Myc represses differentiation-induced p21CIP1 expression viaMiz-1-dependent interaction with the p21 core promoter

Siqin Wu1, Cihan Cetinkaya1, Maria J Munoz-Alonso2, Natalie von der Lehr1, Fuad Bahram1,Vincent Beuger3, Martin Eilers3, Javier Leon2 and Lars-Gunnar Larsson*,1,4

1Department of Plant Biology, Swedish University of Agricultural Sciences, Uppsala, Sweden; 2Departamento de Biologia Molecularand Unidad Asociada-CSIC, Universidad de Cantabria, Santander, Spain; 3Institute for Molecular Biology and Tumour Research,Marburg, Germany

Inhibition of cellular differentiation is one of the well-known biological activities of c-Myc-family proteins. Weshow here that Myc represses differentiation-inducedexpression of the cyclin-dependent kinase (CDK) inhibitorp21CIP1 (CDKN1A, p21), known to play an importantrole in cell fate decisions during growth and differentia-tion, in hematopoietic cells. Our results demonstrate thatthe c-Myc-responsive region is situated in the p21 corepromoter. c-Myc binds to this region in vitro and in vivothrough interaction with the initiator-binding Zn-fingertranscription factor Miz-1, which associates directly withthe promoter. Association of Myc with the promoter invivo correlates inversely with p21 expression. Usingmutants of c-Myc with impaired binding to Miz-1, ourresults further show that repression of p21 promoter/reporters as well as the endogenous p21 gene by Mycdepends on interaction with Miz-1. Expression of Miz-1increases during hematopoietic differentiation andMiz-1 activates the p21 promoter under conditions oflow Myc levels, indicating a positive role for free Miz-1 inthis process. In conclusion, repression of differentia-tion-induced p21 expression through Miz-1 may bean important mechanism by which Myc blocks diffe-rentiation.Oncogene (2003) 22, 351–360. doi:10.1038/sj.onc.1206145

Keywords: Myc; Miz-1; p21Cip1; differentiation; cellcycle; transcription

Introduction

Proto-oncogenes of the MYC family encode transcrip-tion factors that play an important role in the control ofcellular growth, differentiation and apoptosis. Deregu-lated expression of MYC genes as a result of chromo-somal rearrangements contributes to the developmentof a variety of cancers. Two regions of c-Myc areabsolutely required for these biological effects: (i) the

C-terminal basic region, helix–loop–helix and leucinezipper domains (bHLHZip), involved in specific DNA-binding to E-box recognition sequences via interactionwith the obligatory partner Max; and (ii) the N-terminaltransactivation domain (TAD) including the evolution-ary conserved Myc Box 1 and Myc Box 2 (MB1 andMB2) (for a review, see Grandori et al., 2000). MB2associates with the cofactor TRRAP, thereby recruitingSAGA-like complexes containing the histone acetyltransferase (HAT) activity to target promoters (McMa-hon et al., 1998, 2000; Bouchard et al., 2001; Frank et al.,2001; for a review, see Amati et al., 2001). These play arole in modulating chromatin structure during geneactivation.An increasing number of target genes activated by

c-Myc, involved in cell growth, metabolism andproliferation, have been identified (for a review, seeLevens, 2002). Activation of target genes throughE-boxes does, however, not account for all biologicalactivities of Myc (James and Eisenman, 2002). c-Mycalso represses transcription of certain genes of relevancefor its biological effects, including the CDK inhibitorsp15INK4B, p21CIP1 and p27KIP1, as well as genesinvolved in cellular differentiation and metabolism (Liet al., 1994; Mitchell and El-Deiry, 1999; Warner et al.,1999; Wu et al., 1999; Claassen and Hann, 2000; Colleret al., 2000; O’Hagan et al., 2000; Gartel et al., 2001;Yang et al., 2001). The mechanism for repression byMyc is less clear, but the repression of p15INK4B byMyc has recently been shown to involve direct interac-tion with the Zn-finger transcription factor Miz-1, whichbinds to the initiator (Inr) sequence of the p15 promoter(Seoane et al., 2001; Staller et al., 2001).Downregulation of c-Myc is an essential part of the

antiproliferative response to differentiation signals (for areview, see Grandori et al., 2000). We have shownpreviously that constitutive expression of v-Myc blocksinduced differentiation and growth arrest of human U-937 monoblasts (Larsson et al., 1988; Bahram et al.,1999). Since the rapid induction of the CDK inhibitorp21Cip1, here referred to as p21, by differentiation signalsseems to play an important decisive role for thedifferentiation process in a number of cell types,including U-937 cells (Jiang et al., 1994; Steinmanet al., 1994; Liu et al., 1996; Missero et al., 1996; Zhanget al., 1999; Cheng et al., 2000; for reviews, see Sherr and

Received 27 August 2002; revised 10 October 2002; accepted 16October 2002

*Correspondence: Lars-Gunnar Larsson, Uppsala Genetic Center,Department of Plant Biology, Swedish University of AgriculturalSciences, Box 7080, Uppsala 750 07, Sweden.E-mail: lars-gunnar.larsson@vbiol.slu.se

Oncogene (2003) 22, 351–360& 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00

www.nature.com/onc

Roberts, 1999; Steinman, 2002), we hypothesized thatinterference with induced p21 expression may be onemechanism by which v-Myc blocks cellular differentia-tion. p21 belongs to the CIP/KIP family of CDKinhibitors that regulates several types of cyclin/CDKcomplexes and plays an important role in cell cyclearrest, differentiation, DNA repair, cell senescence andapoptosis (for a review, see Sherr and Roberts, 1999).Induction of the p21 gene expression during differentia-tion- or growth inhibition is regulated by severaltranscription factors of importance for these processes,such as p53, VDR, RAR, C/EBP, STATs, Smads, Sp1,Hox10 among others (for a review, see Gartel andTyner, 1999). Here we show that Myc repressesdifferentiation-induced p21 expression via Miz-1-depen-dent interaction with the p21 core promoter.

