BRCA1-hapoinsufficiency: Unraveling the molecular and cellular basis for tissue-specific cancer

abstract

Over the past 20 years tremendous progress has been made in understanding the function of BRCA1 gene products. Yet one question still remains: why is mutation of BRCA1 typically associated with preferential development of breast and ovarian cancers and not tumors in other tissues? Here we discuss recent evidence documenting the effect of BRCA1-haploinsufficiency in different cells and tissues and synthesize a model for how mutations in a single BRCA1 allele in human cells might preferentially confer increased cancer risk in breast epithelial cells.

KEYWORDS:

• BRCA1; haploinsufficiency; tumor suppression

Introduction

The Breast Cancer Associated 1 gene (BRCA1) has been of great interest to the scientific and medical communities since its discovery in the early 1990s in connection with families that have a high incidence of breast and ovarian cancers. Over the past 20 years, tremendous progress has been made in understanding the function of BRCA1 gene products. The complex and ubiquitous role of BRCA1 in DNA damage repair (DDR) has been uncovered, as has its involvement in chromatin organization, gene transcription, protein stability and cell division.^1-6 The versatility of BRCA1 comes from its ability to interact with a number of different proteins, to act as a scaffold for various complexes, and to regulate post-translational modifications of many binding partners including itself.^5,7,8 Although the number of studies on BRCA1 is ever increasing, one question still remains: which one of these functions is necessary for BRCA1 tumor suppression in such a tissue specific manner?

BRCA1 is defined as a chromatin-interacting E3 ubiquitin ligase that is involved in the homologous recombination type of DNA damage repair (DDR) – a process deemed to be error-free compared to the alternative, non-homologous end-joining.^7,9,10,11 Because all tumor cells with BRCA1- dysfunction display genomic instability associated with impaired DDR machinery, it is thought that this particular function of BRCA1 is critical for tumor suppression.^12,13,14 Since genomic maintenance is essential for the viability of all cells, this implies that loss of BRCA1 function, and thereby increased genomic instability, would lead to tumor formation in multiple tissues. However, this is not the case: increased cancer risk in BRCA1-mutation carriers is typically associated with preferential development of breast and ovarian cancers and not tumors in other tissues.

What then is so special about the function of BRCA1 in breast and ovarian tissues? Addressing this problem over the past 20 years has proven very challenging. Several hypotheses have been proposed to explain tissue specific tumor suppression. First, it was postulated that breast and ovarian epithelium specifically require BRCA1 for DDR while other epithelial cells might compensate using other pathways. Other hypotheses included the possibility that breast and ovarian cells might exhibit delayed apoptosis or a different rate of loss of heterozygosity (LOH) compared to other tissues.¹⁵ However, many of these theories have not been formally tested. This has been further complicated by the fact that modeling the human disease associated with BRCA1-halpoinsuficiency in mice has been particularly difficult. Deletion of both BRCA1 alleles is embryonic lethal and BRCA1-heterozygous mice do not develop mammary or ovarian tumors spontaneously. This latter observation implies that there is considerable difference in the function of BRCA1 in human versus mouse cells; and more recently doubts have been raised as to whether studying BRCA1 function in mouse mammary epithelium is even suitable.^16,17

Here we aim to discuss the most recent studies examining the consequence of BRCA1-haploinsufficiency in different cells and tissues and synthesize a working model for how mutations in a single BRCA1 allele in human cells might preferentially confer increased cancer risk in breast epithelial cells.

Unresolved questions about BRCA1

There are several unusual observations involving the phenotype as well as the progression of BRCA1-associated cancers that until recently could not be resolved based on the prior understanding of BRCA1 function. BRCA1-associated cancers have an early and rapid onset compared to sporadic breast and ovarian cancers. In addition to being tissue specific, the majority of BRCA1-associated breast cancers are of the basal subtype, while sporadic breast cancers tend to be of the luminal subtype. Also, unlike other inherited cancer predisposing syndromes, loss of the remaining wild-type BRCA1 allele does not appear to be a rate-limiting step in BRCA1-associated breast tumor progression.

