Развивающиеся и Созревающие Остеобласты

Runx2: A Master Organizer of Gene Transcription in Developing and Maturing Osteoblasts Tania M. Schroeder, Eric D. Jensen, and Jennifer J. Westendorf Birth Defects Research (Part C) 75:213-225, 2005. Wiley-Liss, Inc.
Runx2 важен для развития остеобластов и собственно формирования кости. Член семейства Runt доменовых транскрипционных факторов, Runx2 соединяется со специфическими последовательностями ДНК, чтобы регулировать транскрипцию многочисленных генов и тем самым контролировать разви тие остеобластов из мезенхимных стволовых клеток и созревание в остеоциты. Хотя и необходимый для транскрипции генов и развития остеобластов, Runx2 не достаточен для оптимальной экспрессии генов или формирования кости. Runx2 кооперируется с многочисленными белками, включая транскрипционные факторы и кофакторы, тем самым модифицируется пост-трансляционно и ассоциирует с ядерным матриксом, чтобы интегрировать разнообразные сигналы и обеспечивать критические события во время развития и созревания остеобластов. В соответствии с этой ролью мастера организатора, альтерации уровней экспрессии Runx2 связаны со скелетными болезнями. Случаи гапло-недостаточности Runx2 характеризуются cleidocranial dysplasia, тогда как избыточная экспрессия Runx2 является общим признаком для большинства метастазирующих в кости раковых опухолей. В обзоре рассматриваются молекулярные механизмы, с помощью которых Runx2 интегрирует сигналы посредством ко-регуляторных взаимодействий и обсуждается его роль как мастера организатора, которая может меняться в зависимости от структуры промотора, онтогенетических сигналов и клеточного содержимого.	RUNT DOMAIN PROTEINS: DEVELOPMENTAL ROLES Runx2 is one of three mammalian genes that encode proteins homologous to Drosophila Runt, and that are crucial for proper embryonic development. Runx2 (Cbfal) is required for mesenchymal condensation, osteoblast differentiation from mesenchymal stem cells, chondrocyte hypertrophy, and vascular invasion of developing skeletons (Komori et al., 1997; Otto et al., 1997). Introduction of Runx2 into mesenchymal stem cells stimulates differentiation along the osteoblast lineage (Byers and Garcia, 2004). In contrast, overexpression of Runx2 in type I collagen-positive and lineage-restricted osteoblasts blocks terminal differentiation and increases bone resorption (Geoffroy et al., 2002; Liu et al., 2001). These data indicate that Runx2 Is necessary for osteoblast lineage specification from mesenchymal precursors, but Runx2 levels must be tightly regulated in lineage-determined osseous cells. Factors influencing Runx2 expnression and function are discussed in this review. The other Runx factors also have roles in skeletal development. Runxl and Runx3 contribute to endochondral bone formation (Brenner et -al., 2004; Uan et al., 2003; Yoshida et al., 2004); however, their essential functions during em-bryogenesis are manifested in other tissues. Runxl is essential for definitive hematopoiesis, and is frequently mutated in cancer stem cells that give rise to acute leuke-mias (Miyamoto et al., 2000; Miyoshi et al., 1991; Okada et al., 1998). Runx3 is needed for the development of dorsal root pmg'ia proprioceptive neurons and proper control of gastric epithelium growth (Levanon et al., 2002; Li et al., 2002). The important roles of Runx2 in human development are evident in CCD, a rare autosomal dominant disease. Heterozygous loss of function of Runx2 through a variety of mutations (e.g., deletions, point mutations, insertions, and mis-sense mutations) segregate with abnormal skeletal phenotypes in CCD families (Otto etal., 2002; Yoshida et aL, 2002b). These alterations affect just one allele of Runx2 on chromosome 6q21 and Figure 1. Runx2 contains multiple functional domains. Functional activities of Runx2 domains are shown at the top of the figure. Alternative amino-termini (MASNS and MRIPV) are generated from two promoters. Shapes below the Runx2 diagram summarize the various types of mutations found in the CCD families in these regions. Diamond, missense mutation; triangle pointing up, insertion/ frameshift; triangle pointing down, deletion; circle, splice; square, nonsense mutation. The bottom part of the figure shows the relative positions of protein/cofactor binding to Runx2. Interaction domains for proteins listed below the dotted line have not bean defined. would be predicted to decrease wild-type Runx2 levels and activity by one half. Runx2 haploinsuffi-ciency in mice causes similar phe-notypes as observed in CCD families, with a primary defect in intramembranous bone formation (Komori et al., 1997; Otto et al., 1997). As one may suspect given the importance of DNA binding to a transcription factor, many mutations in CCD individuals are found in the DN A binding domain of Runx2 (Fig. 1). However, other mutations are present in regions proximal and distal to the DNA binding domain. These mutations suggest important roles for other parts of Runx2 in skeletal development. Indeed, CCD mutations in the Runx2 carboxy-tenmi-nus impair Smad interaction and transcriptional activation (Zhang et al., 2000). Moreover, transgenic mice lacking the carboxy-terminus of Runx2 develop a CCD-like pheno-type (Choi et al., 2001). Runx2 was originally cloned from calf thymus nuclei as core binding factor alpha (Cbfal) and later in osteoblasts as nuclear matrix protein 2 (NMP2), osteoblast-specific complex (OBSC), and osteoblast-specific factor 2 (OSF2) (Banerjee et al., 1996; Geoffroy et al., 1995; Merriman et al., 1995; Wang and Speck, 1992). It is also commonly referred to as polyoma enhancer binding protein 2aA (PEBP2aA) and acute myeloid leukemia protein 3 (AML3) (Kamachi et al., 1990; Levanon et al., 1994). The Runx nomenclature was established to unify-these and other names given by various labs around the world to the same developmentally essential proteins. The Runx terminology reflects the homology of mammalian proteins to the founding member of the family, Runt, and the crucial roles of these proteins in developmental processes (van Wi-jnen et al., 2004). While knockout experiments and temporal and spatial expression patterns demonstrate nonredundant functions of the Runx genes in the skeleton and' other tissues during embryogenesis and postnatal development (Levanon et al., 2001; Lian et al., 2003), numerous cell types simultaneously express multiple Runx proteins. An important unresolved issue is whether Runx proteins have functionally identical roles#when they are coexpressed. In vitro molecular and biochemical experiments indicate that all Runx proteins bind the same DNA sequence and have similar overall structure and functional-domains. However, attempts to rescue Runxl-deficient embryos with Runxl-Runx2 or Runxl-Runx3 chimeras revealed that the carboxy-terminal regions of Runx factors, especially Runx2, convey distinct roles during the development of some tissues (Fukushima-Nakase et al., 2005). Numerous proteins associate with Runx2, and a subset interacts with the carboxy-terminus. In this review, we summarize the biochemical interactions that affect Runx2 transcriptional activity in osteoblasts. STRUCTURAL AND FUNCTIONAL DOMAINS OF RUNX2 A highly conserved DNA binding and protein-protein interaction motif, termed the Runt domain, is a characteristic and essential feature of Runx proteins (Fig. 1). Structural studies of the Runxl Runt domain show that it forms an s-type immunoglobulin fold (Backstrom et al., 2002; Warren et al., 2000). This 128-amino acid domain mediates binding of all Runx proteins to the same DNA sequence, 5'-PuAC-CPuCA-3', or its complement, 5'-TGPyGGTPy-3' (Kamachi et al., 1990; Meyers et al., 1993; Wang et al., 1993). The Runt domain also binds other proteins. CBF/3 was the first Runt domain-interacting protein identified (Kamachi et al., 1990; Wang and Speck, 1992). CBF/3 does not contact DNA itself, but enhances the ability of Runt domain proteins to bind DNA by inducing structural changes that unmask their DNA interaction surface (Tahirov et al., 2001). CBF/3 also stabilizes Runt domain proteins by protecting them from ubiquitin-me-diated degradation (Huang et al., 2001). Point mutations that map to Runt domain residues along the Runx2-CBFЈ dimer interface are found in individuals with CCD (Otto et al., 2002; Warren et al., 2000). These data hinted that CBF0 might be essential for osteogenesis. CBF/3-deficient mice died at ED 12 with similar phenotypes as Runxl -deficient animals (Niki et al„ 1997; Wang et al., 1996). Because skeletal development begins after this time, tissue-specific depletion of CBFfi was needed to demonstrate that it is es* sential for proper skeletal development (Kundu et al., 2002; Miller et al., 2002; Yoshida et al., 2002a). NUCLEAR AND SUBNUCLEAR LOCALIZATION OF RUNX2 Runx2 transcriptional activity is dependent on Its localization to the nucleus and to discrete subnuclear foci. Two distinct domains direct Runx proteins to the nucleus and nuclear matrix (Fig. l).The nuclear localization signal (NLS) is a short basic stretch of nine amino acids that are immediately carboxy-ter-mina) to the Runt domain (Kanno et al., 1998; Thirunavukkarasu et al., 1998). The nuclear matrix targeting signal (NMTS) is 31 amino acids in length, located near the carboxy-terminus of the protein, and is the only region other than the Runt domain whose structure has been solved. The NMTS forms a helix-loop-helix structure and directs Runx factors to subnuclear foci (Kanno et al., 1998; Tang et al., 1999; Zeng et al., 1997, 1998). This localization is necessary for Runx2-dependent transactivation (Zaidi et al., 2001). Deletion of the NMTS prevents Runx2'localization to nuclear foci and increases the mobility of Runx proteins (Harrington et al., 2002). Moreover, deletion of or mutations in the NMTS affect the size, packing, and spatial; randomness of Runx2 subnuclear. foci (Young et al., 2004). In vivo, mice homozygous for Runx2 lacking the NMTS and residues car-boxy-terminal to it failed to develop mineralized tissue due to a matura-tional arrest of osteoblasts (Choi et al., 2001). Specific point mutations within the NMTS also convey a dominant-negative function in Runx2-positive breast cancer cells (Javed et al., 2005). The mechanisms by which the NMTS mediates punctate distribution and functional effects are not known, but likely involve interactions with other nuclear proteins. As will be discussed below, numerous proteins including histone deacetylase (HDAC) 6, Smads, and Yes-associated protein (YAP) associate with residues in or near the NMTS. ACTIVATION DOMAINS OF RUNX2 Runx2 contains multiple activation and repression domains