ZRF1 controls oncogene-induced senescence through the INK4-ARF locus

ORIGINAL ARTICLE

ZRF1 controls oncogene-induced senescence throughthe INK4-ARF locusJD Ribeiro1, L Morey1, A Mas1, A Gutierrez1, NM Luis1, S Mejetta1, H Richly1, SA Benitah1,2, WM Keyes1 and L Di Croce1,2

The reactivation of the INK4-ARF locus, which is epigenetically repressed by Polycomb proteins in healthy cells, is a hallmark ofsenescence. One mechanism of reactivating Polycomb-silenced genes is mediated by the epigenetic factor ZRF1, which associateswith ubiquitinated histone H2A. We show that cells undergoing senescence following oncogenic Ras expression have increasedZRF1 levels, and that this binds to the p15INK4b, ARF and p16INK4a promoters. Furthermore, ZRF1 depletion in oncogenic Ras-expressing cells restores proliferation by preventing Arf and p16Ink4a expression, consequently bypassing senescence. Thus, ZRF1regulates the INK4-ARF locus during cellular proliferation and senescence, and alterations in ZRF1 may contribute to tumorigenesis.

Oncogene (2013) 32, 2161–2168; doi:10.1038/onc.2012.241; published online 25 June 2012

Keywords: chromatin; Polycomb; senescence; ZRF1

INTRODUCTIONThe INK4-ARF locus, which encodes the three related genesp15INK4b, ARF and p16INK4a, has a key surveillance functionagainst tumorigenesis.1,2 The p16INK4a and p15INK4b proteins arecyclin-dependent kinase inhibitors that prevent the CDK4/CDK6-directed phosphorylation of the retinoblastoma protein, andconsequently prevent cell cycle progression; in contrast, ARFprimarily regulates the activation of the tumor suppressor p53 bynegatively regulating MDM2, an E3-ubiquitin ligase of p53.2 Allthree products of the INK4-ARF locus are tumor suppressors thatact synergistically to generate a protective response againstoncogenic insults and other stresses, and they have pivotal rolesin activating the senescence response program.3–5

Cellular senescence is a stress response induced by a variety ofstimuli, including DNA damage, progressive telomere shortening,oxidative stress and activated oncogenes.5,6 After induction,senescent cells halt proliferation and express critical mediatorsof tumor suppression, including p15INK4b, p16INK4a and ARF.3

Thus, senescence is a potent tumor suppressor mechanism in pre-malignant lesions that must be overcome to allow progression tofull-blown malignancy.7 Mediators of the senescence response,not all of which have been characterized, are also likely to haveroles in tumor suppression.

Chromatin regulation has been strongly implicated in engagingthe cellular senescence program, although many details remainunknown. Among the best identified changes are epigeneticalterations in the INK4-ARF locus.8,9 The Polycomb group proteins(PcG) and the Trithorax group proteins (TrxG) have recentlyemerged as important regulators of INK4-ARF expression in cellularsenescence, with opposing roles: while TrxG proteins facilitatelocus activation, PcG proteins within the polycomb repressivecomplex (PRC)1 and PRC2 complexes maintain locus repressionunder normal proliferative conditions.10,11 During senescence, theH3K27 trimethyl (H3K27me3) repressive mark is lost, and PRC1 isdisplaced from the INK4-ARF locus, though the exact molecularmechanisms that regulate this change remain unknown.

Furthermore, the loss of certain PcG members can inducepremature senescence, due in part to derepression of the INK4-ARF locus.12–14 On the other hand, following senescence stimuli,the TrxG protein MLL1 and the histone demethylase JMJD3 arerecruited to the INK4-ARF locus, where they facilitate derepressionof the locus by depositing H3K4me3 activation marks andremoving H3K27me3 repressive marks, respectively.15–17 Thisfurther highlights the complexity of the epigenetic regulation ofthis locus. Epigenetic deregulation of the INK4-ARF locus is alsoknown to have a major impact on tumorigenesis.1 Silencing of thep16INK4a and p15INK4b genes by promoter hypermethylation is ahighly frequent and early event in numerous types of humancancer, including squamous cell carcinoma, leukemia andmyelodysplasia.18 Previous reports have shown that micedeficient for all three open reading frames of the INK4-ARF locusare highly tumor prone and develop predominantly skin tumorsand soft-tissue sarcomas.4

Recently, we characterized ZRF1 as a chromatin-associatedprotein that recognizes the H2A mono-ubiquitin mark at lysine119 (H2AK119ub) and displaces RING1B (PRC1) from chromatinduring differentiation of the human teratocarcinoma cell lineNTERA (NT2/D1).19,20 The gene expression profile of ZRF1-depleted NT2/D1 cells indicated that ZRF1 might directlyregulate the INK4-ARF locus. These results, together with the factthat PRC1 and PRC2 regulate this locus in NT2/D1 cells, promptedus to hypothesize that ZRF1 regulates gene expression from theINK4-ARF locus. We now report here that ZRF1 is a novel mediatorin the activation of the INK4-ARF locus, and that it functions bycompeting with Polycomb-mediated epigenetic silencing in thetumor suppressor mechanism of oncogene-induced senescence.