Early and rapid onset

Early onset of breast cancer is observed in germline mutation carriers of various high and moderate susceptibility genes such as: BRCA2, TP53, PTEN, STK11 and ATM, CHEK2, BRIP1 and PALB2, respectively. Because most of these genes function in genome surveillance and repair, mutation in these gatekeeper genes leads to highly unstable genomes creating a mutator phenotype, which results in accelerated accumulation of additional mutations and thus early onset cancer.^18,19 Consistent with these other early onset inherited cancer-causing predispositions, average BRCA1-mutation carriers are younger at the time of breast cancer diagnosis compared to non-carriers, and BRCA1-mutation carriers may even develop breast cancer as early as 18 years of age. It is estimated that by age 40, the cumulative risk of breast cancer in carriers is 10%, whereas in non-carriers it is 0.2%. By age 80, the estimated cumulative risk in carriers is 73.5%, whereas in non-carriers is 6.8%, implying that the lifetime risk of developing breast cancer in BRCA1-mutation carriers is overwhelmingly high compared to the general population.^20,21 Additionally, several studies that compared the age of onset between mothers and daughters that are BRCA1/2-carriers found a statistically significant earlier age of onset in daughters (vs their mothers) that carried BRCA1 mutation.²² While the underlying basis for this generational acceleration of cancer onset is still not fully understood,²³ showed that with each successive generation telomere shortening was due to BRCA mutation.

While early onset of breast cancer is typical in germline mutation carriers of BRCA1-associated as well as other cancer predisposing syndromes, BRCA1-associated breast cancers often develop rapidly between screens (> 6 months), and at the time of diagnosis are quite aggressive.^24,25 Several studies have reported that nearly 50% of BRCA1-mutation carriers who undergo close surveillance develop malignant disease in less than a year after having normal findings on screening mammography.^26,27,28,29 This rapid and earlier onset in successive generations is perplexing as it implies the multistep process of tumorigenesis in BRCA1 carriers might not proceed in a similar manner or the same rate as that in other germline mutation carriers of other high and moderate susceptibility genes.

Tissue specificity

As mentioned earlier, why would mutation of a gene that is important for DNA repair be associated with an increased risk primarily in only the breast and ovary? Since BRCA1 is important in DNA damage repair, the increased risk of developing cancer should be generalized such with other tumor suppressor genes (eg., TP53, PTEN, STK11, ATM, etc). However, this is not the case.

Several ideas have been proposed including the notion that BRCA1 associates with (and thereby potentially regulates) XIST RNA, a non-coding RNA that is involved in coating and inactivation of X chromosome (Xi).^30,31 In this model, BRCA1 maintains heterochromatin and suppresses redundant gene expression from the Xi through XIST RNA in female cells. However, several subsequent studies suggest that data in support of the role of BRCA1 in XIST RNA regulation are inconsistent (i.e. conflicting results are seen in different cell lines) and potentially a result of a broader effect on heterochromatin maintenance.^32,33 In addition, analysis of X-chromosome status in BRCA1 primary tumors (using single cell and global genomic approaches) observed considerable intratumoral and intertumoral variability of the number of XIST RNA domains, which were consistent with chromosomal genetic abnormalities, including gains, losses, reduplication and rearrangements rather than BRCA1 status.^34,35

Another possibility is that BRCA1 performs a unique function in breast and ovarian epithelial cells and this function is only relevant for tumor suppression. BRCA1's role in downregulation of estrogen and progesterone receptor would fit the requirements for the tissue specificity.^36,37,38 However, the majority of BRCA1-tumors are estrogen and progesterone receptor negative and these observations present a challenge for this hypothesis.¹⁵

