The molecular basis of heterochromatin assembly and epigenetic inheritance

Summary

Heterochromatin plays a fundamental role in gene regulation, genome integrity, and silencing of repetitive DNA elements. Histone modifications are essential for the establishment of heterochromatin domains, which is initiated by the recruitment of histone-modifying enzymes to nucleation sites. This leads to the deposition of histone H3 lysine-9 methylation (H3K9me), which provides the foundation for building high-concentration territories of heterochromatin proteins and the spread of heterochromatin across extended domains. Moreover, heterochromatin can be epigenetically inherited during cell division in a self-templating manner. This involves a ‘read-write’ mechanism where pre-existing modified histones, such as tri-methylated H3K9 (H3K9me3), support chromatin association of the histone methyltransferase to promote further deposition of H3K9me. Recent studies suggest that a critical density of H3K9me3 and its associated factors is necessary for the propagation of heterochromatin domains across multiple generations. In this review, I discuss the key experiments that have highlighted the importance of modified histones for epigenetic inheritance.

eTOC blurb

In this review article, Grewal elucidates the mechanisms of heterochromatin assembly that are essential for enforcing developmental gene expression patterns and maintaining eukaryotic genome stability. He highlights key findings that have significantly advanced our understanding of epigenetic propagation of heterochromatin domains via modified histones.

Introduction

The organization of eukaryotic genomes into distinct structural and functional domains is critical for the regulation and transduction of genetic information. DNA is folded together with histones and non-histone proteins to form chromatin. Chromatin is broadly organized into heterochromatin and euchromatin via mechanisms involving posttranslational modifications of histones, chromatin remodeling activities, and DNA methylation.^1–3 Heterochromatin domains are relatively less accessible to transcriptional machinery when compared to euchromatin regions. Heterochromatin is divided into two types, constitutive and facultative. Constitutive heterochromatin, which preferentially targets regions containing repetitive DNA elements, such as at centromeres and telomeres, is in general stably propagated.⁴ By contrast, facultative heterochromatin is relatively dynamic and can assemble or disassemble in response to developmental or environmental signals, enabling rapid reprogramming of gene expression.

Heterochromatin domains are a major feature of complex mammalian genomes, and their assembly is essential for diverse chromosomal processes. To support various functions, heterochromatin proteins recruit effectors, including enzymatic activities that carry out transcriptional and posttranscriptional silencing to prevent inappropriate expression of target loci.⁵ Heterochromatic structures spread across extended domains and can be epigenetically inherited in a self-templating manner during cell division. This is required to maintain stable gene repression or silencing. The cell type-specific megabase-sized heterochromatin domains that are present in differentiated human cells enforce lineage-specific gene expression patterns and form a barrier to cellular reprogramming.⁶ Heterochromatin also protects the host genome from the deleterious effects of “selfish” transposable elements. Transposons and their remnants are packaged into repressive heterochromatin, preventing their expression, and prohibiting illegitimate recombination, thus safeguarding genome integrity.⁷

Genetic and biochemical studies from diverse model systems have contributed to our current understanding of heterochromatin assembly and its epigenetic inheritance. The core heterochromatin assembly pathway involving histone modifications is well-preserved from fission yeast Schizosaccharomyces pombe to mammals, although certain aspects of heterochromatin regulation have evolved to assume greater or lesser roles in the various organisms studied.^4,8 The genetically tractable S. pombe, which contains facultative heterochromatin islands and constitutive heterochromatin domains,⁹ has been a key model organism in defining heterochromatin assembly pathways.^5,10 In this review, I discuss the advances in our understanding of the mechanisms by which heterochromatin domains are assembled and epigenetically inherited. The major focus is on insights gained from S. pombe, with brief comparisons to other systems.

Factors involved in heterochromatin assembly

Histone modifications are essential for the formation of heterochromatic structures. Different from euchromatic regions, heterochromatin is characterized by hypoacetylated histones and high levels of H3K9me.^9,11–13 Histone deacetylase (HDAC) and methyltransferase activities are critical in this process. In higher eukaryotes, H3K9me deposition is catalyzed by multiple histone methyltransferases, depending on the chromosomal context.^13,14 In contrast, in S. pombe, mono-, di-, and tri-methylation of H3K9 (H3K9me1/2/3) is carried out by a single histone methyltransferase enzyme called cryptic loci regulator 4 (Clr4),¹⁵ which is an ortholog of Drosophila Su(var)3–9 and mammalian SUV39H1 and SUV39H2 (collectively referred to as Suv39h).^16,17 Clr4 is part of a multi-subunit protein complex that includes components such as the E3 ubiquitin ligase Cullin 4.^18–20 Clr4/Suv39h proteins share a similar molecular architecture, which includes a carboxy-terminal SET domain with methyltransferase activity and an amino-terminal chromodomain that binds to H3K9me3.²¹ Binding to existing H3K9me3 further increases the enzymatic activity of Clr4 and Suv39h.^22,23 Moreover, it has been suggested that Clr4 can methylate itself, which is believed to play a role in regulating H3K9me and heterochromatin formation.²⁴

Methylation of H3K9 by Clr4/Suv39h provides the foundation for chromatin association of other chromodomain-containing proteins, including Chp1, Chp2 and Swi6 in S. pombe, and HP1α, HP1β and HP1γ in mammals.^15,25–30 Through their amino-terminal chromodomains, HP1 proteins bind to di- or tri-methylated H3K9, and form dimers through their carboxy-terminal chromo-shadow domain (Figure 1).^31,32 Additionally, HP1 family proteins participate in low-affinity multivalent interactions through their intrinsically disordered regions.^33–36 This combination of stoichiometric and non-stoichiometric interactions leads to high local concentrations of HP1 on chromatin (referred to here as HP1 territories). The existence of two distinct pools of HP1, one engaged in direct histone interactions and another that form bridges in a dynamic fashion, is consistent with protein kinetic studies showing both stable and dynamic populations of HP1.^37–39

Figure 1. HP1 binds to methylated nucleosomes and multimerizes to recruit effector proteins for heterochromatin assembly.

