`Parsons, Ramon; Li, Guo-Min: Longley, Matthew; Modrich, Paul; et al
`Science; May 5, 1995; 26R, 5211; Research Library Core
`pg. 738
`
`Mismatch Repair Deficiency in Phenotypically
`Normal Human Cells
`Ramon Parsons, Guo-Min Li, Matthew Longley, Paul Modrich,
`Bo Liu, Theresa Berk, Stanley R. Hamilton, Kenneth W. Kinzler,
`Bert Vogelstein*
`
`Tumor cells in patients with hereditary nonpolyposis colorectal cancer (HNPCC) are
`characterized by a genetic hypermutablfit y caused by defects In DNA m ism atch re(cid:173)
`pai r. A su bset of HNPCC patients was found to have w id esp read mutations not on ly
`in their tumors, but also In their non-neoplastic cells. Although these patients had
`numerous mutations in all tissues examined, they had very few tumors . The hyper(cid:173)
`mutability was associated with a profound defect in mismatch repair at the biochemical
`level. These results have implications for the relation between mutagenesis and car(cid:173)
`cinogenesis, and they suggest that mismatch repair deficiency is compatible with
`normal human development.
`
`Hereditary nonpolyposis colorectal cancer
`can be caused by germline mutations of the
`mismmch repair (MMR) gene~ hMSH2,
`hMLHI, hPMSI, and hPMS2 (1-4). The
`disease is inherireJ in an autosomal Jnmi(cid:173)
`nant fashion, and non-neoplastic cells of
`affected patients cont<Jin one mutant nnd
`one wild-type allele (3 ). These non-neoplas(cid:173)
`tic cells are phenotypically normal and have
`heen assumed to be MMR proficient, as they
`have shown no evidence of genetic instabil(cid:173)
`ity or biochemically measurable MMR defi(cid:173)
`ciency in previous studies (I, 5). Indeed, the
`successful linkage of 1-INPCC to anonymous
`markers on chromosome 2p and chromo(cid:173)
`some 3p was dependent on the stable inher(cid:173)
`itance of microsatetlite DNA (6) . During
`tumor development, the wild-type copy of
`the allele inherited from the unaffected par(cid:173)
`ent is lost or mutated (3, 7). This event is
`thought to render the neoplastic cells com(cid:173)
`pletely MMR deficient, leading to a rapid
`accumulation of mutations and an acceler(cid:173)
`ated rate of neoplastic progression (3, 8).
`We wondered whether the non-neoplas(cid:173)
`tic cells of HNPCC patients might harbor a
`MMR defect. The stability of short, repeat(cid:173)
`ed sequences (microsarellites) provides an
`excellent indicator of repair proficiency, as
`these sequences are prone to misalignment
`during DNA replication (9, 10). We devel(cid:173)
`oped a more sensitive strategy for analyzing
`microsatellite changes, reasoning that if a
`rare cell in a population harbored microsat(cid:173)
`etlite alterations, the new microsatetlite al-
`
`R. Parsons, B. Liu , K. W. Kinzler, B. Vogelstein, Johns
`Hopkins Oncology Center, Baltimore, MD 21231, USA.
`G ·M . Li, M. Longley, P Modrich, Department or Bio(cid:173)
`chemistry, Howard Hughes Medical Institute, Duke Uni(cid:173)
`versity Medical Center, Durham, NC 27710, USA
`T. Berk, Familial Gl Cancer Registry, Mount Sinai Hospi·
`tal, l oronto, Ontario, Canada M5G 1 X5 .
`S R. Hamilton, DepartmenlsoiPathologyandOncology,
`Jolms Hopkins University School of Medicine, Baltimom,
`MD 21205, USA.