Other theories argue that tissue specificity is due to profound differences in DNA damage repair pathways utilized, apoptosis triggers and tolerance, and even the rate at which LOH might occur in different tissues. For example, in unaffected tissues other proteins might serve to compensate for improper BRCA1 function in the context of DNA damage repair. If this were the case, then such compensation for BRCA1 loss would be absent only in ovarian and breast epithelial cells. However, severe embryonic lethality in BRCA1^−/− mice is inconsistent with this hypothesis.¹² Also, it has been suggested that breast and ovarian cells exhibit a delay in apoptosis compared to other rapidly dividing cells. Therefore, loss of the WT BRCA1 allele might be tolerated longer in these tissues compared to unaffected tissues.³⁹ However, no reports testing this particular hypothesis exist as of yet, and it is unclear whether breast and ovarian cells actually exhibit a delay in apoptosis compared to other tissues. Lastly, it is possible that these particular tissues exhibit different rates of loss of the 2^nd WT allele. Several reports indicate that mutation spectrum and frequency differ in a tissue-specific way in mice.^40,41,42,43 In addition, a difference in the rate of mitotic recombination has been observed in human lymphocytes isolated from different individuals. Interestingly, mitotic recombination seems to be higher in females than males, suggesting a possible gender bias for DDR.^15,44 But, this has not been examined in context of BRCA1 tumors or BRCA1-heterozygosity. Thus, it is possible that a combination of these factors maybe contributing to tissue specificity of BRCA1-associated cancers. Research efforts on this front are still in infancy because they require comparative studies of BRCA1 function in different tissues and cell-types, which is challenging to obtain and examine.

Finally, it has been hypothesized that estrogen exposure increases DNA damage and genetic instability leading to increased risk of BRCA1-associated breast and ovarian cancers.⁴⁵ In particular, BRCA1/2 play a role in protecting breast epithelial cells from oxidative DNA damage due to elevated levels of reactive oxygen species – a byproduct of hormonally driven growth and its effect on cell metabolism. Oxidized DNA lesions that persist into S-phase result in double strand breaks would necessitate BRCA1/2 mediated HR repair. While this theory is attractive and supported, other equally hormonally sensitive tissues (eg. uterus, cervix, bone marrow, brain etc.) do not show higher rates of cancer incidence in BRCA1-mutation carriers. Thus, why would estrogen be metabolized differently in the breast and ovary?

Aggressiveness

Another unusual characteristic associated with BRCA1 is that breast cancer from mutation carriers are more likely to exhibit aggressive features compared to cancers developed in non-carriers at the time of diagnosis. Clinical characteristics and outcomes of breast cancers that arise in BRCA1-mutation carriers are significantly more likely to be histologic grade III (100% vs. 59%, p=.04) including having other adverse clinical and histopathological features.^24,25 Also, BRCA1-mutation carriers preferentially develop poorly differentiated, basal subtype of breast cancers.^46,47,48 While the underlying mechanism responsible for BRCA1-associated tumor subtype bias has been reported,⁶⁶ it is unclear whether the other traits associated with BRCA1-associated cancers are a consequence of the basal-tumor phenotype or whether BRCA1 loss affects other traits associated with the aggressiveness.

Is LOH the rate limiting step?

According to the Knudson two-hit hypothesis, loss of heterozygosity in tumor-suppressor genes is the rate-limiting step in tumor development. In support of this hypothesis, LOH is frequently observed in BRCA1-associated cancers and within normal regions around the cancers.^48,52 However, there have been conflicting reports regarding the timing by which the remaining wild-type allele is lost. Some reports found LOH to be rare in disease-free and preneoplastic tissues, while others reported that the number of samples with evidence of LOH was as high as 50%. This difference might be due to tissue sections examined, since in some studies the tissue was obtained from prophylactic mastectomies, while in others the tissue was adjacent to cancer. More recent observations using laser microdissection to isolate cells from various pathologic lesions and corresponding normal tissues, suggest that there is considerable heterogeneity in LOH (affecting wt as well as mutant allele) within and between pre-invasive lesions and invasive cancers from BRCA1-mutation carriers.^52-55 Additionally, several reports have shown that hallmarks of BRCA1-associated cancers (increased genomic instability, inefficient DNA damage repair, PTEN and p53 loss) can be already observed in BRCA1-haploinsufficient cells, and therefore prior to BRCA1 loss.^{52,56,57,58,59} Taken together, these data suggest that LOH is a stochastic (albeit still potentially selected) event and not an obligatory first step for initiation of BRCA1-associated breast tumors.

