cer is often familial in origin, although the
`risks in relatives are not as high as those for
`early-onset breast cancer (10, 1 1). The per-
`centage of such cases that are due to genetic
`susceptibility is unknown.
`Like many other genes involved in fa-
`milial cancer, BRCA1 appears to encode a
`tumor suppressor, a protein that acts as a
`negative regulator of tumor growth. Can-
`cer-predisposing alleles typically carry mu-
`tations that cause loss or reduction of gene
`function. Predisposition to cancer is inher-
`ited as a dominant genetic trait, whereas
`the predisposing allele generally behaves as
`a recessive allele in somatic cells. Thus, a
`single inherited copy of the mutant allele
`causes predisposition, and loss or inactiva-
`tion of the wild-type allele completes one of
`the steps in progression toward malignancy.
`When chromosome loss
`is observed in
`breast and ovarian tumors from patients
`who carry BRCAI predisposing alleles, the
`wild-type copy of BRCA1 is invariably lost
`while the presumptive mutant allele is re-
`tained (12-14). This finding supports the
`hypothesis that BRCA1 is a tumor suppres-
`sor gene and suggests that the functional
`BRCA1 protein is present in normal breast
`and ovarian epithelium tissue and is altered,
`reduced, or absent in some breast and ovar-
`ian tumors.
`Genetic analysis of recombinant chromo-
`somes in members of large kindreds allowed
`localization of BRCA1 initially to a region of
`1 to 2 megabases on chromosome 17q (15-
`17) and, subsequently, to a region of about
`600 kilobase pairs (kb) (18) between markers
`D17S1321 and D17S1325 (19). A physical
`map comprised of overlapping yeast artificial
`chromosomes (YACs), P1, bacterial artificial
`chromosomes (BACs), and cosmid clones was
`generated for this region (18).
`Identification of a strong BRCAI can-
`didate gene. Several strategies were used to
`develop a detailed map of transcripts for the
`600-kb region of 17q21 between D17S1321
`and D17S1325. Sixty-five candidate ex-
`pressed sequences (20) within this region
`were identified. Expressed sequences were
`characterized by DNA sequence, database
`comparison, transcript size, expression pat-
`tern, genomic structure and, most impor-
`tantly, DNA sequence analysis in individu-
`als from kindreds that segregate 17q-linked
`breast and ovarian cancer susceptibility.
`Three expressed sequences eventually were
`merged into a single transcription unit
`whose characteristics strongly suggest that
`it is BRCA1 (21). This transcription unit is
`located in the center of the 600-kb region
`(Fig. 1) spanning D17S855 and will be re-
`ferred to herein as BRCA1.
`A combination of sequences obtained
`from complementary DNA (cDNA) clones,
`hybrid-selected sequences, and amplified
`polymerase chain reaction (PCR) products
`
`Downloaded from
`
` on October 1, 2007
`
`www.sciencemag.org
`
`RESEARCH ARTICLES
`
`A Strong Candidate for the
`Breast and Ovarian Cancer
`Susceptibility Gene BRCA1
`Yoshio Miki, Jeff Swensen, Donna Shattuck-Eidens, P. Andrew Futreal,
`Keith Harshman, Sean Tavtigian, Qingyun Liu, Charles Cochran,
`L. Michelle Bennett, Wei Ding, Russell Bell, Judith Rosenthal,
`Charles Hussey, Thanh Tran, Melody McClure, Cheryl Frye, Tom Hattier,
`Robert Phelps, Astrid Haugen-Strano, Harold Katcher, Kazuko Yakumo,
`Zahra Gholami, Daniel Shaffer, Steven Stone, Steven Bayer, Christian Wray,
`Robert Bogden, Priya Dayananth, John Ward, Patricia Tonin, Steven Narod,
`Pam K. Bristow, Frank H. Norris, Leah Helvering, Paul Morrison,
`Paul Rosteck, Mei Lai, J. Carl Barrett, Cathryn Lewis, Susan Neuhausen,
`Lisa Cannon-Albright, David Goldgar, Roger Wiseman, Alexander Kamb,
`Mark H. Skolnick*
`A strong candidate for the 17q-linked BRCA1 gene, which influences susceptibility to
`breast and ovarian cancer, has been identified by positional cloning methods. Probable
`predisposing mutations have been detected in five of eight kindreds presumed to seg-
`regate BRCA1 susceptibility alleles. The mutations include an 11-base pair deletion, a
`1-base pair insertion, a stop codon, a missense substitution, and an inferred regulatory
`mutation. The BRCA 1 gene is expressed in numerous tissues, including breast and ovary,
`and encodes a predicted protein of 1863 amino acids. This protein contains a zinc finger
`domain in its amino-terminal region, but is otherwise unrelated to previously described
`proteins. Identification of BRCA1 should facilitate early diagnosis of breast and ovarian
`cancer susceptibility in some individuals as well as a better understanding of breast
`cancer biology.
`
`Breast cancer is one of the most common
`and important diseases affecting women.
