Dynamic regulation of histone modifications and long-range chromosomal interactions during postmitotic transcriptional reactivation

In this study, Kang et al. sought to understand the underlying principles that mediate transcriptional memory and reactivation in the daughter cells after mitosis. They used ChIP-seq on synchronized cells at different stages after mitosis to generate genome-wide maps of histone modifications, and combined with EU-RNA-seq and Hi-C analyses, they found that during prometaphase promoters, enhancers, and insulators retain H3K4me3 and H3K4me1, while losing H3K27ac, thus providing new insights into the histone modification landscape during cell division.

Raw read quality was checked using FastQC and contamination was checked with FastqScreen and BLAST. Reads were mapped to an hg19 (human) and dm6 (fly spikein) combined genome using the STAR aligner version 2.5.3b (Dobin et al. 2013).
Mapping was carried out using default parameters (up to 10 mismatches per read, and up to 9 multi-mapping locations per read). The genome index was constructed using the gene annotation supplied with the hg19 and dm6 Illumina iGenomes (iGenomes online. fp,r,a=(|humanp,r,a|+|flyp,r,a|)|flyp=I,r,a|/|flyp,r,a| (|humanp=I,r,a|+|flyp=I,r,a|) Peaks were identified using HOMER (Heinz et al., 2010), using default parameters and input condition as background reference. Normalized read counts for peaks were calculated using the fragments per kilobase per million mapped reads (FPKM) normalization multiplied by the spike-in factors. Low coverage peaks were filtered out (average log2 normalized read counts < 3 across all phases for H3K27ac, H3K4me3, and H3K4me1, and average log2 normalized read counts < 1 across all phases for CTCF, due to lower quality ChIP for CTCF). Then peaks were defined as present in a phase if the average normalized expression in that phase was at least the ¾ the maximum normalized read count across all 3 phases. Peak overlap across cell cycle phases was calculated using HOMER mergePeaks routines with a maximum distance for merging of 1000 bp.

Genomic Element Analysis
Phase-specific genomic elements were defined according to the presence of H3K27ac, H3K4me1, and H3K4me3 normalized peaks in each phase and their proximity to the transcription start sites (TSS) of annotated RefSeq genes. Promoters (Pr) were defined as any regions containing H3K4me3 within 1kb of the TSS of known genes. Primed enhancers (PE) were defined as regions containing only H3K4me1 peaks at a distance of more than 1kb away from the TSS of known genes. Active enhancers (AE) were defined as regions containing both H3K4me1 and H3K27ac peaks at a distance of more than 1kb away from the TSS of known genes. HOMER mergePeaks was used to find all combinations of overlaps between peaks in each phase and HOMER annotatePeaks was used to determine the distance to the nearest TSS.
To calculate the distribution of normalized read coverage around the center of genomic elements (Pr, AE, PE), HOMER annotatePeaks was used with a window size of +/-3kb and a bin size of 100 bp. Spike-in normalization factors were applied and the average read coverage of replicates was shown as a histogram.
The set of promoters and enhancers maintained during prometaphase and those lost in prometaphase, then regained in anaphase/telophase, was determined with HOMER mergePeaks applied to the elements directly. The known motif enrichment analysis was carried out using HOMER findMotifsGenome.pl with -size given.

EU-RNA Analysis
Raw read quality was checked using FastQC and contamination was checked with FastqScreen and BLAST. Reads were mapped to an hg19 (human) using the STAR aligner version 2.5.3b (Dobin et al. 2013). FPKM (fragments per kilobase per million mapped reads) gene expression was quantified across the entire gene with HOMER analyzeRepeats. In addition, reads mapping to spike-in sequences were used to calculate the normalization slope based on the expected concentrations of the spike-ins as was done previously (Palozola et al. 2017). Expression values were further log2 transformed, averaged across replicates, and NoEU control expression values were subtracted from EU values at each time point. Next, time point-specific genes were identified as genes whose expression exceeds 50% of the average expression across all times (excluding asynchronous) and continues to exceed 50% for the duration of the time course. Asynchronous genes were defined as those that were not time point specific and whose values were at least twice as high in the asynchronous time point compared to any of the other time points. Heatmaps were generated using R libraries gplots (heatmap.2 function), or plot.matrix (plot function) with scaling across conditions. An absolute log2fold change of 1 was used to determine genes up or down with A-485.
To correlate EU expression with the ChIP-seq results, we used the merged peaks from the replicates of each ChIP antibody and cell-cycle phase to quantify the EU read counts at each time point averaged across both replicates and normalized for peak size.
These values were then correlated to the average ChIP read counts for that antibody and cell-cycle phase normalized to the peak size. Gene Ontology functional enrichment analysis was performed using DAVID (Dennis et al. 2003).

Hi-C data analysis
Sequencing reads generated from samples of eight time points with two replicates were subjected to alignment and processing as previously described (Dixon et