(Left) S. pombe Swi6/HP1 is concentrated at specific foci within the nucleus. Blue DAPI staining shows the nucleus of an S. pombe cell, while the red foci show high concentrations of Swi6/HP1 protein bound to heterochromatic loci. (Center) Methylation of histone H3 lysine 9 (H3K9me) by Clr4/Suv39h provides the foundation for recruiting and concentrating heterochromatin proteins such as Swi6/HP1. Swi6/HP1 molecules stoichiometrically bind to H3K9 methylated nucleosomes via their amino-terminal chromodomain and form dimers (curved dotted lines) through their carboxy-terminal chromoshadow domain. In addition, HP1 proteins are believed to engage in non-stoichiometric multivalent interactions (straight dotted lines) to form high-concentration spatial nuclear territories, which provides a multifunctional scaffold to amplify the chromatin association of various effector proteins. Specifically, the formation of high-concentration HP1 territories helps to attract and retain otherwise diffusible proteins required for a multitude of heterochromatin functions, including factors involved in histone deacetylation (HDACs), histone methylation (HMTs), histone chaperones (FACT), RNA interference (RNAi), RNA processing and peripheral tethering (RIXC), and chromosome architecture (SMC), among many others.

The formation of high-concentration HP1 territories likely offers an increased interface for concentrating its associated effectors (Figure 1). HP1 acts as a platform to recruit and distribute a range of regulatory proteins, such as those responsible for the proper segregation of chromosomes,^40,41 the organization of the genome^42,43 and directing developmentally-controlled long-range chromatin interactions enabling cell type-switching.⁴⁴ Additionally, HP1 proteins anchor activities involved in transcriptional and post-transcriptional silencing, including components of the RNA interference (RNAi) machinery that converts heterochromatic transcripts into small interfering RNAs (siRNAs), as well as chromatin modifiers such as HDACs, which modify properties of chromatin to assemble silenced domains. For instance, the S. pombe Snf2/HDAC repressor complex (SHREC), an effector for heterochromatic transcriptional gene silencing,⁴⁵ requires HP1 proteins for its distribution across heterochromatin domains.^46–49 Like its mammalian counterpart the nucleosome remodeling and deacetylase (NuRD) complex,⁵⁰ SHREC contains the Mi2-like protein Mit1 and the class II HDAC Clr3 that deacetylates histones to assemble repressive chromatin, thereby denying transcriptional machinery access to the underlying DNA.^45,51

In broad terms, the use of methylated histones as a molecular foundation for building a multivalent interaction network of proteins that attract diverse effectors is expected to be a widespread strategy in eukaryotes. Like H3K9me-HP1, histone H3 lysine-27 trimethylation (H327me3) is recognized by the Polycomb-repressive complex that engages in multivalency driven interactions and is critical for silencing target loci.^52–54 H3K27me also provides binding sites for the Bromo-adjacent homology (BAH) domain protein, which mediates heterochromatic silencing through its multifunctional scaffolding role and recruitment of HDACs.⁵⁵ In mammals and other organisms, heterochromatic silencing requires cross-talk between histone methylation and DNA methylation, which are likely complementary marks used for targeting gene silencing activities.^56–59 Indeed, DNA methylation provides binding sites for methyl-CpG-binding domain (MBD) proteins containing intrinsically disordered regions, which also recruit HDAC like proteins bound to methylated histones.⁶⁰

Heterochromatin nucleation

The assembly of heterochromatin domains is a multi-step process. It begins with the nucleation of heterochromatin at specific genomic sites, and then spreads to the surrounding sequences. The strategies used to nucleate heterochromatin vary depending on the chromosomal context. However, a common element is that nucleation mechanisms help to bring together and retain otherwise diffusible heterochromatin assembly factors, thus creating a high local concentration at specific genomic sites. Specifically, the factors involved in nucleation help to concentrate histone-modifying activities and other factors above the critical threshold needed to initiate a cascade of heterochromatin assembly activity.

In some cases, proteins bind to specific DNA elements, referred to as silencer elements, to nucleate heterochromatin (Figure 2A). For example, mammalian Krüppel-associated box zinc finger proteins (KRAB-ZFPs) bind DNA and, in combination with the corepressor KAP1, recruit histone-modifying enzymes, including H3K9 methyltransferases and the NuRD complex.^61,62 In other cases, coding and non-coding transcripts, including those derived from repeat elements, recruit heterochromatin assembly machinery.⁴ It is paradoxical that target loci transcription and RNAs are required to assemble repressive heterochromatic structures. However, multiple copies of RNAs, when locally retained on chromatin, might help amplify the recruitment of chromatin modifiers to surpass the critical threshold necessary for heterochromatin assembly. In this regard, RNAs generated from multicopy sequences, such as arrays of repetitive DNA elements or transcripts containing multiple binding sites for factors that target heterochromatin proteins, are especially suited to nucleate heterochromatin.

Figure 2. RNA- and DNA-based mechanisms guide the nucleation of heterochromatin domains.

(A) DNA binding proteins (DBP) bind to specific DNA sequences and through adaptor proteins recruit histone-modifying activities, such as the histone methyltransferase Clr4/Suv39h and HDACs, to assemble heterochromatin. (B) In the RNAi-dependent pathway, the RNA-induced transcriptional silencing (RITS) complex containing Ago1, Tas3 and Chp1, is a key component of a self-reinforcing loop mechanism that couples the generation of siRNAs to heterochromatin assembly. Long non-coding RNAs (lncRNAs) generated by RNAPII are processed by RNAi machinery to generate siRNAs. The siRNAs guide RITS to nascent transcripts where it helps target the Clr4 complex (named ClrC) via the Stc1 adaptor protein to methylate H3K9. H3K9me in turn promotes stable chromatin association of RITS via the Chp1 chromodomain protein. RITS acts together with the Rdp1-containing complex (RDRC) and Dicer (Dcr1) to process nascent transcripts into siRNAs. siRNA biogenesis is facilitated by Ers1 that connects RDRC with Swi6/HP1. The siRNAs further promote H3K9me by Clr4 to efficiently assemble heterochromatin and silence target loci. (C) In the RNAi-independent pathway, nuclear RNA elimination factors and RNAPII termination factors target the Clr4 complex to nucleate heterochromatin. The YTH family RNA-binding protein Mmi1 binds to transcripts containing determinant of selective removal (DSR) elements and recruits the cleavage and polyadenylation (CPF) complex to trigger RNAPII termination at non-canonical termination sites via a mechanism that also requires the 5’→ 3’ exoribonuclease Xrn2 (named Dhp1 in S. pombe). On the other hand, Mmi1 associates with the Enhancer of Rudimentary Homolog Erh1 (ERH in mammals). Mmi1-Erh1 recruits MTREC (PAXT in mammals), which promotes RNA degradation through the 3’ → 5’ exoribonuclease activity of Rrp6/exosome. MTREC and termination factors act together to recruit the Clr4 complex to methylate H3K9 and nucleate heterochromatin. In this mechanism, premature termination at non-canonical sites is coupled to RNA degradation and heterochromatin assembly.

