Maintenance of genomic integrity by p53: complementary roles foractivated and non-activated p53

Nils Albrechtsen1, Irene Dornreiter1, Frank Grosse2, Ella Kim1, Lisa WiesmuÈ ller1 andWolfgang Deppert*,1

1Heinrich-Pette-Institut fuÈr Experimentelle Virologie und Immunologie an der UniversitaÈt Hamburg, Martinistrasse 52, D-20251Hamburg, Germany; 2Institut fuÈr Molekulare Biotechnologie, *Abt. Biochemie, Beutenbergstr. 11, D-07745 Jena, Germany

In this review we describe the multiple functions of p53 inresponse to DNA damage, with an emphasis on p53'srole in DNA repair. We summarize data demonstratingthat p53, through its various biochemical activities andvia its ability to interact with components of the repairand recombination machinery, actively participates invarious processes of DNA repair and DNA recombina-tion. An important aspect in evaluating p53 functionsarises from the ®nding that the p53 core domain harborstwo mutually exclusive biochemical activities, sequence-speci®c DNA binding, required for its transactivationfunction, and 3'-45' exonuclease activity, possiblyinvolved in various aspects of DNA repair. As modi®ca-tions of p53 that lead to activation of its sequence-speci®c DNA-binding activity result in inactivation of its3'-4 5' exonuclease activity, we propose that p53 exertsits functions as a `guardian of the genome' at variouslevels: in its non-induced state, p53 should not beregarded as a non-functional protein, but might beactively involved in prevention and repair of endogenousDNA damage, for example via its exonuclease activity.Upon induction through exogenous DNA damage, p53will exert its well-documented functions as a superiorresponse element in various types of cellular stress. Thedual role model for p53 in maintaining genomic integritysigni®cantly enhances p53's possibilities as a guardian ofthe genome.

Keywords: p53; DNA repair; DNA damage; DNArecombination; DNA replication; sequence-speci®cDNA binding

Introduction

Twenty years after its discovery, the tumor suppressorp53 has become one of the most prominent moleculesin the area of cancer research. Initially classi®ed as anoncogene (reviewed in Deppert, 1994) p53 meanwhile isregarded as a `master tumor suppressor' which ensuresthe integrity of the cells' genome by protecting it fromadverse e�ects of DNA damage. Such damage triggersthe activation and accumulation of p53 (Fritsche et al.,1993; Kastan et al., 1991; Nelson and Kastan, 1994).The best studied function of an activated p53 is that ofa transcription factor, mediating expression of a varietyof genes that induce a growth arrest or apoptosis. As a

result, DNA damage will be repaired during a growtharrest, or the damaged cells will be eliminated, therebypreventing ®xation of DNA damage as mutations. Thisfunction of p53 led to the now famous coining of p53as the `guardian of the genome' by Lane (1992).

Although the main features of p53's role inmaintaining the integrity of the genome seem to beoutlined, there is still a lot to be learned as to how p53exactly acts to prevent the accumulation of mutationalevents within a cell. Despite some uncertaintiesregarding the various mechanisms of its induction,elimination of damaged cells by p53-induced apoptosisis an easily conceived mechanism for achieving thisgoal. Much less is known about the role of p53 inDNA repair. So far, the main emphasis has been givento pathways leading to the activation of p53 by signalsemanating from damaged DNA, with p53 integratingthese signals and triggering a cascade of responsesleading to either growth arrest or apoptosis. Thesemechanisms have been summarized in detail in severalrecent reviews (for example see Cox and Lane, 1995;Gottlieb and Oren, 1996; Ko and Prives, 1996; Levine,1997; Bates and Vousden, 1999), thus this topic will beaddressed here only brie¯y. The next steps in thiscascade, namely the activation of the repair pathwaysthemselves, and the role p53 plays in their activationare far less clear. Furthermore, it is still not known,whether and how p53 directly participates in DNArepair processes, despite some evidence pointing to thispossibility. Last but not least, a possible role of p53 inthe control of genomic integrity in its non-inducedstate, i.e., in the absence of signals indicating DNAdamage or other potentially harmful conditions, so farhas not yet been considered at all. In contrast, thegeneral assumption is that a non-activated p53 is anon-functional one (Hupp and Lane, 1994; Vogelsteinand Kinzler, 1992).

In this review we will focus on p53-inducedpathways that lead to or are part of DNA repairprocesses, with an emphasis on the known biochemicalactivities of p53, which might play a direct role inrepair processes. Recent evidence from our laboratoryshowed that at least one of these activities, the 3'-45'exonuclease activity, is regulated in a manner oppositeto that of sequence-speci®c DNA binding (Janus et al.,1999). Based on this ®nding we propose a modelaccording to which p53 in its non-induced state, i.e. inthe absence of conditions which lead to its activation,exerts basic functions in maintaining genomic integrity.Upon various cellular stress conditions, such ashypoxia, nutrient depletion, or DNA damage, thewell known functions of p53 in integrating therespective signals become activated, triggering the*Correspondence: W Deppert

Oncogene (1999) 18, 7706 ± 7717ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $15.00

http://www.stockton-press.co.uk/onc

cascade of p53 responses which lead to growth arrest,DNA repair, or apoptosis (Hall et al., 1996).

Activation of p53 upon DNA damage

Cell cycle check point control and apoptosis

Eukaryotic cells have developed a network of highlyconserved surveillance mechanisms (checkpoints) whichensure that damaged chromosomes are repaired beforethey are replicated or segregated. These mechanismsare essential for maintaining genomic integrity and cellviability. The tumor suppressor p53 is one of thecritical mediators of cellular responses to various typesof genotoxic stress (reviewed e.g. in Hall et al., 1996),acting at di�erent levels of control during the cell cycle.

In response to DNA damage, like ionizing irradia-tion (IR) (Kastan et al., 1991), ultraviolet (UV)irradiation (Maltzman and Czyzyk, 1984; Nelson andKastan, 1994), hypoxia (Graeber et al., 1994), andribonucleoside triphosphate depletion (Linke et al.,1996), p53 becomes activated as a transcription factorvia speci®c post-translational modi®cations. The p53protein is phosphorylated by a wide variety of proteinkinases: in its transactivation domain by casein kinase(Ck) I, DNA-PK, ATM, JNK, and MAP kinases, andat its C terminus by cyclin dependent kinases (Cdks),PKC, and Ck II (see Martinez et al., 1997; Meek, 1994and references herein). The di�erent protein kinasescan modify the sequence-speci®c DNA binding activityof p53 and may thereby modulate the relative e�ciencyof activation of di�erent p53 target genes (Hecker etal., 1996; Hupp and Lane, 1994; Hupp et al., 1992;Takenaka et al., 1995; Wang and Prives, 1995).Phosphorylation by DNA-PK within the transactiva-tion domain of the protein (Shieh et al., 1997) alsoabrogates its interaction with MDM2, a key player innegative regulation of the transcriptional activity andthe level of p53 (reviewed in Lane and Hall, 1997).DNA-PK phosphorylation on serine 15 and possiblyserine 37 within (human) p53 correlates with theenhanced transcription of downstream p53 targetgenes. The ability of p53 to transactivate downstreamtarget genes results in cell cycle arrest at speci®c pointsin the cell cycle. As a consequence, the cell cycle stopseither before DNA replication in G1, or before mitosisin G2, which leads to a severe decrease in the amountof S phase cells.

G1 arrest The p53-dependent G1 arrest results mainlyfrom transactivation of the waf1 gene that codes forthe small kinase inhibitor p21Waf1. p21Waf1 interferes withcell-cycle progression and prevents S phase entry byblocking the activity of Cdks (Dulic et al., 1994; El-Deiry et al., 1993; Harper et al., 1993; Xiong et al.,1993). Irradiated G1 phase cells accumulate high levelsof Cdk2/cyclin E complexes which are inactivated bythe association with p21Waf1. Inhibition of G1 phase-speci®c kinase activity maintains a hypophosphorylatedretinoblastoma susceptibility gene product pRb, whichrepresses E2F-speci®c transcription of genes that arerequired for entry into S phase, and thereby inhibit cellcycle progression.

p53 may also be able to induce a G1-arrest by non-transcriptional mechanisms. Recently, it has been

observed that p53 binds cyclin H, which is part ofthe Cdk7/cyclinH/Mat1 Cdk activating kinase complex(CAK). CAK plays a crucial role in activating cellcycle progression by phosphorylation and activation ofCdk2, and, as part of the TFHII complex, bycontrolling transcriptional activity of RNA polymer-ase II. In the TFHII complex, CAK phosphorylates thecarboxy-terminal repeat domain (CTD) of RNApolymerase II, which is required for the transcriptionalelongation by RNA polymerase II. Binding of p53 toCAK results in strong reduction of CAK activity bothtowards cdk2 and towards the CTD of RNApolymerase II, which could lead to either growtharrest or apoptosis (Ko et al., 1997; Schneider et al.,1998).

