NIRF/UHRF2 occupies a central position in the cell cycle ... · Review NIRF/UHRF2 occupies a...

FEBS Letters 586 (2012) 1570–1583

journal homepage: www.FEBSLetters .org

Review

NIRF/UHRF2 occupies a central position in the cell cycle network and allowscoupling with the epigenetic landscape

Tsutomu Mori a,1,⇑, Daisuke D. Ikeda b,1, Yoshiki Yamaguchi c,1, Motoko Unoki d,1, NIRF Project 1

a Department of Human Lifesciences, Fukushima Medical University School of Nursing, Fukushima, Japanb Kitajima-cho, Itano-gun, Tokushima, Japanc RIKEN, Advanced Science Institute, Chemical Biology Department, Systems Glycobiology Research Group, Structural Glycobiology Team, Wako, Saitama, Japand Division of Epigenomics, Department of Molecular Genetics, Medical Institute of Bioregulation, Kyushu University, Fukuoka, Japan

a r t i c l e i n f o a b s t r a c t

Open access under CC BY-NC-ND license.

Article history:Received 21 March 2012Revised 18 April 2012Accepted 18 April 2012Available online 30 April 2012

Edited by Wilhelm Just

Keywords:NIRFUHRF2Tumor suppressorCell cycleSystems biologyUbiquitin-proteasome systemEpigeneticsCancerPolyglutamine diseaseCyclin D1Cyclin E

0014-5793 � 2012 Federation of European Biochemichttp://dx.doi.org/10.1016/j.febslet.2012.04.038

Abbreviations: CAN gene, candidate cancer gene;CDK, cyclin-dependent kinase; COPA, cancer outliermethyltransferase; GRN, gene regulatory network;H3K9, Histone H3 lysine 9; ICBP90, inverted CCAAT bNIRF, Np95/ICBP90-like RING finger protein; NSCLC, nNp95, 95-kDa nuclear protein; PCNP, PEST proteolyprotein; PDB, protein data bank; PHD, plant homeodinteraction network; PQC, protein quality control;gene; RNAi, RNA interference; SCF, Skp1/Cul1/F-box ppolymorphism; SRA, SET and RING associated; TTD,ubiquitin-like; UHRF, ubiquitin-like with PHD and rin⇑ Corresponding author.

E-mail address: [email protected] (T. Mori).1 A Multi-Affiliation Research Group for NIRF (http:/

As predicted by systems biology, a paradigm shift will emerge through the integration of informa-tion about different layers of cellular processes. The cell cycle network is at the heart of the cellularcomputing system, and orchestrates versatile cellular functions. The NIRF/UHRF2 ubiquitin ligase isan ‘‘intermodular hub’’ that occupies a central position in the network, and facilitates coordinationamong the cell cycle machinery, the ubiquitin–proteasome system, and the epigenetic system. NIRFinteracts with cyclins, CDKs, p53, pRB, PCNA, HDAC1, DNMTs, G9a, methylated histone H3 lysine 9,and methylated DNA. NIRF ubiquitinates cyclins D1 and E1, and induces G1 arrest. The NIRF geneis frequently lost in tumors and is a candidate tumor suppressor, while its paralog, the UHRF1 gene,is hardly altered. Thus, investigations of NIRF are essential to understand the entire biological sys-tems. Through integration of the enormous information flows, NIRF may contribute to the couplingbetween the cell cycle network and the epigenetic landscape. We propose the new paradigm thatNIRF produces the extreme diversity in the network wiring that helps the diversity of Waddington’scanals.

� 2012 Federation of European Biochemical Societies. Published by Elsevier B.V.

1. Introduction

An organism is a highly complicated ‘‘system’’ that is con-structed through interactions of a large number of distinct compo-nent subsystems [1]. Systems biology predicts that the propertiesof a whole biological network are not a simple extension of its

al Societies. Published by Elsevier

CD, cluster of differentiation;profile analysis; DNMT, DNAHDAC, histone deacetylase;

ox binding protein of 90 kDa;on-small cell lung carcinoma;tic signal-containing nuclearomain; PPIN, protein–proteinRING, really interesting newrotein; SNP, single nucleotidetandem Tudor domain; UBL,g finger domains

/www.nirf.jp).

parts, but instead exhibit a new ‘‘emergent’’ state that is totallydifferent from those of the individual subnetwork modules [2].The pursuit of these novel properties will provide new objectives,new methodologies, and new insights, which will result in aparadigm shift in biomedical sciences.

The Np95/ICBP90-like RING finger protein (NIRF) ubiquitin li-gase, which is also called ubiquitin-like with PHD and ring fingerdomains 2 (UHRF2), occupies a central position in the cell cyclenetwork, and has the capacity to ubiquitinate both cyclins D1and E1 [3]. It is also an ‘‘intermodular hub’’ protein that connectsthe cell cycle machinery, the ubiquitin–proteasome system, andthe epigenetic system, and can thus be expected to provide the ba-sis for a new paradigm in the next few years. In sharp contrast toits paralog ubiquitin-like with PHD and ring finger domains 1(UHRF1), the NIRF gene is frequently lost in tumors [3], and will at-tract particular attention with respect to its relevance to tumorpathophysiology.

Here, we summarize the current knowledge about the biologi-cal properties of NIRF, with a focus on its linkage with tumors,

B.V. Open access under CC BY-NC-ND license.

http://dx.doi.org/10.1016/j.febslet.2012.04.038

mailto:[email protected]

http://www.nirf.jp

http://dx.doi.org/10.1016/j.febslet.2012.04.038

http://www.FEBSLetters.org

http://creativecommons.org/licenses/by-nc-nd/3.0/

http://creativecommons.org/licenses/by-nc-nd/3.0/

T. Mori et al. / FEBS Letters 586 (2012) 1570–1583 1571

and discuss future prospects for research on the NIRF genenetwork.

2. Overview

The cell cycle network, a signaling network that governs cell cy-cle progression, is positioned at the heart of the cellular informa-tion processing system.

Into this network, a multitude of information such as mitogenicstimuli, survival signals, stress conditions, epigenetic information,and nutrient statuses converge from the outside and inside of cells[4]. The network is a highly sophisticated system comprising largenumbers of network modules that integrates the flood of informa-tion and makes appropriate decisions, thereby orchestrating manyfunctional modules that direct cell proliferation, differentiation,stress management, senescence, and survival [4]. Defects in theproper coordination among many network modules cause variouscellular dysfunctions, such as cell cycle deregulation and genomicinstability, ultimately leading to the development of diseases,including cancer. However, the network architecture and molecu-lar basis that ensure the correct functioning of the network as asophisticated computing system are largely unknown.

Recent systems biology analyses have revealed the generalproperties of the information networks underpinning complex bio-logical systems. While most proteins have only a few interactingpartners, a small number of ‘‘hub’’ proteins have a very large num-ber of partners [5]. This ‘‘scale-free’’ property indicates the central-ity and importance of the hub proteins. In fact, many hub proteinsare essential for the viability of organisms [6]. Each hub proteinserves as a functional and structural core of an individual module,which comprises a group of proteins related to one another in theirinteraction partners, functions, or expression profiles. The wholebiological network is not constructed as a random network, butas an ordered assembly of network modules. Of note, there is ahierarchical order among the different modules [5], which alsomeans a hierarchy among the hub proteins. Furthermore, the hubproteins are classified into intramodular hubs (party hubs) andintermodular hubs (date hubs) [7,8]. Intramodular hubs interactwith their partners simultaneously, and act inside functional mod-ules. In contrast, intermodular hubs bind their different partnersasynchronously, and organize the proteome by connecting the net-work modules to one another, thereby contributing to the mainte-nance of the integrity and functions of the whole network [7]. As aresult, mutations of intermodular hubs lead to alterations in themodularity of the whole network, causing serious defects in cellu-lar functions. Importantly, intermodular hubs are frequently asso-ciated with tumorigenesis [9]. Therefore, one of the main streamsof current systems biology is to identify the intermodular hubswith the highest hierarchy level, and to reveal their relevance todiseases (particularly cancer), followed by assessment of theiravailability as drug targets.

Information processing by the cell cycle network is chiefly car-ried out within the G1 period of the cell cycle [4]. Through the inte-gration of many signals, the network computes an appropriatecellular decision for whether to progress through the G1/S bound-ary, or progress toward alternative cell fates, e.g., differentiation,death, or senescence. According to the above considerations in sys-tems biology, we can expect the existence of an intermodular hubthat is involved in the multifaceted cellular decision in the G1phase. Importantly, defects in G1 control are essential for tumori-genesis [10]. Therefore, the putative intermodular hub involved inG1 control would be intimately linked to tumorigenesis.

The ubiquitin–proteasome system and the epigenetic systemare two representative network modules that participate in thecellular decision through the regulation of large numbers ofgenes by influencing their transcription, translation, and post-

translational modifications [11,12]. Thus, it is possible that theaforementioned intermodular hub may be related to these sys-tems. The discovery of such a presumptive intermodular hub willbring about a great paradigm shift in biomedical sciences.

Modern molecular biology has established the intimate rela-tionships between the cell cycle network and serious human dis-eases, especially cancer. However, there have been fewindications of the network architecture and molecular propertiesunderlying the higher computation functions of the network. Forthe above reasons, we attempted to identify a new cell cycle gene,with consideration of the probability of the existence of an un-known multifunctional gene involved in the cellular decisions to-ward various directions.

3. Molecular characterization of NIRF

3.1. Cloning of NIRF cDNA

The ubiquitin–proteasome system plays key roles in virtually allcellular activities by tightly regulating the degradation of a largeset of proteins [11]. This system cooperates with the cell cyclemachinery as a major network module, thereby contributing to cellcycle regulation and multiple cellular decision processes. Consider-ing the possibility that the putative determination factor may berelated to the ubiquitin–proteasome system, we searched for a nu-clear protein exhibiting a physical interaction with the ubiquitina-tion-prone protein, PEST proteolytic signal-containing nuclearprotein (PCNP), which was newly identified at that time [13].

PCNP is an abundant nuclear protein containing two remark-able PEST sequences, which are rich in proline (P), glutamic acid(E), serine (S), and threonine (T). PEST sequences generally confersusceptibility to ubiquitin-mediated proteolysis [14]. Many PESTproteins in the nucleus (e.g., c-Myc, p53, cyclin D1, and cyclinE1) play important regulatory roles in various processes includingcell cycle control. We observed that knockdown of PCNP by RNAinterference (RNAi) treatment significantly reduced cell prolifera-tion (unpublished data), while the PCNP mRNA level was reducedupon cell differentiation [15].

We performed a yeast two-hybrid screening to search for pro-teins interacting with PCNP. We found NIRF, on the basis of itscomplex formation with PCNP [16]. We further revealed that NIRFis a ubiquitin ligase and ubiquitinates PCNP [13].

3.2. Structural characteristics of NIRF

NIRF is a nuclear protein with 802 amino acid residues. NIRF is amultidomain protein that shows high structural similarity toUHRF1, which is also called Np95 and ICBP90 (Fig. 1A). NIRF andUHRF1 share at least five functional domains, namely the ubiqui-tin-like (UBL) domain (also called NIRF_N domain) at the N-termi-nus, tandem Tudor domain (TTD), plant homeodomain (PHD)finger domain, SET and RING associated (SRA) domain, and reallyinteresting new gene (RING) finger domain at the C-terminus. Re-cent X-ray crystallographic and NMR analyses have revealed thestructures of these domains and their recognition modes for themodification statuses of DNA and histone H3 lysine residues.