RESULTSZRF1 directly regulates the INK4-ARF locus in NT2/D1 cellsWe recently reported that ZRF1 is recruited to H2AK119ub, andthat it specifically displaces the PRC1 complex from chromatin

1Department of Differentiation and Cancer, Center for Genomic Regulation (CRG), and Universitat Pompeu Fabra (UPF), Barcelona, Spain and 2Institucio Catalana de Recerca iEstudis Avancats (ICREA), Barcelona, Spain. Correspondence: Dr L Di Croce, Department of Differentiation and Cancer, Center for Genomic Regulation (CRG), and UniversitatPompeu Fabra (UPF), 08003 Barcelona, SpainE-mail: [email protected] 18 September 2011; revised 14 March 2012; accepted 15 April 2012; published online 25 June 2012

Oncogene (2013) 32, 2161–2168& 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13

www.nature.com/onc

http://dx.doi.org/10.1038/onc.2012.241

mailto:[email protected]

http://www.nature.com/ONC

during retinoic acid (RA)-induced neuronal differentiation of theNT2/D1 cell line.19 This promotes the transcriptional activation ofPolycomb target genes during differentiation.19,21 Previouspublications indicated that gene expression from the INK4-ARFlocus is induced in differentiating NT2/D1 cells.22,23 Geneexpression profile analyses in NT2/D1 cells have shownp15INK4b to be a putative target gene regulated by ZRF1.19

Interestingly, Polycomb complexes in NT2/D1 cells control theINK4-ARF locus, as chromatin immunoprecipitation (ChIP)-on-chipanalysis indicated that the locus is a RING1B and H2AK119ub-target.19 Therefore, we asked whether ZRF1 could be directlyimplicated in regulating the locus in NT2/D1 cells. To address this,we generated stable NT2/D1 control and ZRF1-depleted cell linesand treated them with RA to induce differentiation. RA treatmentdid not alter the expression of ZRF1 at either the mRNA or theprotein level during the time points studied (Figure 1b andSupplementary Figure S1a). However, the RA-induced expressionof p15INK4b, ARF and p16INK4a was compromised in ZRF1-depleted cells (Figure 1a). To strengthen our observation, weperformed western blot analysis from total protein extracts inboth control and ZRF1-knockdown cells. Importantly, p15INK4b,ARF and p16INK4a protein levels were also compromised uponZRF1 depletion (Figure 1b). As a control, we also analyzed knownRA-responsive genes in both control and ZRF1-depleted NT2 cells(Supplementary Figure S1b).19,24 Expression of non-ZRF1 target

genes, such as MEIS2 and OAT, was not affected in ZRF1-depletedcells. On the other hand, the expression of ZRF1 target genesHOXA4 and HOXA5 was reduced (Supplementary Figure S1c).

We next asked whether ZRF1 might directly regulate thederepression of the INK4-ARF locus upon RA stimulus. TreatingNT2/D1 cells with RA led to ZRF1 recruitment to the p15INK4b, ARFand p16INK4a promoters (Figures 1c and d). We additionallyobserved RING1B and SUZ12 displacement, with a consequentloss of H3K27me3, from the ARF and p16INK4a promoters (Figures1c and d). These results indicate that, following RA treatment,ZRF1 directly regulates the expression of the INK4A-ARF locus byfacilitating the displacement of PRC1.

ZRF1/MIDA1 expression is induced by oncogenic H-RasThe INK4-ARF locus has been described as a key regulator of bothreplicative senescence and oncogene/stress-induced senes-cence.25,26 Therefore, our observation that ZRF1 directly regulatesthe INK4-ARF locus in NT2/D1 prompted us to hypothesize that ZRF1has a potential role not only in differentiation but also duringsenescence. Polycomb complexes bind and regulate the INK4-ARFlocus in proliferating human and mouse fibroblasts.13 Duringoncogene-induced senescence (OIS), both Polycomb proteins andthe repressive H3K27me3 mark must be removed from this locus inorder to reactivate its transcription.13,16

Figure 1. The INK4-ARF locus is regulated by ZRF1. (a) qRT–PCR analysis of the INK4-ARF locus in NT2/D1 cells with a ZRF1 knockdownfollowing administration of 1 mM of RA over 4 days. Values represent the mean±s.d. Expression data was normalized to PUM1 levels.(b) Western blot analysis of p15INK4b, p14ARF and p16INK4a protein levels in control and ZRF1-knockdown NT2/D1 cells, following RAadministration over 4 days. (c) Schematic representation of the INK4-ARF locus. Localization of primer pairs used in the study is shownschematically. (d) ChIP experiments performed with chromatin obtained from NT2/D1 cells either not induced (control, CTR) or induced with1mM of RA for 4 days (þ RA). Occupancy at the different regions across the locus was tested by qRT–PCR, with the CCND1 promoter used as anegative control region. Values represent the mean±s.d.

ZRF1 controls oncogene-induced senescenceJD Ribeiro et al

2162

Oncogene (2013) 2161 – 2168 & 2013 Macmillan Publishers Limited

We investigated the role of ZRF1 in activating the INK4-ARFlocus during OIS, using mouse embryonic fibroblasts (MEFs) thatexpress oncogenic H-Ras (hereafter referred to as Ras).25 As earlyas 2 days after puromycin selection (that is, Ras expression), themRNA transcripts of p15Ink4b, Arf and p16Ink4a were upregulated,as previously described (Figure 2a).25,27 Strikingly, the mouseortholog of ZRF1, MIDA1, which shares 93% identity with ZRF1,28

was upregulated at both the mRNA and protein levels upon Rasexpression (Figures 2b and c), coinciding with an accumulation ofp53 and p16Ink4a (Figure 2c). Interestingly, the increased MIDA1expression remained constant over time in senescent cells(Supplementary Figure 1d). Furthermore, as ZRF1 is known tolocalize to both the cytoplasm and the nuclei of mammaliancells,19 we performed subcellular fractionation of MEFs expressingRas. This revealed an accumulation of MIDA1 mainly in the nuclearfraction (Figure 2d), suggesting a role for MIDA1 predominantly ingene regulation during OIS.