`Current estimates indicate that one in eight
`American women who reach age 95 will
`develop breast cancer (1). Treatment of
`advanced breast cancer is often futile and
`disfiguring, making early detection a high
`priority in medical management of the dis-
`ease. Ovarian cancer, although less frequent
`than breast cancer, is often rapidly fatal and
`is the fourth most common cause of cancer
`mortality in American women.
`Y. Miki, J. Swensen, K. Yakumo, C. Lewis, S. Neu-
`hausen, and D. Goldgar are in the Department of Medical
`Informatics, University of Utah Medical Center, Salt Lake
`City, UT 84132, USA. D. Shattuck-Eidens, K. Harshman,
`S. Tavtigian, Q. Liu, W. Ding, R. Bell, J. Rosenthal, C.
`Hussey, T. Tran, M. McClure, C. Frye, T. Hattier, R.
`Phelps, H. Katcher, Z. Gholami, D. Shaffer, S. Stone, S.
`Bayer, C. Wray, R. Bogden, P. Dayananth, and A. Kamb
`are at Myriad Genetics, 421 Wakara Way, Salt Lake City,
`UT 84108, USA. P. A. Futreal, C. Cochran, L. M. Bennett,
`A. Huagen-Strano, J. C. Barrett, and R. Wiseman are at
`the Laboratory of Molecular Carcinogenesis, National In-
`stitute of Environmental Health Sciences, National Insti-
`tutes of Health, Research Trangle Park, NC 27709, USA.
`J. Ward and L. Cannon-Albnght are in the Department of
`Internal Medicine, University of Utah Medical Center, Salt
`Lake City, UT 84132, USA. P. Tonin and S. Narod are in
`the Department of Medical Genetics, McGill University,
`Montreal, Quebec, H3G 1A4, Canada. P. K. Bristow, F.
`H. Norris, L. Helvering, P. Morrison, P. Rosteck, and M.
`Lai are at Lilly Research Laboratories, Eli Lilly and Com-
`pany, Indianapolis, IN 46285, USA. M. H. Skolnick is in
`the Department of Medical Informatics, University of Utah
`Medical Center, and Myriad Genetics, Salt Lake City, UT
`84108, USA.
`*To whom correspondence should be addressed.
`
`66
`
`Genetic factors contribute to an ill-de-
`fined proportion of breast cancer incidence,
`estimated to be about 5% of all cases but
`approximately 25% of cases diagnosed be-
`fore age 30 (2). Breast cancer has been
`subdivided into two types, early-onset and
`late-onset, a division that is based on an
`in the age-specific incidence
`inflection
`curve around age 50. Mutation of one gene,
`BRCA1, is thought to account for approxi-
`mately 45% of families with significantly
`high breast cancer incidence and at least
`80% of families with increased incidence of
`both early-onset breast cancer and ovarian
`cancer (3). Intense efforts to isolate the
`BRCA1 gene have proceeded since it was
`first mapped to chromosome arm 17q in
`1990 (4, 5). A second locus, BRCA2, re-
`cently mapped to chromosome arm 13q (6),
`appears to account for a proportion of early-
`onset breast cancer roughly equal to that
`resulting from BRCA1. Unlike BRCA1,
`however, BRCA2 may not influence ovari-
`an cancer risk. The remaining susceptibility
`to early-onset breast cancer is likely attrib-
`utable to unmapped genes for familial can-
`cer and rare germline mutations in genes
`such as TP53, which encodes the tumor
`suppressor protein p53 (7). It has also been
`suggested that heterozygote carriers of de-
`fective forms of the gene predisposing to
`ataxia telangiectasia are at higher risk for
`breast cancer (8, 9). Late-onset breast can-
`
`SCIENCE * VOL. 266
`
`*
`
`7 OCTOBER 1994
`
`GeneDX 1005, pg. 1
`
`

`

`ing is coordinated with alternative splicing
`farther downstream, and whether all the
`splice variants produce proteins with an
`identical NH2-terminus, are questions that
`remain to be explored.
`We also probed genomic DNA samples
`from several different species with BRCA1
`sequences devoid of the zinc finger region.
`Low-stringency blots revealed strongly hy-
`bridizing fragments in tissues from humans,
`mice, rats, rabbits, sheep, and pigs, but not
`chickens (Fig. 5). These results suggest that
`
`BRCA1 is conserved in mammals.
