). The P- catenin-dependent pathway is known to enhance osteoblast differentiation, since inactivating mutations in the Wnt co-receptor LRP5 have been shown to decrease bone accrual during growth. Activating mutations cause a phenotype with increased bone mass (350, 351). Furthermore, secreted frizzled-related protein 1 (sFrp-1) is an antagonist of the Wnt/P-catenin canonical pathway and sFrp-1 knockout mice display increased trabecular bone formation, caused by an inhibition of osteoblast/osteocyte apoptosis and enhanced osteoblast proliferation and differentiation (352, 353). In addition, BMP-mediated up-regulation of Alp2 is dependent on P-catenin. Activation of the P- catenin/Lef/Tcf pathway not only stimulates endochondral bone formation, but also participates in BMP-induced osteoblast differentiation, as shown by increased mineralization (354-356).

- 33 - Chapter 1

Figure 5. Interactions of the Wnt and Smad signal transduction pathways. Points of convergence between the Wnt and Smad pathway: I) BMPs induce the expression of Wnt signal transduction genes including Wnt, Frizzled and Lef1, II) Smad-4 is a co-activator of P-catenin, III) Lef/P-catenin regulates activity of Cbfa1, and IV) b-catenin regulates the activity of Smad binding proteins. See text for details. See page 136 for full color figure

The effects of P-catenin-dependent Wnt signaling on chondroblast differentiation are still controversial. It has been shown that constitutively active P-catenin induces ectopic endochondral bone formation, but on the other hand, over-expression of P-catenin-dependent Wnts results in inhibition of chondroblast differentiation both in vivo and in vitro (356, 357). Moreover, Wnt- dependent P-catenin and Ca2+ pathways can inhibit each other’s activity, which suggests that the relative levels of activity of these two pathways might be an important factor in the decision of cell fates (358). Interaction of the Wnt and BMP/Smad signal transduction pathways occurs at various levels (Figure 5). BMPs induce the expression of Wnt and Frizzled during osteoblast differentiation, indicating that a Wnt autocrine loop may be operative during BMP-mediated osteoblast differentiation (355). It is not yet known, however, whether this expression is regulated by Smads or by MAP-kinases, or a combination of these two pathways. Furthermore, Lef1 expression has been shown to be induced by BMPs during tooth development, a process very similar to osteoblast

- 34 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

differentiation (359), which suggests that Lef/Tcf transcription factors could also be induced by BMPs during osteoblast differentiation. In addition, Smad-4 functions as a co-activator of P-catenin in Wnt-signaling. It has been shown that R-Smads and Smad-4 can interact with Lef/Tcf transcription factors and thereby mediate cooperative signaling between the TGF-P and Wnt pathways (360-362). Furthermore, Lef/P-catenin complexes have been shown to bind Cbfa1, thereby inhibiting the Cbfa1-dependent activation of the Ocn promoter (34). This indicates that various genes of which the expression is regulated by Cbfa1, are also regulated by the Wnt/P-catenin pathway. Finally, the Wnt/P-catenin and Smad pathway may converge at the level of regulation of Smad-binding proteins, similarly to the convergence of the MAP-kinase and Smad pathways. In tumor cells, for instance, P-catenin has been shown to induce the transcription of Bambi, a negative regulator of BMP signaling, resulting in an inhibition of growth arrest in these cells (363). At the moment, it is not known, however, whether Wnt/P-catenin signaling also mediates the BMP-induced expression of Bambi in osteoblastic cells.

1.4.3 The Notch pathway

Notch is a single transmembrane protein receptor with a large extracellular domain containing multiple epidermal growth factor (EGF)-like repeats and three notch-specific repeats (LNRs). The small intracellular domain of Notch contains several functional domains, including ankyrin repeats flanked by nuclear localization repeats and a transactivating domain (TAD) (for review see ref. 364). Signaling-competent Notch is synthesized from a precursor protein by cleavage of the extracellular domain from the rest of the protein by a furin protease in the Golgi complex. The resulting parts of Notch are non-covalently linked by a calcium-dependent interaction in the absence of ligand binding (Figure 6). In addition to the Notch receptor, its ligands Delta and Serrate/Jagged are also single transmembrane proteins, requiring cell-cell contact for optimal signal transduction through the Notch receptors (Figure 6). Soluble ligands are known, but they have been shown to function as dominant negative proteins (364). Upon ligand binding, the Notch receptor is subject to two additional proteolytic steps for complete activation. First, the bulk of the extracellular domain is cleaved off by the metalloprotease TACE (TNF-a-converting enzyme), leaving a membrane-tethered intracellular domain. Second, intramembranous cleavage by presinilin-dependent y-secretase releases the Notch intracellular domain (NICD), which subsequently translocates to the nucleus and binds to the transcription factor Cbf1/Su(H)/Rbp-J (Suppressor of Hairless/recombining binding protein suppressor of hairless) (Figure 6). In the absence of Notch signaling, Cbf1 represses gene transcription through the recruitment of the co-repressor Smrt, a histone deacetylase (HDAC), and Skip (Ski interacting protein). NICD binds to the Cbf1/Smrt/HDAC/Skip complex and thereby displaces Smrt and its associated HDAC, resulting in a relieve of transcriptional repression (for reviews see refs. 364, 365). In addition, the NICD/Cbf1 complex recruits the co-activator Maml (Mastermind-like), which binds the transcriptional activator protein p300/CBP to form a NICD/Cbf1/Maml/p300 complex. The subsequent recruitment of P/CAF to this complex enhances Notch signaling (366, 367) (Figure 6).

- 35 - Chapter 1

Delta, Serrâte/Jagged

direct target genes Hes-1 and Hey1

Figure 6. Interaction of the Notch and Smad signal transduction pathway. The Notch intracellular domain (NICD) interacts with R-Smad/Co-Smad complexes in the nucleus thereby inducing the expression o f Notch transcription factors such as Hes-1. See text for details. See page 137 for full color figure

In vertebrates, the family of Notch receptors consists of four highly homologous proteins (Notch 1-4), and the ligands Delta and Serrate/Jagged consist of three and two family members, respectively (Delta-like-1, -2 and -4; Jagged-1, and -2). Activity of the Notch receptor is regulated by glycosylation at O-fucose residues through the action of the so-called fringe gene products. In vertebrates, the family of fringe genes consists of three members, lunatic fringe (Lfng), manic fringe (Mfng) and radical fringe (Rfng). O-fucosylation at the EGF-like repeats of Notch is essential for signaling, since loss of the fucose transferring protein (O-FucT-1) causes embryonic lethality of mice at mid-gestation as a result of severe defects in somitogenesis, vasculogenesis, cardiogenesis, and neurogenesis. Furthermore, inactivation of Lfng in mice causes defects in somitogenesis, skeletal defects and perinatal death (368-371). Strikingly, it has been suggested that the distribution of fringe proteins, rather than that of Notch receptors or its ligands, is the key determinant in the control of Notch signaling (372).

- 36 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

Because inactivation of fringe genes causes skeletal defects in mice, it is likely that the Notch signal transduction pathway plays an important role in bone formation. However, reports on the effects of the various ligands and receptors on Notch signaling in osteoblast differentiation are still contradictory (373-375). On the one hand, it has been reported that over-expression of the NICD in osteoblastic cells impairs osteoblast differentiation (373, 374, 376), while others have reported that over-expression of the NICD stimulates osteoblast differentiation (375). Members of the Hes/Herp (hairy and enhancer of split) family are primary target genes of Notch signaling and either homodimerize or heterodimerize to exhibit their effects (377, 378). Hes-1 has been shown to be a negative regulator of osteoblast differentiation probably through interactions with the Groucho family of transcriptional repressors, which are also known to be involved in regulating Wnt signaling (379-382). On the other hand, the osteoblast specific transcription factor Cbfa1 has also been shown to bind Hes1, thereby releasing the inhibitory Groucho protein. Association of Hesl with Cbfal has been shown to stimulate the transactivating function of Cbfal on Ocn transcription (35). Reports on the interaction of the Notch signal transduction pathway with the Smad-pathway are only scarce. It has been shown, however, that the NICD can interact with both TGF-P- and BMP- specific Smads (Figure 6). TGF-P can rapidly induce the expression of Hes1 in a NICD-and Smad- dependent manner. Furthermore it has been shown that NICD can bind to Smad-3 at its MH2-domain. This interaction is necessary for the activation of promoters containing both Notch- and Smad- binding elements (383). Moreover, it has been shown that Smad-1 and NICD also interact functionally to inhibit both myogenesis and BMP-induced osteogenesis of the C2C12 mesenchymal progenitor cell line, probably through up-regulation of specific Notch target genes (384). In addition to the cross talk between TGF-P and Notch signal transduction pathways, it has been described that the Notch and Wnt pathways also functionally interact. Gsk-3, a negative regulator of Wnt/P-catenin signaling (see above), can phosphorylate the NICD and thereby inhibits the expression of Notch target genes. Moreover, Wnt-induced inactivation of Gsk-3 leads to an up- regulation of Hes-1 transcription (385). Furthermore, it has been shown that the NICD and Lef-1, an effector of Wnt signaling, can interact in vivo resulting in the activation of specific, Wnt/P-catenin independent, promoters (386). The importance of these interactions for bone development is not yet known, but certainly requires further investigation.

1.5 Clinical skeletal disorders associated with osteoblast differentiation

BMPs have critical functions during embryonic development in mice with respect to mesoderm formation and patterning, left-right asymmetry, neural patterning, somite patterning, skeletal development and limb patterning as well as in the development of nearly all organs including tooth, lung, kidney, gut, skin and gonads (108). Although it may be expected that BMPs exhibit similar functions in humans, it was only recently that clinical disorders have been described which are associated with BMPs. These disorders include fibrodysplasia ossificans progressiva (FOP) and a number of chondrodysplasias such as Hunter-Thompson Chondrodysplasia (HTC), Grebe Syndrome and Brachydactyly C (for review see ref. 387). FOP is a severely disabling, autosomal dominant disorder, which only occurs sporadically. FOP is progressive with age and no treatments are currently available. FOP has three distinguishing

- 37 - Chapter 1

features: 1) congenital malformations of the great toes; 2) heterotopic ossifications of soft connective tissues such as tendon, ligament, fascia and skeletal muscle; and 3) temporal progression of osteogenesis in characteristic anatomic patterns (for review see ref. 388). Patients with FOP have in fact two skeletons: one that is grossly normal and which is formed during embryogenesis, and a heterotopic one which is formed after birth and fuses with the normal skeleton, although it is formed independently of the normal skeleton. This fusion of the heterotopic skeleton with the normal skeleton results in impaired muscle movement. Heterotopic bone formation usually begins in the first decade of life and by the time the patient is in his/her thirties, he/she is unable to move independently anymore, except for some small movements such as chewing and grasping. The heterotopic bone is formed through the process of endochondral ossification, and heralded by large, painful, tender, highly vascularised, soft-tissue swellings (lesions) that develop rapidly and spontaneously, but can also develop following minor traumas such as cuts, bruises and intramuscular injections (388). The lesions consist of many cell types including lymphocytic cells, mast cells and smooth muscle cells of vascular origin (389-391). According to current ideas, the combination of symptoms in FOP patients is caused by alterations in the BMP signaling cascade. It has been shown that in lymphocytic cells of FOP lesions expression of BMP-4 is up-regulated due to enhanced transcription of the gene (389, 392), but this probably reflects more a consequence than the cause of the disease, since the gene loci of FOP (see below) have not been mapped to the BMP-4 locus on human chromosome 14 (393). In addition, no mutations in the BMP-4 gene have been described in FOP patients (394). One of the gene loci for FOP has been mapped to human chromosome 4q27-31, which includes genes for BMP signaling intermediates such as Smad-1 and BMPR1B as well as interleukin 15 (IL-15) and Lef1 (395). Another locus for FOP has been mapped to human chromosome 17q21- 22, which includes the gene for noggin, an inhibitor of BMP signaling (396). In support of this observation, several mutations in the noggin gene have been identified in non-related FOP patients (397, 398). These mutations all result in the inability of the noggin protein to become acetylated at amino acid residues 89 ^ 94, which probably results in impaired protein activity (398). Furthermore, it has been shown that the elevated BMP-4 activity in FOP lesional cells is not primarily caused by the over-expression of BMP-4 itself, but more by the lack of active noggin, which renders the cells responsive to BMP-4 for a longer period of time than observed normally (399). Moreover, in a mouse model of BMP-4-induced heterotopic ossification, restoration of the noggin-BMP feedback loop inhibited in vivo ossification (400). These data suggest that in FOP patients the feedback loop for BMP signaling is defect, specifically in lesional cells, causing an attenuated BMP response and induction of endochondral bone formation. However, the question why these patients develop spontaneous lesions, and why the effect of a defect BMP-feedback loop is so devastating particularly in these lesions, but not in other cells, has not yet been solved. The Hunter-Thompson Chondrodysplasia (HTC), Grebe Syndrome and Brachydactyly C disorders belong to the acromesomelic type of chondrodysplasias, where the skeletal phenotype exhibits abnormalities predominantly in the middle and distal segments of the upper and lower limbs (387). Both HTC and Grebe Syndrome are characterized by severe long bone shortening which progresses in a proximodistal fashion: the tips of hands and feet are more affected than the arms and legs, and the feet and legs are more affected than the hands and arms. However, patients with Grebe Syndrome are more affected than HTC patients. In Brachydactyly C, the individual bones in the hands and feet are altered in size and/or shape, but the rest of the skeleton is hardly affected (387). It has been shown that all these diseases result from mutations in the gene for CDMP-1 (cartilage-derived morphogenic protein-1), also known as Gdf-5 in mice, which belongs to the Gdf-

- 38 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

subfamily of BMPs. Gdf proteins are mainly involved in cartilage development, and their bone- promoting activity both in vivo and in vitro is relatively limited when compared with BMPs. Moreover, the expression pattern of CDMP-1/Gdf-5 predominates in and around the developing limbs, with striking expression in the joint interzones (387). HTC patients display a homozygous mutation in the CDMP-1 gene which is caused by a 22- bp insertion, resulting in an altered open-reading frame after the third cysteine codon, thereby completely disrupting the tri-dimensional form of the protein, which consequently results in loss of protein function (401). Grebe Syndrome patients have a tyrosine-for-cysteine mutation at amino acid 400 (C400Y), which impairs secretion of the protein. Moreover, the mutated protein also behaves as a dominant-negative form, which suppresses not only the other CDMP-1 allele, but also selectively interferes with the secretion of other BMPs, which might explain the severity of Grebe Syndrome in comparison with HTC (402). In Brachydactyly C patients, various mutations in CDMP-1 have been observed that all lead to functional haploinsufficiency of the protein, causing a less severe type of chondrodysplasia (403, 404).

1.6 Aim of the thesis

The aim of the present research was to identify genes that act as key- or co-regulators of BMP-induced osteoblast differentiation and that may function as focal points for the interaction of the BMP pathway with other signal transduction pathways relevant in osteoblast differentiation. A characterization of BMP-induced regulation of the differentiation of MSCs to osteoblasts is very important for understanding the bone forming part of bone homeostasis. Such knowledge might facilitate the identification of new putative drug targets that promote bone formation by enhancing osteoblast differentiation. Ultimately, these new drug targets might enable the development of novel therapies to treat serious clinical disorders such as osteoporosis in which bone homeostasis is disturbed. To address these questions, a detailed description of the molecular processes by which BMPs regulate osteoblast differentiation is needed. In this context, the gene expression regulation of several thousands of genes was studied during BMP-induced osteoblast differentiation using the mouse mesenchymal C2C12 cell line. This cell line was used as a model system since it differentiates into osteoblasts upon BMP stimulation, whereas it differentiates towards myoblasts in the absence of BMPs. Moreover, TGF-P has been shown to block both differentiation pathways in these cells (111, 405). To determine gene expression profiles, the “gene expression microarray” (GEM) technology was used. This approach allows the simultaneous monitoring of expression of several thousands of genes under different experimental conditions (406-410). In Chapter 2, gene expression profiles were determined of C2C12 cells stimulated for two hours by either BMP2, TGF-P or activin A. This enables differential expression analysis of immediate early response genes as induced by these growth factors. The growth factors tested are all members of the TGF-P superfamily, but they exhibit different capacities in regulating osteoblast differentiation. Four transcripts, one gene and 3 ESTs, were identified to be regulated by BMP2 and TGF-P in an opposite manner, indicating that these transcripts might be important novel regulators of growth factor-induced osteoblast differentiation. In Chapter 3, regulators of osteoblast differentiation were identified by BMP-induced gene expression pattern analysis during the early phases of osteoblast differentiation. A large set of genes (184 genes and ESTs) was identified that could be subdivided into several clusters according to their

- 39 - Chapter 1

time-dependent expression patterns. These clusters contain functionally related genes, representing the various phases of osteoblast differentiation. Signal transduction regulatory factors are mainly represented in the clustered subset of immediate early genes, whereas the intermediate early and late early clusters consisted primarily of genes related to processes that modulate morphology, basement membrane formation and synthesis of extracellular calcified matrix. Two genes were identified that exhibited very similar expression characteristics to the osteoblast-specific type II isoform of Cbfal, one belonging to the Wnt signal transduction pathway (Tcf7) and one belonging to the Notch signal transduction pathway (Heyl). Subsequent transfection experiments showed that ectopic expression of Tcf-7 regulates the expression of osteoblast differentiation marker genes which suggests that Tcf-7 enhances BMP-2 induced osteoblast differentiation. In Chapter 4, gene expression profiles were determined to identify novel genes that are differentially regulated by BMP-2 and TGF-P during early phase osteoblast differentiation. Genes that are highly induced by BMP-2, but not by TGF-P or are repressed by TGF-P, and those of which the BMP-induced expression levels are inhibited by TGF-P might be important regulators of BMP- induced osteoblast differentiation. A small set of genes exhibiting these expression profiles was identified, including several genes involved in the Notch signal transduction pathway such as Heyl and Lfng. This result suggests that the Notch signal transduction pathway is a major regulator of early phase osteoblast differentiation. In addition, potential focal points of interaction of the BMP and Notch signal transduction pathways were identified. Osteoblast differentiation is not only regulated by the BMP/Smad signal transduction pathway, but also by other signal transduction pathways such as Erk1/2 MAP-kinase and Notch pathways. In Chapter 5, a model for the time-dependent influence of the different pathways on osteoblast differentiation is presented. This model is based on the results of inhibition of the Erk1/2 pathway and transfection of C2C12 cells with various Smads, Heyl and Lfng on expression profiles of genes that are representative for the different phases of osteoblast differentiation. The implications of the results of our research for the identification of drug targets for bone diseases are discussed in Chapter 6.

References

1. M. H. Kaufman, and J. B. L. Bard (1999), The anatomical basis of mouse development, 1 ed., Academic Press, San Diego, California. 2. R. Baron (2003), General Principles of Bone Biology, in Primer on the Metabolic Bone Diseases and Disorders of Mineral Metabolism, Favus, M. J., (Ed.), American Society for Bone and Mineral Research, Washington, DC, pp. 1-8. 3. D. J. Prockop (1997), Marrow stromal cells as stem cells for nonhematopoietic tissues, Science 276: 71-74. 4. J. B. Lian, G. S. Stein, and J. E. Aubin (2003), Bone Formation: Maturation and Functional Activities of Osteoblast Lineage Cells, in Primer on the Metabolic Bone Diseases and Disorders of Mineral Metabolism, Favus, M. J., (Ed.), American Society for Bone and Mineral Research, Washington, DC, pp. 13-28. 5. D. Metcalf (2003), The unsolved enigmas of leukemia inhibitory factor, Stem Cells 21: 5-14. 6. M. Bonyadi, S. D. Waldman, D. Liu, J. E. Aubin, M. D. Grynpas, and W. L. Stanford (2003), Mesenchymal progenitor self-renewal deficiency leads to age-dependent osteoporosis in Sca-1/Ly-6A null mice, Proc Natl Acad Sci USA 100: 5840-5845. 7. J. Cornish, M. T. Gillespie, K. E. Callon, N. J. Horwood, J. M. Moseley, and I. R. Reid (2003), Interleukin-18 is a novel mitogen of osteogenic and chondrogenic cells, Endocrinology 144: 1194-1201. 8. M. S. Lee, G. N. Lowe, D. D. Strong, J. E. Wergedal, and C. A. Glackin (1999), TWIST, a basic helix-loop-helix transcription factor, can regulate the human osteogenic lineage, J Cell Biochem 75: 566-577. 9. T. Ogata, and M. Noda (1991), Expression of Id, a negative regulator of helix-loop-helix DNA binding proteins, is down-regulated at confluence and enhanced by dexamethasone in a mouse osteoblastic cell line, MC3T3E1, Biochem Biophys Res Commun 180: 1194-1199. 10. J. Asp, M. Thornemo, S. Inerot, and A. Lindahl (1998), The helix-loop-helix transcription factors Id1 and Id3 have a functional role in control of cell division in human normal and neoplastic chondrocytes, FEBS Lett 438: 85-90.

- 40 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

11. C. Tribioli, M. Frasch, and T. Lufkin (1997), B apxl: an evolutionary conserved homologue of the Drosophila bagpipe homeobox gene is expressed in splanchnic mesoderm and the embryonic skeleton, Mech Dev 65: 145-162. 12. L. A. Lettice, L. A. Purdie, G. J. Carlson, F. Kilanowski, J. Dorin, and R. E. Hill (1999), The mouse bagpipe gene controls development of axial skeleton, skull, and spleen, Proc Natl Acad Sci USA 96: 9695-9700. 13. C. Tribioli, and T. Lufkin (1999), The murine Bapx1 homeobox gene plays a critical role in embryonic development of the axial skeleton and spleen, Development 126: 5699-5711. 14. I. Satokata, and R. Maas (1994), Msx1 deficient mice exhibit cleft palate and abnormalities of craniofacial and tooth development, Nat Genet 6: 348-356. 15. I. Satokata, L. Ma, H. Ohshima, M. Bei, I. Woo, K. Nishizawa, T. Maeda, Y. Takano, M. Uchiyama, S. Heaney, H. Peters, Z. Tang, R. Maxson, and R. Maas (2000), Msx2 deficiency in mice causes pleiotropic defects in bone growth and ectodermal organ formation, Nat Genet 24: 391-395. 16. S. L. Cheng, J. S. Shao, N. Charlton-Kachigian, A. P. Loewy, and D. A. Towler (2003), MSX2 promotes osteogenesis and suppresses adipogenic differentiation of multipotent mesenchymal progenitors, J Biol Chem 278: 45969-45977. 17. A. Mackenzie, G. L. Leeming, A. K. Jowett, M. W. Ferguson, and P. T. Sharpe (1991), The homeobox gene Hox 7.1 has specific regional and temporal expression patterns during early murine craniofacial embryogenesis, especially tooth development in vivo and in vitro, Development 111: 269-285. 18. M. Mina, J. Gluhak, W. B. Upholt, E. J. Kollar, and B. Rogers (1995), Experimental analysis of Msx-1 and Msx-2 gene expression during chick mandibular morphogenesis, Dev Dyn 202: 195-214. 19. S. Orestes-Cardoso, J. R. Nefussi, F. Lezot, M. Oboeuf, M. Pereira, M. Mesbah, B. Robert, and A. Berdal (2002), Msx1 is a regulator of bone formation during development and postnatal growth: in vivo investigations in a transgenic mouse model, Connect Tissue Res 43: 153-160. 20. B. Kanzler, S. J. Kuschert, Y. H. Liu, and M. Mallo (1998), Hoxa-2 restricts the chondrogenic domain and inhibits bone formation during development of the branchial area, Development 125: 2587-2597. 21. A. Simeone, D. Acampora, M. Pannese, M. D'Esposito, A. Stornaiuolo, M. Gulisano, A. Mallamaci, K. Kastury, T. Druck, K. Huebner, and et al. (1994), Cloning and characterization of two members of the vertebrate Dlx gene family, Proc Natl Acad Sci USA 91: 2250-2254. 22. D. Ferrari, and R. A. Kosher (2002), Dlx5 is a positive regulator of chondrocyte differentiation during endochondral ossification, Dev Biol 252: 257-270. 23. K. Miyama, G. Yamada, T. S. Yamamoto, C. Takagi, K. Miyado, M. Sakai, N. Ueno, and H. Shibuya (1999), A BMP- inducible gene, , regulates osteoblast differentiation and mesoderm induction, Dev Biol 208: 123-133. 24. T. Tadic, M. Dodig, I. Erceg, I. Marijanovic, M. Mina, Z. Kalajzic, D. Velonis, M. S. Kronenberg, R. A. Kosher, D. Ferrari, and A. C. Lichtler (2002), Overexpression of Dlx5 in chicken calvarial cells accelerates osteoblastic differentiation, J Bone Miner Res 17: 1008-1014. 25. M. D. Benson, J. L. Bargeon, G. Xiao, P. E. Thomas, A. Kim, Y. Cui, and R. T. Franceschi (2000), Identification of a homeodomain binding element in the bone sialoprotein gene promoter that is required for its osteoblast-selective expression, J Biol Chem 275: 13907-13917. 26. M. H. Lee, Y. J. Kim, H. J. Kim, H. D. Park, A. R. Kang, H. M. Kyung, J. H. Sung, J. M. Wozney, and H. M. Ryoo (2003), BMP-2-induced Runx2 expression is mediated by Dlx5, and TGF-beta 1 opposes the BMP-2-induced osteoblast differentiation by suppression of Dlx5 expression, J Biol Chem 278: 34387-34394. 27. M. H. Lee, T. G. Kwon, H. S. Park, J. M. Wozney, and H. M. Ryoo (2003), BMP-2-induced Osterix expression is mediated by Dlx5 but is independent of Runx2, Biochem Biophys Res Commun 309: 689-694. 28. P. Ducy, R. Zhang, V. Geoffroy, A. L. Ridall, and G. Karsenty (1997), Osf2/Cbfa1: a transcriptional activator of osteoblast differentiation, Cell 89: 747-754. 29. C. Banerjee, L. R. McCabe, J. Y. Choi, S. W. Hiebert, J. L. Stein, G. S. Stein, and J. B. Lian (1997), Runt homology domain proteins in osteoblast differentiation: AML3/CBFA1 is a major component of a bone-specific complex, J Cell Biochem 66: 1-8. 30. H. Harada, S. Tagashira, M. Fujiwara, S. Ogawa, T. Katsumata, A. Yamaguchi, T. Komori, and M. Nakatsuka (1999), Cbfa1 isoforms exert functional differences in osteoblast differentiation, J Biol Chem 274: 6972-6978. 31. C. Banerjee, A. Javed, J. Y. Choi, J. Green, V. Rosen, A. J. van Wijnen, J. L. Stein, J. B. Lian, and G. S. Stein (2001), Differential regulation of the two principal Runx2/Cbfa1 n-terminal isoforms in response to bone morphogenetic protein-2 during development of the osteoblast phenotype, Endocrinology 142: 4026-4039. 32. H. Drissi, Q. Luc, R. Shakoori, S. Chuva De Sousa Lopes, J. Y. Choi, A. Terry, M. Hu, S. Jones, J. C. Neil, J. B. Lian, J. L. Stein, A. J. Van Wijnen, and G. S. Stein (2000), Transcriptional autoregulation of the bone related CBFA1/RUNX2 gene, J Cell Physiol 184: 341-350. 33. G. Xiao, D. Jiang, P. Thomas, M. D. Benson, K. Guan, G. Karsenty, and R. T. Franceschi (2000), MAPK pathways activate and phosphorylate the osteoblast-specific transcription factor, Cbfa1, J Biol Chem 275: 4453-4459. 34. R. A. Kahler, and J. J. Westendorf (2003), Lymphoid enhancer factor-1 and beta-catenin inhibit Runx2-dependent transcriptional activation of the osteocalcin promoter, J Biol Chem 278: 11937-11944. 35. K. W. McLarren, R. Lo, D. Grbavec, K. Thirunavukkarasu, G. Karsenty, and S. Stifani (2000), The mammalian basic helix loop helix protein HES-1 binds to and modulates the transactivating function of the runt-related factor Cbfa1, J Biol Chem 275: 530-538. 36. T. Komori, H. Yagi, S. Nomura, A. Yamaguchi, K. Sasaki, K. Deguchi, Y. Shimizu, R. T. Bronson, Y. H. Gao, M. Inada, M. Sato, R. Okamoto, Y. Kitamura, S. Yoshiki, and T. Kishimoto (1997), Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts, Cell 89: 755-764. 37. B. Kern, J. Shen, M. Starbuck, and G. Karsenty (2001), Cbfa1 contributes to the osteoblast-specific expression of type I collagen genes, J Biol Chem 276: 7101-7107.

- 41 - Chapter 1

38. A. Javed, G. L. Barnes, B. O. Jasanya, J. L. Stein, L. Gerstenfeld, J. B. Lian, and G. S. Stein (2001), runt homology domain transcription factors (Runx, Cbfa, and AML) mediate repression of the bone sialoprotein promoter: evidence for promoter context-dependent activity of Cbfa proteins, M ol Cell Biol 21: 2891-2905. 39. Z. S. Xiao, A. B. Hjelmeland, and L. D. Quarles (2004), Selective deficiency of the "bone-related" Runx2-II unexpectedly preserves osteoblast-mediated skeletogenesis, J Biol Chem 279: 20307-20313. 40. K. Nakashima, X. Zhou, G. Kunkel, Z. Zhang, J. M. Deng, R. R. Behringer, and B. de Crombrugghe (2002), The novel zinc finger-containing transcription factor osterix is required for osteoblast differentiation and bone formation, Cell 108: 17-29. 41. M. A. Milona, J. E. Gough, and A. J. Edgar (2003), Expression of alternatively spliced isoforms of human Sp7 in osteoblast-like cells, BMC Genomics 4: 43. 42. G. He, T. Dahl, A. Veis, and A. George (2003), Nucleation of apatite crystals in vitro by self-assembled dentin matrix protein 1, Nat Mater 2: 552-558. 43. M. Takagi, T. Yamada, N. Kamiya, T. Kumagai, and A. Yamaguchi (1999), Effects of bone morphogenetic protein-2 and transforming growth factor-beta1 on gene expression of decorin and biglycan by cultured osteoblastic cells, Histochem J 31: 403-409. 44. Y. Mochida, W. R. Duarte, H. Tanzawa, E. P. Paschalis, and M. Yamauchi (2003), Decorin modulates matrix mineralization in vitro, Biochem Biophys Res Commun 305: 6-9. 45. A. Raouf, B. Ganss, C. McMahon, C. Vary, P. J. Roughley, and A. Seth (2002), Lumican is a major proteoglycan component of the bone matrix, Matrix Biol 21: 361-367. 46. C. H. Damsky (1999), Extracellular matrix-integrin interactions in osteoblast function and tissue remodeling, Bone 25: 95-96. 47. T. Katagiri, and N. Takahashi (2002), Regulatory mechanisms of osteoblast and osteoclast differentiation, Oral Dis 8: 147-159. 48. F. Furlan, F. Lecanda, J. Screen, and R. Civitelli (2001), Proliferation, differentiation and apoptosis in connexin43-null osteoblasts, Cell Commun Adhes 8: 367-371. 49. L. I. Plotkin, S. C. Manolagas, and T. Bellido (2002), Transduction of cell survival signals by connexin-43 hemichannels, J Biol Chem 277: 8648-8657. 50. F. Lecanda, P. M. Warlow, S. Sheikh, F. Furlan, T. H. Steinberg, and R. Civitelli (2000), Connexin43 deficiency causes delayed ossification, craniofacial abnormalities, and osteoblast dysfunction, J Cell Biol 151: 931-944. 51. J. Q. Fen, J. Zhang, S. L. Dallas, Y. Lu, S. Chen, X. Tan, M. Owen, S. E. Harris, and M. MacDougall (2002), Dentin matrix protein 1, a target molecule for Cbfa1 in bone, is a unique bone marker gene, J Bone Miner Res 17: 1822-1831. 52. L. D. Quarles (2003), FGF23, PHEX, and MEPE regulation of phosphate homeostasis and skeletal mineralization, Am J Physiol Endocrinol Metab 285: E1-9. 53. L. C. Gowen, D. N. Petersen, A. L. Mansolf, H. Qi, J. L. Stock, G. T. Tkalcevic, H. A. Simmons, D. T. Crawford, K. L. Chidsey-Frink, H. Z. Ke, J. D. McNeish, and T. A. Brown (2003), Targeted disruption of the osteoblast/osteocyte factor 45 gene (OF45) results in increased bone formation and bone mass, J Biol Chem 278: 1998-2007. 54. D. J. Mason, R. A. Hillam, and T. M. Skerry (1996), Constitutive in vivo mRNA expression by osteocytes of beta­ actin, osteocalcin, connexin-43, IGF-I, c-fos and c-jun, but not TNF-alpha nor tartrate-resistant acid phosphatase, J Bone Miner Res 11: 350-357. 55. M. Hadjiargyrou, E. P. Rightmire, T. Ando, and F. T. Lombardo (2001), The E11 osteoblastic lineage marker is differentially expressed during fracture healing, Bone 29: 149-154. 56. K. Tanaka-Kamioka, H. Kamioka, H. Ris, and S. S. Lim (1998), Osteocyte shape is dependent on actin filaments and osteocyte processes are unique actin-rich projections, J Bone Miner Res 13: 1555-1568. 57. D. G. Winkler, M. K. Sutherland, J. C. Geoghegan, C. Yu, T. Hayes, J. E. Skonier, D. Shpektor, M. Jonas, B. R. Kovacevich, K. Staehling-Hampton, M. Appleby, M. E. Brunkow, and J. A. Latham (2003), Osteocyte control of bone formation via sclerostin, a novel BMP antagonist, EMBO J 22: 6267-6276. 58. R. L. Van Bezooijen, B. A. Roelen, A. Visser, L. Van Der Wee-Pals, E. De Wilt, M. Karperien, H. Hamersma, S. E. Papapoulos, P. Ten Dijke, and C. W. Lowik (2004), Sclerostin Is an Osteocyte-expressed Negative Regulator of Bone Formation, But Not a Classical BMP Antagonist, J Exp Med 199: 805-814. 59. F. Liu, L. Malaval, and J. E. Aubin (1997), The mature osteoblast phenotype is characterized by extensive plasticity, Exp Cell Res 232: 97-105. 60. G. A. Candeliere, F. Liu, and J. E. Aubin (2001), Individual osteoblasts in the developing calvaria express different gene repertoires, Bone 28: 351-361. 61. M. Amling, S. Herden, M. Posl, M. Hahn, H. Ritzel, and G. Delling (1996), Heterogeneity of the skeleton: comparison of the trabecular microarchitecture of the spine, the iliac crest, the femur, and the calcaneus, J Bone Miner Res 11: 36­ 45. 62. R. J. Byers, J. Denton, J. A. Hoyland, and A. J. Freemont (1999), Differential patterns of altered bone formation in different bone compartments in established osteoporosis, J Clin Pathol 52: 23-28. 63. D. Miao, B. He, A. C. Karaplis, and D. Goltzman (2002), Parathyroid hormone is essential for normal fetal bone formation, J Clin Invest 109: 1173-1182. 64. V. Krishnan, T. L. Moore, Y. L. Ma, L. M. Helvering, C. A. Frolik, K. M. Valasek, P. Ducy, and A. G. Geiser (2003), Parathyroid hormone bone anabolic action requires Cbfa1/Runx2-dependent signaling, Mol Endocrinol 17: 423-435. 65. J. T. Swarthout, R. C. D'Alonzo, N. Selvamurugan, and N. C. Partridge (2002), Parathyroid hormone-dependent signaling pathways regulating genes in bone cells, Gene 282: 1-17.

- 42 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

66. A. C. Karaplis, A. Luz, J. Glowacki, R. T. Bronson, V. L. Tybulewicz, H. M. Kronenberg, and R. C. Mulligan (1994), Lethal skeletal dysplasia from targeted disruption of the parathyroid hormone-related peptide gene, Genes Dev 8: 277­ 289. 67. U. Chung, and H. M. Kronenberg (2000), Role of Parathyroid Hormone-related Protein and Indian Hedgehog in Skeletal Development, in Skeletal Growth Factors, in Canalis, E., (Ed.), Lippincott Williams & Wilkins, Philadelphia, PA, pp. 355-364. 68. B. St-Jacques, M. Hammerschmidt, and A. P. McMahon (1999), Indian hedgehog signaling regulates proliferation and differentiation of chondrocytes and is essential for bone formation, Genes Dev 13: 2072-2086. 69. S. J. Karp, E. Schipani, B. St-Jacques, J. Hunzelman, H. Kronenberg, and A. P. McMahon (2000), Indian hedgehog coordinates endochondral bone growth and morphogenesis via parathyroid hormone related-protein-dependent and - independent pathways, Development 127: 543-548. 70. M. F. Holick (2003), Vitamin D: Photobiology, Metabolism, Mechanism of Action and Clinical Applications, in Primer on the Metabolic Bone Diseases and Disorders of Mineral Metabolism, in Favus, M. J., (Ed.), American Society for Bone and Mineral Research, Washington, DC, pp. 129-137. 71. T. Yoshizawa, Y. Handa, Y. Uematsu, S. Takeda, K. Sekine, Y. Yoshihara, T. Kawakami, K. Arioka, H. Sato, Y. Uchiyama, S. Masushige, A. Fukamizu, T. Matsumoto, and S. Kato (1997), Mice lacking the vitamin D receptor exhibit impaired bone formation, uterine hypoplasia and growth retardation after weaning, Nat Genet 16: 391-396. 72. J. P. van Leeuwen, M. van Driel, G. J. van den Bemd, and H. A. Pols (2001), Vitamin D control of osteoblast function and bone extracellular matrix mineralization, Crit Rev Eukaryot Gene Expr 11: 199-226. 73. A. D. Weston, L. M. Hoffman, and T. M. Underhill (2003), Revisiting the role of retinoid signaling in skeletal development, Birth Defects Res Part C Embryo Today 69: 156-173. 74. D. M. Kochhar (1967), Teratogenic activity of retinoic acid, Acta Pathol Microbiol Scand 70: 398-404. 75. T. E. Kwasigroch, and D. M. Kochhar (1980), Production of congenital limb defects with retinoic acid: phenomenological evidence of progressive differentiation during limb morphogenesis, Anat Embryol (Berl) 161: 105­ 113. 76. P. Kastner, M. Mark, N. Ghyselinck, W. Krezel, V. Dupe, J. M. Grondona, and P. Chambon (1997), Genetic evidence that the retinoid signal is transduced by heterodimeric RXR/RAR functional units during mouse development, Development 124: 313-326. 77. M. Mark, N. B. Ghyselinck, O. Wendling, V. Dupe, B. Mascrez, P. Kastner, and P. Chambon (1999), A genetic dissection of the retinoid signalling pathway in the mouse, ProcNutr Soc 58: 609-613. 78. M. Leid, P. Kastner, and P. Chambon (1992), Multiplicity generates diversity in the retinoic acid signalling pathways, Trends Biochem Sci 17: 427-433. 79. T. M. Underhill, and A. D. Weston (1998), Retinoids and their receptors in skeletal development, Microsc Res Tech 43: 137-155. 80. P. F. Choong, T. J. Martin, and K. W. Ng (1993), Effects of ascorbic acid, calcitriol, and retinoic acid on the differentiation of preosteoblasts, J Orthop Res 11: 638-647. 81. V. M. Paralkar, W. A. Grasser, A. L. Mansolf, A. P. Baumann, T. A. Owen, S. L. Smock, S. Martinovic, F. Borovecki, S. Vukicevic, H. Z. Ke, and D. D. Thompson (2002), Regulation of BMP-7 expression by retinoic acid and prostaglandin E(2), J Cell Physiol 190: 207-217. 82. J. Skillington, L. Choy, and R. Derynck (2002), Bone morphogenetic protein and retinoic acid signaling cooperate to induce osteoblast differentiation of preadipocytes, J Cell Biol 159: 135-146. 83. A. Raouf, and A. Seth (2000), Ets transcription factors and targets in osteogenesis, Oncogene 19: 6455-6463. 84. D. G. Monroe, and T. C. Spelsberg (2003), Gonadal steroids and receptors, in Primer on the Metabolic Bone Diseases and Disorders of Mineral Metabolism, in Favus, M. J., (Ed.), American Society for Bone and Mineral Research, Washington, DC., pp. 32-38. 85. S. Khosla, and J. P. Bilezikian (2003), The role of estrogens in men and androgens in women, Endocrinol Metab Clin North Am 32: 195-218. 86. D. Vanderschueren, and R. Bouillon (1995), Androgens and bone, Calcif Tissue Int 56: 341-346. 87. J. C. Prior (1990), Progesterone as a bone-trophic hormone, Endocr Rev 11: 386-398. 88. E. Canalis (2003), Mechanisms of glucocorticoid-induced osteoporosis, Curr Opin Rheumatol 15: 454-457. 89. A. M. Delany, and E. Canalis (1995), Transcriptional repression of insulin-like growth factor I by glucocorticoids in rat bone cells, Endocrinology 136: 4776-4781. 90. J. O. Gordeladze, and J. E. Reseland (2003), A unified model for the action of leptin on bone turnover, J Cell Biochem 88: 706-712. 91. J. I. Jones, and D. R. Clemmons (1995), Insulin-like growth factors and their binding proteins: biological actions, Endocr Rev 16: 3-34. 92. V. Hwa, Y. Oh, and R. G. Rosenfeld (1999), The insulin-like growth factor-binding protein (IGFBP) superfamily, Endocr Rev 20: 761-787. 93. J. M. Hock, M. Centrella, and E. Canalis (1988), Insulin-like growth factor I has independent effects on bone matrix formation and cell replication, Endocrinology 122: 254-260. 94. P. A. Hill, A. Tumber, and M. C. Meikle (1997), Multiple extracellular signals promote osteoblast survival and apoptosis, Endocrinology 138: 3849-3858. 95. E. Canalis, S. Rydziel, A. M. Delany, S. Varghese, and J. J. Jeffrey (1995), Insulin-like growth factors inhibit interstitial collagenase synthesis in bone cell cultures, Endocrinology 136: 1348-1354. 96. L. Powell-Braxton, P. Hollingshead, C. Warburton, M. Dowd, S. Pitts-Meek, D. Dalton, N. Gillett, and T. A. Stewart (1993), IGF-I is required for normal embryonic growth in mice, Genes Dev 7: 2609-2617.

- 43 - Chapter 1

97. D. D. Bikle, T. Sakata, C. Leary, H. Elalieh, D. Ginzinger, C. J. Rosen, W. Beamer, S. Majumdar, and B. P. Halloran (2002), Insulin-like growth factor I is required for the anabolic actions of parathyroid hormone on mouse bone, J Bone Miner Res 17: 1570-1578. 98. D. M. Ornitz, and P. J. Marie (2002), FGF signaling pathways in endochondral and intramembranous bone development and human genetic disease, Genes Dev 16: 1446-1465. 99. L. Chen, C. Li, W. Qiao, X. Xu, and C. Deng (2001), A Ser(365)-->Cys mutation of fibroblast growth factor receptor 3 in mouse downregulates Ihh/PTHrP signals and causes severe achondroplasia, Hum Mol Genet 10: 457-465. 100. H. C. Lowe, L. A. Rafty, T. Collins, and L. M. Khachigian (2000), Biology of Platelet-Derived Growth Factor, in Skeletal Growth Factors, in Canalis, E., (Ed.), Lippincott Williams & Wilkins, Philadelphia, PA, pp. 129-151. 101. C. Betsholtz, L. Karlsson, and P. Lindahl (2001), Developmental roles of platelet-derived growth factors, Bioessays 23: 494-507. 102. E. Canalis, and D. M. Ornitz (2000), Platelet-Derived Growth Factor: Skeletal Actions and Regulation, in Skeletal Growth Factors, in Canalis, E., (Ed.), Lippincott Williams & Wilkins, Philadelphia, PA, pp. 153-165. 103. J. Massague (1998), TGF-beta signal transduction, Annu Rev Biochem 67: 753-791. 104. J. Massague (2000), How cells read TGF-beta signals, Nat Rev Mol Cell Biol 1: 169-178. 105. J. Massague, S. W. Blain, and R. S. Lo (2000), TGFbeta signaling in growth control, cancer, and heritable disorders, Cell 103: 295-309. 106. R. Serra, and C. Chang (2003), TGF-beta signaling in human skeletal and patterning disorders, Birth Defects Res Part C Embryo Today 69: 333-351. 107. B. A. Roelen, and P. Dijke (2003), Controlling mesenchymal stem cell differentiation by TGFBeta family members, J Orthop Sci 8: 740-748. 108. B. L. Hogan (1996), Bone morphogenetic proteins: multifunctional regulators of vertebrate development, Genes Dev 10: 1580-1594. 109. Y. Cui, F. Jean, G. Thomas, and J. L. Christian (1998), BMP-4 is proteolytically activated by furin and/or PC6 during vertebrate embryonic development, EMBO J 17: 4735-4743. 110. J. M. Wozney, V. Rosen, A. J. Celeste, L. M. Mitsock, M. J. Whitters, R. W. Kriz, R. M. Hewick, and E. A. Wang (1988), Novel regulators of bone formation: molecular clones and activities, Science 242: 1528-1534. 111. T. Katagiri, S. Akiyama, M. Namiki, M. Komaki, A. Yamaguchi, V. Rosen, J. M. Wozney, A. Fujisawa-Sehara, and T. Suda (1997), Bone morphogenetic protein-2 inhibits terminal differentiation of myogenic cells by suppressing the transcriptional activity of MyoD and myogenin, Exp Cell Res 230: 342-351. 112. P. Leboy, G. Grasso-Knight, M. D'Angelo, S. W. Volk, J. V. Lian, H. Drissi, G. S. Stein, and S. L. Adams (2001), Smad-Runx interactions during chondrocyte maturation, J Bone Joint Surg Am 83-A Suppl 1: S15-22. 113. A. Yamaguchi, T. Ishizuya, N. Kintou, Y. Wada, T. Katagiri, J. M. Wozney, V. Rosen, and S. Yoshiki (1996), Effects of BMP-2, BMP-4, and BMP-6 on osteoblastic differentiation of bone marrow-derived stromal cell lines, ST2 and MC3T3-G2/PA6, Biochem BiophysRes Commun 220: 366-371. 114. T. Iimura, S. Oida, K. Takeda, Y. Maruoka, and S. Sasaki (1994), Changes in homeobox-containing gene expression during ectopic bone formation induced by bone morphogenetic protein, Biochem Biophys Res Commun 201: 980-987. 115. K. Song, Y. Wang, and D. Sassoon (1992), Expression of Hox-7.1 in myoblasts inhibits terminal differentiation and induces cell transformation, Nature 360: 477-481. 116. Y. Jen, H. Weintraub, and R. Benezra (1992), Overexpression of Id protein inhibits the muscle differentiation program: in vivo association of Id with E2A proteins, Genes Dev 6: 1466-1479. 117. A. Hollnagel, V. Oehlmann, J. Heymer, U. Ruther, and A. Nordheim (1999), Id genes are direct targets of bone morphogenetic protein induction in embryonic stem cells, J Biol Chem 274: 19838-19845. 118. T. Ogata, J. M. Wozney, R. Benezra, and M. Noda (1993), Bone morphogenetic protein 2 transiently enhances expression of a gene, Id (inhibitor of differentiation), encoding a helix-loop-helix molecule in osteoblast-like cells, Proc Natl Acad Sci USA 90: 9219-9222. 119. S. C. Xu, M. A. Harris, J. L. Rubenstein, G. R. Mundy, and S. E. Harris (2001), Bone morphogenetic protein-2 (BMP- 2) signaling to the Col2alpha1 gene in chondroblasts requires the homeobox gene Dlx-2, DNA Cell Biol 20: 359-365. 120. L. C. Murtaugh, L. Zeng, J. H. Chyung, and A. B. Lassar (2001), The chick transcriptional repressor Nkx3.2 acts downstream of Shh to promote BMP-dependent axial chondrogenesis, Dev Cell 1: 411-422. 121. Y. Watanabe, and N. M. Le Douarin (1996), A role for BMP-4 in the development of subcutaneous cartilage, Mech Dev 57: 69-78. 122. B. K. Zehentner, C. Dony, and H. Burtscher (1999), The transcription factor Sox9 is involved in BMP-2 signaling, J Bone Miner Res 14: 1734-1741. 123. C. D. Grimsrud, P. R. Romano, M. D'Souza, J. E. Puzas, E. M. Schwarz, P. R. Reynolds, R. N. Roiser, and R. J. O'Keefe (2001), BMP signaling stimulates chondrocyte maturation and the expression of Indian hedgehog, J Orthop Res 19: 18-25. 124. L. Zeng, H. Kempf, L. C. Murtaugh, M. E. Sato, and A. B. Lassar (2002), Shh establishes an Nkx3.2/Sox9 autoregulatory loop that is maintained by BMP signals to induce somitic chondrogenesis, Genes Dev 16: 1990-2005. 125. Y. Takazawa, K. Tsuji, A. Nifuji, H. Kurosawa, Y. Ito, and M. Noda (2000), An osteogenesis-related transcription factor, core-binding factor A1, is constitutively expressed in the chondrocytic cell line TC6, and its expression is upregulated by bone morphogenetic protein-2, J Endocrinol 165: 579-586. 126. K. Yagi, K. Tsuji, A. Nifuji, K. Shinomiya, K. Nakashima, B. DeCrombrugghe, and M. Noda (2003), Bone morphogenetic protein-2 enhances osterix gene expression in chondrocytes, J Cell Biochem 88: 1077-1083. 127. J. Wilson, and A. S. Tucker (2004), Fgf and Bmp signals repress the expression of Bapx1 in the mandibular mesenchyme and control the position of the developing jaw joint, Dev Biol 266: 138-150.

- 44 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

128. B. Sevetson, S. Taylor, and Y. Pan (2004), Cbfa1/RUNX2 directs specific expression of the sclerosteosis gene (SOST), JB io l Chem 279: 13849-13858. 129. K. S. Lee, H. J. Kim, Q. L. Li, X. Z. Chi, C. Ueta, T. Komori, J. M. Wozney, E. G. Kim, J. Y. Choi, H. M. Ryoo, and S. C. Bae (2000), Runx2 is a common target of transforming growth factor betal and bone morphogenetic protein 2, and cooperation between and smad5 induces osteoblast-specific gene expression in the pluripotent mesenchymal precursor cell line C2C12, M ol Cell Biol 20: 8783-8792. 130. S. K. Zaidi, A. J. Sullivan, A. J. van Wijnen, J. L. Stein, G. S. Stein, and J. B. Lian (2002), Integration of Runx and Smad regulatory signals at transcriptionally active subnuclear sites, Proc Natl Acad Sci USA 99: 8048-8053. 131. Y. W. Zhang, N. Yasui, K. Ito, G. Huang, M. Fujii, J. Hanai, H. Nogami, T. Ochi, K. Miyazono, and Y. Ito (2000), A RUNX2/PEBP2alpha A/CBFA1 mutation displaying impaired transactivation and Smad interaction in cleidocranial dysplasia, Proc Natl Acad Sci USA 97: 10549-10554. 132. G. Marazzi, Y. Wang, and D. Sassoon (1997), Msx2 is a transcriptional regulator in the BMP4-mediated programmed cell death pathway, Dev Biol 186: 127-138. 133. E. Hay, J. Lemonnier, O. Fromigue, and P. J. Marie (2001), Bone morphogenetic protein-2 promotes osteoblast apoptosis through a Smad-independent, protein kinase C-dependent signaling pathway, J Biol Chem 276: 29028­ 29036. 134. E. Hay, J. Lemonnier, O. Fromigue, H. Guenou, and P. J. Marie (2004), Bone morphogenetic protein receptor IB signaling mediates apoptosis independently of differentiation in osteoblastic cells, J Biol Chem 279: 1650-1658. 135. M. Suzawa, Y. Takeuchi, S. Fukumoto, S. Kato, N. Ueno, K. Miyazono, T. Matsumoto, and T. Fujita (1999), Extracellular matrix-associated bone morphogenetic proteins are essential for differentiation of murine osteoblastic cells in vitro, Endocrinology 140: 2125-2133. 136. D. J. Rickard, L. C. Hofbauer, S. K. Bonde, F. Gori, T. C. Spelsberg, and B. L. Riggs (1998), Bone morphogenetic protein-6 production in human osteoblastic cell lines. Selective regulation by estrogen, J Clin Invest 101: 413-422. 137. D. Chen, M. A. Harris, G. Rossini, C. R. Dunstan, S. L. Dallas, J. Q. Feng, G. R. Mundy, and S. E. Harris (1997), Bone morphogenetic protein 2 (BMP-2) enhances BMP-3, BMP-4, and bone cell differentiation marker gene expression during the induction of mineralized bone matrix formation in cultures of fetal rat calvarial osteoblasts, Calcif TissueInt 60: 283-290. 138. R. C. Pereira, S. Rydziel, and E. Canalis (2000), Bone morphogenetic protein-4 regulates its own expression in cultured osteoblasts, J Cell Physiol 182: 239-246. 139. N. Ghosh-Choudhury, G. G. Choudhury, M. A. Harris, J. Wozney, G. R. Mundy, S. L. Abboud, and S. E. Harris (2001), Autoregulation of mouse BMP-2 gene transcription is directed by the proximal promoter element, Biochem Biophys Res Commun 286: 101-108. 140. L. M. Helvering, R. L. Sharp, X. Ou, and A. G. Geiser (2000), Regulation of the promoters for the human bone morphogenetic protein 2 and 4 genes, Gene 256: 123-138. 141. H. Enomoto, S. Shiojiri, K. Hoshi, T. Furuichi, R. Fukuyama, C. A. Yoshida, N. Kanatani, R. Nakamura, A. Mizuno, A. Zanma, K. Yano, H. Yasuda, K. Higashio, K. Takada, and T. Komori (2003), Induction of osteoclast differentiation by Runx2 through receptor activator of nuclear factor-kappa B ligand (RANKL) and osteoprotegerin regulation and partial rescue of osteoclastogenesis in Runx2-/- mice by RANKL transgene, J Biol Chem 278: 23971-23977. 142. C. A. O'Brien, B. Kern, I. Gubrij, G. Karsenty, and S. C. Manolagas (2002), Cbfa1 does not regulate RANKL gene activity in stromal/osteoblastic cells, Bone 30: 453-462. 143. K. Itoh, N. Udagawa, T. Katagiri, S. Iemura, N. Ueno, H. Yasuda, K. Higashio, J. M. Quinn, M. T. Gillespie, T. J. Martin, T. Suda, and N. Takahashi (2001), Bone morphogenetic protein 2 stimulates osteoclast differentiation and survival supported by receptor activator of nuclear factor-kappaB ligand, Endocrinology 142: 3656-3662. 144. M. Wan, X. Shi, X. Feng, and X. Cao (2001), Transcriptional mechanisms of bone morphogenetic protein-induced osteoprotegrin gene expression, J Biol Chem 276: 10119-10125. 145. S. Varghese, and E. Canalis (1997), Regulation of collagenase-3 by bone morphogenetic protein-2 in bone cell cultures, Endocrinology 138: 1035-1040. 146. L. Gilboa, A. Nohe, T. Geissendorfer, W. Sebald, Y. I. Henis, and P. Knaus (2000), Bone morphogenetic protein receptor complexes on the surface of live cells: a new oligomerization mode for serine/threonine kinase receptors, Mol Biol Cell 11: 1023-1035. 147. A. Nohe, S. Hassel, M. Ehrlich, F. Neubauer, W. Sebald, Y. I. Henis, and P. Knaus (2002), The mode of bone morphogenetic protein (BMP) receptor oligomerization determines different BMP-2 signaling pathways, J Biol Chem 277: 5330-5338. 148. K. Kurozumi, M. Nishita, K. Yamaguchi, T. Fujita, N. Ueno, and H. Shibuya (1998), BRAM1, a BMP receptor- associated molecule involved in BMP signalling, Genes Cells 3: 257-264. 149. K. Yamaguchi, S. Nagai, J. Ninomiya-Tsuji, M. Nishita, K. Tamai, K. Irie, N. Ueno, E. Nishida, H. Shibuya, and K. Matsumoto (1999), XIAP, a cellular member of the inhibitor of apoptosis protein family, links the receptors to TAB1- TAK1 in the BMP signaling pathway, EMBO J 18: 179-187. 150. C. H. Heldin, K. Miyazono, and P. ten Dijke (1997), TGF-beta signalling from cell membrane to nucleus through SMAD proteins, Nature 390: 465-471. 151. A. Zwijsen, K. Verschueren, and D. Huylebroeck (2003), New intracellular components of bone morphogenetic protein/Smad signaling cascades, FEBSLett 546: 133-139. 152. M. Kretzschmar, F. Liu, A. Hata, J. Doody, and J. Massague (1997), The TGF-beta family mediator Smad1 is phosphorylated directly and activated functionally by the BMP receptor kinase, Genes Dev 11: 984-995.

