Chromatin domains rich in inheritance

See allHide authors and affiliations

Science  06 Jul 2018:
Vol. 361, Issue 6397, pp. 33-34
DOI: 10.1126/science.aat7871

Epigenetic phenomena are heritable changes to gene expression that occur without changes to the DNA sequence and that include posttranslational modifications (PTMs) to the histones that package DNA into chromatin. These PTMs are deposited on histones by enzymes in response to an “initiator,” ultimately altering chromatin structure and, accordingly, gene expression. In multicellular organisms, cellular identity is established by master regulators (initiators) that can activate or repress transcription through their sequence-specific DNA binding activity. The accurate transmission of distinct gene expression profiles during cell division is essential for preserving the properties of cell lineages. Thus, a key feature of the epigenetic process is that after the initiator subsides, these informative chromatin PTMs must be inherited by subsequent cell generations. Numerous histone PTMs can occur, but can they all convey epigenetic information? We discuss the few histone PTMs that qualify as epigenetic and the distinct features of the enzymes that deposit them that account for their epigenetic status.

Histones undergo many PTMs, primarily on their amino-terminal “tails,” which are about 40 amino acids in length, are nonstructured, and protrude away from the nucleosome units that comprise chromatin. When unmodified, the positively charged histone tails can engage in electrostatic interactions with DNA and neighboring nucleosomes, potentially forming compacted chromatin by default. How is the open structure of transcriptionally competent chromatin (euchromatin) established? Among the many histone PTMs that occur, acetylation of lysine 16 of histone H4 (H4K16ac), which is conserved from yeast to humans, functions to establish euchromatin (1). The subsequent combinatorial patterns of histone PTMs mostly operate to promote transcription. Sustaining transcription entails the constant presence of transcriptional activators (initiators) that bind to DNA regulatory sequences (see the figure). These activators recruit coactivators, including histone acetyltransferases that further acetylate the histones near promoter sequences. Additionally, distinct phosphorylation events of the carboxyl-terminal domain of RNA polymerase II (Ser5-P and Ser2-P) recruit specific methyltransferases that independently methylate H3K4 and H3K36 to further facilitate transcription (2).

We speculate that so-called “open” chromatin that is accessible to the transcription machinery is not per se inherited and, thus, not epigenetic. It is reestablished by H4K16ac after each round of DNA replication and/or mitosis, and gene expression requires an initiator. H4K16ac might be epigenetic, but currently there is no supporting evidence. Instead, a few modifications that promote repressed, compacted chromatin domains exhibit epigenetic properties. Regions of repressed transcription can foster the deposition of specific histone PTMs that provide the platform for chromatin compaction. Upon subsidence of the initiator, this compacted chromatin must contain information that transmits its status to daughter cells after cell division, as its accurate inheritance preserves the integrity of repressed gene expression in a cell lineage.

Studies in baker's yeast, Saccharomyces cerevisiae, provide important evidence demonstrating that the local deacetylation of H4K16ac by the nicotinamide adenine dinucleotide (NAD)–dependent histone deacetylase, silent information regulator 2 (Sir2), is necessary to establish compacted (repressed) chromatin domains that can be maintained through multiple generations at mating-type loci, ribosomal DNA repeat sequences, and telomeres (3). Deacetylation of H4K16ac at these loci requires the initial recruitment of the Sir complex (Sir2-Sir3-Sir4). In the case of mating-type loci, the Sir complex is recruited through direct interaction of the Sir1 adaptor and an initiator, the origin recognition complex (Orc). Sir2 deacetylates H4K16ac, which is subsequently bound by Sir3. Sir3 interaction with Sir2 and Sir4 expands the deacetylated state. Importantly, the Sir1-Orc initiator is no longer required for persistence of the deacetylated state. Instead, the Sir2 histone deacetylase maintains the repressed chromatin domain through multiple cell divisions—a bona fide epigenetic phenomenon in yeast; it is unknown if this is also the case in higher eukaryotes.

Studies in higher eukaryotes show that the only heritable chromatin domains comprise trimethylation of H3K9 (H3K9me3) and of H3K27 (H3K27me3), which are associated with repressed chromatin. These PTMs provide binding sites for other factors that compact chromatin, forming large chromatin domains of constitutive (H3K9me3) or facultative (H3K27me3) heterochromatin in specific chromosome locations initiated by master regulators (4, 5). Although still under study, these PTMs likely segregate to the appropriate daughter chromatin domains during DNA replication such that nucleosomes containing histones with parental PTMs are mixed with those containing newly synthesized, unmodified histones. The machinery that deposits H3K9me3 and H3K27me3 has a distinct “write-and-read” mechanism that allows these PTMs to qualify as epigenetic: The enzyme or complex that deposits the PTM can also bind to the PTM. This binding stimulates the enzyme to deposit the PTM on adjacent naïve nucleosomes—positive feedback that reinforces and establishes repressive chromatin domains in daughter cells. The methyltransferases SUV39H1 and Polycomb repressive complex 2 (PRC2) exhibit this self-contained write-and-read mechanism: The enzymatic writing module that trimethylates H3K9 and H3K27 (SET domain) is activated when the reading module (a chromodomain in SUV39H1 and an aromatic cage in PRC2) binds their respective enzymatic product. Thus, the presence of only the histone methyltransferase and the parental PTM suffice to restore the PTM to appropriate domains in daughter cells.