`
`·To whom correspondence should be addresse<J
`
`leles would not he detectable amid the large
`background of no rmal alleles. To increase
`sensitivity, we diluted DNA samples so that
`the genomes of only one-half to three cells
`(0.5 to 3 cell equivalents) were used as
`remph1r.es for each of sever81 polymerase
`chain reactions (PCRs) ( 1 I). Because any
`altered micro~atetlite allele would represent
`greater than 15% of the alleles in such
`samples, we could detect alterations present
`in a small fraction of cells.
`We evaluated Epstein-Barr virus-trans(cid:173)
`formed lymphoblast~ cultured from HNPCC
`and control patients, initially examining two
`microsntell ite markers: D2S 123, consisting
`ob (CA}, rcp ·at, anu BAT25, c:m'!si~Lingof
`an (A), repea t· ( II ). Two conrrol indi vtdua ls
`(not fru m HN PCC 11! mil! es) had no a!tcr(cid:173)
`ations in either microsar.ellite, nor did pa(cid:173)
`tients Pl (hMSH2 mutation) or P3 (hPMSI
`mutation) (Table 1 ). However, patients P2
`(hPMS2 mutation) and P4 (hMLH 1 muta(cid:173)
`tion) exhibited a number of alterations in
`both markers (Fig. I and Table 1 ). More
`than 20% of the diluted DNA samples from
`patient P2, for example, contained a novel
`microsatellite allele. A third, randomly cho(cid:173)
`sen microsatetlite marker (RAT40), consist(cid:173)
`ing of an (A}" repeat, was evahmted, and this
`
`..
`..
`
`..
`
`marker showed a similar number of alter(cid:173)
`ations (Table I). Because of their heteroge(cid:173)
`neity, these alterations were not detectable
`in undiluted DNA samples (1 2).
`The hMLHl Hnd hPMS2 gene products
`form a heterodimeric complex (I, 13 ). Al(cid:173)
`though the mic rnsar.etlite alterat ions were
`observed in lymphohlasts of pauems with
`mutations in hMLH I or hPMS2, mutations
`in these genes were nor. always associated
`with such instability. For example, patient
`PS, who had a different mutation of hMLH I
`than did patient P4, did not have microsat(cid:173)
`ellite alterations in lymphoblast DNA sam(cid:173)
`ples (Table l ). Ncv ·rc hdcss, rhc ticker o.~p
`l'le~Hed to be gcncnca lly detcnnmo::d. ru; P6. ;1
`sibling nf pat em P2 sharlnR the :1~1mc gcrn\(cid:173)
`lmc mu tation of I!PMS2, had a high l · ve l of
`variation (Table I).
`To determine whether uncultured celts
`from HNPCC patients contained similar
`alterations, we examined DNA from non(cid:173)
`neop lnstic colon n ssue ( 14). The tissue was
`microdissecred into two fmctions, un" com(cid:173)
`pose<.! predominantly of epithelial cells
`(mucosa) and the other of muscular ami
`connective
`tissue cells
`(submucosa and
`muscuh1ris propria). The two normal indi(cid:173)
`viduals and patient Pl had few or no mi(cid:173)
`in either
`fraction ,
`crosatellite variants
`whereas patients P2, P4, and P6 ha<.l nu mer(cid:173)
`ous alterations (Tahle 1 ). These alteratio ns
`were considerably more prevalent in the
`epn·helial fraction than in the n~mcpithelial
`mction (for P2 and P6, probability P <
`0.005 by x2 ) . Numerous alterations were
`also observed in DNA from epithelial ce lls
`of the urinary tract of patient P6 (Table 1 ).
