REVIEWS

BRCA1 AND BRCA2: 1994 AND BEYOND

Steven A. Narod* and William D. Foulkes ‡
Abstract | The discovery of the first gene associated with hereditary breast cancer, BRCA1, was
anticipated to greatly increase our understanding of both hereditary and sporadic forms of breast
cancer, and to lead to therapeutic and preventive breakthroughs. Much has been learned during
the past decade about the genetic epidemiology of breast cancer, the ethnic distribution and
clinical consequences of BRCA1 and BRCA2 mutations, and the central role of DNA repair in
breast cancer susceptibility. The ability to translate this knowledge into novel treatments,
however, remains elusive.

A decade has passed since Mark Skolnick and his col- such as insurance and occupational discrimination, would
leagues at Myriad Genetics in Salt Lake City (Utah, USA) be discussed. Many researchers therefore concluded that
announced in 1994 that they had cloned the BRCA1 genetic testing should only be carried out in research set-
gene1. BRCA1 had been named three years earlier by tings. In an editorial in the New England Journal of
Mary-Claire King when she and her group assigned it to Medicine in January 1996, Francis Collins, Director of
chromosome 17 by linkage analysis using a large group of the National Center for Human Genome Research at the
families with cases of early-onset breast cancer2; however, United States National Institutes of Health said “the uncer-
the identification of truncating mutations in the coding tain risks and benefits lead most observers to believe that
sequence of BRCA1 in families with multiple cases of testing, whether in Jewish or non-Jewish women, should
breast cancer was the conclusive step1. Predictions were now be done only in a research setting, with a protocol
soon made about new biological insights and potential approved by an institutional review board and full
therapies. Families with a high incidence of male breast informed consent”6.
cancer, however, were found not to carry BRCA1 muta- In retrospect, few of these initial fears have been
tions3, leading to the search for other breast cancer genes. realized. Caryn Lerman studied the psychological
BRCA2 was linked to chromosome 13 in 1994 (REF. 4) and consequences of genetic testing soon after it was
was cloned only a year later by the same group5 (BOX 1). introduced and reported that it did not lead to undue
Genetic testing for cancer susceptibility quickly increases in anxiety or depression7. There is now com-
followed. Testing was initially viewed with a mixture of pelling evidence that both preventive mastectomy and
exuberance and caution. Fears were expressed by social preventive oophorectomy can markedly reduce cancer
*Centre for Research on scientists and feminists, and breast cancer advocates risk 8,9. In the past decade, women have been increas-
Women’s Health,
Sunnybrook and Womens expressed concern about the possible negative conse- ingly accepting of preventive surgery, although
College Health Sciences quences of testing and its rapid commercialization. They chemoprevention has remained unpopular. Kelly
Center, 790 Bay Street, felt that scientists might be paying undue attention to Metcalfe surveyed women who had recently received a
Toronto, Ontario M5G 1N8, inherited causes of cancer that were beyond the control of positive result after genetic testing in North America
Canada.
‡ individual women and were therefore not amenable to and found that 60% underwent preventive oophorec-
Program in Cancer
Genetics, Departments of public policy and legislation, as opposed to the elimina- tomy and 25% opted for prophylactic mastectomy,
Oncology and Human tion of environmental toxins and the promotion of a whereas only 12% had taken tamoxifen. At present,
Genetics, McGill University, healthy lifestyle. Psychologists recommended that women genetic testing is offered in many centres in North
Montreal, Quebec H3G 1A4, should complete a psychological evaluation before America, Europe, Australia and Israel. Several
Canada.
Correspondence to S.A.N. undergoing genetic testing. Genetic counsellors advised mutation surveys have been conducted in Asian coun-
e-mail: [email protected] that women should attend several comprehensive pre-test tries10, but genetic testing for cancer is still mainly a
doi:10.1038/nrc1431 counselling sessions, where a host of possible concerns, feature only of western medicine. Whether or not

NATURE REVIEWS | C ANCER VOLUME 4 | SEPTEMBER 2004 | 6 6 5

REVIEWS

Summary The Breast Cancer Information Core (BIC; see online

links box) was established in 1995 to catalogue the range
• In the ten years since the discovery of BRCA1 and BRCA2, genetic testing for breast and frequency of germline BRCA1 and BRCA2 muta-
and ovarian cancer susceptibility has become integrated into the practice of clinical tions. Because of the preponderance of protein-truncat-
oncology. ing mutations, the research community was quick to
• Attempts to identify a third breast cancer susceptibility locus (BRCA3) have so far been adopt the in vitro translation technique — also known as
unsuccessful. This is probably because no single gene can account for the remainder of the protein-truncation test (PTT) — which was adapted
families that show a high incidence of breast cancer that is not associated with BRCA1 by Hogervorst in 1995 (REF. 15). The test was quick and
or BRCA2. inexpensive, and reagents were available in kit form. PTT
• In general, the genes that have been identified as being associated with hereditary is still widely used to detect BRCA1 and BRCA2 muta-
breast cancer (BRCA1, BRCA2, TP53, CHK2 and ATM) are involved in the tions. The main limitation of PTT is that it is not effec-
maintenance of genomic integrity and DNA repair. tive for screening small exons using genomic DNA sam-
• The risk of developing cancer is not identical for all carriers of BRCA1 and BRCA2 ples. Some laboratories have restricted its use to the
mutations. Risk can be influenced by allelic heterogeneity, modifier genes, and screening of large exons of BRCA1 (exon 11) and BRCA2
environmental and hormonal cofactors. (exons 10 and 11), and other laboratories have adapted it
for the screening of cDNAs that are generated by reverse
transcription. To ensure a comprehensive and sensitive
‘gold standard’, Myriad Genetics developed a robotic
genetic testing will be perceived to be useful in Asia, sequencing technique to screen for mutations on a com-
Africa and South America, where it would compete for mercial basis. Other more rapid and less expensive tech-
scarce health-care resources, is a question for the next niques have been developed to identify mutations16–18
decade. Even in North America, interest in genetic testing and, in general, these perform well19,20 (TABLE 1).
is greater among white than non-white populations11. If a family contains more than two cases of early-
Because BRCA mutations occur at a frequency of onset breast cancer and at least two cases of ovarian
about 1 in 250 women, there are probably 250,000 cancer, it is likely that a BRCA mutation will be found21.
women in the United States who are carriers. However, in most families, it is only possible to identify
However, it is likely that fewer than 10,000 of these BRCA1 or BRCA2 mutations from a small proportion
women have been identified. This might be because of of women who receive genetic counselling. Factors that
the perceived lack of effective preventive measures. It predict that a mutation will be found include the num-
might also be the case that only a minority of women ber of affected relatives who have breast or ovarian can-
with mutations have a family history of cancer that is cer, their ages at the time of diagnosis of breast cancer
sufficiently strong to attract notice and result in refer- (but not ovarian cancer), Jewish ancestry and certain
ral to a genetic-testing centre. Some women might be pathological features of the cases of breast and ovarian
reluctant to pursue testing because of concerns about cancer that have occurred.
confidentiality and discrimination, whereas others A great deal of effort has been expended over the past
might not have access to testing facilities. But what has decade in identifying families with a history of breast
the hunt for breast-cancer-associated genes taught us cancer that can be accounted for by mutations in these
about breast cancer itself? two genes. Of course, small numbers of breast cancer
cases in a single family might occur by chance, but
BRCA1 and BRCA2 mutations chance alone cannot explain the many cases that occur in
To some extent, the types of mutation that have been some families that do not carry BRCA mutations. These
reported reflect the ease with which they are detected cases could be caused either by large deletions or loss-
and the unambiguous nature of their effects on the of-function mutations in BRCA1 or BRCA2 that are not
BRCA1 protein. For this reason, most of the mutations detected by conventional screening techniques, or by
that were first reported result in protein truncations; mutations in other genes.
these are either small insertions or deletions, or are A few researchers have systematically attempted to
nonsense mutations that lead to the introduction of a find large deletions in the genomes of cancer families.
stop codon1,12–14. These mutations invariably generate a These mutations would usually be missed by conven-
shortened, non-functional BRCA1 protein. tional sequencing or by the PTT. These studies have been
aided by recent technical developments, such as MULTIPLEX
22
LIGATION-DEPENDENT PROBE AMPLIFICATION . Genomic dele-
Box 1 | BRCA mutations in women with breast and ovarian cancer
tions in BRCA1 are infrequent, accounting for only
Studies indicate that it is worthwhile to screen all patients with invasive ovarian cancer or 5–10% of all germline mutations, and these mutations
certain types of breast cancer, as more than 10% of tests will identify a BRCA1 or BRCA2 are probably even less common in BRCA2 (REFS 23,24).
mutation (see table below). Complex rearrangements that involve BRCA1 — which
often involve repetitive elements, such as A SEQUENCES —
LU
Group Proportion with BRCA mutations
have been reported25,26, but are also likely to be rare. If the
Women with invasive ovarian cancer (all ages) 12% proportion of mutations that are due to genomic dele-
Jewish women with breast cancer (all ages) 11% tions and rearrangements proves to be as high as 10% of
Families with two or more cases of breast 12% the total, then these assays could become part of the
cancer in women under 50 years of age complete mutational analysis of BRCA1.

666 | SEPTEMBER 2004 | VOLUME 4 www.nature.com/reviews/cancer

 REVIEWS

Table 1 | Techniques used to detect mutations in BRCA1 and BRCA2

Technique Advantages Disadvantages
Protein-truncation test Cheap; rapid; allows detection of Mutations can be missed when gene product is
genomic deletions very short; does not detect missense mutations;
RNA required to examine small exons
Single-strand-conformation Simple, well-established technique Low sensitivity; labour-intensive; does not
analysis of genomic DNA detect exon deletions*
Denaturing high-performance Detects almost all intra-exonic and Expensive equipment required; does not detect
liquid chromatography splice-site mutations; rapid exon deletions*
DNA chips Can potentially identify all sequence Expensive equipment required; high cost
variants; very rapid per chip
Direct sequencing Identifies most intra-exonic and Expensive; exon deletions can be missed if
splice-site mutations detailed single-nucleotide-polymorphism
analysis is not carried out
Multiplex ligation-dependent Detects all exon deletions Cannot detect intra-exonic mutations
probe amplification
*This type of deletion is thought to be rare in most populations.

