A novel chromatin immunoprecipitation and array (CIA) analysis identifies a 460-kb CENP-A-binding neocentromere DNA

Abstract

Centromere protein A (CENP-A) is an essential histone H3-related protein that constitutes the specialized chromatin of an active centromere. It has been suggested that this protein plays a key role in the epigenetic marking and transformation of noncentromeric genomic DNA into functional neocentromeres. Neocentromeres have been identified on more than two-thirds of the human chromosomes, presumably involving different noncentromeric DNA sequences, but it is unclear whether some generalized sequence properties account for these neocentromeric sites. Using a novel method combining chromatin immunoprecipitation and genomic array hybridization, we have identified a 460-kb CENP-A-binding DNA domain of a neocentromere derived from the 20p12 region of an invdup (20p) human marker chromosome. Detailed sequence analysis indicates that this domain contains no centromeric alpha-satellite, classical satellites, or other known pericentric repetitive sequence motifs. Putative gene loci are detected, suggesting that their presence does not preclude neocentromere formation. The sequence is not significantly different from surrounding non-CENP-A-binding DNA in terms of the prevalence of various interspersed repeats and binding sites for DNA-interacting proteins (Topoisomerase II and High-Mobility-Group protein I). Notable variations include a higher AT content similar to that seen in human alpha-satellite DNA and a reduced prevalence of long terminal repeats (LTRs), short interspersed repeats (SINEs), and Alus. The significance of these features in neocentromerization is discussed.

Figure 1

20p12 BAC/PAC array. (A) Ideogram of the invdup(20p) marker chromosome showing the inverted duplication breakpoint in band 20p11.2 (arrow) and the activation of a neocentromere at 20p12. p′ and q′ indicate the short and long arms of the marker chromosome relative to the neocentromere, respectively. (B) BAC/PAC array constructed from public databases around the neocentromere. Distances (in Mb units) within the array are plotted relative to PAC dJ811H13 on the p′ arm of the marker chromosome. The size of each horizontal bar is proportional to the corresponding insert size of the BAC or PAC. Addresses of the BAC/PACs are listed on the right. The prefix “dJ” denotes a PAC while “bA” indicates a BAC. PACs dJ1009E24 and dJ905G11 map outside the array and are indicated by broken lines. BAC/PACs that were not used in the CIA analysis (see Methods) are shown in italics and recessed. Shaded box represents the CENP-A binding region (see Results).

Figure 2

Immunofluorescence/FISH analysis of 20p12 neocentromere. Immunofluorescence was performed using CREST#6 (red) to mark the position of the neocentromere (solid arrow). FISH was performed using PACs dJ1009E24 (A), dJ811H13 (B), dJ727I10 (C), dJ742J24 (D), and dJ905G11 (E). Two sets of green FISH signals are observed on the marker chromosome invdup (20p) because of the inverted duplication of the probed segment; the signal set at the distal q′ arm is indicated by an open arrow. The combined image in (i) shows the relative positions of the signals on the marker chromosome (blue); (ii) shows only the images for the green and red signals in (i). Colocalization of the two signals appears as yellow in (ii).

Figure 3

Schematic representation of the combined chromatin immunoprecipitation and array (CIA) analysis.

Figure 4

Hybridization of 20p12 BAC/PAC array (X-axis) using DNA extracted from CENP-A immunoprecipitated chromatin (bound) fractions from the patient (filled circles) and a normal (open square) lymphoblast cell line N2. Y-axis shows the percentage difference of the ratio of enhancement (R) of bound fractions over preimmunoprecipitation (input) fractions compared with a normal control cell line N1: The BAC/PAC array covers ∼18 Mb of the 20p12 region of interest. Genome distances (Mb) for the 8-Mb region shown are measured relative to PAC dJ811H13. PAC dJ1009E24 and dJ905G11 map outside this region (indicated by the broken bars on the X-axis) and are 6.7 and 11.0 Mb from dJ811H13, respectively. The datapoint [mean ± standard error of mean (SEM) from eight experiments for the patient and four experiments for the normal N2] marks the midposition of each BAC/PAC. The four BAC/PACs that deviate significantly from the baseline values for surrounding BAC/PACs (P < 0.01) are listed next to the datapoint and indicated by an asterisk. The remaining BAC/PACs are listed on the top of the graph.