3.2.1. NIRF_N domain (UBL domain)Although many UBL domain-containing proteins interact with

the 26S proteasome, which is responsible for ubiquitin-mediatedintracellular proteolysis, it has not been structurally clarified howthe UBL domains in NIRF and UHRF1 are involved in these interac-tions. Ubiquitin forms a five-stranded b-sheet with a single helix. Itis assumed that most UBL domains, including those in NIRF andUHRF1, could have a similar ubiquitin fold [17] (Fig. 1B).

Fig. 1. NIRF and UHRF1 are multidomain proteins providing diverse recognition platforms. (A) Domain organizations of NIRF (UHRF2) and UHRF1, including a ubiquitin-like(UBL) domain, tandem Tudor domain (TTD), plant homeodomain (PHD) finger domain, SET and RING associated (SRA) domain, and really interesting new gene (RING) fingerdomain. The domain structures were created using DOG1.0 software [119]. (B) Solution structure of the human NIRF UBL domain (PDB code: 1WY8) and crystal structure ofthe human UHRF1 UBL domain (PDB code 2FAZ). (C) Solution structure of the human TTD in complex with histone H3 trimethylated on lysine 9 (H3K9me3) (PDB ID: 2L3R)[19]. (D) Solution structure of the human NIRF PHD finger domain (PDB ID: 2E6S) and crystal structure of the human UHRF1 PHD finger domain in complex with unmodifiedhistone H3 peptide (H3R2) (PDB ID: 3SOU) [22]. (E) Crystal structures of the human NIRF SRA domain (PDB ID: 3OLN) and of the human UHRF1 SRA domain in complex withhemimethylated DNA (PDB ID: 3CLZ) [25]. (F) Crystal structures of the human NIRF RING finger domain (PDB code 1Z6U) and the human UHRF1 RING finger domain (PDB ID:3FL2). In all cases, the proteins and bound ligands are shown in ribbon and stick models, respectively, and Zn ions are shown in gray spheres. The figure was prepared usingPymol software (http://www.pymol.org/).

1572 T. Mori et al. / FEBS Letters 586 (2012) 1570–1583

3.2.2. Tandem Tudor domain (TTD)NIRF contains a TTD. Although the TTD itself does not possess

high binding affinity for histone H3 tail peptides harboring di-and tri-methylated lysine 9 (H3K9me2/me3), the combined TTD–PHD region binds to H3K9me2/me3-containing peptides with highaffinity [18]. Recent solution NMR and crystallographic studieshave revealed that the two Tudor domains in UHRF1 are tightlypacked to each other [19]. Each Tudor domain has a five-strandedb-barrel fold seen in various histone-binding proteins, such as p53

binding protein 1 (53BP1), which binds to histone H4K20me2 [20].In the TTD–histone peptide complex, the first Tudor domain recog-nizes histone H3K9me3 through a conserved aromatic cage(Fig. 1C). No structural information for the NIRF TTD is currentlyavailable.

3.2.3. PHD finger domainA PHD finger is a small domain that is often found in nuclear

proteins involved in chromatin-mediated gene regulation. Many

http://www.pymol.org/


PHD finger-containing proteins have been reported to recognizehistone tails. Although the PHD finger of NIRF does not possessbinding affinity toward histone H3 tail peptides, the combinedTTD–PHD region of NIRF has higher binding affinity to H3K9me2/me3 than the single TTD as described above [18]. This cooperationis attributed to the highly conserved linker region that connectsthe TTD and PHD finger. In contrast, the same authors showed thatthe combined TTD–PHD region of UHRF1 does not exhibit en-hanced binding to the histone H3 peptides [18]. However, Xie etal. [21] showed that the combined TTD–PHD domains of UHRF1act together to recognize H3K9me3. Therefore, there is still roomfor argument regarding the function of the TTD–PHD region. ThePHD finger of UHRF1 is known to interact with unmodified histoneH3R2 (Fig. 1D). The binding mode was revealed from the crystalstructure of the UHRF1 PHD finger in a complex with the histoneH3 peptide [22]. A guanidinium group of the unmodified histoneH3R2 residue forms extensive intermolecular contacts with thenegatively charged surface groove on the PHD finger.

3.2.4. SRA domainAlthough it remains controversial, the NIRF SRA domain may be

involved in interactions with hemimethylated DNA [23]. The crys-tal structure of the UHRF1 SRA domain was reported in a complexwith double-stranded DNA containing C5-methylated cytosine inone strand and unmethylated cytosine in the other strand [24–26]. The SRA domain contains two twisted b-sheets packed to-gether, and binds to DNA using the basic concave surface. The mostcharacteristic feature of the UHRF1 SRA domain is recognition of aflipped methylated cytosine (Fig. 1E). In the crystal structure, themethylated cytosine is completely flipped out from the DNA dou-ble helix, and the methyl group is packed against a small cavitywithin a conserved pocket in the SRA domain. Base flipping is aconserved mechanism that is often used by DNA/RNA-modifyingenzymes. The SRA domain is the first identified non-enzymatic se-quence-specific DNA-binding protein that uses base flipping in itsinteraction with DNA. The detailed ligand-binding mode of theNIRF SRA domain has not been reported, although the crystal struc-ture of the ligand-free SRA domain is available. According to a re-port, NIRF moderately binds to fully methylated DNA in addition toits strong binding to hemimethylated DNA, similar to the case forUHRF1 [23].

3.2.5. RING finger domainA RING finger domain is a small zinc-binding domain that often

possesses ubiquitin ligase activity. The RING finger domain con-tains two zinc ions in a unique ‘‘cross-brace’’ arrangement througha defined motif composed of cysteine and histidine residues(Fig. 1F). This arrangement endows the RING finger domain witha globular conformation, characterized by a central a-helix andvariable-length loops separated by several small b-strands [27].Iwata et al. [28] revealed that NIRF ubiquitinates nuclear polyglu-tamine aggregates via its RING finger domain. The exact amino acidresidues in this domain of NIRF that determine the substrate spec-ificity have not been identified.

3.3. Subcellular localization and expression profiles of NIRF

NIRF is enriched at pericentromeric heterochromatin indepen-dently of the cell cycle, as described later in Section 5.3 [23]. Thissubcellular localization depends on an intact TTD and histoneH3K9me3.

NIRF is ubiquitously expressed in almost all tissues (http://biog-ps.org; [29]) (Fig. 2, left). NIRF is upregulated during the course ofdifferentiation [18], and expressed in terminally differentiated tis-sues including the brain, lung, and kidney (unpublished data;

[18,29]). These expression profiles are in sharp contrast to thoseof UHRF1. UHRF1 is exclusively expressed in proliferating and/orundifferentiated cells, but not in quiescent cells [30–32] (Fig. 2,right), and is downregulated accompanied by differentiation[18,31–33]. The reciprocal expression between NIRF and UHRF1suggests different functional roles of these proteins in differentia-tion [18]. These observations are consistent with the concept thatthey have opposing functions in cell proliferation [30,34].

B- and T-lymphocytes have specific properties in that they aresubject to somatic recombination of antigen receptor genes duringdevelopment. Both types of lymphocytes comprise many differentlineages and subsets of cells with differing sets of active transcrip-tion networks, reflecting their different epigenetic landscapes [35].Interestingly, there is a prominent difference in the expression pro-files of NIRF and UHRF1 in lymphoid cells (Fig. 2). NIRF is highly ex-pressed in lymph nodes, especially in germinal center B cells [36],as well as in CD4+ T cells, CD8+ T cells, and CD4+CD25+ regulatory Tcells [29]. In contrast, UHRF1 is highly expressed in the thymus,but not in lymph nodes [32]. UHRF1 shows very little, if any,expression in CD4+ T cells, CD8+ T cells, and CD4+CD25+ regulatoryT cells [29]. These characteristic expression profiles with large dif-ferences in different lineages may be associated with the epige-netic diversity in lymphocytes. Otherwise, they may have arelationship with the somatic recombination event that is accom-panied by gross rearrangement of the genome and epigenome.

4. NIRF occupies a central position in the cell cycle network

Recently, we found that NIRF interacts with core cell cycle pro-teins including cyclins (A2, B1, D1, and E1), cyclin-dependent ki-nases (CDK2, CDK4, and CDK6), retinoblastoma 1 protein (pRB),tumor protein p53 (p53), and proliferating cell nuclear antigen(PCNA). Since all of these interactors are core components of thecell cycle machinery, NIRF must have a close relationship with cellcycle control. Thus, NIRF is considered to be a novel core compo-nent of the cell cycle machinery [3].

Notably, all of these interactors are themselves hub proteinsthat have many interacting partners of their own, and form indi-vidual network modules (Fig. 3). NIRF is positioned at the centerof these modules, thereby connecting the individual modules toone another. When a network analysis was carried out for this NIRFinteraction network, the calculated multiple centrality indexes forNIRF, i.e., betweenness centrality, closeness centrality, and degreecentrality, were all significantly high [3]. In addition to these find-ings, the expression profiles of the NIRF interactors, namely cyclins,CDKs, pRB, p53, and PCNA, were mutually distinct, indicating thatNIRF is an ‘‘intermodular hub’’ of the cell cycle network with func-tional as well as structural centralities. Thus, NIRF occupies a crit-ical position within the hierarchical modular architecture of thenetwork [3]. Therefore, NIRF is presumed to be involved in multi-faceted cellular processes as described earlier in Section 2.

The cell cycle network plays very important roles in coordinat-ing the activities of diverse subnetwork modules to ensure theproper execution of cell cycle events. In addition to the cell cyclemachinery, the ubiquitin–proteasome system and the epigeneticsystem constitute the two major functional modules of the net-work. In this regard, NIRF has a characteristic multidomain compo-sition that mediates interactions with these systems. Therefore, wepropose that NIRF constitutes a nodal point in the cell cycle net-work that is particularly involved in coordination among the cellcycle machinery, the ubiquitin–proteasome system, and the epige-netic system.

Taken together, these situations reinforce the concept that NIRFis a central factor of the whole network system of the cell. In agree-ment with this notion, it is becoming increasingly clear that defects

http://biogps.org

http://biogps.org

Fig. 2. NIRF and UHRF1 are expressed in different tissues and organs [29]. mRNA expression values for each gene were standardized into Z-scores with barcode approach[120]. A Z-score more of than 5 means that the gene is expressed in that tissue. NIRF is ubiquitously expressed in most differentiated tissues, while UHRF1 is mainly expressedin proliferating and/or undifferentiated tissues. With respect to the lymphatic system, NIRF is highly expressed in lymph nodes as well as different subsets of B and T cells. Incontrast, UHRF1 is not expressed in lymph nodes except in germinal center B cells (centrocytes and centroblasts).


in the NIRF gene are involved in cell cycle deregulation, leading totumorigenesis [3]. Thus, the next major target of this research fieldwill be to investigate how NIRF contributes to the proper couplingbetween the modules around NIRF, and how defects in this coordi-nation function are associated with various human diseases, espe-cially cancer.