Since they were first reported in 1994, there have both cheap and reliable in these countries, population-
been difficulties in interpreting the effects of missense based screening should be feasible33,34. Founder muta-
mutations on BRCA1 function. These mutations tions also exist in other countries, but the number of
change a single amino acid, but otherwise leave the pro- additional variants of BRCA genes in these popula-
tein intact. In some cases, these are thought to disrupt tions decreases the efficacy of screening. In some coun-
protein function, but in other cases they are neutral tries, it might be useful to identify common mutations
polymorphic variants. More than 300 missense before undertaking an extensive (and expensive)
sequence variants in BRCA1 have been submitted to the genomic search for cancer-causing mutations.
BIC database. It is challenging for geneticists to assess However, in countries with ethnically mixed popula-
the risk of developing cancer for a woman who is found tions — such as the United Kingdom, Canada and the
to carry an unclassified variant, and it is unsatisfactory United States — the range of genetic variation is wide,
for the patient to be given an ambiguous result. Factors so it has never been possible to narrow the search for
MULTIPLEX LIGATION-
that can be used to assess the pathogenicity of variants cancer-causing genes, except in patients of Jewish
DEPENDENT PROBE include the relative frequency of the variant in cases ver- ancestry. Because one of the three founder mutations
AMPLIFICATION sus healthy controls and the co-segregation of the vari- is present in 2% of Ashkenazi Jews, this group has been
A technique used to determine ant with disease in the family. There have been attempts studied extensively, and much of our knowledge about
the copy number of multiple
to classify variants by measuring their effect on protein the penetrance, pathology and natural history of
specific sequences in a single
reaction. Two probes are function in vitro 27–29, but there is currently no test that hereditary cancer has been derived from this relatively
hybridized to the target sequence can be readily adapted for clinical purposes, and those small group.
and are joined together by that have been proposed have yet to be validated in large
ligation to make a copy of that numbers of patients. Penetrance
sequence. The probes are
designed so that all the products
The penetrance of BRCA mutations is still a matter of
can be amplified using the same Founder effects intense research 10 years after the discovery of these
primer pair. The relative In the years immediately after the identification of genes (BOX 2). It is likely that more effort has gone into
quantity of each product BRCA1, several research teams undertook the catego- estimating the penetrance of BRCA1 mutations than
establishes the copy number of
rization of mutations in different populations. In the for mutations of any other gene. This investment is
the target sequence.
first wave of number, Simard and colleagues identi- rational, given the high frequency of mutations and
ALU SEQUENCES fied recurrent mutations in a small series of families the obligation to communicate risk with accuracy
Short interspersed nuclear from Quebec, Canada 12. Two mutations were seen prior to offering drastic preventive options, such as
elements present at a high more than once (185delAG and 5382insC), and fami- prophylactic mastectomy. It is perhaps disappointing
frequency in primate genomes.
Alu sequences are amplified in
lies with these mutations were found to be of Jewish that there is still controversy regarding which estimates
the genome by ancestry. Two years later, Offit and Neuhausen identi- of penetrance should be used to counsel women with
retrotransposition. A complete fied a third mutation (BRCA2 6174delT) that was BRCA1 and BRCA2 mutations; however, it is probably
Alu sequence is approximately also associated with Jewish ancestry30. Investigators not surprising, as different studies continue to generate
300 bp long and contains an
have shown that if a Jewish woman does not carrry different figures. Penetrance — the lifetime risk of
A-rich region near the centre
and a stretch of As at the 3′ end. one of these three FOUNDER MUTATIONS it is highly developing breast or ovarian cancer — is usually
unlikely that a different mutation will be found. defined as the risk up to the age of 70 years . Both
FOUNDER MUTATIONS Many researchers therefore believe that genetic test- BRCA1 and BRCA2 mutations seem to have pene-
Specific mutations that appear ing in Jewish women could be limited to testing only trance values for breast cancer of about 80%. It is
repeatedly in ethnically defined
groups because of a shared
for these three mutations31,32. widely accepted that the risk of ovarian cancer among
common ancestry and, typically, Founder mutations have also been identified in carriers of BRCA1 mutations (about 40%) exceeds that
rapid population growth. Icelandic and Polish populations. As genetic testing is for carriers of BRCA2 mutations (about 20%), and

NATURE REVIEWS | C ANCER VOLUME 4 | SEPTEMBER 2004 | 6 6 7

REVIEWS

RAD51 FOCI that BRCA2 carriers who develop ovarian cancer typi- known functions of BRCA1 might underlie its role in
Discrete nuclear foci comprised cally do so at an older age35. The risk of ovarian cancer carcinogenesis. These roles include DNA repair, cell-
of DNA-repair complexes that is not the same for all BRCA2 mutations; women with cycle-checkpoint control, protein ubiquitylation and
accumulate after endogenous or
a mutation in the central part of the gene — the ovar- chromatin remodelling (FIG. 2). These complex issues
induced DNA damage. BRCA2
is a component of these foci and ian cancer cluster region (OCCR) — probably have a have been discussed in detail in several reviews45–47.
delivers RAD51 to the sites of higher lifetime risk36. This is puzzling, given that these
DNA damage. BRCA1 might are all truncating mutations and that it was initially DNA repair. Both BRCA1 and BRCA2 are implicated in
also be required to complete predicted that they would all have similar conse- the repair of DNA by homologous recombination (FIG. 2).
these foci.
quences for protein function. Conversly, male carriers BRCA1 associates with RAD51 in subnuclear clusters48.
of BRCA2 mutations that lie outside the OCCR have RAD51 is a key component of the mechanism in which
been proposed to have an increased risk of developing DNA damage is repaired by homologous recombination.
prostate cancer 36,37. When DNA is damaged, both BRCA1 and RAD51 local-
The BRCA2 mutation that is typically found in Jewish ize to to the damaged region, and BRCA1 is phosphory-
women (6174delT) is about as frequent in the general lated during this process. The nature of the interaction
Jewish population as the 185delAG mutation38, but is between RAD51 and BRCA1 is unknown, whereas it is
found less often than 185delAG in cases of breast cancer known that BRCA2 can interact directly with RAD51,
across all populations39. 6174delT is also found much less both through its BRC repeats and through a domain in its
frequently in families with multiple cases40. The explana- carboxyl terminus49–51. BRCA2 forms a complex with
tion for this discrepancy is that the penetrance of the RAD51, holding it in an inactive state, and when BRCA2
6174delT mutation is lower for breast cancer than that of is absent, RAD51 FOCI do not form after DNA damage.
185delAG. Warner and colleagues estimated the breast Cells that are defective for BRCA1 or BRCA2 are
cancer penetrance of the BRCA2 6174delT mutation to hypersensitive to agents that crosslink DNA strands or
be low (28%, compared to 80% for BRCA2 mutations in that produce breaks in double-stranded DNA, such as
general). Again the basis for this difference is unknown, cisplatin and mitomycin C52–54. In these cells, double-
but it is interesting that this mutation lies in the OCCR. strand breaks are repaired by an error-prone mechanism
It should also be remembered that estimates of pene- — such as non-homologous end joining — and errors
trance vary between countries for reasons that are unre- can lead to chromosomal rearrangements55,56. It is
lated to genetics. For example, oral-contraceptive use, thought that the resulting chromosomal instability is a
oophorectomy and parity all influence the risks of ovar- crucial feature of carcinogenesis. When cells are exposed
ian and breast cancer41, and these factors vary between to ionizing radiation, both BRCA1 and BRCA2 (together
countries. Furthermore, as more women become aware with RAD51) initiate homologous recombination and
that they are carriers of BRCA mutations, the pene- the repair of double-strand breaks54. Unsurprisingly, cells
trance values that are associated with each mutation will that express mutated BRCA1 and BRCA2 are hypersensi-
decrease, due to preventive efforts such as prophylactic tive to ionizing radiation and show error-prone repair.
oophorectomy and prophylactic mastectomy 8,9,42–44. The levels of expression of BRCA1, BRCA2 and RAD51
increase in cells when they enter S phase, indicating that
Functions of BRCA1 and BRCA2 they function during or after DNA replication. So,
Not all of the functions of the BRCA1 and BRCA2 pro- BRCA1 and BRCA2 function in a common pathway that
teins have been established, although many have been is responsible for the integrity of the genome and the
discovered during the past decade. BRCA2 is the larger of maintenance of chromosomal stability 46.
the two proteins and consists of 3,418 amino acids (FIG. 1). As well as being involved in the repair of double-
BRCA2 is involved in homologous recombination, but strand breaks, BRCA1 has also been implicated in
little else is known about its function. By contrast, several nucleotide-excision repair. This involves two different
mechanisms — transcription-coupled repair, in which
the transcribed strand is preferentially repaired, and
global genome repair, which does not show strand bias.
Box 2 | Estimating penetrance
BRCA1 might have a role in both transcription-coupled
There is still controversy regarding which epidemiological method is best for repair57 and in global genome repair58. The effect of
estimating the cancer risk that is associated with a particular mutation. Scholars are in BRCA1 deficiency on transcription-coupled repair
two camps — the first of these (which holds most of the original members of the Breast might be limited to the blockage of theRNA polymerase
Cancer Linkage Consortium) hold that the risk of developing breast cancer is high for II transcription machinery at the site of repair of
mutation carriers, and is largely independent of the ascertainment scheme by which 8-oxoguanine residues. This would be expected to be
the family was identified. This approach relies on the study of families that have a accompanied by specific patterns of somatic mutation
history of breast cancer as the proper and logical unit of study. Members of the other (G>T transversions) in cancers that arise in carriers of
camp argue that penetrance estimates that are derived from studying large families will BRCA1 mutations. In the study that indicated a role for
be too high and can generate unnecessary fear in the general population. They BRCA1 in global genome repair58, the effect of BRCA1
therefore promote population-based studies (the inclusion of women in a study deficiency was shown to be independent of p53 status,
regardless of their family history). Most genetic testing, however, is performed on
and no effect on transcription-coupled repair was seen.
women who ask to be tested, because of a family history of cancer. It is hoped that
Unfortunately, progress in this area has been slowed by
prospective studies will provide the most effective analysis yet, but these studies require
retractions of key publications, lack of replication studies
large numbers of patients and extended periods of follow-up.
and the use of different delivery and test systems.