Clues to tissue-specificity: BRCA1-haploinsufficiency phenotype

Given these unresolved questions about the unusual characteristics of BRCA1-associated cancers, considerable effort has been made toward re-examining the paradigms of BRCA1 function and tumor suppression. Thus, investigations have been turned toward the effect of BRCA1-haploinsufficiency, rather than loss in cells. A number of recent studies have emerged with data directly comparing non-cancerous cells and tissues from BRCA1-mutation carriers. Such studies have identified differences in BRCA1 gene dosage, deficiencies in DNA damage repair, and defects in lineage commitment due to BRCA1-haploinsufficiency that have started to shed light on tissue-specific roles of BRCA1. These recent findings strongly support the notion that molecular and cellular features associated with BRCA1-haploinsufficiency and not loss, may be unique to breast epithelial cells and may help explain the increased propensity for specific cell types for neoplastic transformation.

BRCA1 expression levels

Whether inheriting a mutant BRCA1 allele actually results in a difference in gene dosage has been difficult to determine. How much BRCA1 is expressed in cells of mutation carriers (a half or 2/3 of wildtype levels, or is there no significant difference in the protein levels)? Do expression levels vary depending on the type of mutation? What is the expression level of mutated or truncated versions of BRCA1 as well as alternative BRCA1 transcripts? Is there a difference in BRCA1 level across different cell-types and across different tissues? These questions have yet to be fully answered and need to be in order to understand the major contributory factors during BRCA1-associated tumorigenesis.

However, it has been presumed that harboring a deleterious mutation in BRCA1 results in decreased protein expression (perhaps to a half) of BRCA1 WT cells, but only a few recent studies have examined this.⁵⁶ detected a decrease in BRCA1 protein levels in whole-cell extracts of lymphoid cell-lines derived from three BRCA1-mutation carriers in comparison to two BRCA1 WT samples. Also,⁶⁵ noted a decrease in BRCA1 levels in primary breast epithelial cells as well as skin fibroblasts from a relatively few different patient samples. However, two other studies failed to detect any differences at the level of mRNA or protein when BRCA1 levels were compared in whole-cell extracts of primary breast and skin epithelial and fibroblast cells.^59,66 Since none of these studies examined large sample sizes, additional analyses are necessary to establish whether inheriting a mutant BRCA1 allele affects its expression and whether certain mutations exhibit greater gene dosage effects than others.

Another possibility that may not depend on dosage is that mutant BRCA1 may interfere with the function of wildtype BRCA1 in BRCA1-heterozygous cells. While this issue has not been extensively examined, a few studies propose that certain full-length mutant BRCA1 proteins may function incorrectly in the cell. In particular, Fan and colleagues (2001) found that C-terminal truncated BRCA1 proteins could abrogate certain functions of WT BRCA1 such as chemosensitivity, susceptibility to apoptosis, and inhibition of estrogen receptor transcriptional activity. In addition,⁶¹ reported that mutations in the 3′ region of the BRCA1 gene enhance its recruitment to chromatin and chromatin unfolding. Therefore, it is possible that some mutant BRCA1 proteins may act as dominant negatives thereby actively promote tumorigenesis, but further research into this topic is needed. Furthermore, it has been found that BRCA1 has a number of alternative transcripts and that they contribute to BRCA1 function,^62,63,64 However, additional studies are needed to further define and describe these mechanisms especially in the context of BRCA1-haploinsufficiency.

DNA damage repair and genomic instability

No obvious developmental phenotype has been reported in humans harboring heterozygous mutations in BRCA1. However, increased allelic imbalance (or LOH) in pre-malignant breast tissue samples from BRCA1-mutation carriers has been observed.^53,54,57,67 reported that the types of genomic aberrations frequently found in BRCA1-mutation carriers include low copy number gains and losses. Some of these gene copy number changes were similar across samples from different patients and linked to transcriptional regulation and DNA binding.