5. NIRF mediates interactions between the network modules

In this section, we describe the three network modules withinteractions mediated by NIRF, i.e., the cell cycle machinery, theubiquitin–proteasome system, and the epigenetic system. In addi-tion, we discuss the possible interrelationships between thesemodules in light of the systems-level integration of the networkmodules.

5.1. The cell cycle machinery

The cell cycle machinery is the central module of the cell cyclenetwork. It is composed of multiple combinations of cyclin/CDKcomplexes, where each cyclin is the activation subunit of its cog-nate CDK [37]. Meanwhile, the activity of the cell cycle machineryis under the highest control, since deregulation of the machineryoften leads to the development of cancer [10].

The cell cycle is a complex process whereby DNA replicationand mitotic cell division take place, and periodic oscillations inthe CDK activities are essential for correct progression throughthe cell cycle as well as maintenance of the genome integrity[38,39]. The oscillations in the CDK activities are brought aboutby fluctuations in the cyclin abundances, which are regulated atthe levels of transcription and proteolysis at particular time points

of the cell cycle [39,40]. The different cyclin/CDK complexes phos-phorylate distinct groups of substrates in each cell cycle phase,thereby providing a driving force for the progression of cell cycleevents [41]. The G1 cyclins (D and E) are important in the G1/Stransition, where they determine whether a cell enters anotherround of cell division [37,42,43]. In response to mitogenic stimuli,the amount of cyclin D1, a prototypical cyclin D, is increased andforms complexes with CDK4/6, which then target pRB for phos-phorylation [43]. Upon phosphorylation, pRB loses its ability to in-hibit G1/S progression and releases the transcription factor E2F[44,45]. As an essential protein for cell cycle progression, E2F in-creases the amount of cyclin E, which binds to CDK2 and triggersthe phosphorylation of many substrates, including pRB, requiredfor G1/S progression [46]. Among the mitotic cyclins (A and B), cy-clin A is expressed from S phase to early M phase while cyclin B isexpressed from G2 phase to M phase. Cyclin A forms complexeswith CDK1 and CDK2, while cyclin B forms a complex with CDK1.The cyclin A/CDK2 complex controls DNA replication, while cyclinA/CDK1 and cyclin B/CDK1 complexes cooperate with each otherduring the initiation and completion of mitotic cell division [47].

For appropriate modulation of the cell cycle machinery activity,the cell cycle network performs a complex process of computation.The network integrates diverse inputs (e.g., growth factor stimuli,matrix attachment, nutrient sufficiency, checkpoint status, andepigenetic landscape), and creates correct outputs by affecting,for example, the transcription, phosphorylation, and ubiquitinationof cell cycle proteins including cyclins, CDKs, and CDK inhibitors[4]. This integration process involves two different layers of regu-latory networks that act cooperatively: the protein–protein inter-action network (PPIN) and the gene regulatory network (GRN)[48]. The PPIN depends on various modes of interactions (e.g., sig-naling, regulatory, or biochemical) between the cell cycle proteins.

Fig. 3. The NIRF protein–protein interaction network. NIRF interacts with the core cell cycle proteins (cyclin A2, cyclin B1, cyclin D1, cyclin E1, pRB, and p53) and othernuclear proteins. These core cell cycle proteins are hub proteins with many interacting partners, and thereby constitute individual network modules. NIRF is positioned at thecenter of these modules, thereby connecting the modules to one another. The symbols used are as follows: yellow, cyclins; yellow–green, CDKs; blue, two major tumorsuppressors; green, NIRF-interacting proteins retrieved from a database [121]. The thicker black lines indicate interactions demonstrated by more than two methods.(Reproduced with permission from reference [3]).


On the other hand, the GRN represents interactions of DNA seg-ments with other substances in the cell, thereby regulating thetranscription of the cell cycle genes [49]. NIRF participates in boththe PPIN and GRN through its direct interactions with the cell cycleproteins and by affecting the transcription of the cell cycle genes,respectively [3,50]. This two-sided function of NIRF makes it ahighly attractive molecule that may play a critical role in imposingstrict controls on the cell cycle machinery.

On the one hand, NIRF takes part in protein–protein interac-tions with the core components of the cell cycle machinery, there-by defining a crucial nodal point of the PPIN (Fig. 3) [3]. To ourknowledge, no other ubiquitin ligases have been shown to exhibitsuch wide accessibility to both cell cycle-promoting factors (cyc-lins A, B, D, and E, CDK2, CDK4, CDK6, and PCNA) and cell cycle-inhibitory tumor suppressors (pRB and p53) [3]. NIRF interactswith both the G1 cyclins (D1 and E1) and the mitotic cyclins (A2and B1), suggesting versatile roles for NIRF in a wide range of cellcycle phenomena, such as cell cycle entry, restriction (R)-pointtransition, DNA replication, mitotic cell division, and checkpointcontrol.

On the other hand, NIRF has the potential to be engaged in theGRN, through complex formation with chromatin, chromatin-asso-ciated proteins, and/or DNA [3,18,23]. The two major tumor sup-pressors, pRB and p53, are representative chromatin-associated

proteins that affect a vast array of target genes. These tumor sup-pressors cooperatively impose strict surveillance of the activitiesof the cell cycle machinery to hinder cells from undergoing malig-nant transformation [10]. NIRF interacts with these tumor suppres-sors as well as two different G1 cyclins. All of these molecules areimportant regulators of the G1 checkpoint system, which is themost critical barrier for inhibiting cell transformation [51]. Theseinteractions suggest intense functional interplay between NIRFand these G1 control proteins to ensure the precise functioningof the G1 checkpoint system. NIRF may cooperate with pRB andp53 to effectively induce G1 arrest in response to various insultsthat trigger DNA damage.

Since UHRF1 has a similar domain composition to NIRF, it mayalso interact with the cell cycle machinery. In contrast to NIRF,UHRF1 promotes G1/S transition, and p53 downregulates theexpression levels of UHRF1 [52]. Currently, it is considered thatUHRF1 regulates transcription through recognition of chromatinmarks and interactions with proteins that modulate the histonemodification status such as HDAC1 and G9a [53], but not throughinteractions with the cell cycle ‘‘hub’’ proteins like NIRF [3]. How-ever, further analyses are required.

Finally, contrary to the sequence of events whereby NIRF regu-lates cyclin/CDK complexes, NIRF may be subject to regulation viaphosphorylation by CDKs. Indeed, NIRF is phosphorylated by the


cyclin E/CDK2 complex [34]. In this connection, it was reportedthat UHRF1 is phosphorylated by several kinases including proteinkinase A [54], casein kinase 2 [55], cyclin A2/CDK2 complex [56],and cyclin B/CDK1 complex [57]. Phosphorylation of UHRF1 bythe cyclin A2/CDK2 complex is required for zebrafish embryogen-esis [56], and M phase phosphorylation of UHRF1, possibly bythe cyclin B/CDK1 complex, regulates its physical association withthe deubiquitinase USP7 and stability, thereby contributing to cellcycle progression [57]. Phosphorylation by CDKs may represent animportant input mechanism for both NIRF and UHRF1 and theirnetwork activities.

5.2. The ubiquitin–proteasome system

The ubiquitin-mediated regulation of the cell cycle machineryunderlies numerous cellular decision processes. Notably, virtuallyall of the core cell cycle proteins are subject to control via ubiqui-tination, demonstrating that the ubiquitin–proteasome system is amajor network module that occupies a pivotal position in the cellcycle network. Not surprisingly, there is a strong association be-tween deregulated ubiquitination and human cancers [58–60].

The ubiquitin–proteasome system targets cyclins for regulateddestruction, which is critical for the maintenance of orderly pro-gression through the cell cycle [58,61]. Ubiquitination is catalyzedthrough successive reactions of three enzymes: a ubiquitin-acti-vating enzyme (E1), a ubiquitin-conjugating enzyme (E2), and aubiquitin ligase (E3). The polyubiquitinated cyclins are then re-cruited to a large ATP-dependent protease complex called the26S proteasome, where they are subject to rapid destruction[11]. Among the components of the ubiquitination system, the E3ubiquitin ligase is the primary determinant of target recognitiondepending on its binding capacity to the substrate [62]. The G1cyclins and the mitotic cyclins are basically degraded by theS-phase kinase-associated protein 1 (Skp1)/cullin 1 (Cul1)/F-boxprotein (SCF) complex and the anaphase-promoting complex/cyclosome (APC/C), respectively [58,63]. The SCF complexes usedifferent F-box proteins to recognize distinct substrates, whilethe APC/C uses only two substrate adapters to target many sub-strates [63].

With respect to the association with human cancers, it is mostimportant to note that impaired ubiquitination of either G1 cyclincauses its aberrant overexpression, resulting in deregulation of theG1/S transition and ultimately leading to cancer [59,64]. Therefore,complete understanding of the ubiquitin ligases for G1 cyclins is ofgreat importance. According to the classical view, F-box and WDrepeat domain containing 4 (FBXW4), F-box and WD repeat do-main containing 8 (FBXW8), and F-box protein 31 (FBXO31) targetcyclin D1, while F-box and WD repeat domain containing 7(FBXW7) recognizes cyclin E [58]. These SCF ubiquitin ligases areconsidered to contribute to the control at the G1/S boundary bypromoting the degradation of specific G1 cyclins. Consistent withthis, FBXW7 is a tumor suppressor whose loss of function is seenin a number of carcinomas [65].

Surprisingly, however, the identity of the ubiquitin ligase for cy-clin D1 is still under debate [66,67]. This situation is amazing, be-came abnormal overexpression of cyclin D1 caused by defectiveubiquitination is one of the most frequent causes of many typesof tumors [43]. Furthermore, it has been demonstrated that neitherSCF nor APC/C complexes contribute to the control of cyclin D1 sta-bility [67]. This controversy suggests the existence of other ubiqui-tin ligases that target cyclin D1 [67]. In this context, NIRF is anexample of a single-subunit, non-SCF type ubiquitin ligase for cy-clin D1.

We found that NIRF is capable of ubiquitinating two differentclasses of G1 cyclins, D1 and E1 [3]. In sharp contrast, the SCF ubiq-uitin ligases are only able to target either cyclin D1 or E, but not

both [58], thereby allowing individual control for each cyclin. Toour knowledge, no other ubiquitin ligases have been reported totarget both G1 cyclins. This particular property suggests a uniquemechanism underlying the substrate recognition by NIRF, whichcould provide coordinated regulation at the G1/S boundary. Inaddition, this observation highlights a possible characteristic prop-erty of NIRF as a hub-type ubiquitin ligase targeting multiple sub-strates for ubiquitination. Thus, in cases with NIRF abnormalities,the control of both cyclins D1 and E1 may become coincidentallyabrogated, which could readily result in impaired control at theG1/S boundary. It will be intriguing to clarify whether NIRF is alsocapable of targeting other mitotic cyclins for ubiquitination.