S phase arrest In another checkpoint response path-way which operates in S phase and slows down the rateof DNA replication, p21Waf1 binds to the proliferating-cell nuclear antigen (PCNA) and blocks its activity,interfering with cell-cycle progression by blocking theelongation step in DNA replication (Waga et al., 1994).Stillman and coworkers suggested a dual role forPCNA in DNA replication and DNA repair whichallows p21Waf1 to arrest DNA replication whilepermitting active DNA repair (Li et al., 1994a). InDNA replication, PCNA together with replicationfactor C (RFC) recognizes a primer-template junctionand promotes loading of polymerase d (pol d). ThePCNA-RFC-pol d complex also enhances the proces-sivity of pol d during the elongation step of DNAreplication (Melendy and Stillman, 1991). The directbinding of p21Waf1 to PCNA causes a rapid dissociationof the PCNA-RFC-pol d complex from the replicationfork, stalling replicative DNA synthesis. In repairDNA synthesis, PCNA helps the polymerases d or eto localize the junction of incised DNA, but is notneeded during DNA synthesis (Li et al., 1994a).Thereby, repair DNA synthesis could proceed also inthe presence of an upregulated p21Waf1.

G2 arrest The third and the last line of defenseagainst induced chromatid damage occurs in G2. Apossible role for p53 in the G2/M checkpoint wassuggested by Li et al. (1994b) who reported that p21Waf1

can associate with cyclin A and cyclin B complexesduring the later phases of the cell cycle, suggesting afunctional interaction with the respective associatedkinases. Furthermore, a bimodal periodicity for waf1mRNA levels in human ®broblasts with peaks in G1and G2/M was observed, proposing that p21Waf1 mayplay a role at the onset of mitosis. p21Waf1 is absentfrom the nucleus during S phase and transientlyreenters the nucleus during late G2 phase. In late G2half of the Cdk2/cyclin A is complexed with p21Waf1.p21Waf1 may thereby either directly inhibit the activekinase or prevent its activation by the Cdk-activatingkinase CAK (Poon et al., 1996). Another possibility isthat complex formation with p21Waf1 may blockinteraction of substrates with Cdk2/cylin A (Adamset al., 1996). In Xenopus egg extracts Cdk2 serves as apositive regulator for the activation of Cdk1/cyclin Bcomplexes and prevents entry into mitosis whencomplexed to p21Waf1 (Guadagno and Newport, 1996).Nuclear accumulation of p21Waf1 and the resultinginactivation of Cdk-cyclin complexes at the onset of

Dual role model for p53 functionN Albrechtsen et al

7707

mitosis might be part of a p21Waf1 dependent mitoticattenuation mechanism (Dulic et al., 1998). However,p21Waf1 accumulation resulting from p53 activationseems to be mainly required for the DNA damage-induced G1 arrest, as p21Waf1 clearly is not essential forthe immediate G2 checkpoint response (Brugarolas etal., 1995; Levedakou et al., 1995; Paules et al., 1995).Nevertheless, several studies have suggested that theG2/M arrest following DNA damage is also p53-dependent (Agarwal et al., 1995; Aloni-Grinstein et al.,1995; Goi et al., 1997; Schwartz et al., 1997; Stewart etal., 1995).

Vogelstein and colleagues used a wild-type p53expressing human colorectal cancer cell line to analysegene expression following g-irradiation. The cellsarrested mostly in G2, and the block was accompaniedby changes in gene expression. Quantitative analysis ofgene expression patterns revealed a strong induction ofthe 14.3.3s gene. The induction of 14.3.3s is mediatedby a p53-responsive element in the 14.3.3s gene(Hermeking et al., 1997). The 14.3.3-family proteinsare found in a wide variety of mammalian tissues andin other eukaryotic organisms including plants andyeast (Aitken et al., 1992), and diverse biochemicalproperties have been ascribed to them. Mammaliancells contain a minimum of seven 14.3.3 isoforms(Aitken et al., 1995; Wang and Shakes, 1996). Studiesin yeast indicate that 14.3.3 proteins have a role indetermining the timing of mitosis (Ford et al., 1994).The induction of 14.3.3s by p53 might be part of p53'scontrol of G2/M transition. The link between DNAdamage control and cell cycle checkpoint control atG2/M could be further substantiated by the followingobservations. In irradiated cells the protein kinaseChk1, which is required for this DNA damagecheckpoint (Sanchez et al., 1997), undergoes a Rad3-dependent phosphorylation (Rad3 is a kinase related tothe ATM protein) (Ford et al., 1994). The ultimatetarget of this checkpoint signal is thought to be Cdk1,the cyclin-dependent kinase that induces mitosis (Pines,1991), as Chk1 directly phopshorylates Cdc25C, aphosphatase that regulates Cdk1 activity (Furnari etal., 1997; Peng et al., 1997; Sanchez et al., 1997). Thisphosphorylation event promotes the binding of 14.3.3to Cdc25C and thereby its sequestration. In this stateCdc25C cannot dephosphorylate and activate Cdk1.Thus cell cycle progression after g-irradiation (andpossibly other DNA damage inducing events) isblocked by independent p53-mediated e�ector path-ways: Entry into S phase is inhibited by p21Waf1,whereas 14.3.3s prevents cells that have completed Sphase from entering mitosis. Therefore the p53-dependent induction of 14.3.3s connects DNAdamage with G2/M progression driven by Cdk1 inthe same way as the p53-dependent induction of p21Waf1

connects DNA damage with the Cdks required for G1/S progression.

Apoptosis In response to DNA damage, p53 can alsoinduce apoptosis. Induction of apoptosis by p53 iscomplex and not yet understood in detail. Transcrip-tional activation of a variety of di�erent genes involvedin mediating an apoptotic response (e.g. bax, Fas/Apo,Killer/DR5, and redox regulator genes [PIGs]) againseems to be the major mechanism of p53-mediatedapoptosis, but p53 is able to induce apoptosis also in a

transcription-independent mode (Caelles et al., 1994;Haupt et al., 1995; reviewed in Bates and Vousden,1999). The interaction of p53 with CAK alreadydescribed above could be one such mechanism, butother interactions, mediated by the N-terminal proline-rich domain of p53 which serves as a docking place forproteins containing SH3-domains, have also beenimplicated.

Even less understood is what determines the`decision' to induce either a growth arrest or apoptosisafter p53 activation. The outcome de®nitely is cell typespeci®c, and is also regulated by the severity of DNAdamage, with heavier damage promoting apoptosisover growth arrest. Induction of growth arrest ratherthan apoptosis on the other hand is favored by thepresence of survival factors. In this respect, it isinteresting that IGF-BP3 is a target for transcriptionalactivation by p53. Overexpression of IGF-BP3 caninhibit the mitogenic activity of IGF receptor signaling,while in other systems it can inhibit the survival signalsprovided by IGF, thereby amplifying apoptosis. Thusthe extracellular environment and intracellular signalsstrongly modulate the p53 response (reviewed in Batesand Vousden, 1999).

Selectivity of transcriptional regulation by p53 As theoutcome of the cellular response upon activation ofp53 to a great extent is dependent on transactivation ofthe appropriate target genes, the question arises howselectivity of transactivation is secured. So far,approximately 20 more or less con®rmed target genesof p53 have been described, which will induce quitedi�erently and even opposing or mutually exclusivecellular responses. Thus it is necessary that p53 is ableto discriminate between those genes, activatingtranscription from only those target genes whichmediate the appropriate response in a given cellularenvironment and physiological situation. Several waysto achieve this goal can be envisioned. It has beenproposed that di�erent posttranslational modi®cationsof the p53 protein will lead to the preferentialactivation of certain target genes (reviewed in Jayara-man and Prives, 1999). Similarly, the association of p53with di�erent co-activator molecules will conferspeci®city and selectivity to p53-mediated transactiva-tion. For example, association of p33ING with p53seems to be required for the e�cient transactivation ofthe waf1-promoter by p53 (Garkavtsev et al., 1998). Inaddition to these mechanisms our laboratory recentlysuggested a novel way to ensure speci®city andselectivity of transactivation by p53. The cognateDNA sequences in p53-responsive promoters arecharacterized by signi®cant sequence-heterogeneity, asthe consensus sequence, consisting of two 5'-PuPu-PuC(A/T)(TA)GPyPyPy-3' half-sites separated by 0 ±13 bp (EI-Deiry et al., 1992), by itself allows for manyvariations. In addition, analysis of promoter sequencestransactivated by p53 showed that even the presence ofseveral bases not conforming to the consensus istolerated (Bargonetti et al., 1993; Foord et al., 1993).This amazing heterogeneity within p53 cognate sitesasks for additional mechanisms providing speci®cityand selectivity to sequence-speci®c transactivation byp53. Indeed, close examination of the various naturaland arti®cial DNA sequences recognized by p53demonstrated that, despite their sequence-heterogene-

Dual role model for p53 functionN Albrechtsen et al

7708

ity, many of them display an internal symmetry as acommon feature. Internal symmetry is a hallmark ofDNA sequences able to assume a non-B-DNAconformation upon topological stress. We thereforespeculated whether p53 promoter DNA might assumedi�erent conformational states, and whether p53 mightbe able to interact with such DNA. Using a variety ofarti®cial DNA substrates displaying the same p53target sequence in di�erent conformational states, wewere able to demonstrate that p53 indeed is able torecognize isoforms of the same target sequencepresented in di�erent conformations. Interestingly,recognition of such isoforms strongly depended onp53 protein conformation: whereas recognition ofisoforms in duplex B-DNA conformation requiredactivation of p53 by addition of PAb421, recognitionof isoforms in non-B-DNA conformations did not. Onthe contrary, binding of p53 to some of such isoformswas inhibited by PAb421. We concluded that sequence-speci®c DNA binding of p53 can be modulated bothby p53 protein conformation and by the conformationof its cognate DNA sequence (Kim et al., 1997).

