CL-8¡513Bri'N-sh Joumal o[ Surgery 1998, 85,1460-1467

Review

The p53 tumour suppressor geneR. J. C. STEELE, A. M. THOMPSON, P. A. HALL and D. P. LANE

Dundee Cancer Research Institute, Dundee Teaching Hospitals Trust and University of Dundee, Dundee, UKCo"espondence lo: Professor R. J. C. Steele, Department of Surgery, Ninewells Hospital and Medical School, Dundee DDl 9S:i; UK

Background Abnormalities of the p53 tumour suppressor gene are thought to be central to thedevelopment of a high proportion of human tumours. This article reviews current understanding ofits function and potential clinical significance.

Methods Material was identified from previous review articles, references cited in original papers, aMedline search of the literature over the 12 months to January 1998, and by scanning the latestissues of relevant joumals.

Results and conclusion p53 is considered to be a stress response gene, its product (the p53 protein)acting to induce cell cycle arrest or apoptosis in response to DNA damage, thereby maintaininggenetic stability in the organismo These functions are executed by a complex and incompletelyunderstood series of steps known as the 'p53 pathway', part of which involves induction of theexpression of a number of other genes. As p53 is the most cornrnonly mutated gene in humancancer, it has attracted a great deal of interest as a prognostic factor, diagnostic tool andtherapeutic target. However, despite many promising studies, li§.-Potential in practical~~~agement has still to be realizeL

In recent years the rather prosaic term 'p53' has reachedthe attention of most clinicians but many have only thehaziest notion of what it represents. Molecular biologistsnow believe that the p53 protein has a major Tole incellular function and homoeostasis, and that defects in thesystem to which it is central occur in most if not allhuman cancers. Its importance is reflected in ayer 9000p53-related papers published since 1992. The purpose ofthis article is to provide an explanation of currentunderstanding of the physiological and clinical significanceof p53. Given the size of the literature an exhaustivereview has not been attempted; references have beenrestricted to key articles and representative exampleswhere work has been replicated.

Oncogenes and tumour suppressor genesThe development of cancel is now seen as a complex,multi-step process which depends on both externalcarcinogenic influences and subcellular genetic defects.The genetic defects may be caused directly by mutageniccarcinog~ns, but they may also be inherited or may occursporadically (perhaps induced by background radiation).Indeed, not all carcinogenic stimuli produce mutations;they may merely enhance cellular proliferation or survivalsuch that the likelihood of a dangerous mutationoccurring and persisting is increased. It is generallyaccepted, however, that genetic mutations are necessarybefore cancer can arise1.

The genes that are associated with the development ofmalignancy when dysfunctional are broadly categorized asoncogenes, or tumour suppressor genes. Although thisclassification may be imperfect, it is a useful means ofthinking about the genetic basÍs of cancer.

OncogenesOncogenes were first identified when it was realized thatthe tumorigenicity of many retroviruses could beattributed to specific genes, and the first of the&e to becloned was v-src from the Rous sarcoma virus whichcauses sarcomas in chickens. It was then discovered thatchicken DNA contains a very clase relative of v-src, andthat similar versions of the same gene are present in theDNA of other vertebrates2,3. Thus it was realized that theretroviral genes which can transform normal cells intocancer cells (i.e. oncogenes) are actually derived fromnormal cellular genes. These normal genes are nowknown as proto-oncogenes, and may become oncogeneseither by incorporation of a retrovirus into the geneticmaterial or, more commonly, by mutation at their normalsite of residence within the cellular DNA.

Because an oncogene by definition confers malignantproperties anta a cell, mutation of a proto-oncogenegenerally results in gain of function; this may occur byamplification where the affected gene overproduces aprotein that drives cell proliferation or enhances survival,or it may occur by production of a mutant protein whichescapes control mechanisms that normally constrain itsproliferative activity. It follows that proto-oncogenesencode proteins that stimulate cellular growth or survival.These may be broadly categorized as growth factors,growth factor receptors, intracellular signal transducers(which transmit the signal from an activated receptor tothe nucleus) and transcription factors (which induceprotein synthesis by stimulating the DNA in the nucleusto produce messenger RNA).

An oncogene (i.e. a mutated proto-oncogene) typicallyacts as a dominant gene, and so a mutation in one allelewill be sufficient for it to become manifest. However, withfew exceptions4, oncogenes are not inherited and usuallycontribute to the pathogenesis of cancer by somaticmutations within the cells of the target tissues.Paper accepted 30 April1998

1460 @ 1998 Blackwell Science Ltd

THE P53 TUMOUR SUPPRESSOR GENE 1461

Heat shock

¡'" /p53-

Hyp~xia Metabolic changes

DNA damage C~okines".---

Tumour suppressor genes

In contrast to oncogenes, tumour suppressor genes, intheir normal sta te, encode proteins that act to maintaincell numbers by suppressing proliferation or promotingloss. These genes become involved in the tumorigenicprocess when they sustain mutations that result in loss offunction. In this case the normal gene tends to act in adominant fashion and only when both alleles are damagedwill the effect of the mutant gene be apparent. Because ofthis, mutations in single alleles of tumour suppressorgenes may be passed through the germline; virtually all ofthe genes that have been identified as responsible forinherited cancer belong to this categoryl. Sporadic loss oftumour suppressor function in a genetically normalindividual can and does occur but, as this requires bothalleles in a single cell to malfunction due to eithermutation or deletion (loss), it occurs much less frequentlythan in an individual who has an inherited defect in allcells.

p53 belongs to the category of tumour suppressors andit appears to occupy a pivotal Tole in deciding the fate ofcells that have been stressed. Thus elimination of cellsthat have sustained genetic damage depends on functionalp53, and damage to the p53 gene itself may allow cellsbearing other mutant genes to proliferate unchecked.

