Proc. Natl. Acad. Sci. USAVol. 90, pp. 10914-10921, December 1993

Review

Antioncogenes and human cancerAlfred G. KnudsonInstftute for Cancer Research, Fox Chase Cancer Center, Philadelphia, PA 19111

ABSTRACT The antioncogenes, or tu-mor suppressor genes, as negative regula-tors of cell division, stand in contrast tooncogenes. For most human cancers, themore frequently mutated genes are theantioncogenes, the principal exception be-ing the leukemias and lymphomas. Personsheterozygous for germ-line mutations inantioncogenes are strongly predisposed toone or more kinds of cancer, and mostdominantly inherited cancer is attributableto such heterozygosity. Seven antionco-genes have been cloned through the study ofthese persons, and several others have beenmapped. An eighth one was mapped andcloned through the investigation of tumorsand is not yet known in hereditary form.Three dominantly inherited forms of can-cer are not attributable to mutations inantioncogenes. The corresponding nonhe-reditary forms of most cancers generallyreveal abnormalities of the same antionco-genes that are found in the hereditaryforms but may also show additional ones.Some cancers, especially the embryonaltumors ofchildren, have a small number ofantioncogene mutations; some others, suchas most sarcomas, have more, and thecommon carcinomas have the most, reflect-ing a hierarchy of controls over growth ofstem cell populations. StiUl more membersof this gene category remain to be mappedand cloned through the study of cancerfamilies and oftumors. The genes that havebeen cloned act at diverse points in thesignal transduction pathway in ceUls, fromthe outer cell membranes to sites of genetanscription, in some cases as negativeregulators of oncogene expression.

Virtually every human cancer can occurin genetically predisposed individuals.The most striking form of genetic suscep-tibility involves Mendelian dominant in-heritance with high penetrance and ap-pearance of cancer at earlier than usualage, as shown for colon cancer in personswith familial adenomatous polyposis. Inthis example, the heterozygous state ofthe germ-line mutation imparts a high riskforjust one form of cancer, while in otherexamples, such as the Li-Fraumeni syn-drome (LFS), it predisposes to severalkinds of cancer, although never to allforms. The most frequent cancer in LFS iscarcinoma of the breast, although it doesnot afflict all female carriers. This incom-plete penetrance of a gene for a particular

cancer typifies the heterozygous state fora familial cancer gene; its presence is nota sufficient condition for cancer.A simple explanation for this incom-

plete penetrance is that oncogenesis re-quires a somatic mutation in some targetcell, an event that may never occur insome heterozygous carriers; two muta-tions, one germinal and one somatic,would be needed (1). This hypothesisalso relates the hereditary and nonhered-itary forms of a cancer by a commonmechanism; the same two mutationswould operate in both, the first mutationbeing germinal in the former and somaticin the latter. In the hereditary case all ofthe somatic cells would carry a first"hit," whereas in the nonhereditary caseonly a clone of somatic cells would do so.What might the targets of these two

hits be? The simplest answer is the twocopies ofsome autosomal gene; oncogen-esis would be recessive at the cellularlevel in both the hereditary and nonhe-reditary cases (2, 3). The presence ofonenormal (wild type) copy of the genewould interfere with oncogenesis, andthe normal allele could therefore be con-sidered as antioncogenic. Such genes areknown as antioncogenes, or tumor sup-pressor genes. If the 50 or so differenthereditary cancers all arise by this mech-anism, then there should be a corre-sponding number of antioncogenes.The past several years have brought

important discoveries regarding the ge-netic predisposition to cancer. Antionco-genes do exist and in fact play a majorrole in human carcinogenesis. Seven an-tioncogenes whose mutations in the germline predispose their hosts to cancer havebeen cloned, and several putative genesofthis class have been mapped to specificchromosomal bands (Table 1). These dis-coveries have illuminated our under-standing of human cancer considerablyand have revealed a group ofgenes that isimportant in cellular and developmentalbiology, making this a propitious time toreview progress on the subject and totake note of some future prospects forextending this understanding.

Retinoblastoma: The PrototypicHereditary Cancer

Although uncommon, occurring at a rateof approximately 5 cases per 100,000

children in most parts of the world, reti-noblastoma has been a prototype in thestudy of hereditary cancer (1, 4). About40% of cases are attributable to a germ-line mutation of the RBI gene, which islocated in chromosomal band 13q14.2.Most of these affected persons have noprevious family history of the tumor, afact that is explained by new mutationsoccurring at a rate of 8 x 10-6 per locusper generation. However, z50% of theoffspring of these newly mutant caseswill develop the tumor. The remaining60% of cases are nonhereditary, indicat-ing that the incidence oftumor in childrenwithout the germ-line mutation is 3 per100,000. These cases affect just one eye,whereas in the majority of germ-linecases both eyes are affected.The number of tumors in germ-line

cases is distributed in nearly Poissonfashion, with a mean of three tumors, andwith the unaffected carrier class beingapproximately, as expected, e-3, or 5%.Since the number of transformable cellsderived from the normal target cells, theembryonic retinoblasts, is of the order ofmagnitude of 107, the probability that asecond, tumor-forming event will occurcan be estimated as no less than 3 x 10-7per cell division (5). If this frequencyapplies to each of the two cell divisionsnecessary to form a nonhereditary tu-mor, and if we take into account themultiplication of once-hit retinoblasts,the expected incidence would be close to3 x 10-5, as observed. In addition, mostpersons in a population would have atleast one once-hit cell that differentiatesinto a normal postmitotic cell before asecond event can produce a tumor. Theincidences of both the hereditary andnonhereditary forms of retinoblastomacan thus be accounted for by germ-lineand somatic mutation rates that are sim-ilar to known "spontaneous" mutationrates, and there is no need to invokeinduced mutations to account for theincidence of retinoblastoma in mostcountries.About 5-10% of the hereditary cases

show a constitutional deletion of part orall of chromosomal band 13q14, a phe-nomenon that revealed the site of theretinoblastoma gene (RB)) (6, 7). The useof syntenic polymorphic genetic markers(restriction fragment length polymor-phisms, RFLPs) permitted demonstra-

10914

Dow

nloa

ded

by g

uest

on

Dec

embe

r 27

, 202

0

Proc. Natl. Acad. Sci. USA 90 (1993) 10915

Table 1. Characteristic neoplasms associated with germ-line mutations inselected antioncogenes

Neoplasm

Chromosomal Sarcoma, NeurectodermalAntioncogene location Wilms tumor and endocrine Carcinoma

ClonedRBI 13q14 Sarcoma RetinoblastomaWTI 11p13 Wilms tumorTP53 17pl3 Sarcoma Glioma BreastNFl 17qll Sarcoma GliomaNF2 22q12 SchwannomaVHL 3p25 Pheochromocytoma KidneyAPC 5q21 Colon

UnclonedNBI 1p36 NeuroblastomaMLM 9p21 MelanomaMEN) 11q13 Pituitary adenomaBCNS 9q31 Medulloblastoma SkinRCC 3p14 KidneyBRCAI 17q21 Breast, Ovary

tion of the recessiveness of the gene inoncogenesis (8) and of the mechanismspredicted (4) for second events-namely,local mutation, deletion, chromosomalnondisjunction, and somatic recombina-tion. Loss of heterozygosity of at leastsome syntenic markers is a feature of allof these mechanisms except the first.RFLPs were also used for the positionalcloning of the RBI gene (9-11). The nor-mal cDNA of RBI was then shown tocause reversion of the tumorigenic prop-erties of cultivated tumor cells that weremutant for RBI (12). RBI thus becamethe first human antioncogene, or tumorsuppressor gene, to be discovered.The RBI gene itself has been the sub-

ject of intense investigation (reviewed inref. 13). The retinoblastoma protein Rb isphosphorylated during the cell cycle. Inits unphosphorylated form it blocks pas-sage through G1 to S, apparently by com-plexing with a transcription factor suchas E2F, which normally can activate im-portant cell cycle genes, perhaps includ-ing MYCN, which is overexpressed infetal retina and retinoblastoma; an an-tioncogene may oppose the action of anoncogene. Phosphorylated Rb proteindoes not inhibit the factor and is thereforepermissive for this passage. Since Rbexerts such an important effect and sinceit is expressed in all cells, its associationprimarily with a rare tumor of children ispuzzling. Perhaps its action is circum-vented by other regulatory mechanismsin tissues other than retina.