To investigate whether the mechanism of ZRF1 in regulatingthe INK4-ARF locus in senescent cells is conserved betweenhuman and mouse, we next expressed Ras in primary humanfibroblasts (IMR90). Indeed, we also observed an increase in ZRF1upon Ras expression in IMR90 cells (Figure 2e) concomitant withan upregulation of the p15INK4b and p16INK4a transcripts

(Figure 2f). As previously reported for IMR90 cells, we did notdetect an upregulation of the ARF transcript29 (Figure 2f).Furthermore, ZRF1 upregulation was also detected by westernblot in IMR90 transduced cells with Ras (Figure 2g). We alsoobserved a decrease in the mRNA and protein levels of thePRC2-subunit EZH2 (Supplementary Figure 1e), as previouslydescribed.15

ZRF1/MIDA1 is recruited to the INK4-ARF locus duringH-Ras-induced OISWe next asked whether MIDA1 directly regulates the mouseInk4-Arf locus via a Polycomb-dependent mechanism. To elucidatewhether MIDA1 is recruited to the Ink4-Arf locus during itsreactivation upon OIS, we performed ChIP experiments using ananti-ZRF1 antibody in MEFs infected with either a control vector orone that expresses Ras. We found that MIDA1 was consistentlyrecruited to the locus following Ras overexpression, and that thiscorrelated with a decreased binding of Ring1B and reducedH3K27me3 levels (Figure 3a). Interestingly, the most pronouncedchanges regarding the recruitment of MIDA1 occurred at thepromoter region of p15INK4b and around the promoter and firstexons of Arf and p16Ink4a (Figure 3a). Oncogenic Ras expression isalso followed by RING1B displacement and by the reduction of therepressive mark H3K27me3 from Arf and p16Ink4a (Figure 3a).Additionally, ChIP analysis of ZRF1 in Ras-expressing IMR90 humancells indicated that ZRF1 was also recruited to the promoter andfirst exon regions of p15INK4b and p16INK4a (Figure 3b), whilechanges in RING1B, SUZ12 and H3K27me3 levels were observedonly at p16INK4a, suggesting a more complex regulatorymechanism of this locus in human as compared with mouse cells(Figure 3b). Together, these results indicate that ZRF1/MIDA1directly binds the INK4-ARF during oncogene-induced stress.

Ectopic expression of ZRF1/MIDA1 leads to activation of thep16INK4a and inhibits proliferationTo investigate whether ZRF1 has a functional role in regulating theINK4-ARF locus, we ectopically expressed either wild-type ZRF1 ora ZRF1 mutant (ZRF1DUBD), which cannot bind to H2AK119ub,19

in IMR90 cells. We analyzed whether ectopically expressed ZRF1was sufficient to induce derepression of the INK4-ARF locus, andwhether it required H2AK119ub. Overexpression of wild-type ZRF1led to a significant upregulation of the p15INK4b and p16INK4atranscripts but not of ARF (Figure 4a). Conversely, expression ofZRF1DUBD was not sufficient to derepress p16INK4a, whilep15INK4b expression was significantly lower when compared withwild-type ZRF1 (Figure 4a). We next investigated whetherupregulation of p15INK4b and p16INK4a upon ZRF1 overexpres-sion was sufficient to induce a senescence-like arrest. Interestingly,ectopic expression of wild-type ZRF1 led to a significantproliferation defect of IMR90 cells, as determined by reducedBromodeoxyuridine (BrdU) incorporation (Figure 4b) and anincreased activity of senescence-associated b-galactosidase(Figure 4c). However, in agreement with the gene expressiondata, we observed that overexpression of ZRF1DUBD led to partialinduction of senescence, as compared with wild-type ZRF1(Figures 4b and c), which could be explained though theincomplete derepression of p15INK4B, which was previouslyreported to trigger senescence when ectopically expressed inhuman fibroblasts.30 In proliferating IMR90 cells, RING1B is notdetected by ChIP analysis at p15INK4B regulatory regions.Nevertheless, ZRF1 is necessary to activate p15INK4B expression,apparently in a H2AK119ub-independent manner.

Strikingly, MEFs overexpressing wild-type ZRF1 also upregu-lated p16INK4a expression (Figure 4d), which was mirrored byincreased senescence-associated b-galactosidase positive cells(Figure 4e). These findings indicated that ZRF1/MIDA1 is necessaryto activate p16INK4A, both in human and mouse fibroblasts.

Figure 2. ZRF1/MIDA1 expression is induced by oncogenic H-Ras.(a–g) Cells were transduced either with an oncogenic H-Ras–encoding retrovirus (RAS) or a control vector (CTR). (a) p15Ink4b, Arfand p16Ink4a expression levels in MEFs, as determined by qRT–PCR.Values represent the mean±s.d. (b) Expression levels of MIDA1 inMEF cells, as determined by qRT–PCR. Expression data in (a) and(b) was normalized to Actin levels. Values represent means±s.d.(c) Western blot for the protein levels of MIDA1, p16Ink4a and p53 inMEFs transduced with oncogenic H-Ras–encoding retrovirus (RAS)or a CTR vector, 2 days after selection removal for cells over-expressing oncogenic H-Ras. (d) Western blot showing the distribu-tion of MIDA1 in the cytoplasmic (Cytop) and nuclear fraction(Nucleus) extracts from MEFs. (e) ZRF1 expression levels in IMR90cells, as determined by qRT–PCR. Values represent means±s.d.(f ) Expression levels of p15INK4b, ARF and p16INK4a for IMR90 cells,as determined by qRT–PCR. Expression data in (e) and (f ) wasnormalized to actin levels. Values represent means±s.d. (g) Proteinlevels of ZRF1 and p16INK4a in IMR90 cells.