that contribute to Its transcriptional activity. The primary activation domain of Runx proteins is a proline-serine-threonine (PST)-rich region that is distal to the NLS and immediately proximal to the NMTS (Kanno et al., 1998; Thirunavukkarasu et a I., 1998). This domain induces the transcriptional activation of a heterologous promoter when fused to an appropriate DNA binding domain. It contributes to interactions of Runx2 with coactivator proteins, notably p300, MOZ, and MORF (Pel-letier et al., 2002; Sierra et al., 2003). The amino-termini of Runx proteins may also be necessary for maximal transcriptional activation, but they are not sufficient to activate a heterologous promoter (Kanno et al., 1998; Thirunavukkarasu et al., 1998). This region of Runx2 contains a polyglutamine and polyalanine (QA) domain that is not present in Runxl or Runx3. The QA region is frequently altered by genetic mutations in CCD families (Otto et al., 2002). Polymorphic expansions or contractions of the QA domain may also influence skeletal strength and morphology in humans and other species (Fondon and Garner, 2004; Vaughan et al., 2002). In addition to its activation potential, the amino-terminus of Runx2 has repressive activity and can inhibit DNA binding (Inman et al., 2005; Schroeder et al., 2004), suggesting it may be a context-dependent regulator of Runx2 activity. REPRESSION DOMAINS OF RUNX2 Relative to their functions as transcriptional activators, the role of mammalian Runx proteins as transcriptional repressors has been underappreciated. The repressive function of mammalian Runx proteins should not be surprising because the Drosophila protein, Runt, was originally described as a transcriptional repressor and is now recognized as a context-dependent activator or repressor (Manoukian and Krause, 1993; Swantek and Gergen, 2004; Tsai and Gergen, 1994). Numerous regions of Runx2 contribute to transcriptional repression. The last five amino acids of Runx proteins, VWRPY, are perfectly conserved and mediate interactions with several members of the transducin-like enhancer of split (TLE)/Groucho (Grg) family of proteins (Gasperowicz and Otto, 2005). This region associates with TLE2, but it is dispensable for interactions with TLE1 and TLE3 (Javed et al., 2000; McLarren et al., 2000; Thirunavukkarasu et al., 1998). A second repression domain in Runx2 that may contribute to some TLE interactions is found in and around the NMTS (Kanno et al., 1998; Thirunavukkarasu et al., 1998). This region also associates with HDAC6 (Westendorf et al., 2002). Like their amino-termini, the carboxy-termini of Runx proteins inhibit the interaction between the Runt domain and DNA or other proteins (Inman et al., 2005; Kanno et al., 1998): A third repression domain is contained within the first 94 amino acids of Runx2 and mediates binding to HDAC3 (Schroeder et al., 2004). Finally, the Runt domain of Runx2 contributes to interactions with compressors and transcription factors, including HDAC4 and Lefl (Kahler and Westendorf, 2003; Vega et al., 2004), which may prevent DNA binding. Thus, multiple regions of Runx2 are involved in the recruitment of cofactors that repress gene transcription. RUNX2 ISOFORMS The single Runx2 gene in mammalian genomes encodes multiple transcripts, which are derived from two promoters and alternative splicing. Together, differential promoter usage and alternative splicing creates at least 12 distinct iso-forms (Levanon and Groner, 2004; Stock and Otto, 2005; Terry et al., 2004). Therefore, cells may express a complex array of Runx2 proteins that lack certain domains described above and exhibit specialized functions during osteoblast differentiation and maturation. Two promoters, PI (distal) and P2 (proximal), drive expression of the major Runx2 isoforms, type II and type I, respectively. This double promoter structure is conserved in human and murine Runx2 genes, as well as the other mammalian Runx genes, Runxl and Runx3 (Levanon and Groner, 2004). Runx2-I, containing 513 amino acids and beginning with MRIPV, is constitutively expressed in non-osseous mesenchymal cells, os-teoprogenitors, chondrocytes, and thymocytes (Banerjee et al., 2001). Osteoblasts express Runx2-I at consistent levels throughout osteoblast .differentiation (Banerjee et al., 2001). Runx2-H contains 528 amino adds, beginning with MASNS. Runx2-II expression is restricted to osseous cells, induced by bone mor-phogenetic protein (BMP) 2 in osteo-progenitors, and is increased during osteoblast differentiation (Banerjee etal., 2001; Ducy etal., 1997). During development, Runx2-II is expressed in mesenchymal condensations and mature chondrocytes of the developing axial skeleton (Lengner et al., 2002). Kunx2-H-deficient mice have a milder skeletal pheno-type than complete Runx2-null mice, exhibiting dwarfism due to defects in endochondral bone formation and osteopenia caused by reduced osteoblast maturation (Xiao et al., 2004, 2005). These results suggest that Runx2-I is sufficient for early osteoblast differentiation and in-tramembranous bone formation, and Runx2-II is necessary for complete osteoblast maturation and endochondral bone formation. The mechanism(s) for the isoform-spe-cific developmental roles are not known, but may include preferential promoter induction and/or binding of other proteins to the alternative ami-no-termini of the Runx2 proteins. A third N-terminal isoform, OSF2, encodes a protein in mice that is 68 residues longer than Runx2-II and begins with amino acids MLHSPH (Ducy et al., 1997); however, this sequence is not conserved in humans. Alternative splicing of Runx2 ex-ons also generates multiple protein products (Stewart et al., 1997; Geoffroy et al., 1998; Terry et al., 2004). Some splice products antagonize full-length Runx2 proteins by competing for DNA binding sites. It is likely that other Runx2 isoforms will be capable of interacting with unique sets of cofactors that will influence their function. The tissue-specific expression patterns of all 12 Runx2 isoforms are not yet clear, but given the conservation of the Runx2 gene structure, it seems likely that at least some of these isoforms will have distinct developmental roles. RUNX2 PROTEIN PARTNERS As we have discussed, Runt domain proteins either activate or repress transcription of genes by binding the DNA sequence, TGPyGGPy. The Afunctional role of Runt domain proteins is conserved throughout evolution. It is best exemplified by the activation or repression of a Drosophila gene, slpl, by Runt depending on its combinatorial inter- action with one of two differentially expressed transcription factors, Opa or Ftz (Swantek and Gergen, 2004). Interactions with other proteins are equally as important for Runx2-dependent regulation of osteoblast development and differentiation. Although the biphasic temporal regulation of a single osteoblast , gene by Runx2 during development • has not yet been molecularly defined 1 as it has in Drosophila embryos, there are numerous reports describing agonistic or antagonistic effects of other proteins on Runx2-depen-dent activation of specific genes. Many Runx2 target genes are absent in cells that express Runx2, suggesting that Runx2 transcriptional activity is under complex and tight regulation in osseous cells, and is influenced by the temporal and spatial expression of other factors. TRANSCRIPTION FACTOR PARTNERS OF RUNX2 Numerous transcription factors are known to interact with Runx2 (Figs. 1 and 2). Some provide costimula-tory signals, while others directly repress Runx2 function by affecting its DNA binding activity and/or transactivation potential. DNA binding proteins that interact and cooperate with Runx2 to activate gene expression include API (c-Fos and c-Jun), BMP-responsive Smads (Smadl and Smad5), Etsl, androgen and glucocorticoid receptors (AR, GR), several CAATT enhancer binding proteins (C/EBPs), Dlx5, Hesl, Menrn, and Oct-1 (Wotton et al., 1994; Sato et aL, 1998; Hanai et al., 1999; Ning and Robins, 1999; Lee etal., 2000; McLarren et al., 2000; Shirakabe et al., 2001; D'Alonzo et al., 2002; Gutierrez et al., 2002; Nishimura et at., 2002; Hassan et al., 2004; Sowa et al., 2004; Hata et al., 2005; Inman et al., 2005). Most of these proteins interact with either the DNA binding domain or the activation domain of Runx2, although the binding sites for some have not been defined (Fig. 1). It is generally believed that these transcription factors cooperate with Runx2 to facilitate the recruitment of coactivators and the assembly of higher-order transactvation Figure 2. Runx2 is an organizing hub for gene expression. Runx2 binds a specific DNA sequence to regulate gene transcription and is localized to discrete subcellular foci by the nuclear matrix. Runx2 interacts with CBF0, as well as other proteins ttiat are classified as compressors, coactivators, or transcription factors. Runx2 responds to numerous extracellular signals and is posttranslationally modified. omplexes. Some proteins, including Hesl, may block the binding of TLE compressors to Runx2 (McLamen et al., 2000). Others (e.g., Oct-1) may relieve negative constraints on DNA binding imposed by the carboxy-terminus (Inman et al., 2005). Still other cooperating transcription factors, such as API, steroid hormone receptors, Smadl, and Smad5, integrate Runx2 with cell signaling pathways and to the extracellular environment (Selvamurugan et al., 1998; Ning and Robins, 1999; Afzal et al., 2005). Mutations that affect Runx2-Smad Interactions are found in CCD patients and inhibit the ability of Runx2 to induce osteoblast differentiation after BMP stimulation (Zhang et al., 2000). Transcription factors that inhibit Runx2-dependent activation in mesenchymal cells or osteoblasts include C/EBP5, Dlx3, Lefl, Msx2, PPARy, Smad3, Statl, and Twist (McCarthy et al.,- 2000; Alliston et al., 2001; Shirakabe et al., 2001; Jeon et al., 2003; Kahler and Wes-tendorf, 2003; Kim et al., 2003b; Bialek et al., 2004; Hassan et al., 2004). These proteins, repress Runx2 via several mechanisms, including binding the Runt domain and preventing DNA binding (e.g., Lefl, PPARy, Twist) (Jeon et al., 2003; Kahler and Westendorf, 2003; Bialek et al., 2004), by sequestering Runx2 in the cytoplasm (e.g., C/EBP6, Statl) (McCarthy et al., 2000; Kim et al., 2003b), or by unknown mechanisms that involve binding in or around the nuclear matrix targeting domain of Runx2 (e.g., Dlx3) (Hassan et al., 2004). CO-ACTIVATORS OF RUNX2 While transcription factors bind specific DNA sequences to provide specificity for gene regulation, they typically lack enzymatic activities that are" required to modify chromatin structure and regulate