However, studies using 'knockout' rnice (in which specificgenes have been inactivated) have shown that p53 nullmice, while viable, display a high rate of developmentalabnormalities, especialIy of the nervous system9,IO. Thissuggests that p53 might protect against teratogenesis andexperiments using benzo[ a ]pyrene have shown increasedteratogenicity in p53 null micell. In addition, after in uterairradiation p53 null mice have a high incidence ofanomalies and a low death rate, whereas normal rnicehave a low incidence of anomalies and a high death ratel2.Thus p53 mar have evolved primarily as a suppressor ofteratogenesis, with tumour suppression being a secondaryproperty; it has been suggested that it mar have an evenmore fundamental Tole as a facilitator of developmentalcomplexity in higher animalsl3.

Control of p53 and the 'p53 pathway'As indicate

1462 R. J. C. STEELE, A. M. THOMPSON, P. A. HALL and D. P. LANE

Fig. 2 Human skin immunostained for p53 protein afterexposure to sufficient ultraviolet irradiation to cause mildsunbum. The dark staining indicates overexpression of p53 in thenuclei

Mdm-2 is a 491-amino-acid phosphoprotein which canbind to p5317; it not only blocks its biological activity, butalso targets p53 for destruction via the ubiquitinproteosome pathway18.19. Mdm-2 mar be thought of as acomponent of the downstream pathway as its transcriptionis increased by p5320 (Table 1) and it thereby acts as anegative feedback mechanism controlling p53 levels (Fig.3). That this is vitally important has been illustrated byknockout experiments showing that mdm-2 null mice arenot viable unless they are also null for p53, indicating thatmdm-2 is necessary to prevent unregulated p53 activity21 ,

On the basis of these findings it is possible to speculatethat mdm-2 mar be important in controlling the response

Table 1 Key genes transcriptionaIly activated by p53 whichmediate the pathway leading to cell cycle arrest or apoptosis

-

Gene

of p53 to stress, and that phosphorylation of mdm-2 orp53 itself by stress-activated protein kinases could preventtheir interaction and hence allow accumulation of p5322,23.CertainIy other influences that affect the survival of cellsmar act through mdm-2, as evidenced by the finding thatbasic fibroblast growth factor induces mdm-2independently of p53 and so renders the cell insensitive tostimuli that would normally trigger apoptosis through thep53 pathway24.

Whereas the mechanism of the upstream pathway isstilI largely speculative, the downstream pathway is betterunderstood. p53 is a transcription factor; it can bind tospecific DNA sequences and activate the transcription ofgenes containing such binding sites in their promoterregulatory regions25,26. The number of genes thought to beactivated in this way is growing rapidly and there is anequally impressive range of genes which seem to havetheir function repressed by p537.27. Some of the morerobust candidates for the downstream p53 pathway areshown in Table 1; it can be seen how activation of thesegenes can affect the celI cycle and the late of apoptosis. Ithas become clear, however, that p53 mayalso act in otherways not related to transcriptional activity. For instance,recent evidence indicates that it mar inhibit nuclear DNAreplication directly28 and that p53-dependent apoptosiscan occur without transcriptional activation of p53 targetgenes29.30.

Despite the vast amount of data now available, it mustbe stressed that little of the mechanism of p53 function iscertain; the accumulation of knowledge in this field iscurrently very rapid. Recently, homolo~es (i.e. distinctproteins with shared functions) of p5331- and an mdm-2-related protein (mdm-x)35 have been described, and theseare likely to have important implications. It is possiblethat the p53 homologues make up a family of moleculeswith similar effects but induced by different signals andtherefore play fundamentally different roles in celIularhomoeostasis.

It must also be appreciated that much of OUT knowlegeof p53 function comes from in vitro work and that the invivo situation presents a whole new set of problems. Inparticular, it has become clear that the p53 responsevaries not onIy according to the insulting stimulus but alsoaccording to the tissue and celI type involved36.37. Thesediscoveries, coupled with the increasingly complexinterrelated pathways that are emerging, leave room forsignificant changes in OUT view of the precise significanceof p53 over the next few years.

p21/WAFl/Cipl

GADD45BaxIGF-BP3

Function of gene product

Binds to and inactivates p53, forming anautoregulatory loopArrests cell cycle by inhibiting cyclin-cyclin-dependent kinase complexes and binding toPCNAArrests cell cycle by binding to PCNAPromotes apoptosisEnhances apoptosis by blocking the mitoticactivity of insulin-like growth factor

PCNA, proliferating cellular nuclear antigen

Role of p53 in cancerThe main reason for the phenomenal recent interest inp53 is the finding that a high proportion of human cancers(up to 50 per cent) contain mutations of the p53 gene38(Table 2). Most commonly these genetic changes aremissense mutations in one allele, although deletions orchain-termination mutations can occur. The mutationalspectra at the p53 locus indicate that many difierentenvironmental mutagens are likely to be involved8, butthese have yet to be identified specifically. The actual siteof the mutation is also important; as p53 acts mainly as atranscription factor, mutations in the DNA-bindingdomain have fue greatest effect on function. Mostmissense mutations in cancers are located in this domainand these lead to fue production of p53 protein whichfails to bind to DNA in the normal sequence-specific

Stress

C'l" ",

p53~\ '"