Although RNAs can directly target heterochromatin machinery, studies in S. pombe have implicated nuclear RNA processing and transcription termination factors in mediating the targeting of heterochromatin assembly proteins. Two pathways have been identified that use RNA polymerase II (RNAPII) transcription to initiate heterochromatin formation: the RNA interference (RNAi)-dependent pathway that relies on the RNAi machinery, and the RNAi-independent pathway that involves nuclear RNA elimination factors (Figure 2B and 2C).

(a). RNA-based RNAi-dependent pathway:

The RNAi pathway, originally identified as a posttranscriptional silencing mechanism,⁶³ plays an important role in targeting heterochromatin in S. pombe.^64–66 RNAPII transcription of heterochromatic repeats, which occurs mainly during a short window in S phase,^67,68 facilitates RNAi-mediated heterochromatin assembly.^69,70 Repeat transcripts are processed into siRNAs by the RNAi factors Argonaute (Ago1), Dicer (Dcr1) and RNA-dependent RNA polymerase (Rdp1) (Figure 2B). The siRNAs are loaded onto the RNA-induced transcriptional silencing (RITS) complex, which includes Ago1, Tas3, and the chromodomain protein Chp1.⁷¹ The siRNAs guide RITS to nascent transcripts,⁷² where RITS along with other factors recruits the RNA-dependent RNA polymerase complex (RDRC) to generate double-stranded RNA and amplify siRNA production.^73–75 RITS, in conjunction with the adaptor protein Stc1, also recruits the Clr4 complex to methylate H3K9, which nucleates heterochromatin.^21,76 Since siRNAs can target RITS along the length of transcripts, this mechanism can recruit multiple Clr4/Suv39h molecules to reach the threshold for heterochromatin assembly.

RNAi-mediated heterochromatin assembly involves a self-reinforcing loop, wherein siRNAs program RITS to target H3K9me and heterochromatin machinery, which in turn allows RNAi pathway proteins to stably bind to chromatin for efficient processing of transcripts to siRNAs (Figure 2B).^74,77 In addition to RITS binding to H3K9me via the Chp1 chromodomain,^29,77,78 Swi6/HP1 bound to H3K9me stabilizes chromatin association of RDRC to generate siRNAs in cis.^74,79,80 This self-reinforcing loop mechanism, which amplifies signals for targeting heterochromatin, can promote heritable gene silencing in multiple organisms, including S. pombe^81,82 and higher eukaryotes.⁸³ Moreover, the RNAi machinery also functions in heterochromatin assembly in several different organisms including Tetrahymena, Drosophila, C. elegans and plants.^84,85

A question remains as to how siRNAs are initially generated to trigger the RNAi self-reinforcing loop. Processing of RNA degradation products and double-stranded RNA generated by secondary structures may contribute to this process.^86,87 Recent studies have suggested that Pir2, the S. pombe homolog of mammalian ARS2, may play a role in siRNA production.⁸⁸ Pir2 cooperates with splicing factors implicated in siRNA production,^89,90 to enable RNAi-mediated processing of transcripts containing inefficiently spliced cryptic introns to generate siRNAs.⁸⁸

(b). RNA-based RNAi-independent pathway:

Studies in S. pombe have uncovered another heterochromatin assembly pathway that also requires RNAPII transcription but does not involve the production of small RNAs (Figure 2C).^91,92 Among other factors, this pathway involves the Mtl1-Red1 core complex (MTREC),^93,94 which is similar to the mammalian poly(A) exosome targeting (PAXT) complex.⁹⁵ Red1 forms dimers and links the Mtl1 RNA helicase with other factors, including the serine and proline rich protein Pir1 (also called Iss10),^96,97 involved in RNA degradation by the nuclear exosome.⁹⁸ The recruitment of RNA elimination machinery to specific transcripts is mediated by the YTH domain RNA-binding protein Mmi1.⁹⁹ Unlike mammalian YTH proteins that recognize transcripts carrying the N6-methyladenosine (m6A),¹⁰⁰ Mmi1 directly binds hexameric sequence motifs, called determinants of selective removal (DSR).¹⁰¹ Multiple DSRs in target transcripts facilitate efficient loading of Mmi1. Mmi1 forms a stochiometric complex with an ortholog of human Enhancer of Rudimentary (ERH) to engage MTREC,^102,103 which in addition to RNA decay promotes targeting of heterochromatin (Figure 2C).⁹³ Mmi1 also recruits the cleavage and polyadenylation factor (CPF) complex and the 5’→3’ exonuclease Dhp1/Xrn2, which trigger RNAPII termination at non-canonical termination sites within the gene body and mediate heterochromatin assembly.^104–108 Both MTREC and Dhp1/Xrn2 associate with the Clr4 complex,^91,104 which methylates H3K9 to assemble heterochromatin. Thus, Mmi1-mediated recruitment of RNAPII termination and RNA elimination factors links premature termination to co-transcriptional RNA degradation and is functionally coupled to heterochromatin targeting.

This RNAi-independent mechanism plays an important role in facultative heterochromatin assembly and preventing inappropriate expression of regulated genes, including gametogenic genes.^107,109 Signaling cascades that respond to environmental and developmental signals target the RNA elimination network to control heterochromatin formation and mRNA decay. The target of rapamycin (TOR) pathway, a key regulator of eukaryotic cell growth, phosphorylates the MTREC-associated Pir1 protein, thereby protecting it from degradation by the ubiquitin-proteasome system.¹¹⁰ By controlling the levels of Pir1, which connects MTREC to the Mmi1-Erh1 complex bound to RNA, TOR directs RNA degradation and heterochromatin assembly to regulate gene expression.¹¹⁰ Besides silencing genes, the RNAi-independent RNA-based heterochromatin assembly pathway acts parallel to RNAi to target heterochromatin at repetitive DNA elements, including centromeric repeats.^104,107,111

Evidence suggests that a conserved RNA-based RNAi-independent pathway exists in metazoans. For example, components involved in pre-mRNA 3’-end processing and termination have been found to affect gene regulatory processes.^112,113 Additionally, RNA elimination factors have been implicated in heterochromatic silencing.¹¹⁴ In mammals, ERH is the global regulator of H3K9me and silencing of developmental genes,¹¹⁵ as it is in S. pombe.¹⁰² Interestingly, XIST long noncoding RNA (lncRNA), which mediates X-inactivation via heterochromatin formation in female mammals,^116–118 is a key target of the RNA elimination factors, including the YTH domain containing protein YTHDC1.¹¹⁹ In line with the above-mentioned studies linking the YTH family protein Mmi1 to heterochromatin assembly in S. pombe, an intriguing possibility is that similar mechanisms recruit silencing activities for X-inactivation in mammals.

