Jun Dimerization Protein 2 Controls Senescence and Differentiation via Regulating Histone Modification

Transcription factor, Jun dimerization protein 2 (JDP2), binds directly to histones and DNAs and then inhibits the p300-mediated acetylation both of core histones and of reconstituted nucleosomes that contain JDP2 recognition DNA sequences. JDP2 plays a key role as a repressor of adipocyte differentiation by regulation of the expression of the gene C/EBPδ via inhibition of histone acetylation. Moreover, JDP2-deficient mouse embryonic fibroblasts (JDP2−/− MEFs) are resistant to replicative senescence. JDP2 inhibits the recruitment of polycomb repressive complexes (PRC1 and PRC2) to the promoter of the gene encoding p16Ink4a, resulting from the inhibition of methylation of lysine 27 of histone H3 (H3K27). Therefore, it seems that chromatin-remodeling factors, including the PRC complex controlled by JDP2, may be important players in the senescence program. The novel mechanisms that underline the action of JDP2 in inducing cellular senescence and suppressing adipocyte differentiation are reviewed.


Introduction
The structure of chromatin, which influences numerous DNA-associated phenomena, such as transcription, replication, recombination, and repair, is controlled by a complex combination of histone modifications, ATP-dependent chromatin-remodeling enzymes, and nucleosome-assembly factors [1,2]. The modification of histones such as acetylation, phosphorylation, methylation, ubiquitination, sumoylation, and ADP-ribosylation can regulate the gene expression [1][2][3][4]. The chromatin consists of structural units known as nucleosomes. Each nucleosome consists of two histone H2A-H2B dimers, a histone (H3-H4) 2 tetramer, and DNA that is wrapped around the resultant histone octamers.
During chromatin assembly, a histone (H3-H4) 2 tetramer is formed before the two heterodimers of histones H2A and H2B are incorporated to form a nucleosome [5,6]. The regulation of transcription is associated with alterations in chromatin structure that include histone modifications and changes in nucleosome structure [7][8][9][10]. Compaction of the chromatin and organization of nuclesomes represent a barrier that has to be overcome prior to the activation of transcription. The N-terminal histone tails that protrude from nucleosomes do not play a significant role in nuclesome formation but, rather, they appear to act as docking sites for other proteins and protein complexes to regulate chromatin compaction [9].
The structure of chromatin changes to allow greater accessibility by transcription factors when a gene is to be activated [10]. It has been suggested that the change to a more accessible state not only involves the modification of histones and alterations in nucleosomal arrays but also results from changes in nucleosome integrity that are due to displacement of histones [11]. Furthermore, it has been demonstrated that histone chaperones play a critical role in these processes [12][13][14][15]. Thus, it is tempting to speculate that histone chaperones might be important for the compaction of chromatin, and it is now important to determine whether certain corepressors of transcription might influence the deposition and assembly of nucleosomes through the regulation of histone-chaperone activity. The transcription factor Jun dimerization protein 2 (JDP2) is a member of AP-1 family that binds to both AP-1 site and cAMP responsive element (CRE) site in various cis-elements of the target genes. It is generally accepted that the transcription factors have DNA-binding activities and then control their transcriptional activities by DNA binding. However, we found that JDP2 has not only DNA-binding activity but also histone-binding activity. Moreover, JDP2 bound the nucleosome in both DNA sequence-dependent or -independent manner. JDP2 also has the nucleosome assembly activities and the activities of inhibition of histone acetyltransferase and histone methyltransferase. Here, we describe the role of JDP2 on the adipocyte differentiation and replicative senescence in mouse embryonic fibroblasts through histone modification.