mRNA production. Transcriptional activation by Runx proteins and other tis- sue-restricted transcription factors requires the recruitment of RNA polymerase II and general transcription factors to target promoters. This is typically accomplished indirectly through large proteins known, as coactivators, which are defined as proteins that do not bind DNA, but are required for transcriptional activation of gene expression (Roeder, 2005). Coactivators interact with a large number of transcription factors and some possess histone acetyltransferase (HAT) activity, which marks lysine residues in histones and other proteins with a removable acetyl group to facilitate chromatin relaxation and the recruitment of other cofactors and general transcription proteins. Runx2 interacts physically and/or functionally with several well-characterized coactivators, including p300, CBP, MOZ, and MORF. p300 interacts with the activation domain of Runx2 and enhances Runx2-dependent activation of the osteocalcin gene (Sierra et a!., 2003). This may involve the bridging of different promoter elements, including the vitamin D response element/within the osteocalcin promoter. Interestingly, the HAT domain is not required for p300 coactivatton of Runx2, but another reactivator possessing HAT activity, P/CAF, has an additive effect on Runx2-dependent osteocalcin expression when coexpressed with p300 (Sierra et al., 2003). The p300-related protein, CBP, also synergistically enhances Runx2 activity (J.J.W., unpublished data). Neither CBP nor P/CAF has been co-precipitated with Runx2. Two members of the MYST family of HATs, MOZ and MORF, however, do interact with the activation domain of Runx2 and enhance activation of the osteocalcin promoter (Pelletier et al., 2002). This interaction may be functionally relevant to in-tramembranous bone formation, as mice with a mutation in the MYST gene have craniofacial defects (Thomas et al., 2000). Grg5, pRb, and TAZ (transcriptional coactiva-tor with PD2 binding motif) are other proteins that enhance Runx2-mediated transactivation (Thomas et al., 2001; Cui et al., 2003; Wang et al., 2004). Although most Grg/ TLE proteins are compressors, Grg5 appears to be a dominant-negative form of longer Grg/TLE proteins, and thereby enhances Runx2 activity in vivo (Wang et al., 2004). CO-REPRESSORS OF RUNX2 Transcriptional repression requires I the recruitment of protein com- ? plexes that condense chromatin structure. Runt domain proteins and other transcription factors recruit HDACs, which remove acetyl groups from lysine residues on histone tails and contribute to a transcriptionally restrictive chromatin conformation. General HDAC inhibitors, such as trichostatin A, increase Runx2-mediated activation, and Runx2 associates with several HDACs, including HDAC3, HDAC4, and HDAC6 (Westendorf et al., 2002; Schroederetal., 2004; Vega et al., 2004). HDAC3 interacts with the amino-terminus of Runx2. Suppression of HDAC3 expression in differentiating MC3T3-E1 cells accelerates matrix mineralization and the expression of bone marker genes such as osteopontin, bone sialopro-tein, and osteocalcin (Schroeder et al., 2004). HDAC4 binds the Runt domain of Runx2 and blocks Its DNA binding activity in prehypertrophic chondrocytes (Vega et al., 2004). HD4C4-null mice display premature ossification Hue to early onset chondrocyte hypertrophy, and overex-pression of HDAC4 inhibits chondrocyte hypertrophy, suggesting that Runx2 activity Is controlled by HDAC4 in prehypertrophic chondrocytes (Vega et al., 2004). HDAC4 does not appear to be expressed at high levels in most osseous cells, but we detected It in an osteosarcoma cell line, UMR 106 (Schroeder et al., 2004; Vega et al., 2004). HDAC6 interacts with the carboxy-terminus of Runx2, and is recruited from the cytoplasm to the nucleus by Runx2 (Westendorf et al., 2002). TLE/Grg proteins and mSin3a also bind Runx2 and are components of large complexes that contain HDAC activity (Thirunavukkarasu et al., 1998; Chen and Courey, 2000; Javed et al., 2000; Lutterbach et al., 2000; McLarren et al., 2000; Sflverstein and Ekwall, 2005). It is not yet known whether Runx2 associates with multiple corepressor complexes, or whether aft of the compressors mentioned above are components of the same complex. Rynx2 also interacts with YAP, a mediator of Src/Yes signaling, in the cytoplasm and translocates it to the nuclear matrix where YAP represses Runx2-mediated activation of the osteocalcin promoter (Zaidi et al., 2004a). Finally, the proteosome degradation pathway decreases Runx2 protein levels to slow osteoblast differentiation (Tintut et al., 1999). Runx2 interacts with the E3 ubiq-uitin ligase, Smurfl (Zhao et.al., 2003). Transgenic overexpression of Smurfl in murine osteoblasts suppresses their differentiation and bone formation, while Smurfl -deficient mice develop age-dependent increases in bone mass (Zhao et al., 2004; Yamashita et al., 2005). The lysine residue(s) that is tagged with a polyubiquitin chain by Smurfl to target Runx2 to the proteosome has not been defined. RUNX2IN DEVELOPMENTAL SIGNALING PATHWAYS Five major signaling pathways (BMP/transforming growth factor-β TGFβ), FGF/EGF, Hedgehog, Notch, and Wnt) are extensively used during embryonic development. Runx2 is subject to multiple levels of control by most of these signaling pathways, but is also regulated by a variety of endocrine signals (e.g., parathyroid hormone (PTH), vitamin D, and androgen) and integrin cross-frnkage. Several of these Stimuli affect the Runx2 promoters and contribute to regulation of Runx2 expression. In addition, signaling pathways affect Runx2 function by activating enzymes that posttranslationally modify Runx2 and/or by mobilizing cofactors that interact with Runx2 to either positively or negatively regulate its DNA binding or transcriptional activities. Despite the well-recognized and documented importance of Runx2 in the developing skeleton, our understandlng of how Runx2 expression is regulated at the transcriptional level is just beginning. In osseous cells, Runx2 mRNA levels fluctuate during the cell cycle, peaking in Gl and dropping during 5 phase and mitosis (Galindo et al., 2005). Transcription of Runx2 is enhanced by BMPs, FGFs, and retinoic acid, as well as the hormones tamoxifen and dexamethasone (Tsuji et al., 1998; Gori et al., 1999; Lee et al., 1999, 2000; Zhou et al., 2000; Banerjee et al., 2001; Jimenez et al., 2001; Prince et al., 2001; Tou et al., 2001, 2003; Viereck et al., 2002; Zhang et al., 2002; Drissi et al., 2003; Kim et al., 2003a; Aberg et aL, 2004). Many of these studies only examined total Runx2 levels and did not discriminate between activation of its two promoters. However, BMP-2 appears to stimulate only type II (MASNS) Runx2 expression (Banerjee et al., 2001). The regulation of Runx2 expression during development is also subject to feedback regulation, as Runx2 increases BMP-2 expression in response to FGF (Choi et al., 2005) and negatively regulates its own expression (Ducy et al., 1999; Drissi et al., 2000). Runx2 levels are also negatively regulated by vitamin D3 (Drissi et al., 2002). tgf-β exerts different effects on Runx2 expression depending upon the cell type examined (Lee et al., 2000; Alliston et al., 2001; Viereck et al., 2002). The multitudes of FGFs, BMPs, and other cytokines, as well as their respective receptors that are present during development, and the potential for differential regulation of two Runx2 promoters contribute to the complex regulation and multifaceted roles of Runx2. BMP/TGFβ In addition to being transcriptionally regulated, Runx2 is functionally affected by cofactors that are direct mediators of developmental signaling pathways. The importance of BMP/TGF-β signaling for osteogenesis has long been recognized and was recently reviewed (Wan and Cao, 2005). Runx2 cooperates with BMPs or overexpressed Smads to induce osteogenic differentiation markers in various mesenchymal cell types (Lee et al., 2000; Selvamurugan et al., 2004). Smad proteins are intracellular effectors of BMP and TGF-/3 signals. R-Smads, are phosphorylated in response to extracellular ligand binding to the BMP receptor complex, dimerize with the coactivator Smad4, translocate to the nucleus, and modulate expression of target genes. BMP. signaling is mediated by R-Smads 1, 5, and 8, while TGF-β signaling operates through R-Smads 2 and 3. As previously mentioned, R-Smads bind to the carboxy^terminus of Runx2 within the NMTS (Hanai et at., 1999; Zhang et al., 2000; Afzal et al., 2005). Mitogen activated protein (MAP) kinase-dependent phosphorylation of Runx2 is required for interactions with Smads (Afzal et al., 2005). In cells that express Runx2, Smad complexes are directed to Runx2-containing nuclear matrix-bound foci, which are sites of active transcription (Zaidi et al., 2002; Afzal et al., 2005). The colo-calization of Runx, Smads, and other regulatory proteins to specific locations within the nucleus provide a mechanism for the integration of various signaling inputs into properly modulated gene expression. WNTS Wnt signaling pathways are crucial regulators of bone formation (Westendorf et al., 2004). Activating mutations in the Wnt coreceptor LrpS are linked to increased bone mass and osteoblast proliferation, while perturbation of Lrp5 function correlates with low bone mass and impaired osteoblast development (Kato et al., 2002; Gong et al., 2001). Wnts and activated forms of 0-catenin, an intermediate mediator of the canonical Wnt signaling pathway, synergize with BMP-2 to promote osteoblast differentiation and in vivo bone formation (Glass et al., 2005; Mbalaviele et al., 2005). Although Wnt signaling does not affect Runx2 expression, it may regulate Runx2 activity on some promoters. Lefl and TCFs, the downstream nuclear mediators of canonical Wnt signaling, bind to the Runt domain and inhibit gene activation by Runx factors (Kahler and Westendorf, 2003; Li et al., 2004). PTH Runx2 activity is also modulated by PTH and parathyroid-related peptides (PTHrPs). Therapeutically, PTH has drastically different effects on bone formation depending on its concentration and frequency of administration (Qin et al., 2004). During normal physiological situations, PTH is a major regulator of calcium homeostasis. The parathyroid gland secretes PTH in response to low serum calcium levels. PTH increases calcium resorption by the kidneys and intestine (mediated indirectly via increased synthesis of l,25(OH)2 vitamin D3) and stimulates bone resorption to liberate calcium from skeletal reservoirs. High levels of PTH and/or prolonged exposure to PTH decreases bone density, stimulates