brca2 and tp53 Collaborate in Tumorigenesis in Zebrafish

Germline mutations in the tumor suppressor genes BRCA2 and TP53 significantly influence human cancer risk, and cancers from humans who inherit one mutant allele for BRCA2 or TP53 often display loss of the wildtype allele. In addition, BRCA2-associated cancers often exhibit mutations in TP53. To determine the relationship between germline heterozygous mutation (haploinsufficiency) and somatic loss of heterozygosity (LOH) for BRCA2 and TP53 in carcinogenesis, we analyzed zebrafish with heritable mutations in these two genes. Tumor-bearing zebrafish were examined by histology, and normal and neoplastic tissues were collected by laser-capture microdissection for LOH analyses. Zebrafish on a heterozygous tp53M214K background had a high incidence of malignant tumors. The brca2Q658X mutation status determined both the incidence of LOH and the malignant tumor phenotype. LOH for tp53 occurred in the majority of malignant tumors from brca2 wildtype and heterozygous mutant zebrafish, and most of these were malignant peripheral nerve sheath tumors. Malignant tumors in zebrafish with heterozygous mutations in both brca2 and tp53 frequently displayed LOH for both genes. In contrast, LOH for tp53 was uncommon in malignant tumors from brca2 homozygotes, and these tumors were primarily undifferentiated sarcomas. Thus, carcinogenesis in zebrafish with combined mutations in tp53 and brca2 typically requires biallelic mutation or loss of at least one of these genes, and the specific combination of inherited mutations influences the development of LOH and the tumor phenotype. These results provide insight into cancer development associated with heritable BRCA2 and TP53 mutations.


Introduction
Tumor suppressor genes provide a key barrier to neoplastic transformation by repressing survival and proliferation of abnormal cells. Germline mutations in the tumor suppressor genes BRCA2 and TP53 influence tumor susceptibility in many vertebrate species. In humans who inherit one mutated copy of BRCA2 or TP53, tumor development is often associated with loss of the wildtype allele, indicating that somatic loss of heterozygosity (LOH) is important for neoplastic transformation [1][2][3][4]. However, tumors can occur in human carriers of BRCA2 or TP53 mutations without somatic LOH, suggesting that haploinsufficiency for either gene can lead to tumorigenesis [1,5,6].
BRCA2 mutation in humans is associated with two distinct cancer susceptibility syndromes. Individuals who inherit one mutant allele for BRCA2 experience an increased risk for breast and ovarian cancer in adulthood [7,8], while individuals who inherit biallelic BRCA2 mutations have a high incidence of malignancies during childhood [9]. TP53 mutation may have a synergistic effect on tumorigenesis in BRCA2-associated cancers. Breast and ovarian cancers from BRCA2-heterozygous patients often develop TP53 mutations [10][11][12], which may precede loss of the wildtype BRCA2 allele [13], and are not attributable to a generalized or random increase in genetic mutations [10].
Although TP53 dysfunction is important in BRCA2-associated carcinogenesis, the effect of coincident disruptions in these genes on tumorigenesis is not well defined. In this study, we investigated the relationship between inherited mutations in brca2 and tp53, and somatic LOH for these genes, in tumorigenesis in zebrafish [16,18]. Zebrafish with heterozygous tp53 M214K mutation displayed a high incidence of malignant tumors. LOH for tp53 was an important factor in malignant tumors from brca2 wildtype and brca2 Q658X heterozygous zebrafish, but was less important in brca2 Q658X homozygous mutant zebrafish. Furthermore, the brca2 mutation status influenced the age at tumor onset, tumor number, and tumor type. Lastly, malignant peripheral nerve sheath tumors consistently exhibited LOH for tp53, while undifferentiated sarcomas more commonly exhibited loss of brca2 via homozygous mutation. These findings indicate that mutations in brca2 and tp53 enhance tumorigenesis in zebrafish, as seen in humans, and that the incidence and type of tumor depends upon the particular combination of mutations.
To determine if the risk of malignant tumor development correlated with the brca2 genotype in tp53+/m zebrafish, the numbers of zebrafish that developed at least one malignant tumor were determined for brca2+/+;tp53+/m, brca2+/m;tp53+/m, and brca2 m/m;tp53+/m cohorts (Table 1). Tumors with clear histologic evidence of tissue invasion and destruction were classified as malignant, while tumors that exhibited expansile but noninvasive growth were classified as benign. In all three cohorts, most tumorbearing zebrafish developed malignant tumors, rather than benign tumors (Table 1 and Table 2). The proportion of zebrafish that developed at least one malignant tumor was not significantly different between any of the three cohorts ( Figure 1B and Table 1). In comparison to the other cohorts, a greater proportion of brca2 m/m;tp53+/m zebrafish developed benign tumors ( Figure 1B), due to an increased incidence of benign testicular tumors ( Table 2). The most common malignant tumor types observed in this study were malignant peripheral nerve sheath tumors ( Figure 1C and Table 2) and undifferentiated sarcomas ( Figure 1D and Table 2), the latter of which lacked sufficient histologic differentiation for more specific classification. Some zebrafish developed two or more histologically distinct and anatomically discrete tumors ( Figure 1E). Over 50% of brca2 m/m;tp53+/m zebrafish developed more than one tumor; however, less than 20% of brca2+/+;tp53+/m and brca2+/m;tp53+/m zebrafish developed multiple tumors (Table 1). When multiple tumors occurred, they were distinguishable by histologic features, and were typically distinctly different tumor types ( Figure 1E).
To determine the background level of tumor development over time in zebrafish with brca2 mutation alone, we screened a small group of aged brca2+/+;tp53+/+, brca2+/m;tp53+/+, and brca2 m/ m;tp53+/+ zebrafish for tumor development (Table S1). The overall tumor incidence was higher in the brca2 m/m;tp53+/+ cohort (Table S1 and Figure S2) compared to the other cohorts. This difference was attributable to an increased incidence of testicular tumors in brca2 m/m;tp53+/+ zebrafish, as observed previously [16]. The development of multiple tumors in one animal was uncommon in all cohorts (Table S1).