Results

Differentiation-induced p21 expression is inhibited in v-Myc-transformed U-937 human monoblasts

U-937 monoblasts were induced to undergo terminalmonocytic differentiation by the addition of phorbolester (TPA), as exemplified by increased expression ofthe monocytic differentiation marker CD11c and the

reduced incorporation of 3H-labelled thymidine(Figure 1a). This process was completely blocked inU-937-myc-6 cells constitutively expressing v-myc, con-firming previous results (Larsson et al., 1988). As anapproach to elucidate the mechanism behind this block,we investigated the expression of the CDK inhibitorp21, known to play a decisive role during differentiation(Jiang et al., 1994; Steinman et al., 1994; Liu et al.,1996; Missero et al., 1996; Zhang et al., 1999; Chenget al., 2000; for reviews, see Sherr and Roberts,1999; Steinman, 2002). p21 mRNA was induced within2 h of TPA treatment in U-937-1 cells, whereas onlytraces of p21 transcripts were observed in U-937-myc-6cells (Figure 1b). p21 protein synthesis increasedgradually in U-937-1 cells, whereas only a faint p21band appeared after 4 h and then faded in TPA-stimulated U-937-myc-6 cells (Figure 1c). Exogenousv-Myc expression remained unaltered in U-937-myc-6cells, as previously reported (Larsson et al., 1988;Bahram et al., 1999).

c-Myc represses basal and TPA-induced transcriptionfrom the human p21 core promoter

These results suggested that v-Myc directly or indirectlyrepressed TPA-induced expression of p21. Transient

Figure 1 Myc represses basal and differentiation-induced expression of p21 in U-937 monoblasts. (a) Differentiation (CD11cexpression) and growth (3H-TdR incorporation) of U-937-1 and v-Myc-expressing U-937-myc-6 cells in response to TPA. (b) Northernblot analysis of TPA-induced p21mRNA expression. gapdh was used as a reference gene. (c) Immunoprecipitation of 35S in vivo-labeledp21, c- and v-Myc proteins in response to TPA. (d and e) Analysis of basal and TPA-induced p21 promoter activity in U-937-1 cells bytransient cotransfection assays using the p21 promoter/luciferase reporter 0-Luc and the CMV-Myc expression vector. (d) Titration ofCMV-Myc concentrations. (e) Cotransfection of c-Myc with the indicated promoter/luciferase constructs. The data presented in (d)and (e) are based on at least three separate experiments

Myc represses p21CIP1 expression via Miz-1S Wu et al

352

Oncogene

transfection experiments using a p21 promoter/lucifer-ase reporter gene construct in U-937 cells showed thatexpression of c-Myc repressed both basal and TPA-induced p21 promoter activity approximately two fold,but did not affect the activity of two control promoterconstructs, CMV-Luc and pUHC13.3 (Figure 1d and e),suggesting that c-Myc affects p21 expression at least inpart at the level of transcription.To map the Myc-responsive region of the p21

promoter, a series of p21 promoter deletion mutants(depicted in Figure 2a) were utilized. Deletions fromnucleotide positions �2326 bp to �94 relative to thetranscriptional start site did not abolish the relativerepression by c-Myc or activation by TPA (Figure 2b),suggesting that the c-Myc response region localizedbetween positions �94 to +16. To further map the c-Myc-responsive region, we generated two exchangemutants, which replace either nucleotides �94/�50 ofthe p21 promoter (generating CMV/p21-Luc) or nucleo-tides �49/+16 (generating p21/CMV-Luc) with thecorresponding regions of the c-Myc-nonresponsiveCMV IE promoter (Figure 2a). The upper regioncontains four Sp1/3 binding sites, some of which arerequired for p21 induction by multiple signals (Garteland Tyner, 1999). Figure 2c shows that c-Myc repressedthe activity of the CMV/p21-Luc construct, but not thep21/CMV-Luc construct (Figure 2c). These resultsindicated that the c-Myc responsive region is situatedbetween -49 and +16 of the p21 core promoter.

The c-Myc HLHZip domain and Myc Box 2 are requiredfor p21 repression

To map domains of c-Myc required for repression, apanel of c-Myc deletion mutants was used (Figure 2d).Figure 2e shows that deletions of the conserved HLH,Zip or MB2 domains eliminated the transcriptionalrepression of the 10-Luc reporter, while deletion of MB1had a partial effect. In contrast, deletion of the BR didnot affect transcriptional repression, indicating thatdirect binding of c-Myc to DNA is not required.