This ®nding opens the way for a better under-standing of target selectivity of p53 sequence-speci®ctransactivation, as it is becoming increasingly acceptedthat chromatin con®guration, speci®cally promoterDNA conformation, is an important determinant ofpromoter activation. Accordingly, the promoter DNAcan be present in an active conformation, which allowstranscription, and an inactive conformation, whichprohibits transcription. We suggest that p53-reactivepromoters can adopt such conformations, and thatpromoter DNA topology and p53 protein conforma-tion (and/or interaction with auxiliary proteins)determine the interaction of p53 with target genepromoters. In support of this model, we were able toshow that e.g. recognition of the mdm2-promoter byp53 is impaired, when this promoter assumes a non-BDNA conformation in superhelical DNA. Relaxationof superhelicity by topoisomerase treatment restoredp53 binding and increased the level of sequence-speci®ctransactivation by p53 (Kim et al., submitted). In vivo,conformational changes in chromatin are mediated byproteins modeling chromatin architecture (Nickerson,1998; Zlatanova and van Holde, 1998). With regard toour model it therefore may be more than coincidencethat p53 interacts with at least two proteins involved insuch processes, HMG1 (Jayaraman et al., 1998), andtopoisomerase I (Gobert et al., 1996). Another proteinthat is involved in chromatin remodeling and p53binding is the transcriptional co-activator CBP/p300.This protein forms a complex with p53, which in turnis able to enhance transcription of p53-dependent genes(Avantaggiati et al., 1997; Gu and Roeder, 1997). Theinteraction of CBP and p53 is strongly stimulated afterphosphorylation of p53 at serine 15 (Lambert et al.,1998). CBP/p300 bears an intrinsic histone acetyltrans-ferase (HAT) activity that is thought to be involved inhistone acetylation and chromatin decondensation(Bannister and Kouzarides, 1996; Ogryzko et al.,1996). The HAT activity of CBP/p300 is controlledby cell cycle-dependent kinases and the oncoproteinE1A (Ait-Si-Ali et al., 1998). The HAT activity alsophosphorylates p53 in vitro and thereby increases itstranscriptional potential (Sakaguchi et al., 1998). HAT-mediated acetylation of p53 seems also to occur in vivo,

particularly in response to DNA damage (Liu et al.,1999). Hence it appears that there exists anotherregulatory circuit that consists of HATs, p53, and thereorganization of the chromatin structure, that mayrespond to physiological situations such as the ongoingcell cycle and also to pathological threats, such asDNA damage.

DNA repair after activation of p53 by DNA damage

As outlined above, activation of p53 after DNAdamage leads to a p53-induced growth arrest orapoptosis. Whereas induction of apoptosis bypassesany need to repair DNA damage, a p53-inducedtransient growth arrest will ful®ll its intention only, ifit achieves an accurate repair of the damaged DNA.As sequence-speci®c transactivation of p53 targetgenes is regarded as the major function of p53 afterits activation, upregulation of genes involved in DNArepair might be envisioned as a ®rst step to initiaterepair of DNA damage. However, with the exceptionof the gadd45 gene (Kastan et al., 1992), and the p48xeroderma pigmentosum gene (Hwang et al., 1999),which have been linked to DNA repair, no suchdirect e�ect of p53 is known. One thus has toconclude that DNA repair during a p53-inducedgrowth arrest occurs in a p53-independent manner, oralternatively, that p53 participates in DNA repair byother activities. Although there is de®nitely p53-independent DNA repair (as can be demonstratedin p53-de®cient cells), there is also evidence for aparticipation of p53 in repair processes, and for ap53-dependent activation of DNA repair. Theassumption of a direct involvement of p53 in DNArepair leads to the prediction that mutations in p53or the lack of p53 should lead to a defect in DNArepair. If p53 actively participates in DNA repair,p53 should interact with proteins that are part ofDNA repair pathways and/or possess biochemicalactivities by which it could be part of repairpathways.

To test for a direct involvement of p53 in DNArepair, Ford and Hanawalt examined Li-Fraumeni®broblasts which were homozygous for a mutatedp53 and found that these cells were indeed de®cient inthe rate and extent of nuclear excision repair (NER) ofgenomic DNA (Ford and Hanawalt, 1995, 1997). Incontrast, they did not observe a defect in transcription-coupled NER. This point, however, remains somewhatcontroversial, since other investigators found ade®ciency in transcription-coupled NER (Mirzayanset al., 1996; Wang et al., 1995). Reduced NER was alsoobserved when the normal p53 function was disruptedby targeting p53 for degradation by the humanpapillomavirus E6 oncogene, or by expression of adominant-negative mutant p53 (Smith et al., 1995).Furthermore, Li-Fraumeni-Syndrome cells with mu-tated p53 are impaired in their recovery for both RNAand DNA synthesis after UV-treatment, an anomalywhich they share with the DNA repair disordersxeroderma pigmentosum and Cockayne's syndrome(Mirzayans et al., 1996). In this respect, p53 behaveslike other proteins that are directly involved in DNArepair. But not only does a defect in p53 function resultin a defective DNA repair, the induction of p53 after a

Dual role model for p53 functionN Albrechtsen et al

7709

treatment mimicking UV-damage even leads to anenhancement of DNA repair (Eller et al., 1997).

Association of p53 with components of repair pathways

Data showing a speci®c interaction of p53 withcomponents of repair pathways provided clues as tohow p53 is involved in DNA repair. Most of theseproteins are members of the transcription factor IIH(TFIIH) multiprotein complex, which initiates basaltranscription of RNA polymerase II and couplestranscription with NER) (Leveillard et al., 1996;Wang et al., 1995; Xiao et al., 1994). p53 interactswith three components (XPB=ERCC3,XPD=ERCC2 and p62) of the TFIIH proteincomplex. p53 inhibits the helicase activity of XPBand XPD, probably through its strand-reannealingproperties (Leveillard et al., 1996; Wang et al., 1995).The e�ect of TFIIH on p53 function is not yet known.It is possible that the interaction of TFIIH componentswith p53 serves to colocalize p53 at sites of DNArepair or transcriptional initiation.

p53 also binds to Cockayne syndrome B protein(CSB), which was identi®ed as a repair protein byphenotypic complementation of an excision repair-de®cient rodent cell line (Troelstra et al., 1990). Theencoded protein is functionally defect in humanCockayne syndrome type B, the most common formof this disease. A�icted individuals are mentallyretarded and sensitive to sunlight. Wang andcolleagues were able to show that in vitro translatedCSB also binds speci®cally to GST-wild type p53fusion protein (Wang et al., 1995). CSB, like ERCC2and ERCC3 also possesses an unwinding activity.Therefore, binding of p53 to helicases involved in NER(and possibly their inhibition) seems to be a commonfeature of p53, although no mechanistic explanationhas been provided so far. Another p53-binding partnerinvolved in DNA repair is the single-stranded DNAbinding protein RPA. UV irradiation disrupts theRPA-p53 complex, correlating with an activation ofp53, but only in cells which are capable to carry outglobal nucleotide excision repair (Abramova et al.,1997). Since UV-irradiation is a major signal for theactivation and induction of p53, it seems likely thatbinding of p53 to RPA is involved in upstreamregulation of p53-dependent damage response. Com-bined with the fact that a defect in p53 function leadsto a defect in global NER, it may be more thancoincidence that UV-irradiation leads to a disruptionof the binding of RPA to p53 only in cells capable ofglobal NER. A plausible model is that RPA in non-irradiated cells sequesters p53, but releases it afterDNA damage. The released p53 could then act as atranscriptional activator, or/and could directly partici-pate in global NER.