LOH for brca2 and/or tp53 is common in malignant zebrafish tumors
To identify LOH in zebrafish samples, we analyzed 31 histologically malignant tumors, four histologically benign tumors, and 28 matched normal tissue specimens from tp53+/m zebrafish that were brca2+/+, brca2+/m, or brca2 m/m (Table S2). Tissue samples were isolated by laser-capture microdissection (LCM) (Figures 2A-C) for DNA extraction, PCR amplification, and sequencing (Methods S1 and Table S3). Multiple tumors in individual zebrafish were isolated as separate samples ( Figure 2B).
Sequence analyses revealed that LOH for brca2 and tp53 occurred frequently in malignant tumor specimens, but was not observed in normal tissues (Figures 2D and 2E and Table S2). Interestingly, different tumors analyzed from a single zebrafish did not necessarily exhibit the same LOH profile ( Figure 2E and Table  S2). Although tp53 LOH always involved loss of the wildtype allele, two tumors from brca2+/m;tp53+/m zebrafish exhibited loss of the mutant brca2 allele (Table S2). Loss of the mutant BRCA2 allele has been reported in human tumors, but the functional significance of this change is unclear [6]. No LOH was detected in three of four histologically benign tumors or in any normal tissue specimens (Table S2).
To investigate a second mechanism for brca2 inactivation in malignant tumors from brca2+/m;tp53+/m zebrafish, we analyzed methylation status of the putative promoter region for brca2 in four tumor specimens that retained the brca2 wildtype allele. Significant methylation was not detected in tumor or matched control samples (Table S4).