`To quantitate the microsatetlite instabil(cid:173)
`ity at the cellular level. we examined indi(cid:173)
`vidual clones of lymphnhl::~sts ( 15). Twenty(cid:173)
`four clones from patient P6 were compared
`with 18 clones from P7, a patient with
`fami lial adenomarous polyposis, a heredi(cid:173)
`tary colorectal cancer syndrome not associ(cid:173)
`ated with MMR gene defects (16). The
`DNA of each clone was not diluted, so that
`only the predominant pattern in each clone
`was observed . This analysis rev 'Hh:d sub(cid:173)
`stantial alterations in the P6 clones for each
`
`Fig. 1. PCR analysis or lymphoblasts
`derived from HNPCC patients. PCR
`reactions, each containing 0.5 to 3
`cellular equivalents of DNA, were
`amplified with primers for the micro(cid:173)
`satellite marker D2S 123. Arrow(cid:173)
`heads indicate the positions of the
`major PCR products from undiluted
`template DNAs. N1 is a non-HNPCC
`control patient; P2, P4, and P6 are
`HNPCC patients with germline mu-
`lations of either hPMS2 or hMLH1
`(Table 1). Fragments with abnormal
`mobilities were present in the P2, P4, and P61anes but not in the N1 lanes. As expected from a Poisson
`distribution, some lanes contained zero or only one allele. giving rise to zero or one band, respectively,
`instead of the two that were always generated from undiluted DNA templates.
`
`..
`
`P2
`
`738
`
`SCIENCE
`
`• VOl .. 268
`
`• 5 MAY l995
`
`Reproduced with permission of the copyright owner. Further reproduction prohibited without permission .
`
`The Johns Hopkins University Exhibit JHU2007 - Page 1 of 3
`
`
`
`-·
`
`D2S123
`
`BAT26
`
`•' (-·· ,.,. •• 'T ....
`
`•.•.- I ··:r. 1.'·
`
`,.
`
`'
`
`P6 clones
`
`P6 clones
`
`..
`..
`
`'.
`~ i=.·'~J ~
`
`~. ,. ....... ~ ·· ~ .
`P7 clones
`·-~.·~~~.~~~ ~.'fi'J>l(:11;;...,,r,;..Jj,
`.
`
`-
`
`·~~ ... ~f ~ ....
`
`• -.
`
`Fig. 2. Microsatellite alterations in individual clones of lymphoblasts. DNA (undiluted) from P6 or P7
`clones was amplified with primers specific for microsatellite markers D2S123 and BAT26 (70, 14).
`Arrowheads indicate the position of alleles derived from DNA of uncloned P6 and P7 cells.
`
`of the five markers tested. Markers BAT25,
`D2Sl23, BAT26, D18S58, and AP6.3 re(cid:173)
`vealed alterations in 75, 38, 92, 67, and
`54% of P6 clones, respectively (Fig. 2). No
`alterations were observed in P7 clones with
`four markers, and only one alteration with
`the fifth (BAT25). Thb comparison be(cid:173)
`tween clones P6 and P7 was statistically
`significant (P < 0.005 by x2).
`To determine whether the observed in(cid:173)
`stability was due to a defect in MMR, we
`measured MMR activity in extracts from
`the
`lymphoblasts. Two DNA substrates
`were used for these assays, one containing a
`single GT mispair and one containing a CA
`dinucleotide insertion in one strand (17).
`Extracts from H6 cells, a tumor cell line
`containing a homozygous mutation of
`hMLHl and no wild-type hMLHJ gene (4,
`5), displayed no measurable MMR activity
`
`(Table 2 and Fig. 3). Extracts from SO cells,
`a colorectal cancer cell line without micro(cid:173)
`satellite instability (5), had MMR activity
`that was at least 20 times that of H6. Sub(cid:173)
`stantial MMR activity was also observed in
`extracts from three lymphoblastoid lines de(cid:173)
`rived from HNPCC patients P8, P9, and
`PlO, each with an hMSH2 mutation not
`resulting
`in microsatellite alterations in
`non-neoplastic cells. Extracts from the well(cid:173)
`studied non-HNPCC lymphoblastoid cell
`line TK6 (I 8) also had substantial activity
`in these assays. However, the three patients
`(P2, P4, and P6) with microsatellite insta(cid:173)
`bility in their non-neoplastic cells had little
`or no measurable activity in identically pre(cid:173)
`pared extracts (Table 2 and Fig. 3 ).
