An Alternative Approach to ChIP-Seq Normalization Enables Detection of Genome-Wide Changes in Histone H3 Lysine 27 Trimethylation upon EZH2 Inhibition

Abstract

Chromatin immunoprecipitation and DNA sequencing (ChIP-seq) has been instrumental in inferring the roles of histone post-translational modifications in the regulation of transcription, chromatin compaction and other cellular processes that require modulation of chromatin structure. However, analysis of ChIP-seq data is challenging when the manipulation of a chromatin-modifying enzyme significantly affects global levels of histone post-translational modifications. For example, small molecule inhibition of the methyltransferase EZH2 reduces global levels of histone H3 lysine 27 trimethylation (H3K27me3). However, standard ChIP-seq normalization and analysis methods fail to detect a decrease upon EZH2 inhibitor treatment. We overcome this challenge by employing an alternative normalization approach that is based on the addition of Drosophila melanogaster chromatin and a D. melanogaster-specific antibody into standard ChIP reactions. Specifically, the use of an antibody that exclusively recognizes the D. melanogaster histone variant H2Av enables precipitation of D. melanogaster chromatin as a minor fraction of the total ChIP DNA. The D. melanogaster ChIP-seq tags are used to normalize the human ChIP-seq data from DMSO and EZH2 inhibitor-treated samples. Employing this strategy, a substantial reduction in H3K27me3 signal is now observed in ChIP-seq data from EZH2 inhibitor treated samples.

Comparative ChIP-seq (Comp-ChIP-seq): a practical guideline for experimental design and a novel computational methodology

“…In the previous strategies, the sample and the spike-in material are captured using the same antibody, limiting both strategies to using antigens that are highly conserved between both classes of material (for instance, histone modifications). To circumvent this problem, Trojer and colleagues introduced a smart solution by using a second antibody for a fly-specific histone variant (H2Av) to capture the spike-in material [23]. This strategy aims to avoid the cross-140 reactivity constraint of the experimental antibody and to reduce any potential variability due to competition between the spike-in control and the experimental material, which usually exceeds the amount of spike-in material by far.…”

“…Drosophila melanogaster) to correct the overall ChIP-seq signal from the main organism studied (e.g. human) [21,23]. Although initially appealing, this fold-change correction presents in our opinion important shortcomings, such as: i) the spike-in reads mapped not only 200 along the ChIP-seq peaks but also over background regions are used for computing the correction factor; and ii) the correction factor is uniformly applied to all experimental reads in the actual experiment, treating both non-specific and specific signal loci with the same correction value.…”

“…Considering the benefits and limitations of the different strategies (reviewed in Table 1), we propose the addition of exogenous xenogenic fly cells, whenever possible, and the use of a second antibody against a fly-specific histone variant, as a best practice for comparative ChIPseq normalization for mammalian genomes. We recommend using fly material as the spike-in 160 because: 1) the genome sequence has been extensively assembled; 2) the fly chromatin has been largely characterized at epigenetic levels; 3) the evolutionary distance between fly and mammalian genomes is sufficient to allow an unambiguous alignment of the reads [21,23]; and 4) fly cells are relatively easy to culture with standard tissue culture procedures and instruments. Moreover, the genomic occupancy profile of the fly-specific H2Av is already 165 characterized [23], which can be extremely useful as an additional control point for assessing ChIP-seq performance.…”

“…These guidelines take into consideration that: (i) spike-in material should be present in all samples at equal amounts at the earliest step during the ChIP-seq procedure; and (ii) spike-in material should be present in a low-enough quantity (giving a significantly lower number of reads as that of the experimental 175 reads) to not interfere with the actual ChIPseq experiment yet still give an accurate normalization in the final sequencing data. As previously reported, the number of spike-in reads in the final sequencing step should be at least one million reads, and approximately, 2%-5% of the experimental genome, to minimize the changes in overall material used for ChIP-seq [21,23]. This final amount of reads can be influenced by the ratio of the mixture as 180 well as by the quality of the antibody and/or the abundance of the target in the experimental condition.…”

Zbtb24 binding protects promoter activity by antagonizing DNA methylation in mESCs

“…ChIP-seq analysis). To reduce the effects of technical variation and sample processing bias, the datasets were Spike-in normalized (Egan et al 2016). To obtain the normalizing factors, the reads were mapped to the mouse genome (build mm10) and the Drosophila melanogaster genome (build dm6) with the Bowtie2 aligner using the "-very-sensitive" parameter.…”