`Germline BRCA1 mutations in 17q-
`linked kindreds. Identification of a candi-
`date gene as BRCAI requires a demonstra-
`tion of potentially disruptive mutations in
`that gene in carrier individuals from kin-
`dreds that segregate 1 7q-linked susceptibil-
`ity to breast and ovarian cancer. Such in-
`dividuals must contain BRCA1 alleles that
`differ from the wild-type sequence. The set
`of DNA samples used in this analysis con-
`sisted of DNA from individuals represent-
`
` on October 1, 2007
`
`www.sciencemag.org
`
`Downloaded from
`
`A
`
`v
`* v
`0 MDLSALRVEEVQNVINAMQKILECPICLELIKEPVSTKCDHIFCKFCMLKLLNQKKGPSQ
`v
`v
`60 CPLCKNDITKRSLQESTRFSQLVEELLKIICAFQLDTGLEYANSYNFAKKENNSPEHLKD
`v
`120 EVSIIQSMGYRNRAKRLLQSEPENPSLQETSLSVQLSNLGTVRTLRTKQRIQPQKTSVYI
`
`180 ELGSDSSEDTVNKATYCSVGDQELLQITPQGTRDEISLDSAKKAACEFSETDVTNTEHHQ
`240 PSNNDLNTTEKRAAERHPEKYQGSSVSNLHVEPCGTNTHASSLQHENSSLLLTKDRMNVE
`300 KAEFCNKSKQPGLARSQHNRWAGSKETCNDRRTPSTEKKVDLNADPLCERKEWNKQKLPC
`360 SENPRDTEDVPWITLNSSIQKVNEWFSRSDELLGSDDSHDGESESNAKVADVLDVLNEVD
`420 EYSGSSEKIDLLASDPHEALICKSDRVHSKSVESNIEDKIFGKTYRKKASLPNLSHVTEN
`480 LIIGAFVSEPQIIQERPLTNKLKRKRRPTSGLHPEDFIKKADLAVQKTPEMINQGTNQTE
`540 QNGQVMNITNSGHENKTKGDSIQNEKNPNPIESLEKESAFKTKAEPISSSISNELELNIM
`600 HNSKAPKKNRLRRKSSTRHIHALELVVSRNLSPPNCTELQIDSCSSSEEIKKKKYNQMPV
`660 RHSRNLQLMEGKEPATGAKKSNKPNEQTSKRHDSDTFPELKLTNAPGSFTKCSNTSELKE
`720 FVNPSLPREEKEEKLETVKVSNNAEDPKDLMLSGERVLQTERSVESSSISLVPGTDYGTQ
`780 ESISLLEVSTLGKAKTEPNKCVSQCAAFENPKGLIHGCSKDNRNDTEGFKYPLGHEVNHS
`840 RETSIEMEESELDAQYLQNTFKVSKRQSFAPFSNPGNAEEECATFSAHSGSLKKQSPKVT
`900 FECEQKEENQGKNESNIKPVQTVNITAGFPVVGQKDKPVDNAKCSIKGGSRFCLSSQFRG
`960 NETGLITPNKHGLLQNPYRIPPLFPIKSFVKTKCKKNLLEENFEEHSMSPEREMGNENIP
`1020 STVSTISRNNIRENVFKEASSSNINEVGSSTNEVGSSINEIGSSDENIQAELGRNRGPKL
`1080 NAMLRLGVLQPEVYKQSLPGSNCKHPEIKKQEYEEVVQTVNTDFSPYLISDNLEQPMGSS
`1140 HASQVCSETPDDLLDDGEIKEDTSFAENDIKESSAVFSKSVQKGELSRSPSPFTHTHLAQ
`1200 GYRRGAKKLESSEENLSSEDEELPCFQHLLFGKVNNIPSQSTRHSTVATECLSKNTEENL
`*
`1260 LSLKNSLNDCSNQVILAKASQEHHLSEETKCSASLFSSQCSELEDLTANTNTQDPFLIGS
`
`1320 SKQMRHQSESQGVGLSDKELVSDDEERGTGLEENNQEEQSMDSNLGEAASGCESETSVSE
`
`1380 DCSGLSSQSDILTTQQRDTMQHNLIKLQQEMAELEAVLEQHGSQPSNSYPSIISDSSALE
`
`1440 DLRNPEQSTSEKVLQTSQKSSEYPISQNPEGXSADKFEVSADSSTSKNKEPGVERSSPSK
`
`1500 CPSLDDRWYMHSCSGSLQNRNYPPQEELIKVVDVEEQQLEESGPHDLTETSYLPRQDLEG
`1560 TPYLESGISLFSDDPESDPSEDRAPESARVGNIPSSTSALKVPQLKVAESAQSPAAAHTT
`v
`1620 DTAGYNAMEESVSREKPELTASTERVNKRMSMVVSGLTPEEFMLVYKFARKHHITLTNLI
`
`1680 TEETTHVVMKTDAEFVCERTLKYFLGIAGGKWVVSYFWVTQSIKERKMLNEHDFEVRGDV
`R V
`+c
`v
`1740 VNGRNHQGPKRARESQDRKIFRGLEICCYGPFTNMPTDQLEWMVQLCGASVVKELSSFTL
`v
`v
`1800 GTGVHPIVVVQPDAWTEDNGFHAIGQMCEAPVVTREWVLDSVALYQCQELDTYLIPQIPH
`1860 SHY
`
`B
`CPICLELIKEPVSTK-CDHIFCKFCMLKLLNQKK--- GPSQCPLCK
`BRCA1
`CPICLELLKEPVSAD-CNHSFCRACITLNYESNRNTDGKGNCPVCR
`RPT1
`CPICLDMLKNTMTTKECLHRFCSDCIVTALRS----GNKECPTCR
`RIN1
`CPVCLQYFAEPMMLD-CGHNICCACLARCWGTAC---TNVSCPQCR
`RFP1
`C- ------------ C-H--C----------------- C--C
`C3HC4 motif
`Fig. 2. Predicted amino acid sequences for BRCA1 (46). (A) Conceptual translation of the BRCA1 open
`reading frame, indicating the approximate positions of introns (triangles above sequence) and the
`locations of germline mutations (boldface residues). The -1 1 bp deletion in kindred 1901 is shown by an
`asterisk; the nonsense mutation in kindred 2082 is shown by a star; the frameshift in kindred 1910 is
`shown by "+c"; and the missense mutation in kindred 2099 is shown by "R". The BRCA1 nucleotide
`sequence is deposited in GenBank with accession number Ul 4680. PCR primer sequences are available
`via anonymous FTP at the following internet address: morgan.med.utah.edu in the directory pub/BRCA1;
`or by fax at the following number: 801-584-3650. (B) Alignment of the BRCA1 zinc finger domain with
`three other zinc finger domains that scored highest in a Smith-Waterman alignment. RPT1 is a protein
`that appears to be a negative regulator of the interleukin-2 receptor in mice (47). RIN1 is a DNA binding
`protein that includes a RING finger motif related to the zinc finger (48). RFP1 is a putative transcription
`factor comprising the NH2-terminal domain of the REToncogene product (49). The C3HC4 motif shows
`the positions of the cystines and the histidine that form the zinc binding pockets.