`The data described here document a
`profound MMR defect in the phenotypi(cid:173)
`cally normal cells of a subset of HNPCC
`
`Table 1. Microsatellite alterations in phenotypically normal tissues. For lymphoblastoid cells the differ(cid:173)
`ences between patients P2, P4, and P6 and the other individuals were significant (x2 test, P < 0.005,
`0.005, and 0.025, respectively). Their colonic epithelial fraction was also significantly different from the
`epithelial fraction of the other tested individuals (P < 0.01 by x 2). Epith., epithelium; Q, Gin; R, Arg; ter,
`termination codon; NO, not done.
`
`Patient
`
`Mutated
`gene
`
`Mutation•
`
`Tissue fraction
`
`% of PCR reactions revealing an
`alteration in microsatellitet
`
`BAT25
`
`D2S123
`
`BAT40
`
`N1
`
`N2
`
`P1
`
`P2
`
`P3
`P4
`
`P5
`
`P6
`
`None
`
`None
`
`hMSH2
`
`Codons 265 to
`314 deleted
`
`hPMS2
`
`R134 ter
`
`hPMS1
`hMLH1
`
`hMLH1
`
`hPMS2
`
`0233 ter
`Codon 618
`deleted
`
`Frame shift at
`codon 347
`R134 ter
`
`Lymphoblastoid
`Colon epith.
`Lymphoblastoid
`Colon epith.
`Lymphoblastoid
`Colon epith.
`Colon nonepith.
`Lymphoblastoid
`Colon epith.
`Colon nonepith.
`Lymphoblastoid
`Lymphoblastoid
`Colon epith.
`Colon nonepith.
`Lymphoblastoid
`
`Lymphoblastoid
`Colon epith.
`Colon nonepith.
`Urinary tract
`epith.:J:
`
`0
`0
`0
`4
`0
`0
`0
`22
`25
`6
`0
`7
`4
`0
`0
`
`14
`20
`6
`20
`
`0
`NO
`0
`NO
`0
`NO
`NO
`29
`NO
`NO
`0
`10
`NO
`NO
`0
`
`33
`NO
`NO
`NO
`
`0
`0
`0
`0
`NO
`0
`NO
`28
`50
`10
`NO
`12
`22
`8
`NO
`
`33
`53
`3
`33
`
`'Described in (4, 23).
`tBecause 0.5to 3 cellular equivalents of DNA were used lor each PCR, the fraction of cells with
`altered alleles can be estimated by dividing the values by a factor of 0.5 to 3.
`+Cells derived from urine sediment.
`
`SCIENCE . VOL. 268 . 5 MAY 1995
`
`50
`
`• 100 50 50
`P6 (119)
`-
`• 50
`- 50
`TK6 (119) 100
`so (119)
`- 100 •
`• 50
`Fig. 3. Nuclear extracts of P6 are deficient in
`MMR. Nuclear extracts of the cell lines TK6, SO,
`and P6 were incubated with 24 fmol of a G-T
`mismatch substrate (arrowhead) containing a sin(cid:173)
`gle-stranded nick 125 bp 5' to the mispair (17).
`MMR restores a Hind Ill endonuclease recognition
`site. Repair was scored by cleavage with Hind Ill
`and Bsp 106. Arrows indicate repair products.
`
`patients. Our results suggest that normal
`human development is compatible with
`greatly reduced levels of MMR. Although
`we have not assessed embryonic cells di(cid:173)
`rectly, we assume that the same defect
`present in the adult cells was also present
`during embryogenesis.