The recognition of G1 cyclins by NIRF is characteristic in fouradditional respects as follows. First, NIRF is a non-SCF type ubiqui-tin ligase for cyclin D1 [3]. This is consistent with the recent indi-cation that there would be a novel ubiquitin ligase targeting cyclinD1 other than the SCF and APC/C complexes [67]. Second, NIRF rec-ognizes the G1 cyclins without their phosphorylation [3]. Thus far,phosphorylation-independent ubiquitination pathways have beenproposed for both cyclins D1 and E, in addition to the well-knownphosphorylation-dependent pathways [68]. Since the SCF ligasestarget the G1 cyclins upon phosphorylation of the substrates, thepresence of the phosphorylation-independent pathways furtherhighlights the involvement of the SCF-independent ubiquitinationmechanisms for both G1 cyclins. Third, NIRF can ubiquitinate thefree forms of G1 cyclins that are not bound to their cognate CDKs[3]. Again, the presence of ubiquitination pathways for the freeforms of the G1 cyclins has been reported for both cyclins D1and E. NIRF represents a novel ubiquitin ligase for the free formsof cyclins D1 and E1, which may serve to regulate the cellular lev-els of free cyclins. With regard to cyclin E, NIRF binds cyclin E1 anddirectly promotes its ubiquitination, while FBXW7 recognizes thecyclin E protein only when it is phosphorylated by the activated cy-clin E/CDK2 complex [58,65]. These differences imply that the cel-lular conditions in which cyclin E is recognized by NIRF or FBXW7may be different. NIRF would restrict the abundance of cyclin E be-fore it reaches the maximal level at the G1/S boundary. Contrary tothis, FBXW7 is thought to target phosphorylated cyclin E after theCDK2 kinase activity reaches the maximal activity, thereby forminga negative feedback regulation of the cyclin E/CDK2 activity [69].The physiological relevance of the ubiquitination pathway for freecyclin D1 remains to be clarified. Fourth, the ubiquitination of cyc-lins by NIRF takes place in the nucleus (unpublished data). Whilethe ubiquitination of cyclin E occurs in both the nucleus and cyto-plasm [70], that of cyclin D1 predominantly takes place in the cyto-plasm [71]. However, there has been a debate regarding thesubcellular localizations for the ubiquitination and degradation ofcyclin D1 [68]. The SCF ligases ubiquitinate threonine 286(T286)-phosphorylated cyclin D1 within the cytoplasm and targetit for degradation [71]. Conversely, free cyclin D1 is ubiquitinatedand degraded in the nucleus, independently of T286 [72]. It willbe important to investigate whether NIRF is involved in the phos-phorylation-independent, intranuclear ubiquitination pathway forfree cyclin D1, and to elucidate the associations, if any, of this path-way with tumorigenesis.

The other ubiquitination substrates of NIRF include PCNP andhuntingtin, and the latter is described later in Section 6.2.

Finally, UHRF1 ubiquitinates histone H3 [73] and DNA methyl-transferase 1 (DNMT1) [74]. H3 ubiquitination is linked with theformation of heterochromatin [75]. On the other hand, ubiquitina-tion of DNMT1 involves acetylation regulation by Tip60/histonedeacetylase 1 (HDAC1) and ubiquitination regulation by UHRF1/USP specific peptidase 7 (USP7), which is also called Herpesvirus-associated ubiquitin-specific protease (HAUSP) [74]. These knowntargets of NIRF and UHRF1 for ubiquitination are totally different,suggesting functional differences between the two proteins.


However, there is still a possibility that UHRF1 may target compo-nents of the cell cycle machinery for ubiquitination, and this willbe interesting to explore in future studies.

5.3. The epigenetic system

Epigenetics refers to heritable changes in phenotype or geneexpression caused by mechanisms other than changes in theunderlying DNA [76]. The epigenetic system regulates DNA meth-ylation and covalent histone modifications, which influence geneexpression, and plays a chief regulatory role in differentiationand development [77]. On the other hand, this system also regu-lates cell cycle progression locally by affecting the expression ofindividual cell cycle genes, and globally by controlling chromatincondensation and chromosome segregation [78]. Conversely, thecell cycle mechanism ensures correct inheritance of epigeneticinformation to the daughter cells [78].

In general, proliferation and differentiation are mutually exclu-sive, although they are cooperatively regulated during the courseof development [79]. This ‘‘proliferation/differentiation switch’’constitutes a cell cycle checkpoint in the G1 phase [80]. This check-point is extremely important for normal development, and itsabrogation is the basis for malignant transformation and cancerdevelopment [79]. However, the precise mechanisms of how cellscoordinate this switch still remain unknown. Systems biologyresearchers have claimed that Waddington’s theory of develop-ment as a canalization of the epigenetic landscape shaped by genenetworks will provide clues toward solving this problem [81].Extrapolating from Waddington’s theory, proliferation and differ-entiation may be seen as solutions of cross-antagonistic networkmodules [76]. In other words, during the course of development,two cross-antagonistic networks would be working with mutuallyreciprocal patterns of expression. Conversely, the anti-correlatedexpression profiles between two network modules are importantindications of their involvement in the ‘‘switch’’ between the twodifferent phases of the cell [81]. Indeed, two distinct network mod-ules exist: the ‘‘proliferation module’’ that accelerates cell prolifer-ation and the ‘‘differentiation module’’ that promotes celldifferentiation. Importantly, as predicted by the theory, thesecross-antagonistic modules are inversely correlated with eachother at the transcription level [81]. Specifically, upon differentia-tion, the ‘‘proliferation module’’ is downregulated while the ‘‘dif-ferentiation module’’ is upregulated.

In this regard, NIRF and UHRF1 are involved in the GRN as epi-genetic factors with mutually reciprocal expression profiles duringthe course of differentiation [18]. UHRF1 is expressed in embryonicstem cells and downregulated during differentiation, whereas NIRFis upregulated and highly expressed in differentiated tissues [18].Moreover, these molecules are not only involved in the GRN, butalso in the PPIN. The concerted actions of these two different layersof networks are supposed to affect cell proliferation and differenti-ation in versatile ways. In that case, what is the difference betweenthe biological roles of the NIRF and UHRF1 gene networks? Is thereany relationship to Waddington’s prediction? Most probably, NIRFand UHRF1 are multifunctional intermodular hub proteins withmany interaction partners, and thereby behave as the central factorof each network. Furthermore, each of them would mediate inter-actions among the cell cycle machinery, ubiquitin–proteasomesystem, and epigenetic system. Therefore, the networks maygreatly amplify the original biological activities of each hub proteinresiding at the center of the respective network. Accordingly, thedifference between the biological functions of NIRF and UHRF1, ifany, would also be significantly amplified through extensive net-working. In more detail, the differences in the molecular struc-tures, functions, or expression profiles between the two hubproteins will bring about considerable differences in the biological

activities of the whole network. This will arise from any differencesat the molecular level between the two hub proteins by variousreasons, such as (1) differences in their repertoires of interactingpartners, (2) different outcomes of their interactions even withthe common interacting partners, (3) differences in the proteomeexpressed at the time and place of their existence, which cause dif-ferences in the composition of their protein complexes, (4) geneticand epigenetic differences in the organization of their promoters,which affect their expression profiles, and (5) prominent differ-ences between each network in the global architecture and func-tions caused by points (1 to 4). Thus, even slight differences inthe properties between NIRF and UHRF1 will bring about large dif-ferences in their physiological functions. Considering the hugenessof each network, the reciprocally correlated expression of such ver-satile hub proteins could be the basis of a ‘‘phase transition’’ at thesystems level, for example, the ‘‘proliferation/differentiationswitch’’.

Then, does the above discussion have merit to interpret themolecular distinction between NIRF and UHRF1? Before discussingtheir differences, we briefly summarize the currently availableinformation for UHRF1. UHRF1 was initially identified as mouseprotein Np95, which is predominantly expressed in the S phasein normal thymocytes [31]. UHRF1 is upregulated in various cancercells, especially malignant undifferentiated cancer cells with highproliferation ability [82,83]. Knockdown of UHRF1 expression incancer cells significantly suppresses cell growth, indicating thatUHRF1 is essential for cancer progression [53]. In S phase, UHRF1binds to PCNA, recognizes hemi-methylated DNA, which appearsat replication forks, recruits DNMT1 to the sites, and methylatesunmethylated cytosines in the hemi-methylated DNA. Therefore,UHRF1 plays a fundamental role for the maintenance of DNA meth-ylation patterns [24–26,84]. In addition, UHRF1 is involved in DNAdamage control [85], and in the replication and reorganization ofpericentromeric heterochromatin [86]. These observations suggestthat UHRF1 belongs to the ‘‘proliferation module’’, and therebydrives a wide range of cellular processes that are centrally involvedin cell proliferation. In this connection, the molecular activities ofthe genes belonging to the ‘‘proliferation module’’ are directed to-ward various aspects of proliferation-associated functions [81].Although UHRF1 has various roles, the cumulative outcomes ofits molecular functions will converge to cell proliferation-associ-ated activity at the cellular level.

On the other hand, slight differences have been noted betweenthe molecular structures and functions of NIRF and UHRF1. Theyexhibit opposite expression patterns during the course of develop-ment and tissue distributions. Similar to UHRF1, NIRF is localizedto the pericentric heterochromatin. NIRF interacts with methylatedhistone H3 lysine 9 (H3K9) through its TTD–PHD region, and prob-ably with hemi-methylated DNA through its SRA domain [18]. Inaddition, both NIRF and UHRF1 bind to DNA methyltransferases(DNMT1, DNMT3A, and DNMT3B) and a histone methyltransferase(G9a) [23]. Despite all these functional conservations, NIRF doesnot maintain the DNA methylation status [18,23]. UHRF1 interactswith DNMT1 in an S phase-dependent manner, while NIRF doesnot. In addition, NIRF and UHRF1 neither interact [3,23] nor colo-calize [23] with each other. Thus, UHRF1 and NIRF are not func-tionally redundant in terms of DNA methylation maintenance.

Collectively from these observations, the differences in NIRFand UHRF1 are slight at the molecular level. However, given thatNIRF and UHRF1 each have significant centrality in a respectivehuge network, even slight differences may lead to large differencesin their systems-level functions. If the differences between NIRFand UHRF1 are really of physiological importance, and if they rep-resent the cross-antagonistic modules of the ‘‘differentiation mod-ule’’ and the ‘‘proliferation module’’, respectively, there will be asignificant difference in their relevance to tumors. Indeed,


transient expression of NIRF induces G1 arrest [34], and there areconsiderable differences in their relationships with cancer, as de-scribed later in Section 6.1.

5.4. Integration of the network modules: involvement in cellulardecisions

We have described the centrality of NIRF in the cell cycle net-work that mediates interactions among three network modules,whereby it occupies a pivotal position in the cellular computingsystem. Here, we present a hypothesis of how the NIRF gene net-work may participate in cellular decision processes.

The cell cycle network is a complex ‘‘system’’ constructed frommultiple interactions between many distinct subnetwork modules[1]. However, the ‘‘emergent’’ properties of the whole network arenot a simple aggregation of those of the constituents. Rather, theyare inherently novel, according to systems biology [2]. NIRF medi-ates interactions among the three aforementioned modules withcompletely distinct properties. This indicates that the moleculesin the network are subject to multiple distinct layers of regulation,such as phosphorylation, ubiquitination, proteolysis, transcriptionregulation, and epigenetic regulation. Such a multidimensionalregulation will allow fine-tuning of the molecular properties ofeach component. Eventually, the whole network constructed fromlarge numbers of these fine-tuned components will exhibit anemergent state. A striking property of this network is the immensecomplexity of the physical and functional interactions betweenmany regulated components, which will allow the production ofa large amount of distinct functional wiring [87]. Waddington’stheory demonstrates that to exhibit distinct phenotypes, cells se-lect the appropriate network modules to be activated, while inac-tivating other modules [76]. Given that the total number ofgenes is limited, reorganization of the network architecture wouldbe an important means for creating a large amount of different net-work wiring to manifest versatile phenotypes. As an intermodularhub, NIRF could be expected to be involved in such a reorganiza-tion process of the network architecture [3].