Survivors of retinoblastoma withgerm-line mutations of RBI are also sus-ceptible to other tumors, notably os-teosarcoma and soft-tissue sarcomas (re-viewed in ref. 13). The penetrance ismuch lower than the 95% or so observedfor retinoblastoma itself, being of themagnitude of 10-15% for each of theseclasses of tumor by the age of 30 years.Some of these tumors were observed in

irradiated orbital bone or soft tissues, butothers were observed far from the eye,compelling the conclusion that geneticpredisposition was not to a single tumor.The radiation-induced tumors do informus that the frequency of second eventscan be increased by environmentalagents. Furthermore, many nonheredi-tary tumors at these same sites, espe-cially osteosarcomas, revealed mutationsat the RBI locus. Even more surpris-ingly, several other tumors not typicallyobserved in retinoblastoma survivorswere mutated at RBI. Especially strikingin this respect was small-cell carcinomaof the lung, where most, perhaps all,tumors are mutant or deleted for bothcopies of RBI; yet almost none showgerm-line mutations ofRBI. The relativerisk that germ-line mutation of RBI im-poses on its host is _105 for retinoblas-toma, 103 for osteosarcoma, and 10 orless for small-cell carcinoma of the lung(14). Why, then, would the penetrance beextremely high for one tumor, interme-diate for others, and very low for stillothers? Small target-cell pools and verylow tissue-specific mutation rates are notlikely explanations, because osteosar-coma and small-cell carcinomas of thelung are much more common than reti-noblastoma. As we shall see later, theapparent explanation is that for theseother tumors, other controls on growthmust be overcome; RBI is not the onlygene that must mutate.Mice that are constitutionally hetero-

zygous for a nonfunctional allele of RBIdo not develop retinoblastomas (15-17).One explanation is that the number oftarget cells-i.e., retinoblasts-is toosmall to lead to a second mutation in anobservable number of animals. On theother hand, the mice do develop tumorsof the pituitary, which are not seen inhumans. These heterozygous mice canalso be mated with each other to produce

homozygously defective animals, whichdie before 16 days of fetal life, withabnormalities of the brain and hemato-poietic system. Some cells in both tissuesundergo too many mitoses and too littledifferentiation. From this experiment onecan conclude that RBI is a developmen-tally critical gene, whose mutations arerecessive in their developmental and on-cogenic effects.

Wilms Tumor

Wilms tumor was the second cancer forwhich a relationship between hereditaryand nonhereditary forms was proposed(18). This tumor occurs in =10 per100,000 children. Here again the discov-ery of cases with constitutional dele-tions-at 11p13-pointed to the locationof the putative WTI gene (19) and so toregional RFLPs that were used to findmolecular mutations and deletions and toclone the gene (20, 21). Patients withconstitutional mutations or deletions ofWTI appear not to be susceptible to othertumors. The gene's specificity fits wellwith the limitation of its expression to thegenitourinary tract, the mesothelial liningof the body cavities, the spleen, and partsof the central nervous system, in contrastto RBI 's ubiquitous expression (re-viewed in ref. 22). Genitourinary anom-alies are common in children who areheterozygous for mutation or deletion ofWTI, and some mutations produce theDenys-Drash syndrome, which is alsopredisposing to Wilms tumor.Mice made heterozygous for a muta-

tion of WTI do not develop Wilms tumor,again probably because, as for RBI mice,the target-cell population is too small(23). Even in humans the penetrance ofWTI mutations for Wilms tumor is lowerthan that ofRBI for retinoblastoma. Thehomozygous mutant state is again lethal,at 13-15 days, with a failure of develop-ment of the metanephric kidney. Lethal-ity is apparently attributable to wide-spread edema secondary to impaired de-velopment of the thoracic mesothelium,heart, and lungs. Like RBI, WTI is adevelopmentally lethal recessive genethat in the heterozygous mutant statepredisposes its human host to cancer.Wilms tumor differs from retinoblas-

toma in another important way. WhereasRBI appears to be the only gene whosemutations predispose to retinoblastoma,only about 10-20o of Wilms tumors canbe attributed to mutation at WTI. A well-recognized but rare dominantly heritabledisorder, the Beckwith-Wiedemann syn-drome (BWS), also predisposes to thistumor. The BWS gene has not beencloned, but clinical features suggest thatBWS may not be an antioncogene. In-fants with the syndrome are large at birthand exhibit features that suggest the pos-

Review: Knudson

Dow

nloa

ded

by g

uest

on

Dec

embe

r 27

, 202

0

Proc. Natl. Acad. Sci. USA 90 (1993)

sibility of heterozygosity for an overex-pressing mutation of an oncogene.BWS has been mapped by linkage anal-

ysis to chromosomal region ilp15, a con-siderable distance from WTI (24). Onegene in this region, that for insulin-likegrowth factor 2 (IGF2), is especially rel-evant because its product is a fetalgrowth factor, expressed strongly in fetalkidney and in Wilms tumor but not in theadult kidney. It is also an imprinted gene,the maternally derived allele being unex-pressed. In BWS, and in some nonhered-itary Wilms tumors, this imprinting isrelaxed and both parental alleles are ex-pressed, thereby presumably causing ex-cessive fetal growth (25-27). In addition,the protein product of WTI can repressthe IGF2 gene in vitro, binding to a 5'regulatory region of the latter. Thesefindings suggest a kind of equivalence ofthe two genes that predispose to Wilmstumor, one mutation affecting an onco-gene, and the other affecting an antion-cogene that regulates it.

LFS and the p53 Gene

A role for the protein product of the TP53gene in carcinogenesis was originally de-duced from the study of tumors inducedby certain DNA viruses (reviewed in ref.28). Following its mapping to humanchromosomal band 17pl3, loss of het-erozygosity of 17p markers in many hu-man tumors, especially carcinomas, im-plicated it as a member of the antionco-gene category. Now it is known that amajority of mutations in the gene involvesingle base changes that produce a pro-tein with an abnormally long half-life.When mutant, this protein, a DNA-binding protein as are the products ofRBJ and WTI, may interfere with thefunction of the product of the normalallele-i.e., it may have a "dominantnegative" effect, which could explain itsoncogene-like behavior. However, mosttumors show abnormality of both alleles,and some tumors show complete deletionof both alleles.

Later it was found that TP53 is consti-tutionally mutant in most cases of LFS(29), a dominantly inherited syndromediscovered through the study of familialrhabdomyosarcoma in children (30). Thefamilies of these children proved to havea high incidence of other cancers, includ-ing especially breast carcinoma, but alsosoft tissue sarcomas (including rhabdo-myosarcoma), osteosarcoma, brain tu-mors, leukemia, adrenocortical carci-noma, and perhaps some other carcino-mas. The breast cancers typically occurin subjects before the age of 50 years,even in their 20s. The incidence of LFS isprobably 2-4 per 100,000 births, and thehigh mortality before the end ofthe age ofreproduction assures that it is maintained

in populations only by recurrent sponta-neous germ-line mutations.A remarkable feature of LFS is that

some carcinomas are not typical. Thus,carcinoma of the colon and small-cellcarcinoma of the lung, both with highincidence of somatic mutations of TP53,are not important tumors in LFS. Whydoes one carcinoma, that of the breast,but not two others, those ofthe colon andlung, develop frequently in LFS sub-jects? Perhaps there is a clue in a com-parison of the tumors that typify theheterozygous carriers of RBI and TP53.In both instances sarcomas of soft tissuesand bone are conspicuous among carriersin the first two decades of life. Mutationsof both genes are common in nonhered-itary tumors ofthese histologies; but theyare also both common among cases ofnonhereditary small-cell carcinoma ofthe lung, a tumor that is not typicallyfound in constitutional heterozygotes formutation at either gene. Interestingly, thesarcomas occur during a time of netgrowth in somatic tissues, which is ac-companied by growth of tissue stemcells. In this respect these tissues par-tially resemble the growth of fetal retinaand kidney, where clones of once-hitcells greatly magnify the number oftargetcells available for mutation or loss of thesecond allele of RBJ or WTI, respec-tively. Breast is another tissue that un-dergoes net growth in adolescence. Onthe other hand, colon and lung are typicalrenewal tissues; development of once-hitclones in a renewal tissue is uncommonunless there is some stimulus to stem cellproliferation, as happens in the colonwith chronic ulcerative colitis, and, pos-sibly, in the lung in response to cigarettesmoking.The TP53 gene, like RBI, can also

