Limitations and possibilities of low cell number ChIP-seq. Homo sapiens

ChIP-seq experiments using low numbers of input cells, scaled down to the point where data quality is unnacceptably compromised, reveals limits of the technique. Overall design: Two-part ChIPseq study using native chromatin (non-crosslinked) generated with MNase from human CD4+ cells. Part 1: Previously published protocol (Barski et al, 2007, Cell 129:823, hereafter called "Benchmark") used with 2x10e7 cells / ChIP with H3K4me3 antibody, compared to modified method ("new") with decreasing input cell numbers spanning 1000-fold range from 2x10e7 cells / IP to 2x10e4 cells/ IP. Part 2: reproducibility of new method studied using triplicate samples at lowest two cell numbers tested: 1x10e5 and 2x10e4 / ChIP, using H3K4me3 and H3K27me3 antisera.