The NIRF gene network is assumed to have numerous inputsand outputs, through multidirectional involvement in the differentnetwork modules. As described, NIRF exhibits a prominent central-ity in the PPIN [3]. In addition, NIRF is presumed to be involved inthe control of expression of many genes through chromatin and/ortranscriptional regulation [23].

As inputs, many events will cooperatively impact on the molec-ular properties of NIRF. These events may include the cell cycle-dependent formation of different protein complexes (e.g., cyclins,CDKs, pRB, p53, PCNA, DNMT1, HDAC1, and G9a), phosphorylationby kinases, including CDKs, and modification by ubiquitination(including autoubiquitination [13]). Notably, phosphorylation atmany serines and threonines in NIRF have been reported (http://www.phosphosite.org; [88]). The phosphorylation of any of thesesites may cause alterations in the molecular characteristics of NIRF.In addition, NIRF would integrate these different inputs with thegenetic and epigenetic information [3]. To integrate the variousinformation flows, it would be advantageous for NIRF to be associ-ated with the chromatin, where the highest priority informationexists. The integration would be performed through reading ofthe epigenetic codes under tight cooperation with the many signal-ing pathways as well as the cell cycle machinery and the ubiqui-tin–proteasome system.

On the other hand, from NIRF to downstream pathways, twodistinct layers of output would be in operation, i.e., NIRF wouldtake part in the GRN and PPIN. For the GRN, NIRF may be involvednot only in cell-type-specific gene regulation through the readingof various epigenetic marks [23], but also in more general gene

regulation through cooperation with diverse chromatin-associatedproteins and/or transcription factors, for instance, p53 or pRB.Moreover, NIRF would be involved in the modification of epige-netic information of histones and DNA, again in intimate cross-talkwith many signaling pathways, the cell cycle machinery, and theubiquitin–proteasome system. This could be carried out throughcooperation with various chromatin modifiers, such as G9a,DNMTs, and HDAC1. It will be interesting to clarify whether NIRFcan ubiquitinate histones and other chromatin-associated proteins.For the PPIN, NIRF forms complexes with the core components ofthe cell cycle machinery [cyclins (A, B, D, and E), CDKs (CDK1,CDK2, CDK4, and CDK6), and PCNA], two major tumor suppressors(p53 and pRB), chromatin proteins, chromatin modifiers (HDAC1and DNMTs), and components of the ubiquitin–proteasome system[3,13]. In addition, NIRF ubiquitinates other nuclear proteins[13,28]. The formation of multiple different protein complexes indifferent contexts may help to target various distinct proteins forubiquitination.

In general, ubiquitin-mediated degradation is rapid and irre-versible, but often transient. Protein phosphorylation is also rapid,but reversible and mostly quite transient. The specificities of thesereactions are determined at the molecular interaction level. In con-trast, writing and reading of the epigenetic marks are relativelytime-consuming processes. The specificity of epigenetic regulationlargely depends on the cellular context. In addition, epigeneticmarks are established during an allotted time period in early em-bryos, and remain stable throughout life in most tissues. Eventu-ally, combined effects of these timely and mechanisticallydistinct regulations would ensure highly sophisticated controlsover every target gene/protein at the molecular level. In turn, thiswould allow the establishment of extensive diversity in the net-work architecture at the systems level. In particular, NIRF ubiquiti-nates two G1 cyclins, while transient transfection of NIRF inducesG1 arrest [3,34]. Thus, ubiquitination of cyclins and the consequentmodulation of CDK activities represent one of the important meansof network reorganization by NIRF.

From the above discussions, it appears evident that NIRF is notonly a central factor in the network, but also a nodal point for thecellular information flows that influence diverse cellular functions.More importantly, the NIRF gene network is supposed to have theability to create a large amount of distinct topologies of networkwiring that may help the diversity of Waddington’s canals[76,89,90]. As a result, the NIRF gene network would facilitate cel-lular decisions toward diverse directions, e.g., cell proliferation, dif-ferentiation, death, or senescence [3,4]. Conversely, abrogation ofNIRF may impair cellular homeostasis, leading to a predispositionto various diseases, including cancer.

6. NIRF and diseases

Each of the three network modules interacting with NIRF, thecell cycle machinery, the ubiquitin–proteasome system, and theepigenetic system, is intimately involved in oncogenic processes[10,58,59,91,92]. As an intermodular hub, NIRF is therefore ex-pected to be associated with tumorigenesis. The molecular proper-ties of NIRF described in the earlier sections favor the concept thatit negatively regulates cell proliferation. Indeed, loss of the NIRFgene is frequently found in tumors. This is in sharp contrast tothe UHRF1 gene, which is scarcely subject to copy number altera-tions in tumors.

Other than tumorigenesis, a possible link of NIRF with the path-ogenesis of polyglutamine disease (Huntington’s disease) has beenproposed [28]. In this case, NIRF has been suggested to be involvedin the nuclear protein quality control (PQC) [93].

http://www.phosphosite.org

http://www.phosphosite.org


6.1. NIRF and cancer

The NIRF gene is localized to a region of chromosome 9p24 thathas been reported to be lost in numerous types of human malig-nancies at the highest frequency (48%) among all chromosome re-gions [94]. Indeed, we detected loss of heterozygosity of the NIRFgene in several kinds of cancers from the stomach, colon, andbreast [3]. We further examined the copy number aberrations ofthe NIRF gene by accessing the Oncomine database (http://www.oncomine.org/) [95] for the reasons described below.

Genome-wide characterization of cancers has revealed enor-mous diversity in the genomic aberrations in tumors among indi-viduals [96]. For this reason, and to avoid bias, accessing widerinformation resources is essential for determining the roles of aparticular gene in carcinogenesis. In addition, modern bioinformat-ics technology has yielded strong analytic algorithms for the pur-pose of such utilization. Cancer outlier profile analysis (COPA) isan innovative bioinformatics method that enables the identifica-tion of cancer genes with great accuracy and sensitivity [97].Among many cancer databases (see [98] for a review), Oncomineis implemented with COPA, and is thus suited for analyses of up-stream genes that can lead to carcinogenesis [97].

When COPA was carried out with Oncomine, DNA copy numberloss of the NIRF gene was revealed at statistically significant levelsin multiple kinds of malignancies originating from the brain,breast, stomach, kidney, hematopoietic tissue, and lung (Fig. 4)[3], while no significant pattern of copy number alteration of theUHRF1 gene was observed. As depicted, the canonical oncogenes(KRAS and EGFR) and tumor suppressor genes (pRB and p53) exhib-ited amplification-dominant (i.e., red in most cancers) and

Fig. 4. Summary of a cancer outlier profile analysis (COPA) [95,97] for NIRF. Dataset numare illustrated. Dark blue and light blue indicate the top 1% and top 5% among the genesamong the genes with amplification, respectively. The COPA shows that the NIRF gene isblue in most cancers, and red in a few cases). This feature is compatible with those of themost cancers). In contrast, the canonical oncogenes, KRAS and EGFR, are amplification doNIRF gene, the UHRF1 gene does not show a definite pattern of copy number aberration

deletion-dominant (i.e., blue in most cancers) appearances, respec-tively. NIRF exhibits the latter condition, i.e., COPA shows that theNIRF gene is preferentially lost in diverse tumors, while its ampli-fication is a rare event. This feature is compatible with those of theclassical tumor suppressor genes pRB and p53. Although UHRF1 isupregulated in various cancers, the mechanism is different fromthe canonical oncogenes KRAS and EGFR. The upregulation maybe secondary to the cell proliferation property of cancer cells, ormay be caused by some epigenetic changes such as structuralchanges to the chromatin of the UHRF1 gene.

With the discovery that the NIRF gene is subject to frequentcopy number aberrations, a more detailed study was performedusing a multigene analysis in Oncomine. This analysis revealed arecurrent microdeletion targeting NIRF in non-small cell lung car-cinoma (NSCLC). As a result, 13.3% of cases were hemizygously de-leted in the NIRF gene [3]. Consistently, the NIRF mRNA wasreduced at statistically significant levels in lung adenocarcinomaand squamous cell lung carcinoma, both of which are subtypes ofNSCLC. These findings suggest that loss of the NIRF gene is associ-ated with tumor formation, and supports the concept that NIRF is atumor suppressor.

In recent years, genome-wide studies have implied an associa-tion between the NIRF gene locus and colorectal cancer. Zankeet al. [99] and Poynter et al. [100] showed that the single nucleo-tide polymorphism (SNP) rs719725, which is located in the imme-diate proximity (47 kb upstream) of the NIRF gene, is associatedwith the risk of colorectal cancer. While both affirmative [101]and negative [102,103] data have been reported from replicationstudies, these researches pointed out the possibility of linkage ofNIRF with cancer susceptibility. Interestingly, rs719725 is localized

bers showing significant loss or amplification of the copy number of the NIRF genewith DNA loss, respectively. Dark red and light red indicate the top 1% and top 5%preferentially lost in diverse tumors, whereas its amplification is a rare event (i.e.,

classical tumor suppressor genes, pRB and p53, which are deletion dominant (blue inminant (red in most cancers). Contrary to the genetic abnormalities observed in thes.

http://www.oncomine.org/

http://www.oncomine.org/


just within the aforementioned commonly deleted region in NSCLC[3]. This indicates that rs719725 is really a target of chromosomalaberration in a particular type of cancer. Thus, the SNP rs719725appears to be associated with a chromosomal aberration inactivat-ing NIRF, which may cause predisposition to multiple types oftumors.

In line with the above observations of genomic aberrations,Sjöblom et al. [104] and Wood et al. [105] reported a list of candi-date cancer genes (CAN genes) that are mutated in breast and colo-rectal cancers. NIRF was included in the list, indicating that thecoding sequence of the NIRF gene is mutated at statistically signif-icant rates in cancers. Varley et al. [106] examined the methylationpatterns of the CAN gene promoters to determine whether theyexhibited tumor-specific methylation. It is well known that thepromoters of various tumor suppressor genes are frequentlyhypermethylated [91,92]. This can cause transcriptional suppres-sion of these tumor suppressor genes, often serving as a ‘‘secondhit’’, thereby leading to malignant transformation. They found thatthe promoter of the NIRF gene exhibited frequent tumor-specifichypermethylation in both breast (58%) and colon (50%) cancers.These findings could suggest that inactivation of the gene is akey step in tumorigenesis in both tissues [106]. Notably, the fre-quency of aberrant methylation approaches the significance ofeven the most common genetic mutations, such as adenomatouspolyposis coli (APC) or p53 mutations, which were reported to occurin 40–80% of tumors [104]. This conforms to the general hypothesisthat epigenetic defects at CAN genes are more frequent than genet-ic mutations [107]. Thus, the described observations collectivelydemonstrate that the NIRF gene is inactivated at significant ratesby either genetic or epigenetic aberrations in many types of cancer.