-* Gel! death or growth arrest

"'"Degradation ~ t mdm-2

Fig. 3 Diagrarnrnatic representation oí the Tole oí rndrn-2 inpreventing unregulated activation oí p53

@ 1998 Blackwell Science Ltd, British Joumal 01 Surgery 1998, 85, 1460-1467

1463THE P53 TUMOUR SUPPRESSOR GENE

Table 2 Frequency of p53 mutations in different human tumourtypes

LungColorectumOesophagusOvaryPancreasSkinStomachHead and neckBladderProstateHepatocellularBrainAdrenalBreastEndometriumKidneyThyroidHaemopoietic systemCarcinoidMelanomaParathyroidCervix

activated by p53 and then targets it for destruction,thereby forming an autoregulatory mechanism. It is nowbelieved that mutant p53 cannot activate the transcriptionof mdm-2 and it is for this reason that p53 accumulates, ahypothesis supported by experimental work which hasshown that p53 disappears from human tumour cellsmicroinjected with mdm-248.

Detecting mutations in the p53 gene is both expensiveand time consuming, and the discovery that accumulationof the protein is associated with mutation has led to aproliferation of studies of archival material from humantumours using immunohistochemical detection of p53 as asurrogate for mutation. Because p53 acts as a protectivemechanism against the perpetuation of genetic damage, ithas been suggested that tumours with functioning p53mar carry a better prognosis than those expressing themutant protein. In support of this theory there arenumerous studies that have demonstrated an associationwith poor survival and accumulation of p53 in breastcancer49.50, colorectal cancer51-53, gastric cancer54, lungcancer55, ovarian cancer56 and several other tumour types.However, there are a number of studies that have notshown this association and it has been concluded that anyadverse prognostic effect is probably sma1157. Indeed,recent studies on colorectal and breast cancer usingrobust reagents have not been able confirm any prognosticsignificance of p53 accumulation even with a largenumber of patients58.59. This is perhaps not surprising asthe fact that malignant transformation has alreadyoccurred indicates a fundamental breakdown of cellularcontrol, and there is no good a priori reason why if a p53mutation has or has not contributed to this process theoutcome should be any different.

It is algo important to appreciate the limitations ofusing immunohistochemical detection of p53 as a markerof p53 function. First, it has been shown that detectableexpression of p53 can occur in the absence of genemutations60.61. This mar be due to increased sensitivity ofmodero reagents, but other explanations such asmalfunction of mdm-2 mar apply in some instances.Second, it is conceivable that some mutations mar notinterfere with mdm-2 transcription and therefore remainsusceptible to the controls exerted on the wild-typeprotein but still be functionally inactive. Third, it isrecognized that viTal proteins, such as the E6 humanpapilloma virus protein, can target wild-type p53 fordestruction in a similar way to mdm-262.

Most important, however, is the concept that it is thep53 pathway that is crucial rather than the protein alone.Some p53 mutations mar have a different downstreameffect from others; for example, some mutants canactiv~te the p21 gene but not the bax gene63 so that theeffects on cell cycle arrest mar be different to those onapoptosis. Furthermore, rather than sustaining p53mutations some tumours, especiaIly sarcomas, inactivatep53 by overexpression of mdm-264. Very recently anothertumour suppressor, p33 (encoded by the INGl gene), hasbeen shown to be obligatory for the ceIl cycle inhibitoryeffects of p5365 and it mar algo cooperate in the initiationof apoptosis. So if p33 is necessary to support the anti-tumour effects of p53, mutations in the p33/INGl genewiIl have to be taken into account. The p53 pathway iscomplex and not fuIly understood. It is possible that most,if not aIl, tumours have sustained some damage to thismechanism, but this can be detected by conventionalmethods only if the p53 gene itself is mutated andconsequently overexpressed.

fashion39. In addition, work with knockout mire has shownthat p53 null animals display a high late of tumourformation40. It has also been shown that the Li-Fraumenicancel family syndrome is caused by a gerrnline mutationin the p53 gene41.42.

Given the involvement of p53 in the cell cycle arrest orapoptotic response to genotoxic damage outlined above,these findings are not surprising. It is reasonable tosuppose that loss of effective p53 would allow damagedcells to survive and open the door to the accumulation ofmutations in other tumour suppressor genes and proto-oncogenes. It has been shown in transgenic mice thatapoptosis slows the growth of a tumour induced by a Tantigen which cannot inactivate p53 function and thatrapid tumour growth with reduced apoptosis occurs inp53 null animals with the same antigen43. In addition, thep53 mutations found in squamous cell carcinoma are alsofound in skin lesions caused by ultraviolet radiation, andin p53 null mice ultraviolet light does not cause the typicalapoptotic changes seen in the skin of normal mice44.

However, it is naive to regard p53 mutation simply asan 'enabling' event in carcinogenesis. While it appears tobe an early event in the development of skin cancer«, incolorectal cancel it has been found to occur late in theadenoma-carcinoma sequence45. The timing of a p53mutation is related to tumour type and must depend on acomplex series of variables. It has been suggested, becausep53 is induced by hypoxia, that the need for angiogenesisis balanced by the elimination of p53 function8. Thus arich blood supply early in tumour development may leavep53 mutation to alise at a later stage.