668 | SEPTEMBER 2004 | VOLUME 4 www.nature.com/reviews/cancer

 REVIEWS

a BRCA1 aa 200–300 aa 1,646–1,859 The BRCT motif at the carboxyl terminus of BRCA1
NLS NLS BRCT domains (FIG. 1) is a common feature of proteins that are involved
aa 8–96 aa 452–1,079 aa 1,280–1,524 ~110 aa in DNA repair and/or cell-cycle checkpoints60. Interest-
each ingly, unlike BRCA1, the checkpoint function is pre-
RING finger DNA binding SCD 1,863 aa
served in BRCA2-deficient primary cells55. However,
both Brca1–/– and Brca2 –/– mice die during early stages of
embryogenesis61,62. Loss of p53 or WAF1 (also known as
BARD1 p53 BASC RAD51 ATM RNA polymerase II
BAP1 CHK2
p21) delays this embryonic lethality by a few days63,
E2F1
RB
SW1/SNF CDK2 p300 indicating that the absence of checkpoint control might
BACH1 be a crucial step in tumorigenesis. Most BRCA1- and
BRCA2-null cells undergo apoptosis because of intact
MYC ZBRK1 HDAC1/HDAC2
checkpoint controls, but cells in which BRCA1 or
GADD45 p53 CtIP BRCA2 are disrupted — as well as those in which key
RB checkpoint proteins such as p53 or WAF1 are inacti-
vated — survive in the presence of genomic instability.
b BRCA2 BRC repeats
80–300 aa spacing (variable)
NLS This results in the typically abnormal karyotypes that
are seen in breast cancers associated with mutations in
3,418 aa BRCA1 and BRCA2 (REFS 64–66). Despite the observation
that most tumours from women with BRCA1 or
BRCA2 mutations show loss of the corresponding wild-
RAD51 DSS1
type allele, some cancers seem to arise in the presence of
Figure 1 | BRCA1 and BRCA2 functional domains, and selected binding partners. an intact wild-type allele. It has been proposed that in
a | BRCA1, which consists of 1,863 amino acids, contains several important functional domains that these cases — and possibly even in cases with two ‘hits’
interact with a range of proteins. The RING-finger domain binds to BARD1 and this binding that affect the same BRCA gene — the second event in
enhances the ubiquitin-ligase function of BRCA1. RING-finger binding to BAP1 and E2F1 has not
tumorigenesis might involve the inactivation of a check-
been confirmed by in vivo studies. p53, MYC, RB and ZBRK1 all bind to a region of BRCA1 that
includes the nuclear localization signals (NLSs). ZBRK1 is a zinc-finger protein that suppresses point gene, rather than loss of the second BRCA1 or
transcription through an interaction with GADD45. BRCA1 is required for this repression. SW1/SNF BRCA2 allele46,67.
binding occurs between amino acids 260 and 553. The DNA-binding domain encompasses amino
acids 452–1,079. It contributes to the DNA-repair-related functions of BRCA1, which are partly Ubiquitylation. Ubiquitylation is the process by which
mediated through proteins that make up the BRCA1-associated surveillance complex (BASC). proteins are tagged for degradation by the PROTEASOME.
Several proteins (including MRE11, RAD50, NBS1, MDC1, ATM, CHK2 and CDK2) bind to the
Many proteins that have ubiquitylation functions con-
central region of BRCA1. SQ sequences (clusters of serine and threonine sequences), known as
SQ-cluster domains (SCDs) are preferred sites of ATM phosphorylation. There are several SQ
tain a RING-FINGER MOTIF. Both BRCA1 and its interacting
sequences between amino acids 1,280 and 1,524, whereas they are rare elsewhere in BRCA1. protein BARD1 have a RING-finger motif near to their
Two regions at the carboxyl terminus — known as the BRCT domains — are each about 110 amino amino termini (FIG. 1), and it has been shown that the
acids long, and comprise amino acids 1,646–1,859. BRCT domains are found in many proteins BRCA1–BARD1 complex functions in the ubiquitylation
involved in DNA-repair pathways, and bind to many proteins, including RNA polymerase II, p300, process68 (FIG. 2). Disease-associated mutations in BARD1,
BACH1, histone deacetylases (HDACs) 1 and 2, p53, CtIP (carboxy-terminal-binding-protein however, are rare in BRCA1- and BRCA2-negative breast
interacting protein) and RB. The RNA polymerase II holoenzyme binds to both the amino and
carboxyl termini of BRCA1; amino-terminal binding (not shown) is through the BARD1–BRCA1
tumours69–71. Interesting recent studies indicate that
complex. RAD51 and BRCA2 also bind BRCA1, and the three proteins colocalize in sub-nuclear BRCA1-mediated ubiquitylation occurs in response to
foci. RAD51 binds BRCA1 directly, as shown. BRCA2 interacts with the BRCT domains; this replication stress72, linking its ubiquitylation function
interaction may be indirect, possibly as part of a complex with RAD51. b | BRCA2 is larger than to the DNA-damage response.
BRCA1 (consisting of 3,418 amino acids), but only contains two known functional domains. The
middle region of the protein contains eight BRC-repeat motifs, which are essential for its function in Chromatin remodelling. Chromatin remodelling occurs
DNA repair and bind to the DNA recombinase RAD51. The DNA-repair activity of BRCA2 is
around double-strand DNA breaks and is thought to
regulated by DSS1, a small, acidic protein that seems to function as a necessary cofactor. DSS1
binds to the carboxy-terminal region of BRCA2, which also includes the NLSs. See REFS facilitate DNA repair. Several multimeric complexes —
46,47,178,179 for further details. including BASC — are involved in this process, and
BRCA1 seems to be a member not only of BASC, but
also of a complex that contains the chromatin-
remodelling proteins SW1 and SNF 73 (FIG. 2). There
Checkpoint control. Another function of BRCA1 is in seems to be a direct interaction between BRCA1 and the
checkpoint control (FIG. 2). BRCA1 can exist as part SW1–SNF complex, indicating that these proteins func-
of the BRCA1-associated genome-surveillance com- tion as a unit in the remodelling of chromatin that
PROTEASOME plex (BASC)59. This complex includes proteins such occurs around sites of DNA damage. Activation of other
An organelle that breaks down
as Nijmegen breakage syndrome 1 (NBS1), the genes that are implicated in the response to DNA
proteins that have been targeted
for degradation by RAD50– MRE11 complex (which has exonuclease damage, such as KU70 and GADD45, results from this
ubiquitylation (by having a activity at double-strand breaks), ataxia telangiectasia interaction. Interestingly, mutations that affect another
ubiquitin tag added to the mutated (ATM; which functions upstream of BRCA1 component of the SW1/SNF complex, SNF5, have
protein). Lack of regulation of in the double-strand-break repair pathway), the been found in some patients with rare paediatric malig-
proteasomal degradation leads,
for example, to loss of control of
MLH1–PMS1 and MSH2– MSH6 complexes, the nancies74. The fact that BRCA1 can also function as a
the cell cycle and seems to be an BLM protein that is affected in Bloom syndrome, and histone deacetylase75 and interacts with other proteins
important step in tumorigenesis. DNA replication factor C. that are implicated in chromatin remodelling, such as

NATURE REVIEWS | C ANCER VOLUME 4 | SEPTEMBER 2004 | 6 6 9

REVIEWS

DNA damage

Sensor

X-chromosome ATM–ATR BACH1

silencing
Kinase Chromatin
SW1/SNF remodelling
CHK2
HDACs
XIST
Non-homologous
Phosphorylation Access to DNA
end-joining

P BLM
BARD1
Heterodimerization MSH2–MSH6 Homologous
BRCA1
recombination
At nuclear MRE11–RAD50–NBS1
Ubiquitin ligation foci

BRCA2 FANCD2 CHK1 Transcriptional

regulation
p53
RAD51
RB
Target? G2/M phase
PLK1 Checkpoint
WAF1 regulation
S phase and
G2 arrest ↑ GADD45 G1/S phase
DNA repair
Homologous
recombination

Figure 2 | The BRCA1 network. BRCA1 is an important component of pathways that regulate DNA repair, cell-cycle progression,
ubiquitylation and transcriptional regulation. DNA damage (shown at the top of the figure) is thought to be one of the key triggers of
BRCA1 activation. Several damage sensors, including ataxia telangiectasia mutated (ATM) and other kinases, are activated in response
to DNA damage. CHK2 is also activated, and prevents cell division by phosphorylating BRCA1 and p53. Downstream targets of
BRCA1 activation include p53 and the retinoblastoma protein (RB). BRCA2 and RAD51 form a complex that is believed to interact with
FANCD2, which binds to BRCA1. This complex promotes S-phase or G2 arrest. BRCA1 forms a heterodimer with BARD1 to activate
the ubiquitin-ligase function of BARD1, although its targets are unknown. DNA repair by homologous recombination is mediated by the
BRCA1-associated surveillance complex (comprised of BLM, MSH2–MSH6 and MRE11–RAD50–NBS1). This complex also regulates
transcription. BRCA1 has been shown to interact with X-inactive specific transcript (XIST) to mediate X-chromosome silencing, and
also to mediate non-homologous end joining during DNA repair. BRCA1 can form complexes with both BACH1 and SW1/SNF to
mediate chromatin remodelling and homologous recombination. HDACs regulate the access of the SW1/SNF–BRCA1 complex to
DNA. Finally, BRCA1 interacts with CHK1 and polo-like kinase 1 (PLK1) to regulate the G2/M and G1/S checkpoints, possibly via
GADD45; thereby linking BRCA1 to the regulation of apoptosis. See REFS 46, 47,178,179 for further details.

BACH1 (REF. 76), emphasizes the importance of BRCA1 (REF. 79). Other studies have shown that, in rare cases, chil-
in processes that regulate DNA repair. Interestingly, a dren with medulloblastoma or Wilms’ tumour also carry
BRCA2-interacting protein, EMSY 77, also has DNA- two truncating BRCA2 mutations80. Homozygosity for
repair functions, and mutations that affect this protein Brca1-inactivating mutations, however, results in embry-
might underlie sporadic breast and ovarian cancers. onic lethality, confirming the functional differences
between the two proteins.
The Fanconi-anaemia connection
Fanconi anaemia is a rare recessive disease of childhood Models of tissue specificity
that features skeletal abnormalities, abnormal skin pig- Why are the cancer phenotypes that are associated
mentation, short stature and microphthalmia. Mutations with BRCA1 mutations so specific if the functions of
in several genes can cause this condition, but all lead to the gene are so general? One possibility is that
chromosomal instability. When fibroblasts from children absence of BRCA1 could exacerbate the action of tis-
with Fanconi anaemia are exposed to mutagens such as sue-specific carcinogens, such as oestrogen. If breast
RING-FINGER MOTIF mitomycin C or diepoxybutane they show an increased cancer is the simple result of abnormal oestrogen
A motif comprised of cysteine frequency of chromosome breaks compared with normal sensitivity, however, then endometrial cancer should
and histidine residues fibroblasts78. This is similar to the chromosomal instabil- also be associated with BRCA mutations. Further-
interspaced with hydrophobic
amino acids. Proteins that
ity that is seen in Brca2-deficient mice55. It was still sur- more, oestrogen exposure has never been proven to
contain this motif usually have prising, however, when a rare form of Fanconi anaemia be required for ovarian carcinogenesis. Monteiro
ubiquitin-ligase functions. was shown to be caused by biallelic mutations in BRCA2 postulated that tissue-specific differences in mitotic