`7 OCTOBER 1994
`SCIENCE * VOL. 266
`
`67
`
`*
`
`a
`
`allowed the construction of a composite,
`full-length BRCAJ cDNA. The cDNA
`clone extending farthest in the 3' direction
`contains a polyadenylate tract preceded by
`a polyadenylation signal. Conceptual trans-
`lation of the cDNA revealed a single, long
`open reading frame with a presumptive ini-
`tiation codon flanked by sequences resem-
`bling the Kozak consensus sequence (22).
`This reading frame encodes a protein of
`1863 amino acids (Fig. 2A). Smith-Water-
`man (23) and BLAST (24) searches iden-
`tified a sequence near the NH2-terminus
`that has considerable similarity to zinc fin-
`ger domains (25) (Fig. 2B). This sequence
`cystine and histidine
`residues
`contains
`present in the consensus Cys3-His-Cys4
`(C3HC4) zinc finger motif and shares many
`other residues with zinc finger proteins in
`the databases. The BRCA1 gene is com-
`posed of 22 coding exons distributed over
`roughly 100 kb of genomic DNA (Fig. 3).
`Hybridization of RNA blots to labeled
`fragments of BRCA1 cDNA revealed a sin-
`gle transcript of 7.8 kb. This transcript is
`most abundant in testis and thymus, but is
`also present in breast and ovary (Fig. 4).
`The cDNA clones derived from the 5' one-
`third of BRCAJ transcripts display a com-
`plex pattern of alternative splicing. Four
`alternative splices were observed down-
`stream of the start codon as independent
`cDNA clones (P3, P4, B31, and B21 in Fig.
`3); three of these patterns were detected in
`breast cDNA (P3, B31, and B21) and two
`in ovary cDNA (P3 and B21). In addition,
`PCR analysis of cDNA samples prepared
`from breast, ovary, testis, and lymphocyte
`messenger RNA (mRNA) indicates that
`there is considerable heterogeneity in splice
`junction usage near the 5' end of BRCAl
`transcripts, upstream of the presumptive
`initiation codon. How this alternative splic-
`
`/JDD17S250
`
`/
`
`fD17S800
`J ,-U177S73b7
`,/ BRCA1 t-D17S855
`
`4
`
`-D17S1327
`-CA 125
`-RNU2
`
`-D17S1325
`
`D17S579
`
`/
`
`I I
`
`Fig. 1. Schematic map of human chromosome
`17. The pertinent region containing BRCA 1 is ex-
`panded to indicate the relative positions of two
`previously identified genes, CA 125 (34) and
`RNU2 (45). D1 7S855 is located within BRCA1.
`
`GeneDX 1005, pg. 2
`
`

`

`MM.
`
`a
`
`6A). Kindred 1910 contains a single nucle-
`otide insertion in coding exon 20 (Gln1756
`frameshift to 1829 Stop) (Fig. 6B), and
`kindred 2099 contains a missense mutation
`in coding exon 21 (Metl775Arg). The
`frameshift and nonsense mutations are like-
`ly to disrupt the function of the BRCA1
`proteins. The protein encoded by the inser-
`tion allele in kindred 1910 would contain
`an altered sequence beginning 107 amino
`acids residues from the wild-type COOH-
`terminus. The effect of the 1 1-bp deletion
`in kindred 1901 would be even more dra-
`matic because it occurs at the twenty-fourth
`codon. This deletion removes the last 11 bp
`of exon 2 and begins at the first cystine of
`the zinc finger motif, thereby removing the
`zinc finger domain. The mutant allele in
`kindred 2082 would encode a protein miss-
`ing 548 residues from the COOH-terminus.