A key discovery in the oncological significance of p53was high levels of the protejo in many tumours. Thisaccumulation results from an increase in p53 stability,which in turn is usually associated with a mutation in thep53 gene46.47, and it was long thought that the mutantprotein was inherently more stable than the wild-typeprotejo. However, recent evidence indicates that p53protejo stability depends not specifically on mutation, buton binding to mdm-222. Mdm-2 is transcriptionally

@ 1998 Blackwell Science Ltd, British Joumal of Surgery 1998, 85, 1460-1467

1464 R. J. C. STEELE, A. M. THOMPSON, P. A. HALL and D. P. LANE

screening tool. However, while p53 mutations can bedetected in gut lavage fluid from patients with colorectalcancer75, effective faecal tests have proved elusive owingto difficulties in extracting appropriate DNA from stool.

Finally, there has been interest in looking for mutantDNA in blood as a marker for minimal residual diseaseafter apparent eradication of primary malignancy76. If thiswere feasible, it might prove useful for selecting patientsfor adjuvant therapy and for monitoring treatment. Thep53 gene is an obvious candidate for this approach,although little real progress has been made in thisdirection.

Therapy

In the treatment of cancel, restoration or modulation ofp53 function has become something of a holy grail, basedon the premise that the tumour cells lacking this functionmight destroy themselves or at least become moresusceptible to the effects of DNA damage inflicted byconventional chemotherapy or radiotherapy. Essentially,three main approaches look promising in this field. First,there is virus-mediated gene transfer in which a viTalgenome is engineered to contain foreign genes which areexpressed in the host cell genome after infection. Second,there is fue use of a cytolytic virus which can replicateonly in cells that lack p53 function, and by targeting suchcells could destroy tumours with mutant p53. Third, thereis the discovery or design of small molecules that caninterfere with the negative regulation of p53, pharma-cologically activating the p53 response.

Potential role oí p53 in the management oí cancer

Prognostic factor

p53 analysis does not appear to have any clinicallysignificant utility as a prognostic factor and is unlikely tosupplant conventional factors, such as stage andhistological grade, in the near future. However, there isgreat interest in the significance of p53 in modulating theresponse to radiotherapy and chemotherapy. As both ofthese forros of cancer treatment cause DNA damage, it isreasonable to assume that p53-dependent apoptosis marbe responsible for at least part of the therapeutic effectand there is evidence to support this hypothesis. Cellstransformed by the viTal oncogene EIA undergo p53-dependent cell death in response to ionizing radiation ortreatment with 5-ftuorouracil, etoposide or doxorubicin66.Work using transgenic mice has shown that p53 mutationsincrease the resistance of haemopoietic cell lineages to )'irradiation67. In addition, X-irradiation at 9,5 days ofgestation produces fewer deaths but more embryonicabnormalities in p53-deficient mice compared with wild-type animals12, reinforcing the notion that normalfunctioning of the p53 pathway is required for efficientDNA repair or cell death after irradiation.

Study of a small series of patients with rectal cancershowed that radiotherapy appeared to increase thenumber of apoptotic cells only in tumours expressing wild-type p5368. In breast cancer p53 mutations are wellcorrelated with resistance to doxorubicin treatment69. It isalgo of interest that human tumours that are generallysensitive to radiotherapy or chemotherapy, suchas testicular cancer or childhood acute lymphoblasticleukaemia, display low rates of p53 mutations66,70, whereastumours that often contain p53 mutations, such ascolorectal and lung cancers, tend to respond less well tochemotherapy or radiotherapy.

The use of p53 mutation analysis to predict sensitivityto potentially toxic therapy is therefore an attractiveproposition, but there is some evidence to suggest thatp53 status mar not always predict outcome reliably. Somereports indicate that inactivation of the p53 gene canactually Tender cells more sensitive to genotoxicdamage71,72, and in a study of human squamous carcinomacells p53 mutations did not corre late with radio-sensitivity73. This is an afea that requires large-scaleclinical studies with careful assessment of p53 function.

Virus-mediated gene transfel: A wild-type p53 DNAfragment has been inserted into a retrovirus (LNSXretroviral vector). Using this to restore the wild-type p53gene to lung cancer cells, it has suppressed the growth ofboth lung cancer cell liDes and human lung cancers inDude mice77.78. Similar findings in colorectal cancer cellsliDes have been reported using the replication detectiveadenovirus Ad5/CMV /p5379; it has also been shown thatthis agent has a synergistic effect when used with cisplatinchemotherapySO.

There is one reported clinical trial using wild-type p53gene transfer in nine patients with non-small cell lungcancer in whom conventional treatment had failed. In thisstudy the LNSX retroviral vector was injected directly intothe tumour either percutaneously with radiologicalguidance or via a bronchoscope81. In situ hybridization andDNA polymerase chain reaction showed vector-p53sequences in post-treatment biopsies, and apoptosis wasmore frequent in post-treatment than in pretreatmentbiopsies. No treatment-related toxicity was noted andtumour regression occurred in three patients. Furtherextensive trials of adenovirus encoding wild-type p53 arecurrently underway.

Cytolytic virus therapy. The DNA tumour virus adeno-virus produces a 55-kDa protein from tbe E1B region ofits genome wbicb binds and inactivates p53. It wasbypotbesized tbat an adenovirus lacking E1B would notbe able to replicate in normal cells but would in cancercells lacking p53 function. For tbis reason, ONYX-015, anE1B gene-attenuated adenovirus was compared witbnormal adenovirus in buman and colonic cancer ceU liDeswith and witbout p53 function. As expected, the ONYX-015 virus replicated as efficiently as the normal virus in

Diagnostic toolThe fact that p53 mutations are expressed by so manyhuman tumours has led to attempts to use them fordiagnostic purposes. In cytology, suspicious casesfrequently occur for which a satisfactory distinctionbetween benign and malignant cannot be made, andimmunocytochemical detection of p53 has been used in awide variety of tumours74. As p53 mutation is not aconsistent finding in aU tumours, the sensitivity of thisapproach is poor, but the specificity is high (in the regionof 97 per cent) and staining for p53 in suspicious cytologyspecimens may prove to be a useful adjunct tomorphological assessment.

