Histone acetylation is generally associated with active chromatin, but most studies have focused on the acetylation of histone tails. Various histone H3 and H4 tail acetylations mark the promoters of active genes1. This includes acetylation of H3 on lysine 27 (H3K27ac), which blocks the deposition of polycomb mediated H3K27me32. H3K27ac is also widely used to identify active enhancers3,4, and the assumption has been that profiling of H3K27ac is a comprehensive way of cataloguing the set of active enhancers in mammalian cell types. Here we show that acetylation of lysine residues in the globular domain of H3 (H3K64ac and H3K122ac) marks active gene promoters and also a subset of active enhancers. Moreover, we find a novel class of active functional enhancers that are marked by H3K122ac but lack H3K27ac. This work suggests that, to identify enhancers, a more comprehensive analysis of histone acetylation is required than was previously considered.

Compared with histone H3, acetylation of H4 tails has not been well studied, especially in mammalian cells. Yet, H4K16 acetylation is of particular interest because of its ability to decompact nucleosomes in vitro and its involvement in dosage compensation in flies. Here we show that, surprisingly, loss of H4K16 acetylation does not alter higher-order chromatin compaction in vivo in mouse embryonic stem cells (ESCs). As well as peaks of acetylated H4K16 and KAT8 histone acetyltransferase at the transcription start sites of expressed genes, we report that acetylation of H4K16 is a new marker of active enhancers in ESCs and that some enhancers are marked by H3K4me1, KAT8, and H4K16ac, but not by acetylated H3K27 or EP300, suggesting that they are novel EP300 independent regulatory elements. Our data suggest a broad role for different histone acetylation marks and for different histone acetyltransferases in long-range gene regulation.

Genome editing occurs in the context of chromatin, which is heterogeneous in structure and function across the genome. Chromatin heterogeneity is thought to affect genome editing efficiency, but this has been challenging to quantify due to the presence of confounding variables. Here, we develop a method that exploits the allele-specific chromatin status of imprinted genes in order to address this problem in cycling mouse embryonic stem cells (mESCs). Because maternal and paternal alleles of imprinted genes have identical DNA sequence and are situated in the same nucleus, allele-specific differences in the frequency and spectrum of mutations induced by CRISPR-Cas9 can be unequivocally attributed to epigenetic mechanisms. We found that heterochromatin can impede mutagenesis, but to a degree that depends on other key experimental parameters. Mutagenesis was impeded by up to 7-fold when Cas9 exposure was brief and when intracellular Cas9 expression was low. In contrast, the outcome of mutagenic DNA repair was unaffected by chromatin state, with similar efficiencies of homology-directed repair (HDR) and deletion spectra on maternal and paternal chromosomes. Combined, our data show that heterochromatin imposes a permeable barrier that influences the kinetics, but not the endpoint, of CRISPR-Cas9 genome editing and suggest that therapeutic applications involving low-level Cas9 exposure will be particularly affected by chromatin status.