The RNA-based mechanisms often work together with DNA-binding proteins to nucleate heterochromatin. For example, at the S. pombe silent mating-type (mat) region, the RNAi machinery nucleates heterochromatin at the cenH element that shares homology with centromeric repeats (Figure 3).^64,77 On the other hand, heterochromatin assembly is independently initiated by DNA binding proteins, such as the ATF/CREB family proteins.^120–123 Both mechanisms target the Clr4/Suv39h methyltransferases to establish local hubs of H3K9me and its associated factors, such as the HP1 proteins and their interaction partners, to seed heterochromatin assembly. RNAi targets H3K9me by directly recruiting the Clr4 complex, while ATF/CREB associate with Swi6/HP1, which engages Clr4 to methylate H3K9.^120,124 Indeed, enhanced heterochromatin formation upon Swi6/HP1 overexpression is dependent on ATF/CREB proteins.¹²⁰ Heterochromatin formation at subtelomeric regions is also accomplished through overlapping mechanisms; for example, the RNAi machinery and the components of the telomere protection complex Shelterin independently recruit the Clr4 complex.^125,126 The combination of two distinct mechanisms to anchor and accumulate multiple heterochromatin factors at nucleation sites ensures efficient heterochromatin assembly. Similar arrangements likely exist in other systems, including in mammals, where transcription factors and transcriptional activity contribute to heterochromatin assembly.^{4,62,127–129}

Figure 3. Nucleation and spreading of heterochromatin.

The formation of the archetypical heterochromatin domain at the silent mat region in S. pombe involves two distinct nucleation mechanisms that independently recruit/retain heterochromatin factors at specific sites. In addition to RNAi machinery that nucleates heterochromatin at the centromere homologous cenH element, the DNA binding proteins (DBPs), including ATF/CREB family proteins, collaborate with Swi6/HP1 and other factors to engage Clr4/Suv39h at a nearby site. Methylation of H3K9 by Clr4/Suv39h not only provides the foundation for forming high concentration HP1 territories but it also serves as an epigenetic template for loading additional Clr4/Suv39h. The ability of Clr4/Suv39h to bind H3K9me3 and catalyze the further deposition of H3K9me allows the concentration gradient of heterochromatin factors to expand and spread across an extended domain surrounded by inverted repeat (IR-R and IR-L) boundary elements (middle). In this read-write mechanism, the chromatin association of Clr4/Suv39h is further strengthened by Swi6/HP1, which also attracts other factors that safeguard the H3K9me3 epigenetic template. Among these, Clr3 HDAC suppresses histone turnover to maintain H3K9me3, and FACT is believed to aid the retention of parental methylated histones during replication. Moreover, the Rix1-containing RNA processing complex RIXC, which localizes to IR boundary elements containing B-boxes (transcription factor TFIIIC binding site), tethers the heterochromatin domain to the nuclear periphery.¹⁶⁸ Tethering restricts the heterochromatin domain to a specific nuclear volume, presumably enabling heterochromatin factors to be efficiently concentrated. However, considering that RIXC binds Swi6/HP1 with high affinity, RIXC bound to boundary elements might help directly concentrate Swi6/HP1 to facilitate heterochromatin assembly.

Spreading of heterochromatin

Early studies of position effect variegation (PEV) suggested that heterochromatin, when juxtaposed with euchromatin by chromosomal rearrangement, can spread and lead to the silencing of neighboring genes.²⁶ Subsequent research has demonstrated that the spread of heterochromatin from nucleation sites to adjacent sequences is necessary for producing naturally occurring silenced domains.^64,130 Spreading permits heterochromatin proteins and their associated effectors to exert control over extended regions in a sequence-independent and region-specific manner.

In general, the spread of heterochromatin from nucleation sites across large domains occurs slowly,^130,131 likely due to the gradual formation of HP1 territories whose dosage can be a limiting factor. Spreading of heterochromatin requires the histone-modifying activities targeted initially by RNA or DNA-binding proteins to be recruited by the modified histones themselves. The binding of histone-modifying activities to modified histones allows local concentration gradients of silencing factors to expand and encapsulate extended domains, in a process that is common to all studied systems, including S. cerevisiae.¹³² However, S. cerevisiae lacks canonical heterochromatin machinery, such as H3K9me and HP1, and therefore studies using S. pombe have provided important insights into heterochromatin spreading. At the silent S. pombe mat region, heterochromatin spreads from nucleation sites to coat a 20-kb domain surrounded by boundary DNA elements (Figure 3).^{11,64,130,133} This process depends on the ability of Clr4/Suv39h to both methylate H3K9 (“write”) and bind to H3K9me3 via its chromodomain (“read”).²¹ Clr4/Suv39h bound to preexisting H3K9me3 catalyzes further deposition of H3K9me on adjacent nucleosomes, thus promoting the spreading out of heterochromatin factors via the “readwrite” mechanism.²¹

Recent work has demonstrated the importance of a critical density of H3K9me3 to support the read-write mechanism for heterochromatin spreading.¹³⁴ This was first revealed when an unbiased genetic screen identified a mutant histone H3, which cannot be methylated, was incorporated into chromatin and observed to reduce the H3K9me3 density and heterochromatin spreading in a dosage-dependent manner. Interestingly, increasing the affinity of Clr4/Suv39h for methylated H3K9 was sufficient to overcome this defect in heterochromatin spreading in mutant cells with low H3K9me3 density.¹³⁴ These findings uncovered an important principle, namely that an effective local concentration of Clr4/Suv39h bound to chromatin is required for the spreading of heterochromatin. Once the concentration threshold is achieved, a feed forward cascade of Clr4/Suv39h read-write activity can support the propagation of heterochromatin across extended domains.

High H3K9me density similarly helps in concentrating HP1, which also plays an important role in heterochromatin spreading.⁶⁴ The oligomerization of Swi6/HP1 on chromatin¹³⁵ and its ability to associate with other proteins enable the formation of a high local concentration gradient of factors required for heterochromatin spreading. Swi6/HP1 is thought to facilitate the chromatin association of Clr4/Suv39h bound to H3K9me3, thus reinforcing its read-write capabilities.^64,124 Moreover, HP1 proteins recruit HDACs, which are also required for heterochromatin spreading.⁴⁶ Deacetylation of histones by HDACs maintains high levels of H3K9me3 by protecting nucleosomes from destabilizing activities (such as chromatin remodelers) that require acetylated histones.¹³⁶ The removal of the acetyl moiety is also believed to facilitate the ubiquitylation of histone H3K14, which stimulates Clr4/Suv39h activity to maintain high levels of H3K9me3 required for heterochromatin spreading.^137,138 Thus, structural heterochromatin proteins and histone-modifying activities can establish positive feedback loops that perpetuate heterochromatin spreading over large distances.