2163

& 2013 Macmillan Publishers Limited Oncogene (2013) 2161 – 2168

Depletion of ZRF1/MIDA1 cooperates with oncogenic H-Ras forbypassing senescenceFrom the data presented above, we conclude that ZRF1/MIDA1contributes to the activation of the INK4-ARF locus. Expression ofoncogenic Ras triggers senescence, which depends on thefunctional activity of the INK4-ARF locus, while impairedexpression of this locus can lead to a bypass of Ras-inducedsenescence and may facilitate tumor development. To function-ally test whether a lack of ZRF1/MIDA1 would prevent OIS, wegenerated control and MIDA1-depleted MEFs with or without Rasoverexpression. In both cases, the MIDA1 knockdown wasefficient, as shown by quantitative real time PCR (qRT–PCR) fromtotal RNA (Figure 5a). Interestingly, depletion of MIDA1 in Ras-expressing cells reduced the induction of p16Ink4a, and led to aslight reduction in the levels of Arf transcript and protein (Figures5b and c). At the functional level, MIDA-depleted, Ras-expressingMEFs continued to proliferate, and formed colonies whenplated at low density (Figure 5d). We next tested this functionalrole for ZRF1/MIDA1 in another cell type, namely, primarymouse keratinocytes, and observed that also in these epithelialcells, knockdown of MIDA1 efficiently inhibited OIS induction(data not shown). This indicated that ZRF1/MIDA1 is required tostably activate the Ink4a-Arf locus during the induction ofsenescence.

Next, using human primary keratinocytes as a well-establishedmodel to study senescence bypass,31 we investigatedwhether ZRF1 is implicated in regulating senescence in humancells. Oncogenic mutations in H-Ras, as well as INK4a-ARFmissregulation, are frequent in skin lesions and contribute to the

onset of squamous cell carcinoma.32 Importantly, we alsoobserved a proliferative advantage in human keratinocytesdepleted for ZRF1 and OIS bypass in keratinocytes when thesewere co-infected with Ras and a short hairpin RNA (shRNA) againstZRF1 (Figure 6a). Senescence evasion in human keratinocytesdeveloped with reduced p16INK4a levels (Figure 6b), corroborat-ing a role for ZRF1 as an activator of p16INK4A in humankeratinocytes.

These results further reinforce the hypothesis that ZRF1/MIDA1associates with the INK4A-ARF locus upon OIS in primary cells andantagonizes PRC1-mediated silencing at the locus (Figure 6c).

DISCUSSIONRegulation of the INK4-ARF locus is a complex and highlycontrolled process that is central to normal physiology and tumorsuppression. As such, understanding how these genes areregulated is critical. We focused our study on ZRF1/MIDA1, arecently characterized epigenetic factor that is specificallyrecruited to chromatin that is decorated with mono-ubiquitinatedhistone H2A, and which thus competes with RING1B at the onsetof differentiation.

Here, we present evidence that ZRF1/MIDA1 regulates theactivation of the INK4-ARF locus not only during the differentiationprocess but also upon OIS. We demonstrate that ZRF1/MIDA1 isupregulated by OIS. However, it remains to be elucidated whetherthe activation of the INK4A-ARF locus by ZRF1 is solely dependenton the displacement of PRC1 from chromatin, or whether otheradditional mechanisms are required. We propose that ZRF1

Figure 3. ZRF1/MIDA1 associates with the INK4-ARF locus upon oncogenic Ras expression. (a) Schematic representation of the Ink4-Arf locus,showing the localization of the primer pairs used in the ChIPs performed in MEF cells. Actin b was used as a negative region for a control. ChIPexperiments used chromatin from MEF cells (a) or IMR90 cells (b) that had been transduced with a control (CTR) or an H-Ras–encodingretrovirus (RAS). Occupancy at the different regions across the locus was tested by qRT–PCR. Values represent means±s.d. The same regionswere tested in (b) are the same as depicted in Figure 1c.


2164


facilitates transcription at the INK4A-ARF locus by recruitingspecific deubiquitinases to chromatin, as previous in vitro experi-ments have suggested.19 In addition, ZRF1 might cooperate withor facilitate the activity of JMJD3 and MLL1, both of which arenecessary for INK4-ARF activation,15–17 through its stabilization atchromatin.

Interestingly, ZRF1 recruitment to p15INK4b in the NT2/D1 cellline and fibroblasts apparently is independent of H2AK119ub. Wespeculate that specific transcription factors could recruit ZRF1 top15INK4b during differentiation or OIS.

Distinct post-translation modifications are likely to be essentialfor regulating the levels of epigenetic modifiers at the INK4-ARFlocus in proliferating and senescent cells. For instance, theubiquitin-specific proteases USP7 and USP11 regulate ubiquitina-tion of the posterior sex combs (PSC) and sex combs extra (Sce)Polycomb proteins; this consequently controls the levels of PRC1at the INK4-ARF locus.33 As ZRF1 is subject to phosphorylationduring mitosis or following RA stimulation,34,35 it will beinteresting to determine in the future whether post-translationalmodifications of ZRF1 affect its recruitment to the INK4-ARF locus,or its ability to displace PRC1 and PRC2 from the locus.

Our study raises the possibility that ZRF1 is implicated in aging,as p16INK4a upregulation and other senescence features areassociated with aging. Intriguingly, a previous study has identifiedZRF1/MIDA1 as part of the aging signature of the skin of agedmice.36 Both senescence and aging correlate with loss of theregenerative capacity of certain adult stem cells. Understandinghow the epigenetic role of ZRF1 in activating the INK4-ARF locus isassociated with the aging of adult stem cells will be a focus offuture research.