- 45 - Chapter 1

153. S. Abdollah, M. Macias-Silva, T. Tsukazaki, H. Hayashi, L. Attisano, and J. L. Wrana (1997), TbetaRI phosphorylation of Smad2 on Ser465 and Ser467 is required for Smad2-Smad4 complex formation and signaling, J Biol Chem 272: 27678-27685. 154. S. Souchelnytskyi, K. Tamaki, U. Engstrom, C. Wernstedt, P. ten Dijke, and C. H. Heldin (1997), Phosphorylation of Ser465 and Ser467 in the C terminus of Smad2 mediates interaction with Smad4 and is required for transforming growth factor-beta signaling, J Biol Chem 272: 28107-28115. 155. R. Derynck, and Y. E. Zhang (2003), Smad-dependent and Smad-independent pathways in TGF-beta family signalling, Nature 425: 577-584. 156. S. Dennler, S. Itoh, D. Vivien, P. ten Dijke, S. Huet, and J. M. Gauthier (1998), Direct binding of Smad3 and Smad4 to critical TGF beta-inducible elements in the promoter of human plasminogen activator inhibitor-type 1 gene, EMBO J 17: 3091-3100. 157. L. J. Jonk, S. Itoh, C. H. Heldin, P. ten Dijke, and W. Kruijer (1998), Identification and functional characterization of a Smad binding element (SBE) in the JunB promoter that acts as a transforming growth factor-beta, activin, and bone morphogenetic protein-inducible enhancer, J Biol Chem 273: 21145-21152. 158. K. A. Henningfeld, S. Rastegar, G. Adler, and W. Knochel (2000), Smad1 and Smad4 are components of the bone morphogenetic protein-4 (BMP-4)-induced transcription complex of the Xvent-2B promoter, J Biol Chem 275: 21827­ 21835. 159. G. T. Park, and M. I. Morasso (2002), Bone morphogenetic protein-2 (BMP-2) transactivates Dlx3 through Smad1 and Smad4: alternative mode for Dlx3 induction in mouse keratinocytes, Nucleic Acids Res 30: 515-522. 160. W. Ishida, T. Hamamoto, K. Kusanagi, K. Yagi, M. Kawabata, K. Takehara, T. K. Sampath, M. Kato, and K. Miyazono (2000), Smad6 is a Smad1/5-induced inhibitor. Characterization of bone morphogenetic protein- responsive element in the mouse Smad6 promoter, J Biol Chem 275: 6075-6079. 161. O. Korchynskyi, and P. ten Dijke (2002), Identification and functional characterization of distinct critically important bone morphogenetic protein-specific response elements in the Id1 promoter, J Biol Chem 277: 4883-4891. 162. T. Lopez-Rovira, E. Chalaux, J. Massague, J. L. Rosa, and F. Ventura (2002), Direct binding of Smad1 and Smad4 to two distinct motifs mediates bone morphogenetic protein-specific transcriptional activation of Id1 gene, J Biol Chem 277: 3176-3185. 163. Y. Shi, Y. F. Wang, L. Jayaraman, H. Yang, J. Massague, and N. P. Pavletich (1998), Crystal structure of a Smad MH1 domain bound to DNA: insights on DNA binding in TGF-beta signaling, Cell 94: 585-594. 164. L. Attisano, and J. L. Wrana (2000), Smads as transcriptional co-modulators, Curr Opin Cell Biol 12: 235-243. 165. R. Janknecht, N. J. Wells, and T. Hunter (1998), TGF-beta-stimulated cooperation of smad proteins with the coactivators CBP/p300, Genes Dev 12: 2114-2119. 166. K. L. Pearson, T. Hunter, and R. Janknecht (1999), Activation of Smad 1-mediated transcription by p300/CBP, Biochim Biophys Acta 1489: 354-364. 167. M. P. de Caestecker, T. Yahata, D. Wang, W. T. Parks, S. Huang, C. S. Hill, T. Shioda, A. B. Roberts, and R. J. Lechleider (2000), The Smad4 activation domain (SAD) is a proline-rich, p300-dependent transcriptional activation domain, J Biol Chem 275: 2115-2122. 168. M. A. Martinez-Balbas, A. J. Bannister, K. Martin, P. Haus-Seuffert, M. Meisterernst, and T. Kouzarides (1998), The acetyltransferase activity of CBP stimulates transcription, EMBO J 17: 2886-2893. 169. S. Itoh, J. Ericsson, J. Nishikawa, C. H. Heldin, and P. ten Dijke (2000), The transcriptional co-activator P/CAF potentiates TGF-beta/Smad signaling, Nucleic Acids Res 28: 4291-4298. 170. X. Shi, X. Yang, D. Chen, Z. Chang, and X. Cao (1999), Smad1 interacts with homeobox DNA-binding proteins in bone morphogenetic protein signaling, J Biol Chem 274: 13711-13717. 171. S. Bai, and X. Cao (2002), A nuclear antagonistic mechanism of inhibitory Smads in transforming growth factor-beta signaling, J Biol Chem 277: 4176-4182. 172. D. W. Kim, and A. B. Lassar (2003), Smad-dependent recruitment of a histone deacetylase/Sin3A complex modulates the bone morphogenetic protein-dependent transcriptional repressor activity of Nkx3.2, M ol Cell Biol 23: 8704-8717. 173. M. Gonzalez-Gaitan (2003), Signal dispersal and transduction through the endocytic pathway, Nat Rev Mol Cell Biol 4: 213-224. 174. T. Tsukazaki, T. A. Chiang, A. F. Davison, L. Attisano, and J. L. Wrana (1998), SARA, a FYVE domain protein that recruits Smad2 to the TGFbeta receptor, Cell 95: 779-791. 175. S. G. Penheiter, H. Mitchell, N. Garamszegi, M. Edens, J. J. Dore, Jr., and E. B. Leof (2002), Internalization- dependent and -independent requirements for transforming growth factor beta receptor signaling via the Smad pathway, M ol Cell Biol 22: 4750-4759. 176. F. Itoh, N. Divecha, L. Brocks, L. Oomen, H. Janssen, J. Calafat, S. Itoh, and P. Dijke (2002), The FYVE domain in Smad anchor for receptor activation (SARA) is sufficient for localization of SARA in early endosomes and regulates TGF-beta/Smad signalling, Genes Cells 7: 321-331. 177. S. Miura, T. Takeshita, H. Asao, Y. Kimura, K. Murata, Y. Sasaki, J. I. Hanai, H. Beppu, T. Tsukazaki, J. L. Wrana, K. Miyazono, and K. Sugamura (2000), Hgs (Hrs), a FYVE domain protein, is involved in Smad signaling through cooperation with SARA, M ol Cell Biol 20: 9346-9355. 178. W. Schwab, F. Galbiati, D. Volonte, U. Hempel, K. W. Wenzel, R. H. Funk, M. P. Lisanti, and M. Kasper (1999), Characterisation of caveolins from cartilage: expression of caveolin-1, -2 and -3 in chondrocytes and in alginate cell culture of the rat tibia, Histochem Cell Biol 112: 41-49. 179. R. A. Lofthouse, J. R. Davis, C. G. Frondoza, R. H. Jinnah, D. S. Hungerford, and J. M. Hare (2001), Identification of caveolae and detection of caveolin in normal human osteoblasts, J Bone Joint Surg Br 83: 124-129.

- 46 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

180. C. Rauch, A. C. Brunet, J. Deleule, and E. Farge (2002), C2C12 myoblast/osteoblast transdifferentiation steps enhanced by epigenetic inhibition of BMP2 endocytosis, Am J Physiol Cell Physiol 283: C235-243. 181. L. B. Zimmerman, J. M. De Jesus-Escobar, and R. M. Harland (1996), The Spemann organizer signal noggin binds and inactivates bone morphogenetic protein 4, Cell 86: 599-606. 182. J. Groppe, J. Greenwald, E. Wiater, J. Rodriguez-Leon, A. N. Economides, W. Kwiatkowski, M. Affolter, W. W. Vale, J. C. Belmonte, and S. Choe (2002), Structural basis of BMP signalling inhibition by the cystine knot protein Noggin, Nature 420: 636-642. 183. W. N. Pappano, I. C. Scott, T. G. Clark, R. L. Eddy, T. B. Shows, and D. S. Greenspan (1998), Coding sequence and expression patterns of mouse chordin and mapping of the cognate mouse chrd and human CHRD genes, Genomics 52: 236-239. 184. N. Nakayama, C. E. Han, S. Scully, R. Nishinakamura, C. He, L. Zeni, H. Yamane, D. Chang, D. Yu, T. Yokota, and D. Wen (2001), A novel chordin-like protein inhibitor for bone morphogenetic proteins expressed preferentially in mesenchymal cell lineages, Dev Biol 232: 372-387. 185. N. Nakayama, C. Y. Han, L. Cam, J. I. Lee, J. Pretorius, S. Fisher, R. Rosenfeld, S. Scully, R. Nishinakamura, D. Duryea, G. Van, B. Bolon, T. Yokota, and K. Zhang (2004), A novel chordin-like BMP inhibitor, CHL2, expressed preferentially in chondrocytes of developing cartilage and osteoarthritic joint cartilage, Development 131: 229-240. 186. H. Sakuta, R. Suzuki, H. Takahashi, A. Kato, T. Shintani, S. Iemura, T. S. Yamamoto, N. Ueno, and M. Noda (2001), Ventroptin: a BMP-4 antagonist expressed in a double-gradient pattern in the retina, Science 293: 111-115. 187. J. J. Pearce, G. Penny, and J. Rossant (1999), A mouse cerberus/Dan-related gene family, Dev Biol 209: 98-110. 188. D. R. Hsu, A. N. Economides, X. Wang, P. M. Eimon, and R. M. Harland (1998), The Xenopus dorsalizing factor Gremlin identifies a novel family of secreted proteins that antagonize BMP activities, M ol Cell 1: 673-683. 189. S. Iemura, T. S. Yamamoto, C. Takagi, H. Uchiyama, T. Natsume, S. Shimasaki, H. Sugino, and N. Ueno (1998), Direct binding of follistatin to a complex of bone-morphogenetic protein and its receptor inhibits ventral and epidermal cell fates in early Xenopus embryo, Proc Natl Acad Sci USA 95: 9337-9342. 190. M. Oelgeschlager, J. Larrain, D. Geissert, and E. M. De Robertis (2000), The evolutionarily conserved BMP-binding protein Twisted gastrulation promotes BMP signalling, Nature 405: 757-763. 191. W. C. Smith, and R. M. Harland (1992), Expression cloning of noggin, a new dorsalizing factor localized to the Spemann organizer in Xenopus embryos, Cell 70: 829-840. 192. P. Aspenberg, C. Jeppsson, and A. N. Economides (2001), The bone morphogenetic proteins antagonist Noggin inhibits membranous ossification, J Bone Miner Res 16: 497-500. 193. C. Chang, and A. Hemmati-Brivanlou (1999), Xenopus GDF6, a new antagonist of noggin and a partner of BMPs, Development 126: 3347-3357. 194. R. Merino, D. Macias, Y. Ganan, A. N. Economides, X. Wang, Q. Wu, N. Stahl, K. T. Sampath, P. Varona, and J. M. Hurle (1999), Expression and function of Gdf-5 during digit skeletogenesis in the embryonic chick leg bud, Dev Biol 206: 33-45. 195. S. Paine-Saunders, B. L. Viviano, A. N. Economides, and S. Saunders (2002), Heparan sulfate proteoglycans retain Noggin at the cell surface: a potential mechanism for shaping bone morphogenetic protein gradients, J Biol Chem 277: 2089-2096. 196. B. L. Viviano, S. Paine-Saunders, N. Gasiunas, J. Gallagher, and S. Saunders (2004), Domain-specific modification of heparan sulfate by Qsulf1 modulates the binding of the bone morphogenetic protein antagonist Noggin, J Biol Chem 279: 5604-5611. Epub 2003 Nov 5625. 197. B. K. Zehentner, A. Haussmann, and H. Burtscher (2002), The bone morphogenetic protein antagonist Noggin is regulated by Sox9 during endochondral differentiation, Dev Growth Differ 44: 1-9. 198. E. Gazzerro, V. Gangji, and E. Canalis (1998), Bone morphogenetic proteins induce the expression of noggin, which limits their activity in cultured rat osteoblasts, J Clin Invest 102: 2106-2114. 199. X. B. Wu, Y. Li, A. Schneider, W. Yu, G. Rajendren, J. Iqbal, M. Yamamoto, M. Alam, L. J. Brunet, H. C. Blair, M. Zaidi, and E. Abe (2003), Impaired osteoblastic differentiation, reduced bone formation, and severe osteoporosis in noggin-overexpressing mice, J Clin Invest 112: 924-934. 200. I. Kalajzic, Z. Kalajzic, M. M. Hurley, A. C. Lichtler, and D. W. Rowe (2003), Stage specific inhibition of osteoblast lineage differentiation by FGF2 and noggin, J Cell Biochem 88: 1168-1176. 201. R. D. Devlin, Z. Du, R. C. Pereira, R. B. Kimble, A. N. Economides, V. Jorgetti, and E. Canalis (2003), Skeletal overexpression of noggin results in osteopenia and reduced bone formation, Endocrinology 144: 1972-1978. 202. Y. Sasai, B. Lu, H. Steinbeisser, D. Geissert, L. K. Gont, and E. M. De Robertis (1994), Xenopus chordin: a novel dorsalizing factor activated by organizer-specific homeobox genes, Cell 79: 779-790. 203. J. Larrain, D. Bachiller, B. Lu, E. Agius, S. Piccolo, and E. M. De Robertis (2000), BMP-binding modules in chordin: a model for signalling regulation in the extracellular space, Development 127: 821-830. 204. S. Piccolo, Y. Sasai, B. Lu, and E. M. De Robertis (1996), Dorsoventral patterning in Xenopus: inhibition of ventral signals by direct binding of chordin to BMP-4, Cell 86: 589-598. 205. D. Bachiller, J. Klingensmith, C. Kemp, J. A. Belo, R. M. Anderson, S. R. May, J. A. McMahon, A. P. McMahon, R. M. Harland, J. Rossant, and E. M. De Robertis (2000), The organizer factors Chordin and Noggin are required for mouse forebrain development, Nature 403: 658-661. 206. K. Takahara, G. E. Lyons, and D. S. Greenspan (1994), Bone morphogenetic protein-1 and a mammalian tolloid homologue (mTld) are encoded by alternatively spliced transcripts which are differentially expressed in some tissues, J Biol Chem 269: 32572-32578.

- 47 - Chapter 1

207. S. Piccolo, E. Agius, B. Lu, S. Goodman, L. Dale, and E. M. De Robertis (1997), Cleavage of Chordin by Xolloid metalloprotease suggests a role for proteolytic processing in the regulation of Spemann organizer activity, Cell 91: 407-416. 208. I. C. Scott, I. L. Blitz, W. N. Pappano, Y. Imamura, T. G. Clark, B. M. Steiglitz, C. L. Thomas, S. A. Maas, K. Takahara, K. W. Cho, and D. S. Greenspan (1999), Mammalian BMP-1/Tolloid-related metalloproteinases, including novel family member mammalian Tolloid-like 2, have differential enzymatic activities and distributions of expression relevant to patterning and skeletogenesis, Dev Biol 213: 283-300. 209. E. Canalis, A. N. Economides, and E. Gazzerro (2003), Bone morphogenetic proteins, their antagonists, and the skeleton, Endocr Rev 24: 218-235. 210. R. Merino, J. Rodriguez-Leon, D. Macias, Y. Ganan, A. N. Economides, and J. M. Hurle (1999), The BMP antagonist Gremlin regulates outgrowth, chondrogenesis and programmed cell death in the developing limb, Development 126: 5515-5522. 211. P. M. Eimon, and R. M. Harland (2001), Xenopus Dan, a member of the Dan gene family of BMP antagonists, is expressed in derivatives of the cranial and trunk neural crest, Mech Dev 107: 187-189. 212. W. Balemans, M. Ebeling, N. Patel, E. Van Hul, P. Olson, M. Dioszegi, C. Lacza, W. Wuyts, J. Van Den Ende, P. Willems, A. F. Paes-Alves, S. Hill, M. Bueno, F. J. Ramos, P. Tacconi, F. G. Dikkers, C. Stratakis, K. Lindpaintner, B. Vickery, D. Foernzler, and W. Van Hul (2001), Increased bone density in sclerosteosis is due to the deficiency of a novel secreted protein (SOST), H umM ol Genet 10: 537-543. 213. M. E. Brunkow, J. C. Gardner, J. Van Ness, B. W. Paeper, B. R. Kovacevich, S. Proll, J. E. Skonier, L. Zhao, P. J. Sabo, Y. Fu, R. S. Alisch, L. Gillett, T. Colbert, P. Tacconi, D. Galas, H. Hamersma, P. Beighton, and J. Mulligan (2001), Bone dysplasia sclerosteosis results from loss of the SOST gene product, a novel cystine knot-containing protein, Am J Hum Genet 68: 577-589. 214. P. Beighton, J. Davidson, L. Durr, and H. Hamersma (1977), Sclerosteosis - an autosomal recessive disorder, Clin Genet 11: 1-7. 215. W. Balemans, N. Patel, M. Ebeling, E. Van Hul, W. Wuyts, C. Lacza, M. Dioszegi, F. G. Dikkers, P. Hildering, P. J. Willems, J. B. Verheij, K. Lindpaintner, B. Vickery, D. Foernzler, and W. Van Hul (2002), Identification of a 52 kb deletion downstream of the SOST gene in patients with van Buchem disease, J Med Genet 39: 91-97. 216. K. Staehling-Hampton, S. Proll, B. W. Paeper, L. Zhao, P. Charmley, A. Brown, J. C. Gardner, D. Galas, R. C. Schatzman, P. Beighton, S. Papapoulos, H. Hamersma, and M. E. Brunkow (2002), A 52-kb deletion in the SOST- MEOX1 intergenic region on 17q12-q21 is associated with van Buchem disease in the Dutch population, Am J Med Genet 110: 144-152. 217. J. E. Wergedal, K. Veskovic, M. Hellan, C. Nyght, W. Balemans, C. Libanati, F. M. Vanhoenacker, J. Tan, D. J. Baylink, and W. Van Hul (2003), Patients with Van Buchem disease, an osteosclerotic genetic disease, have elevated bone formation markers, higher bone density, and greater derived polar moment of inertia than normal, J Clin EndocrinolMetab 88: 5778-5783. 218. A. Hemmati-Brivanlou, O. G. Kelly, and D. A. Melton (1994), Follistatin, an antagonist of activin, is expressed in the Spemann organizer and displays direct neuralizing activity, Cell 77: 283-295. 219. A. Fainsod, K. Deissler, R. Yelin, K. Marom, M. Epstein, G. Pillemer, H. Steinbeisser, and M. Blum (1997), The dorsalizing and neural inducing gene follistatin is an antagonist of BMP-4, Mech Dev 63: 39-50. 220. F. Otsuka, R. K. Moore, S. Iemura, N. Ueno, and S. Shimasaki (2001), Follistatin inhibits the function of the oocyte- derived factor BMP-15, Biochem Biophys Res Commun 289: 961-966. 221. M. Funaba, K. Ogawa, T. Murata, H. Fujimura, E. Murata, M. Abe, M. Takahashi, and K. Torii (1996), Follistatin and activin in bone: expression and localization during endochondral bone development, Endocrinology 137: 4250-4259. 222. D. Gaddy-Kurten, J. K. Coker, E. Abe, R. L. Jilka, and S. C. Manolagas (2002), Inhibin suppresses and activin stimulates osteoblastogenesis and osteoclastogenesis in murine bone marrow cultures, Endocrinology 143: 74-83. 223. M. M. Matzuk, N. Lu, H. Vogel, K. Sellheyer, D. R. Roop, and A. Bradley (1995), Multiple defects and perinatal death in mice deficient in follistatin, Nature 374: 360-363. 224. C. Chang, D. A. Holtzman, S. Chau, T. Chickering, E. A. Woolf, L. M. Holmgren, J. Bodorova, D. P. Gearing, W. E. Holmes, and A. H. Brivanlou (2001), Twisted gastrulation can function as a BMP antagonist, Nature 410: 483-487. 225. J. J. Ross, O. Shimmi, P. Vilmos, A. Petryk, H. Kim, K. Gaudenz, S. Hermanson, S. C. Ekker, M. B. O'Connor, and J. L. Marsh (2001), Twisted gastrulation is a conserved extracellular BMP antagonist, Nature 410: 479-483. 226. I. C. Scott, I. L. Blitz, W. N. Pappano, S. A. Maas, K. W. Cho, and D. S. Greenspan (2001), Homologues of Twisted gastrulation are extracellular cofactors in antagonism of BMP signalling, Nature 410: 475-478. 227. T. Nosaka, S. Morita, H. Kitamura, H. Nakajima, F. Shibata, Y. Morikawa, Y. Kataoka, Y. Ebihara, T. Kawashima, T. Itoh, K. Ozaki, E. Senba, K. Tsuji, F. Makishima, N. Yoshida, and T. Kitamura (2003), Mammalian twisted gastrulation is essential for skeleto-lymphogenesis, M ol Cell Biol 23: 2969-2980. 228. D. Onichtchouk, Y. G. Chen, R. Dosch, V. Gawantka, H. Delius, J. Massague, and C. Niehrs (1999), Silencing of TGF-beta signalling by the pseudoreceptor BAMBI, Nature 401: 480-485. 229. M. Tsang, R. Kim, M. P. de Caestecker, T. Kudoh, A. B. Roberts, and I. B. Dawid (2000), Zebrafish nma is involved in TGFbeta family signaling, Genesis 28: 47-57. 230. W. G. Degen, M. A. Weterman, J. J. van Groningen, I. M. Cornelissen, J. P. Lemmers, M. A. Agterbos, A. Geurts van Kessel, G. W. Swart, and H. P. Bloemers (1996), Expression of nma, a novel gene, inversely correlates with the metastatic potential of human melanoma cell lines and xenografts, Int J Cancer 65: 460-465. 231. L. Grotewold, M. Plum, R. Dildrop, T. Peters, and U. Ruther (2001), Bambi is coexpressed with Bmp-4 during mouse embryogenesis, Mech Dev 100: 327-330.

- 48 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

232. C. Knight, D. Simmons, T. T. Gu, J. Gluhak-Heinrich, D. Pavlin, M. Zeichner-David, and M. MacDougall (2001), Cloning, characterization, and tissue expression pattern of mouse Nma/BAMBI during odontogenesis, J Dent Res 80: 1895-1902. 233. T. Imamura, M. Takase, A. Nishihara, E. Oeda, J. Hanai, M. Kawabata, and K. Miyazono (1997), Smad6 inhibits signalling by the TGF-beta superfamily, Nature 389: 622-626. 234. A. Nakao, M. Afrakhte, A. Moren, T. Nakayama, J. L. Christian, R. Heuchel, S. Itoh, M. Kawabata, N. E. Heldin, C. H. Heldin, and P. ten Dijke (1997), Identification of Smad7, a TGFbeta-inducible antagonist of TGF-beta signalling, Nature 389: 631-635. 235. M. Takase, T. Imamura, T. K. Sampath, K. Takeda, H. Ichijo, K. Miyazono, and M. Kawabata (1998), Induction of Smad6 mRNA by bone morphogenetic proteins, Biochem Biophys Res Commun 244: 26-29. 236. H. Hayashi, S. Abdollah, Y. Qiu, J. Cai, Y. Y. Xu, B. W. Grinnell, M. A. Richardson, J. N. Topper, M. A. Gimbrone, Jr., J. L. Wrana, and D. Falb (1997), The MAD-related protein Smad7 associates with the TGFbeta receptor and functions as an antagonist of TGFbeta signaling, Cell 89: 1165-1173. 237. A. Hata, G. Lagna, J. Massague, and A. Hemmati-Brivanlou (1998), Smad6 inhibits BMP/Smad1 signaling by specifically competing with the Smad4 tumor suppressor, Genes Dev 12: 186-197. 238. S. Bai, X. Shi, X. Yang, and X. Cao (2000), Smad6 as a transcriptional corepressor, J Biol Chem 275: 8267-8270. 239. X. Lin, Y. Y. Liang, B. Sun, M. Liang, Y. Shi, F. C. Brunicardi, and X. H. Feng (2003), Smad6 recruits transcription corepressor CtBP to repress bone morphogenetic protein-induced transcription, M ol Cell Biol 23: 9081-9093. 240. F. Itoh, H. Asao, K. Sugamura, C. H. Heldin, P. ten Dijke, and S. Itoh (2001), Promoting bone morphogenetic protein signaling through negative regulation of inhibitory Smads, EMBO J 20: 4132-4142. 241. A. Hata, J. Seoane, G. Lagna, E. Montalvo, A. Hemmati-Brivanlou, and J. Massague (2000), OAZ uses distinct DNA- and protein-binding zinc fingers in separate BMP-Smad and Olf signaling pathways, Cell 100: 229-240. 242. S. Shim, N. Bae, and J. K. Han (2002), Bone morphogenetic protein-4-induced activation of Xretpos is mediated by Smads and Olf-1/EBF associated zinc finger (OAZ), Nucleic Acids Res 30: 3107-3117. 243. M. C. Ku, S. Stewart, and A. Hata (2003), Poly(ADP-ribose) polymerase 1 interacts with OAZ and regulates BMP- target genes, Biochem Biophys Res Commun 311: 702-707. 244. T. Shioda, R. J. Lechleider, S. L. Dunwoodie, H. Li, T. Yahata, M. P. de Caestecker, M. H. Fenner, A. B. Roberts, and K. J. Isselbacher (1998), Transcriptional activating activity of Smad4: roles of SMAD hetero-oligomerization and enhancement by an associating transactivator, Proc Natl Acad Sci USA 95: 9785-9790. 245. T. Yahata, M. P. de Caestecker, R. J. Lechleider, S. Andriole, A. B. Roberts, K. J. Isselbacher, and T. Shioda (2000), The MSG1 non-DNA-binding transactivator binds to the p300/CBP coactivators, enhancing their functional link to the Smad transcription factors, J Biol Chem 275: 8825-8834. 246. S. L. Dunwoodie, T. A. Rodriguez, and R. S. Beddington (1998), Msg1 and Mrg1, founding members of a gene family, show distinct patterns of gene expression during mouse embryogenesis, Mech Dev 72: 27-40. 247. N. Nomura, S. Sasamoto, S. Ishii, T. Date, M. Matsui, and R. Ishizaki (1989), Isolation of human cDNA clones of ski and the ski-related gene, sno, Nucleic Acids Res 17: 5489-5500. 248. Y. Li, C. M. Turck, J. K. Teumer, and E. Stavnezer (1986), Unique sequence, ski, in Sloan-Kettering avian retroviruses with properties of a new cell-derived oncogene, J Virol 57: 1065-1072. 249. C. Colmenares, P. Sutrave, S. H. Hughes, and E. Stavnezer (1991), Activation of the c-ski oncogene by overexpression, J Virol 65: 4929-4935. 250. T. Nagase, G. Mizuguchi, N. Nomura, R. Ishizaki, Y. Ueno, and S. Ishii (1990), Requirement of protein co-factor for the DNA-binding function of the human ski proto-oncogene product, Nucleic Acids Res 18: 337-343. 251. T. Nagase, N. Nomura, and S. Ishii (1993), Complex formation between proteins encoded by the ski gene family, J Biol Chem 268: 13710-13716. 252. H. C. Heyman, and E. Stavnezer (1994), A carboxyl-terminal region of the ski oncoprotein mediates homodimerization as well as heterodimerization with the related protein SnoN, J Biol Chem 269: 26996-27003. 253. Y. Sun, X. Liu, E. N. Eaton, W. S. Lane, H. F. Lodish, and R. A. Weinberg (1999), Interaction of the Ski oncoprotein with Smad3 regulates TGF-beta signaling, M ol Cell 4: 499-509. 254. S. Akiyoshi, H. Inoue, J. Hanai, K. Kusanagi, N. Nemoto, K. Miyazono, and M. Kawabata (1999), c-Ski acts as a transcriptional co-repressor in transforming growth factor-beta signaling through interaction with smads, J Biol Chem 274: 35269-35277. 255. W. Xu, K. Angelis, D. Danielpour, M. M. Haddad, O. Bischof, J. Campisi, E. Stavnezer, and E. E. Medrano (2000), Ski acts as a co-repressor with Smad2 and Smad3 to regulate the response to type beta transforming growth factor, Proc Natl Acad Sci USA 97: 5924-5929. 256. W. Wang, F. V. Mariani, R. M. Harland, and K. Luo (2000), Ski represses bone morphogenic protein signaling in xenopus and mammalian cells [In Process Citation], Proc Natl Acad Sci USA 97: 14394-14399. 257. K. Luo, S. L. Stroschein, W. Wang, D. Chen, E. Martens, S. Zhou, and Q. Zhou (1999), The Ski oncoprotein interacts with the Smad proteins to repress TGFbeta signaling, Genes Dev 13: 2196-2206. 258. C. Prunier, M. Pessah, N. Ferrand, S. R. Seo, P. Howe, and A. Atfi (2003), The oncoprotein Ski acts as an antagonist of transforming growth factor-beta signaling by suppressing Smad2 phosphorylation, J Biol Chem 278: 26249-26257. 259. J. Harada, K. Kokura, C. Kanei-Ishii, T. Nomura, M. M. Khan, Y. Kim, and S. Ishii (2003), Requirement of the co­ repressor homeodomain-interacting protein kinase 2 for ski-mediated inhibition of bone morphogenetic protein- induced transcriptional activation, J Biol Chem 278: 38998-39005. 260. T. Nomura, M. M. Khan, S. C. Kaul, H. D. Dong, R. Wadhwa, C. Colmenares, I. Kohno, and S. Ishii (1999), Ski is a component of the histone deacetylase complex required for transcriptional repression by Mad and thyroid , Genes Dev 13: 412-423.

- 49 - Chapter 1

261. S. L. Stroschein, W. Wang, S. Zhou, Q. Zhou, and K. Luo (1999), Negative feedback regulation of TGF-beta signaling by the SnoN oncoprotein, Science 286: 771-774. 262. M. Takeda, M. Mizuide, M. Oka, T. Watabe, H. Inoue, H. Suzuki, T. Fujita, T. Imamura, K. Miyazono, and K. Miyazawa (2004), Interaction with Smad4 Is Indispensable for Suppression of BMP Signaling by c-Ski, Mol Biol Cell 15: 963-972. 263. M. Berk, S. Y. Desai, H. C. Heyman, and C. Colmenares (1997), Mice lacking the ski proto-oncogene have defects in neurulation, craniofacial, patterning, and skeletal muscle development, Genes Dev 11: 2029-2039. 264. R. Dahl, B. Wani, and M. J. Hayman (1998), The Ski oncoprotein interacts with Skip, the human homolog of Drosophila Bx42, Oncogene 16: 1579-1586. 265. S. Zhou, M. Fujimuro, J. J. Hsieh, L. Chen, and S. D. Hayward (2000), A role for SKIP in EBNA2 activation of CBF1-repressed promoters, J Virol 74: 1939-1947. 266. G. M. Leong, N. Subramaniam, J. Figueroa, J. L. Flanagan, M. J. Hayman, J. A. Eisman, and A. P. Kouzmenko (2001), Ski-interacting protein interacts with Smad proteins to augment transforming growth factor-beta-dependent transcription, J Biol Chem 276: 18243-18248. 267. S. Matsuda, J. Kawamura-Tsuzuku, M. Ohsugi, M. Yoshida, M. Emi, Y. Nakamura, M. Onda, Y. Yoshida, A. Nishiyama, and T. Yamamoto (1996), Tob, a novel protein that interacts with p185erbB2, is associated with anti­ proliferative activity, Oncogene 12: 705-713. 268. Y. Yoshida, S. Matsuda, and T. Yamamoto (1997), Cloning and characterization of the mouse tob gene, Gene 191: 109-113. 269. Y. Yoshida, A. von Bubnoff, N. Ikematsu, I. L. Blitz, J. K. Tsuzuku, E. H. Yoshida, H. Umemori, K. Miyazono, T. Yamamoto, and K. W. Cho (2003), Tob proteins enhance inhibitory Smad-receptor interactions to repress BMP signaling, Mech Dev 120: 629-637. 270. Y. Yoshida, S. Tanaka, H. Umemori, O. Minowa, M. Usui, N. Ikematsu, E. Hosoda, T. Imamura, J. Kuno, T. Yamashita, K. Miyazono, M. Noda, T. Noda, and T. Yamamoto (2000), Negative regulation of BMP/Smad signaling by Tob in osteoblasts, Cell 103: 1085-1097. 271. G. R. Beck, Jr., B. Zerler, and E. Moran (2001), Gene array analysis of osteoblast differentiation, Cell Growth Differ 12: 61-83. 272. M. Usui, Y. Yoshida, T. Yamashita, K. Tsuji, I. Isao, T. Yamamoto, A. Nifuji, and M. Noda (2002), Enhancing effect of Tob deficiency on bone formation is specific to bone morphogenetic protein-induced osteogenesis, J Bone Miner Res 17: 1026-1033. 273. D. Wotton, R. S. Lo, S. Lee, and J. Massague (1999), A Smad transcriptional corepressor, Cell 97: 29-39. 274. D. Wotton, R. S. Lo, L. A. Swaby, and J. Massague (1999), Multiple modes of repression by the Smad transcriptional corepressor TGIF, J Biol Chem 274: 37105-37110. 275. D. Wotton, P. S. Knoepfler, C. D. Laherty, R. N. Eisenman, and J. Massague (2001), The Smad transcriptional corepressor TGIF recruits mSin3, Cell Growth Differ 12: 457-463. 276. C. A. Hyman, L. Bartholin, S. J. Newfeld, and D. Wotton (2003), Drosophila TGIF proteins are transcriptional activators, M ol Cell Biol 23: 9262-9274. 277. K. Verschueren, J. E. Remacle, C. Collart, H. Kraft, B. S. Baker, P. Tylzanowski, L. Nelles, G. Wuytens, M. T. Su, R. Bodmer, J. C. Smith, and D. Huylebroeck (1999), SIP1, a novel zinc finger/homeodomain repressor, interacts with Smad proteins and binds to 5'-CACCT sequences in candidate target genes, J Biol Chem 274: 20489-20498. 278. D. Murray, P. Precht, R. Balakir, and W. E. Horton, Jr. (2000), The transcription factor deltaEF1 is inversely expressed with type II collagen mRNA and can repress Col2a1 promoter activity in transfected chondrocytes, J Biol Chem 275: 3610-3618. 279. J. E. Remacle, H. Kraft, W. Lerchner, G. Wuytens, C. Collart, K. Verschueren, J. C. Smith, and D. Huylebroeck (1999), New mode of DNA binding of multi-zinc finger transcription factors: deltaEF1 family members bind with two hands to two target sites, EMBO J 18: 5073-5084. 280. A. A. Postigo (2003), Opposing functions of ZEB proteins in the regulation of the TGFbeta/BMP signaling pathway, EMBO J 22: 2443-2452. 281. T. Takagi, H. Moribe, H. Kondoh, and Y. Higashi (1998), DeltaEF1, a zinc finger and homeodomain transcription factor, is required for skeleton patterning in multiple lineages, Development 125: 21-31. 282. P. Tylzanowski, K. Verschueren, D. Huylebroeck, and F. P. Luyten (2001), Smad-interacting protein 1 is a repressor of liver/bone/kidney alkaline phosphatase transcription in bone morphogenetic protein-induced osteogenic differentiation of C2C12 cells, J Biol Chem 276: 40001-40007. 283. A. A. Postigo, J. L. Depp, J. J. Taylor, and K. L. Kroll (2003), Regulation of Smad signaling through a differential recruitment of coactivators and corepressors by ZEB proteins, EMBO J 22: 2453-2462. 284. J. Long, I. Matsuura, D. He, G. Wang, K. Shuai, and F. Liu (2003), Repression of Smad transcriptional activity by PIASy, an inhibitor of activated STAT, Proc Natl Acad Sci USA 100: 9791-9796. 285. J. Long, G. Wang, I. Matsuura, D. He, and F. Liu (2004), Activation of Smad transcriptional activity by protein inhibitor of activated STAT3 (PIAS3), Proc Natl Acad Sci USA 101: 99-104. 286. S. Sturm, M. Koch, and F. A. White (2000), Cloning and analysis of a murine PIAS family member, PIASgamma, in developing skin and neurons, J Mol Neurosci 14: 107-121. 287. S. Osada, S. Y. Ohmori, and M. Taira (2003), XMAN1, an inner nuclear membrane protein, antagonizes BMP signaling by interacting with Smad1 in Xenopus embryos, Development 130: 1783-1794. 288. G. P. Raju, N. Dimova, P. S. Klein, and H. C. Huang (2003), SANE, a novel LEM domain protein, regulates bone morphogenetic protein signaling through interaction with Smad1, J Biol Chem 278: 428-437.

- 50 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

289. S. C. Chandrasekharappa, S. C. Guru, P. Manickam, S. E. Olufemi, F. S. Collins, M. R. Emmert-Buck, L. V. Debelenko, Z. Zhuang, I. A. Lubensky, L. A. Liotta, J. S. Crabtree, Y. Wang, B. A. Roe, J. Weisemann, M. S. Boguski, S. K. Agarwal, M. B. Kester, Y. S. Kim, C. Heppner, Q. Dong, A. M. Spiegel, A. L. Burns, and S. J. Marx (1997), Positional cloning of the gene for multiple endocrine neoplasia-type 1, Science 276: 404-407. 290. S. C. Guru, P. K. Goldsmith, A. L. Burns, S. J. Marx, A. M. Spiegel, F. S. Collins, and S. C. Chandrasekharappa (1998), Menin, the product of the MEN1 gene, is a nuclear protein, Proc Natl Acad Sci USA 95: 1630-1634. 291. C. Stewart, F. Parente, F. Piehl, F. Famebo, D. Quincey, G. Silins, L. Bergman, G. F. Carle, I. Lemmens, S. Grimmond, C. Z. Xian, S. Khodei, B. T. Teh, J. Lagercrantz, P. Siggers, A. Calender, V. Van de Vem, K. Kas, G. Weber, N. Hayward, P. Gaudray, and C. Larsson (1998), Characterization of the mouse Men1 gene and its expression during development, Oncogene 17: 2485-2493. 292. V. Wautot, S. Khodaei, L. Frappart, N. Buisson, E. Baro, G. M. Lenoir, A. Calender, C. X. Zhang, and G. Weber (2000), Expression analysis of endogenous menin, the product of the multiple endocrine neoplasia type 1 gene, in cell lines and human tissues, Int J Cancer 85: 877-881. 293. A. E. Gobl, M. Berg, J. R. Lopez-Egido, K. Oberg, B. Skogseid, and G. Westin (1999), Menin represses JunD- activated transcription by a histone deacetylase-dependent mechanism, Biochim Biophys Acta 1447: 51-56. 294. H. Kaji, L. Canaff, J. J. Lebrun, D. Goltzman, and G. N. Hendy (2001), Inactivation of menin, a Smad3-interacting protein, blocks transforming growth factor type beta signaling, Proc Natl Acad Sci USA 98: 3837-3842. 295. H. Sowa, H. Kaji, L. Canaff, G. N. Hendy, T. Tsukamoto, T. Yamaguchi, K. Miyazono, T. Sugimoto, and K. Chihara (2003), Inactivation of menin, the product of the multiple endocrine neoplasia type 1 gene, inhibits the commitment of multipotential mesenchymal stem cells into the osteoblast lineage, J Biol Chem 278: 21058-21069. 296. J. S. Crabtree, P. C. Scacheri, J. M. Ward, L. Garrett-Beal, M. R. Emmert-Buck, K. A. Edgemon, D. Lorang, S. K. Libutti, S. C. Chandrasekharappa, S. J. Marx, A. M. Spiegel, and F. S. Collins (2001), A mouse model of multiple endocrine neoplasia, type 1, develops multiple endocrine tumors, Proc Natl Acad Sci USA 98: 1118-1123. 297. H. Sowa, H. Kaji, G. N. Hendy, L. Canaff, T. Komori, T. Sugimoto, and K. Chihara (2004), Menin is required for bone morphogenetic protein 2- and transforming growth factor beta-regulated osteoblastic differentiation through interaction with Smads and Runx2., J Biol Chem 279: 40267-40275. 298. R. H. Kim, D. Wang, M. Tsang, J. Martin, C. Huff, M. P. de Caestecker, W. T. Parks, X. Meng, R. J. Lechleider, T. Wang, and A. B. Roberts (2000), A novel smad nuclear interacting protein, SNIP1, suppresses p300-dependent TGF- beta signal transduction, Genes Dev 14: 1605-1616. 299. R. H. Kim, K. C. Flanders, S. Birkey Reffey, L. A. Anderson, C. S. Duckett, N. D. Perkins, and A. B. Roberts (2001), SNIP1 inhibits NF-kappa B signaling by competing for its binding to the C/H1 domain of CBP/p300 transcriptional co-activators, J Biol Chem 276: 46297-46304. 300. K. Jiao, Y. Zhou, and B. L. Hogan (2002), Identification of mZnf8, a mouse Kruppel-like transcriptional repressor, as a novel nuclear interaction partner of Smad1, M ol Cell Biol 22: 7633-7644. 301. K. Kurisaki, A. Kurisaki, U. Valcourt, A. A. Terentiev, K. Pardali, P. Ten Dijke, C. H. Heldin, J. Ericsson, and A. Moustakas (2003), Nuclear factor YY1 inhibits transforming growth factor beta- and bone morphogenetic protein- induced cell differentiation, M ol Cell Biol 23: 4494-4510. 302. R. Y. Bai, C. Koester, T. Ouyang, S. A. Hahn, M. Hammerschmidt, C. Peschel, and J. Duyster (2002), SMIF, a Smad4-interacting protein that functions as a co-activator in TGFbeta signalling, Nat Cell Biol 4: 181-190. 303. M. Alvarez, P. Thunyakitpisal, P. Morrison, J. Onyia, J. Hock, and J. P. Bidwell (1998), PTH-responsive osteoblast nuclear matrix architectural transcription factor binds to the rat type I collagen promoter, J Cell Biochem 69: 336-352. 304. Z. J. Shen, T. Nakamoto, K. Tsuji, A. Nifuji, K. Miyazono, T. Komori, H. Hirai, and M. Noda (2002), Negative regulation of bone morphogenetic protein/Smad signaling by Cas-interacting zinc finger protein in osteoblasts, J Biol Chem 277: 29840-29846. Epub 22002 May 29822. 305. H. Zhu, P. Kavsak, S. Abdollah, J. L. Wrana, and G. H. Thomsen (1999), A SMAD ubiquitin ligase targets the BMP pathway and affects embryonic , Nature 400: 687-693. 306. X. Lin, M. Liang, and X. H. Feng (2000), Smurf2 is a ubiquitin E3 ligase mediating proteasome-dependent degradation of Smad2 in transforming growth factor-beta signaling, J Biol Chem 275: 36818-36822. 307. Y. Zhang, C. Chang, D. J. Gehling, A. Hemmati-Brivanlou, and R. Derynck (2001), Regulation of Smad degradation and activity by Smurf2, an E3 ubiquitin ligase, Proc Natl Acad Sci USA 98: 974-979. 308. S. Bonni, H. R. Wang, C. G. Causing, P. Kavsak, S. L. Stroschein, K. Luo, and J. L. Wrana (2001), TGF-beta induces assembly of a Smad2-Smurf2 ubiquitin ligase complex that targets SnoN for degradation, Nat Cell Biol 3: 587-595. 309. T. Ebisawa, M. Fukuchi, G. Murakami, T. Chiba, K. Tanaka, T. Imamura, and K. Miyazono (2001), Smurf1 interacts with transforming growth factor-beta type I receptor through Smad7 and induces receptor degradation, J Biol Chem 276: 12477-12480. 310. P. Kavsak, R. K. Rasmussen, C. G. Causing, S. Bonni, H. Zhu, G. H. Thomsen, and J. L. Wrana (2000), Smad7 binds to Smurf2 to form an E3 ubiquitin ligase that targets the TGF beta receptor for degradation, M ol Cell 6: 1365-1375. 311. G. Murakami, T. Watabe, K. Takaoka, K. Miyazono, and T. Imamura (2003), Cooperative inhibition of bone morphogenetic protein signaling by Smurf1 and inhibitory Smads, Mol Biol Cell 14: 2809-2817. 312. M. Zhao, M. Qiao, B. O. Oyajobi, G. R. Mundy, and D. Chen (2003), E3 ubiquitin ligase Smurf1 mediates core­ binding factor alpha1/Runx2 degradation and plays a specific role in osteoblast differentiation, J Biol Chem 278: 27939-27944. 313. M. Zhao, M. Qiao, S. E. Harris, B. O. Oyajobi, G. R. Mundy, and D. Chen (2003), Smurf1 inhibits osteoblast differentiation and bone formation in vitro and in vivo, J Biol Chem 29: 29.

- 51 - Chapter 1

314. S. X. Ying, Z. J. Hussain, and Y. E. Zhang (2003), Smurf1 facilitates myogenic differentiation and antagonizes the bone morphogenetic protein-2-induced osteoblast conversion by targeting Smad5 for degradation, J Biol Chem 278: 39029-39036. 315. D. Koinuma, M. Shinozaki, A. Komuro, K. Goto, M. Saitoh, A. Hanyu, M. Ebina, T. Nukiwa, K. Miyazawa, T. Imamura, and K. Miyazono (2003), Arkadia amplifies TGF-beta superfamily signalling through degradation of Smad7, EMBO J 22: 6458-6470. 316. L. Li, H. Xin, X. Xu, M. Huang, X. Zhang, Y. Chen, S. Zhang, X. Y. Fu, and Z. Chang (2004), CHIP mediates degradation of Smad proteins and potentially regulates Smad-induced transcription, M ol Cell Biol 24: 856-864. 317. Y. Lin, J. Martin, C. Gruendler, J. Farley, X. Meng, B. Y. Li, R. Lechleider, C. Huff, R. H. Kim, W. A. Grasser, V. Paralkar, and T. Wang (2002), A novel link between the proteasome pathway and the signal transduction pathway of the bone morphogenetic proteins (BMPs), BMC Cell Biol 3: 15. 318. H. Ichijo (1999), From receptors to stress-activated MAP kinases, Oncogene 18: 6087-6093. 319. L. Chang, and M. Karin (2001), Mammalian MAP kinase signalling cascades, Nature 410: 37-40. 320. C. A. Hazzalin, and L. C. Mahadevan (2002), MAPK-regulated transcription: a continuously variable gene switch?, N at Rev Mol Cell Biol 3: 30-40. 321. E. Shaulian, and M. Karin (2002), AP-1 as a regulator of cell life and death, Nat Cell Biol 4: E131-136. 322. C. F. Lai, and S. L. Cheng (2002), Signal transductions induced by bone morphogenetic protein-2 and transforming growth factor-beta in normal human osteoblastic cells, J Biol Chem 277: 15514-15522. 323. S. Palcy, I. Bolivar, and D. Goltzman (2000), Role of activator protein 1 transcriptional activity in the regulation of gene expression by transforming growth factor beta1 and bone morphogenetic protein 2 in ROS 17/2.8 osteoblast-like cells, J Bone Miner Res 15: 2352-2361. 324. G. Karsenty, and E. F. Wagner (2002), Reaching a genetic and molecular understanding of skeletal development, Dev Cell 2: 389-406. 325. S. Ohta, Y. Hiraki, C. Shigeno, F. Suzuki, R. Kasai, T. Ikeda, H. Kohno, K. Lee, H. Kikuchi, J. Konishi, and et al. (1992), Bone morphogenetic proteins (BMP-2 and BMP-3) induce the late phase expression of the proto-oncogene c­ fos in murine osteoblastic MC3T3-E1 cells, FEBS Lett 314: 356-360. 326. E. Chalaux, T. Lopez-Rovira, J. L. Rosa, R. Bartrons, and F. Ventura (1998), JunB is involved in the inhibition of myogenic differentiation by bone morphogenetic protein-2, J Biol Chem 273: 537-543. 327. N. Kimura, R. Matsuo, H. Shibuya, K. Nakashima, and T. Taga (2000), BMP2-induced apoptosis is mediated by activation of the TAK1-p38 kinase pathway that is negatively regulated by Smad6, J Biol Chem 275: 17647-17652. 328. T. Moriguchi, N. Kuroyanagi, K. Yamaguchi, Y. Gotoh, K. Irie, T. Kano, K. Shirakabe, Y. Muro, H. Shibuya, K. Matsumoto, E. Nishida, and M. Hagiwara (1996), A novel kinase cascade mediated by mitogen-activated protein kinase kinase 6 and MKK3, J Biol Chem 271: 13675-13679. 329. S. Gallea, F. Lallemand, A. Atfi, G. Rawadi, V. Ramez, S. Spinella-Jaegle, S. Kawai, C. Faucheu, L. Huet, R. Baron, and S. Roman-Roman (2001), Activation of mitogen-activated protein kinase cascades is involved in regulation of bone morphogenetic protein-2-induced osteoblast differentiation in pluripotent C2C12 cells, Bone 28: 491-498. 330. A. Suzuki, J. Guicheux, G. Palmer, Y. Miura, Y. Oiso, J. P. Bonjour, and J. Caverzasio (2002), Evidence for a role of p38 MAP kinase in expression of alkaline phosphatase during osteoblastic cell differentiation, Bone 30: 91-98. 331. O. Kozawa, D. Hatakeyama, and T. Uematsu (2002), Divergent regulation by p44/p42 MAP kinase and p38 MAP kinase of bone morphogenetic protein-4-stimulated osteocalcin synthesis in osteoblasts, J Cell Biochem 84: 583-589. 332. U. Noth, R. Tuli, R. Seghatoleslami, M. Howard, A. Shah, D. J. Hall, N. J. Hickok, and R. S. Tuan (2003), Activation of p38 and Smads mediates BMP-2 effects on human trabecular bone-derived osteoblasts, Exp Cell Res 291: 201-211. 333. M. Tazoe, M. Mogi, S. Goto, and A. Togari (2003), Involvement of p38MAP kinase in bone morphogenetic protein-4- induced osteoprotegerin in mouse bone-marrow-derived stromal cells, Arch Oral Biol 48: 615-619. 334. Y. Hu, E. Chan, S. X. Wang, and B. Li (2003), Activation of p38 mitogen-activated protein kinase is required for osteoblast differentiation, Endocrinology 144: 2068-2074. 335. K. S. Lee, S. H. Hong, and S. C. Bae (2002), Both the Smad and p38 MAPK pathways play a crucial role in Runx2 expression following induction by transforming growth factor-beta and bone morphogenetic protein, Oncogene 21: 7156-7163. 336. J. Guicheux, J. Lemonnier, C. Ghayor, A. Suzuki, G. Palmer, and J. Caverzasio (2003), Activation of p38 mitogen- activated protein kinase and c-Jun-NH2-terminal kinase by BMP-2 and their implication in the stimulation of osteoblastic cell differentiation, J Bone Miner Res 18: 2060-2068. 337. C. F. Lai, L. Chaudhary, A. Fausto, L. R. Halstead, D. S. Ory, L. V. Avioli, and S. L. Cheng (2001), Erk is essential for growth, differentiation, integrin expression, and cell function in human osteoblastic cells, J Biol Chem 276: 14443­ 14450. 338. C. Higuchi, A. Myoui, N. Hashimoto, K. Kuriyama, K. Yoshioka, H. Yoshikawa, and K. Itoh (2002), Continuous inhibition of MAPK signaling promotes the early osteoblastic differentiation and mineralization of the extracellular matrix, J Bone Miner Res 17: 1785-1794. 339. T. Hayashida, M. Decaestecker, and H. W. Schnaper (2003), Cross-talk between ERK MAP kinase and Smad signaling pathways enhances TGF-beta-dependent responses in human mesangial cells, Faseb J 17: 1576-1578. 340. M. Kretzschmar, J. Doody, and J. Massague (1997), Opposing BMP and EGF signalling pathways converge on the TGF-beta family mediator Smad1, Nature 389: 618-622. 341. T. Suzuki, K. T. J, R. Ajima, T. Nakamura, Y. Yoshida, and T. Yamamoto (2002), Phosphorylation of three regulatory serines of Tob by Erk1 and Erk2 is required for Ras-mediated cell proliferation and transformation, Genes Dev 16: 1356-1370.