670 | SEPTEMBER 2004 | VOLUME 4 www.nature.com/reviews/cancer

 REVIEWS

recombination might underlie the specificity of 1970s — that the genes that are involved in hereditary
BRCA mutations for breast and ovarian cancers81. cancer syndromes are the same as those that are involved
Elledge and Amon82 proposed that only cells from in non-hereditary cancers, and that BRCA1 would fall
breast and ovarian tissues are able to survive after into the category of a classic tumour suppressor. This
acquiring defects in BRCA1 because of the anti-apop- seemed to be strengthened by the observations of Devilee
totic effects of oestrogen85. Anti-oestrogen therapies and colleagues that loss of heterozygosity for BRCA1
should therefore decrease the incidence of both breast occurs in cells from breast and ovarian tumours of
and ovarian cancer in BRCA1 carriers. Most BRCA1- patients with germline mutations103, that the wild-type
related breast cancer cells, however, are oestrogen allele is ‘lost’ in BRCA1-linked breast cancer kindreds104
receptor (ER)-negative83. We propose a model that is a and that cancer-associated BRCA1 mutations result in
modification and an extension of that of Elledge and loss of function. It was therefore surprising that alter-
Amon, based on the assumption that the breast cancer ations of single base pairs in the BRCA1 coding region are
stem cell is ER-negative, but the surrounding cells are only rarely associated with breast cancer105,106 or ovarian
ER-positive84. These surrounding cells might respond cancer107,108. These findings seem to contradict the classic
to oestrogen and send pro-survival signals to the TWO-HIT MODEL OF TUMORIGENESIS.
ER-negative cancer stem cells45. Possible explanations for the small number of muta-
Breast stem cells have a high intrinsic proliferative tions that are seen include a narrow developmental
capacity. As a woman reaches menopause, oestrogen lev- window during which mutations can result in a recogniz-
els fall and there is a lower probablility that BRCA1-null able phenotype45,109; an intrinsically low mutation rate for
cells will survive. Similarly, tamoxifen and oophorectomy BRCA1 (REF. 110); and the small number of mutation-
are pro-apoptotic86,87, so BRCA1-null cells that are prone breast stem cells111. It is important to note that the
exposed to either of these interventions are more likely to lack of somatic mutations that are associated with spo-
die early on. This could explain the sharp decrease in can- radic breast cancer does not mean that somatic muta-
cer risk that is seen at menopause in carriers of BRCA1 tions do not exist in patients with hereditary breast cancer
mutations88, which — notably — is not seen in carriers of — large-scale studies have not been published. If point
BRCA2 mutations. However, if BRCA1-null cells do mutations in BRCA1 and BRCA2 are rare or absent in all
escape this apoptotic mechanism, they would accumulate forms of breast cancer, this might represent an effect that
new mutations and begin to proliferate rapidly. is related to the spatiotemporal expression of BRCA1 and
BRCA2 in the developing breast, because point mutations
Is there a heterozygote phenotype? in other cancer-related genes — such as TP53 — are
There is no clear phenotype in murine or human carri- common in breast cancer112.
ers of heterozygous mutations in BRCA1 or BRCA2 — The transcription of BRCA1 is in part regulated by
both species seem to develop normally in the presence of the methylation of CpG islands at the 5′ end of the gene,
a single mutant allele. The possibility that a more subtle and several studies have shown that altered methylation
phenotype might exist in the breasts or ovaries of carri- of this region can lead to gene silencing113. In breast can-
ers has led several investigators to study tissues that are cers that are not associated with BRCA1 mutations,
removed at the time of preventive or therapeutic surgery. methylation of this gene seems to be a frequent
Results have been conflicting 89–95, but overall there is event114,115. Interestingly, breast tumours that are
scant evidence for an increase in the frequency of pre- associated with BRCA1 hypermethylation are histo-
malignant conditions in these organs compared to pathologically similar to those that are caused by inher-
women from the general population. Several researchers ited mutations in BRCA1 (TABLE 2), in that they are high-
have attempted to adapt the conventional model of can- grade, infiltrating ductal breast cancers that do not
cer progression — which starts with hyperplasia, leading express ER. Low levels of BRCA1 mRNA in unselected
to in situ carcinoma and finally invasive cancer — to breast cancer specimens also support a role for the
hereditary breast and ovarian cancers, but their findings altered regulation of this gene in non-hereditary forms
have not been replicated93,96,97. of breast cancer116,117 and have been associated with a
Others researchers have studied BRCA heterozygosity poor outcome following breast cancer in some studies118,
in lymphocytes, rather than in healthy breast and ovarian but not in others119. So, alterations in BRCA1 function
TWO-HIT MODEL OF cells98–100. There is some evidence that carriers might be more frequent that is commonly believed, as
TUMORIGENESIS
States that both alleles of a
of BRCA1 mutations have an increased number of chro- loss of function of BRCA1, whether by genetic or epi-
tumour-suppressor gene need to mosome breaks, which is manifested as the presence of genetic mechanisms, tends to result in a recognizable
be inactivated to promote micronuclei in lymphocytes. However, it is not clear if phenotype. Further analysis of BRCA1 methylation and
unregulated tumour-cell these heterozygous cells are genetically unstable, and expression patterns will be a challenge, as this analysis is
growth. A given allele could be
some of these results have been questioned101,102. laborious and requires well-preserved specimens.
inactivated due to inherited
mutation (constitutional), Researchers have consistently been frustrated by the
somatic mutation or epigenetic Molecular carcinogenesis lack of a dependable antibody against BRCA1, which
silencing. Hereditary tumours In the early 1990s, it was anticipated that although would allow them to evaluate protein levels in normal
would be caused by an inherited germline BRCA1 mutations are rare, a much greater pro- and tumour tissues. Several antibodies have been raised
mutation and a somatic
mutation; non-hereditary
portion of breast cancers might be attributable to somatic against various BRCA1 epitopes, but immunohisto-
tumours would be the result of mutations in BRCA1. This prediction was, of course, chemical analyses with these antibodies have not given
two somatic mutations. based on a paradigm introduced by Knudson in the reproducible results.

NATURE REVIEWS | C ANCER VOLUME 4 | SEPTEMBER 2004 | 6 7 1

REVIEWS

Pathology Whether this information will be clinically useful is not

In 1993, the Breast Cancer Linkage Consortium (BCLC; yet known, but one small study found that carriers of
see online links box) was created to facilitate collaborative BRCA1 and BRCA2 mutations are more likely to show
breast cancer linkage studies. Admirably, the group stayed a complete response to preoperative chemotherapy
together following the cloning of the BRCA genes and has than non-carriers134. In general, progress has been ham-
facilitated many clinical and pathological studies. pered by the expense of mutation detection and the
Through the work of the BCLC and other groups, a clini- absence of large, well-characterized patient cohorts and
copathological phenotype for BRCA1-related breast can- appropriate comparison groups. However, such studies
cer has emerged120 (TABLE 2). Sobol and colleagues have are now well underway in The Netherlands and Israel.
suggested that these features are sufficiently specific that The ability to assay for founder mutations in archived
they can be used to identify probable carriers, based on specimens and to link these to data about clinical out-
tumour pathology121. BRCA1-related breast cancers are comes allows large-scale, historical cohort studies to be
usually high-grade infiltrating ductal carcinomas. An carried out in these countries. A large prospective study
atypical medullary phenotype (which is characterized by is now underway in Poland.
syncytial growth patterns, a smooth margin and abun- The progression of BRCA1-associated breast tumours
dant lymphocytic infiltration) is more common in differs from that seen in sporadic cases in at least two
BRCA1-related breast cancer than in matched con- ways. First, among BRCA1-mutation carriers, there is
trols83,120 but occurs in only ~10% of BRCA1-related only a weak relation between the size of the primary can-
tumours. Conventional and molecular karyotyping stud- cer and the number of axillary lymph nodes to which the
ies have shown that the cells of these tumours are usually tumour spreads135. Several studies have reported, for
highly disorganized64,65. They are also usually ER-nega- example, that the poor prognosis that is associated with
tive, particularly in younger women. Notably, the receptor BRCA1 mutations is restricted to women with node-neg-
tyrosine kinase ERBB2 (also known as HER2 or NEU) is ative disease136–138. Large BRCA1-related tumours are also
overexpressed less often, compared with age-matched much less likely to be node-positive than would be
controls122. Many other immunohistochemical markers expected compared with sporadic tumours or tumours
have been studied, most in small series. in women with BRCA2 mutations. The reported poor
Microarray analysis has allowed a more detailed prognosis for BRCA1-related node-negative breast cancer
analysis of the gene-expression patterns of various is surprising, as tumours that have not spread to the
breast tumours. The most important initial categoriza- lymph nodes are usually associated with a significantly
tion of breast cancers is into ER-positive and ER- better outcome than are tumours of a similar size that are
negative subsets123–125. Most ER-positive cancers seem to node-positive. It is possible that node-negative BRCA1-
show a luminal phenotype, as determined by expression related breast cancers might show atypical metastatic
of simple keratins, such as cytokeratins 8 and 18. By routes of dissemination, posing challenges for breast can-
contrast, ER-negative cancers can be classified according cer screening. The second atypical feature of BRCA1-
to whether they overexpress ERBB2. Tumours that are associated cancers is their apparent ability to respond to
both ER- and ERBB2-negative are characterized by the oestrogen blockade, despite being ER-negative.
presence of basal cytokeratins, such as cytokeratins 5, 6 Oophorectomy is associated with a reduction in the inci-
and 14 (REF. 123). It is perhaps unsurprising, therefore, that dence of first and second primary breast cancers in
BRCA1-related breast cancers have a BASAL PHENOTYPE126,127. BRCA1 carriers139,140, and tamoxifen is effective in pre-
This is of some interest, because this phenotype is often venting second primary cancers — most of which are
associated with a specific expression pattern — apart ER-negative140.
from their ER- and ERBB2-negative status, tumours that Identifying a phenotype that characterizes breast
mainly express cytokeratins 5 and 6, rather than cytoker- tumours that are associated with BRCA2 mutations has
atins 8 and 18, also tend to overexpress cyclin E and p53 been more difficult. In general, BRCA2-associated
and to underexpress KIP1 (also known as p27)128. All of tumours cannot be readily distinguished from sporadic
these features have been associated with BRCA1-related cancers on a morphological basis120. Overexpression of
breast cancer. It has been argued that breast cells that cyclin D1 seems to be a useful marker for BRCA2-
express only cytokeratins 5 and 6 are adult stem cells129, related breast cancer141, but further studies are required
and it is tempting to speculate that BRCA1 has some role to confirm this. It is likely that significant differences do
in regulating the function of breast stem cells111. occur, if only because, like BRCA1-related tumours, they
Another important question is whether breast cancers have specific genomic alterations — as indicated by
that are associated with BRCA1 mutations behave more their distinctive profiles obtained by comparative
BASAL PHENOTYPE aggressively than sporadic tumours. Most of the evidence genomic hybridization64.
Describes a relatively rare
indicates that women with BRCA1-associated tumours Ovarian cancers that develop in both BRCA1 and
subtype of breast cancer that can
be defined by have a worse outcome than women with sporadic breast BRCA2 carriers are usually serous papillary carcino-
immunohistochemistry. These cancers130,131, but for those with mutations in BRCA2 the mas, although endometrioid and clear-cell carcinomas
tumours express markers that situation is less clear130. The effect of treatment, however, also occur142. By contrast, mucinous and borderline
are typically seen in normal has rarely been considered, and could be influenced by ovarian carcinomas are rarely seen in carriers of muta-
basal breast and skin epithelium,
such as cytokeratins 5 and 6.
chemotherapy132, as human BRCA1-null breast cancer tions in either gene142–144. Primary cancers of the fallop-
This phenotype is often cells are highly susceptible to this type of treatment. This ian tube and peritoneum are also seen; these tumours
associated with a poor outcome. effect is reversed when BRCA1 is reintroduced133. also have a characteristic serous papillary appearance,

672 | SEPTEMBER 2004 | VOLUME 4 www.nature.com/reviews/cancer

 REVIEWS

Table 2 | Key pathological characteristics of BRCA1- and BRCA2-associated breast cancers