JDP2 Regulates the AP-1-Mediated Activation of Transcription
JDP2 has been identified as a binding partner of c-Jun in yeast two-hybrid screening experiments, based on the recruitment of the SOS system [16]. JDP2 forms heterodimers with c-Jun and represses the AP-1-mediated activation of transcription [16]. Similarly, JDP2 was isolated by yeast two-hybrid screening with activation transcription factor-2 (ATF-2) as the "bait" [17]. JDP2 was also shown to associate with both the CCAT/enhancer-binding protein gamma (C/EBPγ) [18] and the progesterone receptor [19]. JDP2 is expressed in many cell lines and represses the transcriptional activity of AP-1 [20]. Moreover, JDP2 is rapidly phosphorylated at threonine residue 148 when cells are exposed to UV irradiation, oxidative stress, or inhibitor-induced depressed levels of translation by c-Jun NH(2)-terminal kinase (JNK) [21]. Although a novel JNK-docking domain is necessary for the activated kinase (MAPK) p38-mediated phosphorylation of JDP2 at threonine residue 148, this domain is not sufficient for this process [22]. JDP2 binds to both cAMP-responsive element (CRE) and TPA-responsive elements (TREs) on DNA as a homodimer and as a heterodimer with ATF-2 and members of the Jun family, respectively [16,17]. JDP2 encoded an 18 kDa protein that is able to homodimerize as well as to form heterodimers with other AP-1 members, such as c-Jun, JunB, JunD, and ATF-2, and members of the C/EBP family, C/EBPγ and C/EBP homologous protein 10 (CHOP 10) [16,17] as well as Interferon regulatory factor-(IRF-) binding protein 1 [23]. Dimerization occurs through a conserved leucine zipper domain found all members of the AP-1 family. A basic domain located adjacent to the leucine zipper dimerization motif is responsible for the direct association with TRE and CRE [16,17]. In vitro study using the purified JDP2 protein showed that JDP2 forms the homotrimer but not the homodimer in our biochemical condition (unpublished data). JDP2 inhibits UV-induced apoptosis by suppressing the transcription of the p53 gene [24]. Given the roles of AP-1 in cellular transformation and the reported repression of Jun-and ATF-2-mediated transcription by JDP2, we have demonstrated that JDP2 inhibits the oncogenic transformation of chicken embryonic fibroblasts [25]. JDP2 also modulates the expression of cyclin D1 and p21, which have opposing effects on cell-cycle progression. JDP2 interferes with the progression of the cell cycle by reducing the levels of cyclin D1 and at the same time increases the expression of p21 [26,27]. The forced expression of JDP2 promotes the myogenic differentiation of C2C12 cells, which is accompanied by the formation of C2 myotubes and the strong expression of major myogenic markers. Moreover, the ectopic expression of JDP2 in rhabdomyosarcoma cells induces incomplete myogenesis and the incomplete formation of myotubes [27]. These cells become committed to differentiation via the p38-MAPK pathway [21,22]. A similar enhancement of cell differentiation was reported during the induction of osteoclast formation by the receptor activator of the nuclear factor kB (NFkB) ligand (RANKL) [28]. Unlike other members of the AP-1 family, the levels of JDP2 remain constant in response to a large variety of stimuli, such as UV, irradiation, and retinoic acid (RA), which affect the levels of other factors involved in cellcycle control. The induction of JDP2 expression was only observed during the differentiation of F9 cells to muscle cells and osteoclasts. Therefore, JDP2 may provide a threshold for exit from the cell cycle and a commitment to differentiation ( Figure 1). Further studies of the regulation of the cell cycle and the differentiation of cells induced by JDP2 should be very instructive. It is also interesting that JDP2 is one of the candidate oncoproteins that collaborate in the oncogenesis associated with the loss of p27 as the result of insertional mutations [29]. Recent study of tumor cells demonstrated that JDP2 was a tumor suppressor [30]. We have also found that JDP2 is a repressor of the activation of transcription via AP-1 and a negative regulator of the retinoic acid-(RA-) induced differentiation of mouse embryonic F9 cells and adipocyte differentiation [31][32][33].

JDP2 Inhibits Histone Acetyltransferase (HAT) Activity
We have reported previously that JDP2 represses the transactivation mediated by p300 [32]. Both p300 and ATF-2 have HAT activity [34,35]. It was recently shown that p300 acetylates ATF-2 protein in vitro at lysine residues 357 and 374 and that ATF-2 is essential for the acetylation of histones H4 and H2B in vivo [36,37]. We found that acetylation by p300 is inhibited in a dose-dependent manner by JDP2, when added exogenously. We also found that JDP2 was not acetylated by p300 under our experimental conditions. The inhibitory effect of JDP2 was detected on histone acetylation induced by p300, CREB-binding protein (CBP), p300/CBP-associated protein (PCAF), and general control nonrepressive 5 (GCN5). The overexpression of JDP2 apparently represses the RA-induced acetylation of lysines 8 and 16 of histone H4 and some amino terminal lysine residues of histone H3.

