Identification of Recurrent Chromosome Breaks Underlying Structural Rearrangements in Mammary Cancer Cell Lines

Abstract

Cancer genomes are characterized by the accumulation of small-scale somatic mutations as well as large-scale chromosomal deletions, amplifications, and complex structural rearrangements. This characteristic is at least partially dependent on the ability of cancer cells to undergo recurrent chromosome breakage. In order to address the extent to which chromosomal structural rearrangement breakpoints correlate with recurrent DNA double-strand breaks (DSBs), we simultaneously mapped chromosome structural variation breakpoints (using whole-genome DNA-seq) and spontaneous DSB formation (using Break-seq) in the estrogen receptor (ER)-positive breast cancer cell line MCF-7 and a non-cancer control breast epithelium cell line MCF-10A. We identified concurrent DSBs and structural variation breakpoints almost exclusively in the pericentromeric region of chromosome 16q in MCF-7 cells. We fine-tuned the identification of copy number variation breakpoints on 16q. In addition, we detected recurrent DSBs that occurred in both MCF-7 and MCF-10A. We propose a model for DSB-driven chromosome rearrangements that lead to the translocation of 16q, likely with 10q, and the eventual 16q loss that does not involve the pericentromere of 16q. We present evidence from RNA-seq data that select genes, including SHCBP1, ORC6, and MYLK3, which are immediately downstream from the 16q pericentromere, show heightened expression in MCF-7 cell line compared to the control. Data published by The Cancer Genome Atlas show that all three genes have increased expression in breast tumor samples. We found that SHCBP1 and ORC6 are both strong poor prognosis and treatment outcome markers in the ER-positive breast cancer cohort. We suggest that these genes are potential oncogenes for breast cancer progression. The search for tumor suppressor loss that accompanies the 16q loss ought to be augmented by the identification of potential oncogenes that gained expression during chromosomal rearrangements.

Keywords: 16q loss; Break-seq; CNV; MCF-10A; MCF-7; ORC6; SHCBP1; breast cancer; double-strand breaks (DSBs); genome instability; spontaneous chromosome breaks; structural rearrangements.

Figure 1

Experimental design. Work flow for multi-dimensional genomic queries of the MCF-7 and MCF-10A cell lines for the identification of cancer-specific chromosome breakage sites and potentially impacted genes. The color-coded nodes denote the analytical steps utilizing the highlighted computational methods.

Figure 2

Break-seq analysis identifies cancer-specific chromosome breakage sites in the MCF-7 cell line. (A) Number of DSBs identified in each replicate experiment for both MCF-7 and MCF-10A. (B) Venn diagrams of consensus DSBs found in all four replicate experiments for MCF-7 and MCF-10A. (C) Examples of cancer-specific consensus DSBs in MCF-7 cells and not in MCF-10A cells. The genes proximal to the chromosome breaks are BCAR1, CKM, and DOK5, located on chromosome 16q, 19q, and 20q, respectively. (D) Distribution of DSBs overlapping genomic features in each of the five categories as indicated. (E) Distribution of DSBs per chromosome. Those chromosomes with the highest and lowest number of DSBs per Mb of DNA are marked by red and blue asterisks, respectively.

Figure 3

Structural variation and gene expression at the 16q pericentromere. (A) Break-seq profiles of all four replicate experiments in MCF-7 and MCF-10A cells on chr16. (B) Overlaid plots for DSB scores (top plot), DNA copy number (middle plot), and gene expression (bottom plot) for chr16. The DSB score and gene expression levels expressed as Log2 fold change (FC) in transcript level in MCF-7 over that in MCF-10A cells are plotted on the left, Y1, axis. The DNA copy numbers are plotted on the right, Y2, axis. (C) Expanded view of gene cluster immediately downstream of the pericentromeric region of 16q. FDR, false discovery rate.

Figure 4

Copy number variations. Copy number profiles for (A) MCF-10A and (B) MCF-7 cells. Copy number is expressed as Log₂ transformed normalized sequence read counts in 15 kilobasepair (kbp) segments across the autosomes. Copy number profiles were generated after correction for GC content and mappability, followed by segmenting using default parameters in QDNAseq.

Figure 5

Structural variations in MCF-10A and MCF-7 cell lines. (A) Circos displays of paired structural variation events detected by Socrates for MCF-10A (971 events) and MCF-7 (1334 events). Each chromosome is color-coded. Intra-chromosomal breakpoints are represented by the dome above the chromosome; the width of the dome corresponds to the number of events. Inter-chromosomal translocations are represented by ribbons connecting the two translocated chromosomes, with the thickness of the ribbon corresponding to the number of events. The bar graphs beneath the chromosome indicate the relative proportion of intra- (same color of the chromosome) and inter- (color of the connecting chromosome) chromosomal events. (B) Structural variants from paired chromosomal translocations were further classified into seven categories as indicated, and plotted as stacked column plots for each chromosome.

Figure 6

Survival prediction analysis of breast cancer patients. The survival function of progression-free survival (PFS) time is analyzed. (A) Kaplan–Meier (K–M) plots for patients with primary ER-positive breast cancer discriminated by ORC6 expression level into low- and high-risk groups (expression value cut-off = 548). (B) K–M plots for patients with primary ER-positive breast cancer discriminated by SHCBP1 expression level into low- and high-risk groups (expression value cut-off = 157). (C) K–M plots for patients with primary ER-negative breast cancer discriminated by ORC6 expression level into low- and high-risk groups. (D) K–M plots for patients with primary ER-negative breast cancer discriminated by SHCBP1 expression level into low- and high-risk groups. (E) K–M plots for patients with primary ER-positive breast cancer discriminated by ORC6 expression level into low- and high-risk groups (expression value cut-off = 373). Cohort treatment: endocrine therapy + neoadjuvant therapy. (F) K–M plots for patients with primary ER-positive breast cancer discriminated by SHCBP1 expression level into low- and high-risk groups (expression value cut-off = 106). Cohort treatment: endocrine therapy + neoadjuvant therapy. (G) K–M plots for patients with primary ER-positive breast cancer discriminated by ORC6 expression level into low- and high-risk groups. Cohort treatment: endocrine therapy + adjuvant therapy (expression value cut-off = 692). (H) K–M plots for patients with primary ER-positive breast cancer discriminated by ORC6 expression level into low- and high-risk groups (expression value cut-off = 205). Cohort treatment: endocrine therapy + adjuvant therapy. Higher risk (red color line) is associated with higher expression values.