The initiator-binding transcription factor Miz-1 regulatesthe p21 core promoter

The HLH domain of c-Myc interacts with the Inr-binding transcription factor Miz-1 (Peukert et al., 1997),recently shown to be an activator of the p15INK4bpromoter and a target for c-Myc-induced repression(Peukert et al., 1997; Seoane et al., 2001; Staller et al.,2001). To determine if Miz-1 is able to activate the p21promoter, a Miz-1 vector was cotransfected with p21reporters into U-937 cells. Miz-1 activated 10-Luc andCMV/p21 approximately five- and eight-fold, in un-treated and TPA-treated cells, respectively, but had onlylittle effect on p21/CMV-Luc (Figure 3a), suggestingthat Miz-1 activated p21 through the c-Myc-responsivecore promoter region. Three potential Miz-1 bindingsequences with similarities to Miz-1 binding sites inother promoters (Peukert et al., 1997; Seoane et al.,2001; Staller et al., 2001) were identified within this

region (Figure 2a). Substitution of two nucleotideswithin the potential Miz-1 binding sequence at thesethree locations, creating CMV/p21mut-Luc (Figure 2a),abolished most of the Miz-1 response (Figure 3a).Figure 3b shows that c-Myc was able to repress Miz-1-induced transactivation of the 10-Luc and CMV/p21-Luc reporter constructs in a dose-dependent manner inuntreated and TPA-treated cells. These results suggestthat Miz-1 and c-Myc regulate the p21 promoter inopposite ways through the same core promoter region.

Miz-1 expression increases during myeloid differentiation

To address whether Miz-1 might be involved indifferentiation processes, Miz-1 expression was investi-gated during myeloid differentiation. Miz-1 mRNAexpression was upregulated transiently during TPA-induced monocytic/megakaryocytic differentiation ofK562 (Figure 3c), after TPA-induced monocytic differ-entiation HL-60 and U-937 or RA-induced granulocyticdifferentiation of HL-60 cells, while c-Myc was down-regulated (Figure 3d). Expression of multiple forms ofMiz-1 proteins increased gradually during TPA-induceddifferentiation of HL-60 cells, with maximum expressionat late time points (Figure 3e). Similar results wereobtained during induced megakaryocytic and erythro-cytic differentiation of K562 and monocytic differentia-tion of THP-1 cells (data not shown).

c-Myc binds the p21 core promoter in vitro throughinteraction with Miz-1

To investigate whether Miz-1 and c-Myc are able tobind the p21 core promoter, cell extracts from trans-fected Cos-7 cells was used in a DNA pulldown assaywith a double-stranded oligonucleotide representing nt�49 to +16 of the p21 promoter. Figure 4a (left panel)shows that Miz-1 bound the p21 promoter sequence,while c-Myc alone did not. However, c-Myc boundDNA when cotransfected with Miz-1, suggesting that c-Myc can bind the p21 promoter through interactionwith Miz-1. Coimmunoprecipitations using either c-Mycor Miz-1 antisera confirmed that c-Myc and Miz-1formed a complex in solution (Figure 4a, right panel) aspreviously reported (Peukert et al., 1997). The apparentless efficient coimmunoprecipitation of c-Myc with Miz-1 antibodies than the reverse is likely due to differencesin antibody affinities. Interestingly, Miz-1 DNA bindingwas enhanced in the presence of c-Myc, suggesting thatc-Myc may stabilize the Miz-1 interaction with DNA.The binding of Miz-1/Myc to the p21 sequence wasspecific, since it did not bind the inverse sequence or ap21 sequence with mutations in the three potential Miz-1 binding sites (Figure 4b). A c-Myc HLH point mutantwith impaired binding to Miz-1 but normal binding toMax was recently identified (Herold et al., 2002). Toconfirm that c-Myc DNA binding involves HLHinteraction with Miz-1, we utilized the mutantMycV394D in the DNA pulldown assay. Figure 4c (leftpanel) shows that the DNA-binding activity of theV394D mutant was reduced to approximately 25–30%

Myc represses p21CIP1 expression via Miz-1S Wu et al

353

Oncogene

of that of wt c-Myc, as determined by densitometricalscanning. The enhanced binding of Miz-1 to DNA by c-Myc was also impaired. Figure 4c (middle panel)

confirmed that the V394D mutant bound inefficientlyto Miz-1 (approximately 40% of wt Myc in this assay)(Herold et al., 2002). The right panel shows that equal

Figure 2 c-Myc repression maps to the p21 core promoter and requires the Myc HLH, Zip and MB2 regions, but not the DNA-binding domain. (a) Schematic representation of p21 promoter deletion (upper part) and p21/CMV IE promoter (lower part) exchangemutants. Binding sites for transcription factors regulating the p21 promoter are indicated. Lower part: Enlargement of the p21 10-Lucconstruct (in white), part of the CMV IE promoter (in gray) and exchange mutants CMV/p21-Luc and p21/CMV-Luc based on thesetwo sequences. Binding sites for transcription factors are underlined. (b and c) Cotransfection of c-Myc with indicated p21 promoter/reporter deletion mutants (b) and p21/CMV IE promoter exchange mutants (c) into U-937-1 cells. (d) Schematic representation of c-Myc deletion mutants used. MB1: Myc box 1; MB2: Myc box 2; BR: basic region; HLH: helix–loop–helix; LZ: leucine zipper.(e) Cotransfection of 10-luc with indicated c-Myc deletion mutants into U-937-1 cells. The data presented in (b, c and e) are based onat least three separate experiments

Myc represses p21CIP1 expression via Miz-1S Wu et al

354

Oncogene

amounts of c-Myc and Miz-1 proteins were used for theassays. We conclude that binding of c-Myc to the p21promoter depends on interaction with Miz-1, and that c-Myc stabilizes Miz-1 binding to DNA.