JDP2 Has Intrinsic Nucleosome-Assembly Activity In Vitro
The template activation factor-1β protein (TAF-Iβ protein), which is a component of the inhibition of histone acetylase complex (INHAT complex) identified by Seo et al. [38,39], is a histone chaperone that binds directly to core histones and facilitates the assembly of nucleosomes in vitro. JDP2 interacted directly with all the core histones tested and inhibited the p300-mediated acetylation of those histones. To our surprise, JDP2 also introduced supercoils into circular DNA in the presence of core histones, to levels similar to those observed for yeast CCG1-interacting factor 1 protein (yCia1p) and CCG1-interactiing factor (CIA1). Therefore, JDP2 appears to have significant histone chaperone activity in vitro [32]. We have also shown that the HAT-inhibitory activity of JDP2 is involved, to some extent, in the repression of transcription by JDP2, whereas the maximal capacity Journal of Biomedicine and Biotechnology of JDP2 to suppress the RA-mediated activation of the c-Jun promoter [32, Figure 1] and to suppress the adipocyte differentiation [33] requires the recruitment of histone deacetylases (HDACs).

JDP2 Suppresses Adipocyte Differentiation
JDP2 has been shown to play a role in the cellular differentiation of skeletal muscle, osteoclasts, adipocytes [27,33], and F9 cells [31]. It has been reported that JDP2 has an activity of tumor suppressor using prostate cancer cells [30]. Recently, we also found that JDP2-deficient mouse fibroblasts are resistant to replicative senescence [40]. Finally general inhibition of the AP-1 complex by expression of JDP2 specifically in the heart correlates with the induction of atrial dilatation [41].
In mammals, the strict control of adipocyte development, the mass of adipose tissue, the insulin sensitivity of adipocycles, and the appropriate metabolism of glucose and lipids are critical to the maintenance of energy homeostasis [42,43]. Adipogenesis, namely, the process whereby hormonal stimuli induce the differentiation of fibroblasts or mesenchymal cells to adipocytes, requires the organized and controlled expression of a cascade of transcription factors and the modification of the chromatin within preadipocytes [44][45][46][47]. The factors involved in adipocyte differentiation include a nuclear receptor known as peroxisome proliferation-activated receptor gamma (PPARγ) and a group of C/EBPs [48,49]. The rapid and transient induction of the expression of C/EBPβ and C/EBPδ is one of the earliest events in adipogenesis [46]. These transcription factors bind to specific sequences in the promoters of the C/EBPα gene and the PPARγ gene, inducing their expression, which, in turn, activates the full adipogenic program of gene expression [50][51][52][53]. Expression of PPARγ is also induced via a sterol-regulating elementbinding protein-1c-(SREBP-1c-) dependent pathway [54]. Once both PPARγ2 and C/EBPα are activated, "cross-talk" between PPARγ and C/EBPα maintains the expression of each protein during adipocyte differentiation, even in the absence of C/EBPβ and C/EBPδ [55]. We generated JDP2 "knock-out" (KO) mice in order to study the activities of JDP2 in vivo, and we found that JDP2 plays a role as a repressor of adipocyte differentiation. We also found that JDP2 targeted an adipogenesis-related gene, C/EBPδ, and inhibited its expression via regulation of histone acetylation. The C/EBPδ and C/EBPα mRNAs were overexpressed in Jdp2 −/− MEFs during the initial stages of adipocyte differentiation, whereas C/EBPβ and PPARγ mRNAs were less affected by the absence of JDP2 [56]. C/EBPα is known to be the downstream target of C/EBPδ and C/EBPβ [57]; therefore, it can be speculated that the augmentation of expression of the C/EBP gene might have been due to indirect effects, which were probably caused by the enhanced expression of the C/EBPδ gene [33]. In fact, the recruitment of JDP2 to the C/EBPδ and C/EBPβ genes was detected, but there was no obvious recruitment to the C/EBPα gene. Thus, the target of JDP2 is possibly the C/EBPδ and C/EBPβ genes.