- 52 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

342. M. Maekawa, E. Nishida, and T. Tanoue (2002), Identification of the Anti-proliferative protein Tob as a MAPK substrate, J Biol Chem 277: 37783-37787. 343. L. R. McCabe, C. Banerjee, R. Kundu, R. J. Harrison, P. R. Dobner, J. L. Stein, J. B. Lian, and G. S. Stein (1996), Developmental expression and activities of specific fos and jun proteins are functionally related to osteoblast maturation: role of Fra-2 and Jun D during differentiation, Endocrinology 137: 4398-4408. 344. W. Jochum, E. Passegue, and E. F. Wagner (2001), AP-1 in mouse development and tumorigenesis, Oncogene 20: 2401-2412. 345. S. K. Winchester, N. Selvamurugan, R. C. D'Alonzo, and N. C. Partridge (2000), Developmental regulation of collagenase-3 mRNA in normal, differentiating osteoblasts through the activator protein-1 and the runt domain binding sites, J Biol Chem 275: 23310-23318. 346. J. Hess, D. Porte, C. Munz, and P. Angel (2001), AP-1 and Cbfa/runt physically interact and regulate parathyroid hormone-dependent MMP13 expression in osteoblasts through a new osteoblast-specific element 2/AP-1 composite element, J Biol Chem 276: 20029-20038. 347. Y. Zhang, X. H. Feng, and R. Derynck (1998), Smad3 and Smad4 cooperate with c-Jun/c-Fos to mediate TGF-beta- induced transcription, Nature 394: 909-913. 348. A. Bejsovec (2000), Wnt signaling: an embarrassment of receptors, Curr Biol 10: R919-922. 349. Y. Yang (2003), Wnts and wing: Wnt signaling in vertebrate limb development and musculoskeletal morphogenesis, Birth Defects Res Part C Embryo Today 69: 305-317. 350. Y. Gong, R. B. Slee, N. Fukai, G. Rawadi, S. Roman-Roman, A. M. Reginato, H. Wang, T. Cundy, F. H. Glorieux, D. Lev, M. Zacharin, K. Oexle, J. Marcelino, W. Suwairi, S. Heeger, G. Sabatakos, S. Apte, W. N. Adkins, J. Allgrove, M. Arslan-Kirchner, J. A. Batch, P. Beighton, G. C. Black, R. G. Boles, L. M. Boon, C. Borrone, H. G. Brunner, G. F. Carle, B. Dallapiccola, A. De Paepe, B. Floege, M. L. Halfhide, B. Hall, R. C. Hennekam, T. Hirose, A. Jans, H. Juppner, C. A. Kim, K. Keppler-Noreuil, A. Kohlschuetter, D. LaCombe, M. Lambert, E. Lemyre, T. Letteboer, L. Peltonen, R. S. Ramesar, M. Romanengo, H. Somer, E. Steichen-Gersdorf, B. Steinmann, B. Sullivan, A. Superti- Furga, W. Swoboda, M. J. van den Boogaard, W. Van Hul, M. Vikkula, M. Votruba, B. Zabel, T. Garcia, R. Baron, B. R. Olsen, and M. L. Warman (2001), LDL receptor-related protein 5 (LRP5) affects bone accrual and eye development, Cell 107: 513-523. 351. P. Babij, W. Zhao, C. Small, Y. Kharode, P. J. Yaworsky, M. L. Bouxsein, P. S. Reddy, P. V. Bodine, J. A. Robinson, B. Bhat, J. Marzolf, R. A. Moran, and F. Bex (2003), High bone mass in mice expressing a mutant LRP5 gene, J Bone Miner Res 18: 960-974. 352. T. Yabe, T. Shimizu, O. Muraoka, Y. K. Bae, T. Hirata, H. Nojima, A. Kawakami, T. Hirano, and M. Hibi (2003), Ogon/Secreted Frizzled functions as a negative feedback regulator of Bmp signaling, Development 130: 2705-2716. 353. P. V. Bodine, W. Zhao, Y. P. Kharode, F. J. Bex, A. J. Lambert, M. B. Goad, T. Gaur, G. S. Stein, J. B. Lian, and B. S. Komm (2004), The Wnt antagonist secreted frizzled-related protein-1 is a negative regulator of trabecular bone formation in adult mice, Mol Endocrinol 18: 1222-1237. 354. G. Bain, T. Muller, X. Wang, and J. Papkoff (2003), Activated beta-catenin induces osteoblast differentiation of C3H10T1/2 cells and participates in BMP2 mediated signal transduction, Biochem Biophys Res Commun 301: 84-91. 355. G. Rawadi, B. Vayssiere, F. Dunn, R. Baron, and S. Roman-Roman (2003), BMP-2 controls alkaline phosphatase expression and osteoblast mineralization by a Wnt autocrine loop, J Bone Miner Res 18: 1842-1853. 356. J. Kitagaki, M. Iwamoto, J. G. Liu, Y. Tamamura, M. Pacifci, and M. Enomoto-Iwamoto (2003), Activation of beta- catenin-LEF/TCF signal pathway in chondrocytes stimulates ectopic endochondral ossification, Osteoarthritis Cartilage 11: 36-43. 357. J. A. Rudnicki, and A. M. Brown (1997), Inhibition of chondrogenesis by Wnt gene expression in vivo and in vitro, Dev Biol 185: 104-118. 358. L. Topol, X. Jiang, H. Choi, L. Garrett-Beal, P. J. Carolan, and Y. Yang (2003), Wnt-5a inhibits the canonical Wnt pathway by promoting GSK-3-independent beta-catenin degradation, J Cell Biol 162: 899-908. 359. K. Kratochwil, M. Dull, I. Farinas, J. Galceran, and R. Grosschedl (1996), Lef1 expression is activated by BMP-4 and regulates inductive tissue interactions in tooth and hair development, Genes Dev 10: 1382-1394. 360. S. M. Hussein, E. K. Duff, and C. Sirard (2003), Smad4 and beta-catenin co-activators functionally interact with lymphoid-enhancing factor to regulate graded expression of Msx2, J Biol Chem 278: 48805-48814. 361. K. Kaphingst, and S. Kunes (1994), Pattern formation in the visual centers of the Drosophila brain: wingless acts via decapentaplegic to specify the dorsoventral axis, Cell 78: 437-448. 362. E. Labbe, A. Letamendia, and L. Attisano (2000), Association of Smads with lymphoid enhancer binding factor 1/T cell-specific factor mediates cooperative signaling by the transforming growth factor-beta and wnt pathways, Proc Natl Acad Sci USA 97: 8358-8363. 363. T. Sekiya, S. Adachi, K. Kohu, T. Yamada, O. Higuchi, Y. Furukawa, Y. Nakamura, T. Nakamura, K. Tashiro, S. Kuhara, S. Ohwada, and T. Akiyama (2004), Identification of BMP and activin membrane-bound inhibitor (BAMBI), an inhibitor of transforming growth factor-beta signaling, as a target of the beta-catenin pathway in colorectal tumor cells, J Biol Chem 279: 6840-6846. 364. M. Baron, H. Aslam, M. Flasza, M. Fostier, J. E. Higgs, S. L. Mazaleyrat, and M. B. Wilkin (2002), Multiple levels of Notch signal regulation, Mol Membr Biol 19: 27-38. 365. S. Zhou, M. Fujimuro, J. J. Hsieh, L. Chen, A. Miyamoto, G. Weinmaster, and S. D. Hayward (2000), SKIP, a CBF1- associated protein, interacts with the ankyrin repeat domain of NotchIC To facilitate NotchIC function, Mol Cell Biol 20: 2400-2410. 366. L. Wu, T. Sun, K. Kobayashi, P. Gao, and J. D. Griffin (2002), Identification of a family of mastermind-like transcriptional coactivators for mammalian notch receptors, M ol Cell Biol 22: 7688-7700.

- 53 - Chapter 1

367. A. E. Wallberg, K. Pedersen, U. Lendahl, and R. G. Roeder (2002), p300 and PCAF act cooperatively to mediate transcriptional activation from chromatin templates by notch intracellular domains in vitro, Mol Cell Biol 22: 7812­ 7819. 368. V. M. Panin, L. Shao, L. Lei, D. J. Moloney, K. D. Irvine, and R. S. Haltiwanger (2002), Notch ligands are substrates for protein O-fucosyltransferase-1 and Fringe, J Biol Chem 277: 29945-29952. 369. C. Hicks, S. H. Johnston, G. diSibio, A. Collazo, T. F. Vogt, and G. Weinmaster (2000), Fringe differentially modulates Jagged1 and Delta1 signalling through Notch1 and Notch2, Nat Cell Biol 2: 515-520. 370. Y. A. Evrard, Y. Lun, A. Aulehla, L. Gan, and R. L. Johnson (1998), lunatic fringe is an essential mediator of somite and patterning, Nature 394: 377-381. 371. N. Zhang, and T. Gridley (1998), Defects in somite formation in lunatic fringe-deficient mice, Nature 394: 374-377. 372. N. Haines, and K. D. Irvine (2003), Glycosylation regulates Notch signalling, Nat Rev Mol Cell Biol 4: 786-797. 373. K. Shindo, N. Kawashima, K. Sakamoto, A. Yamaguchi, A. Umezawa, M. Takagi, K. Katsube, and H. Suda (2003), Osteogenic differentiation of the mesenchymal progenitor cells, Kusa is suppressed by Notch signaling, Exp Cell Res 290: 370-380. 374. M. Sciaudone, E. Gazzerro, L. Priest, A. M. Delany, and E. Canalis (2003), Notch 1 Impairs Osteoblastic Cell Differentiation, Endocrinology 144: 5631-5639. 375. K. Tezuka, M. Yasuda, N. Watanabe, N. Morimura, K. Kuroda, S. Miyatani, and N. Hozumi (2002), Stimulation of osteoblastic cell differentiation by Notch, J Bone Miner Res 17: 231-239. 376. D. Nofziger, A. Miyamoto, K. M. Lyons, and G. Weinmaster (1999), Notch signaling imposes two distinct blocks in the differentiation of C2C12 myoblasts, Development 126: 1689-1702. 377. T. Iso, L. Kedes, and Y. Hamamori (2003), HES and HERP families: multiple effectors of the , J Cell Physiol 194: 237-255. 378. T. Iso, V. Sartorelli, C. Poizat, S. Iezzi, H. Y. Wu, G. Chung, L. Kedes, and Y. Hamamori (2001), HERP, a novel heterodimer partner of HES/E(spl) in Notch signaling, M ol Cell Biol 21: 6080-6089. 379. M. Matsue, R. Kageyama, D. T. Denhardt, and M. Noda (1997), Helix-loop-helix-type transcription factor (HES-1) is expressed in osteoblastic cells, suppressed by 1,25(OH)2 vitamin D3, and modulates 1,25(OH)2 vitamin D3 enhancement of osteopontin gene expression, Bone 20: 329-334. 380. K. W. McLarren, F. M. Theriault, and S. Stifani (2001), Association with the nuclear matrix and interaction with Groucho and RUNX proteins regulate the transcription repression activity of the basic helix loop helix factor Hes1, J Biol Chem 276: 1578-1584. 381. D. Levanon, R. E. Goldstein, Y. Bernstein, H. Tang, D. Goldenberg, S. Stifani, Z. Paroush, and Y. Groner (1998), Transcriptional repression by AML1 and LEF-1 is mediated by the TLE/Groucho corepressors, Proc Natl Acad Sci USA 95: 11590-11595. 382. R. A. Cavallo, R. T. Cox, M. M. Moline, J. Roose, G. A. Polevoy, H. Clevers, M. Peifer, and A. Bejsovec (1998), Drosophila Tcf and Groucho interact to repress Wingless signalling activity, Nature 395: 604-608. 383. A. Blokzijl, C. Dahlqvist, E. Reissmann, A. Falk, A. Moliner, U. Lendahl, and C. F. Ibanez (2003), Cross-talk between the Notch and TGF-{beta} signaling pathways mediated by interaction of the Notch intracellular domain with Smad3, J Cell Biol 163: 723-728. 384. C. Dahlqvist, A. Blokzijl, G. Chapman, A. Falk, K. Dannaeus, C. F. Ibanez, and U. Lendahl (2003), Functional Notch signaling is required for BMP4-induced inhibition of myogenic differentiation, Development 130: 6089-6099. 385. L. Espinosa, J. Ingles-Esteve, C. Aguilera, and A. Bigas (2003), Phosphorylation by glycogen synthase kinase-3 beta down-regulates Notch activity, a link for Notch and Wnt pathways, J Biol Chem 278: 32227-32235. 386. D. A. Ross, and T. Kadesch (2001), The notch intracellular domain can function as a coactivator for LEF-1, M ol Cell Biol 21: 7537-7544. 387. F. P. Luyten, F. S. Kaplan, and E. M. Shore (2000), Clinical Disorders Associated with Bone Morphogenetic Proteins, in Skeletal Growth Factors, in Canalis, E., (Ed.), Lippincott Williams & Wilkins, Philadelphia, PA, pp. 323-334. 388. S. Mahboubi, D. L. Glaser, E. M. Shore, and F. S. Kaplan (2001), Fibrodysplasia ossificans progressiva, Pediatr Radiol 31: 307-314. 389. A. B. Shafritz, E. M. Shore, F. H. Gannon, M. A. Zasloff, R. Taub, M. Muenke, and F. S. Kaplan (1996), Overexpression of an osteogenic in fibrodysplasia ossificans progressiva, N Engl J Med 335: 555-561. 390. F. H. Gannon, D. Glaser, R. Caron, L. D. Thompson, E. M. Shore, and F. S. Kaplan (2001), Mast cell involvement in fibrodysplasia ossificans progressiva, Hum Pathol 32: 842-848. 391. L. Hegyi, F. H. Gannon, D. L. Glaser, E. M. Shore, F. S. Kaplan, and C. M. Shanahan (2003), Stromal cells of fibrodysplasia ossificans progressiva lesions express smooth muscle lineage markers and the osteogenic transcription factor Runx2/Cbfa-1: clues to a vascular origin of heterotopic ossification?, J Pathol 201: 141-148. 392. E. A. Olmsted, F. S. Kaplan, and E. M. Shore (2003), Bone morphogenetic protein-4 regulation in fibrodysplasia ossificans progressiva, Clin Orthop: 331-343. 393. J. A. Tabas, G. V. Hahn, R. B. Cohen, H. N. Seaunez, W. S. Modi, J. M. Wozney, M. Zasloff, and F. S. Kaplan (1993), Chromosomal assignment of the human gene for bone morphogenetic protein 4, Clin Orthop: 310-316. 394. M. Xu, and E. M. Shore (1998), Mutational screening of the bone morphogenetic protein 4 gene in a family with fibrodysplasia ossificans progressiva, Clin Orthop: 53-58. 395. G. Feldman, M. Li, S. Martin, M. Urbanek, J. A. Urtizberea, M. Fardeau, M. LeMerrer, J. M. Connor, J. Triffitt, R. Smith, M. Muenke, F. S. Kaplan, and E. M. Shore (2000), Fibrodysplasia ossificans progressiva, a heritable disorder of severe heterotopic ossification, maps to human chromosome 4q27-31, Am J Hum Genet 66: 128-135. 396. G. Lucotte, C. Bathelier, G. Mercier, N. Gerard, G. Lenoir, O. Semonin, and K. Fontaine (2000), Localization of the gene for fibrodysplasia ossificans progressiva (FOP) to chromosome 17q21-22, Genet Couns 11: 329-334.

- 54 - Introduction in the Control of BMP Signaling on Osteoblast Differentiation

397. G. Lucotte, O. Semonin, and P. Lutz (1999), A de novo heterozygous deletion of 42 base-pairs in the noggin gene of a fibrodysplasia ossificans progressiva patient, Clin Genet 56: 469-470. 398. O. Semonin, K. Fontaine, C. Daviaud, C. Ayuso, and G. Lucotte (2001), Identification of three novel mutations of the noggin gene in patients with fibrodysplasia ossificans progressiva, Am J Med Genet 102: 314-317. 399. J. Ahn, L. Serrano de la Pena, E. M. Shore, and F. S. Kaplan (2003), Paresis of a bone morphogenetic protein­ antagonist response in a genetic disorder of heterotopic skeletogenesis, J Bone Joint Surg Am 85-A: 667-674. 400. D. L. Glaser, A. N. Economides, L. Wang, X. Liu, R. D. Kimble, J. P. Fandl, J. M. Wilson, N. Stahl, F. S. Kaplan, and E. M. Shore (2003), In vivo somatic cell gene transfer of an engineered Noggin mutein prevents BMP4-induced heterotopic ossification, J Bone Joint Surg Am 85-A: 2332-2342. 401. J. T. Thomas, K. Lin, M. Nandedkar, M. Camargo, J. Cervenka, and F. P. Luyten (1996), A human chondrodysplasia due to a mutation in a TGF-beta superfamily member, N at Genet 12: 315-317. 402. J. T. Thomas, M. W. Kilpatrick, K. Lin, L. Erlacher, P. Lembessis, T. Costa, P. Tsipouras, and F. P. Luyten (1997), Disruption of human limb morphogenesis by a dominant negative mutation in CDMP1, N at Genet 17: 58-64. 403. N. H. Robin, M. Gunay-Aygun, A. Polinkovsky, M. L. Warman, and S. Morrison (1997), Clinical and locus heterogeneity in brachydactyly type C, Am J Med Genet 68: 369-377. 404. A. Polinkovsky, N. H. Robin, J. T. Thomas, M. Irons, A. Lynn, F. R. Goodman, W. Reardon, S. G. Kant, H. G. Brunner, I. van der Burgt, D. Chitayat, J. McGaughran, D. Donnai, F. P. Luyten, and M. L. Warman (1997), Mutations in CDMP1 cause autosomal dominant brachydactyly type C, N at Genet 17: 18-19. 405. T. Katagiri, A. Yamaguchi, M. Komaki, E. Abe, N. Takahashi, T. Ikeda, V. Rosen, J. M. Wozney, A. Fujisawa- Sehara, and T. Suda (1994), Bone morphogenetic protein-2 converts the differentiation pathway of C2C12 myoblasts into the osteoblast lineage, J Cell Biol 127: 1755-1766. 406. M. Schena, D. Shalon, R. W. Davis, and P. O. Brown (1995), Quantitative monitoring of gene expression patterns with a complementary DNA microarray, Science 270: 467-470. 407. M. Schena, D. Shalon, R. Heller, A. Chai, P. O. Brown, and R. W. Davis (1996), Parallel human genome analysis: microarray-based expression monitoring of 1000 genes, Proc Natl Acad Sci USA 93: 10614-10619. 408. D. Shalon, S. J. Smith, and P. O. Brown (1996), A DNA microarray system for analyzing complex DNA samples using two- color fluorescent probe hybridization, Genome Res 6: 639-645. 409. G. Ramsay (1998), DNA chips: state-of-the art, Nat Biotechnol 16: 40-44. 410. C. Debouck, and P. N. Goodfellow (1999), DNA microarrays in drug discovery and development, Nat Genet 21: 48­ 50.

- 55 - - 56 - 2

Microarray analysis of Bone Morphogenetic Protein, Transforming Growth Factor P, and Activin Early Response Genes During Osteoblastic Cell Differentiation

Journal of Bone Mineral Research 2002, 17, 2119-2129

Diana S. de Jong1, Everardus J.J. van Zoelen2, Susanne Bauerschmidt3, Wiebe Olijve1’3, and Wilma T. Steegenga1

1 Department of Applied Biology, University of Nijmegen, 6525 ED Nijmegen, the Netherlands 2 Department of Cell Biology, University of Nijmegen, 6525 ED Nijmegen, the Netherlands 3 N.V. Organon, Target Discovery Unit, 5340 BH Oss, the Netherlands Chapter 2

Abstract Bone morphogenetic protein 2 (BMP2), a member of the TGF-P family, is a potent regulator of osteoblast differentiation. In addition, both TGF-P and activin A can either induce bone formation or inhibit bone formation depending on cell type and differentiation status. Although much is known about the receptors and intracellular second messengers involved in the action of TGF-P family members, still little is known about how selectivity in the biological response of individual family members is controlled. In the present study, we have investigated selective gene induction by BMP2, TGF-P1 and activin A, in relation to their ability to control differentiation of mouse mesenchymal precursor cells C2C12 into osteoblastic cells. TGF-P1 can inhibit BMP2-induced differentiation of these cells, while activin A was found to be without morphogenetic effect. Using a gene expression microarray approach covering 8,636 sequences, we have identified a total of 57 established genes and ESTs that were either up-regulated or down-regulated 2 hours after treatment with at least one of these three stimuli. With respect to the established genes, 15 new target genes for TGF-P family members were thus identified. Furthermore, a set of transcripts was identified, which was oppositely regulated by TGF-P1 and BMP2. Based on the inverse biological effects of TGF-P1 and BMP2 on C2C12 cells, these genes are important candidates for controlling the process of growth factor- induced osteoblast differentiation.

Introduction The transforming growth factor P (TGF-P) family of cytokines, which includes among others TGF-P itself, activins and the bone morphogenetic proteins (BMPs), exerts extensive control over embryonic development and tissue homeostasis. Depending on the genetic make-up and environment of the target cell, many different responses can be triggered. One of the processes in which TGF-P family members play a crucial role, is bone formation and remodeling. BMP2, -4, and -7 have been characterized as important signaling molecules during the development of the skeleton in vertebrates (1, 2), and are able to induce ectopic bone formation in mammals (2-5). The effects of TGF-P1 and activin A are more complex. These cytokines have been shown to induce bone formation when injected near bone (6, 7), while on the other hand both in vitro and in vivo studies have indicated that they may impede bone formation by inhibiting the early differentiation of osteoblasts (8-10). To generate biological responses, TGF-P family members use heteromeric complexes of so- called type II and type I serine/threonine kinase receptors, resulting in the activation of specific receptor-regulated Smads (R-Smads). These include Smad2 and 3, which are specific for TGF-P and activin A signaling, while Smad1, 5 and 8 have been shown to be primarily BMP-responsive R- Smads. In combination with the common-mediator Smad4, the respective R-Smads are assembled into heteromeric complexes, which subsequently translocate to the nucleus, where they act in concert with other proteins as transcription factor complexes (11-13). Although several target genes of BMP, TGF-P and/or activin are known, some of which can be related to bone formation, a detailed analysis of TGF-P family member target genes involved in the early differentiation of osteoblast cells is still lacking. In this study, we have used the murine pluripotent mesenchymal precursor cell line C2C12 as a model system to study the effects of TGF-P family members on the early stages of osteoblast differentiation. In this cell line, BMP induces osteoblast differentiation, whereas TGF-P inhibits this effect (14). Here we show, using a gene microarray approach, that BMP2, TGF-P1, and activin A induce different sets of early response genes in these cells. By comparing these sets in relation to the osteoblast differentiating capacity of the three

- 58 - Microarray analysis of BMP, TGF P, and Activin Early Response Genes During Osteoblastic Cell Differentiation

growth factors, various novel transcripts were identified that may play a direct role in bone development.

Materials and methods Tissue culture and growth factors The C2C12 murine mesenchymal progenitor cell line (obtained from the American Type Culture Collection) was maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% newborn calf serum (NCS), 44 mM NaHCO3, penicillin (100 U/ml) and streptomycin (100 |ig/ml) at 37oC in a humidified atmosphere at 7.5% CO2. For the various experiments, cells were plated at a density of 5.7x104 cells/cm2 and grown for two days until confluence. Recombinant human BMP2 was kindly provided by Genetics Institute (Andover, MA), recombinant human TGF-P1 and recombinant human activin A were purchased from R&D Systems (Minneapolis, MN).

Alkaline phosphatase assay Cells were seeded in 96-well tissue culture plates (Nalge Nunc International, Denmark) and grown as described above. Subsequently, the medium was changed to DMEM containing 0.5% NCS (low serum medium), supplemented with increasing doses of BMP2, TGF-P1, activin A, or combinations of these factors. Alkaline phosphatase activity was measured 48h after stimulation as described by Van der Plas and coworkers (15). Alkaline phosphatase activity was corrected for differences in cell number based on a measurement of neutral red staining (16).

Stimulation of cells and RNA isolation Cells were seeded and grown in 145 cm2 culture dishes (Nalge Nunc International, Denmark) as described above. Subsequently, the medium was changed to low serum medium and cells were treated for 0.5, 1, 1.5 or 2 hours with BMP2 , TGF-P1 , or activin A . Non-treated controls were kept on low serum medium for 2 hours. Total RNA was extracted according to an acid guanidium thiocyanate-phenol-chloroform method (TriPure® Isolation Reagent, Boehringer Mannheim, Germany) and RNA concentrations were determined by measuring the absorbance at 260 nm. Poly A+ RNA was prepared from total RNA using the Oligotex™ method (Qiagen, Valencia, CA). Quantification of poly A+ RNA was carried out according to the Ribogreen™ RNA quantitation assay (Molecular Probes, Eugene, OR).

Reverse Transcriptase-PCR Complementary DNA (cDNA) was synthesized from 2 |ig of total RNA from non-stimulated cells, using the SUPERSCRIPTTM II reverse transcriptase kit (GibcoBRL, Breda, the Netherlands). Subsequently, PCR reactions were performed (Geneamp 2400 PCR system, Perkin Elmer, Boston, MA) under the following conditions: initial denaturation for 2 min at 94oC followed by 30 cycles consisting of 1 min at 94oC, 1 min at 58oC for the receptors ActRIA, BMPRIA, ActRIB, ActRIIA, ActRIIB, BMPRII, and TPRII or at 62oC for TPRI, BMPRIB, and GAPDH, followed by 1.5 min at 72oC and a final extension of 10 min at 72oC. For all PCR primers, shown in Table 1, control reactions without SUPERSCRIPTTMII reverse transcriptase were performed to exclude genomic DNA contamination as a source of amplified signal. All these samples were negative for receptor signal (data not shown). Aliquots of first-strand cDNA (2 |il) were used in the PCR reactions containing 20 ng/|il of forward and reverse primers in PCR buffer (GibcoBRL) supplemented with 2 mM MgCl2 and 3% DMSO, 200 mM dNTPs (MBI Fermentas, Vilnius, Lithuania) and 0.5 U Taq polymerase

- 59 - Chapter 2

(GibcoBRL) in a final volume of 50 pl. The PCR products were visualized on ethidium bromide containing 1% agarose gels using the Geldoc2000 photosystem (Biorad, Hercules, CA).

Table 1. Primers for Reverse Transcriptase-PCR analysis T a ble 2. Primers for real-time quantitative RT-PCR analysis

Gene Primers (5’^ 3 ’) size Gene Primers (5’^ 3 ’) size (bp) (bp) Timp3 F : gca att gca aga tca ag tcc tg 52 ActRIA F: gga gta atg agc ctt cct gtg c 512 R : ttg gag gtc aca aaa caa ggc R: ttg gtg gtg atg agc cct tc CXCR4 F : atc agc ctg gac cgg tac c 52 BMPRIA F: gat ggg taa aat gcg tgg 522 R : tgg cct ttg act gtt ggt g R: gat gta ggg ctg gaa atg gt AI466638 F : atg aga cca acc ttt cac agc a 95 ActRIB F: cac tga cac cat aga cat tgc tcc a 331 R : cca agg gag cat acc tgt tct t R: gca tca tct ttc cca tca ctc gca a F, forward primer; R, reverse primer TßRI F: ctg cca cct ctg tac aaa gg 721 R: cac agc tgc gtc cat gtc c BMPRIB F: gac act ccc att cct cat c 357 R: gct att gtc ctt tgg acc ag ActRIIA F: tct ctt gct ctt cag gtg ct 479 R: gaa caa gta cag gag ggt ag ActRIIB F: ctg ctg gct aga tga ctt c 698 R: ttc cac gtg atg atg ttc c BMPRII F: tca aga acg gct gtg tgc at 587 R: cgc tca tcc agg gaa cct tt TßRII F: aca tca acc aca aca cgg agc 672 R: tca ttt ccc aga gta cca gag GAPDH F: ccc ttc att cac ctc acc tac atg g 853 R: ggt cca cca ccc tgt tgc tgt agc c

ActRIA, activin receptor type IA; BMPRIA, BMP receptor type IA; ActRIB, activin receptor type IB; TpRI, TGF-p receptor type I; BMPRIB, BMP receptor type IB; ActRIIA, activin receptor type IIA; ActRIIB, activin receptor type IIB; BMPRII, BMP receptor type II; TpRII, TGF-p receptor type II; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; F, forward primer; R, reverse primer.

Real-time quantitative RT-PCR For real-time quantitative RT-PCR, ABI Prism Sequence Detection System 5700 and Primer Express™ (PE Biosystems) were used. For each gene, a set of primers was designed using sequences obtained from Genbank™ and amplicons of 50-100 basepairs with TM between 58 and 60°C were selected (Table 2). Prior to complementary DNA (cDNA) synthesis, 2 ^g of total RNA or 20 ng of poly A+ RNA were DNase treated for 15 min at 37°C, after which RNA was reverse transcribed from random hexamer primers using the SUPERSCRIPT™ II reverse transcriptase kit (GibcoBRL) (50 |il total volume). Subsequently, aliquots of first-strand cDNA (2|il) were amplified using SYBR Green® PCR Mastermix (Applied Biosystems) under the following conditions: initial denaturation for 10 min at 95°C followed by 40 cycles consisting of 15 sec at 94°C and 1 min at 60°C. Fold inductions and expression ratios were calculated from differences in threshold cycles at which an increase in reporter fluorescence above a baseline signal could first be detected (CT) between two samples, and averaged for duplicate experiments.

- 60 - Microarray analysis of BMP, TGF P, and Activin Early Response Genes During Osteoblastic Cell Differentiation

Northern Blotting Samples of 20 |ig of total RNA or 200 ng poly A+ RNA were separated electrophoretically on 1% sodium phosphate-glyoxal-agarose gels and subsequently transferred to Hybond N+ membranes (Amersham Pharmacia, Piscataway, NJ) using 25 mM sodium phosphate buffer, pH 6.5. RNA was crosslinked to the membrane by baking the blots for two hours at 80oC. Prehybridization (30 min.) and hybridization (15 h) were performed in Church Buffer (0.5 M sodium phosphate buffer, pH7.2, 7% SDS, 1 mM EDTA) at 65oC. The following mouse probes were used: JunB (17), Smad7 (18), Tieg (GenBank Accession Number AA476158), Myd118 (GenBank Accession Number W64388), Vdup1-pending (GenBank Accession Number W34721), and GAPDH (PCR product). Probes were labeled with 32P-a-dCTP using a random priming labeling kit (Ready-To-Go™ DNA labeling Beads, Amersham Pharmacia). For visualization, membranes were exposed to Kodak X-AR films at -80oC using intensifying screens and then quantified using the GS-700 Imaging Densitometer (Biorad), or the Molecular Imager FX (Biorad).

Gene Microarray and Statistical Analysis Isolated poly A+ RNA was labeled by reverse transcriptase with nucleotides containing either a Cy3 or Cy5 fluorophore. The following hybridization experiments were performed in duplicate with a fluor reversal to minimize possible bias caused by the molecular structure of the Cy3 and Cy5 dyes: 1) BMP2 treated versus control, 2) TGF-P1 treated versus control and 3) activin A treated versus control. For these hybridization experiments, the Mouse GEM™ 1 array from Incyte Genomics (Incyte Genomics Inc., Palo Alto, CA) was used. Incyte Genomics performed both the labeling and the microarray hybridizations according to their standard protocols. Incyte GEMtools™ software was used for image analysis. Background-subtracted element signals were used to calculate Cy3:Cy5 ratios. Genes were selected for further analysis, if the signal to background ratio for a particular gene was > 2.5 and at least 40% of the area on chip could be used for signal computation in at least one of the two channels. To account for systematic errors caused by non-linear differences in signal intensities of the Cy3 or Cy5 channel, a regression analysis was performed by fitting a parabola to the experimentally observed expression values. Using these fitted values, genes with potentially differential expression were selected according to the following selection method: a normal distribution of non-differentially expressed genes was assumed. Next, the inter-quartile range of the distribution was calculated as a measure for the spread of the non-differentially expressed genes. To each edge of this range, 1.41 times the inter-quartile range was added, corresponding to a p-value of 0.01. Genes were considered to be significantly regulated, if they fell outside this range in both of the inverse duplicate experiments and if the intensity ratio was indeed inverse for both experiments. This selection was performed per gene for each growth factor separately. Using this technique, we were able to identify significantly regulated genes for the entire data set with a fold regulation as measured on array down to 1.3 for BMP2 and 1.5 for TGF-P1 and for activin A

Results C2C12 cells can mediate signal transduction by BMP2, TGF-fil and activin A To examine gene regulation induced by BMP2 during the early phase of osteoblast differentiation, the murine mesenchymal progenitor cell line C2C12 was used as a model system. In this study, we have compared the activity of BMP2 with that of two other members of the TGF-P family, namely activin A and TGF-P1. Since it is known that BMP-induced osteoblast differentiation is paralleled by the induction of alkaline phosphatase (ALP) activity, we first tested the dose-

- 61 - Chapter 2

dependence of the effect of these three factors on osteoblast differentiation by measuring ALP activity under culture conditions similar to the ones we have subsequently used for the microarray experiments. As is shown in Figure 1A, strong up-regulation of this activity was observed when the cells were cultured for two days in the presence of BMP2 (IC50 = 150 ng/ml), whereas TGF-p1 and activin A on their own had no effect on ALP activity. TGF-p1 was able to inhibit the effect of BMP2 on differentiation (IC50 = 0.7 ng/ml), as indicated by the inhibition of ALP activity, while activin A had no significant effect on the BMP2-stimulated activity (Figure 1B). In short, BMP2 has a stimulatory effect on differentiation of C2C12 cells, activin A has a neutral and TGF-p 1 a negative effect on BMP2-induced differentiation.

Figure 1. Biological activity of BMP2, TGF-P1 and activin A in C2C12 cells. (A) Dose dependent induction of alkaline phosphatase (ALP) activity 48 hours after treatment with BMP2 (•), TGF-p1 (■) and activin A (0). (B) Inhibition of BMP2 activity (100 ng BMP2/ml) by increasing doses of TGF-p 1 (■) and activin A (0). ALP activity was corrected for differences in cell density based on neutral red staining.

In order to examine the receptors that mediate the effects of these three TGF-P family members, we tested the cells for the expression of type I and type II serine/threonine kinases by reverse transcription (RT)-PCR. As is shown in Figure 2A, expression of the type II receptors ActRIIA, BMPRII and TPRII was observed, while no ActRIIB expression could be detected. In addition, all tested type I receptors were expressed, although expression of the BMPRIB and the ActRIB seemed to be lower than that of TPRI, BMPRIA and ActRIA. Together, these results show that C2C12 cells express receptors relevant for signal transduction of BMP2 (BMPRII or ActRIIA in combination with BMPRIA, BMPRIB or ActRIA), TGF-P1 (TPRII together with TPRI) as well as activin A (ActRIIA together with ActRIA or ActRIB). The lack of activity of activin A on C2C12 differentiation therefore does not seem to be caused by the absence of receptor expression.

Figure 2. C2C12 cells are able to mediate signal transduction by BMP2, TGF-P 1 and activin A. (A) C2C12 cells express the receptors required for signal transduction of BMP2, TGF-P1 and activin A. RNA of non­ stimulated cells was used for RT-PCR analysis. Primers for mouse sequences of TGF-P receptors I (TflRI; 721 bp) and II (TPRII; 672 bp), BMP receptors IA (BMPRIA; 522 bp), IB (BMPRIB; 357 bp), and II (BMPRII; 587 bp), and activin receptors IA (ActRIA; 512 bp), IB (ActRIB; 331 bp), IIA (ActRIIA; 479 bp) and IIB (ActRIIB; 698 bp) were used. (b,C) Northern blot analysis of JunB and Smad7 expression. (B) Total RNA from C2C12 cells treated for the indicated times with BMP2 (300 ng/ml) or TGF-P1 (5 ng/ml) was used. (C) Total RNA from C2C12 cells treated for 2 hours with BMP2 (300 ng/ml), TGF-P 1 (5 ng/ml) or activin A (20 ng/ml) was used. As a control RNA was taken from cells treated with only low serum medium.

- 62 - Microarray analysis of BMP, TGF P, and Activin Early Response Genes During Osteoblastic Cell Differentiation

We subsequently determined the optimal incubation time for studying early gene responses induced by TGF-P family members in our assay system.. As we observed earlier, BMP2 (300 ng/ml) strongly induced ALP activity after incubation for 2 days (Figure 1A), while TGF-P1 (5 ng/ml) strongly inhibited BMP2-induced ALP activity (Figure 1B), so these concentrations were used in the subsequent experiments. Activin A, which had no significant effect on BMP2-induced ALP activity (Figure 1B), was applied at a concentration of 20 ng/ml similarly as used elsewhere (19). To determine the optimal incubation time, expression regulation of two previously identified TGF-P1 and BMP2 early response genes, JunB and Smad7, was examined (18-21). As can be seen in Figure 2B, expression of both genes was up-regulated during the first 2 hours after addition of BMP2 or TGF-P1, such that these factors displayed a 5-fold and a 9-fold induction of JunB expression respectively, whereas activin A gave only a 2-fold induction of this gene (Figure 2C). Also expression of Smad7 was significantly up-regulated after 2 hours of incubation with BMP2 (3 fold) and TGF-P1 (3 fold). These data show that a 2-hour treatment of C2C12 cells with all three TGF-P family members results in significant up-regulation of early response genes.

BMP2, TGF-01 and activin A-induced gene expression profiles Despite the rapid progress in understanding the signaling mechanisms downstream of receptors for TGF-P-related growth factors, the relation between the biological effects of TGF-P family members on bone differentiation and their induced early genes is as yet poorly characterized. In addition, a systematic search for early target genes of BMPs, activins and TGF-Ps in osteoblasts is still lacking. In this study, we examined the early gene expression patterns induced by BMP2, TGF- P1 and activin A in C2C12 cells by using cDNA microarray analysis based on a large part of the available data from the mouse genome. For this analysis, we used the Mouse GEM 1 array, which consists of 8,636 sequences (Incyte Genomics). Similar as above, we studied expression profiles 2 hours after addition of BMP2, TGF-P1 and activin A to the cells (Figure 2C). To examine specific regulation by these three factors, competitive hybridizations with stimulated versus non-stimulated samples were performed. To minimize possible bias caused by the molecular structure of the Cy3 and Cy5 dyes, each experiment was performed in duplicate with a fluor reversal. Genes were considered to be significantly regulated by one of the growth factors, if they exhibited inverse expression ratios per duplicate hybridization and if these ratios had a p-value < 0.01 according to our statistical analysis (see Material and Methods). Of the 8,636 sequences spotted on the microarray, 57 sequences were identified that were significantly regulated by at least one of the three growth factors (1.3-fold modulation for BMP2 and 1.5 fold modulation for TGF-P1 and activin A). Subsequently, these 57 genes were further categorized depending on their extend of co-regulation by these factors. In our experience, regulation levels of 1.2 or more on microarray can generally be confirmed by other techniques like Northern Blot analysis and quantitative RT-PCR (see also Figure 3). Regulation levels between 1.0 and 1.2 cannot always be confirmed by these other techniques and this group of genes generally contains a number of false positives. We therefore used a threshold value of 1.2 for grouping the significantly regulated genes into the 10 categories shown in Table 3. A group of 44 genes was up-regulated by both BMP2 and TGF-P 1, or by all three factors (categories II and III), while in addition, 3 genes were down-regulated by both BMP2 and TGF-P1 or by all three factors (categories VIII and IX). Furthermore, 4 transcripts were up- or down-regulated by BMP2 only (categories I and X) and 2 transcripts were identified that were specifically regulated by activin A and/or TGF-P1, but not by BMP2 (categories IV and V). Moreover, 2 genes were up- regulated by BMP2 but down-regulated by TGF-P1 (category VI), and 2 genes were down-regulated

- 63 - Chapter 2

by BMP2 while up-regulated by TGF-p1 and activin A (category VII). Table 3 also shows that of the 57 genes identified, 31 were ESTs and 26 were already established genes. Based on their functions, these 26 genes can be subdivided into 14 transcription regulators, 5 growth factors and extracellular matrix proteins, 3 receptors, 1 neurotransmitter, 1 inorganic phosphate transporter, and 2 enzymes (Table 4).

Table 3. BMP2, TGF-(31 and activin A exhibit different sets of significantly regulated early genes (p < 0.01). Genbank Category Accession no. Gene Name BMP2 T G F-pi activin A I BÎ AA002322 EST 1.0 1.0 AI894068 Mm. Fos-like antigen 2 1.0 1.0 II B,TÎ W 88094 Mm. Chemokine (C-X-C) receptor 4 1.0 A A 1I8626 EST 2.0 2.3 1.1 A A060806 Mm. TSC22-related inducible lb 1.9 1.7 1.0 W 82201 EST 1.7 1.3 1.0 W 13229 EST 1.6 1.7 1.1 A A 1I9479 Mm. Lymphoid enhancer binding factor 1 1.6 1.7 1.1 A I5I00I0 EST 1.6 1.7 1.0 AA259873 EST 1.6 1.3 1.1 AI466603 EST 1.5 1.6 1.0 AA547162 Mm. Forkhead box Ol 1.5 1.5 1.1 A A033308 EST 1.5 1.5 1.1 W 98395 Mm. Discoidin domain receptor 1 1.5 1.3 1.0 A A423149 Mm. Cyr61 1.5 1.4 1.0 A A061644 Mm. Smad 7 1.4 1.5 1.0 W 14184 EST 1.4 1.5 1.0 W 852I7 EST 1.3 1.6 1.0 A A184015 EST 1.3 1.5 1.1 A A 12539I EST 1.1 III B,T,AÎ AA036380 Mm. Snail homolog A A144094 EST 2.2 3.2 1.9 AA184207 EST 1.8 2.7 1.5 AA060863 EST 1.8 1.5 1.3 W 85374 EST 1.8 1.5 1.3 A A098166 Mm. Placental growth factor 1.7 1.5 1.2 W 64388 Mm. M y d ll8 1.5 3.5 1.4 AI595201 Mm. HB-EGF 1.5 2.8 1.3 W 45975 Mm. TNF receptor lb 1.5 2.0 1.2 AA271373 Mm. Heparan sulfate 6-O-sulfotransferase 1 1.5 1.4 1.2 A A 06424I Mm. Stimulated by retinoic acid 1.4 3.6 1.9 A A476158 Mm. Tieg 1.4 2.2 1.5 W 36541 Mm. Connective tissue growth factor 1.4 1.9 1.2 A A245505 Mm. Citokine inducible SH2 protein 2 1.4 1.4 1.2 A A177949 Mm. Solute carrier family 20, member 1 1.3 1.9 1.3 A A472299 EST 1.3 1.8 1.4 A A 44462I EST 1.3 1.7 1.3 AI509105 EST 1.3 1.7 1.2 AA036347 Mm. Kruppel-like factor 9 1.3 1.6 1.3 AA386857 EST 1.3 1.6 1.3 AA275737 Mm. Retinal short-chain dehydrogenase/reductase 1 1.3 1.6 1.2 W 66845 EST 1.3 1.6 1.2 A A145784 Mm. Timp3 1.3 1.5 1.2 A A260654 Mm. Tg interacting factor 1.2 2.4 1.6 AA030427 EST 1.2 1.5 1.3 W 98074 EST IV t ,a T AI322575 Mm. Cocaine and amphetamine regulated transcript 1.1 V TÎ AI390830 EST 1.0 1.1 VI b Tt ^ A A 27133I EST -1.2 1.0 W 87955 EST -1.3 -1.1 VII B ^ T ,A t AI552496 EST -1.4 W 64999 Mm. H esl -1.6 VIII B,T^ AI466638 EST -1.2 -1.7 -1.1 IX b ,t ,a 4. W 34721 Mm. Vitamin D upregulated protein 1 -1.3 -1.6 -1.3 A A432818 EST -1.2 -1.6 -1.2 X bI AA174312 EST -1.3 1.0 1.0 AI604140 EST -1.4 1.0 1.0 Genes regulated at least 1.3 times by BMP2 (300ng/ml), 1.5 times by TGF-P1 (5ng/ml) or 1.5 times by activin A (20 ng/ml) are shown, categorized according to expression profiles, based on regulation values of at least 1.2 for all growth factors. Green, up- regulated; red, down-regulated; black, not regulated; B, BMP2; T, TGF-P1; A, activin A; T, up-regulated; i , down-regulated. Each value is the average of two reciprocal hybridizations. See page 138 for full color figure

- 64 - Microarray analysis of BMP, TGF p, and Activin Early Response Genes During Osteoblastic Cell Differentiation

Validation of gene expression regulation by Northern blot analysis Because the dynamic range of these cDNA microarrays is usually lower than of Northern blot analysis or real time quantitative RT-PCR (39), we used these latter two techniques to verify and validate the expression of some of the genes identified by our microarray analysis experiments. Figure 3 shows the expression profiles of six genes, analyzed by Northern blot analysis (Figure 3A) or quantitative RT-PCR (Figure 3B). The genes were chosen because they displayed different levels of expression regulation by BMP2, TGF-P1 and activin A that were representative of the dynamic range of the microarray experiment. The genes TGF-P inducible early growth response (TIEG), Myogenic differentiation 118 (Myd118), Tissue inhibitor if metalloproteinase 3 (Timp3) and Chemokine receptor 4 (CXCR4) were chosen to represent the up-regulated genes and Vitamin D- upregulated protein 1 (VDup-1) and the EST AI466638 were chosen to represent the down-regulated genes. Qualitatively, the expression profiles of the genes tested confirmed the microarray results. However, the level of modulation of these four genes according to Northern blot analysis and RT- PCR was 1-3 times stronger than obtained by microarray analysis. So, although the qualitative expression profiles of the microarray results were confirmed by two independent other methods, the extent of expression regulation differed between the techniques used.

Figure 3. Validation of selected gene expression values. (A) Northern Blot analysis. Poly A+ RNA similarly as in the microarray experiment was used to measure expression levels for two stimulated genes and one repressed gene. Expression levels (fold change ) were corrected for GAPDH and compared to microarray data. (B) Quantitative RT-PCR analysis. poly A+ RNA from the original microarray experiment and RNA from a biological duplo were used to measure expression levels for two stimulated genes and one repressed gene. Mean expression levels were corrected for GAPDH and the serum control and compared to microarray data. Error bars indicate differences between duplo experiments. C, 0.5% serum control; B, BMP2 (300 ng/ml); T, TGF-P 1 (5 ng/ml); A, activin A (20 ng/ml).

Discussion In this study, cDNA microarray analysis was used to correlate early gene expression by members of the TGF-P family with their ability to control osteoblast differentiation. Significant up- regulation of two previously reported early response genes junB and Smad7 was observed after 2 hrs of treatment with BMP2, TGF-P1 or activin A (Figure 2B + C). Therefore, this time point was used to search for additional early response genes. Based on analysis of 8,636 mouse genes and ESTs selected to represent the mouse genome, we identified a group of 57 transcripts that were significantly (p<0.01) regulated for at least one of the three stimuli 2 hours after treatment with BMP2, TGF-P1 or activin A. The 57 transcripts identified here consisted of 31 ESTs and 26 established genes of which only 11 had previously been described

- 65 - Chapter 2

as target genes for TGF-P family members. A total of 15 new target genes were therefore identified in this study, including various transcription factors, enzymes, a growth factor and a neurotransmitter (Table 4).

Table 4. Functions of the regulated genes Previously It has been reported Accession nr. Gene name Category established TGF-p previously that the dynamic range of family target gene these cDNA microarrays is usually Transcription factor lower than more conventional AA245505 Mm. Cish2 B ,T,A t W64999 Mm. Hes-1 B iT ,A t methods like Northern Blot analysis AA547162 Mm. Foxo1 B ,T t or real-time quantitative RT-PCR AI894068 Mm. Fosl-2 (Fra-2) B t (39). To verify our results, we AA036347 Mm. Klf-9 B,T,A t checked the expression regulation of AA119479 Mm. Lef-1 B ,T t (22) six of the 57 genes and ESTs W64388 Mm. MyD118 B ,T,A t (23,24) identified, (TIEG, Myd118, VDup-1, AA061644 Mm. Smad 7 B ,T t (18,21,25) Timp3, CXCR4 and AI466638) by AA036380 Mm. Snail (Sna) B,T,A t either one of these two methods AA064241 Mm. Stra13 B,T,A t (Figure 3A and B). These six genes AA260654 Mm. Tgif T ,A t were chosen so as to represent the AA476158 Mm. TIEG B,T,A t (26-28) dynamic range of this microarray W34721 Mm. Vdup-1 B X A i experiment. The results obtained AA060806 Mm. Tilz-1 B ,T t showed that the qualitative expression Receptor profiles could be confirmed, but that, W88094 Mm. CXCR4 B ,T t (29) as reported previously, the microarray W98395 Mm. Ddr1 B ,T t data underestimated the fold W45975 Mm. Tnfr1b B ,T,A t (30,31) regulation observed with alternative Growth factor techniques. AA423149 Mm. Cyr61 B ,T t (32) BMP2 is a strong inducer of W36541 Mm. Fisp12 B ,T,A t (33) osteoblast differentiation, whereas AI595201 Mm. HB-EGF B ,T,A t (34) TGF-P1 and activin A are not capable AA098166 Mm. Pgf B,T,A t of inducing osteoblast differentiation Enzyme according to alkaline phosphatase AA271373 Mm. Hs6st-1 B,T,A t activity (Fig 1A). In concurrence with AA275737 Mm. Rsdr-1 B ,T t this, our microarray analysis AA177949 Mm. Slc20a1 B,T,A t (35) identified 4 transcripts that are Extracellular matrix protein specifically regulated by BMP2 but AA145784 Mm. Timp3 B ,T t (36-38) not by TGF-P1 or activin A. These

Neurotransmitter include the transcription factor Fosl-2 AI322575 Mm. Cart T ,A t (or Fra-2) and 3 ESTs (Table 3, categories I and X). Interestingly, the In this table the functions of the regulated genes are listed as were found by literature search. In addition, genes previously described as target genes of Fosl-2 gene has not been described BMP-2 and TGF-p are marked previously as a BMP2 early response gene, although it is known to be involved in osteoblast differentiation (40) and capable to activate the osteocalcin promoter (41). In addition, transcripts inversely regulated by BMP2 and TGF-P1 are interesting, since, in agreement with previous observations (14), we found that TGF-P1 can inhibit the BMP2-induced

- 66 - Microarray analysis of BMP, TGF P, and Activin Early Response Genes During Osteoblastic Cell Differentiation

osteoblast differentiation of the C2C12 cells (Figure 1A). In this respect, we found 2 ESTs that were activated by BMP2 but repressed by TGF-P1 (category VI), and we found the Hes1 gene and another EST to be inhibited by BMP2 but up-regulated by TGF-P1 and activin A (category VII). Since activin A cannot inhibit the BMP2-induced osteoblast differentiation it seems less likely that the genes in this latter category are key modulators of the osteoblast differentiation process. On the other hand, it has been shown that the Hes1 protein is a regulator of the osteoblast specific transcription factor Cbfa1/Osf2 (42, 43) and a repressor of osteopontin expression in osteoblasts (44). Therefore, although the specificity of the Hes1 activity is not exactly clear, its functioning in the BMP2-induced differentiation process deserves further investigation. It has been described by Katagiri and coworkers that BMP2 can inhibit the spontaneous myoblast differentiation of C2C12 cells at low serum concentration by suppressing the transcriptional activity of myogenic factors such as MyoD and myogenin (45). In parallel with these observations, we show here that BMP2 can activate the expression of the discoidin domain receptor 1 (Ddr-1), which has been reported to be a myoblast differentiation inhibitor (46). The Ddr-1 gene was also found to be induced by TGF-P1, which is in agreement with previous publications showing that TGF- P1 mimics BMP2 in its ability to inhibit myoblast differentiation of C2C12 cells (14, 20). TGF-P family members initiate signaling by assembling receptor complexes that activate Smad transcription factors. Recent data have indicated that the DNA binding affinity of Smad complexes for Smad binding element (SBE) consensus sequences is generally quite low in the absence of DNA-binding cofactors, which allow additional DNA contacts (13). For some of the TGF-P family member target genes identified in this study, promoter sequences have been described in the literature, and these regions contain a variety of SBEs and related sequences as well as consensus sequences for various known Smad cofactors (data not shown). For example, the promoter of the Vdup-1 gene contains 5 copies of the consensus sequence for the Sip1 transcription factor. This transcription factor has previously been reported to be involved in repression of gene expression (47, 48), and since expression of the Vdup-1 gene was found to be down-regulated by all three growth factors in our study, the repressive effect of TGF-P family members on the Vdup-1 expression is most likely mediated by Sip1. However, analysis of these various promoter regions for the presence of SBEs and consensus sequences of DNA-binding cofactors did not allow us to determine a minimum requirement for BMP2, TGF-P1 or activin A responsiveness, particularly since no comparison with TGF-P family member insensitive genes could be made. Mutation analysis of the putative binding elements involved, as described for example for the Smad7 promoter (49), therefore remains the most straightforward approach, although new bio-informatic methods to perform in silico promoter analysis, as recently described for yeast (50, 51), hold great potential for the future. Of the 26 target genes identified in this study, 14 encode a transcription factor. The observation that these transcription factors are differentially regulated by BMP2, TGF-P1 and activin A indicates that these three growth factors are able to control their own set of transcriptional regulators, thereby inducing independent intracellular pathways and finally giving rise to different biological responses. A more detailed analysis of these transcriptional regulators might contribute to the further elucidation of the signal transduction pathways induced by different TGF-P family members. Previous studies have also indicated that TGF-P family members can negatively control their own signaling efficiency by inducing expression of inhibitory Smads, such as Smad7 (18, 21, 52). Our current results show that TGF-P1 and activin A are also able to induce the TG interacting factor (Tgif) gene, another known Smad co-repressor. This shows that TGF-P family members induce expression of multiple genes involved in an auto-regulatory feed back mechanism.

