Rapid high-resolution mapping of balanced chromosomal rearrangements on tiling CGH arrays

Abstract

The diagnosis and classification of many cancers depends in part on the identification of large-scale genomic aberrations such as chromosomal deletions, duplications, and balanced translocations. Array-based comparative genomic hybridization (array CGH) can detect chromosomal imbalances on a genome-wide scale but cannot reliably identify balanced chromosomal rearrangements. We describe a simple modification of array CGH that enables simultaneous identification of recurrent balanced rearrangements and genomic imbalances on the same microarray. Using custom tiling oligonucleotide arrays and gene-specific linear amplification primers, translocation CGH (tCGH) maps balanced rearrangements to ∼100-base resolution and facilitates the rapid cloning and sequencing of novel rearrangement breakpoints. As proof of principle, we used tCGH to characterize nine of the most common gene fusions in mature B-cell neoplasms and myeloid leukemias. Because tCGH can be performed in any CGH-capable laboratory and can screen for multiple recurrent translocations and genome-wide imbalances, it should be of broad utility in the diagnosis and classification of various types of lymphomas, leukemias, and solid tumors.

Figure 1

Overview of translocation-comparative genomic hybridization (tCGH). In a standard tCGH experiment, tumor and reference DNA are amplified in parallel using identical linear amplification primers and conditions. Amplified tumor and reference genomic DNA are enzymatically labeled with Cy5-dUTP or Cy3-dUTP, respectively, and are cohybridized to a tiling-density oligo array representing potential fusion partners. Cy3-labeled reference amplicons (green) bind exclusively to probes derived from the primary (1°) fusion partner. If a rearrangement is present, the Cy5-labeled tumor amplicons (red) span the rearrangement breakpoint (upper left) and thus bind at their 5′ ends to primary fusion partner probes (lower right) and at their 3′ ends to secondary (2°) fusion partner probes (upper right). If no rearrangement is present, both tumor-derived and reference-derived amplicons in the primary fusion partner are similar and there is no net change in the Cy5/Cy3 ratio and no visible signal on the log-ratio plots. Likewise, tumor- and reference-derived off-target amplicons tend to be similar and therefore are invisible on log-ratio plots unless mock-amplified reference DNA is used (as described under Results). In general, rearrangement breakpoints can be distinguished from true genomic imbalances based on their characteristic size range of several kilobases and their asymmetry when visualized on log-ratio plots. Breakpoint profiles in primary fusion partners tend to have lower amplitudes than secondary partner profiles, because of the presence of additional tumor-derived amplicons derived from the nonrearranged allele of the primary fusion partner.

Figure 2

Identification of balanced IGH@ translocations. A: Recurrent translocations in B-cell lymphomas and plasma cell myeloma pair the IGH@ locus with various partner oncogenes. B–F: Balanced IGH@ translocations involving the BCL2, MYC, and CCND1 loci. Balanced IGH@-BCL2 translocation in lymphoma cell line DHL16 identified by linear amplification with the consensus J_H primer (+J_H) (B) or multiplex linear amplification using a mixture of D_H primers (+D_H) (C). J_H-primed amplification also identifies a known IGH@-MYC translocation in MC116 (D) and two novel IGH@-CCND1 translocation breakpoints in primary mantle cell lymphomas, located 34 kb and 175 kb upstream of the CCND1 gene on chromosome 11 (E and F). Vertical lines indicate translocation breakpoints and arrowheads indicate the direction of linear amplification.

Figure 3

Simultaneous identification of balanced IGH@ translocations and genomic imbalances in lymphoma cell lines RL7 (A, C, and E) and MO2058 (B, D, F, and G). A and D: Cohybridization of J_H-amplified cell line DNA with J_H-amplified normal DNA (+J_H) identifies both IGH@ translocations and genomic imbalances, including an IGH@-BCL2 breakpoint and a 167-kb deletion [dashed vertical lines (A, C, and E)] in RL7 (A) and, in MO2058 (B), an IGH@-CCND1 breakpoint at the MTC, a known 1.9-kb deletion in the CCND1 3′ untranslated region (3′UTR) [open arrowhead (B and D)], and a previously unidentified duplication extending from the IGH@-CCND1 breakpoint through the CCND1 gene toward the 11q telomere. C and D: Cohybridization of mock-amplified (−) RL7 DNA (C) or mock-amplified (−) MO2058 DNA (D) with mock-amplified (−) normal DNA reveals the genomic imbalances but not the balanced translocations. E and F: Cohybridization of J_H-amplified and mock-amplified (−) RL7 DNA (E) or MO2058 DNA (F) reveals the IGH@ translocations but not genomic imbalances. G: Cohybridization of J_H-amplified and mock-amplified (−) normal DNA reveals off-target linear amplification products that are reproducible (arrows in panels F and G; see also Supplemental Figure S6D at http://jmd.amjpathol.org) and that vary with the amplification primer (data not shown). Solid vertical lines indicate translocation breakpoints; solid arrowheads indicate the direction of linear amplification.

Figure 4

Identification of complex IGH@ rearrangements in OCI-Ly8 (Ly8). A and B: Cohybridization of J_H-amplified OCI-Ly8 (A) or D_H-amplified OCI-Ly8 (B) with identically amplified normal DNA reveals an unbalanced IGH@-BCL2 MBR translocation, a 6.3-kb deletion between the J_H and D_H breakpoints, and a distinct 135-kb intronic BCL2 deletion. The lower amplitude of the intronic deletion suggests that it is present in only a subset of OCI-Ly8 cells. Dashed vertical lines indicate the intronic deletion breakpoints. C–E: SγF-primed amplification reveals a previously unidentified Sγ3-BCL6 fusion (C), and SγR amplification identifies both the reciprocal Sγ3-BCL6 fusion (E) and a novel Sγ2-MYC fusion (D) that appears to represent a balanced rearrangement based on the absence of duplications or deletions at the MYC breakpoint. F and G: The Sγ3-BCL6 (G) and Sγ2-MYC (F) fusions were independently identified using linear amplification primers targeting translocation breakpoint clusters in intron 1 of the BCL6 gene and upstream of MYC exon 1, respectively. Solid vertical lines indicate translocation breakpoints and solid arrowheads indicate the direction of linear amplification (A–G).