JDP2 was able to bind to the C/EBPδ promoter to repress the transcription of the C/EBPδ gene. Our JDP2-deficient mice did not have abundant adipose cells, in other words, they did not have a thick layer of fat tissue, perhaps because the mass of adipose tissue might be determined by more complex factors, such as cytokines and hormones, via as yet unknown mechanisms, and by energy status. We found that the adipose tissue of scapulae from young Jdp2 −/− mice consisted mostly of white adipocytes, whereas the majority of cells were brown adipocytes in WT mice. This observation suggested that JDP2 might play a role in adipogenesis in vivo.
Recently, it has been reported that the activation of Wnt signalling inhibited the development of brown adipocytes by mediating the peroxisome proliferation-activated receptor gamma-1 coactivator-1-1α-(PGC1-1α)-) uncoupling pro-tein1 (UCP1) cascade [48]. In addition to C/EBPs as shown in the present studies, JDP2 might be a target of such signal networks. Other report suggested that the absence of Rb switched the differentiation of mouse embryonic fibroblast (MEF) cells from white to brown adipocytes [58], suggesting that the differentiation of WT MEFs to adipocytes is a good model for white-adipocyte differentiation but not for brown-adipocyte differentiation. We also observed that PGC-1α marker of brown adipocyte was not induced after induction of adipocyte differentiation in Jdp2 −/− MEFs and WT MEFs (unpublished). Then, we analyzed the levels of mRNA of PGC-1α in Adeno-JDP2-infected 3T3-L1 cells. 3T3-L1 cells are also known to be a model for white-adipocyte differentiation, however, we could detect small amount of PGC-1α RNA by real-time RT-PCR as the others reported elsewhere [59] probably because quite minor population of 3T3-L1 cells can differentiate to brown adipocyte. The expression of PGC-1α was increased after adipocyte differentiation, and the levels of expression were not remarkably different between Adeno-JDP2-and Adeno-lacZ -infected 3T3-L1 cells. So, we speculate that JDP2 might inhibit the differentiation of white adipocytes exclusively, but not brown adipocytes. In consequence, brown adipocytes are preferentially developed in WT mice scapulae, whereas white adipocytes appear in Jdp2 −/− mice. However, further studies are required to clarify the role of JDP2 in the molecular mechanism that directs fibroblast and mesenchymal cells to differentiate into white or brown adipocytes.

JDP2 Controls Replicative Senescence
We analyzed the aging-dependent proliferation of MEFs from Jdp2 −/− ice in the presence of environmental (20%) or low (3%) oxygen [40]. The Jdp2 −/− MEFs continued to divide, even after six weeks, whereas the wild-type MEFs almost stopped proliferating and entered senescence under environmental oxygen. Conversely, neither the wild-type MEFs nor the Jdp2 −/− MEFs succumbed to replicative senescence at lower oxidative stress. These results demonstrate that MEFs lacking JDP2 can escape from the irreversible growth arrest caused by environmental oxygen. The expressions of p16 Ink4a  Arf compared with those in 20% oxygen, whereas Jdp2 −/− MEFs maintained low-level expression of p16 Ink4a and Arf ( Figure 3). These observations indicate that the agingassociated expression of p16 Ink4a and Arf is dependent on oxygen stress and that JDP2 controls the expression of both p16 Ink4a and Arf. We found no dramatic downregulation of the upstream repressors of p16 Ink4a /Arf, B lymphocyte molony murine leukemia virus insertion region 1 homology (Bmi1), and enhancer of zesta homolog 2 (Ezh2), in the absence of JDP2, suggesting that JDP2 does not regulate their expression. Interestingly, JDP2 expression in wild-type MEFs increased in the presence of 20% oxygen, but not in the presence of 3% oxygen, suggesting that its expression depends on oxygenic stress and that accumulated JDP2 may play a role in the transcriptional activation of p16 Ink4a /Arf (Figure 2). Studies based on ChIP have demonstrated that the methylation of H3K27 at the p16 Ink4a /Arf locus was higher in Jdp2 −/− MEFs than in wild-type MEFs and that the binding of polycomb repressor complex 1(PRC1) and polycomb repressor complex 2 (PRC2) to the p16 Ink4a and Arf promoters was more efficient in Jdp2 −/− MEFs than in wild-type MEFs. These observations suggest that, in the absence of JDP2, H3K27 is methylated by PRC2, and the p16 Ink4a /Arf locus is silenced by PRC1, whereas the increased expression of JDP2 helps to release PRC1 and PRC2 from the p16 Ink4a /Arf locus, thereby reducing H3K27 methylation. Our data demonstrate that JDP2 is an important factor regulating cellular senescence. The loss of JDP2 allows MEFs to escape senescence, and conversely, the overexpression of JDP2 induces cell-cycle arrest. The absence of JDP2 reduces the expression of both p16 Ink4a and Arf, which inhibit cell-cycle progression.