We demonstrate that depletion of MIDA1 in MEFs and primarykeratinocytes that express oncogenic H-Ras impairs the activationof p16Ink4a and Arf, contributing to senescence bypass. Cellularsenescence is a block to tumor progression and an early event intumorigenesis.7 Interestingly, analysis of the Oncomine database(www.oncomine.org) points to a reduced expression of ZRF1 inbreast cancer and chronic lymphocytic leukemia. Loss ofheterozygosity at ZRF1 has been reported in several cancer celllines, according to the CONAN (copy number analysis) database(http://www.sanger.ac.uk/cgi-bin/genetics/CGP/conan/search.cgi).In contrast, other studies have suggested that ZRF1 might have anoncogenic role in head and neck squamous cancer (HNSCC).37

However, most of these studies are more descriptive in nature andwere performed on established cell lines in which importantpathways in cellular senescence and tumorigenesis are abrogated.Based on our observations, functional studies are needed toaddress whether increased levels of ZRF1 are linked to benign andpre-malignant lesions in tissues in which senescence occurs, andwhether progression towards malignancy is associated with a lossof ZRF1. Overall, our data support a possible role of ZRF1 as atumor suppressor that is critical for establishing a response to OIS,and suggests that derailing this response might be involved intumorigenesis.

MATERIALS AND METHODSAntibodies used for ChIP and western blot analysisThe following antibodies were used: anti-tubulin (Abcam 7291, Abcam,Cambridge, UK), anti-p15 (Santa Cruz k-18, Santa Cruz Biotechnology, Inc.,Santa Cruz, CA, USA), anti-p16 (Santa Cruz C-20 and M-156), anti-RING1B

Figure 4. Overexpression of ZRF1 leads to increased expression of p16INK4a in IMR90 and MEFs. (a) IMR90 cells were infected with retroviralvectors expressing human ZRF1 (ZRF1), ZRF1 mutated in its ubiquitin-binding domain (ZRF1DUBD), or a control vector (CTR). Expression levelsof ZRF1, p15INK4b, ARF and p16INK4a were determined by qRT–PCR. Expression data was normalized to ACTIN. ZRF1 (note that the ZRF1mutant has a lower molecular weight) and p16INK4a protein levels were determined by western blot. Values represent mean±s.d. (b) IMR90cells overexpressing ZRF1, ZRF1DUBD, or a control vector were assayed for proliferation by measuring the percentage of BrdU-positive cells.Values represent mean±s.d. (c) IMR90 cells overexpressing ZRF1, ZRF1DUBD, or a control were assayed for SA-b-galactosidase activity. Valuesrepresent means±s.d (n¼ 6 fields). (d) MIDA1 and p16Ink4a protein levels were determined by western blot, and p16Ink4a mRNA levels weredetermined by qRT–PCR, in MEFs infected with retroviral vectors expressing human ZRF1 (ZRF1) or a CTR vector. Expression data wasnormalized to actin. Values represent the means±s.d. (e) MEF cells overexpressing ZRF1 or a control vector were assayed for SA-b-gal. Valuesrepresent the means±s.d. (n¼ 6 fields).


2165


www.oncomine.org

http://www.sanger.ac.uk/cgi-bin/genetics/CGP/conan/search.cgi

(Di Croce laboratory, Barcelona, Spain), anti-ZRF1 (Di Croce laboratory),anti-p53 (Novocrasta–Leica Microsystems, Newcastle Upon Tyne, UK),anti-p19ARF (Santa Cruz no.1063), anti-H-Ras (Santa Cruz C-20), anti-H3(Abcam 1791), anti-EZH2 (Cell Signaling 3147, Cell Signaling Technology,Inc., Danvers, MA, USA) and anti-SUZ12 (Abcam 12073).

Cell culture, transfections and viral infectionsMEFs were prepared from embryonic day 13.5 embryos of the C57BL/6inbred strain. MEFs, IMR90, NTERA (NT2-D1), HEK293T and PhoenixEcotropic/Ampho were cultured with Dulbecco’s modified Eagle’s medium(Gibco/Invitrogen, Life Technologies Ltd., Paisley, UK) supplemented with10% fetal bovine serum at 37 1C with 5% CO2.

In all, 2� 106 HEK293T cells were transfected with 7 mg of pLKO(mammalian expression lentiviral RNAi vector pLKO)-shRNA (Sigma-Aldrich,St Louis, MO, USA), 5 mg of pCMV-VSV-G (envelope vector for pseudotypingof Moloney Murine Leukemia virus (MMLV)-based retroviral vectors), and6mg of pCMVDR-8.91 plasmids to produce lentivirus containing LKO-shRNA. After 72 h, NT2/D1 cells were infected overnight with the virus-containing media filtered with 5 mg per ml of polybrene (Sigma). Cells wereselected 24 h after infection with 2.5mg per ml of puromycin (Sigma) for 2days. The pLKO sequence is shown in Supplementary Table S1(Supplementary Table S1).

Phoenix Ecotropic/Ampho packaging cell line cells were transfected withthe retroviral vectors pBabe-H-RASG12V, pMSCV (murine stem cell virusvector) Empty, pMSCV-ZRF1, pMSCV-ZRF1DUBD or pRETROSUPER (NKI-AVL, Amsterdam, Netherlands) containing the shRNA against MIDA1(Supplementary Table S1), pp16Ink4a and p5338 to produce retroviruscontaining pRS (mammalian expression retroviral RNAi pRETROSUPERvector)-shRNA. Virus-containing media was collected 48 h aftertransfection, and MEFs were infected with 5mg/ml polybrene. Cells were

selected 24 h after infection with 2.5 mg per ml of puromycin (Sigma) and100mg per ml of hygromycin (Sigma).