In addition to the gene inactivation in tumor samples, biologicaldata supporting the tumor suppressor role of NIRF have been re-ported. NIRF forms complexes with the cell cycle machinery, andubiquitinates both cyclins D1 and E1 [3]. Consistently, transientexpression of NIRF induces G1 arrest [3,34]. Furthermore, as de-scribed in Section 5.3, the expression profile of NIRF favors the con-cept that it belongs to the ‘‘differentiation module’’, which has arole in promoting cell differentiation while antagonizing cell prolif-eration. In conclusion, NIRF displays typical characteristics of a tu-mor suppressor gene that functions at the G1/S boundary to inhibitinappropriate cell cycle progression.

Although many reports, as well as our results, suggest that NIRFcan work as a tumor suppressor, other reports suggest that NIRFmay also have oncogenic functions. These contradictory resultsare mainly from in vitro experiments using cancer cell lines. NIRFwas found in amplicons at chromosome 9p24 in human esopha-geal and breast cancer cell lines. Knockdown of NIRF expressioninhibited the growth of breast cancer cells with amplification. Con-versely, ectopic overexpression of NIRF in non-tumorigenicMCF10A cells promoted their proliferation with downregulationof key cell-cycle inhibitors, such as p16INK4a, p21Waf1/Cip1, andp27Kip1 [50]. Another two studies indicated that NIRF is a targetof the microRNA let-7a [108,109], which is an important tumorsuppressor [110]. NIRF is required for let-7a-mediated elevationof p21Waf1/Cip1, and thus the growth-inhibitory effect of let-7a onA549 lung cancer cells may be explained by let-7a-induced sup-pression of NIRF and elevation of p21Waf1/Cip1 [108]. NIRF expres-sion was aberrantly increased in colorectal cancer tissues andassociated with poor overall survival. Knockdown of NIRF led to re-duced colorectal cell proliferation owing to cell cycle arrest at theG0/G1 phase and reduced cell migration [109]. Thus, these findingsindicate tumorigenic roles of NIRF in several cancer cell lines andtumor tissues, similar to the observation that TGF-b has tumor-suppressive functions in some types of cancer cells and oncogenicfunctions in other types of cancer cells.

In summary, the above observations demonstrate that NIRFplays a tumor suppressor role in many original tumors from pa-tients. In contrast, it also plays an oncogenic role, albeit mainlyin in vitro experiments. This apparent contradiction may arise be-cause NIRF is a multifunctional protein that is involved in the coor-dination of three different network modules, the cell cyclemachinery, the ubiquitin–proteasome system, and the epigeneticsystem. In view of the inherent complex nature of the NIRF genenetwork, it will be difficult to explore the precise roles of NIRF intumor biology. Conversely, the degree of controversy on this sub-ject is a strong indication of the importance of NIRF as the chiefdirector of cellular decision processes toward various directions.Thus, studying various different aspects of NIRF in tumor biologywill significantly improve our knowledge of what the bona fide bar-rier is between normal cells and tumor cells.

In further studies, it will be beneficial to consider the issueslisted below. First, the incidence of NIRF gene loss greatly exceedsthat of its amplification in original tumors from various tissues(Fig. 4). Such predominance probably reflects essential gene func-tions in vivo. Therefore, we favor the concept that NIRF behavesas a tumor suppressor in most tumor cases. However, given thatNIRF interacts with both the pRB and p53 tumor suppressors, NIRFmay promote tumor progression through ubiquitin-mediated deg-radation of either of these suppressors. In addition, although NIRFrapidly ubiquitinates cyclins D1 and E1 [3], it also (possibly later)downregulates p16INK4a, p21Waf1/Cip1, and p27Kip1 through epigeneticmechanisms [50]. This may cause bidirectional effects on the cellcycle progression. Second, apparently paradoxical upregulation oftumor suppressors (e.g., pRB [111], p14ARF [112], p16INK4a [113],HIC1 [114], Rad51 [115], BRCA2 [115], miR-449a, and miR-449b[116]) is often observed, reflecting negative-feedback loops thatconstitute failsafe antiproliferative mechanisms [111,114–116].This points out the need for a bioinformatic analysis, such as COPA,that helps toward discrimination between specific and non-spe-cific upregulation of mRNA [97]. Third, prolonged in vitro growthof tumor cell lines sometimes causes loss of representativity. Thisarises because of the evolution of novel genetic and epigenetic fea-tures that support cell proliferation and survival [117]. In addition,such situations may lead to significant differences between the re-sults of transient and stable transfection experiments. To overcomethe caveat of adaptation caused by stable transfection, we per-formed transient expression experiments. In contrast to stabletransfection, transient transfection of NIRF effectively broughtabout growth suppression ([3,34]; unpublished data). Finally, theantibody specificity should be carefully taken into account forthe interpretation of results. We have created mouse monoclonalantibodies that show no cross-reactivity against UHRF1 (manu-script in preparation). We think that Nirf-knockout mice will pro-vide critical information regarding the roles of Nirf intumorigenesis, and we are now ready for these analyses.

6.2. NIRF and neurodegenerative diseases

Huntington’s disease and its related autosomal-dominant poly-glutamine neurodegenerative diseases are characterized by intran-euronal accumulation of protein aggregates. The nucleus reliesentirely on the ubiquitin–proteasome system for PQC. Iwata et al.[28] identified potential roles for NIRF in nuclear PQC. NIRF targetsnuclear polyglutamine aggregates for ubiquitin-mediated degrada-tion, thereby rescuing cultured cells and primary neurons frompolyglutamine-induced cytotoxicity. A toxic expanded polygluta-mine tract version of truncated huntingtin is more preferablydegraded than the non-toxic normal polyglutamine tract version.[28]. More information is needed for how the nuclear PQCsystem selectively recognizes the disease-causing aggregates. The


potential role of NIRF in such a discrimination mechanism awaitsfurther investigation.

Given that NIRF is well expressed in the brain [29], its functionin nuclear protein degradation may have another role in the main-tenance of neuronal homeostasis. In terminally differentiated neu-rons, aberrant re-entry into the cell cycle triggers neuronal deathinstead of proliferation, and is a common pathway shared by someacquired and neurodegenerative disorders [118]. NIRF may repressthe abnormal expression of cell cycle proteins through ubiquitin-mediated degradation and/or epigenetic silencing, which may pro-tect neurons against unwanted cell death. Thus, it can be hypoth-esized that the abrogation of NIRF may be involved in thepathogenesis of various neurological disorders.

6.3. NIRF and cancer diagnosis and treatment

At present, various genetic and epigenetic abnormalities of NIRFhave been found to be associated with cancer. For diagnostic pur-poses, epigenetic abnormalities may be promising. The promoterhypermethylation of NIRF in breast and colon cancers showed bothsensitivity and specificity, making it a strong classifier of cancercells versus normal cells [106]. In addition, it is a good candidatefor a biomarker for distinguishing cancer subtypes or as a diagnos-tic biomarker usable in peripheral specimens. Furthermore, geneticaberrations involving the NIRF gene may also be a predictor of dis-ease susceptibility. The linkage with various diseases should besystematically analyzed with regard to SNPs, copy number altera-tions, mutations, and expression levels of the NIRF gene.

Finally, it should be remembered that the NIRF gene networkcould be associated with almost every means of cancer treatment.Further investigations of NIRF are therefore necessary for futureimprovements to our strategies for fighting cancer.

7. Conclusions

The cell cycle network itself is a nodal point of the cellular com-puting system. NIRF is a prototypical intermodular hub proteinthat ensures coordination among three fundamental subnetworkmodules in the cell cycle network. Therefore, NIRF occupies a cen-tral position not only in the cell cycle network but also in thewhole information network of the cell. Currently, many lines ofevidence have certified the association between NIRF and tumors.Further investigations of the NIRF gene network will provide criti-cal insights into how the cell cycle network regulates the epige-netic landscape, thereby influencing the cellular identity anddifferentiation programs, and conversely, how the epigenetic land-scape affects the cell cycle network, thereby contributing to thecellular decisions toward various directions, e.g., cell proliferation,differentiation, death, and senescence. In addition, we will be ableto obtain novel insights into how the ubiquitin–proteasome sys-tem is integrated into these cellular decision machineries.

We expect that further developments will yield an efficientmolecular biology method for analyzing the multifaceted activitiesof the NIRF gene network in vivo. However, the enormous com-plexity of the dynamic network wiring will pose many new prob-lems. First, the total understanding of such an intricate networkwill critically depend on the development of systems biology. Sec-ond, since NIRF is localized just at the branching point toward mul-tiple cell fates, functional analyses of NIRF will be very difficult. Ofcourse, overcoming these difficulties will help toward the emer-gence of a great paradigm shift in the long run.

Hopefully, we will be able to obtain a critical understanding ofthe dynamic nature of the network, which will help us to choosethe optimal network to focus on as a specific therapeutic target,to achieve maximum beneficial effects with minimum adverse

effects. We believe that joint research studies on NIRF throughmolecular and systems biology will facilitate the understandingof the pathophysiology as well as the development of therapeuticstrategies for many diseases, including neoplastic, neurological,and immunological diseases.

References

[1] Rind, D. (1999) Complexity and climate. Science 284, 105–107.[2] Levesque, M.P. and Benfey, P.N. (2004) Systems biology. Curr. Biol. 14, R179–

R180.[3] Mori, T., Ikeda, D.D., Fukushima, T., Takenoshita, S. and Kochi, H. (2011) NIRF

constitutes a nodal point in the cell cycle network and is a candidate tumorsuppressor. Cell Cycle 10, 3284–3299.

[4] Massague, J. (2004) G1 cell-cycle control and cancer. Nature 432, 298–306.[5] Barabasi, A.L. and Oltvai, Z.N. (2004) Network biology: understanding the

cell’s functional organization. Nat. Rev. Genet. 5, 101–113.[6] He, X. and Zhang, J. (2006) Why do hubs tend to be essential in protein

networks? PLoS Genet. 2, e88.[7] Han, J.D. et al. (2004) Evidence for dynamically organized modularity in the

yeast protein-protein interaction network. Nature 430, 88–93.[8] Vidal, M., Cusick, M.E. and Barabasi, A.L. (2011) Interactome networks and

human disease. Cell 144, 986–998.[9] Taylor, I.W. et al. (2009) Dynamic modularity in protein interaction networks

predicts breast cancer outcome. Nat. Biotechnol. 27, 199–204.[10] Sherr, C.J. (2004) Principles of tumor suppression. Cell 116, 235–246.[11] Hershko, A. and Ciechanover, A. (1998) The ubiquitin system. Annu. Rev.

Biochem. 67, 425–479.[12] Goldberg, A.D., Allis, C.D. and Bernstein, E. (2007) Epigenetics: a landscape

takes shape. Cell 128, 635–638.[13] Mori, T., Li, Y., Hata, H. and Kochi, H. (2004) NIRF is a ubiquitin ligase that is

capable of ubiquitinating PCNP, a PEST-containing nuclear protein. FEBS Lett.557, 209–214.

[14] Rogers, S., Wells, R. and Rechsteiner, M. (1986) Amino acid sequencescommon to rapidly degraded proteins: the PEST hypothesis. Science 234,364–368.

[15] Nasrabadi, D., Rezaei Larijani, M., Pirhaji, L., Gourabi, H., Shahverdi, A.,Baharvand, H. and Salekdeh, G.H. (2009) Proteomic analysis of monkeyembryonic stem cell during differentiation. J. Proteome Res. 8, 1527–1539.