expression of genes involved in bone remodeling, and inhibits expression of several Runx2-regulated genes involved in bone formation (Swarthout etal., 2002). In contrast, ^ intermittent or low levels ofcPTH induce bone synthesis and increase bone mass. In osteoblasts, PTH activates heterotrimeric G-protein-eou-pled receptors, a variety of kinases (e.g., protein kinase A (PKA), protein kinase C (PKC), MAP kinases), and transcription factors (e.g., CREB, API, and Runx2) to stimulate proliferation (Clohisy etal., 1992; Fang et al., 1992; Tyson et al., 1999; Sel-vamurugan et al., 2000). Activation of Runx2 by PTH is at least partially mediated by the phosphorylation of a serine residue in the Runx2 activation domain by PKA (Selvamurugan etal., 2000). PTH-induced transcription of collagenase-3 requires physical interactions between API and Runx2 transcription factors (Selvamurugan et al., 1998; Hess et al., 2001; D'Alonzo etal., 2002). FGFs FGF proteins are also important regulators of Runx2 activity and skeletal development (Marie, 2003). FGF2 promotes bone formation in vivo and premature transformation of cartilage to bone (Coffin et al., 1995), whereas disruption of FGF2 decreases bone mass (Montero et al., 2000). FGF signaling in osteoblasts activates PKC and the ERK and p38 classes of MAP kinases (Xiao et al., 2002; Kim et al., 2003a). Activation of these pathways leads to phosphorylation of the carboxy-terminus of Runx2, increased DNA binding by Runx2, and enhanced expression from the osteocalcin promoter (Xiao et al., 2002; Kim et al., 2003a). ECM Runx2 transcriptional activity may also be modulated by cell contact with the extracellular matrix (ECM). Osteoblast differentiation requires a type I collagen-containing ECM. The MAP kinase cascade transmits signals from integrins at the cell surface to Runx2 in the nucleus. MEK/ERKs phosphorylate a serine residue in the activation domain of Runx2 (Xiaoetal., 2000). ECM production in MC3T3-E1 cells stimulates a dramatic increase in DNA binding activity by Runx2 and activation of the osteocalcin promoter (Xiao et al., 1997, 1998). Smalt-molecule inhibitors of MAP kinase signaling pathways impair osteocalcin activation by ECM (Xiao et al., 2002). Integrin signaling through MAP kinase cascades to Runx2 has also been implicated in cellular response to bone loading and mechanical stress (Wang et al., 2002; Ziros et al., 2002). INTEGRATION OF RUNX2 SIGNALS AT THE NUCLEAR MATRIX An intriguing question is how does Runx2 integrate all the aforementioned proteins and signal transduction cascades into functional units that regulate gene expression at the proper time and place during development. The answer may lie with its focal localization to the nuclear matrix. Runx2 is concentrated in approximately 300 discrete nuclear foci in nonmitotic cells (Zaidi et al., 2003; Young et al., 2004). As previously mentioned, a 31-amino acid region, termed the UMTS, is required for this interaction (Zaidi et al., 2001). The nuclear matrix is a protein-based scaffold that is resistant to detergent and high-salt extractions. It contributes to nuclear structure and participates in many cellular processes by concentrating specific proteins to certain locations (Bidwell et al.,- 1998; Zaidi et al., 2004b). The dynamic nature of interactions between the nuclear matrix and regulatory proteins during bone formation was revealed by the observation that sets of proteins associated with the osteoblast nuclear matrix change as cells undergo the characteristic transitions from proliferation to ECM synthesis to mineralization (Dworetzky et al., 1990). This observation implied that nuclear matrix-bound regulatory proteins direct the differentiation process. Bidwell et al. (1993) subsequently identified two nuclear matrix proteins that were exclusively expressed in osseous cells and bound to the osteocalcin promoter, one of which they referred to as NMP-2. NMP-2 associated with DNA sequences that resembled the Runx family consensus binding site (Mer-riman et al., 1995; Banerjee et al., 1996). The similarity of the NMP-2 and Runx binding sites suggested that NMP-2 was composed of one or more Runx proteins, a hypothesis that was confirmed through antibody supershift experiments, which Showed that Runx2 is the primary component of NMP*2 (Banerjee et al., 1997). Nearly all Runx proteins detected in cells are incorporated into discrete foci associated with the nuclear matrix (Undenmuth et al., 1997; Zeng et al., 1997). A large fraction of these foci are sites of active transcription, containing RNA polymerase II iso-forms and colocalizing with sites of BrUTP incorporation (Zeng et al., 1998). The in vivo formation of Runx-containing foci was examined in living cells with GFP-Runx2 fusion proteins and fluorescence-recovery-after-photobleaching (FRAP) experiments (Harrington et al., 2002). GFP-Runx2 fusion proteins associate with stable foci with a 'half-time of recovery of 10 seconds.; Irr contrast, GFP-Runx2 fusion proteins lacking the NMTS are diffusely organized and have a much greater mobility within the nucleus, comparable to that of GFP alone. Young et al. (2004) recently quantified the spatial organization of Runx2 foci with a bioinformatics approach. Runx2-dependent gene regulation requires the recruitment of co-activators or compressors to target promoters. Thus, it is unsurprising that Runx binding cofactors (e.g., HDAC6, Smad, YAP), as well as elements of core transcription machinery, colocalize with Runx proteins in the nuclear matrix-associated foci (Westendorf et al., 2002; Zaidi et al., 2002, 2004a). Smads and YAP are recruited to the nuclear matrix by Runx proteins, but cannot interact directly with the nuclear matrix (Zaidi et al., 2002, 2004a). Other cofactors (e.g., TLE1, TLE2) bind to the nuclear matrix independently of their association with Runx proteins (Javed et al., 2000). Deletion of theTLE binding domain at the carboxy-terminus of Runx2 does not disrupt binding of Runx2, TLE1, or TLE2 to the nuclear matrix, but does disrupt Runx/TLE colocalization and impairs repression of the osteocalcin reporter by TLEs (Javed et al., 2000). As described above, the association of Runx proteins with the nuclear matrix is essential for in vivo biological activity of Runx proteins. Transgenic mice were made in which the wild-type allele of Runx2 was replaced with a mutant allele containing a premature stop codon; consequently, the gene product lacks the NMTS as well as the car-boxy-terminal transcriptional repression domain (Choi et al., 2001). The phenotype of this mouse is essentially identical to that of complete knockout alleles (Komori et al., 1997; Otto et al., 1997). RUNX2 IN CANCER Runx2 is expressed at high levels in breast and prostate cancers, which have high propensity to metastasize to the skeleton (Mundy, 2002; Roodman, 2004). These metastases are of considerable clinical significance, as they can cause serious symptoms, including severe pain, hypercalcemia, nerve compression, and pathologic fractures. Breast cancer metastases are most often osteolytic, causing bone destruction via stimulatory interactions with osteoclasts and impaired osteoblast differentiation. Interestingly, introduction of Runx2 proteins with point mutations in the NMTS into breast cancer cells attenuates their ability to inhibit in vitro osteogenic differentiation of mouse marrow stromal cells and their ability in vivo to generate osteolytic lesions in bone (Barnes et al., 2004; Javed et al., 2005). SUMMARY AND FUTURE DIRECTIONS Runx2 is essential for mesenchymal condensation, osteoblast development, and osteoblast maturation. Like all developmental^ important proteins, multiple mechanisms actively control Runx2 activity. These mechanisms include those that we have discussed (i.e., transcriptional regulation, alternative splicing, post-translational modifications, sub-nuclear localization, and interactions with cofactors), as well as some we have not mentioned, notably translation a I regulation. A basal level of Runx2 is required for proper osteoblast proliferation and differentiation. A variety of in vitro studies, transgenic mice, and human genetic disorders demonstrate that excessive or insufficient levels of functional Runx2 have devastating consequences on osteoblasts and skeletal structure (Komori et aU 1997; Otto et al., 1997; Choi et al., 2001; Uu et al., 2001; Geoffroy et al., 2002; Yo-shida et al., 2002b; Xiao et al., 2004; Galindo et al., 2005). Two Runx2 promoters are being actively studied to gain a better understanding of how Runx2 is regulated in osteoblasts. It is equally important to consider the expression patterns of transcription factors and cofactors that interact with Runx2 at specific developmental stages. Runx2 is not sufficient to activate gene expression and is expressed at times when some of its target genes, most notably osteocalcin, are not produced. The recent findings that Twist and Dlx3 affect Runx2 transcriptional activity in early mesenchymal progenitors and preosteoblasts, but not in mature osteoblasts, demonstrate that developmental cues control Runx2 activity (Bialek et al., 2004; Hassan et al., 2004). Thus, developmental regula- E tion of Runx2 and cofactor interactions, as well as the possibility that interactions, especially those with i other transcription factors, may be either stimulatory or inhibitory depending on the promoter/enhancer context, are important considerations. The identification of switches, perhaps posttranslational modifications, that convert Runx2 from an E activator to a repressor (and vice versa) and lead to cofactor recruitment at the proper time and place will provide a better understanding of Runx2 regulation. Finally, insights into the mechanisms by which proteins and developmental signals affect Runx2 subcellular localization to the nuclear matrix will be needed to understand how it regulates gene expression. In conclusion, Runx proteins are essential for the developing embryo, but also have crucial roles in postnatal tissue function and cardnogenesis. Runx2 is a central organizing hub for transcriptional regulation in mesenchymal precursors E and osteoblasts. It receives input from extracellular signals and intracellular proteins to promote or suppress the- expression of other genes. Additional proteins will be added to the list of Runx2-interacting proteins in the next decade. Efforts will also be made to understand how the interactions are regulated by extracellular stimuli. The advancement of genomic, proteomic, and bioinformatics fields will allow for the integration of E these signals and a better understanding of how Runx2 organizes gene expression.