influence passage through the G1 phase ofthe cell cycle (31). Especially interestingis the observation that normal cells, butnot those mutant at TP53, are arrestedtemporarily in G1 by ionizing radiation(32). It is therefore something of a sur-prise that mice transgenic for expressingmutant p53 alleles (33) or homozygouslydefective for p53 (34) develop normally,although they have a high incidence oftumors, especially of lymphoid tissue. Itseems that p53 regulation of the cell cycleis in some way conditional, and muta-tions are therefore permissive for normaldevelopment. This situation is compati-ble with the observation that overexpres-sion of TP53 in normal cells elicits aquiescent state in vitro in associationwith a reduction in available guanosinetriphosphate (GTP), a key compound inthe transduction of growth signals (35).One scenario (36) raises the possibilitythat tumors themselves, at some point intheir histories, elicit a host defense in theform of increased production of p53,which is inhibitory to tumor growth. Only

then would mutation or loss of TP53impart a selective advantage to a cell.Thus, mutation at RBI might initiate anosteosarcoma cell, but its malignancywould depend upon loss of normal p53activity. In small-cell lung cancer, a tu-mor of a typical renewal tissue, yet an-other negative control might inhibitgrowth until it too is mutated. There seemto be three kinds of tissues: (i) embryonaltissues whose normal growth is uncondi-tional once proliferation is initiated; (ii)"conditional tissues" like bone, whosegrowth can be induced as by hormones atpuberty; and (iii) renewal tissues, whichhave a capacity for conditional responsebut in addition are programmed for dailyreplenishment (14). TP53 plays an impor-tant role in these latter two tissue cate-gories, but the second category is theprincipal target for oncogenesis in LFS.

Antioncogenes and Tumors of Neuraland Neural Crest Origin

Neuroblastoma. In addition to RB1there are several antioncogenes that pre-dispose to tumors of neural origin withsome specificity. Along with retinoblas-toma and Wilms tumor, neuroblastomawas one of three embryonal tumorstreated as models of hereditary cancerinvolving constitutional heterozygosityfor a mutant gene that is also important inthe nonhereditary form of the same tu-mor (37). The tumors commonly show adeletion that includes all or part of chro-mosomal region lp36, so that site hasbeen suspected as the locus of a NBIgene (38, 39). However, familial neuro-blastoma is rare and linkage analysis hasnot been conducted. Two case reports ofneuroblastoma with constitutional aber-rations of 1p36 enhance the candidacy ofan antioncogene at this site (40, 41). Thefamilies that have been observed withthis tumor typically do not develop othertumors, suggesting a considerable speci-ficity for the neural crest-derived adrenalmedulla and sympathetic nervous chain.This tissue can also be the site of pheo-chromocytoma, but families with neuro-blastoma do not have pheochromocy-toma, and vice versa. Thus, there is spec-ificity not only for an organ but also forthe degree of differentiation within thatorgan.

Neurofibromatosis. The term neurofi-bromatosis is applied to two entities,types 1 and 2 (NF1 and NF2). NF1 fea-tures the benign tumor, neurofibroma,and distinctive cafe-au-lait pigmented le-sions of the skin. One of the most fre-quent dominantly inherited human dis-eases, NF1 affects 30 per 100,000 chil-dren, half of whom are new germ-linemutants. The NFI gene predisposes toseveral tumors: neurofibrosarcoma, ma-lignant schwannoma, glioma, and pheo-chromocytoma, and, at low frequency,

10916 Review: Knudson

Dow

nloa

ded

by g

uest

on

Dec

embe

r 27

, 202

0

Proc. Natl. Acad. Sci. USA 90 (1993) 10917

leukemia, rhabdomyosarcoma, andWilms tumor. The NFl gene was mappedby linkage analysis to chromosomal band17qll and subsequently cloned (42, 43).The gene codes for a large ubiquitouslyexpressed protein, neurofibromin, oneregion of which is homologous to the twoinhibitor genes of RAS found in yeast.The gene's function is similar to that ofthe GAP (GTPase-activating protein)gene product; loss of its activity results infailure of hydrolysis of GTP to GDP bythe RAS protein. This loss of neurofibro-min function is thought to result in ele-vated levels of the GTP-bound RAS pro-tein that transduces signals for cell divi-sion (44-46). The high germinal mutationrate ofNFI may be attributed to its heavymethylation (47). The mutations ob-served in the gene frequently producetruncated, functionless proteins. In a sig-nificant fraction of the characteristic ma-lignant tumors of NFI, the second (nor-mal) copy of the gene appears to bemutated or absent (48-50). Mutations ofTP53 are also observed in some cases ofneurofibrosarcomas (51), as often ob-served in other sarcomas. NFl evidentlyqualifies as an antioncogene, with rela-tively narrow specificity for tumors ofneural and neural crest origin, althoughmutations are found in some tumors thatare not observed in NF1 patients. Wild-type NFl again exemplifies the inhibitionof an oncogene by an antioncogene. Itwill be very interesting to discover whyNFl has the tissue specificity that it does.The existence of both GAP and NFlsuggests redundancy in the regulation ofRAS protein. It is possible that the tis-sues that are susceptible to tumorigenesishave little GAP function, leaving NFI tobe the sole regulator.NF2, a much less common disorder

than NF1, shows a high penetrance andgreat specificity for acoustic nerve tu-mors (vestibular schwannomas) andmeningiomas. Sporadic, nonhereditaryforms of these tumors are often mono-somic for chromosome 22, implicating anantioncogene on this autosome. NF2 hasbeen mapped to chromosomal arm 22q12and has recently been cloned (52, 53). Itsprotein product is unusual for tumor sup-pressor genes in that it is homologouswith proteins found at the interface be-tween the plasma membrane and the cy-toskeleton. Mutations found in the tu-mors typically cause truncation of theprotein product; NF2 clearly qualifies asan antioncogene, although its precisemechanism of action is not yet known.Melanoma. Malignant melanoma, an-

other tumor of neural crest origin, existsin at least one, possibly two or more,hereditary forms. The gene for one ofthese, MLM, has been mapped by link-age studies and in a deletion case near thea- and p-interferon genes, on chromo-somal arm 9p2l, although it has not yet

been cloned (54, 55). A claim has been neurectodermade for another familial melanoma gene hereditary c(on chromosome arm lp (56). Preneoplas- already bee:tic benign nevi usually precede the ap- gliomas, altipearance of melanomas, suggesting one penetrance.somatic event producing the nevus, and quently seenstill further events leading to malignancy. lignancy, anNonhereditary melanomas often show tant cells hasloss of heterozygosity for 9p markers and with progreseven homozygous deletions at band 9p21 of chromoso(57), indicating the importance of the ously as coIMLM antioncogene for both the heredi- remains to btary and nonhereditary forms of mela- fected geneinoma. The 9p2l site has been implicated mutant in hin other tumors, including brain tumors site is appan(58), some leukemias (59), and some lung gression, as icancers (60), but these tumors have not almost univebeen associated with constitutional mu- the very maltation of MLM. An apparently homolo- there is as yegous site on rat chromosome Sq nearly that show halways shows loss of heterozygosity in heritance ofcell lines derived from renal carcinomas erations (forin the Eker rat (61), clearly in association may point twith tumor progression. The possibility formation anmust be considered that not all of the PNETs, v

tumo,rs mentioned involve one and the dren, are ocsame gene. with a predis

Multiple Endocrine Neoplasia. Two an antioncolwell-known syndromes, multiple endo- However, ccrine neoplasia types 1 and 2, that pre- been associdispose to endocrine tumors, some of Gorlin syndneurectodermal and neural crest origin, syndrome (Ehave been mapped by linkage analysis to mapped by Isites at 11q13 (MEN1) and 10qll (MEN2) Carriers of f(62, 63). Heterozygotes for MEN] muta- almost invartions are strongly predisposed to para- cinomas of Ithyroid hyperplasia and to tumors of the half of thesianterior pituitary and pancreatic islets. erozygosityFurthermore, these tumors often show the presenceloss of heterozygosity for llq markers, chromosomgso MEN) appears to qualify as an antion- of persons 'cogene. Such is not the case for MEN2, PNETs, anwhose mutations predispose heterozy- medulloblas'gotes to two neural crest-derived tumors, erozygositypheochromocytoma and medullary car- interesting tcinoma of the thyroid. Loss of heterozy- PNET havegosity (LOH) for markers on 10qll has niospinal irnbeen found rarely in these tumors, even duced largein those persons with the syndrome; on cell carcinorthe other hand, LOH for lp markers is riod, suggesivery common (64). Chromosome arm lp ous "secondhas been implicated also in neuroblasto-mas and melanomas, other tumors of Carcinomasneural crest origin. The responsible geneon chromosome 10qll has recently been Carcinomasidentified as the RET protooncogene gory ofhumg(65), a receptor tyrosine kinase gene, toll of life. 1making it the only example so far of an plicated frononcogene whose mutations in the germ ics. Still, tiline place the host at risk of cancer. A progress in r

plausible scenario for the development of of more to cthese tumors is that the RET mutation The vonproduces the preneoplastic hyperplasia drome and Iof the adrenal and thyroid medullae that dominantly icharacterizes the disease, increasing the acterized b)number of target cells available for trans- features thaiformation to neoplasia by mutation at an Virtually evantioncogene locus on lp. tation devel

