Fig. 1

BORIS expression in K562 establishes a definitive pattern of repeat binding. a A schematic of CTCF and BORIS proteins with the four amino acids residues essential for DNA recognition by each zinc finger (ZFs) showing. The minor differences, indicated in yellow, do not affect the DNA-binding specificity in vitro neither the consensus derived from the genome-wide binding study [55]. b LI-COR image of immunoblotting for BORIS and CTCF proteins in whole-cell protein extracts of K562 (BORIS positive) and HL60 (BORIS negative) cancer cell lines of myeloid origin. Below the immunofluorescent and DNA staining of the two corresponding cell lines. c The left panel shows the enrichment ratio (M) for CTCF and BORIS across all the tiles of the TR microarray. Dots represent microarray tiles enriched ≥4 by either CTCF (red) or BORIS (blue) with lines connecting different tiles belonging to the same repeat. SAM showed that 42,715 tiles were differentially occupied with FDR ranging from 0.103 to 0.245, with 0.75 correlations between CTCF and BORIS arrays. Both measures indicate that a minority of the repeats were differentially bound by the two proteins. The right panel shows the linear fit of BORIS M ratios by CTCF M ratios, with the fit line and 95 % bivariate normal ellipse displayed. d Principal component analysis of ChIP-chip data from K562. PCA was performed with singular value decomposition (SVD), and the first principal component describing the trend of the data was excluded from the analysis. PC2 (42 %) explains the difference between CTCF and BORIS experiments, and PC3 (16 %) explains variance between the replicates. e Smoothed histogram for probe loading kernel density (Y axis) estimate for all probes along the PC2 axis. The mean is indicated, as well as the number of standard deviations from the mean (Z-score). Most probes show no significant loading on PC2, and no significant difference between CTCF and BORIS. Only a fraction of probes show a significant contribution to the PC2 axis