RUNT DOMAIN PROTEINS: DEVELOPMENTAL ROLES

Runx2 is one of three mammalian genes that encode proteins homologous to Drosophila Runt, and that are crucial for proper embryonic development. Runx2 (Cbfal) is required for mesenchymal condensation, osteoblast differentiation from mesenchymal stem cells, chondrocyte hypertrophy, and vascular invasion of developing skeletons (Komori et al., 1997; Otto et al., 1997). Introduction of Runx2 into mesenchymal stem cells stimulates differentiation along the osteoblast lineage (Byers and Garcia, 2004). In contrast, overexpression of Runx2 in type I collagen-positive and lineage-restricted osteoblasts blocks terminal differentiation and increases bone resorption (Geoffroy et al., 2002; Liu et al., 2001). These data indicate that Runx2 Is necessary for osteoblast lineage specification from mesenchymal precursors, but Runx2 levels must be tightly regulated in lineage-determined osseous cells. Factors influencing Runx2 expnression and function are discussed in this review.

The other Runx factors also have roles in skeletal development. Runxl and Runx3 contribute to endochondral bone formation (Brenner et -al., 2004; Uan et al., 2003; Yoshida et al., 2004); however, their essential functions during em-bryogenesis are manifested in other tissues. Runxl is essential for definitive hematopoiesis, and is frequently mutated in cancer stem cells that give rise to acute leuke-mias (Miyamoto et al., 2000; Miyoshi et al., 1991; Okada et al., 1998). Runx3 is needed for the development of dorsal root pmg'ia proprioceptive neurons and proper control of gastric epithelium growth (Levanon et al., 2002; Li et al., 2002).

The important roles of Runx2 in human development are evident in CCD, a rare autosomal dominant disease. Heterozygous loss of function of Runx2 through a variety of mutations (e.g., deletions, point mutations, insertions, and mis-sense mutations) segregate with abnormal skeletal phenotypes in CCD families (Otto etal., 2002; Yoshida et aL, 2002b). These alterations affect just one allele of Runx2 on chromosome 6q21 and

Figure 1. Runx2 contains multiple functional domains. Functional activities of Runx2 domains are shown at the top of the figure. Alternative amino-termini (MASNS and MRIPV) are generated from two promoters. Shapes below the Runx2 diagram summarize the various types of mutations found in the CCD families in these regions. Diamond, missense mutation; triangle pointing up, insertion/ frameshift; triangle pointing down, deletion; circle, splice; square, nonsense mutation. The bottom part of the figure shows the relative positions of protein/cofactor binding to Runx2. Interaction domains for proteins listed below the dotted line have not bean defined.

would be predicted to decrease wild-type Runx2 levels and activity by one half. Runx2 haploinsuffi-ciency in mice causes similar phe-notypes as observed in CCD families, with a primary defect in intramembranous bone formation (Komori et al., 1997; Otto et al., 1997). As one may suspect given the importance of DNA binding to a transcription factor, many mutations in CCD individuals are found in the DN A binding domain of Runx2 (Fig. 1). However, other mutations are present in regions proximal and distal to the DNA binding domain. These mutations suggest important roles for other parts of Runx2 in skeletal development. Indeed, CCD mutations in the Runx2 carboxy-tenmi-nus impair Smad interaction and transcriptional activation (Zhang et al., 2000). Moreover, transgenic mice lacking the carboxy-terminus of Runx2 develop a CCD-like pheno-type (Choi et al., 2001).

Runx2 was originally cloned from calf thymus nuclei as core binding factor alpha (Cbfal) and later in osteoblasts as nuclear matrix protein 2 (NMP2), osteoblast-specific complex (OBSC), and osteoblast-specific factor 2 (OSF2) (Banerjee et al., 1996; Geoffroy et al., 1995; Merriman et al., 1995; Wang and Speck, 1992). It is also commonly referred to as polyoma enhancer binding protein 2aA (PEBP2aA) and acute myeloid leukemia protein 3 (AML3) (Kamachi et al., 1990; Levanon et al., 1994). The Runx nomenclature was established to unify-these and other names given by various labs around the world to the same developmentally essential proteins. The Runx terminology reflects the homology of mammalian proteins to the founding member of the family, Runt, and the crucial roles of these proteins in developmental processes (van Wi-jnen et al., 2004).

While knockout experiments and temporal and spatial expression patterns demonstrate nonredundant functions of the Runx genes in the skeleton and' other tissues during embryogenesis and postnatal development (Levanon et al., 2001; Lian et al., 2003), numerous cell types simultaneously express multiple Runx proteins. An important unresolved issue is whether Runx proteins have functionally identical roles#when they are coexpressed. In vitro molecular and biochemical experiments indicate that all Runx proteins bind the same DNA sequence and have similar overall structure and functional-domains. However, attempts to rescue Runxl-deficient embryos with Runxl-Runx2 or Runxl-Runx3 chimeras revealed that the carboxy-terminal regions of Runx factors, especially Runx2, convey distinct roles during the development of some tissues (Fukushima-Nakase et al., 2005). Numerous proteins associate with Runx2, and a subset interacts with the carboxy-terminus. In this review, we summarize the biochemical interactions that affect Runx2 transcriptional activity in osteoblasts.

STRUCTURAL AND FUNCTIONAL DOMAINS OF RUNX2

A highly conserved DNA binding and protein-protein interaction motif, termed the Runt domain, is a characteristic and essential feature of Runx proteins (Fig. 1). Structural studies of the Runxl Runt domain show that it forms an s-type immunoglobulin fold (Backstrom et al., 2002; Warren et al., 2000). This 128-amino acid domain mediates binding of all Runx proteins to the same DNA sequence, 5'-PuAC-CPuCA-3', or its complement, 5'-TGPyGGTPy-3' (Kamachi et al., 1990; Meyers et al., 1993; Wang et al., 1993). The Runt domain also binds other proteins. CBF/3 was the first Runt domain-interacting protein identified (Kamachi et al., 1990; Wang and Speck, 1992). CBF/3 does not contact DNA itself, but enhances the ability of Runt domain proteins to bind DNA by inducing structural changes that unmask their DNA interaction surface (Tahirov et al., 2001). CBF/3 also stabilizes Runt domain proteins by protecting them from ubiquitin-me-diated degradation (Huang et al., 2001). Point mutations that map to Runt domain residues along the Runx2-CBFЈ dimer interface are found in individuals with CCD (Otto et al., 2002; Warren et al., 2000). These data hinted that CBF0 might be essential for osteogenesis. CBF/3-deficient mice died at ED 12 with similar phenotypes as Runxl -deficient animals (Niki et al„ 1997; Wang et al., 1996). Because skeletal development begins after this time, tissue-specific depletion of CBFfi was needed to demonstrate that it is es* sential for proper skeletal development (Kundu et al., 2002; Miller et al., 2002; Yoshida et al., 2002a).

NUCLEAR AND SUBNUCLEAR LOCALIZATION OF RUNX2

Runx2 transcriptional activity is dependent on Its localization to the nucleus and to discrete subnuclear foci. Two distinct domains direct Runx proteins to the nucleus and nuclear matrix (Fig. l).The nuclear localization signal (NLS) is a short basic stretch of nine amino acids that are immediately carboxy-ter-mina) to the Runt domain (Kanno et al., 1998; Thirunavukkarasu et al., 1998). The nuclear matrix targeting signal (NMTS) is 31 amino acids in length, located near the carboxy-terminus of the protein, and is the only region other than the Runt domain whose structure has been solved. The NMTS forms a helix-loop-helix structure and directs Runx factors to subnuclear foci (Kanno et al., 1998; Tang et al., 1999; Zeng et al., 1997, 1998). This localization is necessary for Runx2-dependent transactivation (Zaidi et al., 2001). Deletion of the NMTS prevents Runx2'localization to nuclear foci and increases the mobility of Runx proteins (Harrington et al., 2002). Moreover, deletion of or mutations in the NMTS affect the size, packing, and spatial; randomness of Runx2 subnuclear. foci (Young et al., 2004). In vivo, mice homozygous for Runx2 lacking the NMTS and residues car-boxy-terminal to it failed to develop mineralized tissue due to a matura-tional arrest of osteoblasts (Choi et al., 2001). Specific point mutations within the NMTS also convey a dominant-negative function in Runx2-positive breast cancer cells (Javed et al., 2005). The mechanisms by which the NMTS mediates punctate distribution and functional effects are not known, but likely involve interactions with other nuclear proteins. As will be discussed below, numerous proteins including histone deacetylase (HDAC) 6, Smads, and Yes-associated protein (YAP) associate with residues in or near the NMTS.

ACTIVATION DOMAINS OF RUNX2

Runx2 contains multiple activation and repression domains that contribute to Its transcriptional activity. The primary activation domain of Runx proteins is a proline-serine-threonine (PST)-rich region that is distal to the NLS and immediately proximal to the NMTS (Kanno et al., 1998; Thirunavukkarasu et a I., 1998). This domain induces the transcriptional activation of a heterologous promoter when fused to an appropriate DNA binding domain. It contributes to interactions of Runx2 with coactivator proteins, notably p300, MOZ, and MORF (Pel-letier et al., 2002; Sierra et al., 2003). The amino-termini of Runx proteins may also be necessary for maximal transcriptional activation, but they are not sufficient to activate a heterologous promoter (Kanno et al., 1998; Thirunavukkarasu et al., 1998). This region of Runx2 contains a polyglutamine and polyalanine (QA) domain that is not present in Runxl or Runx3. The QA region is frequently altered by genetic mutations in CCD families (Otto et al., 2002). Polymorphic expansions or contractions of the QA domain may also influence skeletal strength and morphology in humans and other species (Fondon and Garner, 2004; Vaughan et al., 2002). In addition to its activation potential, the amino-terminus of Runx2 has repressive activity and can inhibit DNA binding (Inman et al., 2005; Schroeder et al., 2004), suggesting it may be a context-dependent regulator of Runx2 activity.