c-Myc repression of p21 involves Miz-1-interactingresidues within the c-Myc HLH domain

To establish whether c-Myc-induced repression of thep21 promoter depends on interaction with Miz-1 in vivo,transient cotransfections were performed as above usingthe c-MycV394D and Myc(BR)-Mad mutants. In thelatter, the HLHZip domain is replaced by the corre-sponding region in Mad1, enabling the mutant to bindMax and to activate transcription from E-box elements,but unable to bind Miz-1 (Staller et al., 2001). The

repression by the V394D mutant was indeed impairedcompared to wt c-Myc, thus correlating with theimpaired binding to Miz-1 (Figure 4d). The Myc(BR)-Mad mutant was also unable to repress as expected.To demonstrate that repression of the endogenous

p21 gene by Myc was Miz-1 dependent, we selectedstable transfectants of K562 cells containing a Zn2+-inducible V394D mutant as confirmed in Figure 4e(upper panel). p21 expression was induced in parentalK562 cells and in K562KMT control cells (K562containing empty vector) during TPA-stimulated differ-entiation, while it was reduced approximately two foldin the presence of Zn2+ in the wt c-Myc KmycB clone(Figure 4e, middle panel). In contrast, no repression wasobserved in the representative KmycVD7 mutant clone.Similar observations were made in other wt and mutant

Figure 3 Miz-1 activates the p21 core promoter and is induced during hematopoietic differentiation. (a) Cotransfection of Miz-1 andindicated p21 promoter/reporter constructs into U-937-1 cells. (b) Increasing amounts of CMV-Myc were cotransfected with 5 mg ofCMV-Miz-1 and the indicated p21 reporters into U-937-1 cells. The values are normalized to those of untreated or TPA-treated cellstransfected with Miz-1 alone. (c and d) Northern blot analysis of miz-1 mRNA expression during TPA-induced differentiation of K562cells (c), TPA-induced monocytic and RA-induced granulocytic differentiation of HL-60 cells for 3 days, and TPA-induced monocyticdifferentiation of U-937-1 cells for 24 h. (d). (e) Western blot analysis of Miz-1 protein expression during TPA-induced differentiation ofHL-60 cells

Myc represses p21CIP1 expression via Miz-1S Wu et al

355

Oncogene

myc clones (data not shown). The findings stronglysuggest that Myc-induced repression of the endogenousp21 gene is dependent on interaction with Miz-1.

c-Myc and Miz-1 interact with the p21 promoter in vivo

We next investigated whether c-Myc and Miz-1 interactwith the p21 promoter in vivo. ChIP analysis using HL-

60 cells demonstrated that a PCR product representingthe p21 promoter region (�194 to +88) was immuno-precipitated with two Myc antisera, two Miz-1 antiseraand a Max antiserum, while only faint backgroundbands were visible using preimmune and two othercontrol antisera (Figure 5a). In contrast, no PCRproducts were visible using primers for downstream(+3455 to +3771) or upstream (�3723 to �3423)

Figure 4 c-Myc binds to and represses the p21 core promoter via interaction with Miz-1. (a–c) DNA pulldown assay using extractsfrom Cos7 cells transfected with indicated vectors and double-stranded oligonucleotides representing the wt or mutated p21 corepromoter (left panels). In (a) right panel and (c) middle panel, the same extracts were immunoprecipitated with indicated antisera andanalysed by Western blot. (c) Right panel: Western blot analysis of whole cell extracts. (d) Transient cotransfections of K562 cells withthe p21DPst-Luc reporter construct and wt Myc, V394D or Myc(BR)-Mad mutants. (e) Northern blot analysis of p21, c-MYC andexogenous c-MYC (exogen) expression in K562 cells containing Zn2+-inducible wt c-Myc (KmycB), MycV394D (KmycVD7) orempty vector (KMT). The cells were induced with Zn2+ for 16 h followed by TPA for 3 h as indicated. All experiments were repeatedat least three times

Myc represses p21CIP1 expression via Miz-1S Wu et al

356

Oncogene

regions of the p21 gene. p21 gene expression is inducedin HL-60 cells, like in U-937 (Figure 1) and K-562(Figure 4) cells, during TPA-induced differentiation(Jiang et al., 1994; Steinman et al., 1994). Figure 5bshows that in vivo binding of c-Myc to the p21 promoterdecreased during induced HL-60 differentiation,whereas the binding of Miz-1 and Max was only slightlyreduced. In contrast, binding of Myc and Max to thep21 promoter increased after TPA treatment in v-Myc-transformed U-937-myc-6 cells (Figure 5c), whereinduced expression of p21 is blocked (Figure 1). Alsobinding of Miz-1 to the promoter increased in responseto TPA. No PCR products were obtained with primersfor the downstream or upstream regions of the p21 gene

using chromatin from untreated and TPA-treated HL-60 or U-937-myc-6 cells (data not shown). In conclu-sion, c-Myc, Miz-1 and Max indeed interact with thep21 promoter in vivo in a differentiation-dependentmanner, correlating with p21 expression.

