Integrative analyses reveal signaling pathways underlying familial breast cancer susceptibility

Abstract

The signaling events that drive familial breast cancer (FBC) risk remain poorly understood. While the majority of genomic studies have focused on genetic risk variants, known risk variants account for at most 30% of FBC cases. Considering that multiple genes may influence FBC risk, we hypothesized that a pathway-based strategy examining different data types from multiple tissues could elucidate the biological basis for FBC. In this study, we performed integrated analyses of gene expression and exome-sequencing data from peripheral blood mononuclear cells and showed that cell adhesion pathways are significantly and consistently dysregulated in women who develop FBC. The dysregulation of cell adhesion pathways in high-risk women was also identified by pathway-based profiling applied to normal breast tissue data from two independent cohorts. The results of our genomic analyses were validated in normal primary mammary epithelial cells from high-risk and control women, using cell-based functional assays, drug-response assays, fluorescence microscopy, and Western blotting assays. Both genomic and cell-based experiments indicate that cell-cell and cell-extracellular matrix adhesion processes seem to be disrupted in non-malignant cells of women at high risk for FBC and suggest a potential role for these processes in FBC development.

Keywords: breast cancer; cellular adhesion; disease susceptibility; multiomic analysis, signaling pathways.

Figure 1. Flow chart illustrating the experimental design of this study

1. We used pathway‐based analytic approaches to identify biological processes that may be disrupted in women who develop familial breast cancer (FBC). Having collected genomewide data, we filtered the data to include only genes associated with a given pathway. For each pathway, we identified differences between individuals who developed FBC and those who did not, using either the Support Vector Machines algorithm (gene expression data) or Barnard's exact test (DNA variant data). We considered the most statistically significant pathways to be candidates for further investigation.

2. We profiled peripheral blood mononuclear cells using gene expression microarrays and exome sequencing and identified pathways that were consistently significant across these data sets. To reduce the chance that our findings were influenced by treatment effects, we excluded pathways that showed significant differences between familial and non‐familial controls.

3. For the remaining pathways, we identified those that showed significant differences in two gene expression data sets representing primary mammary epithelial cells. To validate these findings, we used cell‐based assays and fluorescence microscopy to profile an additional collection of normal breast cells.

Figure 2. Overview of top pathways for which gene expression levels and mutation status differed significantly between FBC women and controls in the Utah and Ontario cohorts

1. Biological processes associated with pathways that attained a rank P‐value < 0.05.

2. Heatmaps show median expression levels for Utah and Ontario women, respectively, who developed FBC and for women who did not. Only genes that exhibited a consistent fold change across the cohorts are shown.

3. Per‐sample DNA variants observed in these pathways are shown. Black dots indicate samples that carried a likely pathogenic in the genes that are shown. Only genes for which at least one variant was observed are shown.

Figure 3. Summary of pathway‐level results that included non‐FBC controls and normal breast gene expression data

1. Cross‐validated estimates that each patient from the Lim et al and Bellacosa et al cohorts had a family history of breast cancer and/or carried a BRCA1/2 mutation. These estimates were derived from genes in the REACTOME Integrin Cell Surface Interactions pathway.

2. Estimates for the same patients using genes from the KEGG Small cell lung cancer pathway.

Data information: The boxes represent the the interquartile range of the “Genomic model score” values. The whiskers extend the the most extreme data points.

Figure EV1. Western blots of selected proteins

Cell lysates obtained from cultures of primary mammary epithelial cells for women having normal breast reduction (control) or prophylactic breast reduction (high‐risk women) surgeries were immunoblotted with FAK (PTK2), ITGA6, ICAM2, p53, PTEN, ITGA IV, ITGA V, VTN, F‐actin, and β‐actin (loading control). Two lysates were loaded to each gel to compare between blots. Sample names shown in black are from high‐risk women; sample names shown in blue are from normal breast reduction samples; sample names shown in red are from breast reduction normal controls placed on every gene to serve as internal loading/processing controls.

Figure 4. Fluorescence microscopy images of primary breast epithelial cultures for FBC women and controls

1. Primary (non‐malignant) mammary epithelial cells from breast reduction patients with no known family history of breast cancer and from prophylactic mastectomy patients who had a breast cancer family history (“high risk”) were cultured on glass slides for 3–5 days and subsequently fixed and stained for F‐actin (red, phalloidin), focal adhesions (green, vinculin) and nuclei (blue, Dapi). Shown are five cell populations (two fields each) from each group that had been identified in a blinded manner as having distinctive cell phenotypes. Scale bar = 50 μm.

2. Box plots showing the results of quantitative comparisons for all microscope fields (n = ˜10) from each of the samples shown. Samples from high‐risk patients were more spread apart and expressed higher levels of F‐actin. See Results for details about how quantitative metrics were derived. The boxes represent the the interquartile range of the respective values. The whiskers extend the the most extreme data points.

Figure 5. Cell‐based assays show aberrant cell adhesion in normal breast epithelial cells for women with a high risk of breast cancer

1. A cell adhesion assay was used to compare extracellular matrix adhesiveness in normal, primary breast cells in women who did or did not have a family history of breast cancer. Cells from women who had a family history of breast cancer were significantly less adherent than cells from women who did not have a family history of breast cancer.

2. A drug‐response assay was used to evaluate responsiveness to PF573228, a FAK inhibitor in μm concentrations. Normal, breast epithelial cells were obtained from high‐risk women who had undergone prophylactic surgery and compared against cells from women who had undergone non‐risk‐related, breast reduction surgery. The prophylactic samples were significantly more sensitive to PF573228 than breast reduction samples. Response values indicate the drug concentration that induces a response that reaches half of its maximal effect.

Data information: The boxes represent the the interquartile range of the respective values. The whiskers extend the the most extreme data points.

Figure EV2. Gefitinib and afatinib assay results

1. A, B

   Primary breast cells were treated with the (A) EGFR inhibitor gefitinib and (B) tyrosine kinase inhibitor afatinib. Similar responses were observed for women who had a family history of breast cancer and for women who did not have a family history of breast cancer. The boxes represent the the interquartile range of the respective values. The whiskers extend the the most extreme data points.

Figure EV3. Overview of criteria used to filter exome‐sequencing variants

Variants were filtered based on frequency, location within protein‐coding regions, conservation, and effect on protein sequence. Variants were also collapsed to gene‐level values before pathway‐level comparisons were performed. All statistics listed on this diagram are per sample.

Figure EV4. Selection of threshold used to filter frequently mutated genes

Genes that were mutated in a relatively high number of germline samples in TCGA were excluded from the pathway‐level mutation analyses. We calculated the number of genes that would be excluded for thresholds ranging between 0.2% and 10%. A threshold of 1.8% was selected based on the maximal difference in number of excluded genes.

Figure EV5. Relationship between variant status and gene expression for 34 samples profiled using gene expression microarrays and exome sequencing

We sought to identify genes whose expression levels correlated strongly with mutation status. This figure shows data for 373 genes that exhibited the strongest association between the presence of one or more potentially pathogenic variants and expression of the same gene. Red dots indicate samples that carried a mutation within a given gene. Gray dots indicate samples that did not carry a mutation within the same gene. In many cases, germline variants are correlated with a considerable increase or decrease of expression levels for the same gene. Expression values for each gene are standardized to a consistent scale for illustration purposes.