REPRESSION DOMAINS OF RUNX2

Relative to their functions as transcriptional activators, the role of mammalian Runx proteins as transcriptional repressors has been underappreciated. The repressive function of mammalian Runx proteins should not be surprising because the Drosophila protein, Runt, was originally described as a transcriptional repressor and is now recognized as a context-dependent activator or repressor (Manoukian and Krause, 1993; Swantek and Gergen, 2004; Tsai and Gergen, 1994). Numerous regions of Runx2 contribute to transcriptional repression. The last five amino acids of Runx proteins, VWRPY, are perfectly conserved and mediate interactions with several members of the transducin-like enhancer of split (TLE)/Groucho (Grg) family of proteins (Gasperowicz and Otto, 2005). This region associates with TLE2, but it is dispensable for interactions with TLE1 and TLE3 (Javed et al., 2000; McLarren et al., 2000; Thirunavukkarasu et al., 1998). A second repression domain in Runx2 that may contribute to some TLE interactions is found in and around the NMTS (Kanno et al., 1998; Thirunavukkarasu et al., 1998). This region also associates with HDAC6 (Westendorf et al., 2002). Like their amino-termini, the carboxy-termini of Runx proteins inhibit the interaction between the Runt domain and DNA or other proteins (Inman et al., 2005; Kanno et al., 1998): A third repression domain is contained within the first 94 amino acids of Runx2 and mediates binding to HDAC3 (Schroeder et al., 2004). Finally, the Runt domain of Runx2 contributes to interactions with compressors and transcription factors, including HDAC4 and Lefl (Kahler and Westendorf, 2003; Vega et al., 2004), which may prevent DNA binding. Thus, multiple regions of Runx2 are involved in the recruitment of cofactors that repress gene transcription.

RUNX2 ISOFORMS

The single Runx2 gene in mammalian genomes encodes multiple transcripts, which are derived from two promoters and alternative splicing. Together, differential promoter usage and alternative splicing creates at least 12 distinct iso-forms (Levanon and Groner, 2004; Stock and Otto, 2005; Terry et al., 2004). Therefore, cells may express a complex array of Runx2 proteins that lack certain domains described above and exhibit specialized functions during osteoblast differentiation and maturation. Two promoters, PI (distal) and P2 (proximal), drive expression of the major Runx2 isoforms, type II and type I, respectively. This double promoter structure is conserved in human and murine Runx2 genes, as well as the other mammalian Runx genes, Runxl and Runx3 (Levanon and Groner, 2004). Runx2-I, containing 513 amino acids and beginning with MRIPV, is constitutively expressed in non-osseous mesenchymal cells, os-teoprogenitors, chondrocytes, and thymocytes (Banerjee et al., 2001). Osteoblasts express Runx2-I at consistent levels throughout osteoblast .differentiation (Banerjee et al., 2001). Runx2-H contains 528 amino adds, beginning with MASNS. Runx2-II expression is restricted to osseous cells, induced by bone mor-phogenetic protein (BMP) 2 in osteo-progenitors, and is increased during osteoblast differentiation (Banerjee etal., 2001; Ducy etal., 1997). During development, Runx2-II is expressed in mesenchymal condensations and mature chondrocytes of the developing axial skeleton (Lengner et al., 2002). Kunx2-H-deficient mice have a milder skeletal pheno-type than complete Runx2-null mice, exhibiting dwarfism due to defects in endochondral bone formation and osteopenia caused by reduced osteoblast maturation (Xiao et al., 2004, 2005). These results suggest that Runx2-I is sufficient for early osteoblast differentiation and in-tramembranous bone formation, and Runx2-II is necessary for complete osteoblast maturation and endochondral bone formation. The mechanism(s) for the isoform-spe-cific developmental roles are not known, but may include preferential promoter induction and/or binding of other proteins to the alternative ami-no-termini of the Runx2 proteins. A third N-terminal isoform, OSF2, encodes a protein in mice that is 68 residues longer than Runx2-II and begins with amino acids MLHSPH (Ducy et al., 1997); however, this sequence is not conserved in humans.

Alternative splicing of Runx2 ex-ons also generates multiple protein products (Stewart et al., 1997; Geoffroy et al., 1998; Terry et al., 2004). Some splice products antagonize full-length Runx2 proteins by competing for DNA binding sites. It is likely that other Runx2 isoforms will be capable of interacting with unique sets of cofactors that will influence their function. The tissue-specific expression patterns of all 12 Runx2 isoforms are not yet clear, but given the conservation of the Runx2 gene structure, it seems likely that at least some of these isoforms will have distinct developmental roles.

RUNX2 PROTEIN PARTNERS

As we have discussed, Runt domain proteins either activate or repress transcription of genes by binding the DNA sequence, TGPyGGPy. The Afunctional role of Runt domain proteins is conserved throughout evolution. It is best exemplified by the activation or repression of a Drosophila gene, slpl, by Runt depending on its combinatorial inter- action with one of two differentially expressed transcription factors, Opa or Ftz (Swantek and Gergen, 2004). Interactions with other proteins are equally as important for Runx2-dependent regulation of osteoblast development and differentiation. Although the biphasic temporal regulation of a single osteoblast , gene by Runx2 during development • has not yet been molecularly defined 1 as it has in Drosophila embryos, there are numerous reports describing agonistic or antagonistic effects of other proteins on Runx2-depen-dent activation of specific genes. Many Runx2 target genes are absent in cells that express Runx2, suggesting that Runx2 transcriptional activity is under complex and tight regulation in osseous cells, and is influenced by the temporal and spatial expression of other factors.

TRANSCRIPTION FACTOR PARTNERS OF RUNX2

Numerous transcription factors are known to interact with Runx2 (Figs. 1 and 2). Some provide costimula-tory signals, while others directly repress Runx2 function by affecting its DNA binding activity and/or transactivation potential. DNA binding proteins that interact and cooperate with Runx2 to activate gene expression include API (c-Fos and c-Jun), BMP-responsive Smads (Smadl and Smad5), Etsl, androgen and glucocorticoid receptors (AR, GR), several CAATT enhancer binding proteins (C/EBPs), Dlx5, Hesl, Menrn, and Oct-1 (Wotton et al., 1994; Sato et aL, 1998; Hanai et al., 1999; Ning and Robins, 1999; Lee etal., 2000; McLarren et al., 2000; Shirakabe et al., 2001; D'Alonzo et al., 2002; Gutierrez et al., 2002; Nishimura et at., 2002; Hassan et al., 2004; Sowa et al., 2004; Hata et al., 2005; Inman et al., 2005). Most of these proteins interact with either the DNA binding domain or the activation domain of Runx2, although the binding sites for some have not been defined (Fig. 1). It is generally believed that these transcription factors cooperate with Runx2 to facilitate the recruitment of coactivators and the assembly of higher-order transactvation

Figure 2. Runx2 is an organizing hub for gene expression. Runx2 binds a specific DNA sequence to regulate gene transcription and is localized to discrete subcellular foci by the nuclear matrix. Runx2 interacts with CBF0, as well as other proteins ttiat are classified as compressors, coactivators, or transcription factors. Runx2 responds to numerous extracellular signals and is posttranslationally modified.

omplexes. Some proteins, including Hesl, may block the binding of TLE compressors to Runx2 (McLamen et al., 2000). Others (e.g., Oct-1) may relieve negative constraints on DNA binding imposed by the carboxy-terminus (Inman et al., 2005). Still other cooperating transcription factors, such as API, steroid hormone receptors, Smadl, and Smad5, integrate Runx2 with cell signaling pathways and to the extracellular environment (Selvamurugan et al., 1998; Ning and Robins, 1999; Afzal et al., 2005). Mutations that affect Runx2-Smad Interactions are found in CCD patients and inhibit the ability of Runx2 to induce osteoblast differentiation after BMP stimulation (Zhang et al., 2000). Transcription factors that inhibit Runx2-dependent activation in mesenchymal cells or osteoblasts include C/EBP5, Dlx3, Lefl, Msx2, PPARy, Smad3, Statl, and Twist (McCarthy et al.,- 2000; Alliston et al., 2001; Shirakabe et al., 2001; Jeon et al., 2003; Kahler and Wes-tendorf, 2003; Kim et al., 2003b; Bialek et al., 2004; Hassan et al., 2004). These proteins, repress Runx2 via several mechanisms, including binding the Runt domain and preventing DNA binding (e.g., Lefl, PPARy, Twist) (Jeon et al., 2003; Kahler and Westendorf, 2003; Bialek et al., 2004), by sequestering Runx2 in the cytoplasm (e.g., C/EBP6, Statl) (McCarthy et al., 2000; Kim et al., 2003b), or by unknown mechanisms that involve binding in or around the nuclear matrix targeting domain of Runx2 (e.g., Dlx3) (Hassan et al., 2004).