Discussion

Inhibition of cellular differentiation is one of the well-known biological activities of c-Myc-family proteins.We show here that Myc represses differentiation-induced expression of the CDK inhibitor p21. A largebody of evidence suggests that p21 plays an importantrole in cell fate decisions during growth and differentia-tion. p21 expression is induced during differentiation,and interference with p21 expression by gene targetingdisturbs normal development and differentiation ofmany tissues (Jiang et al., 1994; Steinman et al., 1994;Liu et al., 1996; Missero et al., 1996; Zhang et al., 1999;Cheng et al., 2000; for reviews, see Sherr and Roberts,1999; Steinman, 2002). Since cell growth and differ-entiation are closely linked and mutually exclusive at theterminal stage of differentiation, it is likely that theCDK inhibitory activity of p21 plays an important roleduring these processes (Sherr and Roberts, 1999), butother activities of p21 possibly involved directly inpromoting differentiation have also been suggested(Zezula et al., 2001). Repression of differentiation-induced p21 expression may therefore be an importantcomponent of the mechanism by which Myc blocksdifferentiation. This view is also supported by theobservation that forced expression of p21 inducesdifferentiation of several cell types including the U-937monoblasts (Liu et al., 1996) used in the present study.Our study suggests that c-Myc represses both

differentiation-induced and basal p21 expression at leastin part at the level of transcription. This is consistentwith recent reports that c-Myc inhibits TGF-b-, TPA-and p53-induced as well as basal p21 expression in othercell systems (Mitchell and El-Deiry, 1999; Ceballos et al.,2000; Claassen and Hann, 2000; Coller et al., 2000;Gartel et al., 2001). Our present study provides therationale for these observations. Taken together, ourresults from promoter/reporter-, DNA pulldown- andChIP-assays suggest that the c-Myc-responsive region issituated around the transcriptional start site and that c-Myc binds to this region through interaction with Miz-1, which associates directly with the promoter. This isreminiscent of the situation at the p15INK4B promoter,where c-Myc also utilizes Miz-1 for repression (Seoaneet al., 2001; Staller et al., 2001). Using mutants of c-Mycwith impaired binding to Miz-1, our results further showthat repression of the endogenous p21 gene depends onresidues specifically interacting with Miz-1.Miz-1 seems to play a dual role at the p21 promoter.

It acts as an activator of the promoter (Figure 3a, b),and its increased expression during hematopoieticdifferentiation may indicate a positive regulatory rolein this process (Figure 3c, d). However, Myc and Miz-1also seem to cooperate. Myc utilizes Miz-1 for repres-

Figure 5 c-Myc and Miz-1 interact with the p21 promoter in vivo.(a) ChIP analysis of HL-60 cells using indicated antisera and PCRprimers specific for the p21 promoter (upper panel), far upstream(middle) or far downstream (lower panel) regions of the p21 gene.(b and c) ChIP analysis using p21 promoter-specific primers andchromatin prepared from HL-60 (b) or v-Myc-transformed U-937-myc-6 (c) cells stimulated by TPA for 2 days or left untreated

Myc represses p21CIP1 expression via Miz-1S Wu et al

357

Oncogene

sion of p21, and enhances Miz-1 DNA-binding activityin vitro (Figure 4) and possibly also in vivo, since Miz-1binding to the p21 promoter seems to correlate withMyc rather than Miz expression (Figure 5). Myc is alsoreported to facilitate transport of Miz-1 to the nucleus(Peukert et al., 1997). We therefore propose a modelwhere Myc/Miz-1 functions as a growth/differentiationswitch resembling the E2F/Rb switch for G1/S transi-tion (Figure 6). According to this model, Miz-1 repressesp21 when complexed with Myc, thereby promoting cellgrowth and inhibiting differentiation. As c-Myc isdown-regulated by differentiation signals, Miz-1 actstogether with cofactors and other transcription factorsto activate transcription of p21, thereby stimulatingterminal differentiation.It is likely that other transcription factors participate

in the early induction of p21 during differentiation, anda number of factors that mediate differentiation and/orgrowth inhibitory signals binds to upstream elements inthe p21 promoter (Gartel and Tyner, 1999) (Figure 2a).During TGFb-induced activation of the p15 promoter,which is similar in organization to the p21 promoter,Miz-1 interacts directly with Smads, which binds toupstream elements, and p300 (Seoane et al., 2001; Stalleret al., 2001). Smads also interacts with Sp1-sites at thepromoter. It is therefore possible that Myc and Miz-1interact with other transcription factors at the p21promoter (Figure 6). Recent reports suggesting thatc-Myc interacts directly with Sp1 and Smads arecompatible with this suggestion (Gartel et al., 2001;Feng et al., 2002). Although it is conceivable that Mycmay also utilize alternative or cooperative mechanismsfor repression, our results show that Myc binding to andrepression of the p21 promoter relies on interaction withMiz-1 and is not dependent on upstream regulatoryelements. Our hypothesis is therefore that Myc primarilyassociates with a central component of the transcrip-tional machinery at the p21 core promoter, Miz-1,thereby blocking multiple incoming upstream signals foractivation.Our mutational analysis also suggested that MB2 and

the Zip domain were essential, indicating that playersother than Miz-1 are involved in Myc-induced repres-sion of p21 and that repression does not simply involveblocking interactions between Miz-1 and its cofactors.Deletion of the Zip domain abolishes the interactionwith Max, which apparently is part of the Myc/Miz-1complex both in vitro and in vivo. The role of Max in thiscontext is, however, still unclear. MB2 interacts withTRRAP, TIP48, TIP49 and BAF53, which are compo-nents of different Gcn5- and MYST-family HATcomplexes involved in the regulation of chromatinstructure (McMahon et al., 1998, 2000; Bouchard et al.,2001; Frank et al., 2001; for a review, see Amati et al.,2001). It is therefore tempting to speculate that Mycmay recruit complexes with repressor activity to the p21promoter through MB2 (Figure 6). It remains to bedemonstrated whether any of the HAT complexesabove, which are usually associated with transcriptionalactivation, could also play a role in Myc-inducedrepression, or whether distinct corepressor complexes

are involved. Gcn5 was, however, recently reported tobe directly involved in repression of the ARG1 gene inyeast (Ricci et al., 2002).