A read-write mechanism similar to that in S. pombe has been implicated in heterochromatin spreading in other systems. Mammalian Suv39h was found to possess read-write activity, akin to that of Clr4.²³ The read-write mechanism has also been linked to the propagation of H3K27me3 and Polycomb proteins across silenced domains in Drosophila and mammals.^139–142 In these cases, a subunit of the Polycomb repressive complex 2 (PRC2) called extra sex combs (ESC) in Drosophila and ectoderm development (EED) in mammals, recognizes H3K27me3, which in turn activates the SET domain methyltransferase (enhancer of zeste or E(z) in mammals and enhancer of zeste homolog 2 or EZH2 in Drosophila) to propagate the H3K27me3 modification.¹⁴⁰ This process is likely influenced by factors affecting histone acetylation, as in S. pombe, with HDACs being preferentially enriched at Polycomb target loci.¹⁴³

Self-templated heterochromatin inheritance

A fascinating characteristic of heterochromatic silencing is that it is faithfully transmitted to daughter cells during cell division.^64,144 This process guarantees the maintenance of cell-type specific transcriptional profiles by preventing the expression of lineage-inappropriate genes.⁶ Genetic and biochemical studies using S. pombe have provided evidence that this inheritance is accomplished in cis via a self-templating mechanism in which H3K9me3 serves as a carrier of epigenetic information as shown at the endogenous silent mat region^{64,134,144,145} and subsequently at ectopic loci.^146–148 Here, I will review the evidence for the self-propagation of heterochromatin domains and discuss the core principles that govern H3K9me3-dependent heterochromatin heritability.

As mentioned above, heterochromatin establishment at the silent mat region in S. pombe involves both RNAi-dependent and -independent mechanisms: RNAi acts via the cenH element, and ATF/CREB family proteins act together with Swi6/HP1 to nucleate heterochromatin nearby (Figure 3). Wild-type cells exhibit synergistic actions of these two mechanisms to assemble heterochromatin.¹²¹ By contrast, cells lacking cenH or the RNAi machinery infrequently show de novo heterochromatin assembly, as DNA binding proteins and Swi6/HP1 by themselves nucleate heterochromatin inefficiently and in a stochastic manner. However, once established, the heterochromatic state in cells lacking cenH is stably propagated in a clonal fashion.^{64,144,145,149} This results in genetically identical cells displaying alternative chromatin states, differing only in the presence of heterochromatin at the mat region. Cells carrying the expressed state are deficient in heterochromatin, while cells exhibiting the silenced state show enrichment of H3K9me and Swi6/HP1.^64,145 This system, wherein the silenced state cells that carry the same DNA sequence as the expressed state cells can faithfully maintain heterochromatin, revealed that some form of epigenetic memory facilitates heterochromatin propagation.

Further analyses suggested that epigenetic memory is stored in heterochromatin itself.¹⁴⁵ When heterochromatin is disrupted by transient treatment of cells with HDAC inhibitors, the epigenetic memory is erased, and the silenced state is heritably converted to the expressed state. Conversely, heterochromatin established upon transient Swi6/HP1 overexpression is stably propagated for multiple generations. These experiments demonstrate the persistence of heterochromatin in the absence of an initiating stimulus (e.g., Swi6/HP1 overexpression) and provided the key evidence that heterochromatin itself comprises epigenetic memory.

Previously, the role of epigenetic memory in heterochromatin inheritance had been inferred from the position effect variegation (PEV) phenomenon in Drosophila, where the heterochromatic state is clonally propagated.²⁶ However, it was uncertain whether the observed phenotypes were simply due to differences in the concentration of a critical protein between expressed and silenced state cell populations (e.g., HP1 levels affecting PEV). The direct demonstration that heterochromatin promotes its own reassembly in a self-templating manner required a unique genetic strategy. In a foundational experiment, genetically identical silenced- or expressed-state haploid cells, differing solely in the presence or absence of heterochromatin at the silent mat region, were crossed to make a diploid (Figure 4).^64,144 Notably, the distinct heterochromatin patterns defining each epigenetic state were maintained in the same nuclear environment of a diploid cell, and the H3K9me patterns defining silenced (H3K9me enriched) and expressed (H3K9me depleted) epialleles were inherited in cis during mitosis and meiosis (Figure 4).⁶⁴ This suggests that epigenetic inheritance of heterochromatin is not due to differences in the levels of trans-acting factors, but rather that heterochromatin promotes its own inheritance in a self-templating manner. The self-propagation of the heterochromatin during mitosis and meiosis further illustrated that the unit of inheritance, the gene, can constitute more than DNA.^64,145

Figure 4. Heterochromatin is epigenetically inherited in cis in a self-templating manner.

In a foundational experiment, genetically identical haploid cells differing only in the level of heterochromatin at the silent mat region were crossed to make a diploid. Remarkably, this experiment showed that the differential H3K9 methylation patterns are maintained in the shared nuclear environment of the diploid cell. During replication, heterochromatin provides a template for its own reassembly, so not only is DNA replicated but the chromatin also replicates in a self-templating manner, leading to the propagation of differential chromatin states for multiple generations. Importantly, these distinct chromatin states are epigenetically propagated in mitotically dividing cells and are transmitted through gametogenesis into gametes. See Hall et al. (2002)⁶⁴ for details.

The cis inheritance of epigenetic memory is a critical feature of mechanisms that require self-propagation of chromatin states. Experimental approaches that combine active and inactive epigenetic states, defined by alternative chromatin structures, in the same nucleus have now been used in various systems^150,151 and to study the heritability of an ectopic heterochromatin domain in S. pombe.¹⁴⁷ The faithful propagation of the heterochromatic state in cis suggests that self-templated epigenetic inheritance of repressive heterochromatin is a general feature.

Mechanism of self-templated heterochromatin inheritance

DNA methylation, which can influence histone methylation and vice versa, has been shown to transmit epigenetic memory.^152–154 DNMT1, the maintenance DNA methyltransferase, recognizes 5meC in CG dinucleotides and catalyzes methylation of the CG on the complementary strand, thus restoring the fully methylated state.³ However, heterochromatin can be inherited epigenetically independently of DNA methylation. In the case of S. pombe - which lacks DNA methylation - heterochromatin heritability relies on methylated H3K9 as the critical component of the “epigenetic loop” mechanism responsible for the self-propagation of heterochromatin.^21,134 Two key principles govern how modified histones mediate the epigenetic inheritance of heterochromatin (Figure 5).