`
`The missense mutation observed in kindred
`2099 is potentially disruptive as it substi-
`tutes a large, charged amino acid (Arg) for
`a small, hydrophobic amino acid (Met).
`Five common polymorphisms were also
`identified in the BRCA1 coding sequence
`(Table 3).
`The individual studied in kindred 2035
`is likely to carry a regulatory mutation in
`BRCAl. In her cDNA, two polymorphic
`sites (PM1 and PM7) appeared homozy-
`gous, whereas her genomic DNA revealed
`heterozygosity at these positions (Fig. 6C).
`One possible explanation for this observa-
`that mRNA from her mutant
`tion
`is
`BRCA1 allele is absent because of a muta-
`tion that affects RNA production or stabil-
`ity. We explored this possibility further by
`examining three additional polymorphic
`sites (PM6, PM7, and PM2) in the BRCA1
`
`Table 1. Kindred descriptions and associated lod scores (50). Br, breast cancer; Br <50, breast cancer
`diagnosed under age 50; Ov, ovarian cancer.
`
` on October 1, 2007
`
`ing eight different BRCA1 kindreds (Table
`1). The lod scores (likelihood ratios for
`linkage) in these kindreds range from 9.49
`to -0.44 for a set of markers in 17q21. Four
`of the families have convincing lod scores
`for linkage, and four have low positive or
`negative lod scores. The latter kindreds
`were included because they demonstrate
`haplotype sharing at chromosome 17q21 for
`at least three affected members. Further-
`more, all kindreds in the set display early-
`onset breast cancer, and four of the kindreds
`include at least one case of ovarian cancer,
`both hallmarks of BRCAI kindreds. Kin-
`dred 2082 has nearly equal incidence of
`breast and ovarian cancer, an unusual oc-
`currence given the relative rarity of ovarian
`cancer in the population (17). All but two
`of the kindreds were ascertained in Utah.
`Kindred 2035 is from the midwestern Unit-
`ed States. Kindred 2099 is an African
`American kindred from the southern Unit-
`ed States; all other kindreds are Caucasian.
`In the initial screen for predisposing mu-
`tations in BRCA1, DNA from one individ-
`ual carrying the predisposing haplotype
`from each kindred was tested. The 21 cod-
`ing exons and associated splice junctions
`were amplified from either genomic DNA
`samples or cDNA prepared from lympho-
`cyte mRNA (26). When the amplified
`DNA sequences were compared to the wild-
`type sequence, four of the eight kindred
`samples were found to contain sequence
`variants (Table 2). All four sequence vari-
`ants are heterozygous, and each appears in
`only one of the kindreds. Kindred 1901
`contains an 11-base pair (bp) deletion in
`exon 2 (Cys24 frameshift to 36 Stop). Kin-
`dred 2082 contains a nonsense mutation in
`coding exon 11 (Gln1313 to Stop) (Fig.
`
`Kindred
`
`Toetalod
`Br
`
`Cases (n)
`
`Total
`
`Br
`<50
`
`Ov
`O
`
`Sporadic
`
`(n)
`n
`
`L
`~~~~~score
`
`M
`Markers
`
`2082
`2099
`
`2035
`1901
`1925
`1910
`
`31
`22
`
`10
`10
`4
`5
`
`20
`14
`
`8
`7
`3
`4
`
`22
`2*
`
`1 *
`1*
`0
`0
`
`7
`0
`
`0
`0
`0
`0
`
`9.49
`2.36
`
`2.25
`1.50
`0.55
`0.36
`
`D17S1327
`D1 7S800 and
`D1 7S855t
`D1 7S1327
`D17S855
`D1 7S579
`D17S579 and
`D1 7S250t
`D1 7S250
`-0.20
`5
`8
`1911
`0
`1
`1927
`0
`D1 7S250
`-0.44
`5
`1
`4
`*Kindred contains one individual who had both breast and ovarian cancer; this individual is counted as both a breast
`cancer case and as an ovarian cancer case.
`tNumber of women with breast cancer (diagnosed under age 50) or
`tBoth markers were used to
`ovarian cancer (diagnosed at any age) who do not share the BRCA 1 -linked haplotype.
`calculate multipoint lod scores.
`
`www.sciencemag.org
`
`Downloaded from
`
`v v v
`
`BRCA1
`5TO
`P3'A.
`p4,
`Breast CDNAi
`Ovary cDNAi
`TY31
`T
`TE
`B21
`B31
`TY4
`F103
`61
`
`M
`
`F191
`F31
`B9
`
`I
`
`Fig. 3. Diagram of BRCA 1
`mRNA, showing the loca-
`tions of introns and the vari-
`ants of BRCA1 mRNA pro-
`duced by alternative splic-
`ing. The top cDNA (BRCA1)
`is the composite used to
`protein
`generate the
`se-
`quence in Fig. 2. Intron loca-
`tions are shown by filled tri-
`angles, and the exons are
`numbered below the com-
`cDNA.