Figure 6 A Myc/Miz-1 switch model for regulation of p21 geneexpression during differentiation. (a) In undifferentiated cells, Miz-1 represses p21 expression by recruiting Myc to the core promoter.The Myc-recruited corepressor (CoR) is hypothesized based on thedependence of the Myc transactivation domain for repression. (b)Differentiation signals results in c-Myc downregulation, increasedexpression of Miz-1, activation of transcription factors involved indifferentiation (DF: differentiation factors), for example, VDR,RAR, C/EBP etc., acting from upstream regulatory elements,recruiting coactivators (CoA). We hypothesize that these factorsinteract and cooperate with Miz-1 in analogy with the dual inputmodel for TGFb-induced activation of the p15INK4b promoter(Seoane et al., 2001). (c) In Myc-transformed tumor cells, Mycdominantly controls the repressor function of Miz-1 despiteactivation of DF by differentiation signals, thereby precludingactivation of p21. This may involve exclusion of CoA from a Miz-1activator complex (Staller et al., 2001), and/or binding a CoRcomplex

Myc represses p21CIP1 expression via Miz-1S Wu et al

358

Oncogene

Our data thus suggest that the Myc/Miz-1 switch forrepression of target genes such as p21 is utilized forinhibition of differentiation by Myc. It is, however,likely that the activation of target genes through E-boxes also plays a role in maintaining cells in anundifferentiated proliferating state. Differentiation sig-nals could turn off E-box-containing Myc target genesvia induced expression of Mad-family proteins, whichdimerize with Max to actively repress such genes(Grandori et al., 2000). Therefore, Myc/Miz and Myc/Max/Mad switches are probably in operation togetherduring terminal differentiation. Further studies areneeded to clarify the relative importance of these twopathways.

Materials and methods

Cell culture and differentiation

Human U-937 and K562 myeloid cells and subclones werecultured in RPMI-1640 supplemented with 10% fetal calfserum and antibiotics. U-937-1, the v-Myc-expressing U-937-myc-6 clone and assays for monocytic differentiation havebeen described (Larsson et al., 1988; Bahram et al., 1999). Forthe analysis of the CD11c monocytic differentiation marker,LeuM5mAb (Becton and Dickinson, Mountain View, CA,USA) was used. Conditions for TPA-induced differentiationand Zn2+ treatment of selected K562 clones containing Zn2+-inducible c-MYC (KmycB) (Delgado et al., 1995), c-MYCV394D mutant (KmycVD7) or empty vector (KMT)have been described (Delgado et al., 1995).

Northern blot analysis

The Northern blot procedure and probes for human c-MYC,p21 and GAPDH were as described (Larsson et al., 1988;Ceballos et al., 2000). The probe for Miz-1 was a 2.4 kbfragment of human cDNA. In sum, 15mg of total RNA wasused per lane.

Protein analysis

35S-labeling, immunoprecipitation and Western blot analysiswere performed as described previously (Bahram et al., 1999).An equal number of TCA-precipitable counts for 35S-labeledproteins or equal amounts of unlabeled protein, as deter-mined by Bradford assay of cell extracts and Ponceau stainingof Western filters, were used for each sample. IG-9 (Bahramet al., 1999) and C-33 (Santa Cruz Biotechnology) antipan-Myc and 10E2 anti-Miz-1 antibodies were employed. p21antibodies were kindly provided by Dr B Westermark,Uppsala, Sweden.

Transfections, luciferase assays and DNA constructs

Transfection by electroporation and luciferase reporter assayswere performed as described (Bahram et al., 1999; Delgadoet al., 2000). pCMV-LacZ or pRL-TK (Promega) wascotransfected to normalize transfection efficiencies. The hu-man p21 promoter/luciferase constructs 0- to 10-Luc (Zenget al., 1997), depicted in Figure 2a, were kindly provided by DrW El-Deiry (Philadelphia, PA, USA). The p21DPst-Lucreporter was described (Delgado et al., 2000). pUHC13.3contains a minimal CMV promoter with multiple binding sitesfor the rtTA transactivator in front of the luciferase gene. Thestructure of the CMV/p21-Luc and p21/CMV-Luc p21

promoter/CMV IE promoter exchange mutants is outlinedin Figure 2a. These were created by the annealing of twopartially overlapping complementary oligonucleotide pairsdesigned to contain HindIII and BglII sites, respectively (50-CCCAAGCTTCCCCATTGACGCAAATGGGCGGTAGGCGTGTACGGTGGGAGTTGTATATCAGGGCCGCGCTand 50-GGAAGATCTCACAAGGAACTGACTTCGGCAGCTGCTCACACCTCAGCTGGCGCAGCTCAGCGCGGCCCTGATA for CMV/p21-Luc and 50CCCAAGCTTCGAGCGCGGGTCCCGCCTCCTTGAGGCGGGCCCGGGCGG-GGCGGGCCTATATAAGCAGAGCT and 50-GAAGATCTCAAAACAGCGTGGATGGCGTCTCCAGGCGATCTGACGGTTACTAAACGAGCTCTGCTTATATA for p21/CMV-Luc), followed by PCR amplification using the underlinedprimers. The HindIII/BglII-digested products were cloned intoplasmid pXP2-Luc. The c-MYC deletion mutants DHLH, DLZ,DBR, DMB1 and DMB2, the Myc(BR)Mad mutant andpcDNA3-Miz-1 have been described previously (Peukert et al.,1997; Staller et al., 2001). The generation of the MycV394Dmutant is described in Herold et al. (2002).For stable transfections, K562 cells electroporated at 260V