Brain Tumors. The principal tumors that types of caiarise in the brain are gliomas and primitive the brain, p]

rmal tumors (PNETs). Twoonditions, NF1 and LSF, haven observed to predispose tohough neither does so at highMutations at TP53 are fre-

n in gliomas of high-grade ma-id clonal expansion of p53 mu-s been observed in associationssion in glioma (66). Deletionsme 9p have been noted previ-mmon in glial tumors, and it)e determined whether the af-is identical with the one that isereditary melanoma. The 9pently important in tumor pro-is one on chromosome 10q, an;rsally affected chromosome inignant glioblastoma (67). Still,at no explanation for pedigreesighly penetrant dominant in-gliomas, even over three gen-example, see ref. 68), which

to initiating events in gliomaad to a new antioncogene.which usually occur in chil-:casionally found in patientssposing germ-line mutation inigene, as with TP53 or NFI.one condition has regularlyiated with PNETs-namely,Irome, or basal cell nevusBCNS), whose gene has beenlinkage analysis to 9q31 (69).mutation at the BCNS locusriably develop basal-cell car-the skin, and approximatelye tumors reveal loss of het-for 9q markers, suggestinge of an antioncogene on thatal arm (70). A few percentagewith the syndrome developId =25% of nonhereditarytomas also reveal loss of het-for 9q markers (70). It isthat syndrome subjects within some cases received cra-radiation, which in turn in-numbers of cutaneous basalmas, with a short latent pe-iting the induction of numer-I hits" that led to cancer (71).

constitute the largest cate-an cancer and take the largestrhey are also the most com-n the vantage point of genet-here has been considerablerecent years, with a promiseSome in the near future.Hippel-Lindau (VHL) Syn-Renal Cell Carcinoma. Thisinherited syndrome is char-y nonneoplastic phenotypicLt facilitate its identification.,ery carrier of the VHL mu-lops one or more of threencer: hemangioblastoma ofiheochromocytoma, or renal

Review: Knudson

Dow

nloa

ded

by g

uest

on

Dec

embe

r 27

, 202

0

Proc. Natl. Acad. Sci. USA 90 (1993)

cell carcinoma (RCC). Curiously, the lasttwo tumors are uncommon in the samefamily, although bilateral tumors are fre-quent at each site when they do occur(72). The gene was localized by linkageanalysis to chromosomal band 3p25 andrecently cloned (73). Its protein productappears to be a cell surface moleculeinvolved in cell adhesion and signal trans-duction. It will be of interest to discoverwhether the protein interferes with thefunction of an oncogene that is involvedin signal transduction. Examination ofnonhereditary renal carcinomas revealsabnormality of VHL in a high percentageof cases, making it the principal RCCgene. Both of these nonhereditary tu-mors and the tumors found in the syn-drome typically reveal mutation or loss ofthe second copy of the gene, as expectedfor an antioncogene. However, one fam-ily with high penetrance for RCC carrieda constitutional translocation with onebreakpoint at 3p14-21 far from the VHLgene (74). It seems likely that 3p harborsa second RCC gene, although it is possi-ble that the tumors in translocation pa-tients have sustained 3p deletions thatcause loss of the VHL gene. Phenotypi-cally, the tumors in both inherited formsare ofthe common clear cell histology. Inaddition, there are nonclear cell tumorsthat do not show LOH for chromosome3p markers, suggesting the existence ofyet another renal cancer gene. Nonclearcell RCC is known in hereditary form inthe Eker rat, and the gene has beenlocalized to rat chromosome 10q, in aregion that seems to be homologous inpart with 16p in humans (75). In neitherhumans nor rats do predisposed genecarriers develop Wilms tumor, nor docarriers of a Wilms tumor mutation de-velop RCC. Here again there is specific-ity for a developmental state in an organ,as with the adrenal medulla for neuro-blastoma and pheochromocytoma.

Familial Adenomatous Polyposis andColon Cancer. One of the best knownhereditary conditions that predispose to amajor cancer ofadults is familial adenom-atous polyposis (FAP), which occurs inpersons heterozygous for a mutation inthe adenomatous polyposis coli (APC)gene. APC is located at chromosomalband 5q21 and has been cloned (76, 77).Its protein product is located in the cy-toplasm and has features suggesting po-tential for interaction with other proteins,perhaps the product of some oncogene.In typical FAP cases, the colonic mucosais studded with hundreds or even thou-sands of benign adenomatous polyps.These polyps, which begin to appeareven in the first decade of life, are clonalin origin. Polyps of this same type occursporadically in normal individuals; cellsof even very small polyps are mutatedsomatically at the APC locus (78). It hasbeen reported that many, conceivably

even all, polyps have sustained muta-tions or loss of both copies of the APCgene (79, 80).The crypts of the colonic mucosa in

FAP patients show a larger than normalproliferative compartment (81); in effecttheAPC mutation causes an expansion ofthe target cell compartment. This phe-nomenon is not clonal and appears to bea dominant effect of the constitutionalmutation, even though some of the mu-tations involve total deletions of thegene. However, the mutations most oftenintroduce termination codons that causeextreme truncation of the protein prod-uct, which can interact with the normalallele's product, interfering with the lat-ter's function and causing the mutation tobehave in a "dominant negative" man-ner, as with some mutations in TP53 (82).It will be interesting to compare the phe-notypic effects of null and truncatingmutations.

Mutations at the APC gene in themouse produce multiple intestinal neo-plasia (Min)-i.e., polyps and carcino-mas (83-85). Multiple cell lineages in thetumors suggest that the mutation exertsits effect upon a pluripotent stem cell.Homozygotes for the mutation die duringfetal life (85), indicating that the gene hasan important role in normal develop-ment, as is true also for RBI and WTI.Even if mutation at the APC locus is a

necessary condition for carcinoma of thecolon, it does not suffice. Other geneticevents are the rule for this tumor, thetargets being the KRAS oncogene, theTP53 antioncogene (86), and the DCC(Deleted in Colon Cancer) antioncogene(87). KRAS mutations are featured inlarge polyps, where their incidence is-400o. They are rare in very small polyps,and their incidence in carcinomas is aboutthe same as in large polyps. They seem notto be directly involved in malignant trans-formation ofthe latter. On the other hand,TP53 and DCC mutations occur uncom-monly in polyps but in the majority ofcarcinomas. It has not been establishedwhether there is a necessary sequence formutation in these two genes.The DCCgene was identified following

LOH studies that showed frequent lossesfor markers on chromosome 18q (87).The gene encodes a cell surface moleculewith considerable homology to the celladhesion molecule of neural cells (N-CAM) and probably plays a role in self-recognition in colonic epithelium. Thecharacterized mutations usually producetermination codons that in turn lead totruncated, presumably nonfunctional,proteins. No normal copies of DCC re-main in the tumor cells, as expected foran antioncogene. This is the only antion-cogene that has been cloned without theavailability of constitutional mutations.So far there is no hereditary condition inwhich DCC is mutant. DCC, for exam-

ple, is expressed in neural tissue as wellas in the gastrointestinal tract (88), andone patient with a constitutional deletionof 18q, the arm that contains the DCCgene, developed a brain tumor (89).Could it be that germ-line mutations ofDCC would be associated with brain tu-mors rather than with colon cancer?Another hereditary condition, known

as the Lynch cancer family syndrometype 2, imparts susceptibility to numer-ous types of carcinoma, including that ofthe colon. A gene for it (there may bemore than one), LCFS2, has recentlybeen mapped to chromosome 2, a chro-mosome not previously implicated in co-lon carcinogenesis (90). Although thegene has not yet been cloned, one re-markable property of it has been de-scribed. Tumors in heterozygous carriersshowwidespreadrearrangementofmicro-satellite sequences (91-93) but not loss ofheterozygosity for chromosome 2 mark-ers, as expected for an antioncogene.However, they do show incidences ofmutation at KRAS, TP53, and APC thatare comparable with those in colon can-cer in general. If the LCFS2 mutationaccelerates the rates of mutation at APC,TP53, and DCC, formation of polyps andtheir transformation to malignant carci-nomas should be expedited.