Phenotype BRCA1-associated BRCA2-associated
Morphology Ductal, no specific type (75%); Ductal, no specific type (75%); atypical medullary < 5%;
atypical medullary ~10%. lobular or ductal with lobular features more prevalent
than in women with BRCA1 mutations (~10%)
Grade High (grade 3; 75%) Medium (grade 2; 45%) or high (grade 3; 45%)
Oestrogen-receptor expression Negative (75%) Positive (75%)
ERBB2 expression Negative (95%) Negative (95%)
p53 expression Positive (50%) Positive (40%)
Cyclin D1 expression Negative (90%) Positive (60%)
Carcinoma in situ Rare Common

indicating that these two genes restrict lineage choice in result, enthusiasm for this form of gene identification
developing cancers. Microarray studies of ovarian car- seems to be waning; no genome-wide searches for
cinomas have shown that almost all cases have expres- BRCA3 have been published. In Cambridge (United
sion profiles that are similar to either BRCA1- or Kingdom) and Lyon (France), Mike Stratton and David
BRCA2-related cancers, indicating that abnormalities Goldgar continue to assess random markers from a
of one of the pathways involving these genes is essential panel of 138 families in which three or more cases of
for ovarian carcinogenesis145. breast cancer have been diagnosed. At a meeting of the
BCLC in Madrid in 2003, they presented results from the
Other breast cancer genes first 400 markers. There was no compelling evidence of
The positional cloning of BRCA1 was achieved less than linkage to any region of the genome, so it is possible that
four years after linkage was first reported in 1990, and it no single gene underlies the breast cancer cases in these
took little more than a year to identify BRCA2 after its families. The supply of genetic markers is essentially
mapping in 1994. By 1996, it was clear that a substantial unlimited, but families that are ideally suited for linkage
percentage of breast cancer families do not carry muta- studies are still hard to find, so alternative strategies have
tions in either of these two genes, indicating the probable been proposed. In London, Ellen Solomon has collected
existence of additional cancer-susceptibility genes. This data on several hundred pairs of sisters who have both
led Mike Stratton and others to pursue the putative developed breast cancer. Her model-free approach could
BRCA3 gene. However, despite rapid advances in the be powerful if recessive genes underlie forms of inherited
high-throughput processing of DNA samples and the breast cancer that are not BRCA1- or BRCA2-associated,
completion of a comprehensive genetic map in the 1990s as has been indicated by some segregation analyses152. In
and of the sequencing of the human genome in 2000, addition, a study in Pakistan by Liede et al. reported that
BRCA3 remains elusive. If there is a BRCA3, it ought to the parents of young women with breast cancer were
have been found by now. Scientists such as Julian Peto more likely to be in consanguineous marriages153.
have suggested that families with a history of breast can- If it is not possible to define a characteristic BRCA3-
cer, but without BRCA mutations, might carry mutations type family, then it might be possible to identify a
that influence susceptibility in a more subtle or a more BRCA3-associated cancer fingerprint. Conventional
complicated manner, such as through gene–gene or histopathology154, microarray technology, loss of het-
gene–environment interactions146. Other researchers erozygosity and comparative genome-hybridization
believe that no single gene influences cancer risk in these arrays155 have all been used to try to identify a tumour-
families. If so, then studies that are specific for particular specific, BRCA3-associated signature. These studies
ethinic groups (for example, Finns or French Canadians) support the idea that familial breast cancers that are not
might have a greater chance of success — many different associated with mutations in BRCA1 or BRCA2 are prob-
genes might cause familial breast cancer clusters world- ably heterogeneous although, overall, they are less aggres-
wide or within an ethnically mixed population, but only sive than non-familial breast cancer. It is possible that
one or a few genes might contribute to cancer in an ethni- Peto is right, and that no other highly penetrant alleles
cally homogeneous population. Furthermore, a distinc- that predispose to breast cancer exist in ethnically mixed
tive phenotype for a third class of inherited breast cancer western populations. In this case, the clustering of breast
has not emerged. It is important to note that the associa- cancer in families is probably the result of a mixture of
tion between breast and ovarian cancer and the presence many interacting genes and chance. Paul Pharoah and his
of male breast cancers were instrumental in the searches colleagues156 have proposed a model in which risk is not
for BRCA1 and BRCA2, respectively. It is also possible symmetrically distributed — the 50% of the population
that the penetrance of BRCA3 mutations is low. that is at the highest risk would account for almost 90%
Several reports have been published since 1995 that of affected individuals.
indicate the linkage of the breast cancer susceptibility It was initially proposed by Swift in the early 1970s
phenotype to various regions of the genome147–149, but that the first-degree relatives of children with ataxia
none have been replicated in larger series150,151. As a telangiectasia, which is caused by mutations in ATM, have

NATURE REVIEWS | C ANCER VOLUME 4 | SEPTEMBER 2004 | 6 7 3

REVIEWS

an increased risk of developing breast cancer157. In the that is induced by ionizing radiation172–174, and activation
past decade, this hypothesis has been confirmed by more of this protein prevents cells from entering mitosis.
detailed epidemiological studies158 and by surveys of spe- Activated CHK2 phosphorylates BRCA1 and p53,
cific ATM mutations in unselected breast cancer cases159. thereby regulating their functions175–177.
By contrast, a study of the role of ATM mutations in
breast cancer reported no relation between the two160. Future directions
These investigators, however, screened only for mutations Despite the results of the studies described above,
in their population using the PTT. It is possible that trun- mutations in ATM and CHK2 are not sufficiently
cating mutations are not pathogenic, but that other types prevalent or penetrant to make these genes candidates
of mutation — for example, missense mutations — are for a third breast cancer gene — mutations in these
pathogenic161,162. One such mutation, ATM*7271T>G, genes only account for a small percentage of the breast
which probably originated in the Orkney islands (which cancer cases that occur in families that do not carry
are located north of Scotland), has been associated with a BRCA1 or BRCA2 mutations, and few clinicians offer
high risk of breast cancer in a few families159,163, but it is tests for mutations in these genes. During the past
not frequently associated with hereditary breast cancer164. decade, it has become apparent that only two genes are
A more general role for missense mutations could be clearly associated with inherited forms of breast cancer,
explained by a dominant-negative effect, in which the and that no simple Mendelian model will explain the
mutation might confer functions that are not present in remaining familial clusters. At the same time, we have
the wild-type protein. By contrast, when a truncating come to accept that genetic testing for BRCA1 and
mutation is present on one ATM allele but the other is BRCA2 has an important place in medical practice. We
normal, the product of the normal allele might be suffi- have also learned that defective DNA repair is a com-
cient to maintain normal function162. A large-scale study mon characteristic of all of the genes that underlie the
is therefore required to identify ATM mutations in hereditary breast cancer syndromes. It will be impor-
patients with breast cancer. tant to determine the reason that the tumorigenic
The biggest discovery in the field of familial breast effects of these mutations predominantly target breast
cancer research since the discovery of BRCA2 was proba- and ovarian cells and the reason that different muta-
bly the identification of the CHK2 gene. The CHK2 tions in the same gene confer different levels of risk for
founder mutation (1100delC), which abolishes the kinase the same type of cancer.
activity of the gene product165, was found to confer a In future studies, it will be important to determine
moderately increased risk of breast cancer in the roles of ER-negative breast stem cells in tumorigene-
Europeans166. There is a large degree of regional variation sis. Further molecular characterization of BRCA1- and
in the prevalence of this allele, but it seems to be most BRCA2-associated breast tumours should also bring
common in northern Europe (partcularly in The improvements in selecting effective chemotherapy regi-
Netherlands and Finland), and is rare among North mens and predicting prognosis. Finally, knowledge of
Americans167–169. CHK2 encodes the human homologue the specific molecular-genetic signatures of BRCA1-
of the yeast checkpoint kinases Cds1 and Rad53 (REFS and BRCA2-associated breast and ovarian cancers
170,171). CHK2 is activated in response to DNA damage might foster the development of new targeted therapies.

1. Miki, Y. et al. A strong candidate for the breast and ovarian 11. Mancuso, C. et al. Ethnicity, but not cancer family history, is 20. Eng, C. et al. Interpreting epidemiological research: blinded
cancer susceptibility gene BRCA1. Science 266, 66–71 (1994). related to response to a population-based mailed comparison of methods used to estimate the prevalence of
2. Hall, J. M. et al. Linkage of early-onset familial breast questionnaire. Ann. Epidemiol. 14, 36–43 (2004). inherited mutations in BRCA1. J. Med. Genet. 38, 824–833
cancer to chromosome 17q21. Science 250, 1684–1689 12. Simard, J. et al. Common origins of BRCA1 mutations in (2001).
(1990). Canadian breast and ovarian cancer families. Nature Genet. 21. Narod, S. A. et al. Genetic heterogeneity of breast–ovarian
3. Stratton, M. R. et al. Familial male breast cancer is not linked 8, 392–398 (1994). cancer revisited. Breast Cancer Linkage Consortium. Am. J.
to the BRCA1 locus on chromosome 17q. Nature Genet. 7, 13. Castilla, L. H. et al. Mutations in the BRCA1 gene in families Hum. Genet. 57, 957–958 (1995).
103–107 (1994). with early-onset breast and ovarian cancer. Nature Genet. 8, 22. Gad, S. et al. Color bar coding the BRCA1 gene on combed
4. Wooster, R. et al. Localization of a breast cancer 387–391 (1994). DNA: a useful strategy for detecting large gene
susceptibility gene, BRCA2, to chromosome 13q12–13. 14. Friedman, L. S. et al. Confirmation of BRCA1 by rearrangements. Genes Chromosom. Cancer 31, 75–84
Science 265, 2088–2090 (1994). analysis of germline mutations linked to breast and (2001).
5. Wooster, R. et al. Identification of the breast cancer ovarian cancer in ten families. Nature Genet. 8, 23. Gad, S. et al. Bar code screening on combed DNA for
susceptibility gene BRCA2. Nature 378, 789–792 (1995). 399–404 (1994). large rearrangements of the BRCA1 and BRCA2 genes in
These five papers describe the localization and 15. Hogervorst, F. B. et al. Rapid detection of BRCA1 mutations French breast cancer families. J. Med. Genet. 39,
identification of BRCA1 and BRCA2, which was the by the protein truncation test. Nature Genet. 10, 208–212 817–821 (2002).
result of years of collaborative effort from numerous (1995). 24. Puget, N. et al. Screening for germ-line rearrangements and
laboratories around the world. 16. Orita, M., Iwahana, H., Kanazawa, H., Hayashi, K. & Sekiya, T. regulatory mutations in BRCA1 led to the identification of
6. Collins, F. S. BRCA1 — lots of mutations, lots of dilemmas. Detection of polymorphisms of human DNA by gel four new deletions. Cancer Res. 59, 455–461 (1999).
N. Engl. J. Med. 334, 186–188 (1996). electrophoresis as single-strand conformation 25. Rohlfs, E. M. et al. An Alu-mediated 7.1 kb deletion of
7. Lerman, C. et al. BRCA1 testing in families with hereditary polymorphisms. Proc. Natl Acad. Sci. USA 86, 2766–2770 BRCA1 exons 8 and 9 in breast and ovarian cancer families
breast–ovarian cancer. A prospective study of patient decision (1989). that results in alternative splicing of exon 10. Genes
making and outcomes. JAMA 275, 1885–1892 (1996). 17. Borresen, A. L., Hovig, E. & Brogger, A. Detection of base Chromosom. Cancer 28, 300–307 (2000).
8. Rebbeck, T. R. et al. Prophylactic oophorectomy in carriers mutations in genomic DNA using denaturing gradient gel 26. Puget, N. et al. A 1-kb Alu-mediated germ-line deletion
of BRCA1 or BRCA2 mutations. N. Engl. J. Med. 346, electrophoresis (DGGE) followed by transfer and removing BRCA1 exon 17. Cancer Res. 57, 828–831
1616–1622 (2002). hybridization with gene-specific probes. Mutat. Res. 202, (1997).
9. Rebbeck, T. R. et al. Bilateral prophylactic mastectomy 77–83 (1988). 27. Scully, R. et al. Genetic analysis of BRCA1 function in a
reduces breast cancer risk in BRCA1 and BRCA2 mutation 18. Wagner, T. et al. Denaturing high-performance liquid defined tumor cell line. Mol. Cell 4, 1093–1099 (1999).
carriers: the PROSE Study Group. J. Clin. Oncol. 22, chromatography detects reliably BRCA1 and BRCA2 28. Hayes, F., Cayanan, C., Barilla, D. & Monteiro, A. N.
1055–1062 (2004). mutations. Genomics 62, 369–376 (1999). Functional assay for BRCA1: mutagenesis of the
10. Liede, A. & Narod, S. A. Hereditary breast and ovarian 19. Andrulis, I. L. et al. Comparison of DNA- and RNA-based COOH-terminal region reveals critical residues for
cancer in Asia: genetic epidemiology of BRCA1 and methods for detection of truncating BRCA1 mutations. transcription activation. Cancer Res. 60, 2411–2418
BRCA2. Hum. Mutat. 20, 413–424 (2002). Hum. Mutat. 20, 65–73 (2002). (2000).