`posite
`Alternative
`mRNAs identified as cDNA
`clones or in hybrid-selection
`experiments are shown be-
`low the composite. The start
`codon (ATG) and stop codon
`(TGA) are indicated. The zinc
`finger region is denoted noted
`by a double line. "V' lines connecting exons indicate the absence of internal
`exons. All exons are drawn proportionally except exon 11 (indicated with a
`dotted line). Upward-pointing unfilled triangles show the position of a single
`codon (CAG) found at the start of exons 8 and 14 in some cDNAs. Leftward-
`and rightward-pointing unfilled triangles represent partial exons in some
`cDNAs. P3 and P4 are cDNA clones isolated from a placental cDNA library;
`
`W
`
`W
`
`v V-AA v v v v
`_-_
`W_
`18 191 20 21i 22123i!t
`17
`TGA
`
`_
`
`24
`
`3
`
`v
`
`'a
`
`W
`;
`13 t 14
`
`1
`
`15
`
`v
`
`16
`
`W
`W
`392x3F
`-F-1Z T
`ii
`
`i
`
`1
`
`at
`
`TY3 and TE2 are 5' RACE clones from thymus and testis, respectively; B21
`and B9 are cDNA clones from a normal breast cDNA library; B31 is a hybrid-
`selected cDNA clone from breast cDNA; TY4 and TY6 are cDNA clones
`isolated from a thymus cDNA library; and F191, F103, and F3 are cDNA
`clones isolated from a fetal brain library. The BRCA 1 variants labeled breast
`cDNA and ovary cDNA are the major forms detected in these tissues by PCR.
`
`68
`
`SCIENCE * VOL. 266
`
`*
`
`7 OCTOBER 1994
`
`GeneDX 1005, pg. 3
`
`

`

`I
`
`AS
`
`were based on 39 samples used in linkage
`studies and on samples from 20 African
`Americans from Utah (28). None of the
`four potential predisposing mutations tested
`was found in the appropriate control popu-
`lation, indicating that they are rare in the
`general population (Table 2). Thus, both
`important requirements for BRCA1 suscep-
`tibility alleles are fulfilled by the candidate
`predisposing mutations: cosegregation of
`the mutant allele with diseases and ab-
`sence of the mutant allele in controls,
`indicating a low frequency in the general
`population (29).
`Phenotypic expression of BRCA1 mu-
`tations. The effect of the mutations on the
`BRCA1 protein correlates with differences
`in the observed phenotypic expression in
`the BRCA1 kindreds. Most BRCA1 kin-
`dreds have a moderately increased ovarian
`cancer risk, and a smaller subset have a high
`risk of ovarian cancer comparable to that
`for breast cancer (3). Four of the five kin-
`dreds in which BRCA1 mutations were de-
`tected fall into the former category, and the
`fifth (kindred 2082) falls into the group
`with high ovarian cancer risk. The BRCA1
`nonsense mutation found in kindred 2082
`has an interesting phenotype. Kindred 2082
`has a high incidence of ovarian cancer, and
`the mean age of breast cancer diagnosis is
`older than that in the other kindreds ( 17).
`This difference in age of onset could be due
`to an ascertainment bias in the smaller,
`more highly penetrant families, or it could
`reflect tissue-specific differences in the be-
`havior of BRCA1 mutations. The other four
`kindreds that segregate known BRCA1 mu-
`tations have, on average, 1 ovarian cancer
`for every 10 cases of breast cancer, but have
`a high proportion of breast cancer cases
`diagnosed at an early age (late 20s or early
`30s). Kindred 1910, which has a 1-bp inser-
`tion mutation, is noteworthy because three
`of the four affected individuals had bilateral
`breast cancer, and in each case the second
`tumor was diagnosed within a year of the
`first occurrence. Kindred 2035, which seg-
`regates the potential regulatory BRCA1
`mutation, might also be expected to have a
`
` on October 1, 2007
`
`www.sciencemag.org
`
`Downloaded from
`
`including both carriers and noncarriers of
`the predisposing haplotype (Fig. 6). In each
`kindred, the corresponding mutant allele
`was detected only in individuals carrying
`the BRCAl-associated haplotype. In the
`case of the potential regulatory mutation in
`kindred 2035, cDNA and genomic DNA
`from carriers in the kindred were compared
`for heterozygosity at polymorphic sites. In
`every instance, the extinguished allele in
`the cDNA sample was shown to lie on the
`chromosome that carries the BRCA1 pre-
`disposing allele.
`To exclude the possibility that the mu-
`tations were simply common polymor-
`phisms in the population, we used allele-
`specific oligonucleotides (ASOs) for each
`mutation to screen a set of control DNA
`samples (27). The actual mutation in kin-
`dred 2035 has not been identified, so we
`could not determine its frequency in the
`general population. Gene frequency esti-
`mates in Caucasians were based on random
`samples from the Utah population. Gene
`frequency estimates in African Americans
`
`C0o X C geJ9
`
`coding region, which are separated by as
`much as 3.5 kb in the BRCA1 transcript. In
`all cases where her genomic DNA appeared
`heterozygous
`for
`a
`polymorphism,
`her
`cDNA appeared homozygous. In individuals
`from other kindreds and in nonhaplotype
`carriers in kindred 2035, these polymorphic
`sites appeared heterozygous in cDNA, im-
`plying that amplification from cDNA was
`not biased in favor of one allele. This anal-
`ysis indicates that a BRCA1 mutation in
`kindred 2035 either prevents transcription
`or causes instability or aberrant splicing of
`the BRCA1 transcript.