Malignant tumor type correlates to brca2 genotype and LOH profile
Segregation of malignant tumors by tumor type indicated that brca2 mutation status correlated with both tumor type and LOH profile (Table 3 and Figure 3B). These differences were observed for the two most common malignant tumor types observed in this study, malignant peripheral nerve sheath tumors (MPNST) and undifferentiated sarcoma. The proportion of zebrafish that developed MPNST was not significantly different between brca2+/+;tp53+/m and brca2+/m;tp53+/m cohorts (Table 3). However, the proportion of zebrafish that developed MPNST in brca2 m/m;tp53+/m zebrafish was significantly lower than both brca2+/ +;tp53+/m and brca2+/m;tp53+/m cohorts ( Table 3). The proportion of zebrafish that developed undifferentiated sarcomas in the brca2 m/m;tp53+/m cohort was higher than either the brca2+/ +;tp53+/m or brca2+/m;tp53+/m cohorts, but this difference did not reach statistical significance (Table 3).
MPNST and undifferentiated sarcomas were also correlated with different LOH profiles. All MPNST analyzed for LOH had lost the tp53 wildtype allele, regardless of brca2 status (14 of 14) ( Figure 3B). Interestingly, this includes the single MPNST analyzed from the brca2 m/m;tp53+/m population, although tp53 LOH was otherwise uncommon in tumors from this group. In comparison, 60% of undifferentiated sarcomas (6 of 10) had loss of the wildtype brca2 allele, or occurred in brca2 m/m;tp53+/m fish, regardless of the tp53 status ( Figure 3B). However, this trend largely reflected the higher incidence of undifferentiated sarcomas in the brca2 m/m;tp53+/m cohort (Table 3).

