Neuroblastoma (NB) is a tumor that arises from the sympathetic nervous system and accounts for approximately 10% of pediatric cancers and 15% of childhood deaths related to cancer [84
]. Genome-wide association studies (GWAS) in NB have shown that BARD1 acts as a tumor suppressor during its development and that certain variations in a single nucleotide have profound effects on BARD1 protein expression and NB susceptibility (Table 1
). Initially, Capasso et al. identified multiple SNPs in the BARD1
gene from blood samples of NB patients from European American and European populations [85
]. One particular SNP, rs6435862 T > G, which is located in intron 1 of the BARD1
gene and results in splicing of exon 2 and 3 and formation of BARD1β, was most significantly associated with susceptibility to NB. Their data was later replicated in African American, Italian, and Han Chinese populations [86
]. Studies also found that rs6435862 was associated with high-risk NB [85
], suggesting that BARD1β is an oncogene [90
]. In European American blood samples, this variant was present in stage IV, MYCN amplified NB and associated with diagnosis after age 1.5 years [86
]. The Italian cohort showed a similar trend in terms of age at diagnosis, high-risk classification, and MYCN status, but did not reach statistical significance. In the Han Chinese population, rs6435862 was associated with stage IV tumors and adrenal gland as the primary site while another study in Han Chinese showed additional associations with stage III disease and the onset of NB after one year of age in patients homozygous for the G risk allele [88
]. Collectively, these data strongly suggest that rs6435862, which results in the expression of the oncogenic BARD1β isoform, is associated with poor prognosis of NB.
Knockdown of BARD1β in NB cell lines, NLF and Nb-Ebc1, which contain rs6435862, resulted in significant inhibition of proliferation and colony formation [90
]. The enhanced proliferation and evasion of apoptosis in NB caused by this variant were through BARD1β interaction with and stabilization of Aurora kinases A and B. The mechanisms of BARD1β were independent of p53 and BRCA1-dependent HR because silencing of BARD1β did not alter the level of phosphorylated p53 and additional silencing of PARP1 was not lethal to NB cells. Intriguingly, selective serotonin reuptake inhibitors (SSRIs), citalopram and escitalopram, inhibited survival and induced apoptosis of NB cell lines regardless of MYCN status [91
]. It was proposed that the decrease in viability was caused by inhibition of BARD1β, which normally stabilizes Aurora kinase A, which then stabilizes MYCN, a major oncogenic driver in NB. These studies suggest that NBs due to the expression of oncogenic BARD1β can potentially be treated with Aurora kinase inhibitors.
Another intronic SNP associated with the development of NB is rs3768716. Combined tumor samples from the United Kingdom and the United States found that this variant was associated with high-risk NB with an odds ratio of 1.68 [85
]. A Chinese population with this variant also showed a significant risk of developing NB [88
], while a second Chinese study and an African American cohort exhibited increased tendency but did not reach significance [87
]. rs3768716 could indicate an aggressive disease phenotype since the SNP was associated with stage III and IV NB, origination at the adrenal gland and diagnosis after 12 months of age [88
One of the most common SNPs in the BARD1
gene associated with NB is rs17489363, which is located in the promoter region [88
]. This variant decreases the transcription of BARD1-FL. One of the rs17489363 variants, which converts C to T risk allele, was associated with high-risk NB among European Americans, African Americans, Italians, and Spaniards [92
]. High-risk NB had significantly decreased mRNA expression of BARD1-FL compared to intermediate- and low-risk NB [92
]. Knocking down of BARD1-FL in SHSY5Y and SKNSH, two human NB cell lines, led to increased viability and invasion, which is consistent with the tumor suppression function of BARD1-FL in NB. Heat shock factor 1 (HSF1) bound more strongly to the BARD1
gene with homozygous T risk allele of rs17489363 present in the SHSY5Y cell line compared to its homozygous C allele in the SKNAS cell line. rs6720708, another SNP in the BARD1
gene, has the strongest association with NB arising from the adrenal gland [94
]. Since it is in strong linkage disequilibrium with rs17489363, it is likely that rs17489363 is also associated with this site of origin. The second rs17489363 variant, which converts G to A risk allele, was found in an NB susceptible Chinese population [88
]. The A risk allele was associated with origination from the adrenal gland. Furthermore, clones containing the A risk allele decreased transcription of BARD1-FL by 1.2- to 4-fold compared to those containing the G allele. Taken together, the A and T risk alleles demonstrate similar phenotypes. rs17489363 predisposes to NB originating from the adrenal gland by allowing the binding of HSF1 to the BARD1
gene at the promoter region and reducing the transcription of BARD1-FL, thereby affecting its DDR function.