`Cosegregation of BRCAI mutations
`with BRCA1 haplotypes and population
`frequency analysis. In addition to potential
`disruption of protein function, a sequence
`variant must meet two other criteria to
`qualify as a candidate predisposing muta-
`tion: It must be present in members of the
`kindred who carry the predisposing haplo-
`type and absent from other members of the
`kindred, and it must be rare in the general
`population.
`To test for cosegregation of mutations
`with the corresponding BRCA1 susceptibil-
`ity allele, we screened several individuals
`from kindreds 1901, 1910, 2082, and 2099,
`
`ZD
`
`0
`@
`
`C
`
`CO
`
`U)_
`
`-9.50
`-7.50
`
`-4.40
`
`CO
`
`9.S0-
`7.50-
`
`-2.40
`
`4.40-
`Fig. 4. Tissue expression pattern of BRCA 1. The
`blots were obtained from Clontech (Palo Alto, CA)
`and contain RNA from the indicated tissues. Hy-
`bridization conditions were those recommended
`by the manufacturer, and the probe was a BRCA 1
`cDNA fragment corresponding to nucleotides
`3575 to 3874. Note that these tissues are heter-
`ogeneous and the percentage of relevant epithe-
`lial cells in breast and ovary can be variable. Size
`standards are in kilobases.
`
`Fig. 5. Blot showing hybridization of a BRCA1
`probe to genomic DNA fragments from various
`species. DNA was digested with Eco RI, subject-
`ed to electrophoresis through a 0.65% agarose
`gel, and transferred to a nylon membrane, which
`was then hybridized (32) to a probe consisting of
`random-primed, a-32P-labeled BRCA 1 cDNA se-
`quences comprising a total of 4.6 kb. The probe
`excluded the zinc finger region. The final wash
`was at 550C in x2 SSPE and 1 % SDS for 20 min.
`Size standards are in kilobases.
`
`Table 2. Predisposing mutations in BRCA1. NA indicates not applicable, as the regulatory mutation is
`inferred and the position has not been identified.
`
`Table 3. Neutral polymorphisms in BRCA 1. For
`the frequency in control chromosomes, the
`number of chromosomes with a particular base
`at the indicated polymorphic site is shown (A, C,
`G, orT).
`
`Kindred
`
`Codon
`
`1901
`2082
`1910
`2099
`2035
`
`24
`1313
`1756
`1775
`NA
`
`Nucleotide
`change
`-11 bp
`C->T
`Extra C
`T->G
`?
`
`Mutation
`
`Coding
`effect
`Frameshift or splice
`Gln->Stop
`Frameshift
`Met->Arg
`Loss of transcript
`
`Frequency
`in control
`chromosomes
`
`0/180
`0/170
`0/162
`0/120
`NA
`
`Name
`
`PM1
`PM6
`PM7
`PM2
`PM3
`
`SCIENCE * VOL. 266
`
`*
`
`7 OCTOBER 1994
`
`Codon
`loca-
`tion
`
`Base
`in
`codon
`
`Frequency in control
`chromosomes
`C
`G
`
`A
`
`T
`
`317
`878
`1190
`1443
`1619
`
`2
`2
`2
`3
`1
`
`152
`0
`109
`0
`116
`
`0
`55
`0
`115
`0
`
`10
`0
`53
`0
`52
`
`0
`100
`0
`58
`0
`
`GeneDX 1005, pg. 4
`
`

`

`dramatic phenotype.
`Eighty percent of
`breast cancer cases in this kindred occur
`under age 50. This figure is as high as any in
`the set, suggesting that this BRCAJ mutant
`allele has a high penetrance (Table 1). Kin-
`dred 1901 displays a phenotypic pattern
`similar to that of kindred 2035. It is likely
`that the 1 1-bp deletion beginning at codon
`24 carried in kindred 1901 results in a loss
`of gene function similar to the effect of the
`regulatory mutation in kindred 2035.
`Although the mutations described in
`this research article are clearly deleterious,
`causing breast cancer in women at very
`young ages, each of the four kindreds with
`mutations includes at least one woman who
`carried the mutation but lived until age 80
`without developing a malignancy. It will be
`of utmost importance in future studies to
`identify other genetic factors or environ-
`mental factors that may ameliorate the ef-
`fects of BRCA1 mutations. In addition, in
`three of the eight putative BRCAl-linked
`kindreds, potential predisposing mutations
`were not found. All of these kindreds have
`lod scores for BRCAJ -linked markers of less
`than 0.55 and thus may not truly segregate
`BRCAl-predisposing alleles. Alternatively,
`the mutations in these three kindreds may
`lie in noncoding regions of BRCA) and
`therefore have escaped detection.
