Transcriptional consequences of genomic structural aberrations in breast cancer

Abstract

Using a long-span, paired-end deep sequencing strategy, we have comprehensively identified cancer genome rearrangements in eight breast cancer genomes. Herein, we show that 40%-54% of these structural genomic rearrangements result in different forms of fusion transcripts and that 44% are potentially translated. We find that single segmental tandem duplication spanning several genes is a major source of the fusion gene transcripts in both cell lines and primary tumors involving adjacent genes placed in the reverse-order position by the duplication event. Certain other structural mutations, however, tend to attenuate gene expression. From these candidate gene fusions, we have found a fusion transcript (RPS6KB1-VMP1) recurrently expressed in ∼30% of breast cancers associated with potential clinical consequences. This gene fusion is caused by tandem duplication on 17q23 and appears to be an indicator of local genomic instability altering the expression of oncogenic components such as MIR21 and RPS6KB1.

Figure 1.

Overview of experimental framework to identify fusion transcripts in breast cancer genomes. (FG^R) Transcripts in which exons from two distinct RefSeq genes are fused together in the same direction. (3′T-E^R) Fusion transcripts in which the 3′-terminus portion of a given 5′ partner gene is truncated and fused to a non-RefSeq but annotated segment that has evidence for being part of a transcript. (3′T-N^R) Any genomic segment that is in the anti-sense strand of any known gene/transcript or in an unannotated region that has no evidence for a transcript in current databases (see Supplemental Glossary).

Figure 2.

Different structure variation types seen in all genomic structure abnormalities and in only those giving rise to fusion transcripts in breast cancer genomes. Fusion transcripts detected by RNA-PET and validated by RT-PCR (top). Fusion transcripts (FG^R + 3′T-E^R) identified through the RT-PCR screening in three cell lines (middle) and five primary tumors (bottom). (Del) Deletion; (TD) tandem duplication; (U-Inv) unpaired-inversion; (Transloc) isolated translocation; (Inv) inversion; (Ins-Intra) intrachromosomal insertion; (Ins-Inter) interchromosomal insertion; (Complex-Intra and Complex-Inter) intra- and interchromosomal connections in hot spot of genome breakpoints (super cluster size ≥3) (Hillmer et al. 2011).

Figure 3.

Translational index of fusion transcripts. (A) The figure depicts a typical sucrose gradient fractionation profile demonstrating the separation between translationally active polysomal RNA and the nontranslated monosomal RNA. The number of the ribosomes associated is an indication of its translational potential. The table below shows the numbers of candidates, validated transcripts, and results for the polysomal assay in MCF-7. Candidates include fusion point/splicing variants. ORF structures of each category are explained in Supplemental Figure 3A. The definition of translational index is given in Methods. (B) The fraction of translational index (Low, Medium, High) in each category showing a high translational index for in-frame fusion genes.

Figure 4.

Expression of recurrent fusion gene transcript (RPS6KB1-VMP1) in breast cancer. (A) Structure of the fusion gene transcript created by a tandem duplication (TD) in the 17q23 locus in the MCF-7 genome. The genome DNA rearrangement point detected by DNA-PET data was validated by genomic PCR and sequencing. Expression of the fusion transcript in breast cancer cell lines (B) and in primary tumor samples (C) determined by RT-PCR. (D) Fusion pattern of the transcripts in primary tumor samples showing a range of exons included in the fusions. (E) Correlation of the fusion expression and disease-free survival (DFS) in breast cancer patients. Sixty patients with available information were separated into two groups (18 fusion [+] and 42 fusion [−]) and disease-free survival events were analyzed. Expression of the fusion showed a trend toward correlation (P = 0.06) with poor prognosis of the patients. Correlation of the expression of the fusion with those of genes involved in the TD (F) and with those of neighboring genes (G) within a 3-Mb window from the TD in breast cancer primary tumors. Sixty-eight patient samples with available information were separated into 21 fusion (+) and 47 fusion (−), and the expression of the genes were compared. P-values (<0.05) for the difference of each gene expression between the fusion (+) and (−) samples are shown on each gene location. PTRH2 (P = 0.0010), PPM1D (P = 0.0016), C17orf71 (P = 0.0031), MRPS23 (P = 0.0039), and RNFT1 (P = 0.0042) showed significant correlation as well as TUBD1 and RPS6KB1.