The methyltransferase cryptic loci regulator 4 (Clr4) in fission yeast, Schizosaccharomyces pombe, and its mammalian homolog, SUV39H1, establish and maintain H3K9me domains by binding to H3K9me, which in turn bolsters their enzymatic activity, giving rise to further H3K9me catalysis. In fission yeast, establishing H3K9me domains requires multiple factors (3). Yet, the role of Clr4 is pivotal, as inheritance of a heterochromatic state depends on the maintenance of H3K9me at the site and requires both Clr4 H3K9me binding (chromodomain) and methyltransferase activity (3). The artificial recruitment of a Clr4 chromodomain mutant to an active chromatin site is sufficient to establish an H3K9me domain and transcriptional repression in the absence of endogenous Clr4. However, this Clr4 mutant cannot maintain H3K9me3 heterochromatin through subsequent cell generations (3). This points to the epigenetic relevance of the dual write-and-read function. In vitro, Clr4 binds equally well to unmodified and H3K9me3-containing mononucleosomes; however, its binding to a dinucleosome comprising H3K9me3 in one nucleosome activates its methyltransferase activity in a nonallosteric manner (6). Such binding is proposed to optimally position Clr4 for recognition of the neighboring naïve nucleosome, thereby increasing catalysis. Similarly, the self-contained write-and-read modules of SUV39H1 lead to propagation of H3K9me3 in vitro. In this case, the chromodomain of SUV39H1 binds to H3K9me3, resulting in allosteric activation of its SET (methyltransferase) domain (7).

Inheritance of repressive histone PTMs

Initiators must be present to establish and maintain histone PTMs that facilitate transcription on euchromatin. PRC2 and SUV39H1 trimethylate H3K27 and H3K9, respectively. These repressive histone PTMs are epigenetic owing to the distinct write-and-read mechanisms of PRC2 and SUV39H1.


In the case of H3K27me3, PRC2 comprises four core subunits, including supressor of zeste homolog 12 (SUZ12), embryonic ectoderm development (EED), and enhancer of zeste homolog 2 (EZH2). EED binds H3K27me3, this binding allosterically activates the SET domain of EZH2. This allosteric activation of PRC2 set the precedent for the write-and-read mechanism that, in this case, specifies the epigenetic character of H3K27me3 (8). After cells are depleted of EED and thus H3K27me2 or H3K27me3, EED expression can be restored either as wild type or mutant in the EED “read” function (9). Mammalian chromatin sites to which PRC2 is initially recruited (nucleation sites) and establishes de novo H3K27me3 are similar in both cases; however, only wild-type EED establishes H3K27me2 or H3K27me3 through chromatin domains, highlighting the importance of the read function. Another study generated nematode worm (Caenorhabditis elegans) embryos containing some chromosomes with H3K27me and some without (10). In the absence of PRC2, H3K27me disappeared through several rounds of cell division, but, in its presence, H3K27me was transmitted epigenetically throughout embryogenesis. Additionally, PRC2 in plants functions to maintain transcriptional repression not only during mitosis but also transgenerationally (11). Recent studies in the fruitfly, Drosophila melanogaster, suggest that sites of PRC2 recruitment to chromatin (Polycomb response elements) are required for long-term H3K27me3 inheritance, otherwise H3K27me3 is diluted during subsequent rounds of cell division (12, 13). Yet, remarkably, the remaining H3K27me3 is inherited at appropriate daughter chromatin domains, underscoring its epigenetic character and the pivotal role of the PRC2 write-and-read mechanism in restoring H3K27me3. Whether mammalian PRC2 nucleation sites are also integral to H3K27me3 inheritance is yet to be determined.

Lastly, H4K20me is a repressive PTM bound by the malignant brain tumor (MBT) domain of histone methyl-lysine binding protein (L3MBTL1), leading to chromatin compaction (14). Interestingly, methylation of H4K20 is antagonistic with H4K16ac (15). Although potentially epigenetic, a corresponding write-and-read function for methylated H4K20 has yet to be demonstrated.

Although many other writers and readers of histone PTMs exist, for example, proteins with bromodomains that recognize acetylated histones, they are distinct from epigenetic agents in that the reader does not stimulate the activity of the writer. There are some chromatin-modifying enzymes that do bind to their products (4), but whether their binding results in stimulation of enzymatic activity has not been determined. Nonetheless, these enzymes operate on euchromatic regions, where an initiator is required for continued activity.

Why should repressive histone PTMs, but not activating PTMs, be epigenetically inherited? We propose that restraining improper activation of genes might be an evolutionary requirement of multicellularity. Positive feedback loops for gene activation could carry excessive risks as they might result in converting variable stimuli into permanent mistakes in cell fate decisions, with potentially deleterious consequences for the organism. When in doubt, genomes should keep their transcriptome under wraps.

References and Notes

Acknowledgments: We thank K.-J. Armache, S. Berger, R. Bonasio, D. Moazed, C. Desplan, S. Grewal, E. Heard, S. Henikoff, T. Jenuwein, R. Kingston, H. Klein, R. Margueron, R. Martienssen, E. Mazzoni, O. Oksuz, J. Skok, R. Schneider, M.-E. Torres Padilla, P. Voigt, and K. Zaret for comments on the manuscript.
View Abstract

Stay Connected to Science

Navigate This Article