BRCA1: a review of structure and putative functions.

BRCA1 is a complex gene implicated in familial breast and ovarian cancer. Although it is almost certainly a tumour suppressor, it is also essential for the normal growth and development of embryonic cells. BRCA1 is probably involved in DNA damage and repair, in cell cycle regulation, and in differentiation of cells. It remains to be established whether all these functions are subserved by single mechanism or pathway. Since the cloning of BRCA1 in 1994, much has been learned about the function of the gene. However, a great deal more still has to be uncovered. The size of the protein coded by the BRCA1 gene and the variety of transcripts argues for a complexity of function and regulation that will provide intellectual and technical challenges for years to come.

, and carriers have a slightly increased rate of both prostate and colon cancers (Ford et al. 1994).Male breast cancer is rarely a feature of these families.Breast and ovarian cancer susceptibility is inherited as an autosomal dominant trait, with high penetrance in affected families (Newman et ai., 1988).The lifetime risk for mutation carriers is generally quoted as 80-90% for breast cancer and nearly 50% for ovarian cancer (Ford et ai., 1994) though more recent data suggest the figures may be somewhat lower (Levy-Lahad et ai., 1997).The majority of disease causing mutations identified to date result in protein truncation, with only a few missense mutations that lead to amino acid substitutions (Couch et ai., 1996).These missense mutations cluster either to the amino terminal RING finger domain or the carboxyl terminal region (the BRCT domain discussed below), supporting the importance of these two structural elements in the function of the protein .

MOLECULAR ORGANISATION AND REGULATION
The promoter region of BRCA J is complex, with evidence for genetic duplication that has generated a BRCA J pseudogene and promoter shared with the NBR2 gene (Barker et al. , 1996).Two separate promoters exist for BRCA 1, generating transcripts that differ in the first exon (part of the un translated region) (Xu et al., 1995).Both promoters contain oestrogen response elements, suggesting that BRCA I may be involved in hormone signalling pathways (Xu et al. , 1997).BRCA J expression is responsive to oestrogen, although evidence suggests that this may be an indirect and nonspecific effect, related to changes in the cell cycle rather than to interactions between BRCA I control pathways and oestrogen (Spillman and Bowcock, 1996;  Marks et al., 1997 ; Gudas et ai., 1995).Variable methylation has also been proposed as a control mechanism for BRCAI in sporadic cancers (Dobrovic and Simpfendorfer, 1997), as have various post-translational modifications, such as proteolytic cleavage (Jensen el ai. , 1996a), phosphorylation (Chen et al., 1995;Chen et al., 1996b;Zhang el al. , 1997 ;Scully et al. , 1997b) , and differential sub-cellular localization (Chen el aI., 1995;Jensen el al., I 996a;Thakur et al., 1997 ;Wilson et al., 1997 ;Wang et al., 1997).Any of these may be operating independently , or several may act in concert to regulate BRCA I function.
A complex pattern of splicing of the mRNA was noted when BRCA J was first cloned (Miki et al. , 1994), and has been confirmed repeatedly (Xu et al.,1995;Thakur et al .. 1997;Lu et aI., 1996;Wilson et al., 1997;Wang et ai. , 1997).The splice variants have yet to be assigned any biological function , although circumstantial evidence suggests they are important.It has been noted that overexpression of the full length BRCA I protein (i .e., the protein coded by the entire open reading frame) results in cellular toxicity , but a splice variant protein lacking most of exon 11 (BRCA I-t.II b) is not toxic (Wilson et al., 1997).Furthermore, while the expression level of the full length protein was found to be similar in all cell lines tested, BRCA1-t.II b levels were decreased in the most aggressive cancer cell lines, suggesting that splice variants may modulate tumorigenicity.However, it is uncertain whether the decrease in expression of the splice variant is a cause or an effect of the increased tumorigenicity.Splice variants , normally cytosol ic, become nuclear (presumably through an alternative nuclear localization pathway) when cells are serum starved (Wang et ai. , 1997).Finally, the sheer number of reported splice variants, encompassing both the coding region and the 5 ' and 3' untranslated regions (Mikietal., 1994;Xuetal., 1995;Thakuretal., 1997;Luetal., 1996;Wilsonetal., 1997;Wang et al., 1997), suggests that they must make significant contributions to overall BRCA I function.