Another afea in which p53 mutations have been used inan attempt to improve diagnosis is faecal screening forgastrointestinal cancer. As p53 is one of the genes that iscommonly mutated in colorectal cancer, it has beenproposed that detection of p53 mutations along with apanel of other mutated genes might offer a useful

@ 1998 Blackwell Science Ltd, British Joumal of Surgery 1998, 8S, 1460-1467

1465THE P53 TUMOURSUPPRESSOR GENE

the cellline lacking wild-type p53, but not in the line withnormal p53 function82.

Subsequent work showed that, while normal humancells are very resistant to the cytolytic effects of ONYX-015, a wide range of human tumour celllines with eithermutant or normal p53 gene sequences is destroyed83. It isinteresting that tumours with wild-type p53 aresusceptible; this is presumably due either to the presenceof undetected p53 mutations or to malfunction of othercomponents of the p53 pathway (e.g. mdm-2 over-expression). A phase 1 trial of ONYX-015 in patients withadvanced cancer is currently underway and a reportshould be forthcoming in the near future.

Inteiference with the negative regulation o/ p53. Tbeactivity of p53 is dependent, at least in part, on sequence-specific DNA binding and this process is controlled by anegative regulatory domain in the p53 molecule84,85.Neutralization of this domain by a specific antibodyactivates p5386 and it is likely that stress-related factorswhich influence the transcriptional activity of p53 act in asimilar way. The same antibody can activate mutant formsof p53 synthesized by human tumour cells lines87, and thisopens up the possibility of discovering or designing smallmolecules that could activate mutant forms of p53,thereby rescuing p53 function. Similarly, the discoverythat mdm-2 interacts with p53 via a small molecularinterface to block p53 function suggests that smallmolecules could be used to disrupt ibis interaction22. Tbiswould be particularly useful in tumours, especiallysarcomas, in which p53 is normal but where there isoverexpression of mdm-264.

Conclusionp53 research is tantalizingly clase to revolutionizingclinical practice in oncology, but it still has some way togo. The prognostic significance of p53 is uncertain, its Tolein diagnosis is promising but yet to be developed, and itspotential in therapy is just emerging. However, there is nodoubt about the central Tole of p53 in the development ofcancer and one of the main questions to answer is: Whydo a significant proportion of tumours seem to havenormal p53? The answer mar come from intensive studyof the p53 pathway rather than of p53 in isolation, as it isconceivable that all malignant tumours have a defectsomewhere along this pathway. This is the direction beingtaken in oncologically relevant p53 research and, whenthe pathway and its defects are fully understood, thetailoring of therapy for an individual tumour based on itsmolecular and genetic profile will be a practical strategy.

References1 Bishop JM, Weinberg RA (eds). Scientific American

Molecular Oncology. New York: Scientific AmericanIncorporated, 1996.

2 Stehelin D, Vannus HE, Bishop JM, Vogt PK. DNA relatedto the transforming gene(s) of avian sarcoma virus is presentin normal avian DNA. Nature 1976; 260: 170-3.

3 Spector DH, Vannus HE, Bishop JM. Nucleotide sequencesrelated to the transforming gene of avian sarcoma virus arepresent in DNA of uninfected vertebrates. Proc Natl Acad SciUSA 1978; 75: 4102-6.

4 Hofstra RMW, Landsvater RM, Ceccherini 1, Stulp RP,Stelwagen T, Luo Y et al. A mutation in the RET proto-oncogene associated with multiple endocrine neoplasia type

2B and sporadic medullary thyroid carcinoma. Nature 1994;367: 375-6.

5 Lane DP, Crawford LV. T antigen is bound to a host proteinin SV40-transformed cells. Nature 1979; 278: 261-3.

6 Linzer DI, Levine AJ. Characterization oí a 54K daltoncellular SV 40 tumor antigen present in SV 40-transíormedcells and uniníected embryonal carcinoma cells. Cell 1979;17: 43-52.

7 Hall fA, Meek D, Lane DP. p53-integrating the complexity.J Patho11996, 180: 1-5 (Editorial).

8 Levine AJ. p53, the cellular gatekeeper íor growth anddivision. Ce111997; 88: 323-31.

9 Armstrong JF, Kaufman MH, Harrison DJ, Clarke AR.High-frequency development abnormalities in p53-deficientmire. CU1T Bio11995; 5: 931-6.

10 Sah VP, Attardi LD, Mulligan GJ, Williams BO, BronsonRT, Jacks T. A subset oí p53-deficient embryos exhibitexencephaly. Nature Cenet 1995; 10: 175-80.

11 Nicol CJ, Harrison ML, Laposa RR, Gimelshtein IL, WellsPG. A teratologic suppressor role íor p53 in benzo[a]pyrene-treated p53-deficient mire. Nature Cenet 1995; 10: 181-7.

12 Norimura T, Nomoto S, Katsuki M, Gondo Y, Kondo S. p53-dependent apoptosis suppresses radiation-induced terato-genesis. Nature Med 1996; 2: 577-80.