Biochemical activities of p53 related to DNA repair

An important feature of p53 in considering its directinvolvement in DNA repair is its ability to interactwith DNA in many di�erent ways: p53 binds todouble-stranded and single-stranded DNA in a non-sequence-speci®c manner (Steinmeyer and Deppert,1988), to ends of double-strand breaks (Bakalkin etal., 1995), to Holliday junctions (Lee et al., 1997), and

to DNA mismatches leading to DNA-`bulges' (Lee etal., 1995). These activities are important for p53'sability to bind to damaged DNA. p53 thus can `sense'and bind strongly to DNA damaged by ionizingradiation (Reed et al., 1995) and form highly stablecomplexes with insertion/deletion mismatches (Lee etal., 1995). p53 in this respect shows parallels to theDNA repair factor MSH2. Binding to damaged DNAis mediated by the C-terminal domain of p53, whichhas been shown to be important for the regulation ofp53 function. The `sensing' mechanism forms the basisfor a model in which p53 via its C-terminus recognizesDNA damage. p53 then becomes activated forsequence-speci®c DNA binding and for transactiva-tion of target genes, which then participate in andenhance DNA repair (Jayaraman and Prives, 1995).However, this model is confronted with a majorproblem: it is di�cult to envision how p53 will bindtightly to damaged DNA with a half-life of more than2 h (Lee et al., 1995), and at the same time willtransactivate genes which in all likelihood are locatedin a distant part of the genome, i.e. megabases awayfrom the DNA damage locus. In addition, as alreadyindicated above, so far no p53 target genes have beenidenti®ed, which could account for the global repairdefect when p53 is mutated. Therefore we suggest thatp53, in addition to damage recognition, could bedirectly involved in DNA repair processes by severalbiochemical activities, namely by its non-sequence-speci®c DNA binding activity, its DNA reannealingactivity, its ability to promote DNA strand transfer,and its 3'-45' exonuclease activity. Especially the p53intrinsic 3'-45' exonuclease activity, localized withinthe p53 core domain (Mummenbrauer et al., 1996);could be an important player in repair activities of p53.Exonucleases are required for DNA replication, DNArepair and recombination and often enhance the®delity of these processes. As mutant p53 isexonuclease de®cient (Mummenbrauer et al., 1996),and cells expressing a mutant p53 are defective inglobal NER (Ford and Hanawalt, 1995, 1997; Wangand Prives, 1995; Mirzayans et al., 1996), thiscorrelation might point to a possible role of the p53exonuclease activity in DNA repair. The variousbiochemical activities of p53 and its interactions withviral and cellular proteins are summarized and relatedto p53 structure in Figure 1.

Involvement of p53 in control of homologousrecombination

Recombination processes are subject to complexsurveillance mechanisms ensuring high ®delity ofDNA repair and of genetic transmission duringmeiosis (Haber, 1997; Xu et al., 1997). Regulatorycircuits must also exist in order to establish di�erencesin DNA exchange frequencies between distinct genomicloci during meiosis, and in mitotically growing cells,where homologous recombination must be suppressedby a factor of 1000. p53 accumulates and becomesfunctionally activated (Lutzker and Levine, 1996;Siegel et al., 1995) upon the introduction of doublestrand breaks (DSB) into DNA (Nelson and Kastan,1994). DSB arise spontaneously due to errors inreplication, recombination or mitosis and can be

Dual role model for p53 functionN Albrechtsen et al

7710

induced experimentally by ionizing radiation, radio-mimetic drugs, or by the introduction of restrictionendonucleases. DSB trigger both repair associated andtargeted recombination processes, a connection whichin turn renders p53 a likely candidate for being aregulatory factor in DNA exchange processes. Duringthe last years we (WiesmuÈ ller et al., 1996) and others(Mekeel et al., 1997; Bertrand et al., 1997; Honma etal., 1997) demonstrated that p53 suppresses sponta-neous inter- and intrachromosomal homologousrecombination events at least by one to two orders ofmagnitude. The same groups also demonstrated thattranscriptional transactivation and cell cycle controlactivities are separable from p53's involvement inhomologous recombination by comparing the activ-ities of cancer-related p53 mutants and C-terminallytruncated p53 variants in these processes (DudenhoÈ �eret al., 1999; BS Lopez, Paris, and SN Powell, personalcommunication).

Inhibition of p53 recombination control by SV40

SV40 provides a suitable model system for probingrecombination, taking advantage of the smallchromatin-packaged viral genome, which is ampli-®ed episomally, and which promotes high frequencyexchange rates for easy detection of recombinationevents (WiesmuÈ ller et al., 1996; Jasin et al., 1985;

Snapka et al., 1991; Kawasaki et al., 1994). TheSV40 tumor antigen (T-Ag) was found to be thecausative agent for the elevated recombinationfrequencies of cellular and viral DNAs as well asfor the stimulation of the closely related geneampli®cation events (WiesmuÈ ller et al., 1996; Perryet al., 1992; Ishizaka et al., 1995; Cheng et al.,1997). SV40 T-Ag targets p53 in SV40 infection andtransformation by forming a tight complex with p53.A possible role for p53-T-Ag complex formation isthe elimination of a function of p53 in the controlof recombination. By comparing the recombinationrates between SV40 genomes expressing a wild typeT-Ag or a T-Ag containing a point mutation whichspeci®cally abolishes T-Ag-p53 interactions (Wies-muÈ ller et al., 1996), we found that recombinationfrequencies in cells expressing a T-Ag unable to bindp53 were reduced by at least one order ofmagnitude. Therefore we conclude that suppressionof homologous recombination events by wild typep53 can be alleviated by complex formation withSV40 T-Ag (WiesmuÈ ller et al., 1996).

Elimination of p53-mediated suppression of recom-bination by T-Ag might also explain the ®nding thatimmortalization of human ®broblasts after stabletransformation with T-Ag increases chromosomalrecombination. Other viral binding partners, such asHPV16 E6 (Mekeel et al., 1997; Havre et al., 1995),

Figure 1 p53 landmarks. Roman numerals represent the ®ve regions of p53 that are conserved within p53 from all vertebrates.Known phosphorylation (P) and acetylation (Ac) sites are indicated. The vertical bars, clustered in the center of the p53 molecule,indicate amino acid residues mutated in human tumors (hot spots are identi®ed by amino acid number). Shown below and indicatedby horizontal bars is the current information concerning various domains of p53 for biological activities, p53 DNA interactions, andp53-protein complex formation. Abbreviations: CK 2, casein kinase 2; CSB, Cockayne's syndrome B protein; DNA PK, DNAdependent protein kinase; NLS, main nuclear localization signal; RP-A, replication protein A; SV 40, Simian Virus 40; TAF,transcription activating factor; TBP, TATA-Box binding protein; TF, transcription factor; XPB, xeroderma pigmentosum B protein;XPD, xeroderma pigmentosum D protein

Dual role model for p53 functionN Albrechtsen et al

7711

and possibly the genetically destabilizing Hepatitis Bvirus (HBV) X antigen (Feitelson et al., 1993), mayalso neutralize the inhibitory function of p53 inrecombination, thereby allowing a correlation betweenthe targeting of p53 by these proteins and elevatedrecombination.

Mismatch-recognition: A speci®c role for p53 inrecombination control?

We recently discovered that p53 inhibits DNAexchange most dramatically upon recognition ofcertain mismatches within nascent recombinationintermediates (Dudenhoe�er et al., 1998). Parallelsto the functions of classical mismatch repair factorsin recombination arise, as the mammalian MutScounterpart, MSH2, like p53, binds to Hollidayjunctions (Alani et al., 1997) and abolishes DNAexchange between divergent sequences (De Wind etal., 1995). In support of our idea, p53 was shown toequally a�ect homologous recombination betweendirect and indirect repeat substrates, which allowsthe conclusion that p53 controls recombination atthe stage of strand invasion (BS Lopez, personalcommunication). Therefore, it is tempting tospeculate that p53 monitors the ®delity of recombi-nation in concert with the mismatch repair system.MSH2 and p53 do not seem to perform structuralbut rather regulatory functions during DNA-exchange processes. Consistent with the hypothesisof two surveillance pathways acting in parallel, theinactivation of both pathways causes a synergism intumorigenesis, evidence for which was recently givenby the analysis of Msh2 and p53 double knock-outmice (Cranston et al., 1997). Therefore, MSH2 andp53 must be considered to be members of di�erentgenetic epistasis groups involved in the regulation ofrecombination processes. The view is consistent withour biochemical data suggesting complementary®delity control due to an opposing mismatchspeci®city for MSH2-GT-binding protein complexes,which most e�ciently recognize G-T single basemispairings (Hughes and Jiricny, 1992), and for p53-tetramers, which display highest a�nities for A ±Gmismatches (Dudenhoe�er et al., 1998). It remainsto be established, whether p53 upon encounteringmismatches in heteroduplices transmits signals todownstream molecules, such as p21waf1, or/and blocksthe progress of DNA exchange, possibly even byparticipating in error removal through its 3'-45'exonuclease activity. First indications for a p53-mediated signal transduction mechanism coupled torecombination processes can be derived from thefact that the hRad51-complex partners Brca1 andBrca2 physically and genetically interact with p53(StuÈ rzbecher et al., 1996; Ouchi et al., 1998; Zhanget al., 1998; Lim and Hasty, 1996; Hakem et al.,1997; Ludwig et al., 1997), and that Brca1 seems tostimulate p53-dependent transcription, whereas over-expressed Brca2 and hRad51 display antagonistice�ects (Marmorstein et al., 1998). In line with thisproposal is the observation that puri®ed multiproteincomplexes performing homologous recombination invitro contain DNA polymerase e, DNA ligase andlow levels of 5'-43' and 3'-45' exonuclease activities(Jessberger et al., 1993).