- 67 - Chapter 2

Although a set of new target genes for the TGF-P family was identified in this study, various known target genes remained undetected. For example, we found only few up-regulated extracellular matrix genes in our analyses, in contrast to a recent study by Verrecchia et al. on expression regulation of a selected group of genes in fibroblast cells (53). Gene regulation by TGF-P family members has been reported to be strongly cell type dependent, while in addition gene expression patterns may depend on the time of incubation. In agreement with this consideration, we observed strong up-regulation of various extracellular matrix genes in the later stages of BMP-induced differentiation of C2C12 cells (Vaes et al., personal communication). Moreover, we are aware that various relevant target genes, including JunB, were not present on the microarray used in this study. Finally, although our statistical analysis allowed us to identify target genes down to a regulation level of only 1.3, the possibility of false negatives in the group of non-regulated genes can certainly not be excluded. Previous studies on C2C12 cells have shown that both BMP2 and TGF-P1 can stimulate the expression of the bone specific transcription factor Cbfa1 (54). In addition, Alliston et al. have recently shown that the inhibitory effect of TGF-P1 on osteoblast differentiation is mediated by Smad3 due to a physical interaction with the Cbfa1 protein, which causes repression of its transcriptional activity through the DNA consensus sequence OSE2 (55). Based on these observations, transcripts that are regulated not only by BMP2 but also by the other family members can be directly involved in the process of bone development. Taken together, we identified a number of novel target genes which might directly or indirectly be involved in BMP2-induced osteoblast differentiation. It will be of great interest to sequence and further characterize the regulated ESTs to determine whether they are indeed important novel early modulators in the osteoblast differentiation process.

Acknowledgements We are grateful to dr. W. Kruijer (Department of Genetics, University of Groningen, the Netherlands) for the generous gift of the JunB probe and to P. ten Dijke (Department of Cellular Biochemistry, the Netherlands Cancer Institute, Amsterdam, the Netherlands) for the generous gift of the Smad7 probe. Recombinant human BMP2 was kindly provided by the Genetics Institute (Cambridge, MA). We are grateful to W. Mulder, C. Boersma and K. Dechering for technical support and helpful discussions. This research was supported by the Netherlands Institute for Earth and Life Sciences (NWO-ALW).

References 1. F. J. Hughes, J. Collyer, M. Stanfield, and S. A. Goodman (1995), The effects of bone morphogenetic protein-2, -4, and -6 on differentiation of rat osteoblast cells in vitro, Endocrinology 136: 2671-2677. 2. S. Cheifetz, I. W. Li, C. A. McCulloch, K. Sampath, and J. Sodek (1996), Influence of osteogenic protein-1 (OP- 1;BMP-7) and transforming growth factor-beta 1 on bone formation in vitro, Connect Tissue Res 35: 71-78. 3. M. Ahrens, T. Ankenbauer, D. Schroder, A. Hollnagel, H. Mayer, and G. Gross (1993), Expression of human bone morphogenetic proteins-2 or -4 in murine mesenchymal progenitor C3H10T1/2 cells induces differentiation into distinct mesenchymal cell lineages, DNA Cell Biol 12: 871-880. 4. P. E. Mead, I. H. Brivanlou, C. M. Kelley, and L. I. Zon (1996), BMP-4-responsive regulation of dorsal-ventral patterning by the homeobox protein Mix.1, Nature 382: 357-360. 5. E. A. Wang, V. Rosen, P. Cordes, R. M. Hewick, M. J. Kriz, D. P. Luxenberg, B. S. Sibley, and J. M. Wozney (1988), Purification and characterization of other distinct bone-inducing factors, Proc Natl Acad Sci USA 85: 9484-9488. 6. J. Massague (1990), The transforming growth factor-beta family, Annu Rev Cell Biol 6: 597-641. 7. Y. Oue, H. Kanatani, M. Kiyoki, Y. Eto, E. Ogata, and T. Matsumoto (1994), Effect of local injection of activin A on bone formation in newborn rats, Bone 15: 361-366.

- 68 - Microarray analysis of BMP, TGF p, and Activin Early Response Genes During Osteoblastic Cell Differentiation

8. T. Ikenoue, S. Jingushi, K. Urabe, K. Okazaki, and Y. Iwamoto (1999), Inhibitory effects of activin-A on osteoblast differentiation during cultures of fetal rat calvarial cells, J Cell Biochem 75: 206-214. 9. R. Fujimoto, T. Tanizawa, S. Nishida, N. Yamamoto, S. Soshi, N. Endo, and H. E. Takahashi (1999), Local effects of transforming growth factor-beta1 on rat calvaria: changes depending on the dose and the injection site, J Bone Miner Metab 17: 11-17. 10. E. C. Breen, R. A. Ignotz, L. McCabe, J. L. Stein, G. S. Stein, and J. B. Lian (1994), TGF beta alters growth and differentiation related gene expression in proliferating osteoblasts in vitro, preventing development of the mature bone phenotype, J Cell Physiol 160: 323-335. 11. C. H. Heldin, K. Miyazono, and P. ten Dijke (1997), TGF-beta signalling from cell membrane to nucleus through SMAD proteins, Nature 390: 465-471. 12. J. Massague (1998), TGF-beta signal transduction, Annu Rev Biochem 67: 753-791. 13. J. Massague, and D. Wotton (2000), Transcriptional control by the TGF-beta/Smad signaling system, EMBO J 19: 1745-1754. 14. T. Katagiri, A. Yamaguchi, M. Komaki, E. Abe, N. Takahashi, T. Ikeda, V. Rosen, J. M. Wozney, A. Fujisawa- Sehara, and T. Suda (1994), Bone morphogenetic protein-2 converts the differentiation pathway of C2C12 myoblasts into the osteoblast lineage, J Cell Biol 127: 1755-1766. 15. A. van der Plas, E. M. Aarden, J. H. Feijen, A. H. de Boer, A. Wiltink, M. J. Alblas, L. de Leij, and P. J. Nijweide (1994), Characteristics and properties of osteocytes in culture, J Bone Miner Res 9: 1697-1704. 16. C. W. Lowik, M. J. Alblas, M. van de Ruit, S. E. Papapoulos, and G. van der Pluijm (1993), Quantification of adherent and nonadherent cells cultured in 96-well plates using the supravital stain neutral red, Anal Biochem 213: 426-433. 17. K. Ryder, L. F. Lau, and D. Nathans (1988), A gene activated by growth factors is related to the oncogene v-jun, Proc Natl Acad Sci USA 85: 1487-1491. 18. A. Nakao, M. Afrakhte, A. Moren, T. Nakayama, J. L. Christian, R. Heuchel, S. Itoh, M. Kawabata, N. E. Heldin, C. H. Heldin, and P. ten Dijke (1997), Identification of Smad7, a TGFbeta-inducible antagonist of TGF-beta signalling, Nature 389: 631-635. 19. L. J. Jonk, S. Itoh, C. H. Heldin, P. ten Dijke, and W. Kruijer (1998), Identification and functional characterization of a Smad binding element (SBE) in the JunB promoter that acts as a transforming growth factor-beta, activin, and bone morphogenetic protein-inducible enhancer, J Biol Chem 273: 21145-21152. 20. E. Chalaux, T. Lopez-Rovira, J. L. Rosa, R. Bartrons, and F. Ventura (1998), JunB is involved in the inhibition of myogenic differentiation by bone morphogenetic protein-2, J Biol Chem 273: 537-543. 21. A. Ishisaki, K. Yamato, A. Nakao, K. Nonaka, M. Ohguchi, P. ten Dijke, and T. Nishihara (1998), Smad7 is an activin-inducible inhibitor of activin-induced growth arrest and apoptosis in mouse B cells, J Biol Chem 273: 24293­ 24296. 22. H. R. Dassule, and A. P. McMahon (1998), Analysis of epithelial-mesenchymal interactions in the initial morphogenesis of the mammalian tooth, Dev Biol 202: 215-227. 23. M. Selvakumaran, H. K. Lin, R. T. Sjin, J. C. Reed, D. A. Liebermann, and B. Hoffman (1994), The novel primary response gene MyD118 and the proto-oncogenes , , and bcl-2 modulate transforming growth factor beta 1- induced apoptosis of myeloid leukemia cells, M ol Cell Biol 14: 2352-2360. 24. M. Vairapandi, A. G. Balliet, A. J. Fornace, B. Hoffman, and D. A. Liebermann (1996), The differentiation primary response gene MyD118, related to GADD45, encodes for a nuclear protein which interacts with PCNA and p21WAF1/CIP1, Oncogene 12: 2579-2594. 25. M. Fujii, K. Takeda, T. Imamura, H. Aoki, T. K. Sampath, S. Enomoto, M. Kawabata, M. Kato, H. Ichijo, and K. Miyazono (1999), Roles of bone morphogenetic protein type I receptors and Smad proteins in osteoblast and chondroblast differentiation, M ol Biol Cell 10: 3801-3813. 26. M. Subramaniam, S. A. Harris, M. J. Oursler, K. Rasmussen, B. L. Riggs, and T. C. Spelsberg (1995), Identification of a novel TGF-beta-regulated gene encoding a putative zinc finger protein in human osteoblasts, Nucleic Acids Res 23: 4907-4912. 27. T. E. Hefferan, M. Subramaniam, S. Khosla, B. L. Riggs, and T. C. Spelsberg (2000), Cytokine-specific induction of the TGF-beta inducible early gene (TIEG): regulation by specific members of the TGF-beta family, J Cell Biochem 78: 380-390. 28. T. E. Hefferan, G. G. Reinholz, D. J. Rickard, S. A. Johnsen, K. M. Waters, M. Subramaniam, and T. C. Spelsberg (2000), Overexpression of a nuclear protein, TIEG, mimics transforming growth factor-beta action in human osteoblast cells, J Biol Chem 275: 20255-20259. 29. K. Sato, H. Kawasaki, H. Nagayama, M. Enomoto, C. Morimoto, K. Tadokoro, T. Juji, and T. A. Takahashi (2000), TGF-beta 1 reciprocally controls chemotaxis of human peripheral blood monocyte-derived dendritic cells via chemokine receptors, J Immunol 164: 2285-2295. 30. F. Geissmann, P. Revy, A. Regnault, Y. Lepelletier, M. Dy, N. Brousse, S. Amigorena, O. Hermine, and A. Durandy (1999), TGF-beta 1 prevents the noncognate maturation of human dendritic Langerhans cells, J Immunol 162: 4567­ 4575. 31. J. Martel-Pelletier, F. Mineau, F. C. Jolicoeur, and J. P. Pelletier (1995), Modulation of TNFSR55 and TNFSR75 by cytokines and growth factors in human synovial fibroblasts, J Rheumatol Suppl 43: 115-119. 32. A. Hollnagel, V. Oehlmann, J. Heymer, U. Ruther, and A. Nordheim (1999), Id genes are direct targets of bone morphogenetic protein induction in embryonic stem cells, J Biol Chem 274: 19838-19845.

- 69 - Chapter 2

33. T. Nakanishi, Y. Kimura, T. Tamura, H. Ichikawa, Y. Yamaai, T. Sugimoto, and M. Takigawa (1997), Cloning of a mRNA preferentially expressed in chondrocytes by differential display-PCR from a human chondrocytic cell line that is identical with connective tissue growth factor (CTGF) mRNA, Biochem Biophys Res Commun 234: 206-210. 34. N. Bulus, and J. A. Barnard (1999), Heparin binding epidermal growth factor-like growth factor is a transforming growth factor beta-regulated gene in intestinal epithelial cells, Biochem Biophys Res Commun 264: 808-812. 35. G. Palmer, J. Guicheux, J. P. Bonjour, and J. Caverzasio (2000), Transforming growth factor-beta stimulates inorganic phosphate transport and expression of the type III phosphate transporter Glvr-1 in chondrogenic ATDC5 cells, Endocrinology 141: 2236-2243. 36. S. Varghese, and E. Canalis (1997), Regulation of collagenase-3 by bone morphogenetic protein-2 in bone cell cultures, Endocrinology 138: 1035-1040. 37. D. R. Edwards, K. J. Leco, P. P. Beaudry, P. W. Atadja, C. Veillette, and K. T. Riabowol (1996), Differential effects of transforming growth factor-beta 1 on the expression of matrix metalloproteinases and tissue inhibitors of metalloproteinases in young and old human fibroblasts, Exp Gerontol 31: 207-223. 38. S. Su, F. Dehnade, and M. Zafarullah (1996), Regulation of tissue inhibitor of metalloproteinases-3 gene expression by transforming growth factor-beta and dexamethasone in bovine and human articular chondrocytes, DNA Cell Biol 15: 1039-1048. 39. D. Shiffman, T. Mikita, J. T. Tai, D. P. Wade, J. G. Porter, J. J. Seilhamer, R. Somogyi, S. Liang, and R. M. Lawn (2000), Large scale gene expression analysis of cholesterol-loaded macrophages, J Biol Chem 275: 37324-37332. 40. L. R. McCabe, C. Banerjee, R. Kundu, R. J. Harrison, P. R. Dobner, J. L. Stein, J. B. Lian, and G. S. Stein (1996), Developmental expression and activities of specific fos and jun proteins are functionally related to osteoblast maturation: role of Fra-2 and Jun D during differentiation, Endocrinology 137: 4398-4408. 41. C. Banerjee, J. L. Stein, A. J. Van Wijnen, B. Frenkel, J. B. Lian, and G. S. Stein (1996), Transforming growth factor­ beta 1 responsiveness of the rat osteocalcin gene is mediated by an activator protein-1 binding site, Endocrinology 137: 1991-2000. 42. P. Ducy (2000), Cbfa1: a molecular switch in osteoblast biology, Dev Dyn 219: 461-471. 43. K. W. McLarren, R. Lo, D. Grbavec, K. Thirunavukkarasu, G. Karsenty, and S. Stifani (2000), The mammalian basic helix loop helix protein HES-1 binds to and modulates the transactivating function of the runt-related factor Cbfa1, J Biol Chem 275: 530-538. 44. M. Matsue, R. Kageyama, D. T. Denhardt, and M. Noda (1997), Helix-loop-helix-type transcription factor (HES-1) is expressed in osteoblastic cells, suppressed by 1,25(OH)2 vitamin D3, and modulates 1,25(OH)2 vitamin D3 enhancement of osteopontin gene expression, Bone 20: 329-334. 45. T. Katagiri, S. Akiyama, M. Namiki, M. Komaki, A. Yamaguchi, V. Rosen, J. M. Wozney, A. Fujisawa-Sehara, and T. Suda (1997), Bone morphogenetic protein-2 inhibits terminal differentiation of myogenic cells by suppressing the transcriptional activity of MyoD and myogenin, Exp Cell Res 230: 342-351. 46. W. Vogel, C. Brakebusch, R. Fassler, F. Alves, F. Ruggiero, and T. Pawson (2000), Discoidin domain receptor 1 is activated independently of beta(1) integrin, J Biol Chem 275: 5779-5784. 47. K. Verschueren, J. E. Remacle, C. Collart, H. Kraft, B. S. Baker, P. Tylzanowski, L. Nelles, G. Wuytens, M. T. Su, R. Bodmer, J. C. Smith, and D. Huylebroeck (1999), SIP1, a novel zinc finger/homeodomain repressor, interacts with Smad proteins and binds to 5'-CACCT sequences in candidate target genes, J Biol Chem 274: 20489-20498. 48. P. Tylzanowski, K. Verschueren, D. Huylebroeck, and F. P. Luyten (2001), Smad-interacting protein 1 is a repressor of liver/bone/kidney alkaline phosphatase transcription in bone morphogenetic protein-induced osteogenic differentiation of C2C12 cells, J Biol Chem 276: 40001-40007. 49. M. Stopa, D. Anhuf, L. Terstegen, P. Gatsios, A. M. Gressner, and S. Dooley (2000), Participation of smad2, smad3, and smad4 in transforming growth factor beta (TGF-beta )-induced activation of smad7., J Biol Chem 275: 29308­ 29317. 50. S. Tavazoie, J. D. Hughes, M. J. Campbell, R. J. Cho, and G. M. Church (1999), Systematic determination of genetic network architecture, N at Genet 22: 281-285. 51. T. J. Lyons, A. P. Gasch, L. A. Gaither, D. Botstein, P. O. Brown, and D. J. Eide (2000), Genome-wide characterization of the Zap1p zinc-responsive regulon in yeast, Proc Natl Acad Sci USA 97: 7957-7962. 52. S. Bai, X. Shi, X. Yang, and X. Cao (2000), Smad6 as a transcriptional corepressor, J Biol Chem 275: 8267-8270. 53. F. Verrecchia, M. L. Chu, and A. Mauviel (2001), Identification of Novel TGF-beta /Smad Gene Targets in Dermal Fibroblasts using a Combined cDNA Microarray/Promoter Transactivation Approach, J Biol Chem 276: 17058-17062. 54. K. S. Lee, H. J. Kim, Q. L. Li, X. Z. Chi, C. Ueta, T. Komori, J. M. Wozney, E. G. Kim, J. Y. Choi, H. M. Ryoo, and S. C. Bae (2000), Runx2 is a common target of transforming growth factor beta1 and bone morphogenetic protein 2, and cooperation between runx2 and smad5 induces osteoblast-specific gene expression in the pluripotent mesenchymal precursor cell line C2C12, Mol Cell Biol 20: 8783-8792. 55. T. Alliston, L. Choy, P. Ducy, G. Karsenty, and R. Derynck (2001), TGF-beta-induced repression of CBFA1 by Smad3 decreases cbfa1 and osteocalcin expression and inhibits osteoblast differentiation, EMBO J 20: 2254-2272.

- 70 - 3

Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation

Journal of Bone Mineral Research 2004, 19, 947-958

Diana S. de Jong1, Bart L.T. Vaes1, Koen J. Dechering2, Alie Feijen3, Jose M.A. Hendriks1, Ron Wehrens4, Christine L. Mummery3, Everardus J. J. van Zoelen5, Wiebe Olijve1, 2 and Wilma T. Steegenga1, 6

1 Department of Applied Biology, University of Nijmegen, 6525 ED Nijmegen, the Netherlands 2 Target Discovery Unit, N.V. Organon, Oss, the Netherlands 3 Hubrecht Laboratory, Netherlands Institute of Developmental Biology, Utrecht, the Netherlands 4 Department of Analytical Chemistry, University of Nijmegen, Nijmegen, the Netherlands 5 Department of Cell Biology, University of Nijmegen, 6525 ED Nijmegen, the Netherlands 6 Current address: Division of Human Nutrition, Wageningen University, Wageningen, the Netherlands Chapter 3

Micro Abstract Key regulatory components of the BMP-induced osteoblast differentiation cascade remain to be established. Microarray and subsequent expression analyses in mice identified two transcription factors, Heyl and Tcf7, with in vitro and in vivo expression characteristics very similar to Cbfal. Transfection studies suggest that Tcf7 modulates BMP2-induced osteoblast differentiation. This study contributes to a better definition of the onset of BMP-induced osteoblast differentiation.

Abstract Introduction: Elucidation of the genetic cascade guiding mesenchymal stem cells to become osteoblasts is of extreme importance for improving the treatment of bone-related diseases like osteoporosis. The aim of this study is to identify regulators of the early phases of BMP2-induced osteoblast differentiation. Methods: Osteoblast differentiation of mouse C2C12 cells was induced by treatment with BMP2 and regulation of gene expression studied during the subsequent 24 hrs using high-density microarrays. The regulated genes were grouped by means of model-based clustering and protein functions were assigned. Real-time quantitative RT-PCR analysis was used to validate BMP2-induced gene expression patterns in C2C12 cells. Osteoblast specificity was studied by comparing these expression patterns with those in C3H10T1/2 and NIH3T3 cells under similar conditions. In situ hybridization of mRNA in embryos at embryonic day (E)14.5 and E16.5 of gestation and on newborn mouse tails were used to study in vivo expression patterns. Cells constitutively expressing the regulated gene Tcf7 were used to investigate its influence on BMP-induced osteoblast differentiation. Results and Conclusions: 184 genes and expressed sequence tags (ESTs) were differentially expressed in the first 24 hrs following BMP2 treatment and grouped in subsets of immediate early, intermediate early and late early response genes. Signal transduction regulatory factors mainly represented the subset of immediate early genes. Regulation of expression of these genes was direct, independent of de novo protein synthesis and independent of the cell-type studied. The intermediate early and late early genes consisted primarily of genes related to processes that modulate morphology, basement membrane formation and synthesis of extracellular calcified matrix. The late early genes require de novo protein synthesis and show osteoblast specificity. In vivo and in vitro experiments showed that the transcription factors Hey1 and Tcf7 exhibited expression characteristics and cell type specificity very similar to those of the osteoblast specific transcription factor Cbfa1, and constitutive expression of Tcf7 in C2C12 cells differentially regulated osteoblast differentiation marker genes.

Introduction Mesenchymal stem cells have the potential to differentiate into various cell types that include osteoblasts, chondrocytes, adipocytes, myoblasts and fibroblasts (1, 2) depending on the signaling cascades activated during the initial phase of differentiation. Each of these signaling mechanisms is likely to regulate a specific set of early response genes that ultimately determine the final fate of the differentiation process. Bone is a highly specialized tissue, but osteoblasts are nevertheless very similar to fibroblasts in terms of gene expression and only a very limited number of osteoblast specific genes have been identified (3). Among the few established osteoblast specific genes are the transcription factors Core-binding factor 1 (Cbfa1) and Osterix (Osx). Cbfa1 is a key regulator of osteoblast differentiation (4-6) and its regulation and function have been studied extensively over the last decade. Several isoforms of Cbfa1 have now been identified, of which the type II isoform is bone specific (7, 8). The activity of the Cbfa1 protein is modulated through post-translational modifications as well as through protein-

- 72 - Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation

protein interactions (9-12). Most of the well-established bone markers, such as Alkaline phosphatase, Collagen I, Bone Sialoprotein, Osteopontin and Osteocalcin, are known target genes of at least one of the Cbfa1 isoforms (4, 13-15). Moreover, the transcription factor Osx acts downstream of Cbfa1 and both Cbfa1 and Osx knock-out mice exhibit a complete absence of skeletal ossification (6, 16). Osteoblast differentiation can be induced by various stimuli, but the most potent inducers are the Bone Morphogenetic Proteins (BMPs). Although the BMPs were first identified for their ability to induce ectopic bone formation in vivo (17), these proteins appear to be multifunctional regulators of morphogenesis during embryonal development (18). In addition to the BMP-induced Smad signal transduction pathway, several major signal transduction pathways, including the Wnt and Notch pathways, have been implicated in osteoblast differentiation. Nevertheless, only parts of the complete gene regulatory program guiding mesenchymal stem cells to become mature osteoblasts are currently known and many gaps still need to filling. Recently, we have shown that gene expression microarray analysis is a powerful approach to identifying late phase bone marker genes (19) and to examine differential responses induced by BMP2, TGF-P1 and activin A (20). The aim of the present study was to identify novel potential regulators of the osteoblast differentiation process that are regulated during the early phases (within 24 hrs) of BMP-induced osteoblast differentiation using high-density microarrays. Model-based clustering and functional analysis identified two osteoblast-related transcription factors, Hey1 and Tcf7, which exhibited very similar expression characteristics to the osteoblast-specific type II Cbfa1 iso form. In addition, Tcf7 was shown to differentially regulate osteoblast differentiation marker genes.

Materials and Methods. Tissue culture and growth factors. The murine mesenchymal progenitor cell lines C2C12 and C3H10T1/2 and the fibroblast cell line NIH3T3 (obtained from the American Type Culture Collection [ATCC]) were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% newborn calf serum (NCS), penicillin (100 U/ml) and streptomycin (100 |ig/ml) as described previously (20). For the various experiments, cells were plated at a density of 2.0x104 cells/cm2 and grown for 24 hrs in DMEM supplemented with 10% NCS. Subsequently, medium was replaced by DMEM containing 5% NCS either in the presence or in the absence of 300 ng/ml recombinant human BMP2 with or without 50 |j.M cyclohexamide (CHX). Recombinant human BMP2 was kindly provided by Wyeth (Andover). Cyclohexamide was purchased from Sigma.

RNA isolation Total RNA was extracted according to an acid guanidium thiocyanate-phenol-chloroform method (TriPure® Isolation Reagent, Roche Diagnostics) and RNA concentrations were determined by measuring the absorbance at 260 nm. Poly A+ RNA was prepared from total RNA using the Oligotex™ method (Qiagen). Quantification of poly A+ RNA was carried out according to the Ribogreen™ RNA quantitation assay (Molecular Probes).

Microarray and statistical analysis Microarray analysis was performed as described before (20). For each time point, the BMP2- treated samples were hybridized versus control samples that were incubated for the same period in DMEM supplemented with 5% NCS. For the hybridization experiments, the mouse Unigene 1 array from Incyte Genomics (Palo Alto) was used, and the experiments were performed in duplicate with a

- 73 - Chapter 3

fluor reversal to minimize possible bias caused by the molecular structure of the Cy3 and Cy5 dyes. Background-subtracted element signals were used to calculate Cy3/Cy5 ratios. Genes were selected for further analysis if the signal to background ratio for a particular gene was > 2.5 and at least 40% of the area on chip could be used for signal computation in at least one of the two channels. To account for systematic errors caused by non-linear differences in signal intensities of the Cy3 and Cy5 channel, statistical analysis was performed as described before (20). Genes with significant (p<0.01) differential expression regulation for at least one of the time points were selected for further analysis.

Prior to clustering, microarray data were log transformed (base 10) according to: logI = log (ICy3/ICy5 ) if ICy3 > ICy5 and logI = -log (ICy5/ICy3) if ICy5 > ICy3, in which I is the intensity of each spot on microarray. Log values were averaged over duplicate experiments and scaled to unit vector length (normalization). Subsequently, normalized log values were clustered using model-based clustering (21). Briefly, in this method clusters are considered to be groups displaying multivariate normal distributions. Various models were fitted to the data and an estimation was made of the accuracy of the model describing the data by assuming that underlying probability distributions occur for each model. Subsequently, a selection was made based on maximizing Bayesian Information Criterion (BIC) values for the specific model and number of clusters that represented the data best. Optimal clustering was obtained using a clustering method in which the distribution of the expression ratios within a cluster were ellipsoidal, the cluster volumes variable, the cluster shapes equal and the cluster orientations variable (VEV-model). Protein functions of the genes were identified according to Incytes Protein Function Hierarchy.

Real-time RT-PCR For real-time RT-PCR, the ABI Prism Sequence Detection System 5700 and Primer ExpressTM Software (PE Biosystems) were used. For each gene, a set of primers was designed using sequences obtained from Genbank™ and amplicons of 50-100 base pairs with TM between 58 and 60°C were selected. Prior to complementary DNA (cDNA) synthesis, 2 ^g of total RNA were DNase (Invitrogen) treated for 15 min at 37°C, after which RNA was reverse-transcribed using random hexamer primers (Amersham Pharmacia) in a total volume of 50 |il using the SUPERSCRIPT™ II reverse transcriptase kit (Invitrogen). Subsequently, aliquots of first-strand cDNA (2|il) were amplified using SYBR Green® PCR Master mix (Applied Biosystems) under the following conditions: initial denaturation for 10 min at 95°C followed by 40 cycles consisting of 15 sec at 94°C and 1 min at 60°C. Fold inductions and expression ratios were calculated from differences in threshold cycles at which an increase in reporter fluorescence above a baseline signal could first be detected (CT) between two samples, and averaged for duplicate experiments.

In situ hybridization Mouse embryos were collected from F1 females (Cba x C57B16) at embryonic day 14.5 and 16.5 (E14.5 and E16.5), dissected free from the uterus and membranes, fixed overnight at 4oC in 4% paraformaldehyde, dehydrated, embedded in paraffin wax and sectioned (6 |im) onto coated slides (Klinipath, Duiven, the Netherlands). Neonatal mouse tails that were kindly provided by dr. Marcel Karperien (Leiden University, the Netherlands) were collected from Swiss albino mice on day 5 after birth and fixed in 4% paraformaldehyde, decalcified by incubation in 12.5% sodium EDTA (pH 8.0) for 1 week at 4oC, embedded in paraffin wax and sectioned (5 |im) as described previously (22). The slides were dewaxed, rehydrated and pretreated with Proteinase K and acetic anhydride, and hybridized overnight at 55oC essentially as described previously (23). Constructs were cloned into pSKII- in the anti-sense orientation in the T7 promoter. RNA probes were generated by transcription of the T7 RNA polymerase in the presence of [a-35S]-uridine triphosphate (UTP). After hybridization,

- 74 - Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation

the slides were washed at high stringency for 30 min at 62oC, treated with RNase A for 30 min at 37oC, washed again at high stringency, dehydrated and allowed to dry. Autoradiography was performed using Ilford G5 photo emulsion. The slides were exposed for 2 weeks at 4oC. Photographs were made from combined bright-field (blue filter) and dark-field images (red filter).

Plasmids and transfections The mouse Tcf7 cDNA cloned in pcDNA1 was kindly provided by dr. Hans Clevers (Hubrecht Laboratory, the Netherlands), while the mouse Cbfa1 cDNA cloned in pBluescript and the p6OSE2 luciferase reporter construct were kindly provided by dr. Patricia Ducy (Baylor College of Medicine, Houston). The renilla reporter construct pRL-SV40 was purchased from Promega. The Tcf7 coding region was excised from the plasmid by HindIII / XbaI digestion and ligated into HindIII /XbaI-digested pcDNA3.1. The Cbfa1 coding region was excised from the plasmid by EcoRV/XbaI digestion and ligated into EcoR V / XbaI-digested pcDNA3.1. The integrity of the resulting pTcf7-3.1 and pCbfa1-3.1 plasmids was verified by DNA sequence analysis. For stable transfections, pTcf7-3.1 was introduced into C2C12 cells using the Lipofectamin method according to the manufacturer’s specifications (Gibco/BRL). Monoclonal stable transfectants were selected on G418 (400 |ig/ml) and Tcf7 expression levels were analyzed by real-time RT-PCR. For transient transfections, pCbfa1-3.1, pTcf7-3.1, p6OSE2 and pRL-SV40 (Promega) were introduced into C2C12 cells using the Lipofectamin method according to the manufacturer’s specifications (Gibco/BRL). Twenty four hours after transfection, luciferase levels were measured .

Table 1. Protein functions represented on the microarray. Group Subgroup Group Subgroup Signal transduction and Extracellular messengers Adhesion and molecular Adhesion regulation Receptors recognition Antigen recognition Intracellular signaling Transcription factors

Membrane Transport Channels and porins Electron transfer Cytochromes Transporters and pumps Iron-sulphur proteins Flavoproteins Other Protein modification and Proteases Localized and structural Extracellular matrix maintenance Protease inhibitors proteins Cell wall Protein glycosyl transferases Surface structure Protein lipid transferases Membrane Protein acetyl transferases Periplasm Protein isomerases Cytoplasm Chaperones Cytoskeleton Ribosome Chromatin Organelle Nucleic acid synthesis and Polymerases EST modification Ligases Nucleases Methylases Helicases Topoisomerases Recombinases Splicing factors Single stranded binding proteins Other Protein functions of the regulated genes were categorized according to Subgroups (Incytes Protein Function Hierarchy). Per gene only 1 functional group was assigned.

- 75 - Chapter 3

Results BMP2 induced expression profiles Upon BMP2 treatment, C2C12 mesenchymal progenitor cells differentiate into osteoblastic cells, whereas they form myotubes in response to low serum conditions (24). Previously, we have shown that under identical experimental conditions as used here, C2C12 cells differentiate into osteoblasts which strongly express bone markers such as Alkaline phosphatase (ALP) and Osteocalcin (OCN) 2 days after BMP2 treatment (19). To identify novel potential regulators of BMP2-induced differentiation, we studied the onset of the differentiation process by microarray analysis (shown in Figure 1). Of the 9,330 genes spotted on the array, 184 sequences were found to exhibit significant (p<0.01) differential expression in the first 24 hours following BMP2 treatment. Model-based clustering, a method recently described for efficient handling of microarray data (25, 26), was used to group the 184 genes by similarities in their expression profiles resulting in the identification of 10 distinct clusters. When classified into functional groups according to the function of their cognate protein (Table 1), the majority (94%) of the genes in our data set fell in the groups “Signal Transduction”, “Localized and Structural Proteins”, or “Expressed Sequence Tags” (ESTs). Figure 1 shows an overview of the clustering results and the distribution of the genes over the functional categories is indicated per cluster. Figure 1A shows the 6 clusters with genes for which the most notable differential expression is positive and Figure 1B shows the 4 clusters with genes for which this is negative. Globally, genes from clusters 2, 3, 4, 7 and 10 can be categorized as immediate early genes that display distinct regulation 2 hrs after BMP2 treatment. Clusters 5, 6 and 9 represent intermediate early genes that were regulated from 6 hrs onwards, and clusters 1 and 8 represent late early genes for which no expression regulation was observed until 24 hrs after BMP2 treatment. Of the immediate early genes 90% (clusters 2, 3, 4 and 7) showed increased expression, while only 10% (cluster 10) exhibited decreased expression following BMP2 treatment. In contrast, the intermediate early responses showed a balance in up- and down-regulated genes (clusters 5, 6 and cluster 9 respectively). This indicates that the initial response to BMP2 is a transcriptional activation response. Furthermore, a time-dependent shift could be observed in the distribution of gene functions. Specifically, 65% (34 of 52) of the immediate early genes could be classified as Signal Transduction proteins, while only 12% (6 of 52) of these genes represented Structural proteins. During proceeding differentiation, less Signal Transduction related genes were regulated (17% and 28% for intermediate early and late early genes respectively), while an increasing number of Structural proteins were regulated (19% and 38% respectively). These Structural proteins include cell morphology-related genes such as the Troponins C and T and keratin complex genes, and matrix-related genes such as Procollagen IV a1 (Col4a1), Matrix gla protein, Nidogen1 and the Laminins a5 and P3. Using model-based clustering we have thus been able to distinguish different time-dependent expression profiles of the regulated genes that coincide with specific functions of the genes present in each cluster.

- 76 - Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation

Figure 1. Expression profiles of the significantly (p<0.01) regulated BMP2-responsive genes. (A) Clusters of primarily up-regulated genes, (B) Clusters of primarily down-regulated genes. Genes regulated within 24 hrs after BMP2 treatment are shown, grouped according to Model-based clustering of 10log transformed microarray data (normalized on a scale to 1). Induction levels are indicated with different color intensities in red and blue, representing up-and down-regulation respectively. Protein function profile graphs show the number of genes within each cluster with a particular function. Genes with unknown function or other functions were labeled as “Other”. EST, expressed sequence tag; TF, transcription factor; Eme, extracellular messenger; IS, intracellular signaling protein; Rec, receptor; Ema, extracellular matrix protein; CPl, cytoplasmatic protein; CSk, cytoskeletal protein. See page 139 for full color figure

- 77 - Chapter 3

ACCESSION# DESCRIPTION ACCESSION# DESCRIPTION AA793036 Vascular endothelial growth factor AI 390830 ESTs Small inducible cytokine A2 AA124623 Ecotropic viral integration site 2 Gadd45a, growth arrest and DNA-damage-inducible o AI 552167 ESTs L-myc, lung carcinoma myc related oncogene 1 AA619763 Cathepsin L Myog eni c d iff erenti ati on 1 AA738820 Phosphoserine transferase Si alyl transferase 4 A1117310 Flavin containing monooxygenase 1 Cardiac morphogenesis (Xin) AA444581 EST AA067625 Phosph ati date cytidylyl transferase AA068116 Acetyl-Coenzyme A dehydrogenase, medium chain AA386892 Dystrobrevin AA606605 Nidogen 1 AA048101 Sema 4C precursor AA217009 Tumor endothelial marker 8 (anthrax toxin receptor) AA241390 Semaphorin 3E AI607683 Semaphorin 7 A IIII AA684257 ESTs AA596866 ESTs AA879966 Laminina5 W89253 Insulin-like growth factor binding protein 5 AI325851 CD97 antigen W36474 Metallothionein 2 W53231 CD82 antigen ACCESSION#. DESCRIPTION AA105866 Glutathione S-transferase a 4 AI 552496 ESTs AI 510089 EST AI 391322 Protein kinase inhibitor a IAA725965 EST AI 552711 Creatine kinase, muscle AA798948 Caveolin, caveolae protein, 22 kDa AA771647 ESTs AA711869 Epidermal growth factor receptor pathway substrate 8 AA815689 Troponin C, cardiac/slow skeletal W34122 Bridging integrator 1 / myc box dependent interacting protein 1 AA760135 Procollagen, type IV, a 1 AA068464 Ect2 oncogene AI 465074 Mesoderm specific transcript AI323051 Growth associated protein 43 AA822257 Myosin light chain, alkali, fast skeletal muscle AA840053 Actinin alpha 2 associated LIM protein AI 894014 Cholinergic receptor, nicotinic, y-polypeptide AA617310 Zyxin Matrix y-carboxyglutamate (gla) protein W99951 Neurofilament, light polypeptide AA117547 Cyclin D1 W13004 Prostaglandin l2(prostacyclin) synthase AA792499 Ankyrin-like repeat protein AA023538 Semaphorin 6A W 16059 Glutathione S-transf erase co 1 AA413490 Tran sferrinrece ptor AA066615 Glycine amidinotransferase AA510298 Plasminogenactivator, urokinase AA067083 High mobility group protein I AA764163 Phospholipid scramblase 1 AI 326605 Cycl in-dependent kinase inhibitor 1A (p21/WAF1) A1152562 Proliferin AA771566 Troponin C, fast skeletal AA867173 ESTs AA116263 UDP-glucose ceram ide glucosyltransferase AA222190 EIG180 AA185112 ESTs AA516800 Death-associatedkinase2 AA288642 Cholinergic receptor, nicotinic, beta polypeptide 1 (muscle) A1152800 ESTs IAA681235 ESTs AA061902 Apolipoprotein B editing complex 2 AA690387 Xist gene (X inactivation transcript) AA770924 Troponin T3, skeletal, fast Calsequestrin 2 AI 326574 Four and a half LIM domains 1 AA066390 ESTs AI 549977 ESTs AI 428595 3-Phosphoglycerate dehydrogenase AA760161 ESTs AA673869 Mesothelin W76988 VPS 10 domain receptor protein SORCS 2 AA571771 X-linked nuclear protein TF | EMe IS | Rec EMa| CPI | CSk ESTs . .. Localised and 0 2 6 24 ESTs Signal Transduction structural Otnei HOURS FUNCTION PROFILE ClusterlO ACCESSION# DESCRIPTION EXPRESSION PROFILE T~| AA475186 ESTs ______W 14224 Ndr1, N-myc downstream regulated 1 ______AA162217 Pbef, pre-B-cell colony-enhancing factor U m AA437755 ESTs ------AA596753 I at exin n n n n 12 16 20 24 TF 1 EMe[ IS [ Re EMa CPI I CSk I HOURS Localised and ESTs Signal Transductior Structural |0,hel

-2.33 -1.60 -1.25 1.25 1.60 2.33 TF I EMe I IS I Rec EMa| CPI I CSk REGULATION ON MICROARRAY . . . Localised and ESTs Signal Transduction efnirhiral Otnei Figure 1, continued. See page 140 for full color figure

Validation of the microarray data We and others have shown previously that microarray data are highly comparable with results obtained using other methods such as real-time RT-PCR and Northern blotting, but that the ratio of expression regulation measured with these other methods vary probably due to the distinct hybridization conditions of the various methodologies (20, 27). To validate the expression profiles

- 78 - Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation

measured in this microarray experiment, 52 genes were tested as representatives of all 10 clusters both in original RNA material used for the microarray and in RNA material obtained from an identical but independent experiment by real-time RT-PCR. The expression profiles of 92% (48 out of 52) of the genes tested was confirmed (Table 2). Only the expression pattern of genes present in cluster 4, which were relatively weakly regulated and reached a maximum of 1.4-fold expression regulation, could not be confirmed. In conclusion, using real-time RT-PCR, we have been able to confirm both the pattern of the expression profiles as well as the level of differential expression regulation for the majority of the genes in this microarray experiment.

Table 2. Validation of the microarray data Accesion # Description Cluster Expression Accesion # Description Cluster Expression profile profile AI552130 Heyl 1 + AW210425 Slc20a1 3 + AI386062 Car3 1 + W18822 Gadd45y 3 + AA450795 CCK 1 + AA517353 Nfkbia 3 + AA667906 Cyr2 1 + AA982549 Placental Growth 3 + factor AA958693 Tcf7 1 + AA895566 Gadd45p 3 + W81878 Osf2 (fasciclin) 1 + AA036380 Snail 3 + AI510076 OASIS 2 + AA733791 Ddrl 3 + W97877 Prrx2 2 + AA162153 Mcsf 3 + AA145784 Timp3 2 + AA427234 Vegf 3 + AA833402 Similar to Ier3 2 + AA879919 Tissue factor 3 + W08321 Idl 2 + AA536836 bHLHB2 4 - AI322367 Id2 2 + AI595201 HB-EGF 4 - AA066816 Id3 2 + AA832598 EST 4 - AA733600 2 + AA271866 Pla2y 4 - AA466979 Similar to CoA 2 + AA238037 Imap38 5 + dehydrogenase AI427265 Smad6 2 + AA796818 Isg15 5 + AA545284 Smad7 2 + AA549591 Exportin l 6 + AA028420 Sprrl a 2 + AA108600 CPO 6 + AI551955 EST 2 + AA152636 Int4a 6 + AI587984 GPR97 2 + AA796822 Siat4c 7 + AA646509 Tieg 2 + AA501052 Xin 7 + AI604159 Similar to frizzled 2 + AI323202 Scya2 7 + AA415211 Musk 2 + AA437694 MyoD 7 + AI430581 Cytidine deaminase 2 + AA606331 L-myc 7 + AA197529 Lunatic fringe 2 + AA067083 Hmgal 8 + W34122 Binl 9 + W14224 Ndrl 10 + Expression profiles identified by microarray analysis were tested for confirmation by real-time RT-PCR. +, expression pattern confirmed; -, expression pattern not confirmed; Hey1, hairy/enhancer-of-split related with YRPW motif; Car3, carbonic anhydrase 3; CCK, cholecystokinin; Cyr2, cysteine-rich protein 2; Tcf7, T-cell specific transcription factor 7; Osf2 (fasciclin), fasciclin I like osteoblast specific factor 2; OASIS, old astrocyte specifically induced substance; Prrx2, paired related homeobox 2; Timp3, tissue inhibitor of metalloproteinase 3; Ier3, immediate early response 3; Id1, inhibitor of DNA binding 1; Id2, inhibitor of DNA binding 2; Id3, inhibitor of DNA binding 3; Klf4, kruppel like factor 4 (gut); CoA dehydrogenase, coenzyme A dehydrogenase; Smad6, MAD homolog 6; Smad7, MAD homolog 7; Sprr1a, small proline-rich protein 1A; EST, expressed sequence tag; GPR97, G- protein coupled receptor 97; Tieg, TGFp inducible early growth response; Musk, neural fold/somite kinase 2; Slc20a1, solute carrier family 20 member 1; Gadd45p, growth arrest and DNA damage inducible p; Ddr1, cell adhesion kinase 1; Mcsf, macrophage colony stimulating factor; Vegf, vascular endothelial growth factor; bHLHB2, basic helix-loop-helix domain containing class B2; HB-EGF, heparin binging epidermal growth factor; Pla2y, Ca2+"independent phospholipase A2 gamma; Imap38, immunity associated protein 38kDa; Isg15, ubiquitin cross reactive protein 15kDa; CPO, coproporphyrinogen oxidase; Int4a, integrin 4a; Siat4C, sialyltransferase 4C; Xin, caridac myogenesis 1; Scya2, small inducible cytokine 2; MyoD, myogenic differentiation 1; L-myc, lung carcinoma myc related oncogene 1; Hmga1, high mobility group protein 1; Bin1, bridging integrator 1; Ndr1, N-myc downstream regulated 1.

- 79 - Chapter 3

Transcriptional regulators of osteoblast differentiation To identify novel potential regulators of osteoblast differentiation, we primarily focused on the transcription factors with altered expression during differentiation. In the complete set of 184 regulated sequences, 17 transcription factors were identified (Table 3). The real-time RT-PCR analysis described above, confirmed the BMP2-induced differential expression for 15 out of the 17 transcription factors. It should be noted that on the microarray, sequences uniquely representing the bone-specific Cbfa1 isoform (Type II) were not present. Hence, type II Cbfa1 was not included in our microarray data set.

Table 3. Gene expression characteristics of osteoblast differentiation related transcription factors Accession Description Type Cluster Category Expression Cell type CHX # # profile specificity dependent

Cbfal, type IIa late early osteoblast yes

AI552130 Heyl Helix-loop-helix l late early + osteoblast partly AA958693 Tcf7 HMG box l late early + osteoblast yes

AI510076 OASIS Leucine zipper 2 immediate early + no no W97877 Prrx2 Homeo domain 2 immediate early + no no W08321 Idl Helix-loop-helix 2 immediate early + no no AI322367 Id2 Helix-loop-helix 2 immediate early + no no AA066816 Id3 Helix-loop-helix 2 immediate early + no no AI427265 Smad6 MH domain 2 immediate early + no no AA545284 Smad7 MH domain 2 immediate early + no no AA646509 Tieg Zinc finger 2 immediate early + no no AA036380 Snal Zinc finger 3 immediate early + no no

AI549669 TSC22 Leucine zipper 2 immediate early - n.d. n.d. AA536836 bHLHB2 Helix-loop-helix 4 immediate early - n.d. n.d. AA606331 L-myc Helix-loop-helix + 7 immediate early + myoblast n.d. Leucine zipper AA437694 MyoD Helix-loop-helix 7 immediate early + myoblast n.d. AA733600 Klf4 Zinc finger 2 immediate early +b no n.d. AA067083 Hmgal HMG box 8 late early +b no n.d. Data on the 17 transcription factors modulated by BMP in the microarray experiments were compared with type II Cbfa1. Type of transcription factor, cluster number and cluster category are listed. Expression profiles identified by microarray analysis were verified by real-time RT-PCR, +: expression pattern confirmed, -: expression pattern not confirmed. Cell type specific expression regulation (Figure 2) and dependence on de novo protein synthesis indicated. For the de novo protein synthesis experiment, cells were pretreated for half an hour with DMEM containing 5% NCS in the absence or the presence of cyclohexamide (CHX, 50^M) after which cells were stimulated with BMP2 (300 ng/ml) in the absence or presence of cyclohexamide. Expression regulation was analyzed by real-time RT-PCR. Abbreviations used: n.d., not done; Cbfa1, type II, 1, type II isoforms; Hey1, hairy/enhancer-of-split related with YRPW motif 1; Tcf7, T-cell specific transcription factor 7; OASIS, old astrocyte specifically induced substance; Prrx2, paired related homeobox 2; Id1, inhibitor of DNA binding 1; Id2, inhibitor of DNA binding 2; Id3, inhibitor of DNA binding 3; Smad6, MAD homologue 6; Smad7, MAD homologue 7; Tieg, TGFp inducible early growth response; Sna1, snail homologue; TSC22, TGFp stimulated protein 22; bHLHB2, basic helix-loop-helix domain containing class B2; L-myc, lung carcinoma myc related oncogene 1; MyoD, myogenic differentiation 1; Klf4, kruppel like factor 4 (gut); Hmga1, high mobility group protein 1; a, not identified in the microarray experiment, but used as a positive control; b, very low levels of expression regulation.

Transcription factors that are specifically expressed in osteoblasts but not in other cell types are likely to be regulators of osteoblast differentiation. We analyzed the cell type specificity for 13 transcription factors that showed a robust response to BMP2 treatment compared to control medium (Table 3). To this end, gene expression regulation by BMP2 was assessed in mouse C2C12 and C3H10T1/2 mesenchymal progenitor cell lines, which both can differentiate into osteoblasts, and for

- 80 - Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation

comparison, in NIH3T3 fibroblasts, which can not differentiate into osteoblasts (Figure 2). To test the validity of this approach, expression regulation of the bone-specific isoform of Cbfa1 (Cbfa1, type II) was analyzed by real-time RT-PCR in all three cell types and compared to expression regulation of sequences correlating with all isoforms (Cbfa1, bipartite sequences). Figure 2A shows a 5-fold BMP- specific induction for Cbfa1 type II in the osteoblastic cell types after 24 hrs, while no expression regulation was detected in the NIH3T3 fibroblasts. In contrast, the Cbfa1 bipartite sequences displayed low levels of BMP2-induced expression in the C2C12 cells, while no differential expression was observed in the C3H10T1/2 or NIH3T3 cells. These data show that this approach enables us to discriminate between osteoblast-restricted and common BMP responses.

Figure 2. BMP-induced expression profiles in osteoblastic and non-osteoblastic cell lines. Expression profiles of the transcription factors in C2C12, C3H10T1/2 and NIH3T3 cells. (A) Expression profiles of Cbfal isoforms, (B) Expression profiles of the 13 transcription factors. Cells were seeded and grown for 24 hrs upon which the medium was replaced by DMEM containing 5% NCS in the presence or absence of BMP2 (300 ng/ml). RNA was isolated 0 hr, 2 hrs, 6 hrs and 24 hrs after stimulation. Subsequently, expression levels were measured by quantitative real time RT-PCR.