Familial Breast Cancer. The clusteringof cases of breast cancer in families haslong been known. Disentangling thecause of this phenomenon as due to he-redity, environment, or chance has beendifficult. Particularly striking are familiesin which female descendants of an af-fected female are affected through two ormore later generations, often at an earlierthan usual age. Two genes are nowknown to contribute to this phenome-non-namely, TP53, in the LFS, and thebreast cancer 1 (BRCAI) gene. Constitu-tional mutations of this latter gene pre-dispose heterozygous carriers to carcino-mas of the breast and ovary. The gene hasbeen mapped to chromosomal band17q21, and vigorous attempts are under-way to clone it (94, 95). Breast and ova-rian cancers from affected women inthese families reveal a high incidence ofloss of the wild-type allele, as expectedfor an antioncogene (96). The gene prob-ably also plays a major role in nonhered-itary cancer of these tissues. The fre-quency of heterozygous carriers of thegene can be estimated to be as high as 1per 1000 (94). Such a high frequency fora serious single gene condition is extraor-dinary and, if true, would imply virtuallyno mortality before the end of the repro-ductive period. This gene is apparentlythe most common among those that pre-dispose to breast cancer and perhaps themost common that predisposes to anycancer.Lung Cancer. There are no pedigrees

that unequivocally demonstrate segrega-

10918 Review: Knudson

Dow

nloa

ded

by g

uest

on

Dec

embe

r 27

, 202

0

Proc. Natl. Acad. Sci. USA 90 (1993) 10919

tion of a dominantly inherited predispo-sition to lung cancer, possibly becausesuch a very large fraction of all cases isattributable to cigarette smoking. Studiesof cytogenetic and molecular changes intumors point to certain genes or chromo-somal regions. Mutations at RBJ andTP53 are common (97, 98), especially insmall-cell lung cancer (SCLC), although,as noted previously, this cancer is not acharacteristic feature of hereditary retin-oblastoma or of the LFS. A third change,found almost universally in SCLC, and ina majority of cases of non-SCLC, is de-letion ofchromosome 3pl4-p21 (99, 100).There is at least one lung cancer antion-cogene (LCI) on chromosome 3p. SCLCis similar to colon cancer in that threedifferent antioncogenes appear to be crit-ical to both, with p53 being common tothem. A difference is that none of thethree lung cancer genes (RBI, TP53, orLC1) resembles APC by providingstrongly predisposing germ-line muta-tions.

Prostate Cancer. The genetics of pros-tate cancer is poorly understood, but theanalysis of familial cases has led to theconclusion that there is a dominantlyinherited predisposing gene (101). A pos-sible clue to the location of such a genehas been provided by cytogenetic andloss of heterozygosity studies in tumors,which have revealed a few recurrent ab-normalities. Deletion commonly occursat 8p22, and was even homozygous in onecase (102). Deletion at 10q24 has beenreported as the sole anomaly in sometumors and displayed clonal evolution inone patient (103). Perhaps one of thecandidate antioncogenes at these chro-mosomal sites is mutant in the hereditaryform of this cancer. Other somatic mu-tations will probably prove to be impor-tant too, as happens with other carcino-mas.

General Comments on Antioncogenesand the Carcinomas. The common andvery important carcinomas, especially ofbreast, colon, and lung, have been dis-cussed in part, especially in connectionwith the TP53, APC, LCFS2, andBRCAJgenes. The phenomenon, noted for RBIand TP53, that carcinomas can show so-matic mutations at these loci, withoutbelonging to the constellation of tumorsto which germ-line mutations predisposetheir hosts, apparently relates to the factthat carcinomas often show mutations atmultiple loci, and no one of them wouldhave a great impact in the germ line. APCmutations have the unusual property ofcausing an expansion of the renewal pro-liferative compartment and, judging fromstudies on the Min mouse, thereby ex-panding the number of stem cells that areavailable to sustain other oncogenicevents. The colon cells assume proper-ties that are similar to those of prolifer-ating embryonal stem cells, such as retin-

Table 2. Cloned human tumor suppressor genes: protein products and homozygous mutanteffects in miceGene Protein Mutant mouseRBI 110-kDa transcription modulator, Fetal lethal

regulator of G1 -* S transitionWTI 45-kDa transcription factor, zinc finger Fetal lethal

proteinTP53 53-kDa transcription factor, conditional Not lethal; predisposition

regulator G1-+ S to tumorsNFI 327-kDa activator of Ras GTPase activity Not reportedNF2 66-kDa protein at membrane-cytoskeleton Not reported

interfaceVHL 34-kDa cell membrane protein Not reportedAPC 310-kDa cytoplasmic protein Fetal lethalDCC 153-kDa cell adhesion molcule Not reported

oblasts and nephroblasts, and the stemcells that are stimulated by puberty andadolescence. The stem cells of normalrenewal tissues do not expand in number,so some environmental or genetic stim-ulus must be provided. Evidently, germi-nal mutations in TP53 do not providesuch an impetus, which may explain whybreast cancer, but not colon or lung can-cer, is a distinctive feature of the LFS.Embryonal tissues require the fewest on-cogenic events; conditional growth tis-sues, an intermediate number; and re-newal tissues, the most.There are other genes whose mutations

are highly penetrant for carcinomas, butnone of them is so obvious as APC orBRCAI. Thus, rare pedigrees with carci-nomas of the stomach, pancreas, or blad-der have been reported, but there hasbeen no progress in mapping the relevantgenes.

Putative Leukemia and Lymphoma An-tioncogenes. Many leukemias and lym-phomas reveal somatic chromosomaltranslocations that activate or rearrangecellular protooncogenes. This mecha-nism for initiation of cancer is evidentlylimited to hematopoietic and a few un-common neoplasms, such as Ewing sar-coma and alveolar rhabdomyosarcoma.The reason for such a limitation of targetsites is not apparent. However, someleukemias seem to involve the antionco-gene mechanism. For example, leukemiais prominent in LFS. Furthermore, TP53mutations are found in some leukemiasand lymphomas, although it is uncertainwhether they are changes associated withinitiation or with progression. There aresome pedigrees with many cases of leu-kemia or lymphoma, usually rather ho-mogeneous for acute leukemia, chroniclymphocytic leukemia, or non-Hodgkinlymphoma. None ofthese genes has beenmapped, and there is no evidence bearingon their identity as oncogenes or antion-cogenes.