Is there a role for p53 in DNA repair in its non-inducedstate?

The transactivation and the exonuclease function of p53seem to be mutually exclusive

The active participation of p53 in DNA repairprocesses might not be restricted to situationsactivating a p53 response, like DNA damage in¯ictedby exogenous factors. On the contrary, expandingp53's role as `guardian of the genome' by endowing itwith a basic function in DNA repair processes in itsnon-induced state instead of being just a non-functional protein should greatly enhance its possibi-lities to preserve the integrity of the genome. Actually,some inconsistencies regarding the role of the waf1gene as a major target in the p53 response to DNAdamage already point to such a basic function. Ifindeed p53-mediated induction of the waf1 gene, andthe ensuing growth arrest and stop of PCNA-dependent DNA replication, would play a major rolein p53's activities to ensure repair of DNA lesions, onewould have to postulate that waf1 nullizygote miceshould display a similar phenotype regarding tumorpredisposition as p53 null mice. However, this is notthe case, despite the ®nding that cells from waf1nullizygote mice have largely lost the ability to arrest inG1 after DNA damage (Deng et al., 1995). Therefore,we assume that some other function of p53 ispreventing the accumulation of mutations in thesemice. We propose that such a function could be the`basic function' of p53, preventing the accumulation ofmutations by its involvement in the repair ofendogenous DNA damage.

With the exception of sequence-speci®c transactiva-tion of p53 target genes, which seems to be dependenton post-translational modi®cation of p53 after DNAdamage, all of its other activities outlined above could,in principle, be exerted also by a non-induced p53.Possible limitations would only arise from the lowamounts of p53 usually present in normal cells.However, even this argument is questionable, as ithas been demonstrated that the transactivationfunction of p53 can be induced without a detectablerise in p53 levels (Hupp et al., 1995). Although there isno in vivo evidence for such a `basic' function of p53,comparative analysis of the major biochemical activ-ities exerted by the p53 core domain, sequence-speci®cDNA binding and 3'-5' exonuclease activity (Figure 1),prompted us to suggest a direct participation of p53 inthe repair of endogenous DNA damage, in preventingfaulty DNA repair, and possibly also in DNAreplication.

As it seems rather unlikely that p53 will act as anexonuclease and simultaneously bind to DNA in asequence-speci®c manner, the localization of the 3'-45'exonuclease activity of p53 to the same domain thatexerts sequence-speci®c DNA binding poses theproblem of how these activities are regulated. Duringdeletion mapping analyses we found that C-terminaltruncation of the C-terminal thirty amino acids of thep53 molecule activated its exonuclease activity byabout a factor of ten, indicating that the C-terminalregulatory domain of p53 negatively in¯uences the p53exonuclease activity. The same, however, has beenshown for the sequence-speci®c DNA binding activity

Dual role model for p53 functionN Albrechtsen et al

7712

of p53, underscoring the importance of the p53 C-terminus for the regulation of p53 functions. Yet,whereas both activities were negatively regulated by thep53 C-terminus, treatments which a�ected the regula-tion of p53 activities by the p53 C-terminus hadopposing e�ects on sequence-speci®c DNA binding andthe exonuclease activity of p53: treatments, whichactivated sequence-speci®c DNA binding of p53, likeaddition of the monoclonal antibody PAb421 whichbinds to an epitope within the C-terminal regulatorydomain, or enhancing the phosphorylation status ofp53, strongly inhibited the p53 exonuclease activity(Janus et al., 1999). This ®nding points to the excitingpossibility that p53 can exist in at least two di�erentfunctionally active states. Since activation of sequence-speci®c DNA binding is considered to be a hallmark ofp53 activation after DNA damage, we conclude thatp53 looses its exonuclease activity when the sequence-speci®c DNA binding of p53 becomes activated afterDNA damage. Conversely, however, one can alsoconclude that non-activated p53 exerts exonucleaseactivity, implying that a non-activated p53 is not equalto a non-functional one.

Possible functions of the p53 exonuclease activity

What could be the role of the p53 exonuclease activityin DNA repair? Exonucleases are required for nearlyall processes of DNA metabolism, such as DNAreplication, long patch DNA repair, postreplicativemismatch repair, and DNA recombination. Animportant type of error avoidance mechanism requir-ing exonuclease activities is the mismatch repairpathway. In bacteria this system invokes the coordi-nated action of the MutSLH damage recognition/endonuclease complex along with the UvrD helicase,DNA polymerase III, DNA ligase, single-strand DNAbinding protein, and any one of the exonucleasesExoI(3'-45' exo), ExoVII (3'-45' and 5'-43' exo) orRecJ (5'-43' exo) (Modrich, 1987, 1989). An involve-ment of p53's exonuclease in the postreplicativemismatch repair pathway would be of particularinterest because this would link one of the enzymolo-gical functions of p53, i.e., its exonuclease activity, tothe tumor suppressor function of the mammalianMutSLH homologues hMSH2, hMLH1, hPMS1, andhPMS2. A role complementary to these mismatchrecognition proteins has already been proposed for p53in mismatch recognition during recombination events(see above). Functional loss of either of these proteinsleads to an increased incidence of colon carcinomas(for reviews see Bodmer et al., 1994; Modrich, 1994;Radman and Wagner, 1993). Furthermore, mutationsin any of these genes display a generalized increase inspontaneous mutation rates, a replication error positive(RER+) phenotype, and a resistance to alkylatingagents (Bhattacharyya et al., 1994; Leach et al., 1993;Parsons et al., 1993). In this respect, it might be morethan coincidence that p53-de®cient animals show arelative resistance to alkylating agents (Harvey et al.,1993). Moreover, there is an inverse correlationbetween RER+ status and p53 mutation in colorectalcancer cell lines (Cottu et al., 1996), pointing to asynergistic action of gene products involved inmismatch repair with p53. This synergism is furthercorroborated by the ®nding that male mice nullizygous

for both Msh2 and p53 develop tumors signi®cantlyearlier than either of the single mutants (Cranston etal., 1997). Furthermore, there is some evidence that,like the p53 protein itself (see above), an intactmammalian MutSHL system displays an `anti-recom-binogenic' e�ect (Fishel and Kolodner, 1995). In linewith the possibility of a direct involvement of p53 inmismatch repair is the recent demonstration that wtp53 can `sense' DNA mismatch lesions by tightlybinding to DNA containing insertion/deletion mis-matches (Lee et al., 1995).

Another possibility is p53's involvement in erroravoidance by contributing a certain type of `proof-reading' activity. There are at least six di�erent cellularDNA polymerases designated as DNA polymerases a,b, g, d, e, and z that are involved in di�erent aspects ofDNA synthesis. Two of these, namely DNA poly-merases a and b, are devoid of a 3'-45' exonucleaseactivity that excises mismatched nucleotides immedi-ately after the incorporation step. Therefore, DNApolymerases a and b are particularly error prone anduntil now it is not known, how mismatched nucleotidesincorporated by these two polymerases are removed(Kunkel, 1992). Moreover, both polymerases have astrong tendency to introduce frame-shift mutations bymisinsertion or deletion of nucleotides (Huang, 1998).Since p53 recognizes DNA bulges caused by insertion/deletion mismatches (Lee et al., 1995), it might beparticularly suited to excise them via its 3'-45'exonuclease. Last but not least, p53 might act as adirect proofreader for the proofreader de®cient DNApolymerases a and b.