and 1mF were selected with 0.5mg/ml G418. pMT-CB6-MycV394D was constructed by ligation of the 1.3 kb KpnI/XbaI fragment of plasmid pcDNA3-MycV394D into the Zn2+-inducible pMT-CB6+ vector (kindly provided by FJ Rausher,Philadelphia, PA, USA).

DNA pulldown and chromatin immunoprecipitation (ChIP)assays

For DNA pulldown, cells were lysed by sonication in HKMGbuffer as described (Hata et al., 2000). Cell extracts (1–2� 106 cells) were incubated with 1mg of biotinylated double-stranded oligonucleotide, 0.2mg ssDNA and 0.5mg BSA in1ml HKMG buffer for 16 h. DNA-bound proteins werecollected with streptavidin–agarose beads, washed withHKMG buffer and analysed by Western blotting. Thefollowing oligonucleotides pairs were used: p21 wt, biotin-TTGTATATCAGGGCCGCGCTGAGCTGCGCCAGCTGAGGTGTGAGCAGCTGCCGAAGTCAGTTCCTTGTGand 50-CACAAGGAACTGACTTCGGCAGCTGCTCACACCTCAGCTGGCGCAGCTCAGCGCGGCCCTGATATACAA; mutated p21, biotin-TTGTATATCGCGGCCGCGCTGAGCTGCGCCAGCTGTAGTGTGAGCAGCT-GCCGAAGTCAGTTCGATGTG and 50-CACATCGAACTGACTTCGGCAGCTGCTCACACTACAGC-TGGCGCAGCTCAGCGCGGCCGCGATATACAA; reversed p21, biotin-GTGTTCCTTGACTGAA-GCCGTCGACGAGTGTGGAGTCGACCGCGTCGAGTCGCGCCGGGACTATATGTT andAACATAT-AGTCCCGGCGCGACTCGACGCGGTCGACTCCACACTCGTCGACGGCTTCAGTCAAGGAACAC.For ChIP assays, cells were lysed and crosslinked with

formaldehyde as described (Eberhardy et al., 2000). ChromatinDNA was sonicated to an average size of 300–500 bp.Immunoprecipitations were carried out in RIPA buffer withthe following antibodies: a-c-Myc (N-262), a-Max (C-17), a-Gal4 (all Santa Cruz), a-Flag M2 (Sigma), polyclonal a-pan-Myc (IG9) and preimmune serum (IG0), a-Miz-1 10E2mAb,Miz-1 polyclonal antiserum. The immunoprecipitates werecollected with protein G-sepharose beads, washed as described(Bahram et al., 1999), and eluted in TE, 1% SDS. PurifiedDNA samples were analysed by PCR using primers specific forp21 gene regions. Primers used for PCR reactions were:p21 (�194/+88), 50-ACCGGCTGGCCTGCTGGAACT and50-TCTGCCGCCGCTCTCTCACCT; p21 (+3455/+3771),50-ATGTTAGGCAAGTTACTTAACTTA and 50-CTCTTG

Myc represses p21CIP1 expression via Miz-1S Wu et al

359

Oncogene

GTAACTTCACACCAAGTT; p21 (�3723/�3400), 50-CTCTCTTCAAACATTGTACAAGAC and 50-AACCTTTTAACTTCAACTGTCCAA.

Note added in proof

After acceptance of this manuscript, two papers containingpartially overlapping data were published, Seoane et al. (2002).Nature, 419, 727–734, and van de Wetering et al. (2002). Cell,111, 241–250.

Acknowledgements

We thank Drs WS El-Deiry, B Luscher, P Staller, X-F Wang,G Westin, B Westermark and J Ziegelbauer for reagents. Wealso thank Dr J Botling, Asa Boker and A Castell forexperimental support, and Drs D Balciunas, D Grander, BLuscher, A Moustakas, F Oberg, H Ronne, C Svensson and GWestin for valuable discussions. This work was supported bygrants from the Swedish Cancer Society, Deutsche For-schungsmeinschaft, and Grants PM98-109 and FD97-1987from the Spanish Government.

References

Amati B, Frank SR, Donjerkovic D and Taubert S. (2001).Biochim. Biophys. Acta, 1471, M135–M145.

Bahram F, Wu S, Oberg F, Luscher B and Larsson L-G.(1999). Blood, 93, 3900–3912.

Bouchard C, Dittrich O, Kiermaier A, Dohmann K, MenkelA, Eilers M and Luscher, B. (2001). Genes Dev., 15, 2042–2047.

Ceballos E, Delgado MD, Gutierrez P, Richard C, Muller D,Eilers M, Ehinger M, Gullberg U and Leon J. (2000).Oncogene, 19, 2194–2204.

Cheng T, Rodrigues N, Shen H, Yang Y, Dombkowski D,Sykes M and Scadden DT. (2000). Science, 287, 1804–1808.