CO-ACTIVATORS OF RUNX2

While transcription factors bind specific DNA sequences to provide specificity for gene regulation, they typically lack enzymatic activities that are" required to modify chromatin structure and regulate mRNA production. Transcriptional activation by Runx proteins and other tis- sue-restricted transcription factors requires the recruitment of RNA polymerase II and general transcription factors to target promoters. This is typically accomplished indirectly through large proteins known, as coactivators, which are defined as proteins that do not bind DNA, but are required for transcriptional activation of gene expression (Roeder, 2005). Coactivators interact with a large number of transcription factors and some possess histone acetyltransferase (HAT) activity, which marks lysine residues in histones and other proteins with a removable acetyl group to facilitate chromatin relaxation and the recruitment of other cofactors and general transcription proteins. Runx2 interacts physically and/or functionally with several well-characterized coactivators, including p300, CBP, MOZ, and MORF. p300 interacts with the activation domain of Runx2 and enhances Runx2-dependent activation of the osteocalcin gene (Sierra et a!., 2003). This may involve the bridging of different promoter elements, including the vitamin D response element/within the osteocalcin promoter. Interestingly, the HAT domain is not required for p300 coactivatton of Runx2, but another reactivator possessing HAT activity, P/CAF, has an additive effect on Runx2-dependent osteocalcin expression when coexpressed with p300 (Sierra et al., 2003). The p300-related protein, CBP, also synergistically enhances Runx2 activity (J.J.W., unpublished data). Neither CBP nor P/CAF has been co-precipitated with Runx2. Two members of the MYST family of HATs, MOZ and MORF, however, do interact with the activation domain of Runx2 and enhance activation of the osteocalcin promoter (Pelletier et al., 2002). This interaction may be functionally relevant to in-tramembranous bone formation, as mice with a mutation in the MYST gene have craniofacial defects (Thomas et al., 2000). Grg5, pRb, and TAZ (transcriptional coactiva-tor with PD2 binding motif) are other proteins that enhance Runx2-mediated transactivation (Thomas et al., 2001; Cui et al., 2003; Wang et al., 2004). Although most Grg/ TLE proteins are compressors, Grg5 appears to be a dominant-negative form of longer Grg/TLE proteins, and thereby enhances Runx2 activity in vivo (Wang et al., 2004).

CO-REPRESSORS OF RUNX2

Transcriptional repression requires I the recruitment of protein com- ? plexes that condense chromatin structure. Runt domain proteins and other transcription factors recruit HDACs, which remove acetyl groups from lysine residues on histone tails and contribute to a transcriptionally restrictive chromatin conformation. General HDAC inhibitors, such as trichostatin A, increase Runx2-mediated activation, and Runx2 associates with several HDACs, including HDAC3, HDAC4, and HDAC6 (Westendorf et al., 2002; Schroederetal., 2004; Vega et al., 2004). HDAC3 interacts with the amino-terminus of Runx2. Suppression of HDAC3 expression in differentiating MC3T3-E1 cells accelerates matrix mineralization and the expression of bone marker genes such as osteopontin, bone sialopro-tein, and osteocalcin (Schroeder et al., 2004). HDAC4 binds the Runt domain of Runx2 and blocks Its DNA binding activity in prehypertrophic chondrocytes (Vega et al., 2004). HD4C4-null mice display premature ossification Hue to early onset chondrocyte hypertrophy, and overex-pression of HDAC4 inhibits chondrocyte hypertrophy, suggesting that Runx2 activity Is controlled by HDAC4 in prehypertrophic chondrocytes (Vega et al., 2004). HDAC4 does not appear to be expressed at high levels in most osseous cells, but we detected It in an osteosarcoma cell line, UMR 106 (Schroeder et al., 2004; Vega et al., 2004). HDAC6 interacts with the carboxy-terminus of Runx2, and is recruited from the cytoplasm to the nucleus by Runx2 (Westendorf et al., 2002). TLE/Grg proteins and mSin3a also bind Runx2 and are components of large complexes that contain HDAC activity (Thirunavukkarasu et al., 1998; Chen and Courey, 2000; Javed et al., 2000; Lutterbach et al., 2000; McLarren et al., 2000; Sflverstein and Ekwall, 2005). It is not yet known whether Runx2 associates with multiple corepressor complexes, or whether aft of the compressors mentioned above are components of the same complex. Rynx2 also interacts with YAP, a mediator of Src/Yes signaling, in the cytoplasm and translocates it to the nuclear matrix where YAP represses Runx2-mediated activation of the osteocalcin promoter (Zaidi et al., 2004a).

Finally, the proteosome degradation pathway decreases Runx2 protein levels to slow osteoblast differentiation (Tintut et al., 1999). Runx2 interacts with the E3 ubiq-uitin ligase, Smurfl (Zhao et.al., 2003). Transgenic overexpression of Smurfl in murine osteoblasts suppresses their differentiation and bone formation, while Smurfl -deficient mice develop age-dependent increases in bone mass (Zhao et al., 2004; Yamashita et al., 2005). The lysine residue(s) that is tagged with a polyubiquitin chain by Smurfl to target Runx2 to the proteosome has not been defined.

RUNX2IN DEVELOPMENTAL SIGNALING PATHWAYS

Five major signaling pathways (BMP/transforming growth factor-β TGFβ), FGF/EGF, Hedgehog, Notch, and Wnt) are extensively used during embryonic development. Runx2 is subject to multiple levels of control by most of these signaling pathways, but is also regulated by a variety of endocrine signals (e.g., parathyroid hormone (PTH), vitamin D, and androgen) and integrin cross-frnkage. Several of these Stimuli affect the Runx2 promoters and contribute to regulation of Runx2 expression. In addition, signaling pathways affect Runx2 function by activating enzymes that posttranslationally modify Runx2 and/or by mobilizing cofactors that interact with Runx2 to either positively or negatively regulate its DNA binding or transcriptional activities.

Despite the well-recognized and documented importance of Runx2 in the developing skeleton, our understandlng of how Runx2 expression is regulated at the transcriptional level is just beginning. In osseous cells, Runx2 mRNA levels fluctuate during the cell cycle, peaking in Gl and dropping during 5 phase and mitosis (Galindo et al., 2005). Transcription of Runx2 is enhanced by BMPs, FGFs, and retinoic acid, as well as the hormones tamoxifen and dexamethasone (Tsuji et al., 1998; Gori et al., 1999; Lee et al., 1999, 2000; Zhou et al., 2000; Banerjee et al., 2001; Jimenez et al., 2001; Prince et al., 2001; Tou et al., 2001, 2003; Viereck et al., 2002; Zhang et al., 2002; Drissi et al., 2003; Kim et al., 2003a; Aberg et aL, 2004). Many of these studies only examined total Runx2 levels and did not discriminate between activation of its two promoters. However, BMP-2 appears to stimulate only type II (MASNS) Runx2 expression (Banerjee et al., 2001). The regulation of Runx2 expression during development is also subject to feedback regulation, as Runx2 increases BMP-2 expression in response to FGF (Choi et al., 2005) and negatively regulates its own expression (Ducy et al., 1999; Drissi et al., 2000). Runx2 levels are also negatively regulated by vitamin D3 (Drissi et al., 2002). tgf-β exerts different effects on Runx2 expression depending upon the cell type examined (Lee et al., 2000; Alliston et al., 2001; Viereck et al., 2002). The multitudes of FGFs, BMPs, and other cytokines, as well as their respective receptors that are present during development, and the potential for differential regulation of two Runx2 promoters contribute to the complex regulation and multifaceted roles of Runx2.

BMP/TGFβ

In addition to being transcriptionally regulated, Runx2 is functionally affected by cofactors that are direct mediators of developmental signaling pathways. The importance of BMP/TGF-β signaling for osteogenesis has long been recognized and was recently reviewed (Wan and Cao, 2005). Runx2 cooperates with BMPs or overexpressed Smads to induce osteogenic differentiation markers in various mesenchymal cell types (Lee et al., 2000; Selvamurugan et al., 2004). Smad proteins are intracellular effectors of BMP and TGF-/3 signals. R-Smads, are phosphorylated in response to extracellular ligand binding to the BMP receptor complex, dimerize with the coactivator Smad4, translocate to the nucleus, and modulate expression of target genes. BMP. signaling is mediated by R-Smads 1, 5, and 8, while TGF-β signaling operates through R-Smads 2 and 3. As previously mentioned, R-Smads bind to the carboxy^terminus of Runx2 within the NMTS (Hanai et at., 1999; Zhang et al., 2000; Afzal et al., 2005). Mitogen activated protein (MAP) kinase-dependent phosphorylation of Runx2 is required for interactions with Smads (Afzal et al., 2005). In cells that express Runx2, Smad complexes are directed to Runx2-containing nuclear matrix-bound foci, which are sites of active transcription (Zaidi et al., 2002; Afzal et al., 2005). The colo-calization of Runx, Smads, and other regulatory proteins to specific locations within the nucleus provide a mechanism for the integration of various signaling inputs into properly modulated gene expression.

WNTS

Wnt signaling pathways are crucial regulators of bone formation (Westendorf et al., 2004). Activating mutations in the Wnt coreceptor LrpS are linked to increased bone mass and osteoblast proliferation, while perturbation of Lrp5 function correlates with low bone mass and impaired osteoblast development (Kato et al., 2002; Gong et al., 2001). Wnts and activated forms of 0-catenin, an intermediate mediator of the canonical Wnt signaling pathway, synergize with BMP-2 to promote osteoblast differentiation and in vivo bone formation (Glass et al., 2005; Mbalaviele et al., 2005). Although Wnt signaling does not affect Runx2 expression, it may regulate Runx2 activity on some promoters. Lefl and TCFs, the downstream nuclear mediators of canonical Wnt signaling, bind to the Runt domain and inhibit gene activation by Runx factors (Kahler and Westendorf, 2003; Li et al., 2004).

PTH

Runx2 activity is also modulated by PTH and parathyroid-related peptides (PTHrPs). Therapeutically, PTH has drastically different effects on bone formation depending on its concentration and frequency of administration (Qin et al., 2004). During normal physiological situations, PTH is a major regulator of calcium homeostasis. The parathyroid gland secretes PTH in response to low serum calcium levels. PTH increases calcium resorption by the kidneys and intestine (mediated indirectly via increased synthesis of l,25(OH)2 vitamin D3) and stimulates bone resorption to liberate calcium from skeletal reservoirs. High levels of PTH and/or prolonged exposure to PTH decreases bone density, stimulates expression of genes involved in bone remodeling, and inhibits expression of several Runx2-regulated genes involved in bone formation (Swarthout etal., 2002). In contrast, ^ intermittent or low levels ofcPTH induce bone synthesis and increase bone mass. In osteoblasts, PTH activates heterotrimeric G-protein-eou-pled receptors, a variety of kinases (e.g., protein kinase A (PKA), protein kinase C (PKC), MAP kinases), and transcription factors (e.g., CREB, API, and Runx2) to stimulate proliferation (Clohisy etal., 1992; Fang et al., 1992; Tyson et al., 1999; Sel-vamurugan et al., 2000). Activation of Runx2 by PTH is at least partially mediated by the phosphorylation of a serine residue in the Runx2 activation domain by PKA (Selvamurugan etal., 2000). PTH-induced transcription of collagenase-3 requires physical interactions between API and Runx2 transcription factors (Selvamurugan et al., 1998; Hess et al., 2001; D'Alonzo etal., 2002).