Discussion
BRCA2 and TP53 are well-known tumor suppressor genes that have been linked to defined human cancer syndromes, but the effect of combined genetic disruptions in these genes on carcinogenesis is not well defined. We examined tumor develop- In the absence of tp53 mutation, tumor incidence was increased in brca2 m/m;tp53+/+ zebrafish when compared to brca2+/ +;tp53+/+ and brca2+/m;tp53+/+ zebrafish. However, on a tp53+/m background, zebrafish of all three brca2 genotypes experienced similar tumor incidence. The age at tumor onset was statistically significantly lower, and the proportion of zebrafish with multiple tumors higher, in brca2 m/m;tp53+/m zebrafish when compared to brca2+/+;tp53+/m or brca2+/m;tp53+/m zebrafish. These results indicate that: 1) homozygous brca2 mutation increases tumor development in zebrafish, similar to humans with germline biallellic BRCA2 mutations, and 2) homozygous brca2 mutation enhances carcinogenesis in tp53+/m zebrafish, supporting previous work that indicates a collaborative relationship between BRCA2 and TP53 in human carcinogenesis.
In humans, somatic LOH occurs frequently in tumors from patients who inherit one mutated copy of BRCA2 or TP53 [1][2][3][4]. To investigate the role for LOH in carcinogenesis in zebrafish with brca2 and tp53 mutations, we examined malignant zebrafish tumors and matched normal tissues for evidence of LOH. Loss of the wildtype alleles for tp53 and brca2 was common in malignant zebrafish tumors, implicating LOH as an important contributor to carcinogenesis in this species.
Importantly, development of LOH was dependent on brca2 mutation status. Since all malignant tumors from brca2+/+;tp53+/ m zebrafish developed tp53 LOH, we conclude that LOH for tp53 is necessary for carcinogenesis in tp53+/m zebrafish without brca2 mutation. In contrast, the majority of malignant tumors from brca2 m/m;tp53+/m zebrafish did not develop somatic LOH for tp53. These findings suggest that on a tp53+/m background, the presence or absence of functional brca2 may influence subsequent genetic alterations required for carcinogenesis. While most malignant tumors from brca2+/m;tp53+/m zebrafish developed LOH for tp53, LOH for brca2 was comparatively less common. These results suggest that in brca2+/m;tp53+/m zebrafish, biallelic inactivation or loss of brca2 is either not required for tumorigenesis, occurs late in disease, or is achieved by other mechanisms. These possibilities have been previously postulated to explain why some cancers in humans with heterozygous BRCA2 mutation do not develop BRCA2 LOH or promoter methylation [1,6]. In comparison, TP53 mutation is thought to be a relatively early event in the pathogenesis of BRCA2-associated cancer in humans with heterozygous BRCA2 mutation [13]. The high incidence of tp53 LOH in malignant tumors from brca2+/m;tp53+/ m zebrafish supports the concept that TP53 dysfunction is a critical and potentially early step in BRCA2-associated carcinogenesis.
In addition to brca2 genotype, LOH status also correlated with malignant tumor type. All MPNST evaluated for LOH had lost the tp53 wildtype allele, regardless of brca2 status. Several previous studies in zebrafish support an association between loss of functional tp53 and MPNST [18][19][20][21]. In contrast, undifferentiated sarcomas were most commonly associated with homozygous brca2 mutation. These patterns of gene disruptions may reflect disparate roles for brca2 and tp53 in zebrafish tumorigenesis. Interestingly, human cancer syndromes linked to heritable BRCA2 mutations are also limited to a small range of tumor types [7][8][9].
The methods applied in this study represent a minimum estimate for LOH in malignant zebrafish tumors, as screening for LOH was done by sequencing within the exons containing the brca2 Q658X and tp53 M214K germline mutations. Other mechanisms of gene inactivation may also have contributed to carcinogenesis in this population. When we examined the methylation status of the putative brca2 promoter in a small set of normal and tumor samples from brca2+/m;tp53+/m zebrafish, we did not find evidence of promoter methylation in any sample. While we cannot rule out the possibility that brca2 promoter methylation contributed to tumorigenesis in this study, BRCA2 promoter methylation occurs infrequently in tumor specimens from humans with heterozygous BRCA2 mutations [1].
Human cancers display a remarkable degree of genetic heterogeneity [22,23]. Both inter-and intratumoral heterogeneity are influenced by genetic factors, such as genomic instability, and nongenetic factors, such as tumor microenvironment [23,24]. Two observations from the current study indicate a degree of heterogeneity among malignant zebrafish tumors. First, different tumors from the same animal did not necessarily exhibit the same LOH profile, suggesting that tumor initiation and/or progression could have involved different mechanisms, and may have been influenced by the site of onset and cell of origin. Second, a small number of tumors exhibited partial LOH for brca2 or tp53 (Table  S2), indicating that a subset of tumor cells retained the wildtype allele for these genes. This suggests the presence of genetically diverse subclones within malignant zebrafish tumors, as has been observed in human cancers [22,23]. Further investigation of genetic diversity in zebrafish tumors will be required to understand tumor heterogeneity in this species.
In this study, we demonstrate that LOH for brca2 and tp53 represents a conserved mechanism for carcinogenesis in zebrafish, and suggest that the relative importance of LOH or haploinsufficiency for brca2 and tp53 in driving carcinogenesis is dictated by brca2 genotype (Figure 3C). In zebrafish that are wildtype or heterozygous for brca2 mutation, tp53 LOH appears to be a critical step in driving carcinogenesis. In contrast, cancers associated with homozygous brca2 mutation do not required tp53 LOH. We have previously reported accelerated tumorigenesis in brca2+/m;tp53 m/ m and brca2 m/m;tp53 m/m zebrafish compared to brca2+/+;tp53 m/m zebrafish [16]. From these two studies, we conclude that carcinogenesis in zebrafish with germline mutations in brca2 and tp53 typically requires biallelic inactivation or loss of at least one of these two genes, and this effect is enhanced by haploinsufficiency or biallelic loss of the other gene.
The collaborative effects of BRCA2 and TP53 mutations on carcinogenesis have been previously described in human cancer. Genomic instability is thought to be a significant factor in human carcinogenesis [22,23], and germline BRCA2 mutation is linked to an increased mutation rate in Brca2-mutant mice [25] and in BRCA2-associated human cancer [26,27]. Coincident or subsequent TP53 pathway disruption in BRCA2-deficient cells may permit survival and proliferation of cell populations with significant genetic aberrations, ultimately leading to neoplastic transformation. Further investigation of genetic and/or epigenetic alterations accompanying brca2 mutation in malignant zebrafish tumors may uncover additional factors that contribute to BRCA2associated cancer in humans.