Subsequently, the expression regulation of the l3 transcription factors was tested for cell type specificity (Figure 2B and Table 3). Nine of these transcription factors were regulated in response to BMP2 treatment in all three cell lines to approximately the same extent and will further be referred to as “general BMP-response transcription factors”. The genes L-myc and MyoD were only expressed in C2Cl2 cells. C2Cl2 cells differentiate into myoblasts under low serum conditions whereas C3Hl0Tl/2 cells do not, therefore these transcription factors could be characteristic for the pre- myoblastic phenotype of C2Cl2 cells. Finally, the transcription factors Heyl and Tcf7 showed specific BMP2-induced up-regulation in both C2Cl2 and C3Hl0Tl/2 cells, similar to the osteoblast

- 8l - Chapter 3

specific transcription factor Cbfa1 type II, but were not or very weakly regulated in NIH3T3 cells. These results suggest that the BMP2-induced regulation of Hey1 and Tcf7 is osteoblast specific. To be able to distinguish between direct and indirect effects in response to BMP2, we analyzed gene expression in the presence or absence of cyclohexamide (Chx), an inhibitor of de novo protein synthesis. The data summarized in Table 3 shows that the immediate early general BMP- response transcription factors were not dependent on protein synthesis for their BMP2-induced gene expression, at least at the 2 hr and 6 hr time points. Messenger RNA of some of these genes was actually superinduced in the presence of BMP2 and Chx, probably as a result of an increase in mRNA stability or loss of transcriptional repressors by Chx as previously reported for Smad7 (28). Furthermore, the late early genes Tcf7 and Cbfa1 type II were fully impaired in their BMP2-induced expression in the presence of cyclohexamide. Induction of Hey1 was severely repressed due to Chx treatment, but its expression was still induced over 13 fold at 24 hrs in the presence of the protein synthesis inhibitor, which suggests that regulation of Hey1 might be both direct and indirect. These results indicate that the classification of the genes based on our clustering method correlates with de novo protein synthesis-dependent transcription regulation.

Expression of Hey1 and Tcf7 during mouse development Results obtained from the above described experiments, indicate that Hey1 and Tcf7 could be potential novel regulators of osteoblast differentiation. To study the involvement of both genes in embryonic bone development, expression of these genes was analyzed in mouse embryos at E14.5 and E16.5 of gestation and in neonatal mouse tails by the use of in situ hybridization. Neonatal mouse tails were used because their primary centers of endochondral ossification are still developing in the most distal vertebrae. These vertebrae consist of alkaline phosphatase-negative cartilage whereas the most proximal vertebrae are alkaline phosphatase-positive and mineralized (22). The expression patterns of Tcf7 and Hey1 were compared with that of Cbfa1, which has previously been shown to be expressed during early skeletal development, in cells of mesenchymal condensations, and in the osteoblast lineage. In later development it is expressed in ossification centers of all bones, independent of their embryonic origin and mechanism of ossification (intramembraneous or endochondral), but not in hypertrophic chondrocytes (4). We first confirmed the expression of Cbfa1 at E14.5 in bone in transverse sections of the head at the level of the jaws (Figure 3A), at E16.5 in sagital sections of the head at the level of the nasal cavities (Figure 3D) and in proximal vertebrae of the mouse neonatal tail (Figure 3G). The ossification centers of the developing mandibular and maxillary bones, nasal bone, clavicle, humurus, and scapula as well as proliferating chondroblasts, periosteal osteoblasts and trabecular osteoblasts all showed clear expression of Cbfa1, as expected (4). In agreement with previous data, high levels of Cbfa1 expression were also observed in the mandibular region/dental mesenchyme surrounding the developing tooth (19). The Tcf7 gene showed an expression pattern remarkably similar to that of Cbfa1 in embryos at E14.5 and E16.5 of gestation (Figures 3B and 3E). Except for the C1 vertebra, expression of Tcf7 was observed in all bones analyzed, although restricted to the periosteum/compacted mesenchyme of these bones. In the mouse neonatal tail vertebrae, both the periosteal and trabecular osteoblasts showed high levels of Tcf7 expression similar to Cbfa1, but no expression of Tcf7 was observed in proliferating chondroblasts (Figure 3H). With respect to the Hey1 gene, no expression could be detected in the clavicle, humurus, scapula or C1 vertebra. However, low but significant levels of expression were observed in the mandibular and maxillary bones (Figures 3C and 3F). Specifically, Hey1 expression was restricted to the central parts of the bones (trabecular bone) and was absent in the periosteum. In the newborn tail vertebrae, the Hey1 gene showed specific expression in trabecular osteoblasts, albeit somewhat weaker than the Cbfa1 and Tcf7 genes (Figure 3I). These data show that

- 82 - Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation

the two osteoblastic transcription factors identified in this study are expressed during in vivo bone differentiation, with individual expression patterns similar but not equal to Cbfa1.

Cbfal Tcf7 Hey1

E14.5

E16.5

newborn tail vertebrae

Figure 3. In vivo expression of Hey1 and Tcf7 in the developing mouse skeleton. (A-C) Transverse sections of E14.5 mouse embryos at the level of the jaws for the control osteoblast marker Cbfa1 (A), Tcf7 (B) and Hey1 (C), (D-F) Sagital sections of E16.5 mouse embryos at the level of the nasal cavities for Cbfa1 (D), Tcf7 (E) and Hey1 (F), (G-I) Mouse proximal tail vertebrae for Cbfa1(G), Tcf7 (H) and Hey1 (I). Abbreviations used: bm, bone marrow; C1v, cartilage primordium of C1 vertebra; cl, clavicle; h, cartilage primordium of humurus; hc, hypertrophic chondroblasts; id, intervertebral disc; Jo, Jacobson’s organ; M, Meckel’s cartilage; md, mandible; mx, maxilla; nc, nasal cavity; nca, nasal capsule; P, periosteal bone; pc, proliferating chondroblasts; sc, scapula; T, trabecular bone; tg, tongue; to, tooth; vb, vertebral body. See page 141 for full color figure

Tcf7 differentially affects BMP2-induced osteoblast differentiation To further investigate the role of Tcf7 and Hey1 expression during osteoblast development, we introduced expression vectors for their cognate genes into C2C12. We identified cells expressing high levels of Tcf7 (Figure 4A), but were unable to isolate clones over-expressing Hey1 (results not shown). The differentiation capacity of Tcf7-overexpressing cells (C2C12-Tcf7) was assessed by measuring ALP activity, as well as the expression of the myogenic differentiation markers Myogenin and MyoD and of the osteogenic differentiation markers ALP, Wnt inhibitory factor 1 (Wif1) (19), Cbfa1 and OCN following BMP2 treatment. The data represented in Figures 4B and 4C show that the C2C12-Tcf7 cells were still able to differentiate to the osteoblast lineage as evidenced by the induction of ALP activity and transcripts as well as the induction of the late phase differentiation marker Wif1. The levels of these markers were slightly elevated in the C2C12-Tcf7 cells compared to the control cells. In contrast, the expression of Cbfa1 was slightly lower in C2C12-Tcf7 cells whereas the expression of OCN was completely abrogated. These data suggest that Tcf7 differentially affects osteoblast differentiation.

- 83 - Chapter 3

Figure 4. BMP2-induced osteoblast differentiation in C2C12-Tcf7 cells. (A) Expression of Tcf7 in C2C12-Tcf7 cells. Expression levels were measured by quantitative real-time RT-PCR. (B) Dose-dependent induction of ALP activity 48 hrs after BMP2-treatment, (■) C2C12 cells, (•) C2C12-Tcf7 cells. (C) BMP2-induced expression profiles. mRNA was used to measure the expression of two myogenic and four osteoblast marker genes at the indicated time points. Mean expression levels were corrected for GAPDH and are expressed relative to the basal expression level in unstimulated C2C12 cells. Error bars indicate differences between duplicate experiments. (D) Tcf7 represses Cbfa1- mediated activation of the osteocalcin promoter in C2C12 cells. Cells were transfected with pRL-SV40 (50 ng) and p6OSE2 (100 ng). Expression vectors pCbfa1-3.1 (50 ng) and/or pTcf7-3.1 (100 ng) were added as indicated. Reactions were supplemented with pcDNA3.1 vector DNA to a total of 500 ng DNA in each reaction. Luciferase activation levels were corrected for renilla levels and are represented as fold activation over control. The luciferase values represent the means of triplicate samples ± S.E.

In control C2C12 cells, the expression of the myogenic differentiation markers Myogenin and MyoD was induced during conditions that favor myoblast differentiation and repressed upon BMP2 treatment. In contrast, the C2C12-Tcf7 cells failed to induce Myogenin and MyoD transcription under myogenic differentiation conditions, indicating that differentiation was impaired. The abrogation of OCN expression in the C2C12-Tcf7 cells could be a result of the slightly reduced Cbfa1 levels in these cells or from a direct action of Tcf7 on OCN transcription. The latter notion is in line with published data that show that Lymphoid enhancer binding factor 1 (Lef1), a functional homologue of Tcf7, represses Cbfa1-dependent transcription from the OCN gene promoter (29). The repression of Lef1 is mediated through direct interaction with Cbfa1 and could be visualized on p6OSE2, which contains an artificial promoter consisting of 6 Cbfa1 recognition sites upstream of the basal osteocalcin promoter (-34/+13), and a luciferase reporter gene. To investigate whether a similar mechanism could play a role in our system, we co-transfected combinations of Tcf7, Cbfa1 and p6OSE2 into C2C12 cells. As expected, Cbfa1 but not Tcf7 induced activity of the luciferase reporter (Figure 4D). Interestingly, co-expression of Tcf7 significantly reduced the Cbfa1-induced activity of the reporter. The SV40 promoter driven expression of the renilla-reporter gene that was co-transfected to monitor differences in transfection efficiency was not affected by Tcf7 (results not shown), indicating that the observed effect is specific for Cbfa1 driven transcription.

- 84 - Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation

Discussion Genetic cascades regulated during the initial phase of differentiation guide mesenchymal stem cells to become osteoblasts, and activating and repressing transcription factors are very important in directing this differentiation process. In our data set of 184 regulated genes and ESTs, we observed an immediate up-regulation of transcription factors and other Signal Transduction proteins (2 hrs), but a slight delay in the repression of genes such as those related to myoblast differentiation (6 hrs). The near absence of repression of gene expression 2 hrs after BMP treatment observed in this study might be a reflection of the observation that Smads are transcriptional activators that first need to activate transcriptional repressors to inhibit gene expression (30). In accordance with this, a number of transcription factors have been identified among the immediate early genes, such as the inhibitor of differentiation (Id) genes (Id1, Id2 and Id3) and the Transforming growth factor pi-induced transcript 4 (TSC-22) gene, that can function as transcriptional repressors (31, 32). The immediate early transcription factors may subsequently regulate the expression of genes involved in remodeling of cell architecture (between 6 and 24 hrs). However, commitment to the osteogenic lineage was not apparent until 24 hrs, when the osteoblast specific transcription factor Cbfai and the novel osteoblast- related genes Hey1 and Tcf7 were found to be regulated. Previously, we have shown that the onset of regulation of bone markers such as ALP and OCN starts around 48 hrs of BMP treatment (19). These results are slightly different from data reported recently by Balint and coworkers who showed both activation and repression of differentiation-related genes within the first hours of osteoblast differentiation, followed by commitment and establishment of the bone phenotype in C2C12 cells within 24 hrs of BMP treatment (33). It should be noted here, that there are substantial differences in the experimental design, such as the cell density at the onset of BMP treatment and the absence of serum in their experiments that might explain the observed differences. Our study identified 15 transcription factors that are induced within the first 24 hours of BMP2 induced osteoblast differentiation. We investigated whether additional osteoblast related transcription factors might be identified among the 3 immediate early ESTs identified in this study. However, in silico sequence analysis did not reveal specific DNA binding domains or characteristic transcription activating domains, suggesting that these ESTs are not novel transcription factors (results not shown). Of the 15 known transcription factors in our dataset, 9 immediate early general BMP-response transcription factors did not display osteoblast specific expression regulation, but were specifically induced by BMP2 and therefore might have a function in the regulation of osteoblast differentiation. Indeed, 8 out of these 9 transcription factors have previously been reported to be involved in osteoblast differentiation or function. OASIS has been described to be expressed in pre­ osteoblasts in vivo and its expression pattern coincides with that of Osteopontin (34). Prx2 cooperates with Prx1 in regulating skeletogenesis in the craniofacial region, inner ear and limbs, as was shown in knock-out experiments (35). Transient induction of the Id genes by BMP has previously been described in osteoblastic cells (24, 36-38) and Id1 was found to be able to inhibit OCN gene expression (39). Furthermore, Smad6 and Smad7 are negative regulators of the BMP signal transduction pathway capable of repressing BMP-induced osteoblast differentiation (40, 41). Tieg was first identified in human osteoblastic cells (42) and is able to induce ALP expression and inhibit OCN expression in human osteoblasts similar to TGF-P (43). Until now, direct involvement of Sna1 in osteoblast differentiation has not been reported. However, Sna1 has been described to be an inhibitor of E-cadherin expression in different cell types (44-46), and E-cadherin itself has been reported to be essential for optimal BMP-induced osteoblast differentiation (47), suggesting that Sna1 might be indirectly involved in osteoblast differentiation. Although bone is a highly specialized tissue, osteoblasts have previously been described as sophisticated fibroblasts, exhibiting only a few osteoblast specific transcription factors and marker

- 85 - Chapter 3

genes, such as Cbfal, Osx and OCN (3). It could be postulated that to establish a highly specialized tissue such as bone, the action of several specific key regulatory transcription factors might not be required. Instead, a limited set of osteoblast-specific transcription factors in combination with general BMP-responsive transcription factors might be sufficient for guiding a mesenchymal stem cell to become a mature osteoblast. In our experiments, BMP2-induced transcription of Heyl and Tcf7 was restricted to osteoblastic cells, suggesting that these factors may contribute to further specifying osteoblast fate delineated by bone specific transcription factors like Cbfal and Osx. This notion is further substantiated by our observation that both Heyl and Tcf7 are expressed in developing bone tissues in vivo. Heyl belongs to the Hairy and enhancer of split (Hesl) family of transcription factors and is a primary target of Notch (48), which has been reported to inhibit osteoblast differentiation (49, 50). Heyl functions as an inhibitor of myoblast differentiation in C2Cl2 cells (5l) and is able to associate to its family member Hesl and form stable heterodimers with Hesl upon DNA binding (48). Hesl has been reported to be a negative regulator of osteoblast differentiation (52), but it can also stimulate the trans-activating function of the osteoblast specific transcription factor Cbfal (9). To the best of our knowledge Heyl knock-out mice have not been reported and Heyl has not previously been reported to be involved in osteoblast differentiation. Our in vitro data indicate that Heyl expression by BMP2 is restricted to osteoblasts and not observed in fibroblasts. Furthermore, Heyl is expressed during in vivo bone formation with an expression pattern that partially overlaps with that of Cbfal and Tcf7. We speculate that Heyl might affect osteoblast differentiation in a comparable manner to or in association with Hesl. This putative mechanism is the subject of our current studies. Tcf7 is a member of the T lymphocyte-specific enhancer / Lymphoid enhancer binding factor (Tcf/Lef) family and an effector of Wnt signaling (53). The canonical Wnt pathway has been implicated in the control of bone mass and osteoblast differentiation. Inactivating mutations in the Wnt co-receptor LRP5 decrease bone accrual during growth whereas activating mutations lead to a high bone mass phenotype (54), and stabilized P-catenin in osteoblasts enhances their differentiation capacity (55, 56). However, the downstream effectors of Wnt/p-catenin in osteoblasts remain to be established. Our data indicate that Tcf7 is a BMP2 target gene in osteoblasts but not in fibroblasts. In vivo, expression of Tcf7 has been reported during embryonic bone development up to embryonic day (E)l4.5 (57). Our data confirm and extend these results and show that Tcf7 expression at El4.5 and El6.5 of gestation is highly but not completely similar to Cbfal expression. In the developing mouse tail, Tcf7 expression is limited to osteoblasts and, unlike Cbfal, not seen in hypertrophic chondrocytes. It is tempting to speculate that Tcf7 plays a role as a downstream effector of BMP2/Wnt signaling and provides a molecular access point where these two signaling pathways integrate. In C2Cl2 cells that over-express Tcf7, we observed a down-regulation of Myogenin and MyoD similar to the down-regulation mediated by Cbfal (58), suggesting a role of Tcf7 in suppression of myogenic differentiation. In addition, we observed a BMP2-dependent differential effect on expression of genes that mark osteoblast differentiation. Our observation that Tcf7 suppresses Cbfal-induced transcriptional activity of the p6OSE2 reporter is in perfect agreement with published work that showed that the Tcf7 family member Lefl exerts an identical effect on Cbfal- mediated transcription (29). The exact mechanisms by which Tcf7 modulates osteoblast and/or myogenic differentiation are not clear and the subject of our current investigations. In conclusion, our results show that BMP-induced osteoblast differentiation employs a cascade in which the initial phase of differentiation is based on general BMP-response genes. During the progression of osteoblast differentiation, osteoblast commitment genes and osteoblast specific genes are regulated in concert with general BMP-response genes, which ultimately leads to a functional osteoblast. In addition, Heyl and Tcf7 have been identified with similar expression

- 86 - Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation

characteristics as the osteoblast specific transcription factor Cbfa1, and Tcf7 was shown to modulate BMP2-induced osteoblast differentiation. Further elucidation of the function of these transcription factors in osteoblast differentiation may contribute to drug target research to treat bone diseases such as osteoporosis.

Acknowledgements We thank Wyeth (Andover) for kindly providing recombinant human BMP-2. We are grateful to Dr. H. Clevers (Hubrecht Laboratory, The Netherlands) for providing the Tcf7 pCDNAl construct, to Dr. H.B.J Karperien (Department of Endocrinology, Leiden University, The Netherlands) for providing the mouse neonatal tails and to Dr. Patricia Ducy (Baylor College of Medicine, Houston) for providing the Cbfal cDNA and p6OSE2 luciferase promoter constructs. Furthermore, we would like to thank L. van der Meer for technical assistance and Dr. Ester Piek for support and fruitful discussion on this manuscript. This work was supported by the Netherlands Institute for Earth and Life Sciences (NWO-ALW) grant 809.67.023 (to D.S.J.) and grant 809.67.024 (to A.F.).

References 1. D. J. Prockop (1997), Marrow stromal cells as stem cells for nonhematopoietic tissues, Science 276: 71 -74. 2. A. Yamaguchi (1995), Regulation of differentiation pathway of skeletal mesenchymal cells in cell lines by transforming growth factor-beta superfamily, Semin Cell Biol 6: 165-173. 3. P. Ducy, T. Schinke, and G. Karsenty (2000), The osteoblast: a sophisticated fibroblast under central surveillance, Science 289: 1501-1504. 4. P. Ducy, R. Zhang, V. Geoffroy, A. L. Ridall, and G. Karsenty (1997), Osf2/Cbfa1: a transcriptional activator of osteoblast differentiation, Cell 89: 747-754. 5. C. Banerjee, L. R. McCabe, J. Y. Choi, S. W. Hiebert, J. L. Stein, G. S. Stein, and J. B. Lian (1997), Runt homology domain proteins in osteoblast differentiation: AML3/CBFA1 is a major component of a bone-specific complex, J Cell Biochem 66: 1-8. 6. T. Komori, H. Yagi, S. Nomura, A. Yamaguchi, K. Sasaki, K. Deguchi, Y. Shimizu, R. T. Bronson, Y. H. Gao, M. Inada, M. Sato, R. Okamoto, Y. Kitamura, S. Yoshiki, and T. Kishimoto (1997), Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts, Cell 89: 755-764. 7. C. Banerjee, A. Javed, J. Y. Choi, J. Green, V. Rosen, A. J. van Wijnen, J. L. Stein, J. B. Lian, and G. S. Stein (2001), Differential regulation of the two principal Runx2/Cbfa1 n-terminal isoforms in response to bone morphogenetic protein-2 during development of the osteoblast phenotype, Endocrinology 142: 4026-4039. 8. H. Drissi, Q. Luc, R. Shakoori, S. Chuva De Sousa Lopes, J. Y. Choi, A. Terry, M. Hu, S. Jones, J. C. Neil, J. B. Lian, J. L. Stein, A. J. Van Wijnen, and G. S. Stein (2000), Transcriptional autoregulation of the bone related CBFA1/RUNX2 gene, J Cell Physiol 184: 341-350. 9. K. W. McLarren, R. Lo, D. Grbavec, K. Thirunavukkarasu, G. Karsenty, and S. Stifani (2000), The mammalian basic helix loop helix protein HES-1 binds to and modulates the transactivating function of the runt-related factor Cbfa1, J Biol Chem 275: 530-538. 10. T. Alliston, L. Choy, P. Ducy, G. Karsenty, and R. Derynck (2001), TGF-beta-induced repression of CBFA1 by Smad3 decreases cbfa1 and osteocalcin expression and inhibits osteoblast differentiation, EMBO J 20: 2254-2272. 11. N. Selvamurugan, M. R. Pulumati, D. R. Tyson, and N. C. Partridge (2000), Parathyroid hormone regulation of the rat collagenase-3 promoter by protein kinase A-dependent transactivation of core binding factor alpha1, J Biol Chem 275: 5037-5042. 12. G. Xiao, D. Jiang, P. Thomas, M. D. Benson, K. Guan, G. Karsenty, and R. T. Franceschi (2000), MAPK pathways activate and phosphorylate the osteoblast-specific transcription factor, Cbfa1, J Biol Chem 275: 4453-4459. 13. B. Kern, J. Shen, M. Starbuck, and G. Karsenty (2001), Cbfa1 contributes to the osteoblast-specific expression of type I collagen genes, J Biol Chem 276: 7101-7107. 14. A. Javed, G. L. Barnes, B. O. Jasanya, J. L. Stein, L. Gerstenfeld, J. B. Lian, and G. S. Stein (2001), runt homology domain transcription factors (Runx, Cbfa, and AML) mediate repression of the bone sialoprotein promoter: evidence for promoter context-dependent activity of Cbfa proteins, Mol Cell Biol 21: 2891-2905. 15. H. Harada, S. Tagashira, M. Fujiwara, S. Ogawa, T. Katsumata, A. Yamaguchi, T. Komori, and M. Nakatsuka (1999), Cbfa1 isoforms exert functional differences in osteoblast differentiation, J Biol Chem 274: 6972-6978. 16. K. Nakashima, X. Zhou, G. Kunkel, Z. Zhang, J. M. Deng, R. R. Behringer, and B. de Crombrugghe (2002), The novel zinc finger-containing transcription factor osterix is required for osteoblast differentiation and bone formation, Cell 108: 17-29. 17. J. M. Wozney, V. Rosen, A. J. Celeste, L. M. Mitsock, M. J. Whitters, R. W. Kriz, R. M. Hewick, and E. A. Wang (1988), Novel regulators of bone formation: molecular clones and activities, Science 242: 1528-1534.

- 87 - Chapter 3

18. B. L. Hogan (1996), Bone morphogenetic proteins: multifunctional regulators of vertebrate development, Genes Dev 10: 1580-1594. 19. B. L. Vaes, K. J. Dechering, A. Feijen, J. M. Hendriks, C. Lefevre, C. L. Mummery, W. Olijve, E. J. van Zoelen, and W. T. Steegenga (2002), Comprehensive microarray analysis of bone morphogenetic protein 2-induced osteoblast differentiation resulting in the identification of novel markers for bone development, J Bone Miner Res 17: 2106-2118. 20. D. S. de Jong, E. J. van Zoelen, S. Bauerschmidt, W. Olijve, and W. T. Steegenga (2002), Microarray analysis of bone morphogenetic protein, transforming growth factor beta, and activin early response genes during osteoblastic cell differentiation, J Bone Miner Res 17: 2119-2129. 21. C. Fraley, and A. E. Raftery (2002), Model-based clustering, discriminant analysis and density estimation, Journal of the American Statistical Association 97: 611-631. 22. M. M. Deckers, E. R. Van Beek, G. Van Der Pluijm, A. Wetterwald, L. Van Der Wee-Pals, M. G. Cecchini, S. E. Papapoulos, and C. W. Lowik (2002), Dissociation of angiogenesis and osteoclastogenesis during endochondral bone formation in neonatal mice, J Bone Miner Res 17: 998-1007. 23. A. Feijen, M. J. Goumans, and A. J. van den Eijnden-van Raaij (1994), Expression of activin subunits, activin receptors and follistatin in postimplantation mouse embryos suggests specific developmental functions for different activins, Development 120: 3621-3637. 24. T. Katagiri, A. Yamaguchi, M. Komaki, E. Abe, N. Takahashi, T. Ikeda, V. Rosen, J. M. Wozney, A. Fujisawa- Sehara, and T. Suda (1994), Bone morphogenetic protein-2 converts the differentiation pathway of C2C12 myoblasts into the osteoblast lineage, J Cell Biol 127: 1755-1766. 25. K. Y. Yeung, C. Fraley, A. Murua, A. E. Raftery, and W. L. Ruzzo (2001), Model-based clustering and data transformations for gene expression data, Bioinformatics 17: 977-987. 26. D. Ghosh, and A. M. Chinnaiyan (2002), Mixture modelling of gene expression data from microarray experiments, Bioinformatics 18: 275-286. 27. D. Shiffman, T. Mikita, J. T. Tai, D. P. Wade, J. G. Porter, J. J. Seilhamer, R. Somogyi, S. Liang, and R. M. Lawn (2000), Large scale gene expression analysis of cholesterol-loaded macrophages, J Biol Chem 275: 37324-37332. 28. A. Nakao, M. Afrakhte, A. Moren, T. Nakayama, J. L. Christian, R. Heuchel, S. Itoh, M. Kawabata, N. E. Heldin, C. H. Heldin, and P. ten Dijke (1997), Identification of Smad7, a TGFbeta-inducible antagonist of TGF-beta signalling, Nature 389: 631-635. 29. R. A. Kahler, and J. J. Westendorf (2003), Lymphoid enhancer factor-1 and beta-catenin inhibit Runx2-dependent transcriptional activation of the osteocalcin promoter, J Biol Chem 278: 11937-11944. 30. F. Liu, A. Hata, J. C. Baker, J. Doody, J. Carcamo, R. M. Harland, and J. Massague (1996), A human Mad protein acting as a BMP-regulated transcriptional activator, Nature 381: 620-623. 31. Y. Jen, H. Weintraub, and R. Benezra (1992), Overexpression of Id protein inhibits the muscle differentiation program: in vivo association of Id with E2A proteins, Genes Dev 6: 1466-1479. 32. H. A. Kester, C. Blanchetot, J. den Hertog, P. T. van der Saag, and B. van der Burg (1999), Transforming growth factor-beta-stimulated clone-22 is a member of a family of leucine zipper proteins that can homo- and heterodimerize and has transcriptional repressor activity, J Biol Chem 274: 27439-27447. 33. E. Balint, D. Lapointe, H. Drissi, C. Van Der Meijden, D. W. Young, A. J. Van Wijnen, J. L. Stein, G. S. Stein, and J. B. Lian (2003), Phenotype discovery by gene expression profiling: Mapping of biological processes linked to BMP-2- mediated osteoblast differentiation, J Cell Biochem 89: 401-426. 34. T. Nikaido, S. Yokoya, T. Mori, S. Hagino, K. Iseki, Y. Zhang, M. Takeuchi, H. Takaki, S. Kikuchi, and A. Wanaka (2001), Expression of the novel transcription factor OASIS, which belongs to the CREB/ATF family, in mouse embryo with special reference to bone development, Histochem Cell Biol 116: 141-148. 35. D. ten Berge, A. Brouwer, J. Korving, J. F. Martin, and F. Meijlink (1998), Prx1 and Prx2 in skeletogenesis: roles in the craniofacial region, inner ear and limbs, Development 125: 3831-3842. 36. A. Hollnagel, V. Oehlmann, J. Heymer, U. Ruther, and A. Nordheim (1999), Id genes are direct targets of bone morphogenetic protein induction in embryonic stem cells, J Biol Chem 274: 19838-19845. 37. O. Korchynskyi, and P. ten Dijke (2002), Identification and functional characterization of distinct critically important bone morphogenetic protein-specific response elements in the Id1 promoter, J Biol Chem 277: 4883-4891. 38. T. Ogata, J. M. Wozney, R. Benezra, and M. Noda (1993), Bone morphogenetic protein 2 transiently enhances expression of a gene, Id (inhibitor of differentiation), encoding a helix-loop-helix molecule in osteoblast-like cells, Proc Natl Acad Sci USA 90: 9219-9222. 39. M. Tamura, and M. Noda (1994), Identification of a DNA sequence involved in osteoblast-specific gene expression via interaction with helix-loop-helix (HLH)-type transcription factors, J Cell Biol 126: 773-782. 40. M. Fujii, K. Takeda, T. Imamura, H. Aoki, T. K. Sampath, S. Enomoto, M. Kawabata, M. Kato, H. Ichijo, and K. Miyazono (1999), Roles of bone morphogenetic protein type I receptors and Smad proteins in osteoblast and chondroblast differentiation, M ol Biol Cell 10: 3801-3813. 41. U. Valcourt, J. Gouttenoire, A. Moustakas, D. Herbage, and F. Mallein-Gerin (2002), Functions of transforming growth factor-beta family type I receptors and Smad proteins in the hypertrophic maturation and osteoblastic differentiation of chondrocytes, J Biol Chem 277: 33545-33558. 42. M. Subramaniam, S. A. Harris, M. J. Oursler, K. Rasmussen, B. L. Riggs, and T. C. Spelsberg (1995), Identification of a novel TGF-beta-regulated gene encoding a putative zinc finger protein in human osteoblasts, Nucleic Acids Res 23: 4907-4912. 43. T. E. Hefferan, G. G. Reinholz, D. J. Rickard, S. A. Johnsen, K. M. Waters, M. Subramaniam, and T. C. Spelsberg (2000), Overexpression of a nuclear protein, TIEG, mimics transforming growth factor-beta action in human osteoblast cells, J Biol Chem 275: 20255-20259.

- 88 - Identification of Novel Regulators Associated with Early Phase Osteoblast Differentiation

44. E. Batlle, E. Sancho, C. Franci, D. Dominguez, M. Monfar, J. Baulida, and A. Garcia De Herreros (2000), The transcription factor snail is a repressor of E-cadherin gene expression in epithelial tumour cells, Nat Cell Biol 2: 84-89. 45. E. A. Carver, R. Jiang, Y. Lan, K. F. Oram, and T. Gridley (2001), The mouse snail gene encodes a key regulator of the epithelial-mesenchymal transition, M ol Cell Biol 21: 8184-8188. 46. I. Poser, D. Dominguez, A. G. de Herreros, A. Varnai, R. Buettner, and A. K. Bosserhoff (2001), Loss of E-cadherin expression in melanoma cells involves up-regulation of the transcriptional repressor Snail, J Biol Chem 276: 24661­ 24666. 47. E. Hay, J. Lemonnier, D. Modrowski, A. Lomri, F. Lasmoles, and P. J. Marie (2000), N- and E-cadherin mediate early human calvaria osteoblast differentiation promoted by bone morphogenetic protein-2, J Cell Physiol 183: 117-128. 48. T. Iso, V. Sartorelli, C. Poizat, S. Iezzi, H. Y. Wu, G. Chung, L. Kedes, and Y. Hamamori (2001), HERP, a novel heterodimer partner of HES/E(spl) in Notch signaling, M ol Cell Biol 21: 6080-6089. 49. K. Shindo, N. Kawashima, K. Sakamoto, A. Yamaguchi, A. Umezawa, M. Takagi, K. Katsube, and H. Suda (2003), Osteogenic differentiation of the mesenchymal progenitor cells, Kusa is suppressed by Notch signaling, Exp Cell Res 290: 370-380. 50. M. Sciaudone, E. Gazzerro, L. Priest, A. M. Delany, and E. Canalis (2003), Notch 1 Impairs Osteoblastic Cell Differentiation, Endocrinology 144: 5631-5639. 51. J. Sun, C. N. Kamei, M. D. Layne, M. K. Jain, J. K. Liao, M. E. Lee, and M. T. Chin (2001), Regulation of myogenic terminal differentiation by the hairy-related transcription factor CHF2, J Biol Chem 276: 18591-18596. 52. M. Matsue, R. Kageyama, D. T. Denhardt, and M. Noda (1997), Helix-loop-helix-type transcription factor (HES-1) is expressed in osteoblastic cells, suppressed by 1,25(OH)2 vitamin D3, and modulates 1,25(OH)2 vitamin D3 enhancement of osteopontin gene expression, Bone 20: 329-334. 53. H. Clevers, and M. van de Wetering (1997), TCF/LEF factor earn their wings, Trends Genet 13: 485-489. 54. P. Babij, W. Zhao, C. Small, Y. Kharode, P. J. Yaworsky, M. L. Bouxsein, P. S. Reddy, P. V. Bodine, J. A. Robinson, B. Bhat, J. Marzolf, R. A. Moran, and F. Bex (2003), High bone mass in mice expressing a mutant LRP5 gene, J Bone Miner Res 18: 960-974. 55. G. Bain, T. Muller, X. Wang, and J. Papkoff (2003), Activated beta-catenin induces osteoblast differentiation of C3H10T1/2 cells and participates in BMP2 mediated signal transduction, Biochem Biophys Res Commun 301: 84-91. 56. G. Rawadi, B. Vayssiere, F. Dunn, R. Baron, and S. Roman-Roman (2003), BMP-2 controls alkaline phosphatase expression and osteoblast mineralization by a Wnt autocrine loop, J Bone Miner Res 18: 1842-1853. 57. M. Oosterwegel, M. van de Wetering, J. Timmerman, A. Kruisbeek, O. Destree, F. Meijlink, and H. Clevers (1993), Differential expression of the HMG box factors TCF-1 and LEF-1 during murine embryogenesis, Development 118: 439-448. 58. K. S. Lee, H. J. Kim, Q. L. Li, X. Z. Chi, C. Ueta, T. Komori, J. M. Wozney, E. G. Kim, J. Y. Choi, H. M. Ryoo, and S. C. Bae (2000), Runx2 is a common target of transforming growth factor beta1 and bone morphogenetic protein 2, and cooperation between runx2 and smad5 induces osteoblast-specific gene expression in the pluripotent mesenchymal precursor cell line C2C12, M ol Cell Biol 20: 8783-8792.

- 89 - - 90 - 4

Regulation of Notch signaling genes during BMP2-induced differentiation of osteoblast precursor cells

Biochemical and Biophysical Research Communications 2004, 320, 100-107

D. S. de Jong1, W. T. Steegenga1, 4, J. M. A. Hendriks1, E. J. J. van Zoelen2, W. Olijve1, 3, and K. J. Dechering3

1 Department of Applied Biology, University of Nijmegen, 6525 ED Nijmegen, the Netherlands 2 Department of Cell Biology, University of Nijmegen, 6525 ED Nijmegen, the Netherlands 3 Target Discovery Unit, N.V. Organon, Oss, the Netherlands 4 Current address: Division of Human Nutrition, Wageningen University, Wageningen, the Netherlands Chapter 4

Abstract The bone morphogenetic protein (BMP)-induced Smad signal transduction pathway is an important positive regulator of osteoblast differentiation. BMP and other members of the transforming growth factor P (TGF-P) family have distinct effects on osteoblast differentiation, depending on cell type and cell differentiation status. In C2C12 mesenchymal cells, BMP-induced osteoblast differentiation can be blocked by TGF-p. In a search for key regulators of osteoblast differentiation we have used microarray analysis to identify genes which are differentially regulated by BMP2 and TGF-P. Within the first 24 hours following the onset of differentiation, 61 BMP2- regulated genes were identified of which the BMP2 effect was counteracted by TGF-p. The majority of these differentially expressed transcripts are related to signal transduction. Notably, our data show that three Notch signal transduction pathway genes, Lfng, Hey1 and Hes1, are differentially regulated by BMP2 and TGF-P. This suggests that these genes might function as the focal point for interaction of Smad and Notch signaling during osteoblast differentiation.

Introduction Bone morphogenetic proteins (BMPs) are the most potent of all inducers of osteoblast differentiation. Although the BMPs were first identified for their ability to induce ectopic bone formation in vivo (1), these proteins appear to be multifunctional regulators of morphogenesis during embryonic development (2). Transforming growth factor-P (TGF-P), a protein closely related to the BMPs, is also able to regulate osteoblast differentiation. However, regulation by TGF-P is more complex, since it can both inhibit and stimulate osteoblast differentiation depending on the cell type and its differentiation status (3-5). To generate biological responses, TGF-P family members use heteromeric complexes of so- called type I and type II serine/threonine kinase receptors, resulting in the activation of specific receptor-regulated Smads (R-Smads). These include Smad-2 and -3, which are specific for TGF-P signaling, while Smad-1, -5 and -8 have been shown to be BMP responsive R-Smads. In combination with the common mediator Smad-4, the respective R-Smads are assembled into heteromeric complexes, which subsequently translocate to the nucleus, where they act in concert with other proteins as transcription factor complexes (6-8). In addition to the Smad-mediated BMP pathway, several other major signaling pathways are known to play a role during osteoblast differentiation. These include the Wnt/P-catenin pathway (9­ 12), the Notch pathway (13, 14) and the different MAP-kinase pathways (15-19). However, the interplay between these various pathways in osteoblast differentiation is still poorly understood. In the C2C12 mouse mesenchymal progenitor cell line, osteoblastic differentiation can be induced by BMPs, while TGF-P1 is not able to do so, and even inhibits BMP2-induced differentiation (20). BMP2 and TGF-p partially share the same Smad signal transduction route and have many target genes in common. Given the differential effect of TGF-p and BMP2 on C2C12 differentiation, genes that are differentially affected by the two cytokines might function as important regulators of osteoblast differentiation. Previously, we have shown that the commitment to the osteoblast phenotype is determined within the first 24 hours following stimulation with BMP2 (21). In this study, we have compared BMP2-induced gene expression levels with those of TGF-P and a combination of these two growth factors in a 24 hour time frame. Notably, several members of the Notch signaling pathway were differentially affected by BMP2 and TGF-p, suggesting that BMP- induced modulation of Notch activity may be a key element in regulating osteoblast differentiation.

- 92 - Regulation of Notch signaling genes during BMP2-induced differentiation of osteoblast precursor cells

Materials and Methods Tissue culture and growth factors The C2C12 murine mesenchymal progenitor cell line (obtained from the American Type Culture Collection) was maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% newborn calf serum (NCS), penicillin (100 U/ml) and streptomycin (100 |ig/ml) as described previously (21). For the various experiments, cells were plated at a density of 2.0x104 cells/cm2 and grown for 24 hours in DMEM supplemented with 10% NCS. Subsequently, medium was replaced by DMEM containing 5% NCS (control medium) either in the presence or in the absence of 300 ng/ml recombinant human BMP2 with or without 5 ng/ml human recombinant TGF- P1. Recombinant human BMP2 was kindly provided by Wyeth (Andover, MA) and recombinant human TGF-P1 was purchased from R&D Systems (Minneapolis, MN).

RNA isolation Total RNA was extracted according to an acid guanidium thiocyanate-phenol-chloroform method (TriPure® Isolation Reagent, Roche Diagnostics, Germany) and RNA concentrations were determined by measuring the absorbance at 260 nm. Poly A+ RNA was prepared from total RNA using the Oligotex™ method (Qiagen, Valencia, CA). Quantification of polyA+ RNA was carried out according to the Ribogreen™ RNA quantitation assay (Molecular Probes, Eugene, OR).

Gene Microarray Analysis Microarray analysis was performed as described previously (22). For the hybridization experiments, RNA from growth factor-treated cells was hybridized versus RNA from control cells using the mouse Unigene 1 array from Incyte Genomics (Palo Alto). The hybridization experiments were performed in duplicate with a fluor reversal to minimize possible bias caused by the molecular structure of the Cy3 and Cy5 dyes. Background-subtracted element signals were used to calculate Cy3/Cy5 ratios. Genes were selected for further analysis if the signal to background ratio for a particular gene was >2.5 and at least 40% of the area on chip could be used for signal computation in at least one of the two channels. To account for systematic errors caused by non-linear differences in signal intensities of the Cy3 or Cy5 channel, statistical analysis was performed as described previously (22). Genes with significant (p<0.05, >1.5 fold) differential expression in at least one of the treatments were selected for further analysis. Subsequently, the genes were subdivided into 9 categories according to their differential regulation by BMP2 (B), BMP2 plus TGF-P1 (BT), and TGF-P1 (T) as follows: Category 1. B>BT>T; Category 2. BBT=T; Category 4. BBTT; Category 7. B=BTT and Category 9. B=BT=T, with the assumption that B=BT or BT=T when the abslolute difference between induction levels was less than 0.1. Protein functions of the genes were identified according to the Gene OntologyTM Database.

Real-time RT-PCR For real-time RT-PCR, the ABI Prism Sequence Detection System 5700 and Primer ExpressTM software (PE Biosystems) were used. For each gene, a set of primers was designed using sequences obtained from Genbank™ All primers generated amplicons of 50-100 base pairs with melting temperatures between 58oC and 60oC. Prior to complementary DNA (cDNA) synthesis, 2 |ig of total RNA were DNase (Invitrogen) treated for 15 min at 37oC, after which RNA was reverse transcribed using random hexamer primers (Amersham Pharmacia) in a total volume of 50|il using the SUPERSCRIPTTM II reverse transcriptase kit (Invitrogen). Subsequently, aliquots of first strand

- 93 - Chapter 4

cDNA (2 |il) were amplified using SYBR Green® PCR Mastermix (Applied Biosystems) under the following conditions: initial denaturation for 10 min at 94oC followed by 40 cycles consisting of 15 sec at 94oC and 1 min at 60oC. Fold inductions and expression ratios were calculated from differences in threshold cycles at which an increase in reporter fluorescence above a baseline signal could first be detected (CT) between two samples, and averaged for duplicate experiments.

Results Expression profiles of genes differentially regulated by BMP2 and TGF-fi Upon BMP2 treatment, C2C12 mesenchymal progenitor cells differentiate into cuboidal- shaped osteoblastic cells, whereas they form myotubes in response to low serum conditions without BMP2 (Figure 1). TGF-P1 is unable to induce differentiation of C2C12 cells, but it inhibits BMP2- induced osteoblast differentiation, as well as myotube formation. This is reflected by morphology (Figure 1A) as well as by expression regulation of marker genes. Whereas BMP2 induces the expression of the osteoblast markers alkaline phosphatase (ALP) and osteocalcin (OCN), TGF-P inhibits this BMP2-induced expression (Figure 1B). Our previous experiments have indicated that C2C12 mesenchymal progenitor cells already become committed to osteoblast differentiation within 1 day after BMP treatment, as evidenced by the up-regulation of the osteoblast-specific transcription factor core-binding factor a1 (Cbfa1) within this time frame (21). In this study, we therefore focused on genes which are regulated within the first 24 hours of growth factor treatment to identify potential key regulators of osteoblast differentiation.

0 'J ri 6h 1d 2 d 3d 6d

TIME Figure 1. Differential activities of BMP2 and TGF-p on C2C12 cells. (A) Differential morphological features of C2C12 cells treated with BMP2, TGF-p and combinations of these growth factors. Cells were seeded and grown for 24 hrs upon which the medium was replaced by control medium (CM: DMEM containing 5% NCS), control medium + 300 ng/ml BMP2 (B), control medium + 5 ng/ml TGF-p (T) and control medium + 300 ng/ml BMP2 and 5 ng/ml TGF-p (BT). The figure shows micrographs taken 3 days after addition of growth factors. (B) Differential regulation of expression of the osteoblast marker genes alkaline phosphatase (ALP) and osteocalcin (OCN) by BMP2 and TGF-p. Cells were seeded and grown for 24 hrs upon which the medium was replaced by DMEM containing 5% NCS in the presence or absence of BMP2 (300ng/ml) with or without TGF-p (5 ng/ml). RNA was isolated 0, 2 and 6 hours, 1 day, 2 days, 3 days and 6 days after stimulation. Expression levels were measured by real-time quantitative RT-PCR. Mean expression levels were corrected for GAPDH levels. Error bars indicate standard deviations between duplicate experiments.

- 94 - Regulation of Notch signaling genes during BMP2-induced differentiation of osteoblast precursor cells

Microarray analysis of C2C12 cells 24 hours after treatment with either BMP2, TGF-P or a combination of these factors, showed 625 sequences out of the 9,330 spotted on the array to exhibit significant differential expression for at least one of the conditions tested, as based on our previously described statistical analysis (22). Subsequently, the genes were subdivided into 9 categories according to their expression pattern by BMP2 (B), BMP2 plus TGF-P (BT), and TGF-P (T) as described in Materials and Methods. Because TGF-P inhibits BMP2-induced differentiation, genes with an expression pattern reflecting categories 1 to 4 (B>BT>T, BBT=T and B

Table 1. Expression profiles of the genes differentially regulated by BMP2 and TGF-P. Expression Accession # Abbre- Description Functional Expression levels pattern viation group B BT T B>BT>T AI386062 Car3 Carbonic anhydrase 3 Enz 3.34 1.21 -1.91 AA062324 Dpysl3 Dihydropyrimidinase-like 3 ST 2.90 2.13 1.33 AI157520 Omd Osteomodulin CAd 2.54 1.58 1.00 W97877 Prrx2 Paired related homeobox 2 ST 2.34 1.72 -1.20 AA028346 Krt1-19 Keratin complex 1, acidic, gene 19 Cy 2.33 1.78 -1.21 AI427265 Madh6 MAD homolog 6 ST 2.33 1.56 1.13 AA821963 Dnpep Aspartyl aminopeptidase Enz 2.14 1.50 1.00 Hairy/enhancer-of-split related with YRPW AI552130 2.11 1.21 -1.10 Hey1 motif 1 ST AA197529 Lfng Lunatic fringe ST 2.06 1.56 1.00 AA981032 Emilin2 Elastin microfibril interfacer 2 EM 2.05 1.51 1.18 AA450795 Cck Cholecystokinin ST 2.01 1.89 -1.07 AA596753 Lxn Latexin Enz 1.90 1.23 1.00 AA545032 Ppap2b Phosphatidic acid phosphatase 2a Enz 1.77 1.21 -1.30 AI323002 Nsg1 Neuron specific family member 1 Other 1.76 1.00 -1.15 AA242413 Asgr1 Asialoglycoprotein receptor 1 ST 1.73 1.19 1.00 AI325697 Ly6a Lymphocyte antigen 6 complex Other 1.69 -1.13 -1.54 AI606539 EST 1.69 1.21 1.00 AA415211 Musk muscle, skeletal, receptor tyrosine kinase ST 1.65 1.29 -1.58 AA606242 Ecm1 Extracellular matrix protein 1 EM 1.59 1.34 -1.05 AA517588 ALP Alkaline phosphatase 2, liver Enz 1.53 1.31 1.00 W16221 Col6a1 Procollagen, type VI, alpha 1 EM 1.30 -1.61 -1.94 W18632 EST 1.24 -1.52 -1.86 AA450440 Ddef1 Development and differentiation enhancing Enz 1.00 -1.16 -1.79 AI552157 EST 1.00 -1.25 -1.58 Camk2d Calcium/calmodulin-dependent protein kinase ST AA462389 1.00 -1.40 -1.62 II delta AA067331 EST -1.02 -1.34 -1.62 Genes significantly (P<0.05, >1.5 fold) regulated for either BMP2 (B, 300 ng/ml), BMP2 + TGF-P (BT, 300 ng/ml + 5 ng/ml respectively) or TGF-P (T, 5 ng/ml) are shown for the expression pattern categories 1-4 (B>BT>T, BBT=T, and B

- 95 - Chapter 4

Table 1, continued BBT=T AA067147 Mod1 Malic enzyme, supernatant Enz 1.72 1.35 1.27 AA590599 EST 1.52 1.30 1.31 AA879919 F3 Coagulation factor III ST 1.22 -1.52 -1.58 AA518187 Ogn Osteoglycin ST 1.00 -1.62 -1.61 AA414208 Sdfr1 Stromal cell derived factor receptor 1 ST -1.18 -1.50 -1.49 AA755873 Mgrn1 Mahogunin, ring finger 1 Other -1.19 -1.43 -1.52 Snap25bp Synaptosomal-associated protein 25 binding Other W17626 -1.24 -1.66 -1.66 protein AI425448 Socs5 Supressor of cytokine signaling 5 ST -1.28 -1.56 -1.56 AA106098 Cfh Complement component factor H Other -1.30 -1.60 -1.51 AA413899 Sat1 Spermidine/spermine N1-acetyl transferase Enz -1.30 -1.61 -1.56 B

Expression of Notch pathway genes during osteoblast differentiation We subsequently focussed on signal transduction genes that had similar expression profiles as ALP, i.e. the signal transduction genes present in Category 1 (B>BT>T). As can be seen from Table 1, these genes include dihydropyrimidinase-like 3 (Dpysl3), paired related homeobox 2 (Prrx2), MAD homologue 6 (Madh6), hairy/enhancer of split with YRPW motif 1 (Hey1), lunatic fringe (Lfng) and calcium/calmodulin dependent kinase II delta (Camk2d). To validate the expression profiles measured in this microarray experiment, the expression of these 6 genes was tested by quantitative real time RT-PCR both in original RNA used for the microarray and in RNA obtained from an identical but independent experiment. Figure 2 shows that the expression profiles for 5 out of 6 genes were confirmed. Only for Camk2d the expression regulation for the various growth factor combinations was moderate and statistically not significant. Of the 61 genes and ESTs identified in our study, two belong to the Notch signal transduction pathway: the transcription factor Hey1 and the signal modulator Lfng. Previously, we have identified another transcription factor of this signal transduction pathway, hairy/enhancer o f split 1 (Hes1), to be differentially regulated between BMP2 and TGF-P (22). Because this gene was not present on our current microarray, we validated the differential effect of BMP2 and TGF-P on expression of Hes1 by quantitative real time RT-PCR in both sets of RNA. As can be seen from Figure 2, Hes-1 is down-

- 96 - Regulation of Notch signaling genes during BMP2-induced differentiation of osteoblast precursor cells

regulated by BMP2 and up-regulated by TGF-p. In addition, TGF-P inhibited the BMP2-induced down-regulation and caused an up-regulation of gene expression similar to that induced by TGF-P alone. The combined results indicate that at least three genes of the Notch signal transduction pathway are differentially modulated by BMP2 and TGF-P in C2C12 cells, which suggests that this pathway plays an important role in regulating the early phase of osteoblast differentiation.

Figure 2. BMP2 and TGF-p differentialy regulate expression of signal transduction related genes. Cells were seeded and grown for 24 hrs upon which the medium was replaced by DMEM containing 5% NCS in the presence or absence of BMP2 (300ng/ml) with or without TGF-p (5 ng/ml). RNA was isolated from untreated cells (day 0) and 1 day after stimulation. Expression levels were measured by real-time quantitative RT-PCR. Mean expression levels were corrected for GAPDH levels. Error bars indicate standard deviations between duplicate experiments. Camk2d, calcium/calmodulin dependent kinase II delta; Dpysl3, dihydropyrimidinase-like 3; Heyl, hairy/enhancer of split with YRPW motif 1; Hesl, hairy/enhancer of split 1; Lfng, Lunatic fringe; Madh6, MAD homologue 6; Prrx2, paired related homeobox 2.