The Search for New Antioncogenes

So far eight antioncogenes (RBI, WTI,VHL, APC, DCC, TP53, NFI, and NF2)

have been cloned. All except DCC havebeen found to be mutated in the germlines of persons predisposed to one ormore tumor types. Another seven puta-tive antioncogenes (BRCA1, RCC, NBI,MLM, MEN), BCNS, and LCI) havebeen mapped but not cloned. Germinalmutations in all except LCI are known topredispose to tumors. Obviously theanalysis of hereditary cancer providesexcellent entry to the world of antionco-genes.The numerous reports of familial can-

cers at virtually all sites indicate thatmany more antioncogenes are to be dis-covered. The list of antioncogenes couldreach 50 or so from the source of hered-itary cancer alone. There is a problem,however. The most common conditionsare those that are mapped first, and italready appears that the remaining he-reditary cancers will be so rare as torender linkage analysis difficult. Ofcourse, rare cases with constitutional de-letions have pointed to genetic sites in thepast and will continue to do so. Similarly,LOH studies can be used, as was donewith DCC in colon cancers, although thisis a very difficult process. However,there may be numerous genes whosemutations are rare or lethal in the germline that could be important in the originof tumors. Some of these genes may evenoccur as germ-line mutations but producetumors very different from those found inthe nonhereditary form. Thus, the dis-covery of somatic mutations in RBI insmall-cell carcinoma of the lung wouldhardly have prepared one for its produc-tion of retinoblastoma in carriers of thegerm-line mutation.Another approach to the discovery of

new members of this class of cancer-predisposing gene could involve mecha-nistic analysis of oncogenes and of thesignal transduction pathway. The earlyproposals that antioncogenes might en-code cell membrane molecules (2) or neg-ative transcription factors (3) were bothcorrect in a way. The antioncogenes thathave been cloned include three genes(RBI, WTI, and TP53) whose proteinproducts interact with DNA or with tran-

Dmv-lview: Knudson

Dow

nloa

ded

by g

uest

on

Dec

embe

r 27

, 202

0

Proc. Natl. Acad. Sci. USA 90 (1993)

scription factors to regulate negativelythe expression of other genes, which arein effect oncogenes (Table 2). The APCand NFI gene products appear to func-tion in the cytoplasm, whereas the prod-ucts of NF2, VHL, and DCC seem to belocated at the cell membrane. If oneassumes that positive and negative regu-lation occurs at all points in the signaltransduction pathway, then it may be thatthe study of the control of oncogeneactivity could lead to identification ofnew inhibitory factors that would be can-didates for antioncogenes.

This work was supported in part by NationalInstitutes of Health Grant CA-06927 and agrant from the Lucille P. Markey CharitableTrust.

1. Knudson, A. G. (1971) Proc. Natl.Acad. Sci. USA 68, 820-823.

2. Knudson, A. G. (1973) Adv. CancerRes. 17, 317-352.

3. Comings, D. E. (1973) Proc. Natl.Acad. Sci. USA 70, 3324-3328.

4. Knudson, A. G. (1978) Semin. Oncol.5, 57-60.

5. Hethcote, H. W. & Knudson, A. G.(1978) Proc. Natl. Acad. Sci. USA 75,2453-2457.

6. Knudson, A. G., Meadows, A. T.,Nichols, W. W. & Hill, R. (1976) N.Engl. J. Med. 295, 1120-1123.

7. Francke, U. & Kung, F. (1976) Med.Pediatr. Oncol. 2, 379-385.

8. Cavenee, W. K., Dryja, T. P., Phillips,R. A., Benedict, W. F., Godbout, R.,Gallie, B. L., Murphree, A. L., Strong,L. C. & White, R. L. (1983) Nature(London) 305, 779-784.

9. Friend, S. H., Bernards, R., Rogelj, S.,Weinberg, R. A., Rapaport, J. M., Al-bert, D. M. & Dryja, T. P. (1986) Na-ture (London) 323, 643-646.

10. Lee, W.-H., Shew, J.-Y., Hong, F. D.,Sery, T. W., Donoso, L. A., Young,L.-J., Bookstein, R. & Lee, E. Y.-H. P. (1987) Nature (London) 329, 642-645.

11. Fung, Y.-K. T., Murphree, A. L.,T'Ang, A., Qian, J., Hinrichs, S. H. &Benedict, W. F. (1987) Science 236,1657-1661.

12. Huang, H.-J. S., Yee, J.-K., Shew,J.-Y., Chen, P.-L., Bookstein, R.,Friedman, T., Lee, E. Y.-H. P. & Lee,W.-H. (1988) Science 242, 1563-1566.

13. Goodrich, D. W. & Lee, W.-H. (1993)Biochim. Biophys. Acta 1155, 43-61.

14. Knudson, A. G. (1992) Semin. CancerBiol. 3, 99-106.

15. Clarke, A. R., Maandag, E. R., vanRoon, M., van derLugt, N. M. T., vanderValk, M., Hooper, M. L., Berns, A.& teRiele, H. (1992) Nature (London)359, 328-330.

16. Jacks, T., Fazeli, A., Schmitt, E. M.,Bronson, R. T., Goodell, M. A. &Weinberg, R. A. (1992) Nature (Lon-don) 359, 295-300.

17. Lee, E. Y.-H. P., Chang, C.-Y., Hu,N., Wang, Y.-C. J., Lai, C.-C., Herrup,K., Lee, W.-H. & Bradley, A. (1992)Nature (London) 359, 288-294.

18. Knudson, A. G. & Strong, L. C. (1972)J. Natl. Cancer Inst. 48, 313-324.

19. Riccardi, V. M., Hittner, H. M.,Francke, U., Yunis, J. J., Ledbetter, D.& Borges, W. (1980) Cancer Genet.Cytogenet. 2, 131-137.

20. Call, K. M., Glaser, T., Ito, C. Y.,Buckler, A. J., Pelletier, J., Haber,D. A., Rose, E. A., Kral, A., Yeger,H., Lewis, W. H., Jones, C. & Hous-man, D. E. (1990) Cell 60, 509-520.

21. Gessler, M., Poustka, A., Cavenee, W.,Neve, R. L., Orkin, S. H. & Bruns,G. A. P. (1990) Nature (London) 343,774-778.

22. Rauscher, F. J. (1993) FASEB J. 7,896-903.

23. Kreidberg, J. A., Sariola, H., Loring,J. M., Maeda, M., Pelletier, J., Hous-man, D. & Jaenisch, R. (1993) Cell 74,679-691.

24. Koufos, A., Grundy, P., Morgan, K.,Aleck, K. A., Hadro, T., Lampkin,B. C., Kalbakji, A. & Cavenee, W. K.(1989) Am. J. Hum. Genet. 44, 711-719.

25. Ogawa, O., Eccles, M. R., Szeto, J.,McNoe, L. A., Yun, K., Maw, M. A.,Smith, P. J. & Reeve, A. E. (1993) Na-ture (London) 362, 749-755.

26. Rainier, S., Johnson, L. A., Dobry,C. J., Ping, A. J., Grundy, P. E. &Feinberg, A. P. (1993) Nature (London)362, 747-749.

27. Weksberg, R., Shen, D. R., Fei, Y. L.,Song, Q. L. & Squire, J. (1993) NatureGenet. 5, 143-149.

28. Malkin, D. (1993) Cancer Genet. Cyto-genet. 66, 83-92.

29. Malkin, D., Li, F. P., Strong, L. C.,Fraumeni, J. F., Nelson, C. E., Kim,D. H., Kassel, J., Gryka, M. A.,Bischoff, F. Z., Tainsky, M. A. &Friend, S. H. (1990) Science 250, 1233-1238.

30. Li, F. P. & Fraumeni, J. F. (1969) J.Natl. Cancer Inst. 43, 1365-1373.

31. Diller, L., Kassel, J., Nelson, C. E.,Gryka, M. A., Litwak, G., Gebhardt,M., Bressac, B., Ozturk, M., Baker,S. J., Vogelstein, B. & Friend, S. H.(1990) Mol. Cell. Biol. 10, 5772-5781.

32. Kastan, M. B., Onyekwere, O., Sidran-sky, D., Vogelstein, B. & Craig, R. W.(1991) Cancer Res. 51, 6304-6311.

33. Lavigueur, A., Maltby, V., Mock, D.,Rossant, J., Pawson, T. & Bernstein, A.(1989) Mol. Cell. Biol. 9, 3982-3991.

34. Donehower, L. A., Harvey, M., Slagle,B. L., McArthur, M. J., Montgomery,C. A., Butel, J. S. & Bradley, A. (1992)Nature (London) 356, 214-221.

35. Sherley, J. L. (1991) J. Biol. Chem. 266,24815-24828.

36. Pietenpol, J. A. & Vogelstein, B. (1993)Nature (London) 365, 17-18.

37. Knudson, A. & Strong, L. (1972)Am. J.Hum. Genet. 24, 514-532.

38. Brodeur, G. M., Sekhon, G. S. & Gold-stein, M. N. (1977) Cancer 40, 2256-2263.

39. Weith, A., Martinsson, T., Cziepluch,C., Bruderlein, S., Amler, L. C. &Berthold, F. (1989) Genes Chromo-somes Cancer 1, 159-166.