Figure 5. Fundamental principles underlying self-propagation of heterochromatin domains.

Two principles govern the stable propagation of heterochromatin: (1) Epigenetic self-templated inheritance of heterochromatin domains requires a high density of H3K9me3 to support an effective local concentration of Clr4/Suv39h on chromatin. (2) The ability of the Clr4/Suv39h methyltransferase to both bind to and deposit H3K9me, referred to as the read-write mechanism, is critical for heterochromatin propagation. HDACs recruited by HP1 proteins and other factors suppress histone turnover and maintain a high density of H3K9me3. During DNA replication, just as parental DNA strands act as a template, parental H3K9me3 transferred to sister chromatids and HP1 provide an epigenetic template for the recruitment of Clr4, which in turn modifies new histones to propagate heterochromatin through its read-write activity.

The first principle involves the ability of the histone methyltransferase Clr4/Suv39h to both "read" and "write" H3K9me, which allows for both spreading and epigenetic inheritance of heterochromatin domains during cell division (Figure 5).²¹ This read-write capability of Clr4/Suv39h allows for the propagation of heterochromatin even when it is artificially targeted to an ectopic site.^146–148 When cells undergo DNA replication, H3-H4 tetramers generated upon disassembly of modified parental nucleosomes are distributed to sister chromatids, along with new histones, to restore nucleosome occupancy.^155–157 The faithful transfer of these parental modified histones, which are deposited at their original location on daughter strands,^158,159 requires factors linked to both DNA replication and chromatin assembly proteins. H3K9me3 serves as a "molecular bookmark" to provide an epigenetic template for loading Clr4/Suv39h via binding of its chromodomain to the H3K9me3 (Figure 5).²¹ Additionally, HP1 helps facilitate this process,^64,145 forming a positive feedback loop that underlies the self-propagation of heterochromatin. Clr4/Suv39h, anchored to chromatin via H3K9me3 and Swi6/HP1 (which also requires an H3K9me foundation), then modifies newly assembled neighboring nucleosomes to match the parental methylation pattern, leading to clonal propagation of heterochromatin in cis. The most direct evidence showing that H3K9me3 serves as the carrier of epigenetic information comes from a genetic study demonstrating that a mutation in the histone H3 tail that reduces H3K9me3 levels compromises heterochromatin propagation.¹³⁴ Histone tail mutation could formally result in general defects in heterochromatin assembly. However, simply enhancing the affinity of Clr4 for H3K9me restored heterochromatin propagation in cells expressing mutant histone H3, thus underscoring the importance of H3K9me3 and the read-write activity of Clr4 in self-propagation of heterochromatin.

The second principle of epigenetic inheritance of heterochromatin involves a critical density of H3K9me3, where high levels of H3K9me3 are necessary to support an increased local concentration of Clr4/Suv39h to promote the read-write activity.^134,148 Factors that stimulate Clr4 activity^137,138 likely contribute to maintaining high H3K9me3 levels. Additionally, HDAC enzymes like the S. pombe Clr3 that is recruited to heterochromatin domains by HP1-dependent and -independent mechanisms help to stabilize nucleosomes carrying H3K9me3 to support heterochromatin propagation (Figure 5).^148,160 Indeed, it was found that robust chromatin association of HDAC Clr3 is necessary and sufficient to support epigenetic inheritance of heterochromatin via preexisting H3K9me3.¹⁴⁸ The high concentration of Clr3 bound to chromatin is believed to shift the dynamic balance that exists between the opposing activities of HDACs and histone acetyltransferases (HATs) in favor of HDACs, which ultimately safeguard H3K9me3 modified nucleosomes to transmit epigenetic memory for heterochromatin propagation via the read-write mechanism. These findings explain why heterochromatin targeted to chromosomal regions with low HDAC enrichment is not inherited unless HATs are deleted.¹⁴⁸

Given the requirement for high H3K9me3 density in self-templated heterochromatin inheritance, multiple pathways likely function to preserve this epigenetic modification.^161–167 Among other factors contributing to preserving the H3K9me3 are replication proteins, the histone chaperone FACT, and the SMARCAD1 homolog Fft3,^168–172 which are believed to help retain preexisting histones during replication.^{155–157,173,174} Additionally, heterochromatin propagation is facilitated by the positioning of heterochromatin domains at the nuclear periphery.^168,175 Peripheral tethering is believed to create a specialized nuclear subdomain that favors efficient loading of factors that suppress histone turnover.¹⁶⁸ Specifically, a high local concentration of Swi6/HP1 in the peripheral subdomain helps with the chromatin association of FACT and other proteins essential for preserving epigenetic memory in the form of modified histones. Loss of factors involved in peripheral tethering or FACT increases histone turnover, which ultimately causes a reduction in H3K9me3 density and defective heterochromatin propagation.^168,170,176

The read-write capability of methyltransferases, coupled with the high density of methylated nucleosomes maintained by HDACs and other factors, are likely features of a conserved strategy for epigenetic inheritance of silenced chromatin domains. Evidence for this concept can be found in the fact that chromosome regions silenced epigenetically by Polycomb display low histone turnover.^177,178 HDACs bound to these regions might suppress histone turnover to maintain a high density of H3K27me3, a modification that promotes the read-write activity of Polycomb. Additionally, chromatin modifications such as DNA methylation that target HDACs may also help to enhance the heritability of heterochromatin. For example, in Neurospora, heterochromatin domains enriched for DNA methylation, HP1, and HDACs show low histone turnover.¹⁷⁹ In mammals, X-chromosome inactivation requires Polycomb, HDAC3, and the SPEN protein, which is targeted to the inactive X-chromosome by the XIST lncRNA.^118,180,181 It has been suggested that a high local concentration of SPEN that associates with HDAC3 is critical for gene silencing.^117,118 These findings, in combination with the observations from the S. pombe system, suggest that SPEN-associated HDAC activity may help to stabilize nucleosomes, thus promoting gene silencing and facilitating Polycomb-mediated heterochromatin propagation across the X-chromosome.