p16
Ink4a and the Rb Pathway. Genes that are essential for cell-cycle progression are transcribed at the beginning of G1 phase by transcription factors of the E2F family. E2F is controlled by Rb family of proteins, pRb, p107, and p130 [60,61]. Early in G1, unphosphorylated Rb proteins bind to the E2F family of proteins and inactivate their function [62,63]. During G1, the Rb proteins are inactivated by phosphorylation by Cdk4/6-cyclinD complexes, thereby allowing the transcription of E2F-dependent gene, including cyclin E. Upregulated cyclin E forms a complex with cdk2, that mediates the hyperphosphorylation of the Rb proteins, an essential requirement for the G1/S transition. p16 Ink4a is an allosteric inhibitor of cdk4/6. Binding to p16 Ink4a changes the conformation of cdk4/6, which prevents its interaction with cyclin D [64,65]. Therefore, p16 Ink4a acts as an inhibitor of the cell cycle at G1 by modulating the Rb pathway. p16 Ink4a is often lost in a variety of human malignancies, including glioblastoma, melanoma, and pancreatic adenocarcinoma [66]. In contrast, the upregulation of p16 Ink4a induces the cell-cycle arrest and senescence [65,66].

Arf and the p53
Pathway. p53 is known to mediate cell-cycle arrest, in G1 and G2, and apoptosis. A number of downstream targets of p53 are involved in these processes, including p21 Cip/wa f 1 in G1 arrest [67], 14-3-3 sigma and growth arrest and DNA damage inducible gene 45 (GADD45) in G2 arrest [68,69], and p21, Bax, PI-3 upregulated modulator of apoptosis (PUMA), Fas/Apo1, and Killer/DR5 in apoptosis [70][71][72][73][74]. p53 is regulated at the levels of protein stability and activity, and to some extent at transcription and translation [75,76]. In unstressed cells, p53 protein levels are very low because its degradation is mediated by the E3 ubiquitin ligase activity of murine double minute 2 (MDM2), which targets p53 for ubiquitindependent proteolysis [77]. MDM2 is a transcriptional target of p53, so p53 directly activates the expression of its own negative regulator, producing a potent negative feedback regulatory loop [78]. There are several stressresponsive kinases, which, by phosphorylating p53, inhibit its degradation by MDM2 and increase its transcriptional activity [79][80][81]. DNA damage rapidly activates the ataxia telangiectasia mutated (ATM) and ataxia elangiectasiarelated (ATR) proteins, which phosphorylate the checkpoint kinases 1 and 2 (Chk1 and Chk2), which in turn propagate the signal to downstream effectors such as p53 [82,83]. Both Chk1 and Chk2 phosphorylate p53 at Ser 20, which prevents the efficient recruitment of MDM2. Thus, p53 is stabilized and its expression level is increased in response to stress signaling. Arf is predominantly localized in the nucleoli and is stabilized by binding to nucleophosmin. In response to stress signaling, Arf is released from nucleophosmin and translocates to the nucleoplasm, where it interacts with MDM2, inhibits its E3 ubiquitin ligase activity, and blocks the nucleocytoplasmic shuttling of the MDM2-p53 complex.
Therefore, the consequences of the activation of Arf are the stabilization and activation of p53 [84,85].