[16] Mori, T., Li, Y., Hata, H., Ono, K. and Kochi, H. (2002) NIRF, a novel RING fingerprotein, is involved in cell-cycle regulation. Biochem. Biophys. Res. Commun.296, 530–536.

[17] Upadhya, S.C. and Hegde, A.N. (2003) A potential proteasome-interactingmotif within the ubiquitin-like domain of parkin and other proteins. TrendsBiochem. Sci. 28, 280–283.

[18] Pichler, G. et al. (2011) Cooperative DNA and histone binding by Uhrf2 linksthe two major repressive epigenetic pathways. J. Cell. Biochem. 112, 2585–2593.

[19] Nady, N. et al. (2011) Recognition of multivalent histone states associatedwith heterochromatin by UHRF1 protein. The Journal of biological chemistry286, 24300–24311.

[20] Botuyan, M.V., Lee, J., Ward, I.M., Kim, J.E., Thompson, J.R., Chen, J. and Mer, G.(2006) Structural basis for the methylation state-specific recognition ofhistone H4–K20 by 53BP1 and Crb2 in DNA repair. Cell 127, 1361–1373.

[21] Xie, S., Jakoncic, J. and Qian, C. (2012) UHRF1 double tudor domain and theadjacent PHD finger act together to recognize K9me3-containing histone H3tail. J. Mol. Biol. 415, 318–328.

[22] Rajakumara, E. et al. (2011) PHD finger recognition of unmodified histoneH3R2 links UHRF1 to regulation of euchromatic gene expression. Mol. Cell 43,275–284.

[23] Zhang, J. et al. (2011) S phase-dependent interaction with DNMT1 dictatesthe role of UHRF1 but not UHRF2 in DNA methylation maintenance. Cell Res.21, 1723–1739.

[24] Hashimoto, H., Horton, J.R., Zhang, X., Bostick, M., Jacobsen, S.E. and Cheng, X.(2008) The SRA domain of UHRF1 flips 5-methylcytosine out of the DNAhelix. Nature 455, 826–829.

[25] Avvakumov, G.V., Walker, J.R., Xue, S., Li, Y., Duan, S., Bronner, C., Arrowsmith,C.H. and Dhe-Paganon, S. (2008) Structural basis for recognition of hemi-methylated DNA by the SRA domain of human UHRF1. Nature 455, 822–825.

[26] Arita, K., Ariyoshi, M., Tochio, H., Nakamura, Y. and Shirakawa, M. (2008)Recognition of hemi-methylated DNA by the SRA protein UHRF1 by a base-flipping mechanism. Nature 455, 818–821.

[27] Chasapis, C.T. and Spyroulias, G.A. (2009) RING finger E(3) ubiquitin ligases:structure and drug discovery. Curr. Pharm. Des. 15, 3716–3731.

[28] Iwata, A. et al. (2009) Intranuclear degradation of polyglutamine aggregatesby the ubiquitin-proteasome system. J. Biol. Chem. 284, 9796–9803.

[29] Wu, C. et al. (2009) BioGPS: an extensible and customizable portal forquerying and organizing gene annotation resources. Genome Biol. 10, R130.

[30] Bronner, C., Achour, M., Arima, Y., Chataigneau, T., Saya, H. and Schini-Kerth,V.B. (2007) The UHRF family: oncogenes that are drugable targets for cancertherapy in the near future? Pharmacol. Ther. 115, 419–434.

[31] Muto, M., Utsuyama, M., Horiguchi, T., Kubo, E., Sado, T. and Hirokawa, K.(1995) The characterization of the monoclonal antibody Th-10a, specific for a


nuclear protein appearing in the S phase of the cell cycle in normalthymocytes and its unregulated expression in lymphoma cell lines. CellProlif. 28, 645–657.

[32] Hopfner, R. et al. (2000) ICBP90, a novel human CCAAT binding protein,involved in the regulation of topoisomerase IIalpha expression. Cancer Res.60, 121–128.

[33] Fujimori, A. et al. (1998) Cloning and mapping of Np95 gene which encodes anovel nuclear protein associated with cell proliferation. Mamm. Genome 9,1032–1035.

[34] Li, Y., Mori, T., Hata, H., Homma, Y. and Kochi, H. (2004) NIRF induces G1arrest and associates with Cdk2. Biochem. Biophys. Res. Commun. 319, 464–468.

[35] Reiner, S.L. (2005) Epigenetic control in the immune response. Human Mol.Genetics 14 Spec No 1, R41–6.

[36] Shaknovich, R. et al. (2011) DNA methyltransferase 1 and DNA methylationpatterning contribute to germinal center B-cell differentiation. Blood 118,3559–3569.

[37] Malumbres, M. and Barbacid, M. (2009) Cell cycle, CDKs and cancer: achanging paradigm. Nat. Rev. Cancer 9, 153–166.

[38] Branzei, D. and Foiani, M. (2010) Maintaining genome stability at thereplication fork. Nat. Rev. Mol. Cell Biol. 11, 208–219.

[39] Coudreuse, D. and Nurse, P. (2010) Driving the cell cycle with a minimal CDKcontrol network. Nature 468, 1074–1079.

[40] Murray, A.W. (2004) Recycling the cell cycle: cyclins revisited. Cell 116, 221–234.

[41] Pagliuca, F.W., Collins, M.O., Lichawska, A., Zegerman, P., Choudhary, J.S. andPines, J. (2011) Quantitative proteomics reveals the basis for the biochemicalspecificity of the cell-cycle machinery. Mol. Cell 43, 406–417.

[42] Musgrove, E.A., Caldon, C.E., Barraclough, J., Stone, A. and Sutherland, R.L.(2011) Cyclin D as a therapeutic target in cancer. Nat. Rev. Cancer 11, 558–572.

[43] Alao, J.P. (2007) The regulation of cyclin D1 degradation: roles in cancerdevelopment and the potential for therapeutic invention. Mol. Cancer 6, 24.

[44] Chen, H.Z., Tsai, S.Y. and Leone, G. (2009) Emerging roles of E2Fs in cancer: anexit from cell cycle control. Nat. Rev. Cancer 9, 785–797.

[45] Weinberg, R.A. (1995) The retinoblastoma protein and cell cycle control. Cell81, 323–330.

[46] Hwang, H.C. and Clurman, B.E. (2005) Cyclin E in normal and neoplastic cellcycles. Oncogene 24, 2776–2786.

[47] Sullivan, M. and Morgan, D.O. (2007) Finishing mitosis, one step at a time.Nat. Rev. Mol. Cell Biol. 8, 894–903.

[48] Nariai, N., Tamada, Y., Imoto, S. and Miyano, S. (2005) Estimating generegulatory networks and protein–protein interactions of Saccharomycescerevisiae from multiple genome-wide data. Bioinformatics 21 (Suppl 2),ii206–ii212.

[49] Chaturvedi, I., Sakharkar, M.K. and Rajapakse, J.C. (2007) Validation of generegulatory networks from protein-protein interaction data: application tocell-cycle regulation in: PRIB 2007 (Rajapakse, J.C., Schmidt, B. and Volkert,G., Eds.), pp. 300–310, Springer, Heidelberg.

[50] Wu, J. et al. (2012) Identification and functional analysis of 9p24 amplifiedgenes in human breast cancer. Oncogene 31, 333–341.

[51] Kastan, M.B. and Bartek, J. (2004) Cell-cycle checkpoints and cancer. Nature432, 316–323.

[52] Arima, Y., Hirota, T., Bronner, C., Mousli, M., Fujiwara, T., Niwa, S., Ishikawa, H.and Saya, H. (2004) Down-regulation of nuclear protein ICBP90 by p53/p21Cip1/WAF1-dependent DNA-damage checkpoint signals contributes tocell cycle arrest at G1/S transition. Genes Cells 9, 131–142.

[53] Unoki, M. (2011) Current and potential anticancer drugs targeting membersof the UHRF1 complex including epigenetic modifiers. Recent Pat. Anti-Cancer Drug Discovery 6, 116–130.

[54] Trotzier, M.A. et al. (2004) Phosphorylation of ICBP90 by protein kinase Aenhances topoisomerase IIalpha expression. Biochem. Biophys. Res.Commun. 319, 590–595.

[55] Bronner, C., Trotzier, M.A., Filhol, O., Cochet, C., Rochette-Egly, C., Scholler-Guinard, M., Klein, J.P. and Mousli, M. (2004) The antiapoptoticprotein ICBP90 is a target for protein kinase 2. Ann. N. Y. Acad. Sci. 1030,355–360.

[56] Chu, J. et al. (2012) UHRF1 phosphorylation by cyclin A2/cyclin-dependentkinase 2 is required for zebrafish embryogenesis. Mol. Biol. Cell 23, 59–70.

[57] Ma, H. et al. (2012) M phase phosphorylation of the epigenetic regulatorUHRF1 regulates its physical association with the deubiquitylase USP7 andstability, Proc. Natl. Acad. Sci. USA.

[58] Nakayama, K.I. and Nakayama, K. (2006) Ubiquitin ligases: cell-cycle controland cancer. Nat. Rev. Cancer 6, 369–381.

[59] Kitagawa, K., Kotake, Y. and Kitagawa, M. (2009) Ubiquitin-mediated controlof oncogene and tumor suppressor gene products. Cancer Sci. 100, 1374–1381.

[60] Lipkowitz, S. and Weissman, A.M. (2011) RINGs of good and evil: RING fingerubiquitin ligases at the crossroads of tumour suppression and oncogenesis.Nat. Rev. Cancer 11, 629–643.

[61] Ang, X.L. and Harper, J.W. (2004) Interwoven ubiquitination oscillators andcontrol of cell cycle transitions. Science’s STKE 2004, 31.

[62] Pickart, C.M. (2001) Mechanisms underlying ubiquitination. Annu. Rev.Biochem. 70, 503–533.

[63] Skaar, J.R. and Pagano, M. (2009) Control of cell growth by the SCF and APC/Cubiquitin ligases. Curr. Opin. Cell Biol. 21, 816–824.

[64] Shi, D. and Grossman, S.R. (2010) Ubiquitin becomes ubiquitous in cancer:emerging roles of ubiquitin ligases and deubiquitinases in tumorigenesis andas therapeutic targets. Cancer Biol. Ther. 10, 737–747.

[65] Welcker, M. and Clurman, B.E. (2008) FBW7 ubiquitin ligase: a tumoursuppressor at the crossroads of cell division, growth and differentiation. Nat.Rev. Cancer 8, 83–93.

[66] Silverman, J.S., Skaar, J.R. and Pagano, M. (2012) SCF ubiquitin ligases in themaintenance of genome stability. Trends Biochem. Sci. 37, 66–73.

[67] Kanie, T. et al. (2012) Genetic reevaluation of the role of F-box proteins incyclin D1 degradation. Mol. Cell. Biol. 32, 590–605.

[68] Alao, J.P. (2009) G1 Phase cyclins in cancer development and progression in:Checkpoint Controls and Targets in Cancer Therapy, Cancer Drug Discoveryand Development (Siddik, Z.H., Ed.), pp. 123–153, Humana Press, New York.