674 | SEPTEMBER 2004 | VOLUME 4 www.nature.com/reviews/cancer

 REVIEWS

29. Humphrey, J. S. et al. Human BRCA1 inhibits growth in 59. Wang, Y. et al. BASC, a super complex of BRCA1- 87. Cameron, D. A., Ritchie, A. A. & Miller, W. R. The relative
yeast: potential use in diagnostic testing. Proc. Natl Acad. associated proteins involved in the recognition and repair of importance of proliferation and cell death in breast cancer
Sci. USA 94, 5820–5825 (1997). aberrant DNA structures. Genes Dev. 14, 927–939 (2000). growth and response to tamoxifen. Eur. J. Cancer 37,
30. Oddoux, C. et al. The carrier frequency of the BRCA2 60. Callebaut, I. & Mornon, J. P. From BRCA1 to RAP1: 1545–1553 (2001).
6174delT mutation among Ashkenazi Jewish individuals is a widespread BRCT module closely associated with DNA 88. Antoniou, A. et al. Average risks of breast and ovarian
approximately 1 percent. Nature Genet. 14, 188–190 (1996). repair. FEBS Lett. 400, 25–30 (1997). cancer associated with BRCA1 or BRCA2 mutations
31. Kauff, N. D. et al. Incidence of non-founder BRCA1 and 61. Xu, X. et al. Centrosome amplification and a defective detected in case series unselected for family history:
BRCA2 mutations in high risk Ashkenazi breast and ovarian G2–M cell cycle checkpoint induce genetic instability in a combined analysis of 22 studies. Am. J. Hum. Genet. 72,
cancer families. J. Med. Genet. 39, 611–614 (2002). BRCA1 exon 11 isoform-deficient cells. Mol. Cell 3, 389–395 1117–1130 (2003).
32. Phelan, C. M. et al. A low frequency of non-founder BRCA1 (1999). Provides robust risk estimates and should be required
mutations in Ashkenazi Jewish breast-ovarian cancer 62. Hakem, R. et al. The tumor suppressor gene Brca1 is reading for those who counsel women at risk for
families. Hum. Mutat. 20, 352–357 (2002). required for embryonic cellular proliferation in the mouse. developing BRCA1- and BRCA2-associated cancers.
33. Tulinius, H. et al. The effect of a single BRCA2 mutation on Cell 85, 1009–1023 (1996). 89. Adem, C. et al. Pathologic characteristics of breast
cancer in Iceland. J. Med. Genet. 39, 457–462 (2002). 63. Hakem, R., de la Pompa, J. L., Elia, A., Potter, J. & Mak, T. W. parenchyma in patients with hereditary breast carcinoma,
34. Gorski, B. et al. A high proportion of founder BRCA1 Partial rescue of Brca1 (5-6) early embryonic lethality by p53 including BRCA1 and BRCA2 mutation carriers. Cancer 97,
mutations in Polish breast cancer families. Int. J. Cancer or p21 null mutation. Nature Genet. 16, 298–302 (1997). 1–11 (2003).
110, 683–686 (2004). References 62 and 63 show that complete loss of 90. Hoogerbrugge, N. et al. High prevalence of premalignant
35. Risch, H. A. et al. Prevalence and penetrance of germline BRCA1 protein in the mouse inhibits development, but lesions in prophylactically removed breasts from women at
BRCA1 and BRCA2 mutations in a population series of 649 can be partially rescued by deletion of the genes that hereditary risk for breast cancer. J. Clin. Oncol. 21, 41–45
women with ovarian cancer. Am. J. Hum. Genet. 68, encode its effector proteins. (2003).
700–710 (2001). 64. Tirkkonen, M. et al. Distinct somatic genetic changes 91. Kauff, N. D. et al. Epithelial lesions in prophylactic
36. Thompson, D. & Easton, D. Variation in cancer risks, by associated with tumor progression in carriers of BRCA1 and mastectomy specimens from women with BRCA mutations.
mutation position, in BRCA2 mutation carriers. Am. J. Hum. BRCA2 germ-line mutations. Cancer Res. 57, 1222–1227 Cancer 97, 1601–1608 (2003).
Genet. 68, 410–419 (2001). (1997). 92. Mote, P. A. et al. Germ-line mutations in BRCA1 or BRCA2
37. Edwards, S. M. et al. Two percent of men with early-onset 65. Wessels, L. F. A. et al. Molecular classification of breast in the normal breast are associated with altered expression
prostate cancer harbor germline mutations in the BRCA2 carcinomas by comparative genomic hybridization: of estrogen-responsive proteins and the predominance of
gene. Am. J. Hum. Genet. 72, 1–12 (2003). a specific somatic genetic profile for BRCA1 tumors. progesterone receptor A. Genes Chromosom. Cancer 39,
38. Struewing, J. P. et al. The risk of cancer associated with Cancer Res. 62, 7110–7117 (2002). 236–248 (2004).
specific mutations of BRCA1 and BRCA2 among Ashkenazi 66. Gretarsdottir, S. et al. BRCA2 and p53 mutations in primary 93. Salazar, H. et al. Microscopic benign and invasive malignant
Jews. N. Engl. J. Med. 336, 1401–1408 (1997). breast cancer in relation to genetic instability. Cancer Res. neoplasms and a cancer-prone phenotype in prophylactic
39. Warner, E. et al. Prevalence and penetrance of BRCA1 and 58, 859–862 (1998). oophorectomies. J. Natl Cancer Inst. 88, 1810–1820 (1996).
BRCA2 gene mutations in unselected Ashkenazi Jewish 67. Bertwistle, D. & Ashworth, A. Functions of the BRCA1 and 94. Stratton, J. F., Buckley, C. H., Lowe, D. & Ponder, B. A.
women with breast cancer. J. Natl Cancer Inst. 91, BRCA2 genes. Curr. Opin. Genet. Dev. 8, 14–20 (1998). Comparison of prophylactic oophorectomy specimens from
1241–1247 (1999). 68. Wu, L. C. et al. Identification of a RING protein that can carriers and noncarriers of a BRCA1 or BRCA2 gene
40. Tonin, P. et al. Frequency of recurrent BRCA1 and BRCA2 interact in vivo with the BRCA1 gene product. Nature Genet. mutation. United Kingdom Coordinating Committee on
mutations in Ashkenazi Jewish breast cancer families. 14, 430–440 (1996). Cancer Research (UKCCCR) Familial Ovarian Cancer Study
Nature Med. 2, 1183–1196 (1996). 69. Hashizume, R. et al. The RING heterodimer BRCA1–BARD1 Group. J. Natl Cancer Inst. 91, 626–628 (1999).
41. Narod, S. A. Modifiers of risk of hereditary breast and is a ubiquitin ligase inactivated by a breast cancer-derived 95. Barakat, R. R. et al. Absence of premalignant histologic,
ovarian cancer. Nature Rev. Cancer 2, 113–123 (2002). mutation. J. Biol. Chem. 276, 14537–14540 (2001). molecular, or cell biologic alterations in prophylactic
42. Meijers-Heijboer, H. et al. Breast cancer after prophylactic 70. Ghimenti, C. et al. Germline mutations of the BRCA1- oophorectomy specimens from BRCA1 heterozygotes.
bilateral mastectomy in women with a BRCA1 or BRCA2 associated ring domain (BARD1) gene in breast and Cancer 89, 383–390 (2000).
mutation. N. Engl. J. Med. 345, 159–164 (2001). breast/ovarian families negative for BRCA1 and BRCA2 96. Skolnick, M. H. et al. Inheritance of proliferative breast
43. Kauff, N. D. et al. Risk-reducing salpingo-oophorectomy in alterations. Genes Chromosom. Cancer 33, 235–242 disease in breast cancer kindreds. Science 250, 1715–1720
women with a BRCA1 or BRCA2 mutation. N. Engl. J. Med. (2002). (1990).
346, 1609–1615 (2002). 71. Ishitobi, M. et al. Mutational analysis of BARD1 in familial breast 97. Colgan, T. J., Murphy, J., Cole, D. E., Narod, S. & Rosen, B.
44. Rebbeck, T. R. et al. Breast cancer risk after bilateral cancer patients in Japan. Cancer Lett. 200, 1–7 (2003). Occult carcinoma in prophylactic oophorectomy specimens:
prophylactic oophorectomy in BRCA1 mutation carriers. 72. Morris, J. R. & Solomon, E. BRCA1: BARD1 induces the prevalence and association with BRCA germline mutation