A model for p53 function in the maintenance of geneticintegrity by DNA repair

Di�erent subclasses of p53 may perform di�erentfunctions within the same cell

An important aspect for evaluating the activities of p53in its induced and in its non-induced state is that notnecessarily all p53 molecules must be in the samefunctionally active state, i.e. must exert the samefunction. Functional heterogeneity of multifunctionalregulatory proteins is well known. Probably one of thebest documented examples is the SV40 T-Ag, the majorregulatory protein of SV40. Analyses of its subcellularlocalization and of its biochemical activities in SV40lytically infected cells clearly demonstrated the coex-istence of several functionally di�erent subclasses of T-Ag molecules, involved in various aspects of viralreplication (reviewed in Deppert and Schirmbeck,1995). Assuming a similar situation for the multi-functional p53 molecule, one could envision that evenafter DNA damage only a certain fraction of the p53molecules becomes activated for sequence-speci®ctranscription, whereas other molecules remain in theirnon-induced state. Speci®city for selective activation(or non-activation) of p53 subclasses can be providedby subcellular compartmentalization of the p53 proteinand/or by its association with di�erent cellularproteins. The existence of functionally di�erentsubclasses of p53 within the same cell so far has notbeen proven. However, previous experiments from thislaboratory at least demonstrated the presence of p53 in

Dual role model for p53 functionN Albrechtsen et al

7713

di�erent nuclear compartments (Deppert and Haug,1986). Therefore, it is possible that the p53 populationnot engaged in transcriptional regulation could exertfunctions other than induction of growth arrest orapoptosis, and directly participate in processes of DNArepair via its various biochemical activities describedabove. The exonuclease activity, for example, could beinvolved in repair processes such as DSB repair, whichis thought to require DNA helicases, like the Kuautoantigen (Suwa et al., 1994; Tuteja et al., 1994), andexonucleases. Furthermore, p53 might act as anexternal proofreader for errors produced by cellularDNA polymerases involved in DNA replication andDNA repair also under DNA damage conditions.DNA polymerase a is certainly involved in nuclearDNA replication (Wang, 1991), but also has somefunction in DNA repair (Saxena et al., 1990; Oda etal., 1996; Holmes and Haber, 1999), whereas DNApolymerase b has been solely assigned to DNA repairprocesses (Wood and Shivji, 1997). When DNAdamage has occurred, there is an apparent shut-o� ofPCNA-dependent DNA replication (Wood and Shivji,1997; Flores-Rozas et al., 1994). Although PCNA-dependent DNA-repair synthesis still might continueunder conditions of p21waf1 induction (Li et al., 1994a,see above), DNA-repair synthesis could also beperformed by the PCNA-independent and proofrea-der-free DNA polymerases a and/or b, for which p53could provide the proof-reader function. If oneassumes that after damage di�erent functional sub-classes of p53 will exist within the same cell, then theensuing increase of p53 protein levels not only will

activate the potential of p53 to transcribe p53 targetgenes, leading to growth arrest and to shut-o� ofPCNA-dependent DNA replication, but will alsoincrease the amount of p53 with a 3'-45' proofreadingexonuclease activity. Such p53 then could enhance theaccuracy of DNA repair synthesis performed by themore error-prone DNA polymerases a and b.

The `dual role' model for p53 The major conclusionwhich can be drawn from the above considerations isthat the current picture of p53 as a guardian of thegenome solely activated by cellular stress and/or DNAdamage will be an oversimpli®cation. We assume that anon-induced p53 is not a protein without function. p53in its non-induced state at least performs exonucleaseactivity, which could be involved in a variety ofpossible functions of p53 in DNA repair which allcontribute to avoid mutations in the genome, asoutlined above. We consider that the basic functionof p53 is the repair of endogenous DNA damage, andthe prevention of mutational events resulting from suchdamage. Superimposed are the up to now bettercharacterized functions of p53 as a superior controlelement in integrating cellular stress signals, followedby the induction of either a growth arrest or ofapoptosis. This model of a dual role of p53 as aguardian of the genome is outlined in Figure 2. Ifinduction of p53 will activate only a fraction of p53 forsequence-speci®c transcription, the `basic function' ofp53 must not be restricted to p53 in its non-inducedstate, and other, non-induced subclasses of p53 coulddirectly engage in the various repair processes outlined

Figure 2 Dual role of p53 as guardian of the genome. The current view of p53 as a superior control element in the responses tovarious types of cellular stress is depicted in the right half of the ®gure (adapted from Hall et al., 1996). The left half provides anextension of this model, according to which a non-induced p53 actively participates in the prevention of mutations arising fromendogenous DNA damage. Note that functions of a non-induced p53 can also operate under conditions of cellular stress, if oneassumes that after cellular stress di�erent functional subclasses can exist within the same cell (for details see text)

Dual role model for p53 functionN Albrechtsen et al

7714

above. Thereby p53, after DNA damage then would beable to exert its full range of possible biochemicalactivities.

Clearly, this model is highly speculative, but testable.For example, the postulated proofreader activity of p53predicts a functional association of an exonuclease-active p53 with the proofreader de®cient polymerases aand/or û. Biosensor studies indeed suggest that p53binds to both polymerases with considerable a�nity(KuÈ hn et al., 1998), and a speci®c complex betweenp53 and an immunologically distinct subclass ofpolymerase a could be detected in vivo (I Dornreiter,unpublished observation). Moreover, pol a can bea�nity-puri®ed together with p53 from monoclonalantibody columns directed against pol a (F Grosse,unpublished observations). Furthermore, Huang (1998)recently showed that wt p53, but not mutant p53,

speci®cally enhanced the replication ®delity of poly-merase a in an in vitro replication assay, stronglysupporting the idea that p53 can act as an exogenousproofreader for this replicase. However, furtherexperimentation is necessary to con®rm and extendthese initial ®ndings.

AcknowledgmentsWork from this laboratory was supported by the DrMildred Scheel Stiftung (Deutsche Krebshilfe), theDeutsche Forschungsgemeinschaft (DFG) and the Fondsder Chemischen Industrie. The Heinrich-Pette-Institut is®nancially supported by the Freie und Hansestadt Ham-burg and by the Bundesministerium fuÈ r Gesundheit. Partsof this review have been published in Janus et al., CMLS,55 (1999), 12 ±27.

References

Abramova NA, Russell J, Botchan M and Li R. (1997). Proc.Natl. Acad. Sci. USA, 94, 7186 ± 7191.

Adams PD, Sellers WR, Sharma SK, Wu AD, Nalin CM andDaelin WGJ. (1996). Mol. Cell. Biol., 16, 6623 ± 6633.

Agarwal ML, Agarwal A, Taylor WR and Stark GR. (1995).Proc. Natl. Acad. Sci. USA, 92, 8493 ± 8497.

Aitken A, Amess B, Howell S, Jones D, Martin H, Patel Y,Robinson K and Toker A. (1992). Biochem. Soc. Trans.,20, 607 ± 611.

Aitken A, Howell S, Jones D, Madrazo J, Martin H, Patel Yand Robinson K. (1995).Mol. Cell. Biochem., 149, 41 ± 49.

Ait-Si-Ali S, Ramirez S, Barre FX, Dkhissi F, Magnaghi-Jaulin L, Girault JA, Robin P, Knibiehler M, PritchardLL, Ducommun B, Trouche D and Harel-Bellan A.(1998). Nature, 396, 184 ± 186.

Alani E, Lee S, Kane MF, Gri�th J and Kolodner RD.(1997). J. Mol. Biol., 265, 289 ± 301.

Aloni-Grinstein R, Schwartz D and Rotter V. (1995). EMBOJ., 14, 1392 ± 1401.

Avantaggiati ML, Ogryzko V, Gardner K, Giordano A,Levine AS and Kelly K. (1997). Cell, 89, 1175 ± 1184.

Bakalkin G, Selivanova G, Yakovleva T, Kiseleva E,Kashuba E, Magnusson KP, Szekely L, Klein G,Terenius L and Wiman KG. (1995). Nucl. Acids Res., 23,362 ± 369.

Bannister AJ and Kouzarides T. (1996). Nature, 384, 641 ±643.

Bargonetti J, Manfredi JJ, Chen X, Marshak DR and PrivesC. (1993). Genes Dev., 7, 2565 ± 2574.

Bates S and Vousden KH. (1999). CMLS, 55, 28 ± 37.Bertrand P, Rouillard D, Boulet A, Levalois C, Soussi T and

Lopez BS. (1997). Oncogene, 14, 1117 ± 1122.Bhattacharyya NP, Skandalis A, Ganesh A, Groden J and

Meuth M. (1994). Proc. Natl. Acad. Sci. USA, 91, 6319 ±6323.

Bodmer W, Bishop T and Karran P. (1994). Nat. Genet., 6,217 ± 219.

Brugarolas J, Chandrasekaran C, Gordon JI, Beach D, JacksT and Hannon GJ. (1995). Nature, 377, 552 ± 557.

Caelles C, Helmberg A and Karin M. (1994). Nature, 370,220 ± 223.

Cheng RZ, Shammas MA, Li J and Reis RJS. (1997). Exp.Cell Res., 234, 300 ± 312.

Cottu PH, Muzeau F, Estreicher A, Flejou JF, Iggo R,Thomas G and Hamelin R. (1996). Oncogene, 13, 2727 ±2730.

Cox LS and Lane DP. (1995). BioEssays, 17, 501 ± 508.Cranston A, Bocker T, Reitmair A, Palazzo J, Wilson T,

Mak T and Fishel R. (1997). Nature Genet., 17, 114 ± 118.

De Wind N, Dekker M, Berns A, Radman M and Riele HT.(1995). Cell, 82, 321 ± 330.