`One explanation for the MMR deficien(cid:173)
`cy in these patients is that the wild-type
`allele co-inherited with the mutant MMR
`gene was lost or mutated somatically, as was
`observed in HNPCC tumors. However, this
`seems unlikely because no evidence of mu(cid:173)
`tation or loss of the unaffected allele could
`be detected by se4uencing the relevant
`complementary DNA or searching for trun(cid:173)
`cated proteins ( 12). Alternatively, these pa(cid:173)
`tients might have inherited mutations of
`other genes that participate in MMR, with
`multiple germline mutations leading to a
`reduction of MMR activity. Yet no muta(cid:173)
`tions were detected in the other known
`MMR genes with the same methods. An(cid:173)
`other explanation for the observed deficien-
`
`Table 2. Mismatch repair activity of nuclear ex(cid:173)
`tracts. The extracts were tested for MMR activity
`with 24 fmol of mismatched substrate (18).
`
`Cell line
`
`Repaired substrate (fmol)
`
`3' CA
`
`5' G-T
`
`P8
`P9
`P10
`P4
`P2
`P6
`
`TK6
`
`H6
`so
`
`HNPCC /ymphoblastoid
`4.9
`3.5
`2.3
`<0.3
`<0.3
`<0.3
`Controllymphoblastoid
`7.6
`HNPCC colorectal cancer
`<0.2
`Sporadic co/orectal cancer
`11
`
`8.6
`2.0
`3.5
`<0.3
`<0.3
`<0.3
`
`8.6
`
`<0.2
`
`5.8
`
`739
`
`Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
`
`The Johns Hopkins University Exhibit JHU2007 - Page 2 of 3
`
`
`
`cy would entail a dominant negative effect
`of specific hPMS2 and hMLHJ mutations.
`The mutated proteins resulting from such
`mutations might be able to interact with
`hMSH2 bound to mismatched DNA but
`not be able to recruit the enzymes necessary
`for proper excision and repair. Although
`inhibition of repair was not seen in mixing
`experiments (Fig. 3 ), it is possible that the
`mutant proteins might have been seques(cid:173)
`tered in a preexisting protein complex.
`Given their elevated rates of mutation,
`it was surprising that these patients did
`not have more colorectal cancers (CRCs).
`Patient PZ had no CRC at age 14, P6 had
`one CRC at age 12, and P4 had two CRCs
`at age 31. An exponential increase in
`CRC formation would be expected from
`standard models of tumorigenesis which
`assume that multiple rate-limiting muta(cid:173)
`tions drive the process [reviewed in ( /9)).
`One interpretation of these observations is
`that mutations per se may not be sufficient
`for a high rate of tumorigenesis. lt has
`been argued that exogenous mutagens are
`carcinogenic not only because they induce
`mutations but also because they induce
`them in cells actively regenerating as a
`result of the extensive cellular damage
`associated with mutagenic compounds
`( 20). Although cells in normal environ(cid:173)
`ments undergo apoptosis when damaged
`(21 ), cells in regenerating tissues may
`have their apoptotic pathways repressed,
`allowing cells with mutations in growth(cid:173)
`controlling genes to expand clonally. The
`patients described here h<~t.i a continuously
`high mutation rate, but perhaps their cells
`did not receive the tissue regenerative
`stimulus afforded by exposure to high con(cid:173)
`centrations of mutagens, and this may ex(cid:173)
`plain why they did not Jevelop larger
`numbers of tumors.
`These results also have potential clinical
`implications. Drugs designed to be lethal to
`MMR-deficient tumor cells would likely be
`toxic to the non-neoplastic cells of the
`MMR-deficient patients described here.
`Careful evaluation of the germline defects
`
`in HNPCC patients would therefore seem
`critical once such therapeutic agents are
`developcJ.
`Finally, th results suggest a srraregy for
`making any cell MMR-deficiem by the
`transfer of genes encoJing the mut:anr pro(cid:173)
`le ins found in patients P2 and P4. The
`expression of the altered gene pruJucrs
`could be drivt:n by cell rype--spedfic pro(cid:173)
`moters to crearc highly mumble cell types
`in
`transgenic organisn\s. This strategy
`might he useful for a variety of experimen(cid:173)
`tal purposes.