Senescence and Aging in Human and Mouse
. Cellular senescence appears to be related to organismal aging. Cellular senescence involves processes that include telomere shortening, the accumulation of DNA damage, and the activation of the p16 Ink4a /Arf locus. The contributions of these factors to senescence seem to differ in humans and mice. Cultured mouse fibroblasts undergo senescence even when they have long telomeres and high telomerase activity. Senescence is abrogated by the loss of the p16 Ink4a /Arf locus [86]. In human cell cultures, the ectopic expression of telomerase is sufficient to overcome senescence by maintaining the length of the telomeres [87]. In mice, the maintenance of telomere length is important because telomerase deficiency shortens their lifespan and leads to premature aging [88][89][90][91][92]. The agedependent accumulation of INK4A has been observed in the human kidney and skin [91], as well as in the majority of mouse tissues [92]. In oncogene-induced senescence, there is in vivo evidence that Arf is the important factor in the activation of p53 tumor suppression [91]. However, another study has shown that components of the DNA-damagesignaling cascade, including ATM and Chk2, are critical for the activation of p53 in response to oncogenic signals [93]. These differences between humans and mice could be attributable to species specificity and/or experimental conditions. Cellular senescence appears to be related to organismal aging because the same processes appear to be involved. Genetic variants of the p16 Ink4a /Arf locus are linked to age-associated disorders, such as general frailty, heart failure, and type 2 diabetes [94][95][96][97][98][99]. Mutations in telomerase or in proteins that affect telomerase activity are linked to premature human aging syndromes, including congenital dyskeratosis and aplastic anemia [100]. There are increases in DNA mutations, DNA oxidation, and chromosome loss during organismal aging. It seems reasonable to assume that all three factors, the activation of the p16 Ink4a /Arf locus, telomere shortening, and the accumulation of DNA damage, have cooperative effects on aging in physiological situations. Understanding the mechanisms of cellular senescence is currently of wide interest, and it is important that we identify new components of this process, such as JDP2.

Interaction of Transcription Factors with Histones and Nucleosomes
Similar to JDP2, other transcription factors or chromatin regulators are reported to have the binding affinities to histones and/or nucleosome. Transcription factor IIIA (TFIIIA) is a 40-kDa protein, with nine zinc finger domains, that binds specifically to the internal promoter of the gene for 5S RNA and to the N-terminal tail domains of histones H3 and H4, but not those of H2A and/or H2B, and directly modulates the ability of TFIIIA to bind nucleosomal DNA [101,102].
A new class of HAT-regulatory proteins has been identified. These proteins block HAT activity via binding to and masking of the histone themselves. This class includes Journal of Biomedicine and Biotechnology 7 the subunits of INHAT complex; TAF-1α, TAF-1β, and pp32 as well as ataxin 3; silencing mediator or retinoid receptor corepressor (SMART)/nuclear hormone receptor corepressor (NcoR); proline-glutamic acid-and leucine-rich protein 1 (PELP1) [103][104][105][106][107]. Thanatos-associated protein 7 (THAP7) is known to associate with TAF-1β and to repress transcription by inhibition histone acetylation [108,109]. A novel INHAT repressor (NIR) binds directly to nucleosomes and core histones and prevents acetylation by histone acetyltransferases, thus acting as a bona fide INHAT [110]. PU.1, a member of the Ets family of oncoproteins, inhibits CBP-mediated acetylation of globin transcription factor-1 (GATA-1) and erythroid Krüppel-like factor (EKLF), as well as of histones, and disrupts acetylation-dependent transcriptional events [111]. Moreover, PU.1 recruits the retinoblastoma tumor-suppressor protein, a histone methyltransferase (HMT) Suv39H, and heterochromatin protein 1α (Hp1α) [112]. Similarly, proliferating cell nuclear antigen binds to p300 to inhibit its HAT activity in vitro and to block HAT-dependent transcription in vivo [113]. TAF-1β and pp32 have also been shown to interact with estrogen receptor (ERα), to inhibit the ERα-mediated activation of transcription, and to inhibit the activation of ERα acetylation by p300 [106,107]. Moreover, TAF-1β also interacts with the Sp-1 transcription factor and with KLF5, negatively regulating the binding to DNA and activation of transcription by these factors [114,115]. Transcription factor NF-kB p50 can accommodate distorted, bent DNA within the nucleosome [116], while DNA-binding domain (DBD) of c-Myb binds to the N-terminal tails of H3 and H3.3, and binding of c-Myb facilitates acetylation of histone tails [117]. Moreover, a kinetochore-null protein (KNL-2) with a c-Myblike DNA-binding domain is specifically required for loading of centromere-specific variants of histone H3 (centromere protein A; CENP-A) in both nematodes and mammalian cells [118]. The recruitment of repressive macroH2A nucleosomes requires direct interactions between ATF-2 bound to a nearby AP-1 site and macroH2A, and recruitment is regulated by DNA-induced protein allostery [119]. Thus, the abovementioned sequence-specific DNA-binding factors might regulate transcription either via histone cores or tails, as well as via the structure of nucleosomes in conjunction with other proteins that bind to the chromatin. Indeed JDP2 is the first case to have both nuclesome assembly activity and DNA-binding activity [32] and inhibits the histone acetylation by p300/CBP [32,33] as well as histone methylation [40] as described below.