40. Laureys, G., Speleman, F., Opdenak-ker, G. & Leroy, J. (1990) Genes Chro-mosomes Cancer 2, 252-254.

41. Biegel, J. A., White, P. S., Marshall,H. N., Fujimori, M., Zackai, E. H.,Scher, C. D., Brodeur, G. M. & Eman-uel, B. S. (1993) Am. J. Hum. Genet.52, 176-182.

42. Cawthon, R. M., Weiss, R., Xu, G.,Viskochil, D., Culver, M., Stevens, J.,Robertson, M., Dunn, D., Gesteland,R., O'Connell, P. & White, R. (1990)Cell 62, 193-201.

43. Wallace, M. R., Marchuk, D. A., An-derson, L. B., Letcher, R., Odeh,H. M.,Saulino,A. M.,Fountain,J. W.,Brereton, A., Nicholson, J., Mitchell,A. L., Brownstein, B. H. & Collins,F. S. (1990) Science 249, 181-186.

44. Xu, G., O'Connell, P., Viskochil, D.,Cawthon, R., Robertson, M., Culver,M., Dunn, D., Stevens, J., Gesteland,R., White, R. & Weiss, R. (1990) Cell62, 599-608.

45. Ballester, R., Marchuk, D., Boguski,M., Sauline, A., Letcher, R., Wigler,M. & Collins, F. (1990) Cell 63, 851-859.

46. Martin, G. A., Viskochil, D., Bollag,G., McCabe, P. C., Crosier, W. J.,Haubrock, H., Conroy, L., Clark, R.,O'Connell, P., Cawthon, R. M., Innis,M. A. & McCormick, F. (1990) Cell 63,843-849.

47. Rodenhiser, D. I., Singh, S. M., Coul-ter-Mackie, M. B. & Jung, J. H. (1993)Am. J. Hum. Genet. 53 (Suppl.), 1223(abstr.).

48. Skuse, G. R., Kosciolek, B. A. & Row-ley, P. T. (1989) Genes ChromosomesCancer 1, 36-41.

49. Glover, T. W., Stein, C. K., Legius,E., Anderson, L. B., Brereton, A. &Johnson, S. (1991) Genes Chromo-somes Cancer 3, 62-70.

50. Xu, W., Mulligan, L. M., Ponder,M. A., Liu, L., Smith, B. A., Mathew,C. G. P. & Ponder, B. A. J. (1992)Genes Chromosomes Cancer 4, 337-342.

51. Menon, A. G., Anderson, K. M., Ric-cardi, V. M., Chung, R. Y., Whaley,J. M., Yandell, D. W., Farmer, G. E.,Freiman, R. N., Lee, J. K., Li, F. P.,Barker, D. F., Ledbetter, D. H., Klei-der, A., Martuza, R. L., Gusella, J. F.& Seizinger, B. R. (1990) Proc. Natl.Acad. Sci. USA 87, 5435-5439.

52. Rouleau, G. A., Merel, P., Lutchman,M., Sanson, M., Zucman, J., Marineau,C., Hoang-Xuan, K., Demczuk, S.,Desmaze, C., Plougastel, B., Pulst,S. M., Lenoir, G., Bijlsma, E., Fash-old, R., Dumanski, J., deJong, P.,Parry, D., Eldridge, R., Aurias, A.,Delattre, 0. & Thomas, G. (1993) Na-ture (London) 363, 515-521.

53. Trofatter, J. A., MacCollin, M. M.,Rutter, J. L., Murrell, J. R., Duyao,M. P., Parry, D. M., Eldridge, R.,Kley, N., Menon, A. G., Pulaski, K.,Haase, V. H., Ambrose, C. M., Mun-roe, D., Bove, C., Haines, J. L., Mar-tuza, R. L., MacDonald, M. E., Seiz-inger, B. R., Short, M. P., Buckler,A. J. & Gusella, J. F. (1993) Cell 72,791-800.

54. Cannon-Albright, L. A., Goldgar,D. E., Meyer, L. F., Lewis, C. M., An-derson, D. E., Fountain, J. W., Hegi,

10920 Review: Knudson

Dow

nloa

ded

by g

uest

on

Dec

embe

r 27

, 202

0

Proc. Natl. Acad. Sci. USA 90 (1993) 10921

M. E., Wiseman, R. W., Petty, E. M.,Bale, A. E., Olopade, 0. I., Diaz,M. O., Kwiatkowski, D. J., Piepkorn,M. W., Zone, J. J. & Skolnick, M. H.(1992) Science 258, 1148-1152.

55. Petty, E. M., Gibson, L. H., Fountain,J. W., Bolognia, J. L., Yang-Feng,T. L., Housman, D. E. & Bale, A. E.(1993) Am. J. Hum. Genet. 53, 96-104.

56. Bale, S. J., Dracopoli, N. C., Tucker,M. A., Clark, W. H., Fraser, M. C.,Stanger, B. Z., Green, P., Donis-Keller, H., Housman, D. E. & Greene,M. H. (1989) N. Engl. J. Med. 320,1367-1372.

57. Fountain, J. W., Karayiorgou, M., Em-stoff, M. S., Kirkwood, J. M., Vlock,D. R., Titus-Ernstoff, L., Bouchard,B., Vijayasaradhi, S., Houghton,A. N., Lahti, J., Kidd, V. J., Housman,D. E. & Dracopoli, N. C. (1992) Proc.Natl. Acad. Sci. USA 89, 10557-10561.

58. Miyakoshi, J., Dobler, K. D., Allalu-nis-Turner, J., McKean, J. D. S.,Petruk, K., Allen, P. B. R., Aronyk,K. N., Weir, B., Huyser-Wierenga, D.,Fulton, D., Urtasun, R. C. & Day,R. S. (1990) Cancer Res. 50, 278-283.

59. Diaz, M. O., Ziemin, S., LeBeau,M. M., Pitha, P., Smith, S. D., Chilcote,R. R. & Rowley, J. D. (1988) Proc. Natl.Acad. Sci. USA 85, 5259-5263.

60. Olopade, 0. I., Buchhagen, D. L., Ma-lik, K., Sherman, J., Nobori, T., Bader,S., Nau, M. M., Gazdar, A. F., Minna,J. D. & Diaz, M. 0. (1993) Cancer Res.53, 2410-2415.

61. Hino, O., Klein-Szanto, A. J. P., Freed,J. J., Testa, J. R., Brown, D. Q., Vilen-sky, M., Yeung, R. S., Tartof, K. D. &Knudson, A. G. (1993) Proc. Natl.Acad. Sci. USA 90, 327-331.

62. Larsson, C., Skogseid, B., Oberg, K.,Nakamura, Y. & Nordenskjold, M.(1988) Nature (London) 332, 85-87.

63. Mathew, C. G. P., Chin, K. S., Eas-ton, D. F., Thorpe, K., Carter, C.,Liou, G. I., Fong, S.-L., Bridges,C. D. B., Haak, H., Kruseman,A. C. N., Schifter, S., Hansen, H. H.,Telenius, H., Telenius-Berg, M. & Pon-der, B. A. J. (1987) Nature (London)328, 527-528.

64. Mathew, C. G. P., Smith, B.A.,Thorpe, K., Wong, Z., Royle, N. J.,Jeffreys, A. J. & Ponder, B. A. J.(1987) Nature (London) 328, 524-526.

65. Mulligan, L. M., Kwok, J. B. J., Hea-ley, C. S., Elsdon, M. J., Eng, C.,Gardner, E., Love, D. R., Mole, S. E.,Moore, J. K., Papi, L., Ponder, M. A.,Telenius, H., Tunnacliffe, A. & Ponder,B. A. J. (1993) Nature (London) 363,458-460.

66. Sidransky, D., Mikkelsen, T.,Schweckheimer, K., Rosenblum,M. L., Cavanee, W. & Vogelstein, B.(1992) Nature (London) 355, 846-847.

67. Rasheed, B. K. A., Fuller, G. N.,Friedman, A. H., Bigner, D. D. & Big-ner, S. H. (1992) Genes ChromosomesCancer 5, 75-82.

68. Thuwe, I., Lundstrom, B. & Walinder,J. (1979) Lancet 1, 504.

69. Gailani, M. R., Bale, S. J., Leffell,D. J., DiGiovanni, J. J., Peck, G. L.,Poliak, S., Drum, M. A., Pastakia, B.,

McBride, 0. W., Kase, R., Green, M.,Mulvihill, J. J. & Bale, A. E. (1992) Cell69, 111-117.

