Additional file 2 of A low-input high resolution sequential chromatin immunoprecipitation method captures genome-wide dynamics of bivalent chromatin

Additional file 2: Figure S1. (A) Agarose DNA gel showing chromatin fragmented with MNase versus sonication (B) ChIP-qPCR for single (top) and reChIP (bottom) experiments using IgG (Invitrogen, black), H3K4me3 (CST 9751, green) and H3K27me3 (CST 9733, red) antibodies on sonicated chromatin. An active H3K4me3 region (Gapdh1), inactive H3K27me3 region (Meis2) and 6 bivalent regions are assayed, four of which are novel to this study. (C) Single ChIP-qPCR analysis using IgG (Invitrogen, left), H3K4me3 (Millipore 07-473, middle) and H3K27me3 (Active Motif 91167, right) antibodies with variable number of cells ranging from 1 million (1M), through to 2,000 (2K). Five primer sets were used including three H3K4me3 enriched regions (green) and two H3K27me3 enriched regions (red). Note that below 50K the background increases and specificity of enrichment is no longer detected. For this experiment a different H3K27me3 antibody was used to the sequenced reChIP datasets. (D, E) downsampling analysis for K4-K27 and K27-K4 datasets using the higher coverage replicate 2. Peaks were called separately for each downsampled dataset and classified according to Figure 3A. (D) Total peaks, (E) promoter peaks. (F, G) Downsampling analysis of independent total H3K4me3 and total H3K27me3 from (GSE135841) (7) showing in silico bivalent predictions atall peaks (G) and promoter peaks (H) at different simulated read depths. (H) Pseudocolour density scatterplot showing log2 CPM/bp enrichment at combined H3K4me3 and H3K4me3-IgG peaks for total H3K4me3 single ChIP (x-axis) compared to H3K4me3-IgG reChIP (y-axis). R=0.906. Data reanalysed from Mas et al. 2018 (14). Figure S2. (A) Schematic showing peak calling strategy for in silico bivalent peak prediction (B-F) Genome browser views of (B) H3K4me3-only, (C) high-confidence, (D) K4-biased, (E) K27-biased, and (F) low confidence genes showing H3K4me3 (green), total H3K27me3 (red) and K4-K27 (purple) and K27-K4 (blue) reChIP datasets. Height of peak represents CPM/bp. E-M represents data from independent 10 million cell total H3K4me3 and total H3K27me3 (GSE135841) (7). R1 and R2 are two independent biological replicates from this study. (G) number of bivalent promoters overlapping (dark grey) or not-overlapping (light grey) CpG islands for the four different classes. Genome-wide 51.97% promoters overlap a CpG island using these criteria (H) Overlap of non-promoter bivalent regions with candidate cis-regulatory regions (cCREs) from ENCODE for the four different classes. Figure S3. (A) Classification strategy for calling bivalent promoters in data from Mas et al. 2018 (14). Note only one replicate of the reChIP datasets were generated in this study. Numbers denote number of peaks/promoters at each step. (B) Enrichment of bivalent promoters shared between this study and Mas et al. 2018 (14) (top, n=3593) or unique to our study (bottom, n=1511) in Mas et al. 2018 (14) datasets. (C) log2 fold change in gene expression levels for pluripotency genes (top) and a random set of expressed non-bivalent genes (bottom) across 9 days of embryoid body differentiation. Each gene has been normalised separately across the time series. Top and bottom groups of genes are on different scales. Gene expression data reanalysed from (GSE135841). Figure S4. (A) Classification strategy for calling differential bivalent promoters between wild type (WT) and Dppa2/4 double knockout (DKO) clones. Numbers denote number of peaks/promoters at each step.