Primary human keratinocytes were isolated from adult foreskin andcultured together with a feeder layer of fibroblasts (J2-3T3) as describedpreviously.39 Keratinocytes were infected with pRETROSUPER containingthe shRNA against ZRF1 (Supplementary Table S1), using PhoenixA cells.Keratinocytes were selected for 2 days with 2 mg/ml puromycin and100mg/ml hygromycin. Infected keratinocytes were grown for 5 days afterthe cycle of infection, selection and reseeding for experimentation.

Plasmids and CloningFull-length ZRF1 and DUBD mutant were subcloned from pCMV2ZRF1 andpCMV2ZRF1DUBD19 into the pMSCV vector by digestion with the BamHIand XhoI restriction enzymes (Fermentas GmbH, St Leon-Rot, Germany)and used to produce stable cell lines by retroviral infection of MEF andIMR90 cells.

qRT–PCR analysisRNA was extracted with the RNA extraction kit (Qiagen, Hilden, Germany).complementary DNA (cDNA) was generated from 0.5 mg of RNA with theFirst Strand cDNA Synthesis Kit (Fermentas). cDNA was diluted to 100mlwith water, and 2 ml samples were used for each RT–qPCR reaction, usingSYBR green (Roche, Basel, Switzerland). The primers used in the RT–qPCRassays are listed in Supplementary Table S1.

ChIP assaysNT2/D1, MEFs and IMR90 cells were trypsinized and crosslinked in 1%formaldehyde for 10 min at room temperature. Crosslinking was quenchedwith a 0.125 M glycine solution for 5 min in formaldehyde, and cells werewashed twice with 1� PBS. Pelleted cells were lysed in 1 ml ChIP buffer(1 volume of sodium dodecyl sulfur buffer with 0.5 volumes of Tritonbuffer), and sonicated for 8 min in a Bioruptor (Diagenode, Liege, Belgium).Soluble material was quantified by Bradford assays. ZRF1 was immuno-precipitated from 1000mg of protein and RING1B from 500mg, and 100mgof immunoprecipitated histone/histone modifications were used. Anti-bodies were incubated overnight with the chromatin in a 500ml volume.Immunocomplexes were recovered with 30ml of a protein A bead slurry.Immunoprecipitated material was washed three times with a low-saltbuffer and once with a high-salt buffer. DNA complexes were de-crosslinked in 100ml decrosslink buffer (1% sodium dodecyl sulfur and100 mM NaHCO3) at 65 1C for 3 h, and DNA was then eluted in 100ml ofwater (Braun wate, Braun Medical S.L, Barcelona, Spain) using a PCRpurification kit (Qiagen). DNA (2ml) was used for each qPCR reaction withSYBR green (Roche). The primers and antibodies used are given inSupplementary Table S1.

BrdU incorporation assayIMR90 cells were treated with 10 mM of BrdU solution for 1 h and thenanalyzed for BrdU incorporation by flow cytometry, using the APC BrdUFlow Kit (BD Pharmigen, Franklin Lakes, NJ, USA) according to themanufacture’s protocol.

Senescence-associated-b-galactosidase assayIMR90 and MEFs were fixed with 0.5% glutaraldehyde (Sigma) in PBS andstained for senescence-associated b-galactosidase activity as previouslydescribed.40 Pictures were acquired using a Leica CTR6000 microscope(Leica Microsystems Wetzlar GmbH, Wetzlar, Germany) at the totalmagnification of � 100.

Western blot analysis and nuclear fractionationCell extracts for western blot analysis were prepared in lysis buffer (50 mM

Hepes, pH 7.5, 150 mM NaCl, 1 mM ethylenediaminetetraacetic acid (EDTA),2.5 mM ethylene glycol tetraacetic acid (EGTA), 0.1% Tween 20 andproteinase inhibitors), sonicated 15 s in a Brason Sonifier (BransonUltrasonics, Danbury, CT, USA), and centrifugated 20 min at maximumspeed at 4 1C. Protein suppernatant was diluted in 4� Laemmli buffer andanalyzed with sodium dodecyl sulfur–polyacrylamide gel electrophoresis.

Nuclear extracts for western blots were prepared in hypotonic buffer(5 mM PIPES (piperazine-N,N0-bis(2-ethanesulfonic acid) buffering agent)pH 8, 85 mM KCL, 0.5% NP-40 and proteinase inhibitors), kept on ice for10 min, and centrifugated 5 min at 5000 g. Supernatants were collected

Figure 5. ZRF1/MIDA1 depletion contributes to neoplastic growth inMEF cells. (a) mRNA was isolated from MEF cells with or withoutoverexpressed oncogenic H-Ras, and transduced with a shorthairpin retrovirus control or short hairpin MIDA1, and analyzed byqRT–PCR for MIDA1 transcript levels. Expression data was normal-ized to Actin. Values represent the means±s.d. (b) mRNA isolatedfrom MEF cells as in (a) were analyzed by qRT–PCR for Arf andp16Ink4a transcript levels. Values represent means±s.d. (c) Proteinlevels of MIDA1, p16Ink4a and Arf in MEF cells treated as in (a).(d) Colony formation assays were performed by plating cells at theindicated numbers and staining with crystal violet.


2166


and diluted in 4� Laemmli buffer. Nuclear pellets were resuspendend inlysis buffer, sonicated 15 s in a Brason Sonifier, and microcentrifugated20 min at maximum speed at 4 1C. Protein supernatant was diluted in4� Laemmli buffer, and all fractions were analyzed by sodium dodecylsulfur–polyacrylamide gel electrophoresis.