HP1 condensates in heterochromatin propagation and silencing

HP1 dimers bound to H3K9me can bridge nucleosomes.^{135,182–184} Additionally, HP1 is involved in non-stoichiometric interactions and has been observed to undergo liquid phase separation,^33–35 although it is still a matter of debate whether this happens in vivo.¹⁸⁵ Regardless, the ability of HP1 to engage in non-stochiometric interactions and form high-concentration nuclear territories has important implications. The formation of such HP1 territories is likely made more efficient by tethering heterochromatic loci to nuclear structures, such as the nuclear periphery, which effectively limits the volume in which silencing factors are concentrated. To this end, it is important to note that HP1 is a dosage-critical heterochromatin assembly factor.^26,145,186 Among other functions, it facilitates the accumulation of enzymes, such as HDACs and methyltransferases, which are required for heterochromatin propagation. As mentioned above, HP1 association with HDACs, whose robust chromatin association is required to maintain modified nucleosomes,¹⁴⁸ preserves the critical density of H3K9me3, which in turn serves as an epigenetic template for read-write activity of Clr4/Suv39h. On the other hand, HP1 directly supports the read-write mechanism by maintaining high local concentrations of Clr4/Suv39h on chromatin.^124,187 Thus, HP1 territories serve to enrich the factors required for heterochromatin assembly and propagation. Other HP1-associated factors, such as mammalian ATRX-DAXX that bind to H3/H4 dimers,¹⁸⁸ may benefit from an environment rich in methyltransferase and HDAC enzymes, as exposure of their associated histones to these activities could promote the assembly of nucleosomes bearing heterochromatic modifications and ultimately stabilize heterochromatin. Similarly, the histone chaperone FACT is believed to rely on high-concentration HP1 compartments for its efficient engagement to support the faithful inheritance of epigenetic information at heterochromatic loci.¹⁶⁸

It is postulated that HP1 can create structures, such as liquid condensates, that exclude transcription machinery to trigger gene silencing. In this regard, it is notable that the S. pombe anti-silencing factor Epe1, which promotes RNAPII transcription, is found in heterochromatic HP1 foci.^189,190 Moreover, recent evidence indicates that heterochromatin foci mix with the same nucleoplasm as euchromatin, and that they may not be exclusionary in nature.¹⁸⁵ Thus, the model that HP1 liquid condensates directly create an environment impermeable to activating factors may not be fully supported. Rather, an alternative model might include the view that silencing is determined at the level of nucleosome stability, which is inextricably linked to the local concentration of HP1-associated activities. Among the key factors that determine nucleosome stability are HDAC enzymes, which remove acetylation from histone tails. The removal of histone lysine acetylation that neutralizes the positive charge could directly stabilize nucleosomes. However, a dominant view is that deacetylation of histones prevents the engagement of chromatin remodeling enzymes that bind to the acetylated histone modification via their bromodomain and destabilize nucleosomes.¹³⁶ Indeed, histone deacetylation, which is a conserved feature of silenced chromatin domains from yeast to mammals, maintains heterochromatin in a state that contrasts with euchromatin, where acetylated histone tails are engaged by bromodomain-containing remodeling enzymes to open chromatin (Figure 6A). Supporting this view, the removal of HDACs leads to a higher rate of histone turnover,¹⁶⁰ which may allow greater exposure of the underlying DNA sequence to transcriptional regulatory factors. This HDAC-mediated mechanism likely acts in concert with other factors including the bridging of nucleosomes by HP1, to coordinate various signaling inputs to dynamically regulate chromatin structure.

Figure 6. Robust chromatin association of HDAC activity is required to preserve the H3K9me3 epigenetic template.

(A) Heterochromatin differs dramatically from euchromatin in the rate of histone turnover. Stably propagated heterochromatin domains show preferential enrichment of HDACs recruited by HP1 and DNA-binding proteins. As a result, the dynamic balance between HDACs and HATs is shifted in favor of HDACs. Deacetylation by HDACs preserves methylated nucleosomes in part by suppressing turnover of histones, presumably by preventing the engagement of bromodomain-containing ATP-dependent chromatin remodeling complexes, which have a higher affinity for nucleosomes with acetylated H3 tails.¹³⁶ On the other hand, euchromatic regions enriched for HATs show high histone acetylation and turnover, particularly at regulatory elements (shown in red) such as gene promoters targeted by remodeling enzymes. (B) Propagation of a heterochromatin domain established by artificial recruitment of Clr4. TetR-Clr4 binds to tet operators (tetO) at an ectopic site via a fusion of the Clr4 catalytic domain to the TetR DNA binding domain. Methylation of H3K9 by TetR-Clr4 recruits HP1 and associated HDACs as well as the endogenous Clr4 complex that spreads via the read-write mechanism. Upon the release of TetR-Clr4, epigenetic inheritance of heterochromatin can only occur if the high density of H3K9me3 is maintained by HDACs. (Top) The low level of HDACs at an ectopic heterochromatin domain located in an otherwise euchromatic environment is insufficient to maintain the high density of H3K9me3 and Clr4 required for heterochromatin propagation. (Bottom) However, restoring robust chromatin association of an HDAC (Clr3 in S. pombe) by fusing it to two chromodomains (HDAC-CDx2), which preserves H3K9me3, is sufficient to prime the system to support self-propagation of heterochromatin through the read-write mechanism.¹⁴⁸ Heterochromatin propagation at the ectopic site can also be recapitulated by loss of the major HATs (not shown), further reinforcing the idea that it is the dynamic balance between opposing activities of HDACs and HATs at a given locus that determines heritability of the silenced chromatin domain.

Broader implications of H3K9me3 density for heritability

Heterochromatin domains assembled throughout the genome display marked differences in both their inherent stability and ability to be epigenetically inherited. Chromosomal context-specific differences in the mechanisms that maintain H3K9me3 density are responsible for this variability in heterochromatin inheritance (Figures 6A and 6B). For example, in contrast to the self-propagation of heterochromatin observed at the silent mat region in S. pombe,^64,144,145 heterochromatin targeted through artificial tethering of Clr4/Suv39h to an ectopic site within a euchromatin domain cannot be propagated in otherwise wild-type cells.^146,147 Recent work has shown that the key factor limiting this process is the difference in the recruitment of HDACs, which modify chromatin to ensure a high H3K9me3 density.¹⁴⁸ Specifically, the HDAC Clr3 is enriched at the silent mat region, but not at the euchromatin ectopic site. However, increased chromatin association of this HDAC is sufficient to maintain H3K9me3 and enable self-propagation of the ectopic heterochromatin domain (Figure 6B).^148,189

The discovery of the impact of HDAC enrichment on heterochromatin stability may have far-reaching implications. Whether a heterochromatin domain is constitutive or relatively dynamic in nature correlates with levels of its HDAC enrichment, such that stable heterochromatin domains display much higher HDAC enrichment than dynamic domains. These findings also provide a framework for understanding how various factors influence heterochromatin propagation. For instance, the loss of the anti-silencing factor Epe1 allows for stable heterochromatin propagation,^146,147,161 similar to the increased chromatin association of the Clr3 HDAC.¹⁴⁸ While Epe1 contains a JmjC domain and has been suggested to be an H3K9me demethylase that compromises heterochromatin stability,^146,147 no demethylase activity has yet been demonstrated. Furthermore, mutations that are expected to abolish any such Epe1 demethylase activity impair its association with Swi6/HP1,¹⁸⁹ which recruits this anti-silencing factor to heterochromatic loci.¹⁹⁰ Considering that Epe1 also associates with HATs,¹⁹¹ it is plausible to speculate that Epe1 may affect the balance between HDACs and HATs, counteracting HDACs and destabilizing heterochromatin. In this regard, Epe1 is known to promote histone turnover,¹⁶⁰ which may be stimulated by its associated HAT activity.

