Gene expression variability in breast tumour development

Abstract

Breast cancer is the most common female malignancy and presents a major health issue both worldwide and in New Zealand. Individuals with a family history of either breast or ovarian cancer have an increased lifetime risk of breast cancer. Furthermore, approximately 10% of breast cancers are thought to be caused by rare inherited genetic variants, of which variants in BRCA1 and BRCA2 are the most prominent. Individuals carrying a pathogenic BRCA1 or BRCA2 variant have up to a 87% and 84% lifetime risk of developing breast cancer compared to 12% for the general population. BRCA1 and BRCA2 functions are critical to genomic stability, which is essential to maintaining a healthy cell state in all tissues. However, there is a lack of disease risk in the majority of non-breast and ovarian tissues. Additionally, there are thousands of different sequence variants across each gene that present distinct risk profiles ranging from benign to pathogenic. A large proportion of these are considered variants of uncertain significance (VUS) resulting in no additional information to aid clinical management. The effect of pathogenic variants on gene expression is a potential source of information that can help to identify genes involved in tissue-specific risk and a phenotype that can be used to help classify VUS. Previous studies have explored BRCA1- and BRCA2-associated gene expression profiles, however, there has been a lack of consistency between genes identified as associated with BRCA1- and BRCA2 variant status. These studies have typically focused on identifying differences by comparing the mean level of gene expression, however, the variability of gene expression is also under genetic control and has been under-explored in BRCA1- and BRCA2-associated tissues. BRCA1 and BRCA2-associated gene expression variability was calculated across three familial breast tumours datasets. Additionally, as BRCA1-associated tumours are typically subtyped as basal-like, gene expression variability was calculated between basal and non-basal breast tumours in four datasets. BRCA1- associated and basal-like tumours exhibited greater global gene expression variability compared to familial breast tumour with no BRCA1 or BRCA2 pathogenic variant (BRCAx) and non-basal tumours. By comparison, the mean level of gene expression was similar between all tumour-types. Three genes (DSC3, EN1 and IGF2BP3) were found to be differentially variable in BRCA1-associated and basal-like breast tumours compared to BRCAx and non-basal tumours, respectively. RNA in situ hybridisation (RNA ISH) was used to explore EN1 expression in 503 breast tumour cores, including 151 BRCA1-associated tumours. Consistent with mircoarray analyses, RNA ISH found the expression of EN1 was variable in BRCA1-associated tumours. Two breast cell lines were used to measure gene expression variability in BRCA1 heterozygous models. In contrast to tumour analysis, MCF-10A BRCA1mut/+ cells displayed less global gene expression variability compared to wild type MCF-10A cells. Additionally, BRCA1 knock-down in MCF-7 cells showed no change in gene expression variability. The knock-down of BRCA1 was, however, associated with a decrease in luminal and an increase in basal marker expression. To further explore these observations, CRISPR-Cas genome engineering was used to generate a MCF-7 cell that harboured a BRCA1 pathogenic variant. Similar to BRCA1 knock-down models, MCF-7 BRCA1 mutant cells exhibited a decrease in expression of the luminal marker ESR1. However, immunocytochemistry was unable to detect the basal maker CK5/6 in any of the MCF-7 mutant clones. Rare high risk breast cancer variants displayed an association between genotype and gene expression variability. Therefore, I also investigated the relationship between common risk variants and gene expression variability by conducting variable expression quantitative trait loci analysis (veQTL). This analysis showed that of the 181 published common breast cancer risk variants, 27 had breast-specific veQTL associations with 60 genes. One variant, rs11075995, was associated with four genes (CYP11B1, CYP17A1, HSD3B2 and STAR), all of which were involved in C21-steroid biosyntheisis. These studies successfully demonstrated the potential utility of gene expression variability analysis to identify candidate breast cancer risk associated genes. The application of this approach may lead to a better understanding of the mechanisms involved in the development of disease. Furthermore, these methods may transform how researchers interpret the results form published and future genome-wide association studies.