13 Hall fA, Lane DP. Tumor suppressors: a developing role forp53? CU1T Bio11997; 7: 144-7.

14 Hall fA, McKee PH, Menage HDP, Dove;r R, Lane DP.High levels oí p53 protein in UV-irradiated normal humanskin. Oncogene 1993; 8: 203-7.

15 Kastan MB, Onyekwere O, Sidransky D, Vogelstein B, CraigRW. Participation oí p53 protein in the cellular response toDNA damage. Cancer Res 1991; 51: 6304-11.

16 Siliciano JD, Canman CE, Taya Y, Sakaguchi K, Appella E,Kastan MB. DNA damage induces phosphorylation oí theamino terminus oí p53. Cenes Dev 1997; 11: 3471-81.

17 Momand J, Zambetti GP, Olson DC, George D, Levine AJ.The mdm-2 oncogene product forms a complex with the p53protein and inhibits p53-mediated transactivation. Cell 1992;69: 1237-45.

18 Haupt Y, Maya R, Kazaz A, Oren M. Mdm2 promotes therapid degradation oí p53. Nature 1997; 387: 296-9.

19 Kubbutat MH, Jones SN, Vousden KH. Regulation oí p53stability by Mdm2. Nature 1997; 387: 299-303.

20 Wu x, Bayle JH, Olson D, Levine AJ. The p53-mdm2autoregulatory feedback loop. Cenes Dev 1993; 7: 1126-32.

21 Montes de Oca Luna R, Wagner DS, Lozano G. Rescue oíearly embryonic lethality in mdm2-deficient mice by deletionoíp53. Nature 1995; 378: 203-6.

22 Lane DP, Hall fA. MDM2-arbiter oí p53's destruction.Trends Biochem Sci 1997; 22: 372-4.

23 Mayo LD, Turchi JJ, Berberich SI. Mdm-2 phosphorylationby DNA-dependent protein kinase prevents interaction withp53. Cancer Res 1997; 57: 5013-16.

24 Shaulian E, Resnitzky D, Shifman O, Blandino G,Amsterdam A, Yayon A et al. Induction of Mdm2 andenchancement oí cell survival by bFGF. Oncogene 1997; 15:2717-25.

25 El-Deiry WS. P53, p21WAF1/CIP1 and the control oí cellproliferation. In: Thomas T, ed. Cell Cycle Control andApoptosis in Malignant Disease. London: Bios Scientific, 1995:19-38.

26 Agarwal ML, Taylor WR, Chemov MV, Chemova OB, StarkGR. The p53 network. J Biol Chem 1998; 273: 1-4.

27 Bourdon JC, Deguin-Chambon V, Lelong JC, Dessen P, MayP, Debuire B et al. Further characterisation oí the p53responsive element-identification oí new candidate genes fortrans-activation by p53. Oncogene 1997; 14: 85-94.

28 Cox LS, Hupp T, Midgley CA, Lane DP. A direct effect onactivated human p53 on nuclear DNA replication. EMBO J1995; 14: 2099-105.

29 Caelles C, Helmberg A, Karin M. p53-dependent apoptosis inthe absence oí transcriptional activation of p53-targetgenes.Nature 1994; 370: 220-3.

30 Haupt Y, Rowan S, Shaulian E, Vousden KH, Oren M.

@ 1998 Blackwell Science Ltd, British Jouma! of Su'b'ery 1998, 85, 1460-1467

1466 R. J. C. STEELE, A. M. THOMPSON, P. A. HALL and D. P. LANE

Induction of apoptosis in HeLa cells by trans-activation-deficient p53. Genes Dev 1995; 9: 2170-83.

31 Schmale H, Bamberger C. A novel protein with stronghomology to the tumour suppressor p53. Oncogene 1997; 15:1363-7.

32 Jost CA, Marin MC, Kaelin WG Jr. p73 is ahuman p53-related protein that can induce apoptosis. Nature 1997; 389:191-4.

33 Drane P, Barel M, Balbo M, Frade R. Identification ofRB18A, a 205 kDa new p53 regulatory protein which sharesantigenic and functional properties with p53. Oncogene 1997;15: 3013-24.

34 Bian J, Sun Yi. p53CP, a putative p53 competing protein thatspecifically binds to the consenus p53 DNA binding gires: athird member of the p53 family? Proc Natl Acad Sci USA1997; 94: 14753-8.

35 Shvarts A, Steegenga WT, Riteco N, van Laar T, Dekker P,Bazuine M et al. MDMX. a novel p53-binding protein withsome functional properties of MDM2. Embo J 1996; 15:5349-57.

36 Midgley CA, Owens B, Briscoe CV, Thomas DB, Lane DP,Hall P A. Coupling between gamma irradiation, p53 inductionand the apoptotic response depends upon cell type in vivo. JCell Sci 1995; 108: 1843-8.

37 MacCallum DE, Hupp TR, Midgley CA, Stuart D, CampbellSJ, Harper A et al. The p53 response to ionising radiation inadult and developing murine tissues. Oncogene 1996; 13:2575-87.

38 Holistein M, Rice K, Greenblatt MS, Soussi T, Fuchs R,Sorlie T et al. Database of p53 gene somatic mutations inhuman tumors and cell lines. Nucleic Acids Res 1994; 22:3551-5.

39 Kern SE, Pietenpol JA, Thiagolingam S, Seymour A, KinzlerKW, Vogelstein B. Oncogenic forms of p53 inhibir p53-regulated gene expression. Science 1992; 256: 827-30.