`
`REFERENCES AND NOTES
`
`1. P. Modrich, Science 266, 1959 (1994)
`2. R. Fishel et al., Ce/175, 1027 (1993); F, Palombo, M.
`Hughes, J. Jlrlcny, 0 . Truong, J. HsuM, Natl.l(e 367,
`417 (1994); C. E. Bronner et a/., Ibid. 368, 258
`(1994).
`3. F. S. Leach et al., Ce/175, 1215 (1993)
`4. N. Papadopoulos et al., SCience 263, 1625 (1994);
`N. C. Nicole1des et al .. Nature 371, 75 (1994); B. L1u
`et al., Cancer Res. 54, 4590 (1994).
`5. R. Parsons eta/., Ce/175, 1227 (1993).
`6. P. Peltomtlki at a/., Science 260, 810 (1993); A. Lind·
`blom, P. Tannergard. B. Werellus, M. Nordenskjold,
`Nat. Gene>!. 5, 279 (1993).
`7. A. Hemminki eta/., Nat. Genet. 8, 405 (1 994); B. Uu
`eta/., ibid. 9, 48 (1995).
`8. J. R. Jass and S Stewart, Gut 33, 783 (1992).
`9. Y. M. lonov, A. Peinado, S. Malkhosyan, D. Shibata,
`M. Perucho, Nature 363, 558 (1993).
`10. S. N. Thibodeau, G. Bran, D. Schaid, Science 260,
`816 (1993); L.A. Aattonan er al., ibid., p. 812; G.
`Levinson and G. A. Gutman, Nucleic Acids Res. 15,
`5323 (1 987); M. Strand, T. A. Prolla. R. M. Llskay. T
`D. Petes, Nature 365, 274 (1993).
`11. DNA was diluted to 0.5 to 3 cellular equivalents per
`reaction and used in multiple parallel PCR reactions
`of 40 cycles PCR was perlon!'led 1n 96·well plates
`wit~ 11 :111P-Iabaled primer (> 1 0" opmlfl-91 as de(cid:173)
`r.oribad IJ. Jen et aJ .. N. Engl. J Med. 331 , 213
`( 1994)), Sample~ were er~alyzed by electrophoresos tn
`uree-formamldo fJ(llyacrylamrde gels, fixed In math·
`anol- acellc acid (5% oach) tor 20 min. dried, and
`(!)<posed to frfm. PCR markers lncluchJ 02$123
`(primers 5'-AAACAGGATGCCTGCCTTIA,3' and
`5'·GC">ACTTTOCIICC1ATGGGAG-3'} (22), 018556
`(22), AP~3 (9), BA T25 (5' -TCGCCTCCMGAA TGTA(cid:173)
`AGT-3' and 5' TCTGCATTTTAAGTATGGCTC-3')
`(12). BAT26 (5' -TGACTACTTTTGACTTCAGCC-3'
`and 5'·MCCATTCAACATlTTIAACCC·J'), and
`BAT40 (S' ·AlTAACTTCCTACACCACAAC-3' and
`5'·GTAGAGCAAGACCACCTTG-3'). D2S123 and
`D18S58 contained dinucleotide repeats, and the oth(cid:173)
`er markers contained mononucleotide repeats.
`12. R. Parsons eta/., unpublished data.
`
`13. T. A. Prolla. Q Pang. E. Alani, A. 0 . Kolodner. A.M.
`Uskay, Sclenc8 265. 1091 (1994); G.-M, U and P.
`Modrlch. Proc. Narl Acad. sc,, U.S.A. 92, 1950
`(1995).
`14. Formalin-fixed. pararnn-embadded eectlons were
`prepared and stained wllh hematoxolln and eosin
`The eplthellal and noneplthelial components were
`11eparated wltllltre aid of a dissecting microscope.
