Transcriptional Scaffolds for Heterochromatin Assembly

Abstract

Heterochromatin is dynamically regulated during the cell cycle and in response to developmental signals. Recent findings from diverse systems suggest an extensive role for transcription in the assembly of heterochromatin, highlighting the emerging theme that transcription and noncoding RNAs can provide the initial scaffold for the formation of heterochromatin, which serves as a versatile recruiting platform for diverse factors involved in many cellular processes.

DNA-mediated processes in the eukaryotic nucleus occur on a chromatin template greatly influenced by DNA methylation and covalent histone modifications. Posttranslational modifications of his-tones permit different layers of regulatory control to be imposed to give rise to diverse chromatin states and architectures. Heterochromatin is a unique type of chromatin that is characterized by its transcriptionally repressed state and highly condensed structure. Emerging evidence reveals a convergent view of the molecular mechanisms that underlie heterochromatin assembly in diverse processes that range from pericentromeric heterochromatin formation in fission yeast to RNA-directed DNA methylation in plants to X chromosome inactivation (XCI) and imprinting in mammals. This Essay highlights recent findings that contribute to the evolving view of heterochromatin, with emphasis on the roles of transcription and RNA molecules in its assembly and function.

Forming the Heterochromatin Platform

The assembly of heterochromatin is believed to be a multistep process. Heterochromatin structures are nucleated at specific regulatory sequences and can spread into neighboring sequences, thereby influencing gene expression in a region-specific manner. Importantly, the ability of heterochromatin to propagate far from its original nucleation site provides a molecular platform for the recruitment of effector complexes involved in various chromosomal processes (Gre wal and Jia, 2007). Studies from diverse systems suggest that a common set of structural components contribute toward the construction of the heterochromatin platform. Aside from the budding yeast Saccharomyces cerevisiae, which relies on a different set of protein complexes to assemble heterochromatin, heterochromatin protein HP1 and its binding substrate, histone H3 methylated at lysine 9 (H3K9me), are markers of heterochromatin in most eukaryotes. H3K9me is catalyzed by the histone methyltransferase Suv39h in humans, SU(VAR)3–9 in the fly Drosophila melanogaster, and Clr4 in the fission yeast Schizosaccharomyces pombe. HP1, originally characterized in Drosophila, consists of a family of proteins conserved from fission yeast (Swi6, Chp1, Chp2) to humans (HP1α, HP1β, HP1γ). HP1 proteins bind to H3K9me through their chromodo-mains, thereby facilitating HP1 localization at heterochromatic loci (reviewed in Grewal and Jia, 2007). Together with Clr4, which also binds to H3K9me via its own chromodomain (Zhang et al., 2008), HP1 proteins not only promote the propagation of heterochromatin but also the recruitment of protein complexes with various enzymatic activities essential for the assembly of repressed chromatin (Grewal and Jia, 2007). The presence of numerous effectors that are involved in different chromosomal processes at the heterochromatin domain contributes to the growing appreciation of the versatility of heterochromatin as a dynamic platform. For example, heterochromatin could serve to recruit factors involved in diverse processes such as cell-type switching, sister chromatid cohesion, RNAi, and transcriptional gene silencing (reviewed in Grewal and Jia, 2007).

Transcriptional Scaffolds for Heterochromatin Assembly

Depending upon the chromosomal context, the strategies used to target heterochromatin differ in different organisms. In addition to DNA binding factors that are important for the targeting of heterochromatin, recent evidence suggests that transcription, in particular noncoding products of transcription, play critical roles in heterochromatin assembly. In mammals, transcription of long non-coding RNAs (ncRNAs) such as Xist and Tsix (involved in XCI) and Kcnq1ot1 and Air (implicated in imprinting) are essential for the onset of the silent chromatin state (Nagano et al., 2008; Pandey et al., 2008; Zhao et al., 2008) (see Review by C.P. Ponting, P.L. Oliver, and W. Reik in this issue of Cell). Moreover, recruitment of Polycomb group chromatin remodeling complexes for the initiation and spread of XCI is mediated by direct binding of the Polycomb group complex PRC2 (Polycomb repressive complex 2) to short RNA repeats within Xist (Zhao et al., 2008). Also, Air ncRNA interactions with the H3K9 histone methyltransferase G9a and chromatin at gene promoters are critical for imprinted silencing of the Slc22a3 gene in the mouse placenta (Nagano et al., 2008). A growing body of evidence also points to the importance of both RNAi and transcription in RNA-directed DNA methylation in plants. However, it appears that plants use RNA polymerases distinct from RNA polymerase (Pol) I, II, and III for heterochromatin assembly. In the plant Arabidopsis thaliana, RNA Pol IV and Pol V, which are essential for small interfering RNA (siRNA)-mediated gene silencing of transposons, also facilitate heterochromatin formation and silencing of genes adjacent to transposons (Wierzbicki et al., 2008). Transcription of target loci by RNA Pol IV generates RNAs that are processed into 24 nucleotide siRNAs by RNA interference (RNAi) machinery that includes the RNA-dependent RNA polymerase RDR2 and the DICER homolog Dicer-like 3 (DCL3) (reviewed in Henderson and Jacobsen, 2007). siRNAs are then loaded into the Argonaute family protein AGO4 that, together with RNA Pol V, helps guide DRM2 (domains rearranged methyltransferase 2), a DNA methyltransferase with de novo cytosine methylation activity, to methylate target sequences for epigenetic silencing (reviewed in Henderson and Jacobsen, 2007).