Colony formation assayMEFs were plated at 2500, 5000 or 10 000 cells per well in duplicate during10 days. Cells were fed every 2 days for 10 days. After 10 days, cells werewashed with 1� PBS, fixed with 4% formalin (Sigma) for 15 min, rinsedwith 1� PBS, and stained with crystal violet solution (1% in 1� PBS) for20 min. Cells were then washed 4�with H2O and allowed to air dry.

ACKNOWLEDGEMENTSWe thank VA Raker for help in preparing the manuscript, and to the CRG GenomicUnit. This work was supported by grants from the Spanish ‘Ministerio de Educacion yCiencia’ (CONSOLIDER and BFU2010-18692), from the European Commission FP7project 4DCellFate (277899), and from AICR (10-0177) to LDC. JDR was supported bygrant SFRH/BD/15908/2005 from Foundation for Science and Technology (FCT)Portugal and is a fellow of the Graduate Program in Areas of Basic and AppliedBiology (GABBA), University of Oporto, Portugal; LM was supported by a post-doctoralCRG-Novartis fellowship.

REFERENCES1 Kim WY, Sharpless NE. The regulation of INK4/ARF in cancer and aging. Cell 2006;

127: 265–275.2 Lowe SW, Sherr CJ. Tumor suppression by Ink4a-Arf: progress and puzzles. Curr

Opin Genet Dev 2003; 13: 77–83.3 Collado M, Serrano M. The power and the promise of oncogene-induced senes-

cence markers. Nat Rev Cancer 2006; 6: 472–476.4 Krimpenfort P, Ijpenberg A, Song JY, van der Valk M, Nawijn M, Zevenhoven J

et al. p15Ink4b is a critical tumour suppressor in the absence of p16Ink4a. Nature2007; 448: 943–946.

5 Kuilman T, Michaloglou C, Mooi WJ, Peeper DS. The essence of senescence. GenesDev 2010; 24: 2463–2479.

6 Adams PD. Healing and hurting: molecular mechanisms, functions, and pathol-ogies of cellular senescence. Mol Cell 2009; 36: 2–14.

7 Collado M, Serrano M. Senescence in tumours: evidence from mice and humans.Nat Rev Cancer 2010; 10: 51–57.

8 Gil J, Peters G. Regulation of the INK4b-ARF-INK4a tumour suppressor locus: all forone or one for all. Nat Rev Mol Cell Biol 2006; 7: 667–677.

9 Cruickshanks HA, Adams PD. Chromatin: a molecular interface between cancerand aging. Curr Opin Genet Dev 2011; 21: 100–106.

10 Maertens GN, El Messaoudi-Aubert S, Racek T, Stock JK, Nicholls J, Rodriguez-Niedenfuhr M et al. Several distinct polycomb complexes regulate and co-localizeon the INK4a tumor suppressor locus. PLoS One 2009; 4: e6380.

11 Kia SK, Gorski MM, Giannakopoulos S, Verrijzer CP. SWI/SNF mediates polycombeviction and epigenetic reprogramming of the INK4b-ARF-INK4a locus. Mol CellBiol 2008; 28: 3457–3464.

12 Gil J, Bernard D, Martinez D, Beach D. Polycomb CBX7 has a unifying role incellular lifespan. Nat Cell Biol 2004; 6: 67–72.

13 Bracken AP, Kleine-Kohlbrecher D, Dietrich N, Pasini D, Gargiulo G, Beekman C etal. The Polycomb group proteins bind throughout the INK4A-ARF locus and aredisassociated in senescent cells. Genes Dev 2007; 21: 525–530.

14 Dietrich N, Bracken AP, Trinh E, Schjerling CK, Koseki H, Rappsilber J et al. Bypassof senescence by the polycomb group protein CBX8 through direct binding tothe INK4A-ARF locus. EMBO J 2007; 26: 1637–1648.

15 Agger K, Cloos PA, Rudkjaer L, Williams K, Andersen G, Christensen J et al. TheH3K27me3 demethylase JMJD3 contributes to the activation of the INK4A-ARFlocus in response to oncogene- and stress-induced senescence. Genes Dev 2009;23: 1171–1176.

16 Barradas M, Anderton E, Acosta JC, Li S, Banito A, Rodriguez-Niedenfuhr M et al.Histone demethylase JMJD3 contributes to epigenetic control of INK4a/ARF byoncogenic RAS. Genes Dev 2009; 23: 1177–1182.

17 Kotake Y, Zeng Y, Xiong Y. DDB1-CUL4 and MLL1 mediate oncogene-inducedp16INK4a activation. Cancer Res 2009; 69: 1809–1814.

18 Simboeck E, Ribeiro JD, Teichmann S, Di Croce L. Epigenetics and senescence:Learning from the INK4-ARF locus. Biochem Pharmacol 2011; 82: 1361–1370.

Figure 6. ZRF1 depletion leads to OIS bypass in human keratinocytes. (a) Colony formation of ZRF1-deficient cells (shZRF1) or ZRF1-deficientcells co-infected with oncogenic H-Ras (shZRF1þ RAS) in human epidermal keratinocytes, as compared with the control vector (CTR orCTRþ RAS, respectively). One-thousand cells were seeded in clonal density per well in triplicate and, after 10 days, stained with crystal violet.Pictures are representative of two independent infections of human keratinocytes. Scale bar: 400 mm. (b) Protein levels of ZRF1, Ras, p16INK4aand tubulin of human keratinocytes collected 10 days after infection. (c) Schematic representation of ZRF1’s epigenetic regulation of theINK4A-ARF locus upon Ras OIS. In proliferation conditions, PRC2 specifically trimethylates H3K27 and PRC1 catalyzes H2AK119ub at the INK4A-ARF locus, thereby maintaining it transcriptionally silent. Ras OIS triggers ZRF1 recruitment to the INK4A-ARF locus, displacing PRC1 complexesby interacting with H2AK119ub. After PRC1 and PRC2 removal, the locus is transcriptionally actived: MLL1 specifically trimethylates H3K4, andRNAPII is recruited to the locus.