BRCAI is not mutated in sporadic breast tumours and only rarely in sporadic ovarian cancers (Takahashi et aI., 1996;Foster et aI., 1996).Yet evidence for the involvement of BRCA 1 in sporadic tumours is convincing (Cropp et al., 1994;Borg et al., 1994;Thompson et aI., 1995;Holt et al., 1996;Dobrovic and Simpfendorfer, 1997).It is possible that DNA lesions, not detectable by standard mutation screening, may be responsible for the disruption of function of BRCA I in sporadic tumours.Possible mechanisms include variable methylation (Dobrovic and Simpfendorfer, 1997), large deletion events (Puget et aI., 1997;Petrijbosch et aI., 1997), and downregulated expression (Thompson et al. , 1995).However, it is also possible that BRCA I does not participate in the formation of sporadic breast cancers at all.BRCA I has been found to have a role in embryonic development (at least in mice) (Gowen et aI., 1996;Haken et aI., 1996;Ludwig et aI., 1997) and in differentiation (Marquis et al., 1995;Lane et aI., 1995;Rajan et al., 1996), supporting a possible role in the very early initiation of tumorigenesis, for example, during development ill utero.
Several different domains have been recognised in BRCA I (Fig I ).A coding sequence for a RING finger zinc binding domain was identified in 5' end of the transcript (Miki et aI., 1994), as welI as a carboxyl terminal BRCT (BRCAI ~arboxyllerminal) domain of unknown function (Koonin et aI., 1996).Both regions share the highest level of conservation between species found anywhere in the coding sequence (Abel et al., 1995;Sharan et aI., 1995;Chen et aI., 1996), implying some functional importance.Overall, BRCA I shows very poor inter-species conservation.Other potentially functional regions identified in the BRCAI coding sequence include two nuclear localization sequences (Chen et al., 1996), a granin consensus sequence (Jensen et al., 1996a), and a domain that interacts with Rad51 (Scully et al., 1997c).

BRCAI FUNCTION
Evidence to support the view that BRCA I is a tumour suppressor was apparent even before the gene had been cloned.Loss of heterozygosity has been observed in both sporadic and familial tumours (Cropp et al., 1994;Borg et al., 1994;Cropp et aI., 1993;Lindblom et al., 1993;Foulkes et aI., 1993;Kerangueven et al., 1997;Sato et al., 1991;Futreal et al., 1992;Beckmann et al., 1996;Munn et al., 1996;Papp et aI., 1996) and, when linkage allowed determination of the disease causing alIele, LOH was always found to affect the wild type allele in the familial tumours (Neuhausen and Marshall, 1994;Smith et al., 1992).More concrete evidence for the importance of BRCA I in preventing tumour growth emerged from the finding that overexpression of BRCA 1 in cells in culture is lethal (Wilson et al., 1997) while antisense inhibition of BRCA 1 is reported to cause an increase in growth rates specific for breast and ovarian cells (Thompson et al., 1995), although this has yet to be substantiated by other groups.Antisense inhibition of mouse NIH-3T3 fibroblasts increased growth rate, soft agar growth, and tumorigenicity (Rao et aI., 1996) further supporting a role for BRCA 1 in tumour suppression.
The exact biochemical and cellular functions of BRCA I remain elusive, as do somc basic characteristics, such as the sub-cellular localization.The weight of evidence suggests that BRCAI is localized to the nucleus (Chen el af .. 1995;Aprelikova et af. , 1996;Chen et aI., 1996a;Chen et al., 1996b;Scully et aI., 1996;Shao et al., 1996;Thomas et al., 1996;Scully et al. , 1997c;Wilson et aI., 1997), with a nuclear punctate pattern pattern apparent during Sand M phases (Chen et al., 1996b;Scully et al., 1996;Scully et al. 1997c).However, there is also a claim that BRCAI may be a granin, initially targeted extracellularly and proteolytic ally cleaved to activate its function (Jensen et al., 1996a).These extracellular protein fragments may become localized to the nucleus; however, this would be by a secondary mechanism involving transport, following binding to membrane receptors or other carrier molecules (Jensen et ai.,I 996b ).BRCA I probably subserves a complex set of functions in the cell that mayor may not be linked to a common biochemical mechanism.Targeted disruption of Brcal in mice results in death in utero, with gross anatomical abnormalities and disruption of cell growth in Brcal null animals (Gowen et al., 1996;Haken et ai., 1996;Ludwig et af., 1997).The stage at which death occurs has been reported to be either before E7.5 (Haken et al., 1996) or as late as E 13 (Gowen el af., 1996), suggesting that additional factors (such as the genetic background of the experimental strain or the precise position of the truncating mutation) may modify the effect of Brca I knockout.Mouse models do not shed much light on the effects of BRCA 1 in normal adult mammary tissue; mice heterozygous for a Brca I mutation are viable and completely normal, without apparent susceptibility to any cancers.Although normal Brcal function is essential for early development in the mouse, it appears to be less so in humans.At least one developmentally normal individual (though with early onset breast cancer) has been reported who has inherited identical mutations (2800deIAA) from both parents.(Boyd et ai., 1995).The data thus suggest that BRCA I may serve functions both in development and in cancer progression.The differences between mice and humans highlight the difficulty of extrapolating from animal models to establish the function of specific genes; human BRCA I mutation heterozygotes have greatly increased risk for breast cancer, while heterozygous mice apparently do not.These differences may have a trivial explanation (for example, mice may not live long enough to exhibit Brca 1associated mammary tumours) or they may reflect fundamental differences in the biology of the two species.