J. Natl Cancer Inst. 91, 1475–1479 (1999). formation of conjugated ubiquitin structures, dependent on status. Am. J. Surg. Pathol. 25, 1283–1289 (2001).
45. Scully, R. & Livingston, D. M. In search of the tumour- K6 of ubiquitin, in cells during DNA replication and repair. 98. Foray, N. et al. γ-rays-induced death of human cells carrying
suppressor functions of BRCA1 and BRCA2. Nature 408, Hum. Mol. Genet. 13, 807–817 (2004). mutations of BRCA1 or BRCA2. Oncogene 18, 7334–7342
429–432 (2000). 73. Bochar, D. A. et al. BRCA1 is associated with a human (1999).
46. Venkitaraman, A. R. Cancer susceptibility and the functions SWI/SNF-related complex: linking chromatin remodeling to 99. Baldeyron, C. et al. A single mutated BRCA1 allele leads to
of BRCA1 and BRCA2. Cell 108, 171–182 (2002). breast cancer. Cell 102, 257–265 (2000). impaired fidelity of double strand break end-joining.
47. Venkitaraman, A. R. Tracing the network connecting BRCA 74. Versteege, I. et al. Truncating mutations of hSNF5/INI1 in Oncogene 21, 1401–1410 (2002).
and Fanconi anaemia proteins. Nature Rev. Cancer 4, aggressive paediatric cancer. Nature 394, 203–206 100. Coupier, I. et al. Fidelity of DNA double-strand break repair in
266–276 (2004). (1998). heterozygous cell lines harbouring BRCA1 missense
48. Scully, R. et al. Association of BRCA1 with Rad51 in mitotic 75. Yarden, R. I. & Brody, L. C. BRCA1 interacts with mutations. Oncogene 23, 914–919 (2004).
and meiotic cells. Cell 88, 265–275 (1997). components of the histone deacetylase complex. Proc. Natl 101. Rothfuss, A. et al. Induced micronucleus frequencies in
49. Sharan, S. K. et al. Embryonic lethality and radiation Acad. Sci. USA 96, 4983–4988 (1999). peripheral lymphocytes as a screening test for carriers of a
hypersensitivity mediated by Rad51 in mice lacking Brca2. 76. Cantor, S. B. et al. BACH1, a novel helicase-like protein, BRCA1 mutation in breast cancer families. Cancer Res. 60,
Nature 386, 804–810 (1997). interacts directly with BRCA1 and contributes to its DNA 390–394 (2000).
50. Mizuta, R. et al. RAB22 and RAB163/mouse BRCA2- repair function. Cell 105, 149–160 (2001). 102. Baria, K. et al. Correspondence re: A. Rothfuss et al.
proteins that specifically interact with the rad51 protein. 77. Hughes-Davies, L. et al. EMSY links the BRCA2 pathway to Induced micronucleus frequencies in peripheral blood
Proc. Natl Acad. Sci. USA 94, 6927–6932 (1997). sporadic breast and ovarian cancer. Cell 115, 523–535 lymphocytes as a screening test for carriers of a BRCA1
51. Wong, A. K., Pero, R., Ormonde, P. A., Tavtigian, S. V. & (2003). mutation in breast cancer families. In Cancer Research.
Bartel, P. L. RAD51 interacts with the evolutionarily conserved 78. Tischkowitz, M. D. & Hodgson, S. V. Fanconi anaemia. 60, 390–394, 2000. Cancer Res. 61, 5948–5949 (2001).
BRC motifs in the human breast cancer susceptibility gene J. Med. Genet. 40, 1–10 (2003). 103. Cornelis, R. S. et al. High allele loss rates at 17q12–q21 in
brca2. J. Biol. Chem. 272, 31941–31944 (1997). 79. Howlett, N. G. et al. Biallelic inactivation of BRCA2 in breast and ovarian tumors from BRCA1-linked families. The
References 48–51 show that interactions between Fanconi anemia. Science 297, 606–609 (2002). Breast Cancer Linkage Consortium. Genes Chromosom.
BRCA1, BRCA2 and RAD51 are crucial elements of the 80. Offit, K. et al. Shared genetic susceptibility to breast cancer, Cancer 13, 203–210 (1995).
coordinated response to DNA damage, and illustrate the brain tumors, and Fanconi anemia. J. Natl Cancer Inst. 95, 104. Smith, S. A., Easton, D. F., Evans, D. G. & Ponder, B. A.
consequences of the disruption of these relationships. 1548–1551 (2003). Allele losses in the region 17q12–21 in familial breast and
52. Moynahan, M. E., Cui, T. Y. & Jasin, M. Homology-directed References 79 and 80 showed that homozygous ovarian cancer involve the wild-type chromosome. Nature
DNA repair, mitomycin-c resistance, and chromosome truncating mutations in BRCA2 do not necessarily Genet. 2, 128–131 (1992).
stability is restored with correction of a Brca1 mutation. result in embryonic lethality, but do lead to a severe 105. Futreal, P. A. et al. BRCA1 mutations in primary breast and
Cancer Res. 61, 4842–4850 (2001). form of childhood cancer. Such an effect has not been ovarian carcinomas. Science 266, 120–122 (1994).
53. Tassone, P. et al. BRCA1 expression modulates seen for BRCA1 mutations 106. Sorlie, T., Andersen, T. I., Bukholm, I. & Borresen-Dale, A. L.
chemosensitivity of BRCA1-defective HCC1937 human 81. Monteiro, A. N. BRCA1: the enigma of tissue-specific tumor Mutation screening of BRCA1 using PTT and LOH analysis
breast cancer cells. Br. J. Cancer 88, 1285–1291 (2003). development. Trends Genet. 19, 312–315 (2003). at 17q21 in breast carcinomas from familial and non-familial
54. Yuan, S. S. F. et al. BRCA2 is required for ionizing radiation- 82. Elledge, S. J. & Amon, A. The BRCA1 suppressor hypothesis: cases. Breast Cancer Res. Treat. 48, 259–264 (1998).
induced assembly of rad51 complex in vivo. Cancer Res. an explanation for the tissue-specific tumor development in 107. Merajver, S. D. et al. Somatic mutations in the BRCA1 gene in
59, 3547–3551 (1999). BRCA1 patients. Cancer Cell 1, 129–132 (2002). sporadic ovarian tumours. Nature Genet. 9, 439–443 (1995).
55. Patel, K. J. et al. Involvement of Brca2 in DNA repair. Mol. 83. Chappuis, P. O., Nethercot, V. & Foulkes, W. D. Clinico- 108. Hosking, L. et al. A somatic BRCA1 mutation in an ovarian
Cell 1, 347–357 (1998). pathological characteristics of BRCA1- and BRCA2-related tumour. Nature Genet. 9, 343–344 (1995).
56. Zhong, Q. et al. Association of BRCA1 with the breast cancer. Semin. Surg. Oncol. 18, 287–295 (2000). 109. Haber, D. Roads leading to breast cancer. N. Engl. J. Med.
hRad50–hMre11–p95 complex and the DNA damage 84. Zeps, N., Bentel, J. M., Papadimitriou, J. M., D’Antuono, M. F. 343, 1566–1568 (2000).
response. Science 285, 747–750 (1999). & Dawkins, H. J. Estrogen receptor-negative epithelial cells 110. Narod, S. Roads to breast cancer. N. Engl. J. Med. 344,
57. Le Page, F. et al. BRCA1 and BRCA2 are necessary for in mouse mammary gland development and growth. 937 (2001).
the transcription-coupled repair of the oxidative 8- Differentiation 62, 221–226 (1998). 111. Foulkes, W. D. BRCA1 functions as a breast stem cell
oxoguanine lesion in human cells. Cancer Res. 60, 85. Gompel, A. et al. Hormonal regulation of apoptosis in breast regulator. J. Med. Genet. 41, 1–5 (2004).
5548–5552 (2000). cells and tissues. Steroids 65, 593–598 (2000). 112. Greenblatt, M. S., Bennett, W. P., Hollstein, M. & Harris, C. C.
58. Hartman, A. R., Ford, J. M. BRCA1 induces DNA damage 86. Somai, S. et al. Antiestrogens are pro-apoptotic in normal Mutations in the p53 tumor suppressor gene: clues to
recognition factors and enhances nucleotide excision repair. human breast epithelial cells. Int. J. Cancer 105, 607–612 cancer etiology and molecular pathogenesis. Cancer Res.
Nature Genet. 32, 180–184 (2002). (2003). 54, 4855–4878 (1994).