Mechanism of JDP2-Mediated Regulation of C/EBPδ and p16 Ink4a /p19
Differentiation and senescence are associated with dynamic changes in gene expression, which are regulated by chromatin remodeling. Here, we have shown that the expression of JDP2 regulated the differentiation competent genes and the senescence competent genes such as C/EBPδ and p16 Ink4a and Arf in response to the induced hormone and accumulating oxidative stresses. In the case of suppression of adipocyte differentiation by JDP2, how might JDP2 inhibit transcription of the C/EBPδ gene? It seems possible that, in response to signals that lead to differentiation, the C/EBPδ promoter might recruit transcription factors, including the coactivator p300, that mediate the acetylation of histones associated with the C/EBPδ gene to stimulate transcription of this gene. In the presence of JDP2, the acetylation of histone is inhibited and the expression of the C/EBPδ gene is suppressed during adipocyte differentiation, even in the presence of differentiation-inducing signals. Decreased expression of C/EBPδ results in less effective differentiation to adipocytes, a conclusion that is consistent with a previous report that deletion of the C/EBPδ gene results in the impairment of adipocyte differentiation [59]. By contrast, HDAC3 was not recruited by JDP2 to the promoter region of the C/EBPβ gene, although the recruitment of HDAC3 to the C/EBPβ gene was slightly enhanced during differentiation. In addition, the acetylation of histone H3 on the C/EBPβ gene was less affected by differentiation-inducing stimuli. These observations suggest that histone acetylation might not play an important role in the transcriptional regulation of the C/EBPβ, gene and the HDAC3 might not be involved in such regulation. This hypothesis explains why the presence of JDP2 on the promoter had no effect on transcription.
In conclusion, we propose that JDP2 acts as a negative molecular-switch in some, but not all, types of differentiation via the regulation of the expression of specific genes in concert with histone deacetylase HDAC3, as shown in the present study of adipocyte differentiation.
In the case of induction of cellular senescence by JDP2, we propose a model; the accumulation of oxidative stress and/or other environmental stimuli during aging upregulate JDP2 expression in primary untransformed cells. Increased JDP2 helps to remove PRC1 and PRC2, which are responsible for the methylation of histone H3, from the p16 Ink4a /Arf locus, leading to increased p16 Ink4a and Arf expression and entry into the senescence (Figure 4). There is some evidence that JDP2 acts as a tumor suppressor; JDP2 inhibits the Rasdependent transformation of NIH3T3 cells [29], and JDP2 gene disruptions are often found in the lymphomas induced by insertional mutagenesis caused by the Moloney murine leukemia virus in MYC/Runx2 transgenic mice [27]. Here, we suggest that JDP2 not only inhibits the transformation of cells but also plays a role in the induction of cell senescence. Both functions of JDP2 might be important for its role in inhibiting tumor formation. Our findings also provide new insights into the molecular mechanisms, by which senescence is induced in the context of the epigenetic regulation of the p16 Ink4a /Arf locus. Recently, we reported that JDP2 regulates the expressions of cell-cycle regulators such as cyclin A2, cyclin E2, and p16 Ink4a and also affects the expression of p53 and p21 protein [120]. Thus, it is evident that JDP2 control the expression of cell-cycle regulators to induce the cell-cycle arrest via p53-p21 pathway and RB-p16 pathway.