BRCAl in DNA repair
The BRCT domain has been recognised as common to BRCA I, a p53 binding protein, 53BPI and a number of other proteins (Koonin et al., 1996).Further characterization of the domain led to the identification of its presence in a considerable number of other genes, both characterized and uncharacterized (Bork et al., 1997;Callebaut and Momon, 1997).In most instances the BRCT domain has had no function directly attributed to it but it appears to participate in either DNA binding (for example in RFC 1 and RAP I (Bork et aI., 1997» or in protein-protein interactions (for example, between XRCC4 and DNA Ligase IV (Critchlow et al., 1997), PARP, DNA Ligase III/XRCC 1 (Bork et al., 1997), 53BP I (Koonin et al. 1996), and BRCA I /RNA Polymerase II holoenzyme (Scully et at., 1997a».BRCT domains are thought not to participate directly in the enzymatic functions of the proteins that bear them, however, it has been suggested that enzyme activity may be controlled by the domain (for example in RAPI (Bork et aI., 1997» or that it may modify the specificity of the reaction (for example, in the TdT' s) (Bork et al., 1997).Such interactions have been proposed as a mechanism for recruitment of members of the DN A damage repair family to the site of repair, through BRCT domain interactions; for example XRCC4 recruitment of a suitable DNA ligase to DNA PK repair complexes (Critchlow et al., 1997).The BRCT domain may be involved directly in DNA repair and metabolism since a high proportion of the characterized proteins that contain it participate in such functions (Callebaut and Momon, 1997;Bork et aI., 1997;Critchlow et al., 1997).
BRCA I has been shown to have a specific interaction with Rad51 protein mediated through aBRCA I exon II domain (Scully etal., 1997c).Rad51, in mammals, is involved with the homologous pairing of DNA during the process of recombination and recombinational repair (Sung and Robberson, 1995;Baumann et al., 1996), supporting the proposition that BRCAI is involved in DNA repair processes.The interaction of Rad51 and BRCAI was first identified through the observation of similar staining patterns for the two proteins (discrete nuclear punctate staining during Sand M phases of the cell cycle) (Scully et al., 1996;Scully et aI., 1997c).Although the interaction appears specific, its functional basis has not been proved.There are reports that BRCA I undergoes dramatic changes in phosphorylation and subnuclear localization in response to DNA damage (Scully et aI., 1997b).The same changes in localization apply to Rad51, suggesting that they may have a physiological interaction that is triggered by DNA damage.This interaction with Rad51 , however, cannot be the sole function for BRCA 1; BRCA 1 is highly expressed during spermiogenesis (Zabludoff et aI., 1996) when Rad51 expression is silenced.(Yamamoto et aI., 1996).
Both Rad5 / and BrcaJ nullizygous mice undergo severe disruption of embryogenesis and die in utero, suggesting that similar pathways may be implicated.Brcal and Rad51 nullizygotes share other early developmental characteristics, including the inability of embryonic stem cells to survive in tissue culture and partial rescue of the phenotype in mice with p53 or p21 WallIC ipl null background (Lim and Hasty, 1996; Ludwig et at., 1997;  Hakem et al., 1997).However, the phenotype is not precisely the same for both genes; BRCAJ loss of function results in death near E7.5 (Haken et al., 1996;Ludwig et aI., 1997) or even later at Ell (Gowen et aI., 1996), while Rad51 nullizygosity results in death as early as the morula/blastocyst stage (Tsuzuki et aI., 1996;Lim and Hasty, 1996).For BRCA I, this apparent block on cell growth may apply only to fetal cells, as disruption of expression in adult cells is reported to increase cell growth (Thompson et aI., 1995) and tumorigenicity (Rao et al., 1996).
