End-sequence profiling: sequence-based analysis of aberrant genomes

Abstract

Genome rearrangements are important in evolution, cancer, and other diseases. Precise mapping of the rearrangements is essential for identification of the involved genes, and many techniques have been developed for this purpose. We show here that end-sequence profiling (ESP) is particularly well suited to this purpose. ESP is accomplished by constructing a bacterial artificial chromosome (BAC) library from a test genome, measuring BAC end sequences, and mapping end-sequence pairs onto the normal genome sequence. Plots of BAC end-sequences density identify copy number abnormalities at high resolution. BACs spanning structural aberrations have end pairs that map abnormally far apart on the normal genome sequence. These pairs can then be sequenced to determine the involved genes and breakpoint sequences. ESP analysis of the breast cancer cell line MCF-7 demonstrated its utility for analysis of complex genomes. End sequencing of approximately 8,000 clones (0.37-fold haploid genome clonal coverage) produced a comprehensive genome copy number map of the MCF-7 genome at better than 300-kb resolution and identified 381 genome breakpoints, a subset of which was verified by fluorescence in situ hybridization mapping and sequencing.

Fig. 1.

Genome analysis of MCF-7 using ESP. (A) Schematic representation of the generated types of data. End-sequence density plotted as a function of distance along the genome shows regions of copy number increase and decrease. Horizontal lines connecting BES pairs show regions of the genome that are within ≈141 kb in the test genome. Abnormally long lines indicate regions of the test genome joined by structural changes. Abnormalities that can be detected with this approach include translocations, inversions, deletions, and complex structures that develop during gene amplification. (B) End-sequence density and end-sequence pair plots for the MCF-7 genome. A total of 8,320 end-sequence pairs were mapped onto the normal human genome sequence (represented as a horizontal line along the top). The dark green plot represents the number of end sequences per 1-Mb interval. BAC end pairs with ends mapping to different chromosomes are shown as horizontal red lines. BAC clones with ends in the wrong orientation (not pointing toward each other) are shown in blue. BAC clones with ends mapping >3 SDs farther apart than the average BAC insert are shown in green. Ambiguously mapped end sequences are shown in purple. Blue arrows indicate BAC clones linking inversions and translocations validated by FISH, and red arrows indicate BAC clones detecting complex structural rearrangements associated with gene amplification that were confirmed by FISH and sequencing (see Figs. 3 and 4). (C) Expanded view of chromosome 20. Plot symbols and annotation are as described for B. Copy number data are presented in 100-kb windows. (D) Enlarged view of chromosome 11. Plot symbols and annotation are as described for B.

Fig. 2.

Comparison of measurements of genome copy number along chromosome 20q in MCF-7 cells made by using CGH array and end-sequence density analysis. CGH array was comprised of minimal tiling path BAC clones along 20q. End-sequence density plots were generated by enumerating the numbers of end sequences in 1-Mb-wide bins. Blue diamonds linked by solid blue lines indicate CGH measurements. Pink lines show end-sequence densities measured by using ESP. The concordance between the two assay techniques is remarkable.

Fig. 3.

FISH-based validation of genome rearrangements identified by ESP. Complete metaphase images can be viewed as Figs. 6-16, which are published as supporting information on the PNAS web site. (A) FISH analysis of the hybridization of BAC clone MCF7_1–5H15 to normal human metaphase chromosomes confirms the ESP analysis indicating that this clone connected chromosomes 15q11.2 and 16q22.2. This clone also hybridized to chromosome 1, suggesting a more complex rearrangement (data not shown). (B) FISH analysis of the hybridization of BAC clone MCF7_1–9I10 to normal human metaphase chromosomes. The FISH analysis confirms the ESP determination that this clone joins 11p11.2 to 11q14.3. (C) Hybridization of BAC clone MCF7_1–9I10 to MCF-7 metaphase chromosomes. This clone hybridizes to five chromosomes. (D) Hybridization of BAC clone MCF7_1–5K16 to normal human metaphase chromosomes. The FISH analysis confirms the ESP determination that this clone joins 9q22.3 and 9q34.1. The distal breakpoint is located within the first intron of the ABL oncogene. (E) FISH analysis of the hybridization of BAC MCF7_1–1A11 to normal human metaphase chromosomes. This BAC was determined by ESP to connect amplicons on 20q13.2 and 17q23. Hybridization to 17q23 and 20q13.2 confirms the ESP analysis. (F) Hybridization of BAC MCF7_1–1A11 to MCF-7 metaphase chromosomes reveals multiple loci of amplification. (G) Hybridization of BAC clone MCF7_1–3F5 to MCF-7 metaphase chromosomes detects high-level amplification. ESP mapping shows that this clone has one end sequence at the ZNF217 locus at 20q13.2 and another at 3p14. FISH with BAC MCF7_1–3F5 on normal metaphase chromosomes confirmed the ESP analysis (data not shown). (H) Dual-color FISH using normal BACs spanning BMP7 (red) and ZNF217 (green) to MCF-7 metaphase chromosomes. Yellow FISH signals show coamplification and colocalization of these loci in MCF-7 genome.

Fig. 4.

A graphical representation of the structural organization of BAC clone MCF7_1–3F5 selected to contain ZNF217 at 20q13.2. Red arrows show end sequences 3p14 and 20q13.2. Sequence-tagged site content mapping localized the 5′ region of the BMP7 gene within the BAC. This is 1 of 26 independent clones in the library juxtaposing ZNF217 and BMP7, and 1 of 4 that also contain end sequences in a region of amplification at 3p14. Full sequence analysis of MCF7_1–3F5 identified five widely separated chromosomal regions fused together in the orientations shown. Only ZNF217 is structurally intact. PTPRT, BMP7, and L39 all are truncated. PTPRT intron 6 is fused to BMP7 intron 1 in opposing polarity. L39 is fused to nontranscribed DNA 3′of ZNF217. A large CpG island shared by BMP7 and L39 is structurally intact. genscan and fgenes predict at least two novel genes created by these genome rearrangements (blue arrows). (B) Sequences spanning each genome breakpoint are presented with the fusion site in red. (C) Genome cryptographer (19) plot of the density and classification of repetitive elements (Alu, elements red; L1, green; LTR, blue, etc., details in ref. 19) at each breakpoint shown for each 150-bp window on a scale of 0–100%. Blue arrows across the top show position and orientation of the normal genome reference sequence. Thus, breakpoints 1, 2, and 4 occur in regions of very high repetitive element density, whereas breakpoint 3 occurs in single-copy DNA.