NATURE REVIEWS | C ANCER VOLUME 4 | SEPTEMBER 2004 | 6 7 5

REVIEWS

113. Magdinier, F. et al. Regional methylation of the 5′ end CpG 137. Moller, P. et al. Survival in prospectively ascertained familial 163. Stankovic, T. et al. ATM mutations and phenotypes in
island of BRCA1 is associated with reduced gene expression breast cancer: analysis of a series stratified by tumour ataxia-telangiectasia families in the British Isles: expression
in human somatic cells. FASEB J. 14, 1585–1594 (2000). characteristics, BRCA mutations and oophorectomy. Int. J. of mutant ATM and the risk of leukemia, lymphoma, and
114. Catteau, A., Harris, W. H., Xu, C. F. & Solomon, E. Cancer 101, 555–559 (2002). breast cancer. Am. J. Hum. Genet. 62, 334–345 (1998).
Methylation of the BRCA1 promoter region in sporadic 138. Eerola, H. et al. Survival of breast cancer patients in BRCA1, 164. Szabo, C. I. et al. Are ATM mutations 7271T>G and
breast and ovarian cancer: correlation with disease BRCA2, and non-BRCA1/2 breast cancer families: a relative IVS10-6T>G really high-risk breast cancer-susceptibility
characteristics. Oncogene 18, 1957–1965 (1999). survival analysis from Finland. Int. J. Cancer 93, 368–372 alleles? Cancer Res. 64, 840–843 (2004).
115. Esteller, M. et al. Promoter hypermethylation and BRCA1 (2001). 165. Wu, X., Webster, S. R. & Chen, J. Characterization of tumor-
inactivation in sporadic breast and ovarian tumors. J. Natl 139. Metcalfe, K. et al. Contralateral breast cancer in BRCA1 and associated Chk2 mutations. J. Biol. Chem. 276, 2971–2974
Cancer Inst. 92, 564–569 (2000). BRCA2 mutation carriers. J. Clin. Oncol. 22, 2328–2335 (2001).
116. Thompson, M. E., Jensen, R. A., Obermiller, P. S., Page, D. L. (2004). 166. Meijers-Heijboer, H. et al. Low-penetrance susceptibility to
& Holt, J. T. Decreased expression of BRCA1 accelerates 140. Narod, S. A. et al. Tamoxifen and risk of contralateral breast breast cancer due to CHEK2*1100delC in noncarriers of
growth and is often present during sporadic breast cancer cancer in BRCA1 and BRCA2 mutation carriers: a case- BRCA1 or BRCA2 mutations. Nature Genet. 31, 55–59 (2002).
progression. Nature Genet. 9, 444–450 (1995). control study. Hereditary Breast Cancer Clinical Study 167. Oldenburg, R. A. et al. The CHEK2*1100delC variant acts as
117. Magdinier, F., Ribieras, S., Lenoir, G. M., Frappart, L. & Group. Lancet 356, 1876–1881 (2000). a breast cancer risk modifier in non-BRCA1/BRCA2
Dante, R. Down-regulation of BRCA1 in human sporadic 141. Hedenfalk, I. et al. Gene-expression profiles in hereditary multiple-case families. Cancer Res. 63, 8153–8157 (2003).
breast cancer; analysis of DNA methylation patterns of the breast cancer. N. Engl. J. Med. 344, 539–548 (2001). 168. Vahteristo, P. et al. A CHEK2 genetic variant contributing to
putative promoter region. Oncogene 17, 3169–3176 (1998). 142. Moslehi, R. et al. BRCA1 and BRCA2 mutation analysis of a substantial fraction of familial breast cancer. Am. J. Hum.
118. Seery, L. T. et al. BRCA1 expression levels predict distant 208 Ashkenazi Jewish women with ovarian cancer. Am. J. Genet. 71, 432–438 (2002).
metastasis of sporadic breast cancers. Int. J. Cancer 84, Hum. Genet. 66, 1259–1272 (2000). References 166–168 describe the relation between
258–262 (1999). 143. Boyd, J. et al. Clinicopathologic features of BRCA-linked CHK2 and familial breast cancer and provide important
119. Lambie, H. et al. Prognostic significance of BRCA1 and sporadic ovarian cancer. JAMA 283, 2260–2265 examples of how breast cancer predisposition is likely
expression in sporadic breast carcinomas. J. Pathol. 200, (2000). to be caused by polygenic factors.
207–213 (2003). 144. Gotlieb, W. H. et al. Rates of Jewish ancestral mutations in 169. Offit, K. et al. Frequency of CHEK2*1100delC in New York
120. Lakhani, S. R. et al. Multifactorial analysis of differences BRCA1 and BRCA2 in borderline ovarian tumors. J. Natl breast cancer cases and controls. BMC Med. Genet. 4, 1
between sporadic breast cancers and cancers involving Cancer Inst. 90, 995–1000 (1998). (2003).
BRCA1 and BRCA2 mutations. J. Natl Cancer Inst. 90, 145. Jazaeri, A. A. et al. Gene expression profiles of BRCA1- 170. Matsuoka, S. et al. Ataxia telangiectasia-mutated
1138–1145 (1998). linked, BRCA2-linked, and sporadic ovarian cancers. J. Natl phosphorylates Chk2 in vivo and in vitro. Proc. Natl Acad.
121. Jacquemier, J., Lidereau, R., Birnbaum, D., Eisinger, F. & Cancer Inst. 94, 990–1000 (2002). Sci. USA 97, 10389–10394 (2000).
Sobol, H. Assessing the risk of BRCA1-associated breast 146. Peto, J. Breast cancer susceptibility — a new look at an old 171. Chaturvedi, P. et al. Mammalian Chk2 is a downstream
cancer using individual morphological criteria. Histopathol. model. Cancer Cell 1, 411–412 (2002). effector of the ATM-dependent DNA damage checkpoint
38, 378–379 (2001). 147. Sobol, H., Birnbaum, D. & Eisinger, F. Evidence for a third pathway. Oncogene 18, 4047–4054 (1999).
122. Quenneville, L. A. et al. HER-2/neu status and tumor breast-cancer susceptibility gene. Lancet 344, 1151–1152 172. Ahn, J. Y., Schwarz, J. K., Piwnica-Worms, H. &
morphology of invasive breast carcinomas in Ashkenazi (1994). Canman, C. E. Threonine 68 phosphorylation by ataxia
women with known BRCA1 mutation status in the Ontario 148. Seitz, S. et al. Strong indication for a breast cancer telangiectasia mutated is required for efficient activation of
Familial Breast Cancer Registry. Cancer 95, 2068–2075 susceptibility gene on chromosome 8p12–p22: Chk2 in response to ionizing radiation. Cancer Res. 60,
(2002). linkage analysis in German breast cancer families. 5934–5936 (2000).
123. Perou, C. M. et al. Molecular portraits of human breast Oncogene 14, 741–743 (1997). 173. Falck, J., Mailand, N., Syljuasen, R. G., Bartek, J. & Lukas, J.
tumours. Nature 406, 747–752 (2000). 149. Kainu, T. et al. Somatic deletions in hereditary breast The ATM–Chk2–Cdc25A checkpoint pathway guards against
124. Van’t Veer, L. J. et al. Gene expression profiling predicts cancers implicate 13q21 as a putative novel breast cancer radioresistant DNA synthesis. Nature 410, 842–847 (2001).
clinical outcome of breast cancer. Nature 415, 530–536 susceptibility locus. Proc. Natl Acad. Sci. USA 97, 174. Chehab, N. H., Malikzay, A., Appel, M. & Halazonetis, T. D.
(2002). 9603–9608 (2000). Chk2/hCds1 functions as a DNA damage checkpoint in G1
125. Gruvberger, S. et al. Estrogen receptor status in breast 150. Rahman, N. et al. Absence of evidence for a familial breast by stabilizing p53. Genes Dev. 14, 278–288 (2000).
cancer is associated with remarkably distinct gene cancer susceptibility gene at chromosome 8p12–p22. 175. Shieh, S. Y., Ahn, J., Tamai, K., Taya, Y. & Prives, C. The
expression patterns. Cancer Res. 61, 5979–5984 Oncogene 19, 4170–4173 (2000). human homologs of checkpoint kinases Chk1 and Cds1
(2001). 151. Thompson, D. et al. Evaluation of linkage of breast cancer (Chk2) phosphorylate p53 at multiple DNA damage-
126. Sorlie, T. et al. Repeated observation of breast tumor to the putative BRCA3 locus on chromosome 13q21 in 128 inducible sites. Genes Dev. 14, 289–300 (2000).
subtypes in independent gene expression data sets. multiple case families from the Breast Cancer Linkage 176. Lee, J. S., Collins, K. M., Brown, A. L., Lee, C. H. &
Proc. Natl Acad. Sci. USA 100, 8418–8423 (2003). Consortium. Proc. Natl Acad. Sci. USA 99, 827–831 Chung, J. H. hCds1-mediated phosphorylation of BRCA1
127. Foulkes, W. D. et al. Germline BRCA1 mutations and a basal (2002). regulates the DNA damage response. Nature 404, 201–204
epithelial phenotype in breast cancer. J. Natl Cancer Inst. 152. Cui, J. et al. After BRCA1 and BRCA2 — what next? (2000).
95, 1482–1485 (2003). Multifactorial segregation analyses of three-generation, 177. Bell, D. W. et al. Heterozygous germ line hCHK2
128. Korsching, E. et al. Cytogenetic alterations and cytokeratin population-based Australian families affected by female mutations in Li–Fraumeni syndrome. Science 286,
expression patterns in breast cancer: integrating a new breast cancer. Am. J. Hum. Genet. 68, 420–431 (2001). 2528–2531 (1999).
model of breast differentiation into cytogenetic pathways 153. Liede, A. et al. Contribution of BRCA1 and BRCA2 178. Jasin, M. Homologous repair of DNA damage and
of breast carcinogenesis. Lab. Invest. 82, 1525–1533 mutations to breast and ovarian cancer in Pakistan. Am. J. tumorigenesis: the BRCA connection. Oncogene 21,
(2002). Hum. Genet. 71, 595–606 (2002). 8981–8993 (2002).
129. Bocker, W. et al. Common adult stem cells in the human 154. Lakhani, S. R. et al. The pathology of familial breast cancer: 179. Deng, C. X. & Brodie, S. G., Roles of BRCA1 and its
breast give rise to glandular and myoepithelial cell lineages: histological features of cancers in families not attributable to interacting proteins. Bioessays 22, 728–737 (2000).
a new cell biological concept. Lab. Invest. 82, 737–745 mutations in BRCA1 or BRCA2. Clin. Cancer Res. 6,
(2002). 782–789 (2000). Acknowledgements
References 126–129 report the existence of a basal 155. Hedenfalk, I. et al. Molecular classification of familial non- Work in W.D.F.’s laboratory is funded by the US Army, the Susan G.
subtype of breast cancer, and references 126 and 127 BRCA1/BRCA2 breast cancer. Proc. Natl Acad. Sci. USA Komen Breast Cancer Foundation and the Canadian Breast
show that this tumour phenotype is over-represented 100, 2532–2537 (2003). Cancer Alliance.
in carriers of BRCA1 mutations compared with other 156. Pharoah, P. D. et al. Polygenic susceptibility to breast cancer and
types of breast cancer. implications for prevention. Nature Genet. 31, 33–36 (2002). Competing interests statement
130. Robson, M. E., Boyd, J., Borgen, P. I. & Cody, H. S. Hereditary Looks towards the future of breast cancer genetics — The authors declare no competing financial interests.
breast cancer. Curr. Probl. Surg. 38, 387–480 (2001). there is probably no single BRCA3 gene, but rather
131. Evans, D. G. & Howell, A. Are BRCA1- and BRCA2–related many low-penetrance, low-frequency genes are likely to
breast cancers associated with increased mortality? Breast underlie the remaining cases of familial breast cancer. Online links
Cancer Res. 6, E7 (2004). 157. Swift, M. & Chase, C. Cancer and cardiac deaths in
132. Robson, M. E. et al. A combined analysis of outcome obligatory ataxia-telangiectasia heterozygotes. Lancet 1, DATABASES
following breast cancer: differences in survival based on 1049–1050 (1983). The following terms in this article are linked online to:
BRCA1/BRCA2 mutation status and administration of 158. Easton, D. F. Cancer risks in A-T heterozygotes. Int. J. Cancer.gov: http://www.cancer.gov
adjuvant treatment. Breast Cancer Res. 6, R8–R17 (2004). Radiat. Biol. 66, S177–S182 (1994). breast cancer | endometrial cancer | ovarian cancer | prostate
133. Quinn, J. E. et al. BRCA1 functions as a differential 159. Chenevix-Trench, G. et al. Dominant negative ATM cancer | Wilms’ tumour
modulator of chemotherapy-induced apoptosis. Cancer mutations in breast cancer families. J. Natl Cancer Inst. 94, Entrez Gene:
Res. 63, 6221–6228 (2003). 205–215 (2002). http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene
134. Chappuis, P. O. et al. A significant response to neoadjuvant 160. Fitzgerald, M. G. et al. Heterozygous ATM mutations do not ATM | BARD1 | BRCA1 | BRCA2 | CHK2 | cyclin D1 | cyclin E |
chemotherapy in BRCA1/2 related breast cancer. J. Med. contribute to early onset of breast cancer. Nature Genet. 15, EMSY | ER | ERBB2 | KIP1 | p53 | RAD51 | WAF1
Genet. 39, 608–610 (2002). 307–310 (1997). OMIM: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=OMIM
135. Foulkes, W. D. et al. Disruption of the expected positive 161. Scott, S. P. et al. Missense mutations but not allelic variants Fanconi anaemia
correlation between breast tumor size and lymph node alter the function of ATM by dominant interference in
status in BRCA1-related breast carcinoma. Cancer 98, patients with breast cancer. Proc. Natl Acad. Sci. USA 99, FURTHER INFORMATION
1569–1577 (2003). 925–930 (2002). The Breast Cancer Information Core:
136. Goffin, J. R. et al. Impact of germline BRCA1 mutations and 162. Gatti, R. A., Tward, A. & Concannon, P. Cancer risk in ATM http://research.nhgri.nih.gov/bic/
overexpression of p53 on prognosis and response to heterozygotes: a model of phenotypic and mechanistic The Breast Cancer Linkage Consortium:
treatment following breast carcinoma: 10-year follow up differences between missense and truncating mutations. http://www.humgen.nl/lab-devilee/bclchome.htm
data. Cancer 97, 527–536 (2003). Mol. Genet. Metab. 68, 419–423 (1999). Access to this links box is available online.

676 | SEPTEMBER 2004 | VOLUME 4 www.nature.com/reviews/cancer