Discussion Modulation of the activity of signal transduction pathways during the initial phase of differentiation guides mesenchymal stem cells to become osteoblasts, and activating and repressing transcription factors are very important in directing the differentiation process. In our data set, we observed that “Signal transduction” genes constituted over a quarter of the total set of genes regulated 24 hours after growth factor treatment. Previously, we have shown that commitment of the C2C12 cells to the osteoblastic lineage occurs around 24 hours after BMP treatment, when the osteoblast specific transcription factor Cbfal and the novel osteoblast related genes Tcf7 and Heyl become regulated (21). Genes with expression patterns that associate with the differential effects of BMP2 and TGF-P on osteoblast differentiation in C2C12 cells might be mediators of osteoblast differentiation. Therefore, in search of novel genes implied in the commitment of mesenchymal progenitor cells to osteoblast diferentiation, we have analysed the effects of BMP2 and TGF-P on gene expression at 24 hours post stimulation. Our study identified 5 signal transduction related genes that exhibited similar expression profiles as the osteoblast marker ALP after 24 hours of stimulation. Of these, 4 genes have previously been reported to be involved in osteoblast differentiation or function. Prrx2 cooperates with Prrx1 in regulating the skeletogenesis of the craniofacial region, inner ear and limbs, as was shown in knock­ out experiments (23). Madh6 is a negative regulator of the BMP signal transduction pathway capable of repressing BMP-induced osteoblast differentiation (24, 25). The function of Hey1 and Hes1 in osteoblast differentiation will be discussed below. The observation that four out of the five signal transduction genes identified here have been implicated before in osteoblast differentiation indicates that screening for genes differentially affected by BMP2 and TGF-P is a valid strategy for the

- 97 - Chapter 4

identification of novel regulators of osteoblast differentiation. The dyhydropyrimidinase like protein 3 (Dpysl-3) has not previously been implicated in osteoblast differentiation. Dpysl-3 is a member of a family of dihydropyrimidinase related proteins which have been shown in humans to be intracellular components of a signalling cascade involved in the development of the nervous system (26, 27). Furthermore, studies in which we compared the expression regulation of Dpysl-3 in osteoblastic cells and in non-osteoblastic cells showed that Dpysl-3 is specifically regulated in osteoblastic cells ((21) and data not shown). These results indicate that Dpysl-3 might also be a regulator of osteoblast differentiation. We investigated whether additional osteoblast-related transcription factors might be identified among the 4 ESTs identified in this category. However, in silico sequence analysis did not reveal specific DNA binding domains or characteristic transcription activating domains, suggesting that these ESTs are not novel transcription factors (results not shown). In this study, we identified 3 members of the Notch signalling pathway, Lfng, Heyl and Hesl, to be differentially regulated by BMP2 and TGF-p. Hey1 belongs to the Hairy and enhancer of split (HES) family of transcription factors and is a primary target of Notch (28). Previously, we have shown that Heyl is expressed in bone during mouse embryonic development with an expression pattern that partially overlaps with that of Cbfal (21). Hey1 functions as an inhibitor of myoblast differentiation in C2C12 cells (29). This finding is consistent with our data that show that Hey1 expression is induced when transdifferentiation of C2C12 cells to the osteoblast lineage is induced with BMP2. Hey1 is able to associate with its family member Hes1 and form stable heterodimers with Hes1 upon DNA binding (28). The role of Hes1 in osteoblast differentiation is unclear as it has been reported to be both an inhibitor and activator of osteoblast differentiation (30, 31). Our finding that Hes1 is downregulated by the osteogenic factor BMP2, whereas it is induced by the inhibitory factor TGF-P, would be consistent with an inhibitory role of Hes1 in osteoblast differentiation. In addition to the effect on the Notch target genes Hes1 and Hey1, we observed a differential effect of BMP2 and TGF-P on expression of Lfng. This protein is a modulator of Notch signalling because of its O-fucose-specific glycosyltransferase activity on both the Notch receptors and their ligands. However, no direct evidence has been obtained so far for a functionally relevant modification of the Notch ligands, while fucosylation of the Notch receptors occurs in a cell autonomous fashion (32-35). O-fucosylation is an essential step in Notch signalling, since loss of the protein that transfers fucose to the EGF-like repeats of Notch (O-FucT-1), causes embryonic lethality of mice at midgestation with severe defects in somitogenesis, vasculogenesis, cardiogenesis and neurogenesis. Inactivation of Lfng in mice causes fusion of somites, skeletal defects and perinatal death (36, 37). It has been suggested that Lfng changes the sugar structures on Notch and thereby causes alterations in binding of Jagged and Delta. However, the reports on the effect of Lfng on Notch signaling in osteoblast differentiation as well as the effect of the different ligands on Notch signalling in osteoblast differentiation are contradictory (13, 14, 33, 38, 39). Strikingly, in many cases during embryonic development, it is the distribution of fringe proteins, rather than of the Notch receptors or their ligands, that is the key determinant of where Notch signalling will be active and in which form (40). In our study, we have shown that Lfng and Heyl are regulated in a very similar manner by BMP2 and TGF-P, and that expression regulation of Hesl is negatively correlated with that of Lfng and Heyl. Expression of Lfng was shown to be initiated already 2 hours after BMP2 treatment, while Heyl expression was induced 24 hours after BMP treatment (21). These results suggest that expression of Lfng causes a switch in the Notch signaling pathway such that Heyl becomes up-regulated and Hesl down-regulated. In conclusion, we have shown that several members of the Notch signal transduction pathway are differentially regulated by BMP2 and TGF-P in C2C12 mouse mesenchymal progenitor cells. We hypothesize that Lfng might function as one of the the focal points for interaction of the Smad and

- 98 - Regulation of Notch signaling genes during BMP2-induced differentiation of osteoblast precursor cells

Notch signal transduction pathways during osteoblast differentiation: Lfng expression is induced by BMP2 and subsequently the altered distribution of this gene might induce changes in the activity and form of Notch signalling, which in turn could modulate BMP-induced osteoblast differentiation. Further elucidation of the function of these members of the Notch signalling pathway in osteoblast differentiation may contribute to the identification of new therapeutics for the treatment of bone diseases such as osteoporosis.

Acknowledgements We thank Wyeth (Andover) for kindly providing recombinant human BMP2. Furthermore, we thank Dr. Ester Piek for support and fruitful discussion on this manuscript. This work was supported by the Netherlands Institute for Earth and Life Sciences (NWO-ALW) grant 809.67.023 (to D.S. de Jong)

References 1. J. M. Wozney, V. Rosen, A. J. Celeste, L. M. Mitsock, M. J. Whitters, R. W. Kriz, R. M. Hewick, and E. A. Wang (1988), Novel regulators of bone formation: molecular clones and activities, Science 242: 1528-1534. 2. B. L. Hogan (1996), Bone morphogenetic proteins: multifunctional regulators of vertebrate development, Genes Dev 10: 1580-1594. 3. J. Massague (1990), The transforming growth factor-beta family, Annu Rev Cell Biol 6: 597-641. 4. R. Fujimoto, T. Tanizawa, S. Nishida, N. Yamamoto, S. Soshi, N. Endo, and H. E. Takahashi (1999), Local effects of transforming growth factor-beta1 on rat calvaria: changes depending on the dose and the injection site, J Bone Miner Metab 17: 11-17. 5. E. C. Breen, R. A. Ignotz, L. McCabe, J. L. Stein, G. S. Stein, and J. B. Lian (1994), TGF beta alters growth and differentiation related gene expression in proliferating osteoblasts in vitro, preventing development of the mature bone phenotype, J Cell Physiol 160: 323-335. 6. C. H. Heldin, K. Miyazono, and P. ten Dijke (1997), TGF-beta signalling from cell membrane to nucleus through SMAD proteins, Nature 390: 465-471. 7. J. Massague (1998), TGF-beta signal transduction, Annu Rev Biochem 67: 753-791. 8. J. Massague, and D. Wotton (2000), Transcriptional control by the TGF-beta/Smad signaling system, EMBO J 19: 1745-1754. 9. G. Rawadi, B. Vayssiere, F. Dunn, R. Baron, and S. Roman-Roman (2003), BMP-2 controls alkaline phosphatase expression and osteoblast mineralization by a Wnt autocrine loop, J Bone Miner Res 18: 1842-1853. 10. G. Bain, T. Muller, X. Wang, and J. Papkoff (2003), Activated beta-catenin induces osteoblast differentiation of C3H10T1/2 cells and participates in BMP2 mediated signal transduction, Biochem Biophys Res Commun 301: 84-91. 11. J. Kitagaki, M. Iwamoto, J. G. Liu, Y. Tamamura, M. Pacifci, and M. Enomoto-Iwamoto (2003), Activation of beta- catenin-LEF/TCF signal pathway in chondrocytes stimulates ectopic endochondral ossification, Osteoarthritis Cartilage 11: 36-43. 12. P. Babij, W. Zhao, C. Small, Y. Kharode, P. J. Yaworsky, M. L. Bouxsein, P. S. Reddy, P. V. Bodine, J. A. Robinson, B. Bhat, J. Marzolf, R. A. Moran, and F. Bex (2003), High bone mass in mice expressing a mutant LRP5 gene, J Bone Miner Res 18: 960-974. 13. K. Shindo, N. Kawashima, K. Sakamoto, A. Yamaguchi, A. Umezawa, M. Takagi, K. Katsube, and H. Suda (2003), Osteogenic differentiation of the mesenchymal progenitor cells, Kusa is suppressed by Notch signaling, Exp Cell Res 290: 370-380. 14. M. Sciaudone, E. Gazzerro, L. Priest, A. M. Delany, and E. Canalis (2003), Notch 1 Impairs Osteoblastic Cell Differentiation, Endocrinology 144: 5631-5639. 15. G. Xiao, D. Jiang, P. Thomas, M. D. Benson, K. Guan, G. Karsenty, and R. T. Franceschi (2000), MAPK pathways activate and phosphorylate the osteoblast-specific transcription factor, Cbfa1, J Biol Chem 275: 4453-4459. 16. F. Vinals, T. Lopez-Rovira, J. L. Rosa, and F. Ventura (2002), Inhibition of PI3K/p70 S6K and p38 MAPK cascades increases osteoblastic differentiation induced by BMP-2, FEBSLett 510: 99-104. 17. C. Higuchi, A. Myoui, N. Hashimoto, K. Kuriyama, K. Yoshioka, H. Yoshikawa, and K. Itoh (2002), Continuous inhibition of MAPK signaling promotes the early osteoblastic differentiation and mineralization of the extracellular matrix, J Bone Miner Res 17: 1785-1794. 18. A. Suzuki, J. Guicheux, G. Palmer, Y. Miura, Y. Oiso, J. P. Bonjour, and J. Caverzasio (2002), Evidence for a role of p38 MAP kinase in expression of alkaline phosphatase during osteoblastic cell differentiation, Bone 30: 91-98. 19. S. Gallea, F. Lallemand, A. Atfi, G. Rawadi, V. Ramez, S. Spinella-Jaegle, S. Kawai, C. Faucheu, L. Huet, R. Baron, and S. Roman-Roman (2001), Activation of mitogen-activated protein kinase cascades is involved in regulation of bone morphogenetic protein-2-induced osteoblast differentiation in pluripotent C2C12 cells, Bone 28: 491-498.

- 99 - Chapter 4

20. T. Katagiri, A. Yamaguchi, M. Komaki, E. Abe, N. Takahashi, T. Ikeda, V. Rosen, J. M. Wozney, A. Fujisawa- Sehara, and T. Suda (1994), Bone morphogenetic protein-2 converts the differentiation pathway of C2C12 myoblasts into the osteoblast lineage, J Cell Biol 127: 1755-1766. 21. D. S. de Jong, B. L. T. Vaes, K. J. Dechering, A. Feijen, J. M. A. Hendriks, R. Wehrens, C. L. Mummery, E. J. van Zoelen, W. Olijve, and W. T. Steegenga (2004), Identification of Novel Regulators Associated with Early-Phase Osteoblast Differentiation, J Bone Miner Res 19: 947-958. 22. D. S. de Jong, E. J. van Zoelen, S. Bauerschmidt, W. Olijve, and W. T. Steegenga (2002), Microarray analysis of bone morphogenetic protein, transforming growth factor beta, and activin early response genes during osteoblastic cell differentiation, J Bone Miner Res 17: 2119-2129. 23. D. ten Berge, A. Brouwer, J. Korving, J. F. Martin, and F. Meijlink (1998), Prx1 and Prx2 in skeletogenesis: roles in the craniofacial region, inner ear and limbs, Development 125: 3831-3842. 24. M. Fujii, K. Takeda, T. Imamura, H. Aoki, T. K. Sampath, S. Enomoto, M. Kawabata, M. Kato, H. Ichijo, and K. Miyazono (1999), Roles of bone morphogenetic protein type I receptors and Smad proteins in osteoblast and chondroblast differentiation, M olB iol Cell 10: 3801-3813. 25. U. Valcourt, J. Gouttenoire, A. Moustakas, D. Herbage, and F. Mallein-Gerin (2002), Functions of transforming growth factor-beta family type I receptors and Smad proteins in the hypertrophic maturation and osteoblastic differentiation of chondrocytes, J Biol Chem 277: 33545-33558. 26. N. Hamajima, K. Matsuda, S. Sakata, N. Tamaki, M. Sasaki, and M. Nonaka (1996), A novel gene family defined by human dihydropyrimidinase and three related proteins with differential tissue distribution, Gene 180: 157-163. 27. Y. Goshima, F. Nakamura, P. Strittmatter, and S. M. Strittmatter (1995), Collapsin-induced growth cone collapse mediated by an intracellular protein related to UNC-33, Nature 376: 509-514. 28. T. Iso, V. Sartorelli, G. Chung, T. Shichinohe, L. Kedes, and Y. Hamamori (2001), HERP, a new primary target of Notch regulated by ligand binding, M ol Cell Biol 21: 6071-6079. 29. J. Sun, C. N. Kamei, M. D. Layne, M. K. Jain, J. K. Liao, M. E. Lee, and M. T. Chin (2001), Regulation of myogenic terminal differentiation by the hairy-related transcription factor CHF2, J Biol Chem 276: 18591-18596. 30. M. Matsue, R. Kageyama, D. T. Denhardt, and M. Noda (1997), Helix-loop-helix-type transcription factor (HES-1) is expressed in osteoblastic cells, suppressed by 1,25(OH)2 vitamin D3, and modulates 1,25(OH)2 vitamin D3 enhancement of osteopontin gene expression, Bone 20: 329-334. 31. K. W. McLarren, R. Lo, D. Grbavec, K. Thirunavukkarasu, G. Karsenty, and S. Stifani (2000), The mammalian basic helix loop helix protein HES-1 binds to and modulates the transactivating function of the runt-related factor Cbfa1, J Biol Chem 275: 530-538. 32. K. Bruckner, L. Perez, H. Clausen, and S. Cohen (2000), Glycosyltransferase activity of Fringe modulates Notch- Delta interactions, Nature 406: 411-415. 33. C. Hicks, S. H. Johnston, G. diSibio, A. Collazo, T. F. Vogt, and G. Weinmaster (2000), Fringe differentially modulates Jagged1 and Delta1 signalling through Notch1 and Notch2, Nat Cell Biol 2: 515-520. 34. D. J. Moloney, V. M. Panin, S. H. Johnston, J. Chen, L. Shao, R. Wilson, Y. Wang, P. Stanley, K. D. Irvine, R. S. Haltiwanger, and T. F. Vogt (2000), Fringe is a glycosyltransferase that modifies Notch, Nature 406: 369-375. 35. V. M. Panin, L. Shao, L. Lei, D. J. Moloney, K. D. Irvine, and R. S. Haltiwanger (2002), Notch ligands are substrates for protein O-fucosyltransferase-1 and Fringe, J Biol Chem 277: 29945-29952. 36. Y. A. Evrard, Y. Lun, A. Aulehla, L. Gan, and R. L. Johnson (1998), lunatic fringe is an essential mediator of somite segmentation and patterning, Nature 394: 377-381. 37. N. Zhang, and T. Gridley (1998), Defects in somite formation in lunatic fringe-deficient mice, Nature 394: 374-377. 38. K. Shimizu, S. Chiba, T. Saito, K. Kumano, T. Takahashi, and H. Hirai (2001), Manic fringe and lunatic fringe modify different sites of the Notch2 extracellular region, resulting in different signaling modulation, J Biol Chem 276: 25753­ 25758. 39. K. Tezuka, M. Yasuda, N. Watanabe, N. Morimura, K. Kuroda, S. Miyatani, and N. Hozumi (2002), Stimulation of osteoblastic cell differentiation by Notch, J Bone Miner Res 17: 231-239. 40. N. Haines, and K. D. Irvine (2003), Glycosylation regulates Notch signalling, Nat Rev Mol Cell Biol 4: 786-797.

- 100 - 5

The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation

Diana S. de Jong1, Bart L.T. Vaes1, Olexander Korchynskyi2, Jose Hendriks1, Alie Feijen3, Christine Mummery3, Peter ten Dijke2, Wiebe Olijve1, 4, Wilma Steegenga1, 6, Koen J. Dechering4 and Everardus J.J. van Zoelen5.

1 Department of Applied Biology, University of Nijmegen, Nijmegen, the Netherlands 2 Department of Cellular Biochemistry, Dutch Cancer Institute, Amsterdam, the Netherlands 3 Hubrecht Laboratory, Netherlands Institute of Developmental Biology, Utrecht, the Netherlands 4 Target Discovery Unit, N.V. Organon, Oss, the Netherlands 5 Department of Cell Biology, University of Nijmegen, Nijmegen, the Netherlands 6 Current address: Division of Human Nutrition, Wageningen University, Wageningen, the Netherlands Chapter 5

Abstract Bone morphogenetic proteins (BMPs) are the primary regulators of osteoblast differentiation. BMPs exert their activity through type I and type II serine/threonine kinase receptors and BMP- specific Smad proteins as second messengers. However, activation of Smad signaling alone is not sufficient for full osteoblast differentiation and cooperation with other major signal transduction pathways, such as MAP-kinase, Wnt and Notch signaling, is required. At present, little is known about how these different pathways interact, but most likely specific genes act as focal points for their interaction. To identify such focal points, expression profiles were determined in C2C12 cells of a number of genes that are known to be regulated during different stages of osteoblast differentiation with respect to their regulation by Smads, Erk1/2 MAP-kinase and the Notch signaling modulator Lunatic fringe. The results show that Smad signaling is important but not sufficient to drive full expression of BMP target genes during osteoblast differentiation, whereas Erk1/2 MAP-kinase appears to be particularly important during late phase osteoblast differentiation. Although it could not definitively be proven that Lfng, a modulator of Notch, is expressed in vivo during bone formation, we did find indications that Notch signaling is also important for late phase differentiation. Finally, one immediate early and one intermediate gene, Tieg and Heyl respectively, were found to be regulated by Smad and Erk1/2 MAP-kinase pathways independently, indicating that these two genes may serve as focal points for integration of the different signaling pathways during osteoblast differentiation.

Introduction Bone morphogenetic proteins (BMPs) are the most potent inducers of osteoblast differentiation. Although the BMPs were first identified for their ability to induce ectopic bone formation in vivo (1), these proteins appear to be multifunctional regulators of morphogenesis during embryonic development (2). To generate biological responses, BMP family members use heteromeric complexes of so-called type I and type II serine/threonine kinase receptors. For BMPs, three distinct type II receptors have been identified, in particular BMPR-II (BMP type II receptor), ActR-II (activin type II receptor) and ActR-IIB (activin type IIB receptor). Moreover, three type I receptors, also referred to as activin receptor-like kinases (Alks) have been identified, namely BMPR-IA/Alk3 (BMP type IA receptor), BMPR-IB/Alk6 (BMP type IB receptor) and Alk2 (activin receptor-like kinase 2) (3). Upon ligand binding, a complex is formed between two type I and two type II receptors, upon which the constitutively active kinase of the type II receptor phosphorylates and activates the type I receptor which subsequently activates specific receptor-regulated Smads (R-Smads). Smad-1, -5 and - 8 have been shown to be BMP responsive R-Smads, while Smad-2 and -3 are responsive to TGF-P and activins, both proteins that are closely related to BMPs. In combination with the common mediator Smad-4, the respective R-Smads are assembled into heteromeric complexes, which subsequently translocate to the nucleus, where they act in concert with other proteins as transcription factor complexes (3-5). In addition to the BMP-activated Smad pathway, several other major signaling pathways are known to play a role during osteoblast differentiation. These include the different MAP-kinase pathways (6-10) and the Notch pathway (11, 12). Moreover, it has been shown that BMPs not only activate various MAP-kinase pathways, but also stimulate the activity of various transcriptional mediators of the MAP-kinase pathway (AP-1 members), and regulate the expression of several AP-1 members (13-15). Recently, it has been shown that BMP receptor binding results in a direct activation of the Tab1 (Tak1 binding protein 1)/Tak1 (TGF-P activated kinase 1)/p38 MAP-kinase pathway (16­ 19). Interestingly, p38 MAP-kinase has been shown to be a positive regulator of bone cell

- 102 - The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation

differentiation, since its inhibition results in down-regulation of the genes for the osteoblast markers Cbfal (Core binding factor al), Alp2 (Alkaline phosphatase type liver, bone, kidney), Colla (Collagen type Ia), Ocn (Osteocalcin), Opn (Osteopontin) and Opg (Osteoprotegerin), as well as in decreased matrix mineralization (9, 10, 20-25). In contrast to p38, Erkl/2 MAP-kinase has been shown to be mainly a negative regulator of osteoblast differentiation. Although complete inactivation of Erkl/2 MAP-kinase activity inhibits osteoblast differentiation (26), this effect is mainly due to the finding that Erkl/2 MAP-kinase- mediated phosphorylation of Cbfal is necessary for Cbfal activity (6). Partial inhibition of Erkl/2 MAP-kinase is sufficient for residual Cbfal activity and enhancement of BMP-induced osteoblast differentiation, as shown by the up-regulation of Alp2 activity as well as matrix mineralization. Furthermore, expression of constitutively active MEKl, which activates Erkl/2 MAP-kinase, results in reduced BMP-induced osteoblast differentiation, as shown by repressed Alp2 activity (8). BMP is able to regulate the activity of Notch signaling based on the observation that the Notch intracellular domain (NICD), which functions as a transcription factor, can interact with both TGF-P- and BMP-specific Smads. TGF-P rapidly induces the expression of Hesl in a NICD- and Smad- dependent manner, and it has been shown that NICD can bind to Smad-3 at its MH2-domain. This interaction is necessary for the activation of NICD- and Smad- binding promoters (27). Moreover, it has been shown that interaction of the BMP-specific Smad-l with NICD results in the inhibition of both myogenesis and BMP-induced osteogenesis of the C2Cl2 mesenchymal progenitor cell line, probably through up-regulation of specific Notch target genes (28). However, the interplay between the various pathways in osteoblast differentiation is still poorly understood. Recently, we identified Lunatic fringe (Lfng) and Heyl (Hairy and enhancer of split related with YRPW motif l) as potential focal points for the integration of the Smad and Notch signaling pathways in osteoblast differentiation (29, 30). Lfng belongs to the family of Fringe proteins, which are glycosyltransferases that modulate the activity of the Notch receptors (3l, 32), and Heyl belongs to the Hes/Herp family of transcription factors that function in the Notch signal transduction pathway (33, 34). To investigate the influence of various signal transduction pathways on osteoblast differentiation, we studied the role of Smads, Erkl/2 MAP-kinase and Notch on gene expression regulation during various stages of osteoblast differentiation. Our results show that during BMP- induced osteoblast differentiation, gene expression is regulated, at least in part, by Smads, and that particularly for the expression of late phase bone markers, Smads act in combination with Erkl/2 MAP-kinase and Notch signaling. Moreover, we identified two genes as potential focal points for the integration of the Smad and Erkl/2 signaling pathways.

Materials and Methods Tissue culture and growth factors The C2C12 murine mesenchymal progenitor cell line (the American Type Culture Collection) was maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% newborn calf serum (NCS), penicillin (100 U/ml) and streptomycin (100 |ig/ml), as described previously (35). For the various experiments, cells were plated at a density of 2.0x104 cells/cm2 and grown for 24 hrs in DMEM supplemented with 10% NCS unless stated otherwise. In the Erk1/2 MAP-kinase experiment, the medium was subsequently replaced by DMEM containing 5% NCS either in the presence or in the absence of 5|iM of the MEK inhibitor U0126, and incubated for 2 hrs. After incubation, the medium was replaced by DMEM containing 5% NCS either in the presence or in the absence of 300 ng/ml recombinant human BMP2 with or without 5|iM of the MEK inhibitor U0126.

- 103 - Chapter 5

In the other experiments, the medium was subsequently replaced by DMEM containing 5% NCS either in the presence or in the absence of 300 ng/ml recombinant human BMP2, 5 ng/ml TGF-P or a combination of these growth factors, unless stated otherwise. The MEK1/2 inhibitor U0126 was purchased from Sigma and recombinant human BMP2 was kindly provided by Wyeth (Andover, MA).

Adenoviral infections Cells were plated at density of 2.5x104 cells/cm2 and grown for 12 hrs in DMEM supplemented with 10% fetal calf serum (FCS). Subsequently, the cells were infected with the various virus stocks at a MOI of 2000. Before harvesting, the cells were washed extensively and starved for 10 hrs in serum-free medium supplemented with 0.2% bovine serum albumin (BSA; Sigma). Virus stocks containing expression constructs for LacZ (P-galactosidase expressing control), constitutively active Alk2 receptor (caAlk2), Smad1, Smad4, Smad5 and Smad8 were kindly provided by dr. K. Miyazono (36).

RNA isolation and Real-time RT-PCR Total RNA was extracted according to an acid guanidium thiocyanate-phenol-chloroform method (TriPure® Isolation Reagent, Roche Diagnostics) and RNA concentrations were determined by measuring the absorbance at 260 nm. For real-time RT-PCR, the ABI Prism Sequence Detection System 5700 and Primer Express™ Software (PE Biosystems) were used. For each gene, a set of primers was designed using sequences obtained from Genbank™ and amplicons of 50-100 base pairs with Tm between 58 and 60°C were selected. Prior to complementary DNA (cDNA) synthesis, 2 ^g of total RNA were DNase (Invitrogen) treated for 15 min at 37°C, after which RNA was reverse- transcribed using random hexamer primers (Amersham Pharmacia) in a total volume of 50 |il using the SUPERSCRIPTTM II reverse transcriptase kit (Invitrogen). Subsequently, aliquots of first-strand cDNA (2|il) were amplified using SYBR Green® PCR Mastermix (Applied Biosystems) under the following conditions: initial denaturation for 10 min at 95°C, followed by 40 cycles consisting of 15 sec at 94°C and 1 min at 60°C. Fold inductions and expression ratios were calculated from differences in threshold cycles at which an increase in reporter fluorescence above a baseline signal could first be detected (CT) between two samples, and averaged for duplicate experiments.

Protein Isolation and Western Blotting Cells were lysed in IPB-0.7 buffer (50 mM Tris pH7.5, 0.7 M NaCl, 0.5% NP40, 0.25% sodium deoxycholic acid, 1 mM EDTA) supplemented with 1 |J.M PMSF, 1 ng/ml leupeptin, 1 ng/ml pepstatin, 0.5 ng/ml aprotein, 1 mM Na3VO4 and 1 mM NaF. Proteins from the cell lysates were separated on SDS 10% polyacrylamide gel electrophoresis and transferred to Protran Nitrocellulose membranes (Schleicher & Schuell). Phosphorylated Erk1/2 MAP-kinase was detected using a phospho-p44/42 MAP kinase (Thr202/Tyr204) antibody (Cell Signaling Technology), while total Erk1/2 MAP-kinase was detected using the K-23 antibody (Santa Cruz). All antibodies were used at a 1:1000 dilution. After incubation with HRP-conjugated secondary antibodies (Tebu-bio, 1:2000 dilution), the proteins were visualized by enhanced chemo-luminescence (ECL, Roche Diagnostics).

Plasmids and stable transfections The human LFNG cDNA cloned in pCMV-Sport6 was purchased from the German Research Center for Genome Research (RZPD). The LFNG coding region was excised from the plasmid by Sall/Notl digestion and ligated into Sall/Notl digested pcDNA3.1. The integrity of the resulting pLFNG -3.1 plasmid was verified by DNA sequence analysis. For stable transfections, pLFNG-3.1

- 104 - The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation

was introduced into C2C12 cells using the Lipofectamin 2000 method according to the manufacturer’s specifications (Gibco/BRL). Monoclonal stable transfectants were selected in the presence of G418 (400 |ig/ml) and LFNG expression levels were analyzed by real-time RT-PCR.

Alkaline phosphatase assay Cells were seeded in 96-well tissue culture plates (Nalge Nunc International, Denmark) and grown as described above. Subsequently, the medium was changed to DMEM containing 5% NCS, supplemented with increasing doses of BMP2. Alkaline phosphatase activity was measured 48h after stimulation as described by Van der Plas and coworkers (37). Alkaline phosphatase activity was corrected for differences in cell number based on a measurement of neutral red staining (38).

Calcium deposition assay Cells were seeded in 24-well tissue culture plates (Nalge Nunc International, Denmark) and grown as described above. Subsequently, the medium was changed to DMEM containing 5% NCS in the presence or absence of BMP2 (300 ng/ml) supplemented with P-glycerophosphate (12.5 mM). Calcium deposition was measured 8 days after stimulation using the Sigma calcium release assay system.

In situ hybridization Mouse embryos were collected from F1 females (Cba x C57B16) at embryonic day 14.5 (E14.5), dissected free from the uterus and membranes, fixed overnight at 4oC in 4% paraformaldehyde, dehydrated, embedded in paraffin wax and sectioned (6 |im) onto coated slides (Klinipath, Duiven, the Netherlands). Neonatal mouse tails, kindly provided by Dr. Marcel Karperien (Leiden University Medical Center, the Netherlands), were collected from Swiss albino mice on day 5 after birth and fixed in 4% paraformaldehyde, decalcified by incubation in 12.5% sodium EDTA (pH8.0) for 1 week at 4oC, embedded in paraffin wax and sectioned (5 |im) as described previously (39). The slides were dewaxed, rehydrated and pretreated with Proteinase K and acetic anhydride, and hybridized overnight at 55oC, essentially as described previously (40). Constructs were cloned into pSKII- in the anti-sense orientation in the T7 promoter. RNA probes were generated by transcription of the T7 RNA polymerase in the presence of [a-35S]-uridine-triphosphate (UTP). After hybridization, the slides were washed at high stringency for 30 min at 62oC, treated with RNase A for 30 min at 37oC, washed again at high stringency, dehydrated and allowed to dry. Autoradiography was performed using Ilford G5 photo emulsion. The slides were exposed for 4 weeks at 4oC. Photographs were made from combined bright-field (blue filter) and dark-field (red filter) images.

Results Expression regulation by Smads Previous data have shown that C2C12 mesenchymal cells undergo osteoblastic differentiation upon infection with the genes encoding the constitutively active type I receptors Alk2, Alk3 and Alk6, as shown by the induction of Alp2 activity (36). Furthermore, these BMP type I receptors are known to have different effects on bone formation in vivo (41, 42), but in C2C12 mesenchymal cells no difference could be observed between these three receptors in their ability to induce osteoblast differentiation, as measured by gene expression patterns and induction of Alp2 activity (36, 43). In addition to the type I BMP receptors Alk2, Alk3 and Alk6, C2C12 cells are also able to undergo osteoblastic differentiation upon infection with the genes encoding Smad1 and Smad5, as shown by the induction of Alp2 activity (36). Moreover, co-infection with the gene encoding the common

- 105 - Chapter 5

mediator Smad-4 potentiated the effect of Smad-1 and Smad-5 (36). At present, little is known about the ability of the third BMP-specific R-Smad, Smad-8, to induce osteoblast differentiation. Furthermore, it is not known whether the different BMP-specific R-Smads induce different genes during osteoblast differentiation. In this study, we therefore examined the involvement of the various BMP-specific R-Smads alone and in combination with Smad-4 in the regulation of genes expressed during osteoblast differentiation. To obtain an general overview of the involvement of the Smad signal transduction pathway in osteoblast differentiation, we examined the expression of transcription factor genes known to be regulated during the early phases of osteoblast differentiation (29), as well as a number of established osteoblast markers and several novel late osteoblast differentiation marker genes that we identified in a previous study (44). The genes studied include Id-1, Id-2, Id-3 (Inhibitor of differentiation genes 1­ 3), Oasis (Old astrocyte specifically induced substance), Prrx2 (Paired related homeobox 2), Smad-6, Smad-7 (MAD homologues 6 and 7), Snail and Tieg (TGF-P-induced early growth factor) as immediate early regulated genes; Cbfal, Heyl and Tcf7 (T-cell specific factor 7) as intermediate genes; and Alp2, Ocn, Col6a2 (Collagen type 6 a2), Prdc (Protein related to DAN and Cerberus), Wif-1 (Wnt inhibitory factor 1) and the ESTs (Expressed sequence tags) AA222756, AA709999 and AA794190 as late phase genes. As shown in Figure 1A, expression of several of the immediate early and intermediate genes was induced by over-expression of either a single Smad cDNA alone or in combination with Smad-4, which is considered to be the typical manner in which BMP responsive genes react to Smad signaling. In particular, expression of Smad-7, Tieg and Heyl was moderately induced 24 hrs and 48 hrs after infection with genes encoding Smad-1, Smad-4, Smad-5 or Smad-8, and this effect was strongly enhanced upon co-infection of the R-Smads with the Smad-4 cDNA. Furthermore, expression of Id-1, Id-2 and Id-3 was shown to be induced by the various Smads tested here, but in this case co-infection with Smad-4 did not significantly potentiate expression levels. Intriguingly, the immediate early genes Snail, Prrx2, Oasis and Smad-6, as well as the intermediate genes Tcf-7 and Cbfal, were not induced by infection with Smad genes, even upon co-infection with Smad-4. However, these genes were induced by addition of BMP as well as upon infection with a cDNA encoding caAlk2. These results imply that for Snail, Prrx2, Oasis, Smad-6, Tcf-7 and Cbfal the Smad signal transduction pathway is either not or only weakly involved in their expression regulation. Figure 1B shows that 6 out of 8 late phase genes, in particular Alp2, Prdc, Wif-l and the ESTs AA222756, AA794190 and AA709999, were moderately regulated by infection with only Smad-1, Smad-4, Smad-5 or Smad-8 cDNA, and co-infection with Smad-4 cDNA potentiated this effect not only for Smad-8 (Wif-l) but also for Smad-1 and Smad-5 (Alp2, Prdc, AA222756, AA794190 and AA709999). It should be noted, however, that for 4 of these 6 genes their gene expression levels induced by Smads were considerably lower than those observed after infection with caAlk2 or treatment with BMP2. Most probably this is due to the fact that the transfected Smads were not activated. After infection of caAlk2 or BMP2 treatment, the endogenous levels of activated Smads are probably higher than those achieved by over-expression without ligand stimulus. Furthermore, our data show that in the case of Col6a2, co-infection with Smad-4 cDNA did not significantly potentiate its expression regulation as compared to the single Smad infections. Finally, it was shown that expression of Ocn could not be induced by either single Smads or by co­ infection of the R-Smads with Smad-4. Since Cbfa1 is the major transcription factor that regulates Ocn expression (45), and Cbfal expression was not affected by Smads in the present experiment, it is likely that also expression of Ocn cannot be regulated by infection with Smads.

- 106 - The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation

Figure 1. Smad- mediated expression during osteoblast dif­ ferentiation. C2C12 cells were seeded and grown for 12 hrs after which they were infected with the various virus stocks at an MOI of 2000. Before harves- ing, the cells were washed extensively and starved for 10 hrs in serum-free medium supplemented with 0.2% BSA. RNA was isolated 0 hr, 1 day, 2 days and 3 days after infection. As a control, C2C12 cells were plated at a density of 2.0x104 cells/cm2 and grown for 24 hrs in DMEM supplemented with 10% NCS. Subse­ quently, the medium was replaced by DMEM containing 5% NCS either in the presence or in the absence of BMP2 (300 ng/ml). RNA was isolated 0 hrs, 1 day and 3 days after stimulation. Subse­ quently, expression levels were measured by quantitative real time RT-PCR. (A) Expression regulation of immediate early and intermediate osteoblast differen­ tiation genes, (B) Expression regulation of late phase osteo­ blast differentiation genes. LacZ = P- galactosidase expres­ sing control, caAlk2 = constitutively active Alk2 receptor.

In conclusion, these experiments show that the expression of 55% (5 out of 9) of the immediate early and 75% (6 out of 8) of the late phase genes are regulated by Smads, although it is realized that the obtained expression levels of the late phase genes were sub optimal. Furthermore, for 44% (4 out of 9) of the immediate early genes and 66% (2 out of 3) of the intermediate genes, Smads

- 107 - Chapter 5

alone did not appear sufficient to induce expression of these genes. Finally, these results indicate that most of the genes that can be induced by Smads in C2C12 cells are regulated by all three BMP- specific R-Smads without preference.

Regulation by the Erk1/2 MAP-kinase signal transduction pathway For full differentiation of C2C12 cells into mineralizing osteoblasts, cells have to be treated with BMP2 in the presence of at least 5% serum. At lower serum concentrations, cells die around day 3 (results not shown). Serum has been reported to activate signal transduction pathways such as the Erk1/2 MAP-kinase pathway independently of BMP2 and previous studies have shown that Erk1/2 MAP-kinase plays a role in osteoblast differentiation next to the BMP-induced Smad signal transduction pathway (5, 6, 8, 10). Continuous inhibition of the Erk1/2 MAP-kinase signaling pathway has been shown to enhance BMP-induced osteoblast differentiation and mineralization of the matrix (8), and Erk1/2 MAP-kinase is able to activate the osteoblast-specific transcription factor Cbfa1 (6). Therefore we also examined the involvement of Erk1/2 MAP-kinase in the expression regulation of the above-mentioned genes during BMP-induced osteoblast differentiation.

Figure 2. Influence of inhibition of Erk1/2 MAP-kinase phosphorylation on BMP2-induced gene expression. C2C12 cells were seeded and grown for 24 hrs upon which medium was replaced by DMEM containing 5% NCS in the presence or absence of the MEK inhibitor U0126 (5|iM). After 2 hrs, the cells were stimulated with BMP2 (300 ng/ml) in the presence or absence of the inhibitor. (A) Western Blot analysis of Erk1/2 phosphorylation. Proteins were isolated from whole cell lysates 0 hrs, 15 min, 1 hr, 2.5 hrs, 5 hrs and 24 hrs after treatment, (B) Expression regulation of immediate early and intermediate osteoblast differentiation genes. RNA was isolated 0 hrs, 2 hrs, 6 hrs and 24 hrs after treatment. Subsequently, expression levels were measured by quantitative real-time RT-PCR, (C) Expression regulation of late phase osteoblast differentiation genes. RNA was isolated 0 hrs, 2 hrs, 6 hrs, 1 day and 3 days after treatment. Subsequently, expression levels were measured by quantitative real­ time RT-PCR.

- 108 - The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation

As is shown in Figure 2A, Erk1/2 MAP-kinase is readily phosphorylated by freshly added serum also in the absence of BMP2. Within 24 hrs after BMP2 treatment, this phosphorylation gradually declined. Phosphorylation of Erk1/2 MAP-kinase is completely blocked in the presence of the U0126 MEK inhibitor. We therefore tested the effect of U0126 on BMP2-induced expression regulation of the transcription factors identified in this study. As shown in Figure 2B, no inhibitory effect could be observed on BMP2-induced expression regulation for any of the transcription factors tested. Moreover, BMP-induced regulation was either not affected, or up-regulated on at least one time point in the presence of the MEK inhibitor as observed for Prrx2, Tieg, Heyl and Cbfal. To verify whether U0126 represses Erk1/2 MAP-kinase-mediated transcription regulation under our experimental conditions, we analyzed expression of the cholecystokinin (CCK) gene in more detail. This gene is an intermediate BMP response gene (29) and previous studies have shown that its expression is regulated by Erk1/2 MAP-kinase (46, 47). As shown in Figure 2C, BMP-induced expression regulation of this gene was indeed completely blocked in the presence of U0126. Of the 9 immediate early BMP target genes that encode transcription factors, only the expression of Prrx2 and Tieg appeared to be regulated by Erk1/2 MAP-kinase (Figure 2B). It is known from previous studies that Erk1/2 MAP-kinase inhibition enhances osteoblast differentiation and matrix mineralization (8), and therefore we speculated that Erk1/2 MAP-kinase might affect the expression regulation of the late phase osteoblast differentiation genes more strongly than that of the early phase genes. To investigate this hypothesis, Erk1/2 MAP-kinase-mediated expression regulation was analyzed for the late phase osteoblast marker genes, including Alp2 and Ocn. As shown in Figure 2C, the BMP-induced expression of Ocn was blocked in the presence of the inhibitor in a similar manner as reported previously (6, 10). In addition, the expression of Prdc, Col6a2 and all three ESTs was up-regulated, while expression of Alp2 and Wif-l remained unaffected by the inhibitor. In conclusion, our data indicate that indeed Erk1/2 MAP-kinase induces a more pronounced regulation of the genes expressed in late phase than of those expressed in early phase osteoblast differentiation. The results of these experiments in combination with the Smad infection experiments provide more insight into the question by which signal transduction pathway and underlying mechanism the various osteoblast differentiation-related genes are regulated. Table 1 summarizes these data, from which it is shown that expression of the immediate early genes Smad-7, Id-l, Id-2 and Id-3, as well as of the late phase genes Alp2 and Wif-l can be induced by Smads, whereas inhibition of Erk1/2 MAP- kinase activation has no effect. Furthermore, expression of the immediate early gene Tieg, the intermediate gene Heyl and the late phase gene Prdc as well as the ESTs AA222756, AA794190 and AA709999, can be induced by Smads while inhibition of Erk1/2 activation enhances BMP-induced expression levels, indicating that the Erk1/2 MAP-kinase signal transduction pathway negatively regulates the expression of these genes. Moreover, Table 1 shows that Smads are not sufficient to induce the expression of the immediate early gene Prrx2, the intermediate gene Cbfal and the late phase genes Col6a2 and Ocn, whereas Erk1/2 MAP-kinase negatively regulates the expression of Prrx2, Cbfal and Col6a2, and positively regulates the expression of Ocn. Finally, neither Smads nor the Erk1/2 MAP-kinase signal transduction pathway are by themselves sufficient to regulate the expression of the immediate early genes Snail, Oasis and Smad-6, as well as the intermediate gene Tcf7. This indicates that expression regulation of these genes depends mostly on other signal transduction pathways, thereby suggesting that a combination of multiple signal transduction pathways is needed for gene expression regulation.

- 109 - Chapter 5

Table 1. Expression regulation of BMP-response genes by Smads and Erk1/2 MAP-kinase Gene BMP effect caALK2 Smads U0126/Erk Smad-7 ▲ ▲ ▲ ▲ Id1 ▲ ▲ ▲ ▲ Id2 ▲ ▲ ▲▲ ▲ Id3 ▲ ▲ ▲▲ ▲ Tieg ▲ ▲ ▲ ▲ Prrx2 ▲ ▲ ▲ - ▲ Snail ▲ ▲ - Oasis ▲ ▲▲ ▲ - Smad-6 ▲ ▲ ▲ -

Hey ▲ ▲ ▲▲ ▲ ▲ ▲ ▲ - Cbfa-1 ▲ ▲ ▲ ▲ ▲ ▲ ▲ Tcf7 ▲ ▲ -

Alp2 ▲ Wif-1 ▲ AA222756 ▲ ▲ ▲ ▲ ▲▲ Prdc ▲ ▲ ▲ ▲▲▲ ▲ ▲ AA794190 ▲ ▲▲▲▲ ▲ ▲ ▲▲ AA709999 ▲▲▲▲▲▲ ▲ ▲ Col6a2 ▲ ▲ ▲ - ▲ <

Ocn ▲ ▲ ▲ ▲ - ▼ ▼▼ Summary of Figures 1 and 2. ▲ = upregulation, ▼ = down-regulation, the number of triangles indicate relative levels of expression. See page 142 for full color table

Regulation by the Notch signal transduction pathway One of the signal transduction pathways that are potentially involved in the expression regulation of osteoblast differentiation-related genes, is the Notch signal transduction pathway. Upon ligand binding, the Notch receptor becomes activated and the majority of its intracellular part, the so- called Notch intracellular domain (NICD), is subsequently released by presenilin-dependent cleavage. The NICD then translocates to the nucleus where it functions in concert with the transcription factor Su(H)/RBP-J/CBF1 (Suppressor of Hairless / recombining binding protein suppressor of hairless) in a complex (48, 49) that induces the expression of members of the family of Hairy and Enhancer of Split (Hes/Herp) transcription factors (33, 50). Previously, we have reported that Heyl, Hesl and Lfng (lunatic fringe, a modulator of Notch receptor activity) are specifically regulated by BMP2 during osteoblast differentiation (30). In order to study the role of Notch signaling in the regulation of osteoblast differentiation related genes, it is therefore of interest to investigate the influence of ectopic expression of the NICD as well as of Heyl, Hesl and Lfng on osteoblast differentiation. Ectopic expression of NICD has previously been reported to induce the expression of both Heyl and Hey2 (34, 51). However, since Hey2 was not expressed in our previous experiments (results not shown), we decided to study the involvement of Notch signaling on BMP-induced osteoblast differentiation by focusing on the effect induced by ectopic expression of Heyl and Lfng. Expression vectors containing the human Lfng (LFNG) and human Heyl (HEY1) genes were introduced into C2C12 cells, and stably expressing cells were identified for LFNG (Figure 3A), but could, for unknown reasons, not be established for HEYl. The differentiation capacity of the LFNG- expressing cells (C2C12-LFNG) was assessed by measuring Alp2 activity, Ca2+ deposition, morphological changes, as well as by expression regulation of early, intermediate and late osteoblast markers, and the myoblast differentiation marker myogenin (Myog) upon treatment of the cells with BMP2, TGF-P or a combination of both stimuli. Treatment of the cells with TGF-P and a

- 110 - The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation

combination of TGF-P and BMP2 was included in these experiments to study the effect of ectopic expression of LFNG on the capacity of TGF-P to inhibit osteoblast differentiation. Since we have previously shown that BMP-induced expression of endogenous Lfng was inhibited by TGF-P (30), we postulated that Lfng could be a positive regulator of osteoblast differentiation and might be able to counteract the inhibitory effect of TGF-P on BMP-induced osteoblast differentiation.

Figure 3. BMP2-induced osteoblast differentiation in C2C12-LFNG cells (A) Expression of LFNG in C2C12-LFNG clones. Expression levels were measured by quantitative real-time RT-PCR, (B) Dose- dependent induction of ALP activity 48 hrs after BMP2-treatment in C2C12 (□) and C2C12-LFNG (o) cells, (C) Ca2+ deposition in C2C12-LFNG cells. Cells were seeded and grown for 24 hrs upon which medium was replaced by DMEM containing 5% NCS in the presence or absence of BMP2 (300 ng/ml) supplemented with P-glycerophosphate (12.5 mM). Calcium deposition was measured 8 days after stimulation, (D) Morphological features of C2C12-LFNG cells. Cells were seeded and grown for 24 hrs upon which medium was replaced by DMEM containing 5% NCS in the presence or absence of BMP2 (300 ng/ml), TGF-P (5ng/ml) or a combination of these growth factors. Photographs were taken 3 days after treatment, (E) Gene expression profiles in C2C12-LFNG cells. Cells were seeded and grown for 24 hrs upon which medium was replaced by DMEM containing 5% NCS in the presence or absence of BMP2 (300 ng/ml), TGF-P (5ng/ml) or a combination of these growth factors. Messenger RNA was used to measure the expression of 3 immediate early, 3 intermediate , 2 late phase osteoblast differentiation genes and 1 myogenic marker gene at the indicated time points. Mean expression levels were corrected for GAPDH and are expressed relative to the basal expression level in un-stimulated C2C12 cells. Error bars indicate differences between duplicate experiments.

- 111 - Chapter 5

As shown in Figure 3A, expression of LFNG had no influence on the expression of endogenous Lfng, neither under control conditions nor upon TGF-P stimulation, but the BMP2- induced levels were approximately 2-fold increased in the LFNG expressing clones. Furthermore, the data represented in Figure 3B and 3C show that the C2C12-LFNG cells had lost their ability to differentiate into osteoblast cells, as evidenced by a complete lack of Alp2 activity and Ca2+ deposition, in comparison with control C2C12 cells. Remarkably, the C2C12-LFNG cells were morphologically very similar to normal C2C12 cells, such that they became elongated and fused under myogenic conditions (5% serum), and showed a cuboidal morphology after BMP2 treatment. Moreover, TGF-P was able to inhibit the morphological changes related to both myogenesis and osteogenesis (Figure 3D). To investigate the genomic basis of these findings, the expression regulation of the immediate early BMP2 response genes Smad6, Prrx2 and Id-l, as well as that of the intermediate genes Cbfal, H eyl and Tcf7, was analyzed. In addition, the late osteoblast marker genes Alp2 and Ocn, the non­ osteoblast-specific late phase BMP2-regulated gene decorin (Dcn), and the myoblast marker Myog were analyzed as well. Figure 3E shows that the expression regulation profiles of Smad6, Id-l and Prrx2 were not or only slightly affected in LFNG-expressing cells. These results indicate that the immediate early gene regulation by BMP2 is probably not affected by Lfng. However, the intermediate osteoblast differentiation markers Cbfal and H eyl, as well as the osteoblast markers Alp2 and Ocn, showed strong inhibition of their BMP2-induced gene expression in the C212-LFNG cells. Furthermore, the expression levels of these genes were strongly reduced in the additional presence of TGF-P. These results indicate that Lfng might affect the overall ability of the cells to differentiate into osteoblasts. On the other hand, expression profiles of Tcf7, a transcription factor of the Wnt signaling pathway, were not or only very slightly affected. Expression of the decorin gene varied extensively between duplicate experiments in wild-type C2C12 cells (results not shown), and therefore no solid conclusions could be drawn from these results on the C2C12-LFNG cells. Inhibition of osteoblast differentiation by Lfng might be due to a general inhibition of differentiation capacity of the C2C12 cells. Therefore, we investigated the expression of the myoblast differentiation marker Myog. Figure 3E shows that in wild-type C2C12 cells expression of Myog is extensively induced under myogenic differentiation conditions (control), whereas it is greatly reduced by BMP2 as well as by TGF-P. In the C2C12-LFNG cells, Myog was normally regulated under control conditions. However, inhibition of Myog expression by BMP2 was much less rigorous than in wild-type C2C12 cells, and inhibition of Myog expression by TGF-P was completely abolished. These results indicate that the C2C12-LFNG cells can still normally differentiate into myoblasts, and that the effect of Lfng on osteoblast differentiation is not an effect of a general inhibition of cellular differentiation. In summary, our results indicate that over-expression of Lfng specifically inhibits the functional differentiation of C2C12 cells into osteoblasts. Most likely, Lfng has no influence on the initiation of osteoblast differentiation, but it inhibits the differentiation process at a later stage, such that no functional osteoblasts are formed.

Expression of Lfng during embryonal development Results obtained from the above described experiments implicate that Lfng could play a role in modulating osteoblast differentiation. To study the role of Lfng during embryonic bone development, expression of this gene was analyzed in mouse embryos at E14.5 of gestation and in neonatal mouse tails by the use of in situ hybridization. Neonatal mouse tails were used because their primary centers of endochondral ossification are still developing in the most distal vertebrae. These vertebrae consist of alkaline phosphatase-negative cartilage whereas the most proximal vertebrae are

- 112 - The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation

alkaline phosphatase-positive and mineralized (39). The expression pattern of Lfng was compared to that of Cfbal and Heyl, which have previously been shown to be expressed in embryonic bone development with similar but distinct patterns (35, 45). We first confirmed the expression of Cbfal and Heyl at E14.5 in bone in transverse sections of the head at the level of the jaws (Figure 4D and 4E), and in proximal vertebrae of the mouse neonatal tail (Figure 4G and 4H). The ossification centers of the developing mandibular and maxillary bones, nasal bone, dental mesenchyme surrounding the developing tooth, as well as proliferating chondroblasts, periostal osteoblasts and trabecular osteoblast all showed clear expression of Cbfal as expected (29, 44, 45). In addition, low but significant levels of expression of Heyl were observed in the mandibular and maxillary bones, as well as in the trabecular bone of the mouse neonatal tail vertebrae as expected (29). Remarkably, Lfng did not display any detectable level of expression in the bones analyzed, as shown in Figure 4F and 4I. Possibly the probe used for this experiment may not have been sensitive enough for detection of Lfng expression. However, sagital sections of the head at the level of the ventricles at E14.5 of gestation showed significant levels of expression of Lfng (Figure 4C) and Heyl (Figure 4B) as expected (52-54), whereas Cbfal did not show any expression here (Figure 4A). Therefore, most likely in vivo expression of Lfng in bone is below the detection level of this technique.