Deng C, Zhang P, Harper JW, Elledge SJ and Leder P.(1995). Cell, 82, 675 ± 684.

Deppert W. (1994). Cancer Biol., 5, 187 ± 202.Deppert W and Haug M. (1986). Mol. Cell. Biol., 6, 2233 ±

2240.Deppert W and Schirmbeck R. (1995). Int. Rev. Cytol., 162,

486 ± 537.Dudenhoe�er C, Rohaly G, Will K, Deppert W and

Wiesmueller L. (1998). Mol. Cell. Biol., 18, 5332 ± 5342.DudenhoÈ �er C, Kurth M, Janus F, Deppert W and

WiesmuÈ ller L. (1999). Oncogene, (In press).Dulic V, Kaufmann WK, Wilson SJ, Tlsty TD, Lees E,

Harper JW, Elledge SJ and Reed SI. (1994). Cell, 76,1013 ± 1023.

Dulic V, Stein GH, Far DF and Reed SI. (1998). Mol. Cell.Biol., 18, 546 ± 557.

El-Deiry WS, Kern SE, Pietenpol JA, Kinzler KW andVogelstein B. (1992). Nature Genet., 1, 45 ± 49.

El-Diery WS, Tokino T, Velculescu VE, Levy DB, ParsonsR, Trent JM, Lin D, Mercer WE, Kinzler KW andVogelstein B. (1993). Cell, 75, 817 ± 825.

Eller MS, Maeda T, Magnoni C, Atwal D and Gilchrest BA.(1997). Proc. Natl. Acad. Sci. USA, 94, 12627 ± 12632.

Feitelson MA, Zhu M, Duan LX and London WT. (1993).Oncogene, 8, 1109 ± 1117.

Fishel R and Kolodner RD. (1995). Curr. Opin. Genet. Dev.,5, 382 ± 395.

Flores-Rozas H, Kelman Z, Dean FB, Pan ZQ, Harper JW,Elledge SJ, O'Donnell M and Hurwitz J. (1994). Proc.Natl. Acad. Sci. USA, 91, 8655 ± 8569.

Foord O, Navot N and Rotter V. (1993). Mol. Cell. Biol., 13,1378 ± 1384.

Ford JC, Al-Khodairy F, Fotou E, Sheldrick KS, Gri�thsDJ and Carr AM. (1994). Science, 265, 533 ± 535.

Ford JM and Hanawalt PC. (1995). Proc. Natl. Acad. Sci.USA, 92, 8876 ± 8880.

Ford JM and Hanawalt PC. (1997). J. Biol. Chem., 272,28073 ± 28080.

Fritsche M, Haessler C and Brandner G. (1993).Oncogene, 8,307 ± 318.

Furnari B, Rhind N and Russell P. (1997). Science, 277,1495 ± 1497.

Garkavtsev I, Grigorian IA, Ossovskaya VS, Chernov MV,Chumakov PM and Gudkov AV. (1998). Nature, 391,295 ± 298.

Dual role model for p53 functionN Albrechtsen et al

7715

Gobert C, Bracco L, Rossi F, Olivier M, Tazi J, Lavelle F,Larsen AK and Riou JF. (1996). Biochemistry, 35, 5778 ±5786.

Goi K, Takagi M, Iwata S, Delia D, Asada M, Donghi R,Tsunematsu Y, Nakazawa S, Yamamoto H, Yokota J,Tamura K, Saeki Y, Utsonomiya J, Takahashi T, Ueda R,Ishioka C, Eguchi M, Kamata N and Mizutani S. (1997).Cancer Res., 57, 1895 ± 1902.

Gottlieb TM and Oren M. (1996). Biochim. Biophys. ActaRev. Cancer, 1287, 77 ± 102.

Graeber TG, Peterson JF, Tsai M, Monica K, Fornace AJJand Giaccia AJ. (1994). Mol. Cell. Biol., 14, 6264 ± 6277.

Gu W and Roeder RG. (1997). Cell, 90, 595 ± 606.Guadagno TM and Newport JW. (1996). Cell, 84, 73 ± 82.Haber JE. (1997). Cell, 89, 163 ± 166.Hakem R, de la Pompa JL, Elia A, Potter J and Mak TW.

(1997). Nature Genet., 16, 298 ± 302.Hall PA, Meek D and Lane DP. (1996). J. Pathol., 180, 1 ± 5.Harper JW, Adami GR, Wei M, Keyomarsi K and Elledge

SJ. (1993). Cell, 75, 805 ± 816.Harvey M, McArthur MJ, Montgomery JR CA, Butel JS,

Bradley A and Donehower LA. (1993). Nature Genet., 5,225 ± 229.

Haupt Y, Rowan S, Shaulian E, Vousden KH and Oren M.(1995). Genes Dev., 9, 2170 ± 2183.

Havre PA, Yuan JL, Hedrick L, Cho KR and Glazer PM.(1995). Cancer Res., 55, 4420 ± 4424.

Hecker D, Page G, Lohrum M, Weiland S and ScheidtmannKH. (1996). Oncogene, 12, 953 ± 961.

Hermeking H, Lengauer C, Polyak K, He TC, Zhang L,Thiagalingan S, Kinzler KW and Vogelstein B. (1997).Mol. Cell., 1, 3 ± 11.

Holmes AM and Haber JE. (1999). Cell, 96, 415 ± 424.Honma M, Zhang LS, Hayashi M, Takeshita K, Nakagawa

Y, Tanaka N and Sofuni T. (1997). Mol. Cell. Biol., 17,4774 ± 4781.

Huang P. (1998). Oncogene, 17, 261 ± 270.Hughes MJ and Jiricny J. (1992). J. Biol. Chem., 267, 23876 ±

23882.Hupp TR and Lane DP. (1994). Curr. Biol., 4, 865 ± 875.Hupp TR, Meek DW, Midgley CA and Lane DP. (1992).

Cell, 71, 875 ± 886.Hupp TR, Sparks A and Lane DP. (1995). Cell, 83, 237 ± 245.Hwang BJ, Ford JM, Hanawalt PC and Chu G. (1999). Proc.

Natl. Acad. Sci. USA, 96, 424 ± 428.Ishizaka Y, Chernov MV, Burns CM and Stark GR. (1995).

Proc. Natl. Acad. Sci. USA, 92, 3224 ± 3228.Janus F, Albrechtsen N, Knippschild U, WiesmuÈ ller L,

Grosse F and Deppert W. (1999). Mol. Cell. Biol., 19,2155 ± 2168.

Jasin M, de Villiers J, Weber F and Scha�ner W. (1985). Cell,43, 695 ± 703.

Jayaraman L, Moorthy NC, Murthy KGK, Manley JL,Bustin M and Prives C. (1998). Genes Dev., 12, 462 ± 472.

Jayaraman L and Prives C. (1995). Cell, 81, 1021 ± 1029.Jayaraman L and Prives C. (1999). CMLS, 55, 76 ± 87.Jessberger R, Podust V, HuÈ bscher U and Berg P. (1993). J.

Biol. Chem., 268, 15070 ± 15079.Kastan MB, Onyekwere O, Sidransky D, Vogelstein B and

Craig RW. (1991). Cancer Res., 51, 6304 ± 6311.Kastan MB, Zhan Q, El-Deiry WD, Carrier F, Jacks T,

Walsh WV, Plunkett BS, Vogelstein B and Fornace JR AJ.(1992). Cell, 71, 587 ± 597.

Kawasaki I, Bae YS, Eki T, Kim Y and Ikeda H. (1994).Mol.Cell. Biol., 14, 4173 ± 4182.

Kim E, Albrechtsen N and Deppert W. (1997). Oncogene, 15,857 ± 869.

Ko LJ and Prives C. (1996). Genes Dev., 10, 1054 ± 1072.Ko LJ, Shieh SY, Chen XB, Jayaraman L, Tamai K, Taya Y,

Prives C and Pan ZQ. (1997). Mol. Cell. Biol., 17, 7220 ±7229.

KuÈ hn C, MuÈ ller F, Melle C, Nasheuer H-P, Janus F, DeppertW and Grosse F. (1998). Oncogene, 18, 769 ± 774.

Kunkel TA. (1992). J. Biol. Chem., 267, 18251 ± 18254.Lambert PF, Kashanchi F, Radonovich MF, Shiekhattar R

and Brady JN. (1998). J. Biol. Chem., 273, 33048 ± 33053.Lane DP. (1992). Nature, 358, 15 ± 16.Lane DP and Hall PA. (1997). Trends Biochem. Sci., 22,

372 ± 374.Leach FS, Nicolaides NC, Papadopoulos N, Liu B, Jen J,

Parsons R, Peltomaki P, Sistonen P, Aaltonen LA andNystrom-Lahti M. (1993). Cell, 75, 1215 ± 1225.

Lee S, Elenbaas B, Levine A and Gri�th J. (1995). Cell, 81,1013 ± 1020.