The final evidence supporting the participation ofBRCAI in DNA damage response comes from interactions observed between BRCA 1 and p53 pathways.As mentioned above, a p53 null background partially rescues the Brca I-null lethal phenotype, extending the time of death from E7.5 up to E II (Ludwig et al. , 1997; Hakem et al., 1997).This suggests that BRCA I may be responsible for DNA damage sensing and repair, so that and when its function is lost, DNA damage accumulates and p53 responsive pathways cause cellular arrest and apoptosis .Loss of p53 function in these cells then allows the cells to continue dividing with unrepaired damage until genomic stability is sufficiently compromised to result in cell death.Although this link between the pathways is tenuous , clinical observations support an association between BRCA I associated tumorigenesis and p53.In a recent study, p53 was mutated in all the breast tumours of BRCA I mutation carriers (Crook et aI., 1997), suggesting that loss of p53 function may be an absolute requirement for the progression of cancer.Furthermore, the spectrum of mutations was not as expected for breast cancers; mutations were mainly in exon 5 of p53 (the amino end of the DNA binding domain of p53), while specific mutations that are common in sporadic breast cancer were rare in BRCAI associated cancers.In several cases, multiple mutations were identified in a single allele of p53, sometimes as many as four.This hypermutability of p53 did not appear to result from increased general mutation rates as other genes did not show increased mutability , nor did other regions of p53.The observation of multiple mutations within single alleles of p53 is curious; a single event is enough to impair p53 function, so multiple events should provide no further growth advantage to cells.

BRCAJ in Cell cycle control
In addition to p53 involvement, a recent report has shown p21 Wall!CIP I activation by BRCA I to induce a G 1 to S phase cell cycle block (Somasundaram et a!., 1997).This finding is consistent with observations of the cell cycle regulation ofBRCA I; although expression seems maximal aftertheG/S checkpoint, it is induced in late G 1 (Chen eta!. , 1996b; Gudas  et aI., 1996; Vaughn et aI., 1996; Ruffner and Verma, 1997).The interaction with p21 Wall! C IPI does suggest that a link exists between BRCA I and cell cycle checkpoint control , implying that BRCA 1 may be more than a simple DNA repair protein.It has been proposed that BRCA 1 is a guardian of genomic integrity (Kinzler and Vogelstein, 1997), and such a role could explain why BRCA I mutations are not found in sporadic cancers.
BRCA I protein expression is closely linked to the cell cycle, both in terms of absolute level and of phosphorylation status (Chen et al .. 1996b; Gudas et al., 1996; Vaughn et  al .. 1996; Ruffner and Verma, 1997; Zhang et al.. 1997).Various cell cycle regulated and regulatory proteins, such as cdc2, cdk2, cdk4, cyclin A, cyclin B, cyclin B I, cyclin D, cyclin Dl, and E2F-4, interact with BRCAI (Chen et al.. 1996b;Wang et aI., 1997) , mainly modifying the phosphoryl ation state.Furthermore, an association between BRCAI and erbB/neu proteins (the EGFR family of receptor tyrosine kinases) has been proposed (Zhang et al. . 1997) such that activated tyrosine kinase receptors cause hypo phosphorylation of BRCA I at tyrosi ne residues, leading to functional inactivation of the protein.This hypophosphorylation of BRCA 1 may induce G/M cell cycle arrest (Zhang et al.. 1997).The proposal for a G /M arrest has been supported by observations made with a BRCA I mutant that abrogates such a checkpoint (Larson et al .. 1997).Several reports suggest that BRCA I mediates a G /S phase checkpoint (Vaughn et al..  1996; Somasundaram et al.. 1997).Despite the evidence for BRCA I Ip53 interaction, the G /S checkpoint may be activated directly through the transcriptional activation of p21 Wa l l /e l P I (Somasundaram et al., 1997) without participation of p53 .