2167


19 Richly H, Rocha-Viegas L, Ribeiro JD, Demajo S, Gundem G, Lopez-Bigas N et al.Transcriptional activation of polycomb-repressed genes by ZRF1. Nature 2010;468: 1124–1128.

20 Richly H, Di Croce L. The flip side of the coin: role of ZRF1 andhistone H2A ubiquitination in transcriptional activation. Cell Cycle 2011; 10:745–750.

21 Andrews PW. Retinoic acid induces neuronal differentiation of a cloned humanembryonal carcinoma cell line in vitro. Dev Biol 1984; 103: 285–293.

22 Lois AF, Cooper LT, Geng Y, Nobori T, Carson D. Expression of the p16 and p15cyclin-dependent kinase inhibitors in lymphocyte activation and neuronal dif-ferentiation. Cancer Res 55: 4010–4013 1995.

23 Thullberg M, Bartkova J, Khan S, Hansen K, Ronnstrand L, Lukas J et al. Distinctversus redundant properties among members of the INK4 family of cyclin-dependent kinase inhibitors. FEBS Lett 2000; 470: 161–166.

24 Freemantle SJ, Kerley JS, Olsen SL, Gross RH, Spinella MJ. Developmentally-related candidate retinoic acid target genes regulated early during neuronaldifferentiation of human embryonal carcinoma. Oncogene 2002; 21: 2880–2889.

25 Serrano M, Lin AW, McCurrach ME, Beach D, Lowe SW. Oncogenic ras provokespremature cell senescence associated with accumulation of p53 and p16INK4a.Cell 1997; 88: 593–602.

26 Zindy F, Quelle DE, Roussel MF, Sherr CJ. Expression of the p16INK4a tumorsuppressor versus other INK4 family members during mouse development andaging. Oncogene 1997; 15: 203–211.

27 Malumbres M, Perez De Castro I, Hernandez MI, Jimenez M, Corral T, Pellicer A.Cellular response to oncogenic ras involves induction of the Cdk4 and Cdk6inhibitor p15(INK4b). Mol Cell Biol 2000; 20: 2915–2925.

28 Shoji W, Inoue T, Yamamoto T, Obinata M. MIDA1, a protein associated with Id,regulates cell growth. J Biol Chem 1995; 270: 24818–24825.

29 Ferbeyre G, de Stanchina E, Querido E, Baptiste N, Prives C, Lowe SW. PML isinduced by oncogenic ras and promotes premature senescence. Genes Dev 2000;14: 2015–2027.

30 McConnell BB, Starborg M, Brookes S, Peters G. Inhibitors of cyclin-dependentkinases induce features of replicative senescence in early passage human diploidfibroblasts. Curr Biol 1998; 8: 351–354.

31 Lazarov M, Kubo Y, Cai U, Dajee M, Tarutani M, Lin Q et al. CDK4 coexpression withRas generates malignant human epidermal tumorigenesis. Nature Medicine 2002;8: 1105–1114.

32 Kemp CJ. Multistep skin cancer in mice as a model to study the evolution ofcancer cells. Semin Cancer Biol 2005; 15: 460–473.

33 Maertens GN, El Messaoudi-Aubert S, Elderkin S, Hiom K, Peters G. Ubiquitin-specific proteases 7 and 11 modulate Polycomb regulation of the INK4a tumoursuppressor. EMBO J 2010; 29: 2553–2565.

34 Matsumoto-Taniura N, Pirollet F, Monroe R, Gerace L, Westendorf JM. Identifica-tion of novel M phase phosphoproteins by expression cloning. Mol Biol Cell 1996;7: 1455–1469.

35 Smith JC, Duchesne MA, Tozzi P, Ethier M, Figeys D. A differential phosphopro-teomic analysis of retinoic acid-treated P19 cells. J Proteome Res 2007; 6: 3174–3186.

36 Adler AS, Sinha S, Kawahara TL, Zhang JY, Segal E, Chang HY. Motif module mapreveals enforcement of aging by continual NF-kappaB activity. Genes Dev 2007;21: 3244–3257.

37 Resto VA, Caballero OL, Buta MR, Westra WH, Wu L, Westendorf JM et al. Aputative oncogenic role for MPP11 in head and neck squamous cell cancer.Cancer Res 2000; 60: 5529–5535.

38 Keyes WM, Wu Y, Vogel H, Guo X, Lowe SW, Mills AA. p63 deficiency activates aprogram of cellular senescence and leads to accelerated aging. Genes Dev 2005;19: 1986–1999.

39 Gandarillas A, Watt FM. c-Myc promotes differentiation of human epidermal stemcells. Genes Dev 1997; 11: 2869–2882.

40 Dimri GP, Lee X, Basile G, Acosta M, Scott G, Roskelley C et al. A biomarker thatidentifies senescent human cells in culture and in aging skin in vivo. Proc NatlAcad Sci USA 1995; 92: 9363–9367.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)


2168


http://www.nature.com/onc

Documents

ZRF1 controls oncogene-induced senescence through the INK4-ARF locus