40 Donehower LA, Harvey M, Stagle BL, McArthur MJ,Montgomery CA Jr, Butel JS et al. Mice deficient for p53 aredevelopmentally normal but susceptible to spontaneoustumours. Nature 1992; 356: 215-21.

41 Malkin D, Li FP, Strong LC, Fraumeni JF Jr, Nelson CE,Kim DH et al. Germ line p53 mutations in a íamilialsyndrome oí breast cancer, sarcomas, and other neoplasms.Science 1990; 250: 1233-8.

42 Srivastava S, Zou ZQ, Pirollo K, Blattner W, Chang EH.Germ-line transmission oí a mutated p53 gene in a cancer-prone íamily with Li-Fraumeni syndrome. Nature 1990; 348:747-9.

43 Symonds H, Krall L, Reminton L, Saenz Robles M, Lowe S,Kacks T et al. p53-dependent apoptosis suppresses tumorgrowth and progression in vivo. Ce111994; 78: 703-11.

44 Ziegler A, Jonason AS, Leffell DJ, Simon JA, Sharma HW,Kimmelman JT et al. Sunburn and p53 in the onset oí skincancer. Nature 1994; 372: 773-6.

45 Vogelstein B, Kinzler KW. The multistep nature oí cancer.Trends Genet 1993; 9: 138-41.

46 Iggo R, Gatter K, Bartek J, Lane D, Harris AL. Increasedexpression oí mutant íorms oí p53 oncogene in primary lungcancer. Lancet 1990; 335: 675-9.

47 Hall fA, Lane DP. p53 in tumour pathology: can we trustimmunohistochemistry? -revisited. J Patho11994; 172: 1-4.

48 Midgiey CA, Lane DP. p53 protein stability in tumour cells isnot determined by mutation but is dependent on Mdm2binding. Oncogene 1997; 15: 1179-89.

49 Thor AD, Moore DHII, Edgerton SM, Kawasaki ES,Reihsaus E, Lynch HT et al. Accumulation oí p53 tumorsuppressor gene protein: an independent marker oí prognosisin breast cancers. J Natl Cancer Inst 1992; 84: 845-55.

50 Silvestrini R, Benini E, Daidone MG, Veneroni S, BoracchiP, Cappelletti V et al. p53 as an independent prognosticmarker in lymph node-negative breast cancer patients. J NatlCancer Inst 1993; 85: 965-70.

51 Remvikos Y, Tominaga O, Harnmel P, Laurent-Puig P,Salmon RJ, Dutrillaux B. Increased p53 protein content oí

colorectal tumours correlates with poor survival. Br J Cancer1992; 66: 758-64.

52 Hamelin R, Laurent-Puig P, Olschwang S, Jego N, AsselainB, Remvikos Y et al. Association oí p53 mutations with shortsurvival in colorectal cancer. Gastroenterology 1994; 106:42-8.

53 Bosari S, Viale G, Bossi P, Maggioni M, Coggi G, Murray JJet al. Cytoplasmic accumulation oí p53 protein: an inde-pendent prognostic indicator in colorectal adenocarcinomas.J Natl Cancer Inst 1994; 86: 681-7.

54 Martin HM, Filipe MI, Morris RW, Lane DP, Silvestre F.p53 expression and prognosis in gastric carcinoma. Int JCancer 1992; 50: 859-62.

55 Mitsudomi T, Oyama T, Kusano T, Osaki T, Nakanishi R,Shirakusa T. Mutations oí the p53 gene as a predictor oípoor prognosis in patients with non-small-cell lung cancer. JNatl Cancer Inst 1993; 85: 2018-23.

56 Bosari S, Viale G, Radaelli V, Bossi P, Bonoldi E, Coggi G.p53 accumulation in ovarian carcinomas and its prognosticimplications. Hum Patho11993; 24: 1175-9.

57 Dowell SP, Hall P A. The p53 tumour suppressor gene andtumour prognosis: is there a relationship? J PathoI1995; 177:221-4.

58 Rosen PP, Lesser ML, Arroyo CD, Cranor M, Borgen P,Norton L. P53 in node-negative breast carcinoma: animmuohistochemical study oí epidemiologic risk íactors,histologic íeatures, and prognosis. J Clin Oncol 1995; 13:821-30.

59 Poller DN, Baxter KJ, Shepherd NA. p53 and Rbl proteinexpression: are they prognostically useful in colorectalcancer? Br J Cancer 1997; 75: 87-93.

60 Baas 10, Mulder JW, Offerhaus B, Vogelstein B, HamiltonSR. An evaluation oí six antibodies íor irnmunohisto-chemistry oí mutant p53 gene product in archival colorectalneoplasms. J Patho11994; 172: 5-12.

61 Baas 10, van den Berg FM, Mulder JW, Clement MI, SlebosRJC, Hamilton SR et al. Potential íalse-positive results withantigen enhancement íor immunohistochemistry oí the p53gene product in colorectal neoplasms. J Pathol 1996; 178:264-7.

62 Scheffner M, Wemess BA, Huibregtse JM, Levine AJ,Howley PM. The E6 oncoprotein encoded by humanpapillomavirus types 16 and 18 promotes the degradation oíp53. Ce111990; 63: 1129-36.

63 Friedlander P, Haupt Y, Prives C, Oren M. A mutant p53that discriminates between p53-responsive genes cannotinduce apoptosis. Mol Cell Bio11996; 16: 4961-71.

64 Oliner ID, Kinzler KW, Meltzer PS, George DL, VogelsteinB. Amplification oí a gene encoding a p53-associated proteinin human sarcomas. Nature 1992; 358: 80-3.