Ethics statement
Zebrafish were monitored for clinical and gross evidence of tumor development and humanely euthanized with 50X Tricaine in system water buffered with Sodium Bicarbonate (0.7 grams/ liter). All animal studies were approved by the Intramural Animal Care and Use Committee, National Cancer Institute, National Institutes of Health, Bethesda, MD (Animal Study Protocol #MB-081).

Zebrafish maintenance
Experiments were performed with adult zebrafish from the brca2 hg5 and tp53 zdf1 mutant zebrafish lines carrying the brca2 Q658X [16] and tp53 M214K [18] mutations, respectively. All zebrafish evaluated in this study were related. For additional details, see Methods S1.

Histologic analyses
Zebrafish were processed for histology as previously described [16]. Tumors were classified based on histologic features, and the number of histologically and anatomically distinct tumors was determined for each specimen. Histologic diagnoses were made without knowledge of the brca2 genotype or loss of heterozygosity status. For additional details, see Methods S1.

LCM and DNA extraction from paraffin-embedded zebrafish tissues
Tissue sections were placed on PEN-membrane glass slides, and tumor and normal tissue specimens were individually collected by laser-capture microdissection. Collected specimens were routinely processed for DNA isolation (See Methods S1).

LOH analyses
LOH analyses were achieved by PCR amplification and sequencing over the brca2 Q658X and tp53 M214K mutation sites. See Methods S1, Figure S3, and Table S3 for details.

CpG island identification and pyrosequencing methylation detection assay
The zebrafish brca2 promoter has not been characterized, but a CpG island as determined by NCBI algorithm is predicted to occur 306 base pairs upstream of the 59 position of the translational start codon for brca2 on chromosome 15 (Danio rerio genome version Zv9). DNA isolated from LCM-collected samples was routinely processed for bisulfite conversion, PCR amplification, and Pyrosequencing analysis for methylation status of this CpG island (See Methods S1).

Statistics
Data sets comprised of the proportions of male or female animals that developed tumors were compared by Fisher's exact test (GraphPad Prism, version 6.0b), and P,0.05 was accepted to indicate statistical significance. Data sets comprised of age at tumor diagnosis were compared by unpaired t-test with Welch's correction (GraphPad Prism, version 6.0b), and P,0.05 was accepted to indicate statistical significance. Data sets comprised of the proportions of animals with malignant tumors were compared by Fisher's exact test with Mehta's modification (GraphPad Prism, version 6.0b), and P,0.05 was accepted to indicate statistical significance. For additional details on statistical comparisons, see Methods S1. Figure S1 Overall survival declines rapidly in brca2 m/ m;tp53+/m zebrafish. Kaplan-Meier survival curves for all tp53+/m zebrafish described in this study show that the survival curve for the brca2 m/m;tp53+/m cohort declined rapidly in comparison to brca2+/+;tp53+/m and brca2+/m;tp53+/m cohorts. (TIF) Figure S2 Tumorigenesis is enhanced by brca2 mutation. The percentage of zebrafish that developed tumors (benign or malignant) is higher in the brca2 m/m;tp53+/+ cohort than in brca2+/+;tp53+/+ or brca2+/m;tp53+/m cohorts. Table S1 Characteristics of tumor development in brca2+/+;tp53+/+, brca2+/m;tp53+/+, and brca2 m/ m;tp53+/+ zebrafish.