Claassen GF and Hann SR. (2000). Proc. Natl. Acad. Sci.USA, 97, 9498–9503.

Coller HA, Grandori C, Tamayo P, Colbert T, Lander ES,Eisenman RN and Golub TR. (2000). Proc. Natl. Acad. Sci.USA, 97, 3260–3265.

Delgado MD, Lerga A, Canelles M, Gomez-Casares MT andLeon J. (1995). Oncogene, 10, 1659–1665.

Delgado MD, Vaque JP, Arozarena I, Lopez-Ilasaca MA,Martinez C, Crespo P and Leon J. (2000). Oncogene, 19,783–790.

Eberhardy SR, D’Cunha CA and Farnham PJ. (2000). J. Biol.Chem., 275, 33798–33805.

Feng XH, Liang YY, Liang M, Zhai W and Lin X. (2002).Mol. Cell, 9, 133–143.

Frank SR, Schroeder M, Fernandez P, Taubert S and AmatiB. (2001). Genes Dev., 15, 2069–2082.

Gartel AL and Tyner AL. (1999). Exp. Cell Res., 246, 280–289.Gartel AL, Ye X, Goufman E, Shianov P, Hay N, NajmabadiF and Tyner AL. (2001). Proc. Natl. Acad. Sci. USA, 98,4510–4515.

Grandori C, Cowley SM, James LP and Eisenman RN. (2000).Annu. Rev. Cell Dev. Biol., 16, 653–699.

Hata A, Seoane J, Lagna G, Montalvo E, Hemmati-BrivanlouA and Massague J. (2000). Cell, 100, 229–240.

Herold S, Wanzel M, Beuger V, Frohme C, Beul D, HillukkalaT, Syvaoja J, Saluz H-P, Haenel F and Eilers M. (2002).Mol. Cell, 10, 509–521.

Jiang H, Lin J, Su ZZ, Collart FR, Huberman E and FisherPB. (1994). Oncogene, 9, 3397–3406.

James L and Eisenman RN. (2002). Proc. Natl. Acad. Sci.USA, 99, 10429–10434.

Larsson L-G, Ivhed I, Gidlund M, Pettersson U, Vennstrom Band Nilsson K. (1988). Proc. Natl. Acad. Sci. USA, 85,2638–2642.

Levens D. (2002). Proc. Natl. Acad. Sci. USA, 99, 5757–5759.Li LH, Nerlov C, Prendergast G, MacGregor D and Ziff EB.(1994). EMBO J., 13, 4070–4079.

Liu M, Lee MH, Cohen M, Bommakanti M and FreedmanLP. (1996). Genes Dev., 10, 142–153.

McMahon SB, Van Buskirk HA, Dugan KA, Copeland TDand Cole MD. (1998). Cell, 94, 363–374.

McMahon SB, Wood MA and Cole MD. (2000). Mol. CellBiol., 20, 556–562.

Missero C, Di Cunto F, Kiyokawa H, Koff A and Dotto GP.(1996). Genes Dev., 10, 3065–3075.

Mitchell KO and El-Deiry WS. (1999). Cell Growth Differ., 10,223–230.

O’Hagan RC, Schreiber-Agus N, Chen K, David G, EngelmanJA, Schwab R, Alland L, Thomson C, Ronning DR,Sacchettini JC, Meltzer P and DePinho RA. (2000). Nat.Genet., 24, 113–119.

Peukert K, Staller P, Schneider A, Carmichael G, Hanel F andEilers M. (1997). EMBO J., 16, 5672–5686.

Ricci AR, Genereaux J and Brandl CJ. (2002). Mol. Cell Biol.,22, 4033–4042.

Seoane J, Pouponnot C, Staller P, Schader M, Eilers M andMassague J. (2001). Nat. Cell Biol., 3, 400–408.

Sherr CJ and Roberts JM. (1999). Genes Dev., 13, 1501–1512.Staller P, Peukert K, Kiermaier A, Seoane J, Lukas J,Karsunky H, Moroy T, Bartek J, Massague J, Hanel Fand Eilers M. (2001). Nat. Cell. Biol., 3, 392–399.

Steinman RA, Hoffman B, Iro A, Guillouf C, Liebermann DAand el-Houseini ME. (1994). Oncogene 9, 3389–3396.

Steinman RA. (2002). Oncogene 21, 3403–3413.Warner BJ, Blain SW, Seoane J and Massague J. (1999). Mol.

Cell Biol., 19, 5913–5922.Wu KJ, Polack A and Dalla-Favera R. (1999). Science, 283,676–679.

Yang W, Shen J, Wu M, Arsura M, FitzGerald M, Suldan Z,Kim DW, Hofmann CS, Pianetti S, Romieu-Mourez R,Freedman LP and Sonenshein GE. (2001). Oncogene, 20,1688–1702.

Zeng YX, Somasundaram K and el-Deiry WS. (1997). Nat.Genet., 15, 78–82.

Zezula J, Casaccia-Bonnefil P, Ezhevsky SA, Osterhout DJ,Levine JM, Dowdy SF, Chao MV and Koff A. (2001).EMBO Rep., 2, 27–34.

Zhang P, Wong C, Liu D, Finegold M, Harper JW andElledge SJ. (1999). Genes Dev., 13, 213–224.

Myc represses p21CIP1 expression via Miz-1S Wu et al

360

Oncogene