The requirement to maintain a critical H3K9me3 density explains how multiple pathways, including DNA-binding proteins, can converge to reinforce the heritability of silenced chromatin domains (Figure 7).^{141,142,192,193} The assembly and propagation of a heterochromatin domain depend on a high density of methylated histones, which are maintained by factors that are recruited by a combination of mechanisms. If one or more of these recruitment mechanisms, including those involving DNA-binding proteins, is defective, it will diminish local concentrations of methylated histones and their associated proteins, resulting in compromised heterochromatin assembly (Figure 7). However, defective heterochromatin can be rescued by strengthening other recruitment mechanisms. Indeed, when H3K9me3 is stabilized by increasing the chromatin association of an HDAC, it is sufficient to propagate heterochromatin at an ectopic site in a self-perpetuating manner for multiple generations (Figures 6B and 7).¹⁴⁸ Self-propagation of heterochromatin by stabilized H3K9me3 demonstrates the minimum requirement for epigenetic inheritance occurring with no input from DNA/RNA-based recruitment mechanisms, and without deleting anti-silencing factors such as Epe1.

Figure 7. Overlapping mechanisms maintain a critical density of methylated histones that serve as the carriers of epigenetic information.

Normally, multiple pathways mediate the recruitment of histone-modifying enzymes, such as Clr4/Suv39h and HDACs, which collaborate to maintain a high local concentration of H3K9me3 to support read-write heterochromatin propagation. If RNA- or DNA-based mechanisms, or both, are impaired, the levels of H3K9me3 and its interacting proteins maintained mainly by the read-write mechanism vary depending on the chromosomal context and other conditions affecting nucleosome stability, thus resulting in a metastable state and defective heterochromatin propagation. Defects in DNA-mediated and RNA-mediated mechanisms can be bypassed by enhancing H3K9me-mediated recruitment via alternative mechanisms. For example, when H3K9me3 is stabilized (indicated by dark blue nucleosomes) by enhanced chromatin association of HDACs (e.g. by fusion of Clr3 HDAC to chromodomains), which acts in part by suppressing histone turnover, H3K9me3 modified nucleosomes alone are sufficient to support self-propagation of the heterochromatin domain for multiple cell divisions.

In mammalian cells, H3K9me3 and heterochromatic silencing, which is induced when HP1 is tethered at an ectopic site, can persist for multiple generations even after the initial stimulus has been removed.¹³¹ Heterochromatin propagation can occur independently of DNA methylation, although its stability is enhanced by high levels of DNA methylation. Quantitative modeling suggests that heterochromatin heritability is governed by the dynamic competition between the rate of H3K9 methylation and histone turnover. Therefore, to sustain the epigenetic inheritance of heterochromatin in mammals, it is necessary to suppress histone turnover and maintain high levels of H3K9me3, as in S. pombe.^{148,160,168,170,172} DNA methylation and other factors, such as HDACs, could also contribute to heterochromatin propagation, possibly through the maintenance of a high local concentration of H3K9me3.

Perspective and conclusions

Exploring the mechanisms governing the assembly and epigenetic inheritance of heterochromatin domains has revealed the central role modified histones play in the transmission of epigenetic memory. Maintaining a critical density of modified nucleosomes and HP1 has emerged as an important requirement for parental methylated histones (such as H3K9me3) to promote epigenetic inheritance of heterochromatin via the read-write mechanism. In this scheme, the control of histone turnover and the stability of nucleosomes carrying epigenetic information can be exploited to modulate the heterochromatic state and gene silencing in response to environmental and developmental signals. Indeed, several studies have highlighted the importance of stabilizing heterochromatin domains, which involves changes in the levels of factors known to affect histone turnover, for cellular adaptation to changing growth conditions.^{109,194–196}

In mammals, the levels of H3K9me3 and heterochromatin progressively increase during early development.^6,197 This upsurge in heterochromatin, which is critical to repress lineage-inappropriate gene expression, may also involve a shift in the balance between opposing chromatin-modifying activities affecting histone turnover, leading to an increase in H3K9me3 density. Such a shift could explain differences in heterochromatin heritability observed between naïve and differentiated cells.¹⁹⁸ Furthermore, H3K9me3-coated heterochromatin domains have been shown to act as an impediment to the reprogramming of somatic cell nuclei transferred into oocytes,¹⁹⁹ or the conversion of differentiated cells into induced pluripotent stem cells (iPS).^200,201 Disruption of heterochromatin barriers likely requires a shift in balance to favor nucleosome destabilizing activities and/or demethylases, thus causing a decrease in H3K9me density. Indeed, perturbations of factors implicated in heterochromatin propagation, including the inhibition of histone deacetylases that preserve the H3K9me3 epigenetic template,¹⁴⁸ enhance reprogramming efficiency.²⁰²

The discovery that stably maintained H3K9me3 promotes epigenetic inheritance of heterochromatin through both mitosis and meiosis raises the possibility that modified histones contribute to transgenerational epigenetic inheritance. In this regard, it is important to note that many loci escape the erasure of their epigenetic modifications during global reprogramming in the human germline.²⁰³ These loci are enriched for H3K9me3 and contain binding sites for KRAB-ZFP/KAP1, which is implicated in heterochromatin propagation ²⁰³. Among other functions, KAP1 associates with HDACs,^61,62 which may help maintain a high density of H3K9me3, thus creating a barrier to epigenetic reprogramming. In this respect, proximity to specific DNA elements, such as transposable elements, which are known to be targeted by HDACs,²⁰⁴ likely facilitates epigenetic inheritance by providing a chromatin environment favoring the retention of modified histones and heterochromatin propagation.

Given the importance of heterochromatin in controlling the activity and stability of eukaryotic genomes, further mechanistic advances promise to shed light on how heterochromatic structures are assembled and stably propagated to prevent untimely gene expression and chromosomal abnormalities, which underlie many human diseases, including cancer.