Figure 4. In vivo expression of Heyl and Lfng in the developing mouse skeleton. (A-C) Sagital sections of E14.5 mouse embryos at the level of the ventricles for the control osteoblast marker Cbfal (A), Heyl (B) and Lfng (C), (D-F) Transverse sections of E14.5 mouse embryos at the level of the jaws for Cbfal (D), H eyl (E) and Lfng (F), (G-I) Mouse proximal tail vertebrae for Cbfal (G), H eyl (H) and Lfng (I). Abbreviations used: bm, bone marrow; D, Diencephalon; hc, hypertrophic chondroblasts; id, intervertebral disc; Jo, Jacobson’s organ; md, mandible; Mo, Medulla oblongata; mx, maxilla; nc, nasal cavity; nca, nasal capsule; P, periosteal bone; pc, proliferating chondroblasts; T, trabecular bone; tg, tongue; Sb, Skull bones, V, Ventricular zone; vb, vertebral body. See page 14 2for full color figure

- 11S - Chapter 5

Discussion The contribution of the various BMP-specific R-Smads (Smad-1, Smad-5 and Smad-8) as well as that of other signal transduction pathways, such as Erk1/2 MAP-kinase and Notch, in regulating BMP-induced gene expression during osteoblast differentiation is only known for a couple of genes and has not yet been related to time-dependent expression profiles. In this study, we examined the involvement of the BMP-specific R-Smads alone and in combination with Smad-4 on the expression regulation of various genes regulated during osteoblast differentiation and compared their effects with those of Alk2 and BMP2. Based on their responsiveness to Smads and the synergistic effects of Smad-4, we identified Smad-7, Tieg, Heyl, Alp2, Prdc, Wif-l and the ESTs AA222756, AA794l90 and AA709999 as typical BMP-inducible genes. However, we also identified two groups of genes with distinct Smad responsiveness. The first group, including Id-l, Id-2, Id-3 and Col6a2, consists of genes for which expression was modulated by over-expression of single BMP-specific Smads, without additional effects of co-expressed Smad- 4. In contrast to our results, it has been shown elsewhere that the Id genes are immediate target genes of BMPs (15, 55), and that expression regulation of Id-l by Smads resembles that of typical BMP- inducible genes (56). More experimental data will be required before it can be decided which of the two options is correct. The second group, including Snail, Prrx2, Oasis, Smad-6, Tcf-7, Cbfal and Ocn, consists of genes for which expression was not affected after transfection by either single Smads or combinations with Smad-4. For these genes, the Smad signal transduction pathway is either not involved in their expression regulation, or is not sufficient on its own to induce expression. Tcf-7 is a major transcription factor of the Wnt/P-catenin signal transduction pathway, and it has been shown that both Tcf-7 and L efl, which encodes a family member of Tcf-7, can be induced by BMP (29, 57). It is unknown, however, whether this induction requires the activity of BMP-specific Smads. Expression of the gene for Cbfa1, the major transcription factor of the osteoblast specific marker Ocn (45), is mainly induced by p38 MAP-kinase (25). Furthermore, Smads are able to induce Cbfal expression through a JunB/p38 MAP-kinase cascade (25). In addition, it has recently been shown that PKA/CREB (protein kinase A/cAMP responsive element binding protein) signaling cooperates with BMP-specific R-Smads to enhance the expression of Smad-6, a BMP-specific inhibitory Smad (58). Therefore, it seems likely that expression of Snail, Prrx2, Oasis, Smad-6, Tcf-7, Cbfal and Ocn is regulated by Smads, but that Smads alone are not sufficient and the cooperation of other signal transduction pathways is required to induce their expression. Further research is required to establish whether these genes are indeed induced by Smads. In addition to the BMP-induced Smad signal transduction pathway, the Erk1/2 MAP-kinase pathway has been implicated to play a role in osteoblast differentiation (6, 8, 10). Here, we report that Erk1/2 MAP-kinase has a negative effect on the expression of a small number of the transcription factors tested, in particular Prrx2, Tieg, Heyl and type II Cbfal. In addition, most late phase bone marker genes, including Prdc, Col6a l and the ESTs AA709999, AA794190 and AA222756, were also negatively regulated by Erk1/2 MAP-kinase. In conclusion, our results support the theory that inhibition of Erk1/2 MAP-kinase promotes osteoblastic differentiation, as postulated previously by Higuchi and coworkers (8). Although it is known that the coordinated action of multiple pathways is required for optimal osteoblast differentiation, only few genes are known to act as focal points for interaction of the various pathways. As described above, Smads have been shown to induce expression of the osteoblast specific transcription factor Cbfal in an indirect manner through a JunB/p38 MAP-kinase cascade (25). In addition, Cbfa1 activation is regulated by Erk1/2 MAP-kinase (6). Therefore, Cbfa1 is one of

- 114 - The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation

the focal points for integration of both the p38 and Erk1/2 MAP-kinase pathways in osteoblast differentiation. In this study, we identified a total of six genes (Tieg, H eyl, Prdc, AA222756, AA794190 and AA709999) that were regulated by both the Smad and Erk1/2 MAP-kinase pathways, indicating that these genes also may act as key regulatory targets that integrate the two pathways in the process of osteoblast differentiation. Tieg is an immediate early TGF-P/BMP response gene that has previously been shown to be regulated during BMP-induced osteoblast differentiation (29). Furthermore, it has been shown that Tieg acts as an enhancer of TGF-P signaling and mimics TGF-P action in human MG-63 osteoblastic cells, which for these cells results in induction of osteoblast differentiation (59, 60). Therefore, it seems likely that Tieg has a similar function in BMP-induced osteoblast differentiation. Hey1 is a transcription factor of the Notch signal transduction pathway and has previously been proposed by us to be the focal point of Smad/Notch signaling during osteoblast differentiation (30). Since we have shown in this report that Heyl is also regulated by Erk1/2 MAP-kinase, this gene might be one of the major regulators of osteoblast differentiation. Indeed, Zamurovic and coworkers have recently shown that Hey1 is a negative regulator of osteoblast differentiation, which exerts its effect by binding to Cbfa1 and thereby inhibits its transcriptional activity (61). Prdc and the ESTs AA222756, AA794190 and AA709999 are markers of late osteoblast differentiation (44), which suggests that these genes might be important in regulating the function of the mature osteoblast. Recently, it has been shown that Prdc is a specific BMP antagonist (62), which suggests that Prdc is a negative regulator of osteoblast function. The Notch signal transduction pathway has been implicated in osteoblast differentiation, although contradictory results have been reported (11, 12, 63). In most studies, over-expression of the NICD has been used to measure the effect of Notch signaling on osteoblast differentiation. However, over-expression induces not only the expression of H eyl, but also that of Hey2, which is not expressed under physiological conditions during BMP-induced osteoblast differentiation in C2C12 cells. As a consequence, over-expression of the NICD may result in aberrant regulation of osteoblast differentiation due to the additional presence of Hey2. Therefore we chose instead to use Fringe proteins to examine the effect of Notch signaling on osteoblast differentiation. This is particularly relevant, since in many cases the distribution of Fringe proteins is the key determinant that controls where and in which form Notch signaling is active (64). Our results indicate that over-expression of LFNG specifically inhibits the expression of intermediate and late phase osteoblast markers whereas myogenic differentiation remains unaffected. Therefore, despite the large variation of Fringe expression levels in individual C2C12 clones and our inability to demonstrate Fringe expression in differentiating bone, we hypothesize that LFNG specifically inhibits the functional differentiation of C2C12 cells into osteoblasts. Most likely, Lfng has no influence on the onset of osteoblast differentiation, but inhibits further progression of the differentiation process so that no functional osteoblasts are formed. Previously, we have shown that during osteoblast differentiation genes with specific functions are coordinately regulated in time (29, 44). These data in combination with the results obtained in this study suggest a model in which multiple rounds of transcription regulation, induced and modulated by various signal transduction pathways, are involved in BMP2-mediated osteoblast differentiation as is outlined in the model presented in Figure 5. The basis of this model is formed by a set of immediate early and intermediate early general BMP-response genes that are regulated at the onset of differentiation. These genes are likely to be direct target genes of BMP2 and are primarily Signal Transduction related genes. The transcription factors expressed at this stage of osteoblast differentiation might subsequently regulate the expression of the late early genes in the second layer of transcription regulation. Here, an increasing number of regulated genes encode Structural proteins

- 115 - Chapter 5

that, in combination with a number of osteoblast-specific transcription factors such as Heyl, Tcf7 and Cbfal, reflect the commitment of the cells to the osteogenic lineage. The ultimate signal for osteoblast differentiation is provided by expression of osteoblast marker genes in the third layer as a result of the action of osteoblast-specific transcription factors. During the entire process, gene expression is regulated, at least in part, by BMP-specific R-Smads, often in combination with Erkl/2 MAP-kinase and Notch signaling, particularly for the expression of late phase bone markers. The effect of Erkl/2 MAP-kinase on expression regulation is probably indirect through Erkl/2-dependent phosphorylation of Cbfal (6) and possibly of other transcription factors.

Figure 5. Model of multi-layered BMP induced osteoblast differentiation For details see text, CHX = cyclohexamide. See page 143 for color figure

In conclusion, our results show that BMP-induced osteoblast differentiation employs a multi­ layered cascade in which the initial phase of differentiation is based on general BMP-response genes. During the progression of osteoblast differentiation, osteoblast commitment genes and osteoblast specific genes are regulated in concert with the general BMP-response genes. In addition, Erkl/2 MAP-kinase and Notch signaling become more and more effective in modulating expression profiles, which ultimately leads to a functional osteoblast characterized by the deposition of a calcified extracellular matrix.

Acknowledgements We thank Wyeth (Andover) for kindly providing recombinant human BMP-2. We are grateful to Dr. H.B.J Karperien (Department of Endocrinology, Leiden University, The Netherlands) for providing the mouse neonatal tails. Furthermore, we would like to Dr. Ester Piek for support and fruitful discussions on this manuscript. This work was supported by the Netherlands Institute for Earth and Life Sciences (NWO-ALW) grant 809.67.023 (to D.S.J.) and grant 809.67.024 (to A.F.).

- ll6 - The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation

References

1. J. M. Wozney, V. Rosen, A. J. Celeste, L. M. Mitsock, M. J. Whitters, R. W. Kriz, R. M. Hewick, and E. A. Wang (1988), Novel regulators of bone formation: molecular clones and activities, Science 242: 1528-1534. 2. B. L. Hogan (1996), Bone morphogenetic proteins: multifunctional regulators of vertebrate development, Genes Dev 10: 1580-1594. 3. J. Massague (1998), TGF-beta signal transduction, Annu Rev Biochem 67: 753-791. 4. C. H. Heldin, K. Miyazono, and P. ten Dijke (1997), TGF-beta signalling from cell membrane to nucleus through SMAD proteins, Nature 390: 465-471. 5. J. Massague, and D. Wotton (2000), Transcriptional control by the TGF-beta/Smad signaling system, EMBO J 19: 1745-1754. 6. G. Xiao, D. Jiang, P. Thomas, M. D. Benson, K. Guan, G. Karsenty, and R. T. Franceschi (2000), MAPK pathways activate and phosphorylate the osteoblast-specific transcription factor, Cbfa1, J Biol Chem 275: 4453-4459. 7. F. Vinals, T. Lopez-Rovira, J. L. Rosa, and F. Ventura (2002), Inhibition of PI3K/p70 S6K and p38 MAPK cascades increases osteoblastic differentiation induced by BMP-2, FEBS Lett 510: 99-104. 8. C. Higuchi, A. Myoui, N. Hashimoto, K. Kuriyama, K. Yoshioka, H. Yoshikawa, and K. Itoh (2002), Continuous inhibition of MAPK signaling promotes the early osteoblastic differentiation and mineralization of the extracellular matrix, J Bone Miner Res 17: 1785-1794. 9. A. Suzuki, J. Guicheux, G. Palmer, Y. Miura, Y. Oiso, J. P. Bonjour, and J. Caverzasio (2002), Evidence for a role of p38 MAP kinase in expression of alkaline phosphatase during osteoblastic cell differentiation, Bone 30: 91-98. 10. S. Gallea, F. Lallemand, A. Atfi, G. Rawadi, V. Ramez, S. Spinella-Jaegle, S. Kawai, C. Faucheu, L. Huet, R. Baron, and S. Roman-Roman (2001), Activation of mitogen-activated protein kinase cascades is involved in regulation of bone morphogenetic protein-2-induced osteoblast differentiation in pluripotent C2C12 cells, Bone 28: 491-498. 11. K. Shindo, N. Kawashima, K. Sakamoto, A. Yamaguchi, A. Umezawa, M. Takagi, K. Katsube, and H. Suda (2003), Osteogenic differentiation of the mesenchymal progenitor cells, Kusa is suppressed by Notch signaling, Exp Cell Res 290: 370-380. 12. M. Sciaudone, E. Gazzerro, L. Priest, A. M. Delany, and E. Canalis (2003), Notch 1 Impairs Osteoblastic Cell Differentiation, Endocrinology 144: 5631-5639. 13. S. Ohta, Y. Hiraki, C. Shigeno, F. Suzuki, R. Kasai, T. Ikeda, H. Kohno, K. Lee, H. Kikuchi, J. Konishi, and et al. (1992), Bone morphogenetic proteins (BMP-2 and BMP-3) induce the late phase expression of the proto-oncogene c­ fos in murine osteoblastic MC3T3-E1 cells, FEBS Lett 314: 356-360. 14. E. Chalaux, T. Lopez-Rovira, J. L. Rosa, R. Bartrons, and F. Ventura (1998), JunB is involved in the inhibition of myogenic differentiation by bone morphogenetic protein-2, J Biol Chem 273: 537-543. 15. A. Hollnagel, V. Oehlmann, J. Heymer, U. Ruther, and A. Nordheim (1999), Id genes are direct targets of bone morphogenetic protein induction in embryonic stem cells, J Biol Chem 274: 19838-19845. 16. N. Kimura, R. Matsuo, H. Shibuya, K. Nakashima, and T. Taga (2000), BMP2-induced apoptosis is mediated by activation of the TAK1-p38 kinase pathway that is negatively regulated by Smad6, J Biol Chem 275: 17647-17652. 17. K. Yamaguchi, S. Nagai, J. Ninomiya-Tsuji, M. Nishita, K. Tamai, K. Irie, N. Ueno, E. Nishida, H. Shibuya, and K. Matsumoto (1999), XIAP, a cellular member of the inhibitor of apoptosis protein family, links the receptors to TAB1- TAK1 in the BMP signaling pathway, EMBO J 18: 179-187. 18. T. Moriguchi, N. Kuroyanagi, K. Yamaguchi, Y. Gotoh, K. Irie, T. Kano, K. Shirakabe, Y. Muro, H. Shibuya, K. Matsumoto, E. Nishida, and M. Hagiwara (1996), A novel kinase cascade mediated by mitogen-activated protein kinase kinase 6 and MKK3, J Biol Chem 271: 13675-13679. 19. A. Nohe, S. Hassel, M. Ehrlich, F. Neubauer, W. Sebald, Y. I. Henis, and P. Knaus (2002), The mode of bone morphogenetic protein (BMP) receptor oligomerization determines different BMP-2 signaling pathways, J Biol Chem 277: 5330-5338. 20. O. Kozawa, D. Hatakeyama, and T. Uematsu (2002), Divergent regulation by p44/p42 MAP kinase and p38 MAP kinase of bone morphogenetic protein-4-stimulated osteocalcin synthesis in osteoblasts, J Cell Biochem 84: 583-589. 21. U. Noth, R. Tuli, R. Seghatoleslami, M. Howard, A. Shah, D. J. Hall, N. J. Hickok, and R. S. Tuan (2003), Activation of p38 and Smads mediates BMP-2 effects on human trabecular bone-derived osteoblasts, Exp Cell Res 291: 201-211. 22. C. F. Lai, and S. L. Cheng (2002), Signal transductions induced by bone morphogenetic protein-2 and transforming growth factor-beta in normal human osteoblastic cells, J Biol Chem 277: 15514-15522. 23. M. Tazoe, M. Mogi, S. Goto, and A. Togari (2003), Involvement of p38MAP kinase in bone morphogenetic protein-4- induced osteoprotegerin in mouse bone-marrow-derived stromal cells, Arch Oral Biol 48: 615-619. 24. Y. Hu, E. Chan, S. X. Wang, and B. Li (2003), Activation of p38 mitogen-activated protein kinase is required for osteoblast differentiation, Endocrinology 144: 2068-2074. 25. K. S. Lee, S. H. Hong, and S. C. Bae (2002), Both the Smad and p38 MAPK pathways play a crucial role in Runx2 expression following induction by transforming growth factor-beta and bone morphogenetic protein, Oncogene 21: 7156-7163. 26. C. F. Lai, L. Chaudhary, A. Fausto, L. R. Halstead, D. S. Ory, L. V. Avioli, and S. L. Cheng (2001), Erk is essential for growth, differentiation, integrin expression, and cell function in human osteoblastic cells, J Biol Chem 276: 14443­ 14450. 27. A. Blokzijl, C. Dahlqvist, E. Reissmann, A. Falk, A. Moliner, U. Lendahl, and C. F. Ibanez (2003), Cross-talk between the Notch and TGF-{beta} signaling pathways mediated by interaction of the Notch intracellular domain with Smad3, J Cell Biol 163: 723-728.

- 117 - Chapter 5

28. C. Dahlqvist, A. Blokzijl, G. Chapman, A. Falk, K. Dannaeus, C. F. Ibanez, and U. Lendahl (2003), Functional Notch signaling is required for BMP4-induced inhibition of myogenic differentiation, Development 130: 6089-6099. 29. D. S. de Jong, B. L. T. Vaes, K. J. Dechering, A. Feijen, J. M. A. Hendriks, R. Wehrens, C. L. Mummery, E. J. van Zoelen, W. Olijve, and W. T. Steegenga (2004), Identification of Novel Regulators Associated with Early-Phase Osteoblast Differentiation, J Bone Miner Res 19: 947-958. 30. D. S. de Jong, W. T. Steegenga, J. M. Hendriks, E. J. van Zoelen, W. Olijve, and K. J. Dechering (2004), Regulation of Notch signaling genes during BMP2-induced differentiation of osteoblast precursor cells, Biochem Biophys Res Commun 320: 100-107. 31. C. Hicks, S. H. Johnston, G. diSibio, A. Collazo, T. F. Vogt, and G. Weinmaster (2000), Fringe differentially modulates Jagged1 and Delta1 signalling through Notch1 and Notch2, Nat Cell Biol 2: 515-520. 32. V. M. Panin, L. Shao, L. Lei, D. J. Moloney, K. D. Irvine, and R. S. Haltiwanger (2002), Notch ligands are substrates for protein O-fucosyltransferase-1 and Fringe, J Biol Chem 277: 29945-29952. 33. T. Iso, L. Kedes, and Y. Hamamori (2003), HES and HERP families: multiple effectors of the Notch signaling pathway, J Cell Physiol 194: 237-255. 34. T. Iso, V. Sartorelli, G. Chung, T. Shichinohe, L. Kedes, and Y. Hamamori (2001), HERP, a new primary target of Notch regulated by ligand binding, M ol Cell Biol 21: 6071-6079. 35. D. S. de Jong, E. J. van Zoelen, S. Bauerschmidt, W. Olijve, and W. T. Steegenga (2002), Microarray analysis of bone morphogenetic protein, transforming growth factor beta, and activin early response genes during osteoblastic cell differentiation, J Bone Miner Res 17: 2119-2129. 36. M. Fujii, K. Takeda, T. Imamura, H. Aoki, T. K. Sampath, S. Enomoto, M. Kawabata, M. Kato, H. Ichijo, and K. Miyazono (1999), Roles of bone morphogenetic protein type I receptors and Smad proteins in osteoblast and chondroblast differentiation, M ol Biol Cell 10: 3801-3813. 37. A. van der Plas, E. M. Aarden, J. H. Feijen, A. H. de Boer, A. Wiltink, M. J. Alblas, L. de Leij, and P. J. Nijweide (1994), Characteristics and properties of osteocytes in culture, J Bone Miner Res 9: 1697-1704. 38. C. W. Lowik, M. J. Alblas, M. van de Ruit, S. E. Papapoulos, and G. van der Pluijm (1993), Quantification of adherent and nonadherent cells cultured in 96-well plates using the supravital stain neutral red, Anal Biochem 213: 426-433. 39. M. M. Deckers, E. R. Van Beek, G. Van Der Pluijm, A. Wetterwald, L. Van Der Wee-Pals, M. G. Cecchini, S. E. Papapoulos, and C. W. Lowik (2002), Dissociation of angiogenesis and osteoclastogenesis during endochondral bone formation in neonatal mice, J Bone Miner Res 17: 998-1007. 40. A. Feijen, M. J. Goumans, and A. J. van den Eijnden-van Raaij (1994), Expression of activin subunits, activin receptors and follistatin in postimplantation mouse embryos suggests specific developmental functions for different activins, Development 120: 3621-3637. 41. D. Chen, X. Ji, M. A. Harris, J. Q. Feng, G. Karsenty, A. J. Celeste, V. Rosen, G. R. Mundy, and S. E. Harris (1998), Differential roles for bone morphogenetic protein (BMP) receptor type IB and IA in differentiation and specification of mesenchymal precursor cells to osteoblast and adipocyte lineages, J Cell Biol 142: 295-305. 42. H. Zou, R. Wieser, J. Massague, and L. Niswander (1997), Distinct roles of type I bone morphogenetic protein receptors in the formation and differentiation of cartilage, Genes Dev 11: 2191-2203. 43. O. Korchynskyi, K. J. Dechering, A. M. Sijbers, W. Olijve, and P. ten Dijke (2003), Gene array analysis of bone morphogenetic protein type I receptor-induced osteoblast differentiation, J Bone Miner Res 18: 1177-1185. 44. B. L. Vaes, K. J. Dechering, A. Feijen, J. M. Hendriks, C. Lefevre, C. L. Mummery, W. Olijve, E. J. van Zoelen, and W. T. Steegenga (2002), Comprehensive microarray analysis of bone morphogenetic protein 2-induced osteoblast differentiation resulting in the identification of novel markers for bone development, J Bone Miner Res 17: 2106-2118. 45. P. Ducy, R. Zhang, V. Geoffroy, A. L. Ridall, and G. Karsenty (1997), Osf2/Cbfa1: a transcriptional activator of osteoblast differentiation, Cell 89: 747-754. 46. J. C. Gevrey, M. Cordier-Bussat, E. Nemoz-Gaillard, J. A. Chayvialle, and J. Abello (2002), Co-requirement of cyclic AMP- and calcium-dependent protein kinases for transcriptional activation of cholecystokinin gene by protein hydrolysates, J Biol Chem 277: 22407-22413. 47. T. V. Hansen, J. F. Rehfeld, and F. C. Nielsen (1999), Mitogen-activated protein kinase and protein kinase A signaling pathways stimulate cholecystokinin transcription via activation of cyclic adenosine 3',5'-monophosphate response element-binding protein, Mol Endocrinol 13: 466-475. 48. M. Baron, H. Aslam, M. Flasza, M. Fostier, J. E. Higgs, S. L. Mazaleyrat, and M. B. Wilkin (2002), Multiple levels of Notch signal regulation, Mol Membr Biol 19: 27-38. 49. S. Zhou, M. Fujimuro, J. J. Hsieh, L. Chen, A. Miyamoto, G. Weinmaster, and S. D. Hayward (2000), SKIP, a CBF1- associated protein, interacts with the ankyrin repeat domain of NotchIC To facilitate NotchIC function, Mol Cell Biol 20: 2400-2410. 50. T. Iso, V. Sartorelli, C. Poizat, S. Iezzi, H. Y. Wu, G. Chung, L. Kedes, and Y. Hamamori (2001), HERP, a novel heterodimer partner of HES/E(spl) in Notch signaling, M ol Cell Biol 21: 6080-6089. 51. M. M. Maier, and M. Gessler (2000), Comparative analysis of the human and mouse Hey1 promoter: Hey genes are new Notch target genes, Biochem Biophys Res Commun 275: 652-660. 52. N. Zhang, and T. Gridley (1998), Defects in somite formation in lunatic fringe-deficient mice, Nature 394: 374-377. 53. K. Serth, K. Schuster-Gossler, R. Cordes, and A. Gossler (2003), Transcriptional oscillation of lunatic fringe is essential for somitogenesis, Genes Dev 17: 912-925. 54. C. Leimeister, A. Externbrink, B. Klamt, and M. Gessler (1999), Hey genes: a novel subfamily of hairy- and Enhancer of split related genes specifically expressed during mouse embryogenesis, Mech Dev 85: 173-177.

- 118 - The Smad, Erk1/2 MAP-kinase and Notch signaling pathways in osteoblast differentiation

55. T. Ogata, J. M. Wozney, R. Benezra, and M. Noda (1993), Bone morphogenetic protein 2 transiently enhances expression of a gene, Id (inhibitor of differentiation), encoding a helix-loop-helix molecule in osteoblast-like cells, Proc Natl Acad Sci USA 90: 9219-9222. 56. O. Korchynskyi, and P. ten Dijke (2002), Identification and functional characterization of distinct critically important bone morphogenetic protein-specific response elements in the Id1 promoter, J Biol Chem 277: 4883-4891. 57. K. Kratochwil, M. Dull, I. Farinas, J. Galceran, and R. Grosschedl (1996), Lef1 expression is activated by BMP-4 and regulates inductive tissue interactions in tooth and hair development, Genes Dev 10: 1382-1394. 58. A. M. Ionescu, H. Drissi, E. M. Schwarz, M. Kato, J. E. Puzas, D. J. McCance, R. N. Rosier, M. J. Zuscik, and R. J. O'Keefe (2004), CREB Cooperates with BMP-stimulated Smad signaling to enhance transcription of the Smad6 promoter, J Cell Physiol 198: 428-440. 59. S. A. Johnsen, M. Subramaniam, R. Janknecht, and T. C. Spelsberg (2002), TGFbeta inducible early gene enhances TGFbeta/Smad-dependent transcriptional responses, Oncogene 21: 5783-5790. 60. T. E. Hefferan, G. G. Reinholz, D. J. Rickard, S. A. Johnsen, K. M. Waters, M. Subramaniam, and T. C. Spelsberg (2000), Overexpression of a nuclear protein, TIEG, mimics transforming growth factor-beta action in human osteoblast cells, J Biol Chem 275: 20255-20259. 61. N. Zamurovic, D. Cappellen, D. Rohner, and M. Susa (2004), Coordinated activation of notch, Wnt, and transforming growth factor-beta signaling pathways in bone morphogenic protein 2-induced osteogenesis. Notch target gene Hey1 inhibits mineralization and Runx2 transcriptional activity, J Biol Chem 279: 37704-37715. 62. S. Sudo, O. Avsian-Kretchmer, L. S. Wang, and A. J. Hsueh (2004), Protein related to DAN and cerberus is a bone morphogenetic protein antagonist that participates in ovarian paracrine regulation, J Biol Chem 279: 23134-23141. 63. K. Tezuka, M. Yasuda, N. Watanabe, N. Morimura, K. Kuroda, S. Miyatani, and N. Hozumi (2002), Stimulation of osteoblastic cell differentiation by Notch, J Bone Miner Res 17: 231-239. 64. N. Haines, and K. D. Irvine (2003), Glycosylation regulates Notch signalling, Nat Rev Mol Cell Biol 4: 786-797.

- 119 - - 120 - 6

General Discussion Chapter 6

The skeleton is continually being regenerated by the concerted action of bone forming cells, the osteoblasts, and bone resorbing cells, the osteoclasts (1). In adults, bone formation and degradation are in a delicate balance, and serious clinical skeletal disorders may arise from a disturbed bone homeostasis. For instance, in so-called sclerosing bone diseases such as osteopetrosis, sclerosteosis and Van Buchem’s disease, the balance has been shifted towards the osteoblasts and too much bone is formed, whereas in osteoporosis this balance has been shifted towards the osteoclasts and too much bone is resorbed. To date, no treatment is available for patients suffering from sclerosing bone diseases, while current therapies for osteoporosis focus primarily on inhibiting the function of the osteoclasts. However, therapies that target induction or enhancement of bone formation may also be effective for osteoporosis patients. In this thesis, we studied the regulation of osteoblast differentiation with the aim to discover new drug targets that may be used for induction of bone formation and that are the basis of novel treatments of skeletal disorders. Mesenchymal stem cells (MSCs) have the potential to differentiate into a variety of cell types including osteoblasts, chondroblasts, myoblasts, adipocytes and fibroblasts, depending on the signaling cascades activated during the initial phase of differentiation (for review see ref. 2). The most important inducer for differentiation of MSCs into osteoblasts are the BMPs, which are known to be involved in various differentiation processes during embryonic development (for review see ref. 3). The observation that BMPs and components of the BMP signaling cascade are involved in skeletal disorders suggests that a search for additional components of the BMP signaling pathways may reveal new genes that also play a role in bone-related diseases. In order to develop effective and selective therapies, it is therefore important to understand the molecular mechanisms of BMP signaling in more detail. Currently, only a few key- or co-regulators of BMP-induced osteoblast differentiation are known. Since such genes may be attractive targets for therapy, it is of great importance to identify additional regulators. Particularly transcription factors are potential key- and co-regulators of osteoblast differentiation because these proteins are generally considered to be major regulators of differentiation processes. Using the gene expression microarray (GEM) technique, it is possible to study gene expression patterns of thousands of genes simultaneously during the differentiation process. By comparing the expression profiles of transcription factors regulated during osteoblast differentiation with those in fibroblasts, it is possible to identify specific key- and co-regulators of osteoblast differentiation.

6.1 Regulators of BMP-2-induced osteoblast differentiation

In this thesis, the identification of novel regulators of BMP-2-induced osteoblast differentiation in mouse cells is described. Microarray analysis revealed a set of 184 transcripts to be regulated by BMP-2 during the first 24 hrs of osteoblast differentiation (Chapter 3, Figure 1A and 1B). In addition, 61 transcripts were identified to be differentially regulated by BMP-2 and TGF-P 24 hrs after treatment of the cells with these growth factors (Chapter 4, Table 1). In addition to established osteoblast differentiation-related genes, we identified various ESTs (expressed sequence tags) as well as several genes that have not been implicated before in osteoblast differentiation, such as the Wnt pathway transcription factor gene Tcf7 (T-cell specific transcription factor 7), and the Notch pathway genes Hey1 (Hairy/enhancer of split related with YRPW motif 1) and Lfng (Lunatic fringe) (Chapters 3 and 4). Evidence is provided that Tcf7 is a positive regulator of osteoblast differentiation (Chapter 3), whereas Lfng is most likely a negative regulator of osteoblast differentiation (Chapters 4 and 5).

- 122 - General Discussion

Since MSCs are also able to differentiate into other cell types such as myoblasts and fibroblasts, it cannot be excluded that Lfng is a positive regulator of the differentiation of MSCs towards one of these cell types, thereby indirectly inhibiting osteoblast differentiation. However, we demonstrated in Chapter 5 that over-expression of Lfng has no effect on myoblast differentiation. This result suggests that Lfng inhibits osteoblast differentiation without promoting differentiation towards other cell types. Heyl has recently also been shown to act as a negative regulator of osteoblast differentiation (4). Potentially, also the ESTs identified in our studies may therefore encode regulators of osteoblast differentiation (the immediate early ESTs) or of osteoblast function (the intermediate and late early ESTs). Elucidating the genes from which these ESTs originate, as well as determining the encoded protein, is therefore an important step in identifying potential additional novel regulators of osteoblast differentiation. Regulation of differentiation processes is not only mediated by the level of transcription modulation of key- or co-factors, but also by the activity and stability of the encoded proteins. In Chapter 5 we show that Lfng is still most likely a negative regulator of osteoblast differentiation although the level of Lfng mRNA transcripts in developing bone is below the detection level. It is therefore important to study the regulation of the protein expression of Lfng and other potential key regulators during osteoblast differentiation. Ducy and others have postulated that osteoblasts are essentially “sophisticated fibroblasts”, that can respond selectively to a set of physiological signals to induce differentiation. As a result, osteoblasts originating from different parts of the body exhibit different gene expression patterns (5, 6). In this thesis, we have shown that many genes that are regulated during early phase osteoblast differentiation, encode proteins that function as signal transduction regulators, enzymes and modulators of cell adhesion, cytoskeleton formation and extracellular matrix formation, and that several members of the same protein family are regulated during osteoblast differentiation (Chapters 3 and 4). It can therefore be postulated that, in order to establish a highly specialized tissue such as bone, a limited set of osteoblast-specific transcription regulators is sufficient for guiding MSCs to become mature osteoblasts provided that they act in combination with general BMP-responsive transcriptional regulators. This so-called “all in the balance” theory requires that both the key-regulatory proteins and general proteins are present at the correct level and time to achieve a balance between inductive and repressive differentiation processes. Any shift in this balance during osteoblast differentiation, or disturbance in the inductive and repressive signals could then lead to either more bone formation, or less bone formation, according to the time at which the disturbance occurs. This balance concept also ensures that at different places in the body, osteoblastic cells can respond differently to physiological signals. As a consequence, osteoblast-specific transcriptional regulators are potentially good candidates for drug targeting. We focused in this study on the transcription factors that are regulated during the early phase of osteoblast differentiation and compared the expression profiles of these genes in differentiating osteoblasts with those in BMP-2 treated fibroblasts. In Chapter 3 (Table 3 and Figures 2 and 3) it is shown that the transcription factors Tcf7 and Hey1 are specifically regulated in differentiating osteoblasts, both in vitro and in vivo. Moreover, over-expression of Tcf7 results in increased osteoblast differentiation as measured by ALP (alkaline phosphatase) activity and calcium deposition (Chapter 3, Figure 4). As mentioned above, Hey1 has been shown to be an inhibitor of osteoblast differentiation. Therefore, Tcf7 might be considered a potential drug target for therapies in which osteoblast differentiation should be stimulated (osteoporosis), whereas Hey1 might be a drug target for bone overgrowth diseases. Potential strategies to develop drugs that regulate Heyl and Tcf7 protein activity are described in more detail below in section 6 .2 .2 .

- 123 - Chapter 6

6.2 BMP-induced gene expression profiles in diagnosis, drug target discovery and therapy

6.2.1 Diagnosing bone diseases Traditionally, the diagnosis of bone-related diseases is based on the clinical analysis of skeletal abnormalities. These abnormalities are usually identified by autoradiography of the skeleton and bone mineral density (BMD) measurements (7). BMD is influenced by both genetic and environmental factors. In recent years, genetic research has provided major breakthroughs in the understanding of hereditary clinical conditions that are characterized by increased bone density (for review see ref. 8). Particularly the positional cloning approach, which aims at the identification of disease-causing genes, has benefited greatly from the increasing DNA sequence information available from the Human Genome Project. Disease-causing genes are genes whose cognate protein is either defective or misexpressed ultimately causing the particular disease. Defective proteins are the consequence of mutations in the DNA sequence of these genes, whereas misexpression (either over- or under-expression, or expression at the wrong point in time) of disease-causing genes is usually the result of mutations in the regulatory DNA sequences of the gene. Misexpression of normal genes can also occur due to the altered activity of gene expression regulatory proteins, but in that case, the latter are designated as the disease-causing genes. In positional cloning, linkage analysis among patient families is used to determine the chromosomal localization of potential disease-causing genes. Subsequently, genes located at the identified genome position are tested for mutations. Using this approach, disease-causing genes have recently been identified for several bone diseases including sclerosteosis (the SOST gene), osteopetrosis (with renal tubular acidosis and cerebral calcifications; the carbonic anhydrase II gene) and Blomstrand dysplasia (a neonatal severe osteosclerotic dysplasia; the PTH receptor 1 gene) (8). In contrast to this “from clinic to gene” approach, it is also possible to identify disease-causing genes using the reverse methodology, referred to as “from gene to clinic”. In the latter approach, disease-causing genes are identified by generating knockout mice, or by down-regulating gene expression by RNA interference. When a knockout mouse is presented with a phenotype that resembles a human skeletal disease, the human orthologue of this particular gene can subsequently be screened for mutations that might cause defective proteins or impaired expression of the protein. For example, knockout mice for Cbfa1 (Core binding factor 1/AML1/Runx2/PEBP2aA) were shown to exibit a phenotype resembling the human disease Cleidocranial dysplasia (9). Since the human orthologue of this gene was shown to map to the chromosome locus 6p21 (10, 11), the locus for Cleidocranial dysplasia (12), it was suggested that mutations in Cbfa1 caused this disease (9). Indeed, subsequent analysis in human patients revealed the existence of several mutations in this gene causing a defective protein, and it was shown that different levels of Cbfa1 protein activity corresponded with Cleidocranial dysplasia symptoms (13). The GEM technique may be an important tool specifically for the identification of misexpressed disease-causing genes. Although to our knowledge this approach has not been described for bone diseases, it is an established approach in cancer diagnosis and establishment of the most optimal therapy for treatment (14-18). In human acute lymphomas, gene expression profiling has been described in combination with in silico clustering of the profiles to identify specific gene expression profiles between acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL) as well as all of the subtypes of ALL (T-ALL, hyperdiploid with >50 chromosomes, BCR-ABL, E2A- PBX1, TEL-AML1 and MLL gene rearrangement) (15, 17). Moreover, within each ALL subgroup, distinct expression profiles were identified that predicted relapse after therapy. Importantly, no single

- 124 - General Discussion

gene was identified to be of use to predict relapse, but rather patterns of expression of a combination of genes were found to be predictive (17). In analogy, MSCs from patients can be harvested from bone, myogenic or fatty tissue, and induced to differentiate into osteoblasts (for review see ref. 19). Gene expression profiles of such “diseased” osteoblasts can then be compared to those of “healthy” osteoblasts using the GEM technique, and differentially expressed genes can be identified. Depending on the time point of expression and the function of the identified genes, it would then be possible to determine whether in a patient with a particular bone-related disease, osteoblast differentiation, function or apoptosis is impaired. The GEM technique used in this manner can therefore become of primary importance for the diagnosis and prognosis of individual patients. The GEM technique can also be used to select appropriate drug targets for therapy, as will be discussed in another section of this chapter (see section 6.2.2). In this thesis, gene expression profiles of a “healthy” mouse osteoblast during the first 24 hrs of differentiation upon BMP treatment have been described. This dataset can be used as a reference to identify bone diseases that are caused by impaired or enhanced osteoblast differentiation according to the methodology described above. However, this approach will only be the first step in identifying misexpressed disease-causing genes in the “from gene to clinic” approach. Subsequent research in mouse models in which potential genes are knocked-out or over-expressed, in combination with mutational analysis in patients, are just one of the steps needed to positively identify the disease- causing genes.

6.2.2 Drug target discovery

Gene expression profiling is a relatively new approach to identify potential drug targets. The most common application compares gene expression levels in diseased tissues with those from normal tissues, obtained from either humans or animal model systems, or by analyzing cells or tissues that were drug treated (20). For example, several investigators have compared gene expression profiles of postmortem samples of the frontal lobes of patients with schizophrenia with those of healthy persons. It was shown in these experiments that the expression of synapsin and of genes involved in both presynaptic function as well as postsynaptic responsiveness is down-regulated in patients with schizophrenia. Furthermore, it was found that five specific metabolic pathways previously not implicated to play a role in schizophrenia were also down-regulated (21-23). These results not only contributed to the understanding of the molecular basis of schizophrenia, but also provided a list of genes that might be potential targets for drug development. Although these GEM experiments have greatly attributed to the understanding of schizophrenia, many questions regarding the initial causal changes leading to the disease remain to be answered. Moreover, it should be noted here that although GEM experiments can provide a list of potential drug target genes, additional research in the field of systematic analysis of gene function, proteomics and high troughput screening of potential drugs has to be performed before effective therapies based on GEM data can be realized. Potential drug targets could also be discovered by gene expression profiling of normal tissue or cells. It is important to note here that drugs should be efficient in enhancing or inhibiting the normal function of its target, preferably specific for its target tissue, and easy to administer. Therefore, clinically relevant drugs include small soluble growth factors or cytokines, antibodies and protein antagonists, while clinically relevant drug targets include cell membrane receptors, nuclear receptors, and small soluble growth factors or cytokines. In order to identify potential drug targets from gene expression profiling of normal cells, it is not only important to know the function of the genes, but also whether they are expressed in a transient or sustained manner. Therefore, time-based

- 125 - Chapter 6

expression profiling in combination with in silico data mining is one of the best options to identify potential drug targets from normal cells. In this thesis, we provide a list of genes that are expressed within the first 24 hrs of BMP- induced osteoblast differentiation (Chapter 3, figure 1). Using in silico data mining, we identified 16 genes encoding receptors, 17 genes encoding transcription factors and 9 genes encoding growth factors/cytokines to be regulated by BMP-2 during early phase osteoblast differentiation. Of the genes encoding receptors, 9 were found to be down-regulated while 7 were up-regulated (Chapter 3). These up-regulated genes include Gpr97 (G-protein coupled receptor 97), an EST highly similar to the frizzled receptor, MuSK2 (Neural fold/somite kinase 2), cholecystokinin and TF (Tissue factor/coagulation factor 3), P-selectin (Platelet selectin), and Gadd45y (growth and differentiation factor 45y). For some of these receptors (including GPR97 and Gadd45Y), no function has yet been established in osteoblast differentiation. Using mouse knock-out and ectopic over-expression studies, as well as RNA inhibition approaches in combination with BMP-induced differentiation experiments, a systematic analysis of the functions of these genes during osteoblast differentiation should first be established. If these genes are indeed involved in osteoblast differentiation, they might be used as potential drug targets and research should focus on identifying small molecules that specifically enhance or inhibit their activity, depending on their role in osteoblast differentiation. Cholecystokinin has been shown to be important for dietary needs (24, 25) and both P-selectin and TF are important factors for angiogenesis. Since both diet and angiogenesis influence bone formation, these receptors might be very important in bone formation. Indeed, both cholecystokinin and TF have been shown to regulate bone formation in vivo (26, 27). Cholecystokinin is a potential negative regulator of cartilage formation in vivo, since an antagonist of this receptor was shown to magnify cartilage formation when administered in vivo. Subsequently, it was shown in rabbit primary chondrocytes that this magnification of cartilage formation was not due to increased chondrocyte differentiation, but enhanced chondrocyte proliferation (maturation) (26). A paste of human recombinant tissue factor (hrTF), autologous calvarial bone chips, platelet- rich plasma and tetracyclin has been shown in a clinical study to induce maxillary bone graft regeneration. Histology of the maxillary bone graft revealed a normal vascularization and bone formation in the bone graft (27). Since TF is an important factor in angiogenesis, it is very likely that hrTF influences the regeneration of the bone graft by regulating blood vessel formation, while having no influence on osteoblast differentiation directly. However, since we observed an increase in TF expression in our experiments, it might also be possible that TF directly influences osteoblast differentiation. To establish whether TF directly influences osteoblast differentiation, further research is needed. Finally, the EST exhibiting high similarity to the Frizzled receptor might constitute a novel member of the Frizzled receptor family. This family is known to transduce Wnt-signals, which has been shown to be very important in regulating osteoblast differentiation (28-34). However, due to the fact that Wnt signaling is also very important for normal embryonic development and tissue formation, Frizzled proteins are not very likely candidates for drug targetting. On the other hand, if this EST indeed constitutes a novel member of the Frizzled receptor family, it might also be possible that it functions in an osteoblast specific manner, rendering it potentially suitable to use as a drug target. Therefore, identifying the gene to which this EST localizes and its cognative protein as well as a systematic analysis of its protein function is required first to establish whether this EST might constitute a potential drug target. In addition to receptors, transcription factors are potential drug targets. In this thesis we identified 17 transcription factors that are regulated during osteoblast differentiation of which 14 are up-regulated by BMP-2 (Chapter 3, figure 1). These include several general BMP-response factors,

- 126 - General Discussion

such as the Id genes (Inhibitor of differentiation genes), as well as two osteoblast specific transcription factors, Hey1 and Tcf7 (Chapter 3, figure 1 and Table 3). As mentioned above, Tcf7 has been shown to be an enhancer of BMP-induced osteoblast differentiation, and might therefore be considered a drug target for therapies in which osteoblast differentiation should be stimulated (osteoporosis). In contrast, Heyl has been shown to be an inhibitor of osteoblast differentiation, and might therefore be a drug target for therapies in which osteoblast differentiation should be inhibited (bone overgrowth diseases). In order for Hey1 and Tcf7 to function as drug targets, it will be necessary to regulate their expression levels and activity indirectly through activation of relevant cell surface receptors. Tcf7 is a member of the Lef/Tcf7 transcription factor family that transduces Wnt signals, and the receptors of the Wnts are the Frizzled proteins (30, 35-37). Therefore, small molecules that induce specific Frizzled proteins or that enhance the function of specific Wnts, could function as potential drugs for regulation of Tcf7 activity. However, due to the fact that Wnt signaling is also very important for normal embryonic development and tissue formation, developing drugs that specifically target Tcf7 in osteoblasts will be very challenging. Hey1 is a member of the Hes/Herp family of transcription factors that transduce Notch signals (38). Both the Notch receptor and its ligand are transmembrane proteins (39) and are therefore potential drug targets. However, soluble ligands are known to inhibit Notch signaling instead of activating it (39), and therefore developing drugs that stimulate Notch signaling and hence stimulate the function of Hey1 through its receptors will be very difficult. In contrast to the stimulation of Hey1 activity in therapies in which osteoblast differentiation should be inhibited, down-regulation of Hey1 activity might be used in therapies in which osteoblast differentiation should be stimulated. In these therapies, inhibition of Hey1 activity through the use of soluble ligands might abolish its negative regulation of osteoblast differentiation, and might therefore stimulate osteoblast differentiation. Other potential drug targets are cytokines and other soluble extracellular factors. In this thesis, we identified several of such factors that were regulated during osteoblast differentiation (Chapter 3). These include amongst others Vegf (vascular endothelial growth factor), Plgf (placental growth factor), Mcp-1/Scya2 (Monocyte chemoattractant protein 1/Small inducible cytokine A2) and Mcsf (Macrophage colony stimulating factor). Vegf is known to stimulate angiogenesis and has recently been described to promote bone formation in tissue engineering (see section 6.2.3). As mentioned above, bone formation and bone homeostasis are the net result of a balance between the activities of osteoblasts and osteoclasts. These activities are coupled to each other, since differentiated osteoblasts express factors such as Rankl that promote osteoclast differentiation from monocyte/macrophage precursors (40). In this respect, the three growth factors Plgf, Mcp-1 and Mcsf have all been shown to be expressed by osteoblasts (41-43), and Mcp-1 and Mcsf have been shown to regulate bone formation by positively regulating the differentiation and activity of osteoclasts (42, 43). In addition, Plgf has also been proposed to regulate osteoclast differentiation or function since it has been shown to activate monocytes during inflammation (41, 44). Therefore, the transient up- regulation of these three growth factors by BMP-2 as shown in this thesis (Chapter 3, figure 1), might be linked to the onset of recruitment and differentiation of osteoclast precursors, similarly as happens in vivo. On the other hand, we and others have shown that Mcp-1 is able to enhance BMP-2-induced osteoblast differentiation and function, implying that the transient expression of Plgf, Mcp-1 and M csf might also be necessary for the induction of osteoblast differentiation (M. Kuijer, personal communication; E.A. Kaiser, abstract on ASBMR 2003 meeting). As a consequence, these three factors might be potential drug targets for bone diseases. In order to investigate their role in osteoblast differentiation, mouse knockout and ectopic over-expression experiments, as well as RNA inhibition approaches coupled to BMP-induced osteoblast differentiation experiments are required. The next

- 127 - Chapter 6

step in drug development will then be the development of small molecules resembling Plgf, Mcp-1 or Mcsf, which specifically target osteoblasts and not osteoclasts. To asses the efficiency of potential new drugs BMD measurements can be used in combination with measurements on bone strength (fracture rate). However, these methods require long term follow-up procedures. Therefore, researchers are now focusing on the identification of “biomarkers” that can be tested quickly and efficiently and give a good prognosis on the efficacy of the treatment. Gene expression profiling is one of the methods that might be used to monitor the function of potential drugs on bone formation. In this thesis, we provide a list of genes that are regulated by BMP-2 during the early phase of osteoblast differentiation (Chapter 3, Figure 1). Researchers who want to establish whether a potential drug is able to affect osteoblast differentiation, can use this list as a reference by which they can correlate gene expression profiles of drug-treated ‘healthy’ cells and drug-treated ‘diseased’ cells. Alternatively, the list may be used to select a number of marker genes to test the effect of the drug on osteoblast differentiation (“biomarkers”). However, it should be noted here, that also these new biomarkers should first be validated clinically before they can be used in general to predict efficiency of novel drugs on osteoblast differentiation and function.

6.2.3 Tissue engineering therapies

Millions of people all over the world suffer from skeletal tissue loss due to injury, disease or congenital malformation (45). With a progressively aging population, there is an increasing demand for therapies to regenerate or replace the lost tissue. Bone grafting and joint prostheses have been effective in alleviating much of the disability associated with loss of bone or cartilage, but these therapies suffer from serious limitations such as lack of either autogenic (from the patient itself) or allogenic (from another person) donor tissue, rejection of allogenic or alloplastic (fabricated) donor tissue or susceptibility to infections. Therefore, in recent years, much research has focused on tissue engineering approaches for the development of biological substitutes in tissue replacement. Tissue-engineering strategies utilize specific combinations of bioactive factors, scaffolds, and cells. In skeletal tissue engineering, recombinant BMP-2 and BMP-7/OP-1 have been used successfully to treat open tibia fractures, tibial nonunions and to enhance fusion of the spine (for review see ref. 46). However, supraphysiological doses of these recombinant proteins are required for the induction of bone formation, which renders this procedure inefficient. Moreover, in difficult clinical situations where there is not much bone left, a single dose of recombinant protein may not provide a sufficient osteogenic response (46). Due to the limitations on using bone grafts as scaffolds, current research on tissue engineering focuses on the use of synthetic, preferentially biodegradable, sponge-like scaffolds that provide a solid framework for cell growth and differentiation at a local site, allowing cell attachment and migration. The scaffold may be implanted alone to induce host cell migration to the wound site and initiate tissue regeneration, but more frequently, tissue engineers focus on the inclusion in the scaffolds of cell types capable of differentiating to osteoblasts (for review see ref. 47). Cells can be obtained from a number of sources at various stages of differentiation, and, in many cases, tissue regeneration requires or is enhanced by the introduction of biochemical signals, in the form of growth factors, cytokines or other environmental cues, such as mechanical stimulation, which direct cellular responses including proliferation, differentiation and matrix synthesis (47).

- 128 - General Discussion

6.2.3.1 Stem cells in bone tissue engineering Numerous cell types are available for tissue engineering strategies, including embryonic stem cells (ES cells), mesenchymal stem cells and differentiated osteoblast cells. ES cells are the most optimal type of cells to use for tissue engineering, since they exhibit unlimited proliferation in vitro and have the ability to differentiate into any cell type in the human body. However, due to scientific, legal, political and ethical questions regarding the use of these ES cells, alternative sources such as bone marrow, muscle, fat and brain are currently investigated for obtaining mesenchymal stem cells (MSCs), which are capable to differentiate into osteoblasts (46). Although transplanted stem cells induce tissue regeneration, until recently no prove could be found for the integration of these transplanted stem cells into the new tissue. Recently, however, several groups have shown that transplanted neural progenitor cells as well as bone marrow stromal cells (BMSCs) not only contribute to tissue regeneration, but differentiate and integrate into the newly formed tissue (48-51). Both in animal and human studies BMSCs have been reported to enhance critical size bone fracture repair (for review see ref. 19). Critical size bone fractures are fractures that are too large to heal naturally. Usually, tissue engineering is required for the repair of these fractures. The crucial step in this BMSC-based procedure is the harvesting and ex vivo expansion of the BMSCs. Expansion of the BMSC population is generally required since the harvesting of BMSCs (preferentially autogenic) is a painful procedure and there is a limit on donor sites suitable for harvesting the BMSCs. On the other hand, harvesting MSCs from muscle or fat is a much less invasive procedure and usually yields a larger pool of stem cells. Ex vivo expansion of MSCs generally leads to a progressive decrease of proliferation and loss of the multilineage differentiation potential (19). Therefore, the development of optimal culture conditions to select or isolate a stem cell subpopulation, that can self-renew for a long period of time while maintaining its multi-differentiation potential, is very important for using MSCs in therapy. An approach that has been used to maintain the indefinite, multipotential proliferation of MSCs is to transfect these cells with the human telomerase reverse transcriptase (hTERT) gene to overcome telomere shortening and senescence (19). These hTERT-transduced human MSCs, when implanted in immunodeficient mice, showed enhanced bone formation compared to normal cells (52, 53), indicating that this procedure can be used to stimulate bone formation. However, reports on the carcinogenic capacity of these cells are contradictory (52, 54, 55). Therefore, this approach needs further investigation before it can be implemented in human therapy. The finding that carrier-bound recombinant BMPs can be used to restore bone defects suggests that BMPs are able to recruit autologous mesenchymal stem cells to the site of injury and induce their differentiation into osteoblasts. It is not known whether this process depletes the local stock of stem cells or whether BMPs are able to initiate differentiation of the stem cell population while also maintaining the pool of undifferentiated cells by inducing self-renewal. It has recently been described that BMPs support the self-renewal of ES cells, but the precise mechanism remains unclear (56, 57). In this thesis, we observed that BMP-2 induces the expression of Sca-1/Ly6a (Stem cell antigen 1/lymphocyte antigen 6) 24 hrs after treatment, while this expression was down-regulated upon TGF-P treatment (Chapter 4). Recently, it has been reported that Sca-1/Ly6a is required for the self-renewal of mesenchymal stem cells (58), and therefore our results indicate that BMPs may be capable of maintaining and regulating the pool of undifferentiated stem cells by regulating Sca- 1/Ly6a expression during osteoblast differentiation. In light of these findings, BMPs might possibly be used to optimize the expansion of multi-potent MSC cultures, for example by applying BMP at concentrations that are insufficient to induce osteoblast differentiation. In addition to engineering MSCs to maintain their multi-potentiality and proliferation capacity, MSCs have been transfected with BMP genes and introduced in vivo in order to enhance their own differentiation capacity as well as that of local stem cells, and thereby to enhance the

- 129 - Chapter 6

effectiveness of bone therapy (46). Although it has been shown that these BMP-expressing MSCs are much more efficient in inducing osteoblast differentiation and repairing bone defects than normal MSCs, the efficiency of this approach is less than the often used but very invasive bone grafting therapy. This suggests that the former approach still needs further optimization. Indeed, it has recently been reported that using a pool of BMP-4-expressing MSCs and VEGF-expressing MSCs is much more potent in restoring bone defects than BMP-4-expressing MSCs alone (59). Important to note here is that the ratio of VEGF-expressing cells to BMP-4-expressing cells was critical to ensure the enhancement of bone healing elicited by these cells. In other words, too much expression of VEGF compared to BMP-4 inhibited bone formation instead of stimulating it (59). The use of genetically engineered MSCs should therefore not only be tested for the short-term effects, such as restoration of bone formation within a couple of weeks, but also for long-term effects in order to exclude the occurrence of detrimental side effects. In this thesis, we have identified several genes that have not previously been correlated with osteoblast differentiation (Chapters 3 and 4). Moreover, we have shown that the product of one of these genes, the transcription factor Tcf7, is an enhancer of BMP-induced osteoblast differentiation (Chapter 3). Therefore, this factor might be used to further optimize MSC-based bone therapies, e.g. by engineering MSCs expressing Tcf7 and using these Tcf7-expressing MSCs in combination with the BMP- and VEGF-expressing MSCs. In addition to Tcf7, the lists of genes involved in osteoblast differentiation provided by us could also help researchers to identify additional genes that enhance osteoblast differentiation and might therefore be used in engineering MSCs for various bone therapies. In conclusion, gene expression profiles can help to optimize diagnosis, drug target discovery and tissue engineering therapies of bone diseases. Novel key- and co-regulators of BMP-induced osteoblast differentiation have been identified and some of the many genes regulated during osteoblast differentiation may be novel potential drug targets. To investigate the actual potential of these genes for use in therapies in more detail, a comprehensive analysis of their function during osteoblast differentiation is required first. This may be achieved by using mouse knockout and ectopic over-expression studies, as well as by RNA inhibition strategies in combination with BMP-induced differentiation experiments. Subsequently, MSCs expressing drug target genes might be developed, and tested for efficiency in restoring bone formation in vivo. Furthermore, small biochemical molecules could be engineered that enhance, inhibit or mimic specific functions of drug targets, and subsequently tested for their efficiency in restoring bone formation in vivo.