`The role of BRCA1 in cancer. Most
`
`m
`
`mutant tumor suppressor genes identified to
`date encode proteins that are absent, non-
`functional, or reduced in function. The ma-
`jority of TP53 mutations are missense; some
`of these have been shown to produce ab-
`normal p53 molecules that interfere with
`the function of the wild-type product (30,
`31). A similar dominant-negative mecha-
`nism of action has been proposed for some
`adenomatous polyposis coli (APC) alleles
`that produce truncated molecules (32) and
`for point mutations in the Wilms tumor
`gene (WTJ), which alter DNA binding of
`the WT1 protein (33). The nature of three
`mutations observed in the BRCAI coding
`sequence is consistent with production of
`either dominant-negative proteins or non-
`functional proteins. All three mutations are
`located in the COOH-terminal half of the
`protein. The regulatory mutation inferred
`in kindred 2035 cannot be dominant-nega-
`tive; rather, this mutation likely causes re-
`duction or complete loss of BRCA1 expres-
`sion from the affected allele. Similarly, the
`11-bp deletion in kindred 1901 likely pro-
`duces a nonfunctional product.
`The BRCAI protein contains a C3HC4
`zinc finger domain similar to domains found
`in numerous nucleic acid binding proteins.
`The first 180 amino acids of BRCA1 con-
`tain five more basic residues than acidic
`residues. In contrast, the remainder of the
`
`P
`
`a
`
`b
`
`c
`
`d
`
`e
`
`f
`
`g
`
`B
`
`C
`
`14
`
`A
`1353
`1078-
`872-
`603-
`
`310-
`
`molecule is very acidic, with a net excess of
`70 acidic residues. The excess negative
`charge is particularly concentrated near the
`COOH-terminus. Thus, one possibility is
`that BRCA1 encodes a transcription factor
`with an NH2-terminal DNA binding do-
`main and a COOH-terminal "acidic blob"
`domain with transactivational activity. In-
`terestingly, the product of another familial
`tumor suppressor gene, WTl, also contains
`zinc finger domains (34), and these are
`altered by many cancer-predisposing muta-
`tions in the gene (33-35). The WT1 gene
`encodes a transcription factor, and alterna-
`tive splicing of exons that encode parts of
`the zinc finger domains alters the DNA
`binding properties of WT1 (36). Some al-
`ternatively spliced forms of WT1 mRNA
`generate WT1 proteins that act as tran-
`Differential
`scriptional
`repressors
`(37).
`splicing of BRCA1 may alter the zinc finger
`motif (Fig. 3), raising the possibility that a
`regulatory mechanism similar to that occur-
`ring in WT1 may apply to BRCA1.
`The identification of a gene that (i) falls
`within the interval known from genetic
`studies to include BRCA1 and (ii) contains
`frameshift, nonsense, and regulatory muta-
`tions that cosegregate with predisposing
`BRCA1 alleles strongly indicates that this
`gene is BRCA1. The observation of poten-
`tial predisposing mutations in individuals
`
` on October 1, 2007
`
`www.sciencemag.org
`
`Downloaded from
`
`G A G
`
`A G A G
`
`PM1
`
`r;
`
`r_
`
`7
`
`p
`
`a
`
`1
`
`Fig. 6. Mutation and cosegregation
`analysis in BRCA 1 kindreds. Carrier in-
`dividuals are represented as filled cir-
`cles and squares in the pedigree dia-
`grams. (A) Nonsense mutation in kin-
`dred 2082. P indicates the person orig-
`inally screened; b and c are haplotype
`carriers; a, d, e, f, and g do not carry the
`BRCA 1 haplotype. The C to T mutation
`results in a stop codon and creates a
`site for the restriction enzyme Avr II.
`PCR amplification products were cut
`with this enzyme, subjected to electro-
`phoresis through 1.0% agarose gels, and stained with ethidium bromide. The
`carriers are heterozygous for the site and, therefore, show three bands. The
`PCR products of noncarriers remain uncut by Avr II and, therefore, show one
`band. Size standards are in base pairs. (B) Frameshift mutation in kindred
`1910. Sequencing reactions were loaded side by side as A, C, G, and
`T reactions. The first three lanes for each nucleotide set contain sequence
`ladders from noncarriers. Lanes 1 to 3 contain A ladders from carrier
`individuals. Lane 4 contains the A ladder from a kindred member who does
`not carry the BRCA 1 mutation. The frameshift resulting from the cytosine
`insertion is apparent in lanes 1 through 3. (The diamond shape in the
`pedigree diagram is used to protect the confidentiality of the transmitting
`parent.) (C) Inferred regulatory mutation in kindred 2035. ASO analysis of
`haplotype carriers and noncarriers used two different polymorphisms,
`SCIENCE * VOL. 266
`
`70
`
`l-A -C
`GIIT
`
`PM1 and PM7 (Table 3). Samples were examined for heterozygosity in the
`germ line an