On the other hand, other SNPs in the BARD1
gene have been shown to be negatively associated with NB susceptibility and therefore may play a protective role. rs7585356, located downstream of the BARD1
gene in the 3′ untranslated region (UTR), resulted in overexpression of BARD1-FL [85
]. This variant was negatively associated with high-risk, stage IV NB as well as MYCN amplification, and more commonly diagnosed at an age of less than 18 months in European Americans [86
]. DNA taken from Italian children with NB showed a similar association with disease stage and age of diagnosis. Harboring at least one A allele at rs7585356 was significantly associated with decreased risk of NB in Chinese girls [89
]. rs7585356 by conversion of the G allele to an A allele overexpressed BARD1-FL and resulted in better clinical prognosis. rs1048108, located in exon 1 of the BARD1
gene near the RING domain, was also negatively associated with NB risk in European Americans, African Americans, Italians, and Chinese people [85
]. Pathway analysis indicated that rs1048108 repressed cellular development and regulated apoptosis [95
]. This variant showed no differences in its binding to BRCA1 compared to the wild-type counterpart, indicating that the protective role of the BARD1 SNP, rs1048108, was independent of BRCA1 [88
Ethnic disparities exist amongst NB prevalence and severity. In the United States, NB is more common in children of European descent; however, the high-risk disease is more prevalent in children of African descent [87
]. Latorre et al. identified BARD1 SNPs for the first time in African American patients that were not reported in European Americans. For example, rs16852804 was significantly associated with NB susceptibility, while rs7599060 was significantly associated with high-risk disease. Studies in a Chinese population identified a variant, rs3738888, that was unique to NB susceptibility in Han ethnicity [88
]. This indicates that ethnicity may play an important role in genetic susceptibility to NB. There are many other single nucleotide variants of BARD1 that are significantly associated with NB susceptibility (Supplemental Table S1
), however, the details regarding their role in cancer prognosis still need to be further investigated.
In summary, BARD1-FL acts as a tumor suppressor in the development of NB, primarily by regulating DDR, cellular proliferation, and programmed cell death. BARD1-FL promotes DDR through G2-M checkpoint arrest via downregulation of cyclin B and induction of apoptosis via phosphorylation of p53 [96
]. It is hypothesized that the cell cycle arrest is through the BARD1-BRCA1 heterodimer while cell death was BRCA1-independent, but the mechanisms were not directly tested. SNPs in the BARD1
gene have both oncogenic and tumor-suppressing roles. In support of BARD1 as a tumor suppressor, rs7585356 leads to the overexpression of BARD1-FL while rs1048108 does not affect BRCA1 binding. Both SNPs are associated with a decreased risk of NB. Reduced expression of BARD1-FL (rs17489363) and formation of the oncogenic BARD1β isoform (rs6435862) increase susceptibility to NB and correlate with poor prognostic factors such as stage III and IV disease, older age at diagnosis and tumor originating in the adrenal gland. BARD1β affects DDR in a BRCA1-independent mechanism. The proliferation of NB cell lines and xenograft tumors with mutations or deletions in DNA repair genes, including BARD1, were inhibited by the PARP inhibitor (PARPi), Olaparib [97
]. Thus, SNPs leading to oncogenic isoforms or decreased activity of BARD1-FL result in dysregulation of DDR, and those patients could benefit from treatment with Aurora kinase inhibitors or PARP inhibitors.