References

1. R. Baron (2003), General Principles of Bone Biology, in Primer on the Metabolic Bone Diseases and Disorders of Mineral Metabolism, in Favus, M. J., (Ed.), American Society for Bone and Mineral Research, Washington, DC, pp. 1-8. 2. D. J. Prockop (1997), Marrow stromal cells as stem cells for nonhematopoietic tissues, Science 276: 71-74. 3. B. L. Hogan (1996), Bone morphogenetic proteins: multifunctional regulators of vertebrate development, Genes Dev 10: 1580-1594. 4. N. Zamurovic, D. Cappellen, D. Rohner, and M. Susa (2004), Coordinated activation of notch, Wnt, and transforming growth factor-beta signaling pathways in bone morphogenic protein 2-induced osteogenesis. Notch target gene Hey1 inhibits mineralization and Runx2 transcriptional activity, J Biol Chem 279: 37704-37715. 5. J. B. Lian, G. S. Stein, and J. E. Aubin (2003), Bone Formation: Maturation and Functional Activities of Osteoblast Lineage Cells, in Primer on the Metabolic Bone Diseases and Disorders of Mineral Metabolism, in Favus, M. J., (Ed.), American Society for Bone and Mineral Research, Washington, DC, pp. 13-28. 6. P. Ducy, T. Schinke, and G. Karsenty (2000), The osteoblast: a sophisticated fibroblast under central surveillance, Science 289: 1501-1504. 7. W. Van Hul, F. Vanhoenacker, W. Balemans, K. Janssens, and A. M. De Schepper (2001), Molecular and radiological diagnosis of sclerosing bone dysplasias, Eur J Radiol 40: 198-207. 8. W. Van Hul (2003), Recent progress in the molecular genetics of sclerosing bone dysplasias, Pediatr Pathol Mol Med 22: 11-22.

- 130 - General Discussion

9. T. Komori, H. Yagi, S. Nomura, A. Yamaguchi, K. Sasaki, K. Deguchi, Y. Shimizu, R. T. Bronson, Y. H. Gao, M. Inada, M. Sato, R. Okamoto, Y. Kitamura, S. Yoshiki, and T. Kishimoto (1997), Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts, Cell 89: 755-764. 10. D. Levanon, V. Negreanu, Y. Bernstein, I. Bar-Am, L. Avivi, and Y. Groner (1994), AML1, AML2, and AML3, the human members of the runt domain gene-family: cDNA structure, expression, and chromosomal localization, Genomics 23: 425-432. 11. Y. W. Zhang, S. C. Bae, E. Takahashi, and Y. Ito (1997), The cDNA cloning of the transcripts of human PEBP2alphaA/CBFA1 mapped to 6p12.3-p21.1, the locus for cleidocranial dysplasia, Oncogene 15: 367-371. 12. S. Mundlos, J. B. Mulliken, D. L. Abramson, M. L. Warman, J. H. Knoll, and B. R. Olsen (1995), Genetic mapping of cleidocranial dysplasia and evidence of a microdeletion in one family, Hum Mol Genet 4: 71-75. 13. F. Otto, H. Kanegane, and S. Mundlos (2002), Mutations in the RUNX2 gene in patients with cleidocranial dysplasia, Hum Mutat 19: 209-216. 14. C. M. Perou, S. S. Jeffrey, M. van de Rijn, C. A. Rees, M. B. Eisen, D. T. Ross, A. Pergamenschikov, C. F. Williams, S. X. Zhu, J. C. Lee, D. Lashkari, D. Shalon, P. O. Brown, and D. Botstein (1999), Distinctive gene expression patterns in human mammary epithelial cells and breast cancers, Proc Natl Acad Sci USA 96: 9212-9217. 15. T. R. Golub, D. K. Slonim, P. Tamayo, C. Huard, M. Gaasenbeek, J. P. Mesirov, H. Coller, M. L. Loh, J. R. Downing, M. A. Caligiuri, C. D. Bloomfield, and E. S. Lander (1999), Molecular classification of cancer: class discovery and class prediction by gene expression monitoring, Science 286: 531-537. 16. S. L. Pomeroy, P. Tamayo, M. Gaasenbeek, L. M. Sturla, M. Angelo, M. E. McLaughlin, J. Y. Kim, L. C. Goumnerova, P. M. Black, C. Lau, J. C. Allen, D. Zagzag, J. M. Olson, T. Curran, C. Wetmore, J. A. Biegel, T. Poggio, S. Mukherjee, R. Rifkin, A. Califano, G. Stolovitzky, D. N. Louis, J. P. Mesirov, E. S. Lander, and T. R. Golub (2002), Prediction of central nervous system embryonal tumour outcome based on gene expression, Nature 415: 436-442. 17. E. J. Yeoh, M. E. Ross, S. A. Shurtleff, W. K. Williams, D. Patel, R. Mahfouz, F. G. Behm, S. C. Raimondi, M. V. Relling, A. Patel, C. Cheng, D. Campana, D. Wilkins, X. Zhou, J. Li, H. Liu, C. H. Pui, W. E. Evans, C. Naeve, L. Wong, and J. R. Downing (2002), Classification, subtype discovery, and prediction of outcome in pediatric acute lymphoblastic leukemia by gene expression profiling, Cancer Cell 1: 133-143. 18. U. Scherf, D. T. Ross, M. Waltham, L. H. Smith, J. K. Lee, L. Tanabe, K. W. Kohn, W. C. Reinhold, T. G. Myers, D. T. Andrews, D. A. Scudiero, M. B. Eisen, E. A. Sausville, Y. Pommier, D. Botstein, P. O. Brown, and J. N. Weinstein (2000), A gene expression database for the molecular pharmacology of cancer, Nat Genet 24: 236-244. 19. A. R. Derubeis, and R. Cancedda (2004), Bone marrow stromal cells (BMSCs) in bone engineering: limitations and recent advances, Ann Biomed Eng 32: 160-165. 20. R. Kramer, and D. Cohen (2004), Functional genomics to new drug targets, Nat Rev Drug Discov 3: 965-972. 21. K. Mirnics, F. A. Middleton, A. Marquez, D. A. Lewis, and P. Levitt (2000), Molecular characterization of schizophrenia viewed by microarray analysis of gene expression in prefrontal cortex, Neuron 28: 53-67. 22. Y. Hakak, J. R. Walker, C. Li, W. H. Wong, K. L. Davis, J. D. Buxbaum, V. Haroutunian, and A. A. Fienberg (2001), Genome-wide expression analysis reveals dysregulation of myelination-related genes in chronic schizophrenia, Proc Natl Acad Sci USA 98: 4746-4751. 23. K. Mirnics, F. A. Middleton, D. A. Lewis, and P. Levitt (2001), Analysis of complex brain disorders with gene expression microarrays: schizophrenia as a disease of the synapse, Trends Neurosci 24: 479-486. 24. A. S. Kopin, W. F. Mathes, E. W. McBride, M. Nguyen, W. Al-Haider, F. Schmitz, S. Bonner-Weir, R. Kanarek, and M. Beinborn (1999), The cholecystokinin-A receptor mediates inhibition of food intake yet is not essential for the maintenance of body weight, J Clin Invest 103: 383-391. 25. J. R. Grider (1994), Role of cholecystokinin in the regulation of gastrointestinal motility, J Nutr 124: 1334S-1339S. 26. M. Ochi, K. Kawasaki, H. Kataoka, Y. Uchio, and H. Nishi (2001), AG-041R, a gastrin/CCK-B antagonist, stimulates chondrocyte proliferation and metabolism in vitro, Biochem Biophys Res Commun 283: 1118-1123. 27. P. Philippart, M. Brasseur, D. Hoyaux, and R. Pochet (2003), Human recombinant tissue factor, platelet-rich plasma, and tetracycilne induce a high-quality human bone graft: a 5-year survey, Int J OralMaxillofacImplants 18: 411-416. 28. L. Fischer, G. Boland, and R. S. Tuan (2002), Wnt-3A enhances bone morphogenetic protein-2-mediated chondrogenesis of murine C3H10T1/2 mesenchymal cells, J Biol Chem 277: 30870-30878. 29. P. V. Bodine, W. Zhao, Y. P. Kharode, F. J. Bex, A. J. Lambert, M. B. Goad, T. Gaur, G. S. Stein, J. B. Lian, and B. S. Komm (2004), The Wnt antagonist secreted frizzled-related protein-1 is a negative regulator of trabecular bone formation in adult mice, Mol Endocrinol 18: 1222-1237. 30. Y. Yang (2003), Wnts and wing: Wnt signaling in vertebrate limb development and musculoskeletal morphogenesis, Birth Defects Res Part C Embryo Today 69: 305-317. 31. G. Bain, T. Muller, X. Wang, and J. Papkoff (2003), Activated beta-catenin induces osteoblast differentiation of C3H10T1/2 cells and participates in BMP2 mediated signal transduction, Biochem Biophys Res Commun 301: 84-91. 32. J. Kitagaki, M. Iwamoto, J. G. Liu, Y. Tamamura, M. Pacifci, and M. Enomoto-Iwamoto (2003), Activation of beta- catenin-LEF/TCF signal pathway in chondrocytes stimulates ectopic endochondral ossification, Osteoarthritis Cartilage 11: 36-43. 33. J. Galceran, I. Farinas, M. J. Depew, H. Clevers, and R. Grosschedl (1999), Wnt3a-/--like phenotype and limb deficiency in Lef1(-/-)Tcf1(-/-) mice, Genes Dev 13: 709-717. 34. P. Babij, W. Zhao, C. Small, Y. Kharode, P. J. Yaworsky, M. L. Bouxsein, P. S. Reddy, P. V. Bodine, J. A. Robinson, B. Bhat, J. Marzolf, R. A. Moran, and F. Bex (2003), High bone mass in mice expressing a mutant LRP5 gene, J Bone Miner Res 18: 960-974.

- 131 - Chapter 6

35. M. Kuhl (2004), The WNT/Calcium pathway: biochemical mediators, tools and future requirements, Front Biosci 9: 967-974. 36. A. Bejsovec (2000), Wnt signaling: an embarrassment of receptors, Curr Biol 10: R919-922. 37. H. Clevers, and M. van de Wetering (1997), TCF/LEF factor earn their wings, Trends Genet 13: 485-489. 38. T. Iso, L. Kedes, and Y. Hamamori (2003), HES and HERP families: multiple effectors of the Notch signaling pathway, J Cell Physiol 194: 237-255. 39. M. Baron, H. Aslam, M. Flasza, M. Fostier, J. E. Higgs, S. L. Mazaleyrat, and M. B. Wilkin (2002), Multiple levels of Notch signal regulation, Mol Membr Biol 19: 27-38. 40. T. Katagiri, and N. Takahashi (2002), Regulatory mechanisms of osteoblast and osteoclast differentiation, Oral Dis 8: 147-159. 41. S. Marrony, F. Bassilana, K. Seuwen, and H. Keller (2003), Bone morphogenetic protein 2 induces placental growth factor in mesenchymal stem cells, Bone 33: 426-433. 42. D. T. Graves, Y. Jiang, and A. J. Valente (1999), Regulated expression of MCP-1 by osteoblastic cells in vitro and in vivo, Histol Histopathol 14: 1347-1354. 43. M. G. Cecchini, W. Hofstetter, J. Halasy, A. Wetterwald, and R. Felix (1997), Role of CSF-1 in bone and bone marrow development, MolReprodDev 46: 75-84. 44. N. Perelman, S. K. Selvaraj, S. Batra, L. R. Luck, A. Erdreich-Epstein, T. D. Coates, V. K. Kalra, and P. Malik (2003), Placenta growth factor activates monocytes and correlates with sickle cell disease severity, Blood 102: 1506­ 1514. 45. R. Langer, and J. P. Vacanti (1993), Tissue engineering, Science 260: 920-926. 46. S. C. Gamradt, and J. R. Lieberman (2004), Genetic modification of stem cells to enhance bone repair, Ann Biomed Eng 32: 136-147. 47. B. Sharma, and J. H. Elisseeff (2004), Engineering structurally organized cartilage and bone tissues, Ann Biomed Eng 32: 148-159. 48. M. Tohill, C. Mantovani, M. Wiberg, and G. Terenghi (2004), Rat bone marrow mesenchymal stem cells express glial markers and stimulate nerve regeneration, Neurosci Lett 362: 200-203. 49. Y. Fujiwara, N. Tanaka, O. Ishida, Y. Fujimoto, T. Murakami, H. Kajihara, Y. Yasunaga, and M. Ochi (2004), Intravenously injected neural progenitor cells of transgenic rats can migrate to the injured spinal cord and differentiate into neurons, astrocytes and oligodendrocytes, Neurosci Lett 366: 287-291. 50. Y. Y. Jang, and S. J. Sharkis (2004), Metamorphosis from bone marrow derived primitive stem cells to functional liver cells, Cell Cycle 3: 980-982. 51. J. Liu, Q. Hu, Z. Wang, C. Xu, X. Wang, G. Gong, A. Mansoor, J. Lee, M. Hou, L. Zeng, J. R. Zhang, M. Jerosch- Herold, T. Guo, R. J. Bache, and J. Zhang (2004), Autologous stem cell transplantation for myocardial repair, Am J Physiol Heart Circ Physiol 287: H501-511. 52. J. L. Simonsen, C. Rosada, N. Serakinci, J. Justesen, K. Stenderup, S. I. Rattan, T. G. Jensen, and M. Kassem (2002), Telomerase expression extends the proliferative life-span and maintains the osteogenic potential of human bone marrow stromal cells, Nat Biotechnol 20: 592-596. 53. S. Shi, S. Gronthos, S. Chen, A. Reddi, C. M. Counter, P. G. Robey, and C. Y. Wang (2002), Bone formation by human postnatal bone marrow stromal stem cells is enhanced by telomerase expression, Nat Biotechnol 20: 587-591. 54. N. Serakinci, P. Guldberg, J. S. Burns, B. Abdallah, H. Schrodder, T. Jensen, and M. Kassem (2004), Adult human mesenchymal stem cell as a target for neoplastic transformation, Oncogene 23: 5095-5098. 55. Y. Xiaoxue, C. Zhongqiang, G. Zhaoqing, D. Gengting, M. Qingjun, and W. Shenwu (2004), Immortalization of human osteoblasts by transferring human telomerase reverse transcriptase gene, Biochem Biophys Res Commun 315: 643-651. 56. X. Qi, T. G. Li, J. Hao, J. Hu, J. Wang, H. Simmons, S. Miura, Y. Mishina, and G. Q. Zhao (2004), BMP4 supports self-renewal of embryonic stem cells by inhibiting mitogen-activated protein kinase pathways, Proc Natl Acad Sci USA 101: 6027-6032. 57. Q. L. Ying, J. Nichols, I. Chambers, and A. Smith (2003), BMP induction of Id proteins suppresses differentiation and sustains embryonic stem cell self-renewal in collaboration with STAT3, Cell 115: 281-292. 58. M. Bonyadi, S. D. Waldman, D. Liu, J. E. Aubin, M. D. Grynpas, and W. L. Stanford (2003), Mesenchymal progenitor self-renewal deficiency leads to age-dependent osteoporosis in Sca-1/Ly-6A null mice, Proc Natl Acad Sci USA 100: 5840-5845. 59. H. Peng, V. Wright, A. Usas, B. Gearhart, H. C. Shen, J. Cummins, and J. Huard (2002), Synergistic enhancement of bone formation and healing by stem cell-expressed VEGF and bone morphogenetic protein-4, J Clin Invest 110: 751­ 759.

- 132 - 7

Full color figures Chapter 7

Figure 3 (C hapter 1). Signal transduction via BMP receptors and regulation of BMP-induced signaling. Signal transduction via BMP receptors can be mediated 1) by binding of BMP to two preformed type I/II dimeric complexes, subsequently inducing a conformational change that activates this complex, or 2) by binding of BMP to high affinity type I homo- /heterodimeric receptor complexes, subsequently recruiting unliganded BMPR-II homodimers into the complex. Furthermore, each of the different receptor oligomers may allow association of different cytoplasmatic proteins (indicated as X, Y and Z) which may participate in the regulation of distinct pathways. See text for details. Regulation of BMP-induced signaling is achieved by I) extracellular agonists and antagonists, II) pseudoreceptors, III) inhibitory Smads, IV) Smad binding proteins and IV) proteasomal degradation. See text for details. (Adopted from Canalis et al. (2003), see ref. 209)

- 134 - Full color figures

III Transcription regulation of o-Fos, c-Jun, JunB

Figure 4 (C hapter 1). Interactions of the MAPKinase and Smad signal transduction pathways. To regulate osteoblast differentiation, multiple points of convergence between the MAPKinase and Smad pathways can be identified: I) BMPs differentially activate p38, Erk1/2 and Jnk, II) BMPs activate individual API components, III) BMPs regulate the expression of API components, IV) p38 MAPKinase stimulates Smad-1 phosphorylation, V) Erk1/2 MAPKinase activates the Cbfal protein, VI) Erk1/2 MAPKinase can phosphorylate R-Smads, and VII) Erk1/2 and Jnk MAPKinase can regulate the activity of Smad binding proteins. See text for details.

- 135 - Chapter 7

Figure 5 (C hapter 1). Interactions of the Wnt and Smad signal transduction pathways. Points of convergence between the Wnt and Smad pathway: I) BMPs induce the expression of Wnt signal transduction genes including Wnt, Frizzled and Lef1, II) Smad-4 is a co-activator of p-catenin, III) Lef/p-catenin regulates activity of Cbfa1, and IV) b-catenin regulates the activity of Smad binding proteins. See text for details.

- 136 - Full color figures

Delta, Serrate/Jagged

direct target genes Hes-1 and Hey1

Figure 6 (Chapter 1). Interaction of the Notch and Smad signal transduction pathway. The Notch intracellular domain (NICD) interacts with R-Smad/Co-Smad complexes in the nucleus thereby inducing the expression o f Notch transcription factors such as Hes-1. See text for details.

- 137 - Chapter 7

Table 3 (Chapter 2). BMP2, TGF-(3l and activin A exhibit different sets of significantly regulated early genes (p < 0.01). Genbank Category Accession no. Gene Name BMP2 T G F-pi activin A I B t AA002322 EST 1.5 1.0 1.0 A I894068 VI m. Fos-like antigen 2 1.3 1.0 1.0 II b ,t T W 88094 Mm. Chemokine (C-X-C) receptor 4 2.4 1.7 1.0 A A II8626 EST 2.0 2.3 1.1 A A060806 Mm. TSC22-related inducible leucine zipper lb 1.9 1.7 1.0 W 8220I EST 1.7 1.3 1.0 W 13229 EST 1.6 1.7 1.1 A A II9479 Mm. Lymphoid enhancer binding factor 1 1.6 1.7 1.1 A I510010 EST 1.6 1.7 1.0 AA259873 EST 1.6 1.3 1.1 AI466603 EST 1.5 1.6 1.0 AA547162 Mm. Forkhead box Ol 1.5 1.5 1.1 A A033308 EST 1.5 1.5 1.1 W 98395 Mm. Discoidin domain receptor 1 1.5 1.3 1.0 A A423149 VI m. Cyr61 1.5 1.4 1.0 A A061644 Mm. Smad 7 1.4 1.5 1.0 W 14184 EST 1.4 1.5 1.0 W 852I7 EST 1.3 1.6 1.0 A A I84015 EST 1.3 1.5 1.1 A AI25391 EST 1.3 1.4 1.1 III B,T,a T AA036380 Mm. Snail homolog 2.2 3.7 A A I44094 EST 2.2 3.2 1.9 A A I84207 EST 1.8 2.7 1.5 AA060863 EST 1.8 1.5 1.3 W 85374 EST 1.8 1.5 1.3 A A098166 Mm. Placental growth factor 1.7 1.5 1.2 W 64388 Mm. IYI v(1118 1.5 3.5 1.4 AI595201 Mm. HB-EGF 1.5 2.8 1.3 W 45975 Mm. TNF receptor lb 1.5 2.0 1.2 AA271373 Mm. Heparan sulfate 6-O-sulfotransferase 1 1.5 1.4 1.2 AA064241 Mm. Stimulated by retinoic acid 1.4 3.6 1.9 A A476158 VI m. Tieg 1.4 2.2 1.5 W 3654I Mm. Connective tissue growth factor 1.4 1.9 1.2 A A245505 VI m. Citokine inducible SH2 protein 2 1.4 1.4 1.2 A A I77949 VI m. Solute carrier family 20, member 1 1.3 1.9 1.3 A A472299 EST 1.3 1.8 1.4 AA444621 EST 1.3 1.7 1.3 A I509105 EST 1.3 1.7 1.2 AA036347 VI m. Kruppel-like factor 9 1.3 1.6 1.3 AA386857 EST 1.3 1.6 1.3 AA275737 Mm. Retinal short-chain dehydrogenase/reductase 1 1.3 1.6 1.2 W 66845 EST 1.3 1.6 1.2 A A I45784 Mm. Timp3 1.3 1.5 1.2 A A260654 Mm. Tg interacting factor 1.2 2.4 1.6 AA030427 EST 1.2 1.5 1.3 W 98074 EST 1.2 1.5 IV t ,a T A I322575 Mm. Cocaine and amphetamine regulated transcript 1.1 2.0 V t T A I390830 EST 1.0 1.6 1.1 VI b Tt ^ AA271331 EST 1.4 -1.2 1.0 W 87955 EST 1.6 -1.3 -1.1 VII B^T,AT A I552496 EST -1.4 1.3 W 64999 VI m. H esl -1.6 1.3 VIII B,T^ A I466638 EST -1.2 -1.7 -1.1 IX B,T,A^ W 3472I Mm. Vitamin D upregulated protein 1 -1.3 -1.6 -1.3 A A432818 EST -1.2 -1.6 -1.2 X B^ AA174312 EST -1.3 1.0 1.0 A I604140 EST -1.4 1.0 1.0 Genes regulated at least 1.3 times by BMP2 (300ng/ml), 1.5 times by TGF-^1 (5ng/ml) or 1.5 times by activin A (20 ng/ml) are shown, categorized according to expression profiles, based on regulation values of at least 1.2 for all growth factors. Green, up- regulated; red, down-regulated; black, not regulated; B, BMP2; T, TGF-p1; A, activin A; T, up-regulated; i , down-regulated. Each value is the average o f two reciprocal hybridizations.

- 138 - Full color figures

ACCESSION# DESCRIPTION ACCESSION# DESCRIPTION ■ AA062324 Dlhydropyrimidinase-like 3 _ AA536836 Basic helix-loop-helix domain containing, class B2 (Bhlhb2), I AA067147 Malic enzym e, supernatant " AI595201 Heparin binding epidermal growth factor-like growth factor Ï AI552130 Hairy/enhancer-of-split related 'with YRPW mot ‘ AA832598 ESTs il AA821963 Dnpep, aspartyl ami nopeptidase ‘ AA271866 Mus musculus Ca2+-independent phospholipase A2 long form AA727967 Pigment epithelium-derived factor AA7261G2 Keratin complex 1. acidic, gene 16 AI 386062 Carbonic anhydrase 3 , FUNCTION PROFILE AA738914 Gap junction membrane channel protein a.1 EXPRESSION PROFILE AA667906 Cysteine-rich protein 2 AA473123 UTP-glucose-1-phosphate uridyl yltransf era se W 81878 Fasciclin I like osteoblast specific factor 2 AA028346 Keratin complex 1, acidic, gene 1S AI019837 Laminin p3 I AA981 032 ESTs " AA684158 Keratin complex 1, acidic, gene 12 ¡I AA450795 Choi ecyst oki nin I AA958693 Transcription factor 7, T-cell specific

C lu ste r 5 a c c e s s io n # description FUNCTION PROFILE AA674392 EST s EXPRESSION PROFILE " AI587822 EST s " AA238037 Im m unity-associated protein. 38 KDa ' AA796818 Ubiquitin cross reactive protein (15 kDa) - AA466852 C ysteine rich protein 61 " AI552842 " AA200306

FUNCTION PROFILE

AA415211 Neural fold/somite kinase 2 (musk2) AA152636 Integrin

Clu ste r 3 a c c e s s io n # description FUNCTION PROFILE AW210425 Solute carrier family 20, member 1 _ _ W18822 Gadd45y, growth arrest and DNA-damage-inducible y ___ AA517353 N Fk light chain g en e enhancer in B-cells inhibitor, a ___ AA982549 Placental growth factor ~__ AA895566 Gadd45|3, growth arrest and DNA-damage-inducible p _ _ AA036380 Snail homolog. (drosophila) ___ AA733791 DdH, cell adhesion kinasel ___ AA162153 Colony stimulating factor 1 (macrophage) TF EMe| IS [ R eel EMa| CPI ]c S k | _ _ AA427234 Vascular endothelial growth factor AA879919 Tissue Factor ESTsI Signal Transduction strucuTra|an r"

FUNCTION PROFILE EXPRESSION PROFILE 1

Figure 1 (Chapter 3) . Expression profiles of the significantly (p<0.01) regulated BMP2-responsive genes. (A) Clusters of primarily up-regulated genes, (B) Clusters of primarily down-regulated genes. Genes regulated within 24 hrs after BMP2 treatment are shown, grouped according to Model-based clustering of 10log transformed micro array data (normalized on a scale to 1). Induction levels are indicated with different color intensities in red and blue, representing up-and down-regulation respectively. Protein function profile graphs show the number of genes within each cluster with a particular function. Genes with unknown function or other functions were labeled as “Other”. EST, expressed sequence tag; TF, transcription factor; Eme, extracellular messenger; IS, intracellular signaling protein; Rec, receptor; Ema, extracellular matrix protein; CPI, cytoplasmatic protein; CSk, cytoskeletal protein.

- 139- Chapter 7

ACCESSION# DESCRIPTION Cluster 9 accession# DESCRIPTION AA793036 Vascular endothelial growth factor Al 390830 ESTs Al323202 Small inducible cytokine A2 AA124623 Ecotropic viral integration site 2 AA553242 Gadd45a, growth arrest and DNA-damage-inducible o Al 552167 ESTs AA606331 L-myc, lung carcinoma myc related oncogene 1 AA619763 Cathepsin L AA437694 Myog eni c d iff erenti ati on 1 AA738820 Phosphoserine transferase AA796822 Sialyl transferase 4 Al 117310 Flavin containing monooxygenase 1 AA501052 Cardiac morphogenesis (Xin) AA444581 EST AA067625 Ph osphatidate cytidylyltransferase AA068116 Acetyl-Coenzyme A dehydrogenase, medium chain AA386892 Dystrobrevin AA606605 Nidogen 1 AA048101 Sema 4C precursor AA217009 Tumor endothelial marker 8 (anthrax toxin receptor) AA241390 Semaphorin 3E Al 607683 Semaphorin 7A AA684257 ESTs __ AA596866 ESTs AA879966 Laminin a 5 W89253 Insulin-like growth factor binding protein 5 AI325851 CD97 antigen W36474 Metallothionein 2 W53231 CD82 antigen ACCESSION#. DESCRIPTION AA105866 Glutathione S-transferase a 4 Al 552496 ESTs AI510089 EST Al 391322 Protein kinase inhibitor a AA725965 EST Al 552711 Creatine kinase, muscle AA798948 Caveolin, caveolae protein, 22 kDa AA771647 ESTs ÌAA711869 Epidermal growth factor receptor pathway substrate 8 AA815689 Troponin C, cardiac/slow skeletal W34122 Bridging integrator 1 / myc box dependent interacting protein 1 AA760135 Procollagen, type IV, a 1 AA068464 Ect2 oncogene Al 465074 Mesoderm specific transcript AI323051 Growth associated protein 43 AA822257 Myosin light chain, alkali, fast skeletal muscle AA840053 Actinin alpha 2 associated LIM protein Al 894014 Cholinergic receptor, nicotinic, y-polypeptide AA617310 Zyxin Matrix y-carboxyglutamate (gla) protein W99951 Neurofilament, light polypeptide AA117547 Cyclin D1 W13004 Prostaglandin 12 (prostacyclin) synthase AA792499 Ankyrin-like repeat protein AA023538 Semaphorin 6A W 16059 Glutathione S-transferase co 1 AA413490 Transferrin receptor AA066615 Glycine amidinotransferase AA510298 Plasminogen activator, urokinase AA067083 High mobility group protein I AA764163 Phospholipid scramblase 1 Al 326605 Cycl in-dependent kinase inhibitor 1A (p21/WAF1) Al 152562 Proliferin AA771566 Troponin C, fast skeletal AA867173 ESTs AA116263 UDP-glucose ceram ide glucosyltransferase AA222190 EIG 180 AA185112 ESTs AA516800 Death-associated kinase 2 AA288642 Cholinergic receptor, nicotinic, beta polypeptide 1 (muscle) Al 152800 ESTs AA681235 ESTs AA061902 Apolipoprotein B editing complex 2 IAA690387 Xist gene (X inactivation transcript) AA770924 Troponin T3, skeletal, fast Calsequestrin 2 Four and a half LIM domains 1 ESTs ESTs 3-Phosphoglycerate dehydrogenase ESTs Mesothelin VPS 10 domain receptor protein SORCS 2 X-linked nuclear protein ESTs

Clusterl 0 ACCESSION# DESCRIPTION AA475186 ESTs W 14224 Ndr1, N-myc downstream regulated 1 AA162217 Pbef, pre-B-cell colony-enhancing factor AA437755 ESTs AA596753 I at exin

FUNCTION PROFILE EXPRESSION PROFILE

-2.33 -1.60 -1.25 1.25 1.60 2.33 S -1 REGULATION ON MICROARRAY

Figure 1 (Chapter 3), continued.

- 140 - Full color figures

Figure 3 (Chapter 3). In vivo expression o f Hey1 and Tcf7 in the developing mouse skeleton. (A-C) Transverse sections of E14.5 mouse embryos at the level of the jaws for the control osteoblast marker Cbfa1 (A), Tcf7 (B) and Hey1 (C), (D-F) Sagital sections of E16.5 mouse embryos at the level of the nasal cavities for Cbfa1 (D), Tcf7 (E) and Hey1 (F), (G-I) Mouse proximal tail vertebrae for Cbfa1(G), Tcf7 (H) and Hey1 (I). Abbreviations used: bm, bone marrow; C1v, cartilage primordium of C1 vertebra; cl, clavicle; h, cartilage primordium of humurus; hc, hypertrophic chondroblasts; id, intervertebral disc; Jo, Jacobson’s organ; M, Meckel’s cartilage; md, mandible; mx, maxilla; nc, nasal cavity; nca, nasal capsule; P, periosteal bone; pc, proliferating chondroblasts; sc, scapula; T, trabecular bone; tg, tongue; to, tooth; vb, vertebral body.

- 141 - Chapter 7

Table 1 (Chapter 5). Expression regulation of BMP-response genes by Smads and Erk1/2 MAP-kinase Gene BMP effect caALK2 Smads U0126/Erk Smad-7 ▲ ▲ ▲ ▲ Id1 ▲ ▲ ▲ ▲

Id 2 ▲ ▲ ▲▲ ▲ Id3 ▲ ▲ ▲▲ ▲ Tieg ▲ ▲ ▲ ▲ Prrx2 ▲ ▲ ▲ - ▲ Snail ▲ ▲ - Oasis ▲ ▲▲ ▲ - Smad-6 ▲ ▲ ▲ -

Hey ▲ ▲ ▲▲ ▲ ▲ ▲ ▲ Cbfa-1 ▲ ▲▲ ▲ ▲ - ▲ ▲ Tcf7 ▲ ▲ -

Alp2 ▲ Wif-1 ▲ AA222756 ▲ ▲ ▲ ▲ ▲▲ Prdc ▲ ▲ ▲ ▲▲▲ ▲ ▲ AA794190 ▲ ▲▲▲▲ ▲ ▲ ▲▲ AA709999 ▲ ▲▲▲▲ ▲▲▲▲▲▲ ▲ ▲ Col6a2 ▲ ▲ ▲ - ▲ Ocn ▲ ▲▲ ▲ ▲ - ▼ ▼▼ Summary of Figures 1 and 2. ▲ = upregulation, ▼ = down-regulation, the number of triangles indicate relative levels of expression.

Figure 4 (C hapter 5). In vivo expression of Hey1 and Lfng in the developing mouse skeleton. (A-C) Sagital sections of E14.5 mouse embryos at the level of the ventricles for the control osteoblast marker Cbfa1 (A), Hey1 (B) and Lfng (C), (D-F) Transverse sections of E14.5 mouse embryos at the level of the jaws for Cbfa1 (D), Hey1 (E) and Lfng (F), (G-I) Mouse proximal tail vertebrae for Cbfa1 (G), Hey1 (H) and Lfng (I). Abbreviations used: bm, bone marrow; D, Diencephalon; hc, hypertrophic chondroblasts; id, intervertebral disc; Jo, Jacobson’s organ; md, mandible; Mo, Medulla oblongata; mx, maxilla; nc, nasal cavity; nca, nasal capsule; P, periosteal bone; pc, proliferating chondroblasts; T, trabecular bone; tg, tongue; Sb, Skull bones, V, Ventricular zone; vb, vertebral body.

- 142 - Full color figures

A Signal transduction pathways

Figure 5 (C hapter 5). Model of multi-layered BMP induced osteoblast differentiation For details see text, CHX = cyclohexamide

- 143 - - 144 - 8

Summary

Samenvatting Chapter 8

Summary

Bone development can occur by two mechanisms of ossification: intramembranous ossification and endochondral ossification. In the process of intramembranous ossification, bone growth begins with the condensation of mesenchymal cells that expand and form a membranous structure. Subsequently, these mesenchymal cells differentiate directly into osteoblasts in the primary centers of ossification, and calcification by mature osteoblasts occurs in irregularly distributed patches. During endochondral ossification, the condensed embryonic mesenchymal cells first differentiate into chondroblasts, which secrete the cartilaginous matrix. Subsequently, the chondroblasts proliferate, differentiate into chondrocytes and eventually become hypertrophic and undergo programmed cell death. Osteoclasts then resorb the cartilage matrix and adjacent mesenchymal stem cells differentiate to osteoblasts, which form new bone in the primary center of ossification. Subsequent ossification occurs in a regulated fashion in secondary centers of ossification. Osteoblast differentiation can be regulated by several hormonal and polypeptide growth factors, such as PTH/PTHrP (parathyroid hormone/parathyroid hormone-related protein), Indian hedgehog (IHH), Vitamin D, insulin-like growth factors (IGFs), fibroblast growth factors (FGFs), platelet-derived growth factor (PDGF), transforming growth factor P (TGF-P) and the bone morphogenetic proteins (BMPs). Of these factors, the BMPs are the most potent in inducing osteoblast differentiation. To generate biological responses, BMP family members use heteromeric complexes of so- called type I and type II serine/threonine kinase receptors. Upon ligand binding, a complex is formed between two type I and two type II receptors, upon which the constitutively active kinase of the type II receptor phosphorylates and activates the type I receptor, which subsequently activates specific receptor-regulated Smads (R-Smads). Smad-1, -5 and -8 have been shown to be BMP responsive R- Smads, while Smad-2 and -3 are responsive to TGF-P and activins, both proteins that are closely related to BMPs. In combination with the common mediator Smad-4, the respective R-Smads are assembled into heteromeric complexes, which subsequently translocate to the nucleus, where they act in concert with other proteins as transcription factor complexes. In addition to the BMP-activated Smad pathway, several other major signaling pathways are known to play a role during osteoblast differentiation. These include the different MAP-kinase pathways and the Notch pathway. However, the interplay between the various pathways in osteoblast differentiation is still poorly understood. A detailed characterization of BMP-induced regulation of the differentiation of MSCs to osteoblasts is therefore very important for understanding the bone forming part of bone homeostasis. Such knowledge might facilitate the identification of new putative drug targets that promote bone formation by enhancing osteoblast differentiation. Ultimately, these new drug targets might enable the development of novel therapies to treat serious clinical disorders such as osteoporosis in which bone homeostasis is disturbed. This thesis addressed the identification of (novel) genes that act as key- or co-regulators of BMP-induced osteoblast differentiation and that may function as focal points for the interaction of the BMP signal transduction pathway with other pathways relevant for osteoblast differentiation. The “gene expression microarray” (GEM) technology allows the simultaneous monitoring of expression of several thousands of genes under different experimental conditions. Therefore, we used this approach to obtain a detailed characterization of gene expression regulation during the early phases of osteoblast differentiation (chapter 3). A large set of BMP2-regulated genes (184 genes and ESTs) was identified that could be subdivided into several clusters according to their time-dependent expression patterns. These clusters contain functionally related genes, representing the various phases of osteoblast differentiation. Signal transduction regulatory factors are mainly represented in the

- 146 - Summary

clustered subset of immediate early genes, whereas the intermediate early and late early clusters consisted primarily of genes related to processes that modulate morphology, basement membrane formation and synthesis of extracellular calcified matrix. Interestingly, two genes were identified that exhibited very similar expression characteristics as the osteoblast-specific type II isoform of C bfal: one belonging to the Wnt signal transduction pathway (Tcf7) and one belonging to the Notch signal transduction pathway (Heyl). Moreover, ectopic expression of Tcf-7 was shown to regulate the expression of osteoblast differentiation marker genes which suggests that Tcf-7 enhances BMP-2 induced osteoblast differentiation. Earlier studies revealed that members of the TGF-P family of growth factors exhibit different capacities in regulating osteoblast differentiation. We have subsequently exploited these differences by comparing gene expression profiles of cells stimulated by either BMP2, TGF-P or a combination of these growth factors (chapters 2 and 4). Genes that are highly induced by BMP-2, but not by TGF- P or are repressed by TGF-P, and those of which the BMP-induced expression levels are inhibited by TGF-P might be important regulators of BMP-induced osteoblast differentiation. A small set of genes exhibiting these expression profiles was identified, including several genes involved in the Notch signal transduction pathway such as Heyl, Hes and Lfng. This result suggests that the Notch signal transduction pathway is a major regulator of early phase osteoblast differentiation. In addition, potential focal points of interaction of the BMP and Notch signal transduction pathways were identified. As mentioned above, osteoblast differentiation is not only regulated by the BMP/Smad signal transduction pathway, but also by other signal transduction pathways such as Erk1/2 MAP- kinase and Notch pathways. To investigate the influence of these signal transduction pathways on osteoblast differentiation, we studied the role of Smads, Erk1/2 MAP-kinase and Notch on gene expression regulation during various stages of osteoblast differentiation (chapter 5). Our results showed that during BMP-induced osteoblast differentiation a multi-layered cascade of expression regulation is employed in which the initial phase of differentiation is based on general BMP-response genes. During the progression of osteoblast differentiation, osteoblast commitment genes and osteoblast specific genes are regulated in concert with the general BMP-response genes. In addition, Erk1/2 MAP-kinase and Notch signaling become more and more effective in modulating expression profiles, which ultimately leads to a functional osteoblast characterized by the deposition of a calcified extracellular matrix. Moreover, several genes were identified as potential focal points for the integration of the Smad and Erk1/2 signaling pathways. In short, this thesis provides an overview of the genes regulated during early phase BMP- induced osteoblast differentiation, and provides insights in how various signal transduction pathways interact to regulate osteoblast differentiation.

- 147 - Chapter 8

Samenvatting

Gewervelde dieren hebben als kenmerk dat zij een inwendig skelet hebben. Dit skelet is bij zoogdieren opgebouwd uit bot en kraakbeen. In volgroeide zoogdieren is er normaal gesproken een evenwicht tussen botaanmaak en botafbraak. Bij een verstoring van dit evenwicht kunnen echter ernstige medische klachten optreden, zoals bij osteoporose, waarbij er meer afbraak dan aanmaak van bot plaatsvindt. De huidige medicijnen voor deze ziekte zijn zo ontwikkeld dat zij de botafbraak remmen, maar aanmaak van nieuw bot wordt niet of nauwelijks gestimuleerd. Medicijnen die botvorming stimuleren kunnen echter een belangrijke aanwinst zijn in de behandeling van osteoporose. Voor de ontwikkeling van dit soort medicijnen is een gedetailleerde kennis over de regulatie van botvorming belangrijk. Botvorming kan via twee mechanismen optreden, namelijk via intramembrane ossificatie en via endochondrale ossificatie. Bij het eerste mechanisme begint botvorming met de groei van mesenchymale stamcellen tot een membraanachtige structuur. Vervolgens differentiëren de mesenchymale cellen direct uit tot osteoblasten, de botvormende cellen. Bij endochondrale ossificatie differentiëren de mesenchymale cellen eerst naar kraakbeen vormende cellen, de chondroblasten. Vervolgens wordt het kraakbeen afgebroken en differentiëren andere mesenchymale cellen zich tot osteoblasten die in de vrijgekomen ruimte botweefsel aanmaken. De differentiatie van mesenchymale stamcellen tot osteoblasten kan gereguleerd worden door verschillende hormonen en groeifactoren zoals parathyroid hormoon, vitamine D, “transforming growth factor P” (TGF-P) en de “bone morphogenetic proteins” (BMPs). Van al deze factoren induceren de BMPs de differentiatie naar osteoblasten het beste. Mesenchymale cellen worden door middel van specifieke BMP receptoren op het oppervlak van de cel en door een cascade aan signaaltransductie-eiwitten gestimuleerd om te differentiëren naar osteoblasten. Deze specifiek door BMP gereguleerde cascade van signalen wordt ook wel de BMP/Smad signaaltransductie-route genoemd. Naast deze specifieke door BMPs gereguleerde route is ook een aantal andere signaaltransductie-routes betrokken bij de differentiatie van mesenchymale cellen naar osteoblasten, die niet alleen door BMPs geactiveerd kunnen worden, maar ook door andere groeifactoren en hormonen. Deze omvatten onder andere de MAP-kinase en de Notch signaaltransductie-routes. De samenwerking van deze verschillende routes binnen het proces van osteoblast differentiatie is nog erg onduidelijk. Kennis van de precieze eiwitten die tot expressie komen in de cel en hun interacties met elkaar tijdens het differentiatie proces is erg belangrijk, aangezien dit kan bijdragen aan de ontdekking en ontwikkeling van nieuwe medicijnen die botvorming stimuleren. Deze nieuwe medicijnen zouden dan, naast de al bestaande medicijnen die botafbraak remmen, voor de behandeling van osteoporose gebruikt kunnen worden. Dit proefschrift bespreekt de identificatie van genen, de blauwdrukken van eiwitten, die een sleutelrol spelen in de regulatie van BMP-geïnduceerde osteoblast differentiatie, en die mogelijk functioneren als brandpunt voor de interactie van de verschillende signaaltransductie-routes die werkzaam zijn tijdens osteoblast differentiatie. Om een gedetailleerd overzicht te verkrijgen welke genen gereguleerd worden tijdens osteoblast differentiatie kan de “gene expression microarray” (GEM) techniek gebruikt worden. Deze techniek maakt het mogelijk om in een keer van duizenden genen de genexpressie te meten onder verschillende experimentele omstandigheden of in een tijdsverloop. Aangezien direct na stimulatie van cellen met BMP de meeste regulatie optreedt voor het aansturen van het differentiatieproces, hebben we de GEM techniek gebruikt om een gedetailleerde beschrijving te verkrijgen van genexpressie-regulatie tijdens de vroege fase van osteoblast differentiatie (hoofdstuk 3). Gebleken is dat een groot aantal genen (184) gereguleerd werd, die in verschillende groepen konden worden ingedeeld op basis van hun expressie in de loop

- 148 - Samenvatting

van de tijd. Hierbij gold dat de genen die het vroegst tot expressie kwamen, voornamelijk coderen voor signaaltransductie-eiwitten en dat genen die later in het proces tot expressie kwamen, voornamelijk coderen voor eiwitten die de vorm van cellen veranderen en die gebruikt worden om het daadwerkelijke bot te vormen. Opmerkelijk genoeg werden er twee genen geïdentificeerd, Tcf7 en H eyl, met een expressiepatroon dat erg leek op het expressie patroon van het Cbfal gen. Het eiwit waar Cbfal voor codeert, is een erg belangrijke regulator van osteoblast differentiatie. Bovendien kon worden aangetoond dat Tcf7 de differentiatie van osteoblasten door BMP-2 kan versnellen. Eerdere studies hebben uitgewezen dat de verwante groeifactoren BMP-2 en TGF-P in verschillende mate in staat zijn osteoblast differentiatie te reguleren. Gebruikmakend van deze verschillen hebben wij een vergelijking gemaakt tussen de genexpressie-profielen van cellen die ofwel met BMP-2, ofwel met TGF-P, ofwel met een combinatie van BMP-2 en TGF-P behandeld werden (hoofdstukken 2 en 4). Theoretisch gezien zouden genen die in hun expressie sterk gereguleerd worden door BMP-2, maar niet of nauwelijks door TGF-P, en waarvan de BMP- geïnduceerde expressie geremd kan worden door TGF-P, zeer belangrijke regulatoren kunnen zijn van osteoblast differentiatie. Gebleken is dat een klein aantal genen dergelijke expressieprofielen bezitten. Van deze kleine groep genen behoorden er drie (Heyl, Hes en Lfng) tot de Notch signaaltransductie- route, hetgeen een indicatie is dat deze route een belangrijke rol speelt in de vroege regulatie van osteoblast differentiatie. Verder zouden deze genen ook als brandpunt kunnen dienen voor de interactie van de Notch en BMP signaaltransductie-routes. Osteoblast differentiatie wordt niet alleen gereguleerd door de BMP/Smad signaaltransductie-route, maar ook door andere signaaltransductie-routes zoals de Erk1/2 MAP-kinase en de Notch routes. Om te bepalen in welke mate deze routes bijdragen aan de regulatie van osteoblast differentiatie, hebben we onderzoek gedaan naar de rol van verschillende Smad eiwitten alsmede van het Erk1/2 eiwit en het Notch eiwit op genexpressie regulatie tijdens verschillende stadia van osteoblast differentiatie (hoofdstuk 5). Gebleken is dat tijdens BMP-geïnduceerde osteoblast differentiatie gelaagde expressieregulatie plaatsvindt. In de eerste laag wordt, direct na stimulatie, de genexpressie gereguleerd van genen die een algemene BMP respons weergeven. Vervolgens wordt de expressie gereguleerd van genen die de cellen “vastleggen” in het differentiatieproces, en uiteindelijk wordt de genexpressie gereguleerd van genen die specifiek zijn voor het osteoblast celtype. Ook geldt dat, naarmate het differentiatieproces vordert, de invloed van de Erk1/2 MAP-kinase en Notch signaaltransductie-routes op de regulatie van genexpressie steeds meer toeneemt. Dit geheel aan modulaties op het differentiatieproces leidt uiteindelijk tot een celtype dat botweefsel kan aanmaken, te weten een functionele osteoblast. Bovendien werden er enige genen geïdentificeerd die mogelijk brandpunten vormen van de interactie van de Erk1/2 en BMP/Smad signaaltransductie-routes. Samenvattend geeft dit proefschrift een gedetailleerd overzicht van de genen waarvan de expressie gereguleerd wordt in de vroege stadia van het proces van osteoblast differentiatie. Daarmee draagt het bij aan het begrip van hoe verschillende signaaltransductie-routes met elkaar communiceren om de differentiatie van mesenchymale cellen zo te geleiden dat er functionele osteoblasten ontstaan.

- 149 - - 150 - Dankwoord

Afscheid van de wetenschappelijke wereld. Na een periode van vier en een half jaar in Nijmegen en nog eens 9 maanden in Alkmaar is het nu tijd geworden om een periode af te sluiten. Vol goede moed was ik begonnen aan het AIO-schap en na een flinke dip heb ik alles toch nog met voldoening kunnen afronden. Een groot aantal mensen heeft mij hierbij geholpen. Ten eerste wil ik graag Joop van Zoelen en Wiebe Olijve bedanken voor de mogelijkheid die jullie mij gaven om het onderzoek beschreven in dit proefschrift te verrichten. Uiteindelijk heeft dit onderzoek, na veel volharding, een aantal veelbelovende resultaten opgeleverd en ik hoop dan ook dat het een basis zal vormen voor succesvol vervolgonderzoek. Graag wil ik ook Wilma Steegenga bedanken voor haar inzet 4 jaar lang om ons, haar AIO’s, te begeleiden in het onderzoek en in het gareel te houden. Het zweepje was af en toe best nodig... Koen Dechering wil ik graag bedanken voor zijn enthousiasme voor het onderzoek en zijn vaak zeer nuttige inbreng in het onderzoek. Ron Wehrens wil ik met name bedanken voor zijn enthousiasme om te goochelen met grote datasets en hieruit ook nog eens een biologische logica uit te peuteren! Zonder jou waren de clusteringen een crime geweest. Peter ten Dijke wil ik graag bedanken voor de gesprekjes over TGF-P signalering en de samenwerking tussen zijn lab en Toegepaste Biologie. Ook wil ik graag Christine Mummery, Alie Feijen en Marcel Karperien bedanken voor de samenwerking. Zonder jullie had ik al die mooie in situ’s niet gehad. Verder wil ik graag mijn mede-AIO van de afdeling Toegepaste Biologie Bart Vaes bedanken, niet alleen voor onze discussies over botbiologie en voor je practische hulp bij het onderzoek, maar ook simpelweg voor je aanwezigheid. Het had beslist een stuk saaier geweest zonder jou. Ook wil ik graag Jose Hendriks, Leonie Heuver, Ester Piek en ‘mijn’ studente Maaike Kuijer bedanken voor hun bijdrage aan het onderzoek en de gezelligheid op het lab. Maaike, ik ben ontzettend blij dat wij uit een lastige periode een goede vriendschap hebben kunnen opbouwen. Mijn mede treinreizigers van de eerste jaren, Saskia Oomen, Debbie Ouwens en Karen Pasman, wil ik graag bedanken voor de gezelligheid in de trein, al was het soms maar een klein stukje dat we samen reisden. Iedereen van de afdelingen Celbiologie, Toegepaste Biologie en Moleculaire biologie wil ik graag bedanken voor het bijdragen aan de prettige werksfeer en de gezellige borrels, nachtelijke speurtochten en labuitjes. Buiten het lab gaat mijn speciale dank uit naar mijn dansmaatje Oleg van Schaik. Ik heb ontzettend genoten van onze danslessen bij Elvira en ik ben blij dat jij voor mij zo’n vrolijke uitlaatklep wou zijn. Mijn ouders wil ik graag bedanken voor het simpele feit dat ze in mij geloven en mij door dik en dun steunen. Ook de rest van mijn familie en mijn vrienden wil ik bedanken voor de interesse die jullie getoond hebben in mijn onderzoek. Als laatste gaat mijn speciale dank uit naar Ard Groot. Je kwam precies op tijd mijn leven binnen......

- 151 - - 152 - List of Publications

de Jong DS, Olijve W, van Zoelen EJJ. Osteoblast Differentiation: BMP Signaling in Interaction with MAP-kinase, Wnt and Notch Pathways (review). Int JDevelBiol; submitted. de Jong DS, Steegenga WT, Hendriks JM, van Zoelen EJ, Olijve W, Dechering KJ. Regulation of Notch signaling genes during BMP2-induced differentiation of osteoblast precursor cells. Biochem Biophys Res Commun. 2004 Jul 16;320(1):100-7. de Jong DS, Vaes BL, Dechering KJ, Feijen A, Hendriks JM, Wehrens R, Mummery CL, van Zoelen EJ, Olijve W, Steegenga WT. Identification of novel regulators associated with early-phase osteoblast differentiation. J Bone Miner Res. 2004 Jun;19(6):947-58. de Jong DS, van Zoelen EJ, Bauerschmidt S, Olijve W, Steegenga WT. Microarray analysis of bone morphogenetic protein, transforming growth factor beta, and activin early response genes during osteoblastic cell differentiation. J Bone Miner Res. 2002 Dec;17(12):2119-29. van de Zande L, van Apeldoorn RC, Blijdenstein AF, de Jong DS, van Delden W, Bijlsma R. Micro satellite analysis of population structure and genetic differentiation within and between populations of the root vole, Microtus oeconomus in the Netherlands. Mol Ecol. 2000 Oct;9(10):1651-6.

- 153 - - 154 - Curriculum Vitae

Diana de Jong werd geboren op 27 januari 1976 te Sneek. Na het afronden van het Gymnasium aan de Rijks Scholen Gemeenschap “Magister Alvinus” te Sneek werd in 1994 begonnen met de studie Biologie aan de Rijks Universiteit van Groningen. Gedurende haar opleiding was zij werkzaam als stagiaire bij de vakgroep Populatiegenetica van de afdeling Genetica (dr. L. van der Zande, Rijks Universiteit Groningen) en bij de afdeling Moleculaire Biologie van het Nederlands Kanker Instituut (dr. Denise Barlow, NKI Amsterdam). Na de afronding van haar doctoraal examen begon zij in 1999 als assistent in opleiding (AIO) bij de afdeling Toegepaste Biologie van de Katholieke Universiteit Nijmegen (thans Radboud Universiteit Nijmegen) onder leiding van prof. dr. W. Olijve, prof. dr. E. J. J. van Zoelen en dr. W. T. Steegenga. Met ingang van 23 mei 2005 heeft zij de wetenschap vaarwel gezegd en is zij werkzaam als treindienstleider bij ProRail, afdeling Noord-Holland.

- 155 - - 156 -