In addition to these studies, BRCA1-haploinsufficency leads to defects in DNA damage repair responses and genomic instability in tissues and cells from BRCA1-mutation carriers, as well as in genetically engineered breast epithelial cells.^56,58,68 In lymphoblastoid cell-lines derived from BRCA1-mutation carriers, BRCA1-heterozygous breast cancer cells, as well as genetically engineered BRCA1-heterozygous breast epithelial cells, a substantial deficiency in error-free double-strand end-joining and homologous repair is seen.^56,58 In addition, sensitivity to genotoxic stress and increased rates of spontaneous hyper-recombination in BRCA1-heterozygous cells was found.^56,58,68 More recently, Pathania et al. examined functionality of BRCA1 in DNA damage repair, checkpoint control and genome integrity maintenance. This was done in a collection of primary human BRCA1 WT and BRCA1-haploinsufficient mammary epithelial cells and skin fibroblasts. The group did not find differences in homologous recombination- double-strand break repair, checkpoint functions, centrosome number control and spindle pole formation. However, these same cells, according to the study, exhibited defects in managing replication stress (stalled replication fork repair and/or suppression of fork collapse). Interestingly, when BRCA1-haploinsufficient cells were pre-exposed to replication stress (UV-treated) and then assayed for HR (IR-irradiated), their ability to recruit Rad51 to DSBs, and thereby presumably execute HR-DSBR effectively, was diminished. These observations revealed the first evidence of “conditional haploinsufficiency” in BRCA1-heterozygous cells. That is, basal levels of BRCA1 in haploinsufficient cells can still perform most of its expected functions, but the available pool of this protein is indeed limited; thus, defects become apparent when cells are challenged under conditions that require BRCA1 intervention/involvement.

Why are there conflicting reports regarding HR impairment due to BRCA1 mutation? Certainly one possibility could be attributed to differences in the cells used for the assays –multiple patient derived primary cells ⁶⁵ vs. established laboratory cell-lines.^56,58,68 In addition, BRCA1 levels in these cells might differ as well as the type of assays and biomarkers used to examine HR in the different studies might play a role. Where one study examined the level of recruitment of DNA damage repair proteins involved in HR,⁶⁵ the other studies employed vector-based HR repair reporter assays.^56,58,68 Regardless of their differences, these findings fortify the notion that BRCA1-haploinsufficient cells exhibit defects in DNA repair and maintenance of genomic stability, prior to malignant features and provide strong evidence that a single mutant allele (rather than loss of both alleles) is sufficient to accelerate genomic instability.

While this function of BRCA1-haploinsufficiency is provocative and provides some explanation for the increased risk of neoplastic transformation in the absence of BRCA1 loss, the findings above were not a tissue or cell-type specific phenotype. Hence, the question still remains as to why the breast and ovaries more susceptible to transformation? To begin to answer this question,⁵⁹ examined the effect of BRCA1-haploinsufficiency in disease-free breast and skin tissues from wild-type or BRCA1 mutation carriers to ascertain whether there might be any tissue- or cell-type specific changes due to BRCA1-haploinsufficiency. This group performed karyotype analysis and used semi-quantitative methods to determine the level of DNA damage and activation of the DNA damage response (DDR) pathway in primary breast epithelial cells (HMECs) as well as breast fibroblasts (HMFs). Increased genomic instability and DDR activation was observed in BRCA1-haploinsufficient HMECs but not in HMFs. In addition, HMECs exhibited increased rate of telomere erosion. Concomitant with this, BRCA1-haploinsufficient HMECs but not HMFs exhibited premature senescence that occurred in the absence of LOH—a process termed haploinsufficiency induced senescence (HIS).

The finding that breast epithelial cells suffered increased DNA damage and appeared to be more sensitive to decreased levels of BRCA1 compared to fibroblasts suggested that epithelial cells rely mostly on BRCA1-directed homologous recombination (HR) pathway to repair double strand breaks than other pathways. Therefore, HIS may exists as a response to genetic insults that accumulate in epithelial cells due to inefficient DNA damage repair. It is possible that fibroblasts do not undergo HIS because they are able to compensate for weakened HR pathway by utilizing other DNA damage repair mechanisms such as non-homologous end joining. This finding is consistent with other studies reporting differences in DDR between epithelial cells and fibroblasts.⁶⁹ Furthermore, the difference between epithelial cells and fibroblasts in their response to BRCA1-haploinsufficiency is interesting because it is consistent with the notion that germline mutations in BRCA1 do not predispose to soft tissue sarcomas, but rather carcinomas.