Further evidence for a role in control of the cell cycle is suggested by observations of changes in the subnuclear localization during the cell cycle.BRCA I is detected as diffuse, weak, nuclear staining in cells in most stages of the cell cycle.However, during Sand M phases, a nuclear punctate pattern appears (Scully et al., 1996;Scully et aI. , 1997c).This dramatic change in the localization is also modified by the presence of DNA damage (Scully et aI., 1997b), which results in the release of BRCA I from the complexes with a concurrent change of phosphorylation state and the accumulation of BRCA I on PCNAcontaining structures.These findings seem to indicate a direct interaction between BRCA I and damaged, replicating DNA.Evidence is strong that BRCA I does affect the cell cycle; reduced expression is associated with highly proliferative cells in vivo (Sobol et aI. , 1996) and low BRCA I levels increase growth rates (Thompson etal. , 1995;Raoetal. , 1996); high expression ofBRCA I is toxic to cells (Holt et al., 1996;Wilson et aI. , 1997) and the effect is seen even in yeast cells (Humphrey et al., 1997) which do not have any identifiable BRCAI homologue.It has been proposed that BRCAI is a DNA binding transcription activator (Miki et al., 1994), as supported by the demonstration oftranscriptional activation by GAL4/BRCA I fusions (Chapman and Verma, 1996;Monteiro et aI., 1996) and the association ofBRCA I with the RNA polymerase II holoenzyme (Scully et al., 1997a).The same (BRCT) domain, is both the transcriptional activator and the RNA polymerase II interacting domain, suggesting that BRCA I may recruit the transcriptional machinery to target promoters through interaction with DNA or other proteins.Such evidence suggests a mechanism by which BRCA I may mediate control over the cell cycle.

Development and Differentiation
BRCA I appears to serve a role in development and differentiation .The nullizygous mouse models that have been developed show that such a function is essential for cellular survival during embryogenesis (Gowen et aI., 1996;Hakem et al., 1996;Ludwig e f al. , 1997;Hakem e f al. , 1997).The protein is expressed throughout all tissues of the developing mouse, particularly those that are rapidly proliferating and undergoing differentiation (Marquis et al., 1995).During pregnancy and lactation in the mouse , high levels of BRCA I are expressed at the time of rapid , hormonally induced , growth associated with lactation and differentiation of the mammary gland (Marquis et aI., 1995 ;Lane et al., 1995).Observations on the expression of BRCA I in human spermatocytes support a corresponding role in differentiation in humans (Zabludoff et al. , 1996).

MEDICAL RELEV ANCE OF BRCA I
Beyond screening individuals from families with high risk of cancer, BRCA I may have some potential for use in therapy.Given the cellular toxicity accompanying overexpression of BRCA I (Holt et al., 1996;Wilson et al., 1997), clinical trials have been undertaken to introduce wild type BRCA I to ovarian tumours using a retroviral vector (Tait et aI., 1997).Limitations of the delivery sysems available are likely, however, to restrict this approach to tumours with a large accessible surface, which therefore would exclude most breast cancers.Furthermore, if BRCA I is fundamentally a mediator of genomic stability (Kinzler and Vogelstein , 1997), this treatment strategy may be ill-advised.By stabilizing the tumour genome after a high proliferation rate has been established, many cells could be protected from the accumulation of lethal damage.
Quite apart from attempts at direct gene therapy, careful study of BRCA I and tumour phenotypes may lead to improvements in traditional treatment of BRCA I associated tumours.Mutation status of BRCA I has been found to correlate with several histopathological parameters that collectively determine the grade of the tumour (Goldstein et al. , 1987;Eisinger et ai., 1996;Rubin et al., 1996;Lakhani el ai., 1997).BRCA I-associated tumours tend to be poorly differentiated and highly proliferative.Such changes are intuitively associated with a worse prognosis, however, BRCA I tumours may be more drug-responsive than their unmutated counterparts (Spillman and Bowcock, 1996;Marks et al., 1997;Phillips el ai., 1997).Enhanced drug sensitivity may be a result of increased genomic instability or may simply reflect higher growth rates.Further study should help to clarify this, and contribute to the development of better treatment strategies.In addition, BRCA 1 status appears to influence the clinical outcome of patients (Porter et al., 1993;Malone et al., 1996;Marcus et ai., 1996;Rubin el al. , 1996;Brunet et ai., 1997;Burk, 1997;Cannistra, 1997;10hannsson et al. , 1997;Modan, 1997;Rubin, 1997;Whitmore, 1997).Many BRCA I-linked familial cancers evidently carry a better prognosis than their sporadic counterparts.If true, the reason is still unclear and may be a direct result of the way that BRCA I contributes to tumorigenesis, at the molecular level, or a less direct consequence of the tumour phenotype associated with germline BRCA I mutation.
Figure I -Re presentation of domains and interactions for BRCA I. BRCA I is represented linearly.with exons indicated.Shaded boxes are all the domains that have been identified (named above).Below each domain its demonstrated or inferred function s are shown.