65 Garkavtsev 1, Grigorian lA, Ossovskaya VS, Chemov MV,Chumakov AV, Gudkov AV. The candidate tumoursuppressor p331NG1 cooperates with p53 in cell growthcontrol. Nature 1998; 391: 295-8.

66 Lowe SW, Ruley HE, Jacks T, Housman DE. p53-dependentapoptosis modulates the cytotoxicity oí anticancer agents.Ce111993; 74: 957-68.

67 Lee JM, Bernstein A. p53 mutations increase resistance toionizing radiation. Proc Natl Acad Sci USA 1993; 90: 5742-6.

68 Hamada M, Fujiwara T, Hizuta A, Gochi A, Naomoto Y,Takakura N et al. The p53 gene is a potent determinant oíchemosensitivity and radiosensisitivity in gastric andcolorectal cancers. J Cancer Res Clin Oncol 1996; 122:360-5.

69 Aas T, Borresen AL, Geisler S, Smith-Sorensen B, JohnsenH, Vargaug JE et al. Specific p53 mutations are associatedwith de novo resistance to doxorubicin in breast cancerpatients. NatureMed 1996; 2: 811-14.

70 Lutzker S, Levine AJ. A functionally inactive p53 protein interatocarcinoma cells is activatedby either DNA damage orcellular differentiation. Nature Med 1996; 2: 804-10.

71 Fan S, Smith ML, Rivet DJ 11, Duba D, Zhan Q, Kohn KWet al. Disruption oí p53 function sensitizes breast cancer

@ 1998 Blackwell Science Ltd, British Joul1Ial o[ Surgery 1998, 85, 1460-1467

1467THE P53 TUMOUR SUPPRESSOR GENE

Tanida T et al. Novel combination therapy for human coloncancel with adenvirus-mediated wild-type p53 gene transferand DNA-damaging chemotherapeutic agent. Int J Cancer1997; 73: 367-70.

81 Roth JA, Nguyen D, Lawrence DD, Kemp BL, Carrasco CH,Ferson DZ et al. Retrovirus-mediated wild-type p53 genetransfer to tumors of patients with lung cancel. Nature Med1996; 9: 985-91.

82 Bischoff JR, Kirn DH, Williams A, Heise C, Horn S, MunaM et al. An adenovirus mutant that replicates selectively inp53-deficient human tumour cells. Science 1996; 274: 373-6.

83 Heise C, Sampson-Johannes A, Williams A, McCormick F,Von Hoff DD, Kirn DH. ONYX-015, an EJB gene-attenuated adenoVirus, causes tumor-specific cytolysis andantitumoral efficacy that can be augmented by standardchemotherapeutic agents. Nature Med 1997; 3: 639-45.

84 Hupp TR, Lane DP. Regulation of the cryptic sequence-specific DNA-binding function of p53 by protein kinases.Cold Spring Harbor Symp Biol1994; 59: 195-206.

85 Takenaka 1, Morin F, Seizinger BR, Kley N. Regulation ofthe sequence-specific DNA binding function of p53 byprotein kinase C and protein phosphatases. J Biol Chem1995; 270: 5405-11.

86 Hupp TR, Sparks A, Lane DP. Small peptides activate thelatent sequence-specific DNA binding function of p53. Cell1995; 83: 237-45.

87 Niewolik D, Vojtesek B, Kovarik J. p53 derived from humantumour celllines and containing distinct point mutations canbe activated to bind its consensus target sequence. Oncogene1995; 10: 881-90.

MCF- 7 cells to cisplatin and pentoxifylline. Cancer Res 1995;ss: 1649-54.

72 Xu C, Meikrantz W, Schlegel R, Sager R. The humanpapilloma virus 16E6 gene sensitizes human mammaryepithelial cells to apoptosis induced by DNA damage. ProcNatl Acad Sci USA 1995; 92: 7829-33.

73 lung M, Notario V, Dritschilo A. Mutations in the p53 genein radiation-sensitive and -resistant human squamouscarcinoma cells. Cancer Res 1992; 52: 6390-3.

74 Dowell SP, Wilson POG, Derias NW, Lane DP, Hall FA.Clinical utility of the immunocytochemical detection of p53protein in cytological specimens. Cancer Res 1994; 54:2914-18.

75 Potter MA, Morris RG, Ferguson A, Wyllie AR. Colorectalcancer mutations in whole gut lavage fluido Br J Surg 1997;84: 1590 (Abstract).

76 Mulcahy HE, Croke DT, Farthing MlG. Cancer andmutantDNA in blood plasma. Lancet 1996; 348: 628.

77 Mukhopadhyay T, Tainsky M, Cavender AC, Roth lA.Specific inhibition of K-ras expression and tumorigenicity oflung cancer cells by antisense RNA. Cancer Res 1991; 51:1744-8.

78 Fujiwara T, Caí DW, Georges RN, Mukhopadhyay T, GrirnmEA, Roth JA. Therapeutic effect of a retroviral wild-type p53expression vector in an orthotopic lung cancer modelo J NatlCancer Inst 1994; 86: 1458-62.

79 Spitz FR, Nguyen D, Skibber JM, Cusack J, Roth JA,Cristiano RJ. In vivo adenovirus-mediated p53 tumorsuppressor gene therapy for colorectal cancer. Anticancer Res1996; 16: 3415-22.

80 Ogawa N, Fujiwara T, Kagawa S, Nishizaki M, Morimoto Y,

@ 1998 Blackwell Science Ltd, British Jouma! of Surgery 1998,85, 1460-1467