Concluding Remarks
We characterized a sequence-specific DNA-binding protein as a nucleosome-assembly factor. Our finding should facilitate efforts to understand some aspects of nuclesome assembly and the remodeling of chromatin. Histone H3 seems to be recruited to target sites in a DNA synthesisdependent manner via the interaction of chromatin assembly factor 1 (CAF-1) with proliferation cell nuclear antigen (PCNA); by contrast, a histone chaperone HIRA was reported to incorporate H3.3 in a DNA synthesisindependent manner [15]. The observation that histone H3.3 is found in several other subcomplex suggests that H3.3 might be recruited to different sites by different pathways. Identification of a gene-specific DNA-binding protein, namely JDP2, as a nucleosome-assembly factor suggests that H3.3 could be directly deposited at specific locations by site-specific DNA-binding proteins that also have histone-chaperone activity. It will be of interest to examine the binding preference of JDP2 for histone H3 as compared with H3.3.
Since JDP2 binds to core histone or nuclesome partially in DNA sequence-specific manner or histone subset-specific manner, the histone acetyltransferase or histone methytransferase may not access to the nucleosome in vitro (Figure 4). However, we are not sure the case of in vivo. Chromatin immunoprecipitation assay (ChIP assay) demonstrated that JDP2 inhibited at least the acetylation of histone H4K8 and H4K16 although we cannot determine other precise residues of histone H3 acetylation (31). Moreover, JDP2 associated with histone H3K27 and blocked the methylation of histones (40). Thus, we assume that the interaction of JDP2 with nucleosome is DNA sequence or histone modification specific, and thus, only certain restricted set of histone might be associated with JDP2 in vivo. Addressing these precise functions in the context of epigenesis helps us to understand how senescence and differentiation, in a broader context, are regulated.

Future Prospects
Understanding how JDP2 promotes aging or cell differentiation, whether by inducing cellular senescence or decreasing the frequency of cell-cycle entry, is an important issue. The data demonstrating the increase in expression of p16 Ink4a or C/EBPδ with aging (or oxygen stress) or hormone induction can be reconciled with two different models like "commitment model" and "threshold model." With aging or hormone induction, stochastic activation of p16 Ink4a or C/EBPδ expression could occur on a cell-by-cell basis in selfrenewing compartments to induce commitment (commit- Self-renewal Age Oxygen p16 Ink4a hormone C/EBP Age Oxygen p16 Ink4a hormone C/EBP Figure 5: Model of role of JDP2 in cellular aging and cell differentiation. Self-renewing cells undergo repeated divisions, and p16 Ink4a or C/EBPδ expression increases with age, oxygen, and hormone, as a consequence of undefined stimuli. (a) Commitment model: p16 Ink4a or C/EBPδ expression occurs stochastically in a subpopulation of cells resulting in their senescence. In this model, the capacity for selfrenewal of the noncommitted cells is not affected. (b) Threshold model: expression of p16 Ink4a or C/EBPδ increases or decreases, respectively, uniformly in the tissue specific. Self-renewing cells compartment is compromised over time. Self-renewal is indicated by curved arrows; committed cells are colored blue. ment model), or expression could increase simultaneously with the majority of cells of a self-renewing compartment (threshold model) ( Figure 5). In the latter case, self-renewal would be impaired by p16 Ink4a or C/EBPδ expression by decreased frequency of cell-cycle entry in the absence of senescence or cell differentiation.
Addressing this question has important implications for future "antiaging" therapies or "antidifferentiation" therapies. The model in which cell-cycle entry is decreased (threshold model) suggests that the age-induced defects in proliferation or the hormone-induced defects in proliferation could be improved merely by reducing p16 Ink4a levels or C/EBPδ levels or otherwise increasing cdk4/6 activity in these cells. The commitment model, however, suggests that the defects in self-renewal could only be remedied through more drastic self-renewing cells from an exogenous source. The understanding of the in vivo regulation of the Ink4a/Arf/Ink4b locus with aging (oxidative stress) or C/EBPδ locus with differentiation is needed further.