[69] Gerard, C. and Goldbeter, A. (2009) Temporal self-organization of the cyclin/Cdk network driving the mammalian cell cycle. Proc. Natl. Acad. Sci. U.S.A.106, 21643–21648.

[70] Brandt, Y., Mitchell, T., Wu, Y. and Hartley, R.S. (2011) Developmentaldownregulation of Xenopus cyclin E is phosphorylation and nuclear importdependent and is mediated by ubiquitination. Dev. Biol. 355, 65–76.

[71] Kim, J.K. and Diehl, J.A. (2009) Nuclear cyclin D1: an oncogenic driver inhuman cancer. J. Cell. Physiol. 220, 292–296.

[72] Germain, D., Russell, A., Thompson, A. and Hendley, J. (2000) Ubiquitinationof free cyclin D1 is independent of phosphorylation on threonine 286. J. Biol.Chem. 275, 12074–12079.

[73] Citterio, E., Papait, R., Nicassio, F., Vecchi, M., Gomiero, P., Mantovani, R., DiFiore, P.P. and Bonapace, I.M. (2004) Np95 is a histone-binding proteinendowed with ubiquitin ligase activity. Mol. Cell. Biol. 24, 2526–2535.

[74] Du, Z. et al. (2010) DNMT1 stability is regulated by proteins coordinatingdeubiquitination and acetylation-driven ubiquitination. Sci. Signal. 3, ra80.

[75] Karagianni, P., Amazit, L., Qin, J. and Wong, J. (2008) ICBP90, a novel methylK9 H3 binding protein linking protein ubiquitination with heterochromatinformation. Mol. Cell. Biol. 28, 705–717.

[76] Graf, T. and Enver, T. (2009) Forcing cells to change lineages. Nature 462,587–594.

[77] Gibney, E.R. and Nolan, C.M. (2010) Epigenetics and gene expression.Heredity 105, 4–13.

[78] Bou Kheir, T. and Lund, A.H. (2010) Epigenetic dynamics across the cell cycle.Essays Biochem. 48, 107–120.

[79] Maeda, Y. (2011) Cell-cycle checkpoint for transition from cell division todifferentiation. Dev. Growth Differ. 53, 463–481.

[80] Sherr, C.J. (1996) Cancer cell cycles. Science 274, 1672–1677.[81] Xia, K. et al. (2006) Identification of the proliferation/differentiation switch in

the cellular network of multicellular organisms. PLoS Comput. Biol. 2, e145.[82] Unoki, M., Kelly, J.D., Neal, D.E., Ponder, B.A., Nakamura, Y. and Hamamoto, R.

(2009) UHRF1 is a novel molecular marker for diagnosis and the prognosis ofbladder cancer. Br. J. Cancer 101, 98–105.

[83] Unoki, M., Daigo, Y., Koinuma, J., Tsuchiya, E., Hamamoto, R. and Nakamura,Y. (2010) UHRF1 is a novel diagnostic marker of lung cancer. Br. J. Cancer 103,217–222.

[84] Sharif, J. et al. (2007) The SRA protein Np95 mediates epigenetic inheritanceby recruiting Dnmt1 to methylated DNA. Nature 450, 908–912.

[85] Tien, A.L., Senbanerjee, S., Kulkarni, A., Mudbhary, R., Goudreau, B., Ganesan,S., Sadler, K.C. and Ukomadu, C. (2011) UHRF1 depletion causes a G2/Marrest, activation of DNA damage response and apoptosis. Biochem. J. 435,175–185.

[86] Papait, R., Pistore, C., Negri, D., Pecoraro, D., Cantarini, L. and Bonapace, I.M.(2007) Np95 is implicated in pericentromeric heterochromatin replicationand in major satellite silencing. Mol. Biol. Cell 18, 1098–1106.

[87] Magtanong, L. et al. (2011) Dosage suppression genetic interaction networksenhance functional wiring diagrams of the cell. Nat. Biotechnol. 29, 505–511.

[88] Hornbeck, P.V., Kornhauser, J.M., Tkachev, S., Zhang, B., Skrzypek, E., Murray,B., Latham, V. and Sullivan, M. (2012) PhosphoSitePlus: a comprehensiveresource for investigating the structure and function of experimentallydetermined post-translational modifications in man and mouse. NucleicAcids Res. 40, D261–D270.

[89] Hemberger, M., Dean, W. and Reik, W. (2009) Epigenetic dynamics of stemcells and cell lineage commitment: digging Waddington’s canal. Nat. Rev.Mol. Cell Biol. 10, 526–537.

[90] Waddington, C.H. (1959) Canalization of development and geneticassimilation of acquired characters. Nature 183, 1654–1655.

[91] Rodriguez-Paredes, M. and Esteller, M. (2011) Cancer epigenetics reachesmainstream oncology. Nat. Med. 17, 330–339.

[92] Baylin, S.B. and Jones, P.A. (2011) A decade of exploring the cancerepigenome – biological and translational implications. Nat. Rev. Cancer 11,726–734.

[93] Fredrickson, E.K. and Gardner, R.G. (2012). Selective destruction of abnormalproteins by ubiquitin-mediated protein quality control degradation. SeminCell Dev Biol. http://dx.doi.org/10.1016/j.semcdb.2011.12.006.

[94] Knuutila, S. et al. (1999) DNA copy number losses in human neoplasms. Am. J.Pathol. 155, 683–694.

[95] Rhodes, D.R. et al. (2007) Oncomine 3.0: genes, pathways, and networks in acollection of 18,000 cancer gene expression profiles. Neoplasia 9, 166–180.

[96] Akavia, U.D. et al. (2010) An integrated approach to uncover drivers of cancer.Cell 143, 1005–1017.

[97] MacDonald, J.W. and Ghosh, D. (2006) COPA–cancer outlier profile analysis.Bioinformatics 22, 2950–2951.

http://dx.doi.org/10.1016/j.semcdb.2011.12.006


[98] Gadaleta, E., Lemoine, N.R. and Chelala, C. (2011) Online resources of cancerdata: barriers, benefits and lessons. Briefings Bioinf. 12, 52–63.

[99] Zanke, B.W. et al. (2007) Genome-wide association scan identifies acolorectal cancer susceptibility locus on chromosome 8q24. Nat. Genet. 39,989–994.

[100] Poynter, J.N. et al. (2007) Variants on 9p24 and 8q24 are associated with riskof colorectal cancer: results from the Colon Cancer Family Registry. CancerRes. 67, 11128–11132.

[101] Kocarnik, J.D. et al. (2010) Characterization of 9p24 risk locus and colorectaladenoma and cancer: gene-environment interaction and meta-analysis.Cancer Epidemiol. Biomark. Prev. 19, 3131–3139.

[102] von Holst, S. et al. (2010) Association studies on 11 published colorectalcancer risk loci. Br. J. Cancer 103, 575–580.

[103] Lochhead, P. et al. (2011) Possible association between a geneticpolymorphism at 8q24 and risk of upper gastrointestinal cancer. Eur. J.Cancer Prev. 20, 54–57.

[104] Sjöblom, T. et al. (2006) The consensus coding sequences of human breastand colorectal cancers. Science 314, 268–274.

[105] Wood, L.D. et al. (2007) The genomic landscapes of human breast andcolorectal cancers. Science 318, 1108–1113.

[106] Varley, K.E. and Mitra, R.D. (2010) Bisulfite Patch PCR enables multiplexedsequencing of promoter methylation across cancer samples. Genome Res. 20,1279–1287.

[107] Chan, T.A. et al. (2008) Convergence of mutation and epigenetic alterationsidentifies common genes in cancer that predict for poor prognosis. PLoS Med.5, e114.

[108] He, X., Duan, C., Chen, J., Ou-Yang, X., Zhang, Z., Li, C. and Peng, H. (2009) Let-7a elevates p21(WAF1) levels by targeting of NIRF and suppresses the growthof A549 lung cancer cells. FEBS Lett. 583, 3501–3507.

[109] Wang, F., Zhang, P., Ma, Y., Yang, J., Moyer, M.P., Shi, C., Peng, J. and Qin, H.(2012) NIRF is frequently upregulated in colorectal cancer and itsoncogenicity can be suppressed by let-7a microRNA. Cancer Lett. 314, 223–231.

[110] Brueckner, B., Stresemann, C., Kuner, R., Mund, C., Musch, T., Meister, M.,Sultmann, H. and Lyko, F. (2007) The human let-7a-3 locus contains an

epigenetically regulated microRNA gene with oncogenic function. CancerRes. 67, 1419–1423.

[111] Stevaux, O. and Dyson, N.J. (2002) A revised picture of the E2F transcriptionalnetwork and RB function. Curr. Opin. Cell Biol. 14, 684–691.

[112] Komori, H., Enomoto, M., Nakamura, M., Iwanaga, R. and Ohtani, K. (2005)Distinct E2F-mediated transcriptional program regulates p14ARF geneexpression. EMBO J. 24, 3724–3736.

[113] Lee, C.T., Capodieci, P., Osman, I., Fazzari, M., Ferrara, J., Scher, H.I. andCordon-Cardo, C. (1999) Overexpression of the cyclin-dependent kinaseinhibitor p16 is associated with tumor recurrence in human prostate cancer.Clin. Cancer Res. 5, 977–983.

[114] Jenal, M. et al. (2009) The tumor suppressor gene hypermethylated in cancer1 is transcriptionally regulated by E2F1. Mol. Cancer Res. 7, 916–922.

[115] Klopfleisch, R. and Gruber, A.D. (2009) Increased expression of BRCA2 andRAD51 in lymph node metastases of canine mammary adenocarcinomas.Vet. Pathol. 46, 416–422.

[116] Yang, X., Feng, M., Jiang, X., Wu, Z., Li, Z., Aau, M. and Yu, Q. (2009) MiR-449aand miR-449b are direct transcriptional targets of E2F1 and negativelyregulate pRb-E2F1 activity through a feedback loop by targeting CDK6 andCDC25A. Genes Dev. 23, 2388–2393.

[117] Gisselsson, D., Lindgren, D., Mengelbier, L.H., Ora, I. and Yeger, H. (2010)Genetic bottlenecks and the hazardous game of population reduction in cellline based research. Exp. Cell Res. 316, 3379–3386.

[118] Wang, W., Bu, B., Xie, M., Zhang, M., Yu, Z. and Tao, D. (2009) Neural cell cycledysregulation and central nervous system diseases. Prog. Neurobiol. 89, 1–17.

[119] Ren, J., Wen, L., Gao, X., Jin, C., Xue, Y. and Yao, X. (2009) DOG 1.0: illustratorof protein domain structures. Cell Res. 19, 271–273.

[120] McCall, M.N., Uppal, K., Jaffee, H.A., Zilliox, M.J. and Irizarry, R.A. (2011) TheGene Expression Barcode: leveraging public data repositories to begincataloging the human and murine transcriptomes. Nucleic Acids Res. 39,D1011–D1015.

[121] Cerami, E.G. et al. (2011) Pathway Commons, a web resource for biologicalpathway data. Nucleic Acids Res. 39, D685–D690.

NIRF/UHRF2 occupies a central position in the cell cycle ... · Review NIRF/UHRF2 occupies a...

Documents

Transcript of NIRF/UHRF2 occupies a central position in the cell cycle ... · Review NIRF/UHRF2 occupies a...