4.3. Non-Small Cell Lung Cancer
Lung cancer is the primary cause of cancer-related death worldwide [121
]. Lung cancer is subdivided into two main categories: small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC). NSCLC is comprised of adenocarcinoma, squamous cell carcinoma, and large cell carcinoma. Zhang et al. identified nearly all isoforms (β, γ, δ, ϕ, ε, η, κ, π) of BARD1 from RNA extracted from 20 matched NSCLC tumor and normal adjacent tissue samples [18
]. BARD1-FL and all other detected isoforms were increased in tumors compared to neighboring normal lung tissue with the π isoform being the most significantly upregulated in tumors. BARD1 was undetected in normal healthy controls, suggesting that BARD1 and its isoforms play a key role in cellular transformation and tumor progression in the lung. Stratification by sex indicated that β and κ isoforms were expressed higher in males while the η, γ, and ε isoforms were higher in females but did not reach statistical significance. Additionally, the lncRNA BARD1 9′L, which stimulates the expression of BARD1-FL and isoforms β, δ, and γ through interaction with the 3′ UTR of BARD1
, was significantly upregulated in lung tumors compared to paired cancer-free lung tissue as well as in the human lung adenocarcinoma cell line, A549 [103
IHC staining analyses of 54 NSCLC tumors using antibodies that detect antigens mapped to exon 1 and exon 11 of BARD1 showed the highest expression of BARD1 in large cell carcinoma followed by squamous cell carcinoma then adenocarcinoma [122
]. Staining patterns varied and rarely showed positivity for all antibodies used, indicating the presence of different isoforms in NSCLC [18
]. An additional 100 NSCLC tumors were stained with antibodies recognizing antigens mapped to exons 1, 3, 4, and 11 [18
]. Staining for regions corresponding to exons 1 and 11 correlated and indicated the presence of ϕ, δ, and ε isoforms. The π isoform was also detected as staining for regions mapped to exons 3 and 4 also correlated. Expression of either exon 3, or beginning of exon 4, or both, present in isoforms β, κ, and π, was associated with decreased disease-free survival and overall survival. These isoforms were also the most prevalent and more common in squamous cell and large cell carcinomas compared to adenocarcinoma. When BARD1β was overexpressed in A549 cells, it increased cellular proliferation, inhibited apoptosis, and increased the expression of fibronectin, an inducer of EMT, compared to its BARD1-FL expressing counterparts [123
]. Intriguingly, the expression of BARD1 isoforms did not correlate with tumor grade or stage [18
]. However, induction of lung adenocarcinomas in mice led to the detection of staining patterns positive for antigens corresponding to exon 1 in early disease and exons 3 and 4 in the later disease that were higher in tumors compared to adjacent tissues [18
]. Thus, the π isoform is likely involved in tumor progression and an aggressive phenotype.
Since the BARD1 isoforms are highly expressed in NSCLCs, Pilyguin et al. proposed that the detection of antibodies in patients’ sera against the tumor-associated antigens present on the BARD1 isoforms could be used to predict the presence of lung cancer [124
]. They were able to distinguish between those with lung cancer and healthy individuals with high sensitivity and specificity with a model using 27 antigens and patients’ sera. The same 27 antigens model did not predict NB, ovarian, or breast cancer based on a sensitivity of at least 0.90, although some samples were detected. In addition, applying the ten highest expressed peptides to stages I-III or stage IV disease showed no differences. Therefore, the simple detection of antibodies against BARD1 isoforms in blood sera can accurately predict the presence of lung cancer, regardless of stage.
One proposed treatment modality for patients with NSCLC is the combination of neddylation and PARP inhibitors [125
]. The neddylation E1 inhibitor, MLN4924, decreased BARD1 intensity following laser ablation. The combination treatment of MLN4924 and Olaparib for 72 h significantly inhibited the proliferation of A549 and H1299 NSCLC cells. Irradiation following the combination treatment resulted in defective DNA repair, as indicated by a significantly increased number of cells with γH2AX foci. Cell death was induced in NSCLCs by a combination of PARP and neddylation inhibitors, which caused and prevented repair of DSBs, respectively.
IHC staining of NSCLC tumors showed that the expression of BARD1 did not correlate with that of BRCA1 or p53 [18
], suggesting that disruption of BARD1’s protein-protein interactions plays a role in tumorigenesis. Aurora kinase B was also upregulated in lung tumors [18
]. Although the molecular pathways activated or inhibited by the BARD1 oncogenic isoforms in NSCLC need to be further investigated, inhibition of these mutated proteins, especially isoforms β, κ, and π, may drastically inhibit tumor progression.