`Pools of 10 sectrons of ttle epithelial frachon (mu(cid:173)
`cosa) or the submuoOSill·musaularis propria frac(cid:173)
`Uon were Incubated al 56"C In protemase K (0.5
`mg/ml], I% SOS, 500 mM lns·HCI (pH 8.9). 20 mM
`EDTA, and I 0 rnM NaCI. Samples ware IIUln boiled
`for 10 m•n. axtrectod with phenoi-Cil!Orolorm. eth(cid:173)
`EillOI proclpna tcd, and resuspended in 50 ~~ of 3
`mM Ins-HOI (pH 7.5) and 0.2 JTIM EDT A. Samples
`lypiCally com~:uned 1000 to 1 0.000 genome equlv·
`alents per microliter. Sprlel10·1old dii\Jllon~ of DNA
`were used Ia dfltermlne the DNA concentration
`that would allow amplification of one to six alleles
`per reaction.
`15. The P6 and P7 clonal cells were cultured in 96-well
`plates Jn the presence or GM 18998 Ieeder cons (5 x
`1 o• per wa!O that hed been Irradiated prav,rusly wilh
`40 Gy (absorbed dose of ion1Z1nQ radiation) Clones
`were expanded In the absence of feeder cells before
`DNA WBS purified.
`16. A K. Rustg;, N. Engl. J. Me<l. 331 , 1694 (1994).
`1 7. Repair assays were perfor!T1ed wllh 50 '-'9 of nuclear
`extract and 24 !mol of mlspalrad DNA 1;11 37'C for 15
`min. The 3' CA substrate was a CA dinucleotide in(cid:173)
`sertion heteroduplex containing a nick 1 81 base pairs
`(bp) 3' to the mlsmatct1 (5). The 5' G-T mispair con(cid:173)
`taoned e G T mi!ll natch and a nick 125 bp 5' to the
`mispalr IS.-S. Su, R. S. Lahue. K. G. Au, P. MOdrich,
`J. Bioi. Cham. 263, 6829 (1988)). The repair or the
`substrate restored a restriction endonuclease site
`·me repair emrJency was measured 't:f.J dtgesllon of
`lho raactiorl prQducts wilh I he appropt~Ale ras1ncUon
`enzyme~ and resolving them on en agarose gel. None
`or the cterecuve extracts Inhibited a repair reactlotl
`containing a wild-type nuclear extract.
`18. T. R. Skopeck et al., Biochem. Biophys. Res. Com(cid:173)
`mun. 84, 41111978).
`19. A. G. Knudson, Cancer Res. 45, 1437 (t985); B.
`Vogelstekl and K W. Kim:lar, Trends Genet. 9, 138
`(1993).
`20. B. N, AmesandL S. Gold, Saierl!:e249, 970 (1990);
`S. M. Cohan and L. B. EllweJn, Ibid .. p. 1007; I. B.
`Weinstein, Cancer Res. 48, 4135 (1988).
`2t. A. H. Wyllie, Br. J. Cancer 67,205 (1993).
`22. J. WeJssenbach et al., Nature 359, 794 (1 992).
`23. S. R. Hamilton ef a/ , N. Engl. J. Med., 332, 839
`(1995).
`24. We thank A. Knudson and A. de Ia Chapelle for
`halpful discussions arlo T Gwilllda for p•eparatiOO
`of !he manuscript Suppor1ed by tile Clayton Fund.
`tmns!ntlonaJ funds from tha Duke Comprehensive
`Cancer Center, NIH grants GM45190, CA35494,
`CA62924, and CA09320, and American Cancer So(cid:173)
`ciety (ACS) grant PF·3940. B.V. is an ACS Research
`Professor.
`
`3 February 1995; accepted 17 March 1995
`
`740
`
`SCIENCE
`
`• VOL. 268
`
`• 5 MAY 1995
`
`Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
`
`The Johns Hopkins University Exhibit JHU2007 - Page 3 of 3
`
`