This study also provided the first molecular evidence of a tissue specific difference in BRCA1 function. Although HIS was observed in both epithelial cell types examined, HIS in breast epithelial cells was different in that it was triggered by the unique nature of this cell type; rapid telomere erosion is characteristic of human breast epithelial cells and not other cells and this study found that inactivation of the NAD-dependent deacetylase SIRT1 triggered HIS in HMECs. Further analysis revealed that BRCA1 might be involved in maintenance of telomere stability (i.e., telomeric heterochromatin) specifically in HMECs through regulation of SIRT1. It is possible that this may be an important mechanism in BRCA1-haploinsufficient HMECs leading to telomere fragility and increased genomic instability associated with telomere-end fusions as well as cell-cycle arrest. This increased rate of telomere erosion in combination with deficient DNA damage repair may result in faster accumulation of deleterious mutations in BRCA1-haploinsufficient breast epithelial cells and potentially explain the early onset and aggressive nature of BRCA1-associated cancers.

BRCA1 in breast epithelial cell differentiation

Other phenotypes associated with BRCA1-haploinsufficiency have also been described within the last decade. In 2008 Liu et al. showed that BRCA1 levels increase in mammary progenitor cells upon differentiation. Knockdown of BRCA1 in primary HMECs increased both the number and self-renewal of in vitro stem/progenitor cells. When grown as xenografts, BRCA1 knockdown cells failed to form proper mammary structures suggesting that decreased levels of BRCA1 may block differentiation.⁷⁰

Two subsequent studies also examined properties of breast epithelium derived from BRCA1-mutation carriers.^66,71 Both of these studies described altered mammary progenitor cell numbers in disease-free breast from BRCA1-mutation carriers compared to non-carriers. While Lim et al. reported expansion of luminal progenitors, Proia et al. observed an increase in basal progenitors. The differences between these studies could be explained by the fact that luminal progenitor cells in BRCA1-mutation carriers appear to co-express luminal as well as basal markers, as described by Proia et al. Consistent with this, defects in luminal linage maturation and aberrantly expressed differentiation markers were observed in the breast epithelium of BRCA1-mutation carries; these were also accompanied by notable changes in genes associated with DNA transcription and chromatin remodeling suggesting changes in the epigenetic landscape of breast epithelial cells from BRCA1-mutation carriers.⁵⁹ Collectively, evidence is supporting the notion that mutation in one BRCA1 allele is sufficient to cause significant changes within mammary epithelial cells that affects their differentiation potential and possibly stem cell functions. Whether similar changes are observed in ovarian or fallopian tube epithelial cells is currently not known.

Conclusion

It is clear that elucidating BRCA1-haploinsufficient phenotype is becoming instrumental in understanding BRCA1 function in breast epithelial cells as well as the peculiar observations about BRCA1-associated breast cancers. Several recent findings support the notion that mutations in a single BRCA1 allele are sufficient to alter breast epithelial cells prior to any indications of neoplastic changes. This may hold the clues to the tissue-specific nature of carcinogenesis in BRCA1-mutation carriers. Differences seen in BRCA1-haploinsufficient breast epithelial cells imply that BRCA1 function seems to be essential for DNA damage repair, telomere stability and epithelial cell differentiation in these cells, but possibly not in other cells. These early studies provide a framework for future studies of BRCA1 tumor suppressive function in a cell- and tissue- specific manner. Identifying BRCA1 function and the molecular events associated with it that span different cells types from BRCA1-mutation carriers is a highly innovative approach and will provide insight(s) that may ultimately lead to development of more efficient and less invasive treatments for BRCA1-mutation carriers.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by the Raymond & Beverly Sackler Convergence Lab and grants from ArtBeCAUSE, the Breast Cancer Research Foundation, and the NIH/NCI CA125554 and CA092644.