FGFs

FGF proteins are also important regulators of Runx2 activity and skeletal development (Marie, 2003). FGF2 promotes bone formation in vivo and premature transformation of cartilage to bone (Coffin et al., 1995), whereas disruption of FGF2 decreases bone mass (Montero et al., 2000). FGF signaling in osteoblasts activates PKC and the ERK and p38 classes of MAP kinases (Xiao et al., 2002; Kim et al., 2003a). Activation of these pathways leads to phosphorylation of the carboxy-terminus of Runx2, increased DNA binding by Runx2, and enhanced expression from the osteocalcin promoter (Xiao et al., 2002; Kim et al., 2003a).

ECM

Runx2 transcriptional activity may also be modulated by cell contact with the extracellular matrix (ECM). Osteoblast differentiation requires a type I collagen-containing ECM. The MAP kinase cascade transmits signals from integrins at the cell surface to Runx2 in the nucleus. MEK/ERKs phosphorylate a serine residue in the activation domain of Runx2 (Xiaoetal., 2000). ECM production in MC3T3-E1 cells stimulates a dramatic increase in DNA binding activity by Runx2 and activation of the osteocalcin promoter (Xiao et al., 1997, 1998). Smalt-molecule inhibitors of MAP kinase signaling pathways impair osteocalcin activation by ECM (Xiao et al., 2002). Integrin signaling through MAP kinase cascades to Runx2 has also been implicated in cellular response to bone loading and mechanical stress (Wang et al., 2002; Ziros et al., 2002).

INTEGRATION OF RUNX2 SIGNALS AT THE NUCLEAR MATRIX

An intriguing question is how does Runx2 integrate all the aforementioned proteins and signal transduction cascades into functional units that regulate gene expression at the proper time and place during development. The answer may lie with its focal localization to the nuclear matrix. Runx2 is concentrated in approximately 300 discrete nuclear foci in nonmitotic cells (Zaidi et al., 2003; Young et al., 2004). As previously mentioned, a 31-amino acid region, termed the UMTS, is required for this interaction (Zaidi et al., 2001). The nuclear matrix is a protein-based scaffold that is resistant to detergent and high-salt extractions. It contributes to nuclear structure and participates in many cellular processes by concentrating specific proteins to certain locations (Bidwell et al.,- 1998; Zaidi et al., 2004b). The dynamic nature of interactions between the nuclear matrix and regulatory proteins during bone formation was revealed by the observation that sets of proteins associated with the osteoblast nuclear matrix change as cells undergo the characteristic transitions from proliferation to ECM synthesis to mineralization (Dworetzky et al., 1990). This observation implied that nuclear matrix-bound regulatory proteins direct the differentiation process. Bidwell et al. (1993) subsequently identified two nuclear matrix proteins that were exclusively expressed in osseous cells and bound to the osteocalcin promoter, one of which they referred to as NMP-2. NMP-2 associated with DNA sequences that resembled the Runx family consensus binding site (Mer-riman et al., 1995; Banerjee et al., 1996). The similarity of the NMP-2 and Runx binding sites suggested that NMP-2 was composed of one or more Runx proteins, a hypothesis that was confirmed through antibody supershift experiments, which Showed that Runx2 is the primary component of NMP*2 (Banerjee et al., 1997).

Nearly all Runx proteins detected in cells are incorporated into discrete foci associated with the nuclear matrix (Undenmuth et al., 1997; Zeng et al., 1997). A large fraction of these foci are sites of active transcription, containing RNA polymerase II iso-forms and colocalizing with sites of BrUTP incorporation (Zeng et al., 1998). The in vivo formation of Runx-containing foci was examined in living cells with GFP-Runx2 fusion proteins and fluorescence-recovery-after-photobleaching (FRAP) experiments (Harrington et al., 2002). GFP-Runx2 fusion proteins associate with stable foci with a 'half-time of recovery of 10 seconds.; Irr contrast, GFP-Runx2 fusion proteins lacking the NMTS are diffusely organized and have a much greater mobility within the nucleus, comparable to that of GFP alone. Young et al. (2004) recently quantified the spatial organization of Runx2 foci with a bioinformatics approach.

Runx2-dependent gene regulation requires the recruitment of co-activators or compressors to target promoters. Thus, it is unsurprising that Runx binding cofactors (e.g., HDAC6, Smad, YAP), as well as elements of core transcription machinery, colocalize with Runx proteins in the nuclear matrix-associated foci (Westendorf et al., 2002; Zaidi et al., 2002, 2004a). Smads and YAP are recruited to the nuclear matrix by Runx proteins, but cannot interact directly with the nuclear matrix (Zaidi et al., 2002, 2004a). Other cofactors (e.g., TLE1, TLE2) bind to the nuclear matrix independently of their association with Runx proteins (Javed et al., 2000). Deletion of theTLE binding domain at the carboxy-terminus of Runx2 does not disrupt binding of Runx2, TLE1, or TLE2 to the nuclear matrix, but does disrupt Runx/TLE colocalization and impairs repression of the osteocalcin reporter by TLEs (Javed et al., 2000).

As described above, the association of Runx proteins with the nuclear matrix is essential for in vivo biological activity of Runx proteins. Transgenic mice were made in which the wild-type allele of Runx2 was replaced with a mutant allele containing a premature stop codon; consequently, the gene product lacks the NMTS as well as the car-boxy-terminal transcriptional repression domain (Choi et al., 2001). The phenotype of this mouse is essentially identical to that of complete knockout alleles (Komori et al., 1997; Otto et al., 1997).

RUNX2 IN CANCER

Runx2 is expressed at high levels in breast and prostate cancers, which have high propensity to metastasize to the skeleton (Mundy, 2002; Roodman, 2004). These metastases are of considerable clinical significance, as they can cause serious symptoms, including severe pain, hypercalcemia, nerve compression, and pathologic fractures. Breast cancer metastases are most often osteolytic, causing bone destruction via stimulatory interactions with osteoclasts and impaired osteoblast differentiation. Interestingly, introduction of Runx2 proteins with point mutations in the NMTS into breast cancer cells attenuates their ability to inhibit in vitro osteogenic differentiation of mouse marrow stromal cells and their ability in vivo to generate osteolytic lesions in bone (Barnes et al., 2004; Javed et al., 2005).

SUMMARY AND FUTURE DIRECTIONS

Runx2 is essential for mesenchymal condensation, osteoblast development, and osteoblast maturation. Like all developmental^ important proteins, multiple mechanisms actively control Runx2 activity. These mechanisms include those that we have discussed (i.e., transcriptional regulation, alternative splicing, post-translational modifications, sub-nuclear localization, and interactions with cofactors), as well as some we have not mentioned, notably translation a I regulation. A basal level of Runx2 is required for proper osteoblast proliferation and differentiation. A variety of in vitro studies, transgenic mice, and human genetic disorders demonstrate that excessive or insufficient levels of functional Runx2 have devastating consequences on osteoblasts and skeletal structure (Komori et aU 1997; Otto et al., 1997; Choi et al., 2001; Uu et al., 2001; Geoffroy et al., 2002; Yo-shida et al., 2002b; Xiao et al., 2004; Galindo et al., 2005). Two Runx2 promoters are being actively studied to gain a better understanding of how Runx2 is regulated in osteoblasts. It is equally important to consider the expression patterns of transcription factors and cofactors that interact with Runx2 at specific developmental stages. Runx2 is not sufficient to activate gene expression and is expressed at times when some of its target genes, most notably osteocalcin, are not produced. The recent findings that Twist and Dlx3 affect Runx2 transcriptional activity in early mesenchymal progenitors and preosteoblasts, but not in mature osteoblasts, demonstrate that developmental cues control Runx2 activity (Bialek et al., 2004; Hassan et al., 2004). Thus, developmental regula- E tion of Runx2 and cofactor interactions, as well as the possibility that interactions, especially those with i other transcription factors, may be either stimulatory or inhibitory depending on the promoter/enhancer context, are important considerations. The identification of switches, perhaps posttranslational modifications, that convert Runx2 from an E activator to a repressor (and vice versa) and lead to cofactor recruitment at the proper time and place will provide a better understanding of Runx2 regulation. Finally, insights into the mechanisms by which proteins and developmental signals affect Runx2 subcellular localization to the nuclear matrix will be needed to understand how it regulates gene expression.

In conclusion, Runx proteins are essential for the developing embryo, but also have crucial roles in postnatal tissue function and cardnogenesis. Runx2 is a central organizing hub for transcriptional regulation in mesenchymal precursors E and osteoblasts. It receives input from extracellular signals and intracellular proteins to promote or suppress the- expression of other genes. Additional proteins will be added to the list of Runx2-interacting proteins in the next decade. Efforts will also be made to understand how the interactions are regulated by extracellular stimuli. The advancement of genomic, proteomic, and bioinformatics fields will allow for the integration of E these signals and a better understanding of how Runx2 organizes gene expression.

Runx2: A Master Organizer of Gene Transcription in Developing and Maturing Osteoblasts Tania M. Schroeder, Eric D. Jensen, and Jennifer J. WestendorfBirth Defects Research (Part C) 75:213-225, 2005. Wiley-Liss, Inc.

Runx2: A Master Organizer of Gene Transcription in Developing and Maturing Osteoblasts
Tania M. Schroeder, Eric D. Jensen, and Jennifer J. Westendorf
Birth Defects Research (Part C) 75:213-225, 2005. Wiley-Liss, Inc.