70. Bare, J. W., Bonifas, J. M., Master,S. P., Kerschmann, R. L., Daneshvar,L., Cogan, P. H., Menon, A. G. & Ep-stein, E. (1993) Am. J. Hum. Genet. 53(Suppl.), 274 (abstr.).

71. Strong, L. C. (1977) in Genetics ofHu-man Cancer, eds. Mulvihill, J. J.,Miller, R. W. & Fraumeni, J. F.(Raven, New York), pp. 401-415.

72. Neumann, H. P. H. & Wiestler, 0. D.(1991) Lancet 337, 1052-1054.

73. Latif, F., Tory, K., Gnarra, J., Yao, M.,Duh, F.-M., Orcutt, M. L., Stackhouse,T., Kuzmin, I., Modi, W., Geil, L.,Schmidt, L., Zhou, F., Li, H., Wei,M. H., Chen, F., Glenn, G., Choyke, P.,Walther, M. M., Weng, Y., Duan, D.-S. R., Dean, M., Glavac, D., Richards,F. M., Crossey, P. A., Ferguson-Smith,M. A., LePaslier, D., Chumakov, I., Co-hen, D., Chinault, A. C., Maher, E. R.,Linehan, W. M., Zbar, B. & Lerman,M. I. (1993) Science 260, 1317-1320.

74. Cohen, A. J., Li, F. P., Berg, S., Mar-chetto, D. J., Tsai, S., Jacobs, S. C. &Brown, R. S. (1979) N. Engl. J. Med.301, 592-595.

75. Yeung, R. S., Buetow, K. H., Testa,J. R. & Knudson, A. G. (1993) Proc.Natl. Acad. Sci. USA 90, 8038-8042.

76. Groden, J., Thliveris, A., Samowitz,W., Carlson, M., Gelbert, L., Albert-sen, H., Joslyn, G., Stevens, J., Spirio,L., Robertson, M., Sargeant, L., Krap-cho, K., Wolff, E., Burt, R., Hughes,J. P., Warrington, J., McPherson, J.,Wasmuth, J., Le Paslier, D., Abderra-him, H., Cohen, D., Leppert, M. &White, R. (1991) Cell 66, 589-600.

77. Nishisho, I., Nakamura, Y., Miyoski,Y., Miki, Y., Ando, H., Horii, A.,Koyama, K., Utsunomiya, J., Baba, S.,Hedge, P., Markham, A., Krush, A. J.,Petersen, G., Hamilton, S. R., Nilbert,M. C.,Levy,D. B.,Bryan,T. M.,Prei-singer, A. C., Smith, K. J., Su, L.-K.,Kinzler, K. W. & Vogelstein, B. (1991)Science 253, 665-669.

78. Powell, S. M., Zilz, N., Beazer-Bar-clay, Y., Bryan, T. M., Hamilton,S. R., Thibodeau, S. N., Vogelstein, B.& Kinzler, K. W. (1992) Nature (Lon-don) 359, 235-237.

79. Ichii, S., Horii, A., Nakatsuru, S., Fu-ruyama, J., Utsunomiya, J. & Naka-mura, Y. (1992) Hum. Mol. Genet. 1,387-390.

80. Smith, K. J., Johnson, K. A., Bryan,T. M., Hill, D. E., Markowitz, S., Will-son, J. K. V., Paraskeva, C., Petersen,G. M., Hamilton, S. R., Vogelstein, B.& Kinzler, K. W. (1993) Proc. Natl.Acad. Sci. USA 90, 2846-2850.

81. Deschner, E. E. & Lipkin, M. (1975)Cancer 35, 413-418.

82. Su, L.-K., Johnson, K. A., Smith,K. J., Hill, D. E., Vogelstein, B. &Kinzler, K. W. (1993) Cancer Res. 53,2728-2731.

83. Su, L. K., Kinzler, K. W., Vogelstein,B., Preisinger, A. C., Moser, A. R.,Luongo, C., Gould, K. A. & Dove,W. F. (1992) Science 256, 668-670.

84. Moser, A. R., Dove, W. F., Roth,

K. A. & Gordon, J. I. (1992) J. CellBiol. 116, 1517-1526.

85. Fodde, R., Edelman, W., Yang, K.,Renault, B., Breukel, C., van Leeuwen,C., Lipkin, M., Khan, P. M. &Kucherlapati, R. (1993) Am. J. Hum.Genet. 53 (Suppl.), 297 (abstr.).

86. Fearon, E. R. & Vogelstein, B. (1990)Cell 61, 759-767.

87. Fearon, E. R., Cho, K. R., Nigro,J. M., Kern, S. E., Simons, J. W.,Ruppert, J. M., Hamilton, S. R., Preis-inger, A. C., Thomas, G., Kinzler,K. W. & Vogelstein, B. (1990) Science247, 49-56.

88. Lawlor, K. G. & Narayanan, R. (1992)Cell Growth Differ. 3, 609-616.

89. Faulkner, K. W., Holmes, L. B., Stein-feld, A. & Abrams, I. F. (1983) J. Pe-diatr. 103, 600-602.

90. Peltomaki, P., Aaltonen, L. A., Sis-tonen, P., Pylkkanen, L., Mecklin,J.-P., Jarrinen, H., Green, J. S., Jass,J. R., Weber, J. L., Leach, F. S., Pe-tersen, G. M., Hamilton, S. R., de laChapelle, A. & Vogelstein, B. (1993)Science 260, 810-812.

91. Aaltonen, L. A., Peltomaki, P., Leach,F. S., Sistonen, P., Pylkkanen, L.,Mecklin, J.-P., Jarvinen, H., Powell,S. M., Jen, J., Hamilton, S. R., Peter-sen, G. M., Kinzler, K. W., Vo-gelstein, B. & de la Chapelle, A. (1993)Science 260, 812-816.

92. Thibodeau, S. N., Bren, G. & Schaid,D. (1993) Science 260, 816-819.

93. Ionov, Y., Peinado, M. A., Malkho-syan, S., Shibata, D. & Perucho, M.(1993) Nature (London) 33, 558-561.

94. Hall, J. M., Lee, M. K., Morrow, J.,Newman, B., Anderson, L., Huey, B.& King, M. C. (1990) Science 250,1684-1689.

95. Bishop, D. T. & Easton, D. F. (1993)Am. J. Hum. Genet. 52, 677.

96. Smith, S. A., Easton, D. F., Evans,D. G. R. & Ponder, B. A. J. (1992) Na-ture Genet. 2, 128-131.

97. Harbour, J. W., Lai, S.-L., Whang-Peng,J., Gazdar, A. F., Minna, J. D. & Kaye,F. J. (1988) Science 241, 353-357.

98. Takahashi, T., Nau, M. M., Chiba, I.,Birrer, M. J., Rosenberg, R. K., Vi-nocour, M., Levitt, M., Pass, H., Gaz-dar, A. F. & Minna, J. D. (1989) Sci-ence 296, 491-494.

99. Whang-Peng, J., Kao-Shan, C. S. &Lee, E. C. (1982) Science 215, 181-182.

100. Whang-Peng, J., Knutsen, T., Gazdar,A., Steinberg, S. M., Oie, H., Linnoila,I., Mulshine, J., Nau, M. & Minna,J. D. (1991) Genes Chromosomes Can-cer 3, 168-188.

101. Carter, B. S., Beaty, T. H., Steinberg,G. D., Childs, B. & Walsh, P. C. (1992)Proc. Natl. Acad. Sci. USA 89, 3367-3371.

102. Bova, G. S., Carter, B. S., Bussemak-ers, M. J. G., Emi, M., Fujiwara, Y.,Kyprianou, N., Jacobs, S. C., Robin-son, J. C., Epstein, J. I., Walsh, P. C.& Isaacs, W. B. (1993) Cancer Res. 53,3869-3873.

103. Arps, S., Rodewald, A., Schmalen-berger, B., Carl, P., Bressel, M. &Kastendieck, H. (1993) Cancer Genet.Cytogenet. 66, 93-99.

Review: Knudson

Dow

nloa

ded

by g

uest

on

Dec

embe

r 27

, 202

0