Lee SM, Cavallo L and Gri�th J. (1997). J. Biol. Chem., 272,7532 ± 7539.

Levedakou EN, Kaufmann WK, Alcorta DA, Galloway DAand Paules RS. (1995). Cancer Res., 55, 2500 ± 2502.

Leveillard T, Andera L, Bissonnette N, Schae�er L, BraccoL, Egly JM and Wasylyk B. (1996). EMBO J., 15, 1615 ±1624.

Levine AJ. (1997). Cell, 88, 323 ± 331.Li R, Waga S, Hannon GJ, Beach D and Stillman B. (1994a).

Nature, 371, 534 ± 537.Li Y, Jenkins CW, Nichols MA and Xiong Y. (1994b).

Oncogene, 9, 2261 ± 2268.Lim D-S and Hasty P. (1996). Mol. Cell. Biol., 16, 7133 ±

7143.Linke SP, Clarkin KC, Di Leonardo A, Tsou A and Wahl

GM. (1996). Genes Dev., 10, 934 ± 947.Liu L, Scolnick DM, Trievel RC, Zhang HB, Marmorstein

R, Halazonetis TD and Berger SL. (1999).Mol. Cell. Biol.,19, 1202 ± 1209.

Ludwig T, Chapman DL, Papaioannou VE and EfstratiadisA. (1997). Genes Dev., 11, 1226 ± 1241.

Lutzker SG and Levine AJ. (1996). Nature Med., 2, 804 ±810.

Maltzman W and Czyzyk L. (1984). Mol. Cell. Biol., 4,1689 ± 1694.

Marmorstein LY, Ouchi T and Aaronson SA. (1998). Proc.Natl. Acad. Sci. USA, 95, 13869 ± 13874.

Martinez JD, Craven MT, Joselo� E, Milczarek G andBowden GT. (1997). Oncogene, 14, 2511 ± 2520.

Meek DW. (1994). Cancer Biol., 5, 203 ± 210.Mekeel KL, Tang W, Kachnic LA, Luo CM, DeFrank JS

and Powell SN. (1997). Oncogene, 14, 1847 ± 1857.Melendy T and Stillman B. (1991). J. Biol. Chem., 266,

1942 ± 1949.Mirzayans R, Enns L, Dietrich K, Barley RDC and Paterson

MC. (1996). Carcinogenesis, 17, 691 ± 698.Modrich P. (1987). Annu. Rev. Biochem., 56, 435 ± 466.Modrich P. (1989). J. Biol. Chem., 264, 6597 ± 6600.Modrich P. (1994). Science, 266, 1959 ± 1960.Mummenbrauer T, Janus F, Mueller B, Wiesmueller L,

Deppert W and Grosse F. (1996). Cell, 85, 1089 ± 1099.Nelson WG and Kastan MB. (1994). Mol. Cell. Biol., 14,

1815 ± 1823.Nickerson JA. (1998). J. Cell. Biochem., 70, 172 ± 180.Oda N, Saxena JK, Jenkins TM, Prasad R, Wilson SH and

Ackerman EJ. (1996). J. Biol. Chem., 271, 13816 ± 13820.Ogryzko VV, Schiltz RL, Russanova V, Howard BH and

Nakatami Y. (1996). Cell, 87, 953 ± 959.Ouchi T, Monteiro ANA, August A, Aaronson SA and

Hanafusa H. (1998). Proc. Natl. Acad. Sci. USA, 95,2302 ± 2306.

Parsons R, Li GM, Longley MJ, Fang WH, PapadopoulosN, Jen J, Chapelle ADL, Kinzler KW, Vogelstein B andModrich P. (1993). Cell, 75, 1227 ± 1236.

Paules RS, Levedakou EN, Wilson SJ, Innes CL, Rhodes N,Tisty TD, Galloway DA, Donehower LA, Tainsky MAand Kaufmann WK. (1995). Cancer Res., 55, 1763 ± 1773.

Dual role model for p53 functionN Albrechtsen et al

7716

Peng CY, Graves PR, Thoma RS, Wu Z, Shaw AS andPiwnica-Worms H. (1997). Science, 277, 1501 ± 1505.

Perry ME, Commane M and Stark GR. (1992). Proc. Natl.Acad. Sci. USA, 89, 8112 ± 8116.

Pines J. (1991). Cell Growth Di�er., 2, 305 ± 310.Poon RYC, Jiang W, Toyoshima H and Hunter T. (1996). J.

Biol. Chem., 271, 13283 ± 13291.Radman M and Wagner R. (1993). Nature, 366, 722.Reed M, Woelker B, Wang P, Wang Y, Anderson ME and

Tegtmeyer P. (1995). Proc. Natl. Acad. Sci. USA, 92,9455 ± 9459.

Sakaguchi K, Herrera JE, Saito S, Miki T, Bustin M,Vassilev A, Anderson CW and Appella E. (1998). GenesDev., 12, 2831 ± 2841.

Sanchez Y, Wong C, Thoma RS, Richman R, Wu Z,Piwnica-Worms H and Elledge SJ. (1997). Science, 277,1497 ± 1501.

Saxena JK, Hays JB and Ackerman EJ. (1990). Nucl. AcidsRes., 18, 7425 ± 7432.

Schneider E, Montenarh M andWagner P. (1998).Oncogene,17, 2733 ± 2741.

Schwartz D, Almog N, Peled A, Gold®nger N and Rotter V.(1997). Oncogene, 15, 2597 ± 2607.

Shieh SY, Ikeda M, Taya Y and Prives C. (1997). Cell, 91,325 ± 334.

Siegel J, Fritsche M, Mai S, Brandner G and Hess RD.(1995). Oncogene, 11, 1363 ± 1370.

Smith ML, Chen I-T, Zhan Q, O'Connor PM and FornaceJR AJ. (1995). Oncogene, 10, 1053 ± 1059.

Snapka RM, Shin C-G, Permana P and Strayer J. (1991).Nucl. Acids Res., 19, 5065 ± 5072.

Steinmeyer K and Deppert W. (1988). Oncogene, 3, 501 ±507.

Stewart N, Hicks GG, Paraskevas F and Movat M. (1995).Oncogene, 10, 109 ± 115.

StuÈ rzbecher H-W, Donzelmann B, Henning W, KnippschildU and Buchhop S. (1996). EMBO J., 15, 1992 ± 2002.

Suwa A, Hirakata M, Takeda Y, Jesch SA, Mimori T andHardin JA. (1994). Proc. Natl. Acad. Sci. USA, 91, 6904 ±6908.

Takenaka I, Morin F, Seizinger BR and Kley N. (1995). J.Biol. Chem., 270, 5405 ± 5411.

Troelstra C, Odijk H, Wit JD, Westerveld A, Thompson LH,Bootsma D and Hoeijmakers JH. (1990). Mol. Cell. Biol.,10, 5806 ± 5813.

Tuteja N, Tuteja R, Ochem A, Taneja P, Huang NW,Simoncsits A, Susic S, Rahman K, Marusic L and Chen J.(1994). EMBO J., 13, 4991 ± 5001.

Vogelstein B and Kinzler KW. (1992). Cell, 70, 523 ± 526.Waga S, Hannon GJ, Beach D and Stillman B. (1994).

Nature, 369, 574 ± 578.Wang TS. (1991). Annu. Rev. Biochem., 60, 513 ± 552.Wang W and Shakes DC. (1996). J. Mol. Evol., 43, 384 ± 398.Wang XW, Yeh H, Schae�er L, Roy R, Moncollin V, Egly J-

M, Wang Z, Friedberg EC, Evans MK, Ta�e BG, BohrVA, Weeda G, Hoeijmakers JHJ, Forrester K and HarrisCC. (1995). Nature Genet., 10, 188 ± 195.

Wang Y and Prives C. (1995). Nature, 376, 88 ± 91.WiesmuÈ ller L, Cammenga J and Deppert W. (1996). J. Virol.,

70, 737 ± 744.Wood RD and Shivji MK. (1997). Carcinogenesis, 18, 605 ±

610.Xiao H, Pearson A, Coulombe B, Truant R, Zhang S, Regier

JL, Trizenberg SJ, Reinberg D, Flores O and Ingles CJ.(1994). Mol. Cell. Biol., 14, 7013 ± 7024.

Xiong Y, Hannon GJ, Zhang H, Casso D, Kobayashi R andBeach D. (1993). Nature, 366, 701 ± 704.

Xu L, Weiner BM and Kleckner N. (1997). Genes Dev., 11,106 ± 118.

Zhang HB, Somasundaram K, Peng Y, Tian H, Zhang HX,Bi DK, Weber BL and EI-Deiry WS. (1998). Oncogene, 16,1713 ± 1721.

Zlatanova J and van Holde K. (1998). FASEB J., 12, 421 ±431.

Dual role model for p53 functionN Albrechtsen et al

7717