In fission yeast, transcription of the centromeric DNA repeats is critical for RNAi-mediated assembly of heterochromatin (reviewed in Grewal and Jia, 2007; Buhler and Moazed, 2007). This process involves the RNA-induced transcriptional gene silencing (RITS) complex, which consists of the proteins Ago1 (an Argonaute homolog), Chp1, and Tas3, as well as an RNA-dependent RNA polymerase complex (RDRC) containing Rdp1, Hrr1, and Cid12. These protein complexes coordinate their functions with the Dicer enzyme Dcr1 to mediate processing of centromeric repeat transcripts into siRNAs. They also facilitate the targeting of heterochromatin assembly factors such as the Clr4-containing ClrC complex to centromeric repeat sequences. Importantly, accumulating evidence suggests that RNAi and histone-modifying factors cooperate with RNA Pol II and its associated complexes to nucleate heterochromatin. Indeed, defects in the RNA Pol II machinery severely compromise both the processing of repeat transcripts into siRNAs and the assembly of heterochromatic structures (see for example, Djupedal et al., 2005). It is believed that RITS interacts with nascent RNA Pol II transcripts through Ago1-bound siRNAs (Motamedi et al., 2004) and, together with its associated RDRC complex, contributes to centromeric silencing by mediating the processing of repeat transcripts into siRNAs (reviewed in Buhler and Moazed, 2007; Grewal and Jia, 2007). Interestingly, effective centromeric silencing also requires splicing factors that associate with the RDRC subunit Cid12 and function as a platform for siRNA generation by the RNAi machinery (Bayne et al., 2008).

Considering the involvement of RNAi complexes in posttranscriptional silencing of heterochromatic repeats, it is perhaps not unexpected that these factors are also associated with repeat transcripts. However, the more surprising finding has been that RNA Pol II transcription of repeat loci is closely linked to the initial loading of chromatin-modifying factors involved in heterochromatin assembly (Chen et al., 2008; Zhang et al., 2008). Two components of the ClrC complex, Rik1 and Clr4, play critical roles in integrating transcription and RNAi processes for heterochromatin assembly. The WD-β-propeller domain-containing protein Rik1 bridges the interaction between ClrC and the nascent transcript, presumably through its putative RNA-binding domain and interaction with the RITS complex (Zhang et al., 2008). Clr4, on the other hand, catalyzes the methylation of H3K9 to provide binding sites for HP1 family proteins. Transcription-coupled targeting of ClrC might be analogous to the recruitment of histone methyltransferases by RNA Pol II to modify chromatin in transcribed regions of chromatin (Li et al., 2007).

These findings suggest that what distinguishes heterochromatin may not be so much the lack of transcription, but rather RNA Pol II transcription being “hijacked” for the assembly of heterochromatin. This transcriptional hijacking appears to be highly regulated, involving both the RNA Pol II machinery and its attendant complexes in heterochromatin assembly. A scaffolding role for RNA Pol II transcription in heterochromatin assembly may help explain why a growing set of histone-modifying enzymes commonly associated with euchromatic transcription are also found at heterochromatin.

Linking RNA Synthesis to Heterochromatin Targeting

Findings that we have reported suggest that in fission yeast, a self-reinforcing loop mechanism involving siRNA synthesis drives heterochromatin assembly (Grewal and Jia, 2007 and references therein). In this mechanism, RITS bound to H3K9me through the chromodomain of Chp1 and heterochromatin proteins (such as Swi6) allow the RNAi machinery to act as a stable component of chromatin to pro cess RNA transcripts from centromeric repeats into siRNAs. This continuous siRNA generation functions in a positive feedback loop to further stabilize heterochromatin (Figure 1). Interestingly, RNAi-mediated heterochromatin formation appears to be primarily restricted to sites of siRNA production (reviewed in Buhler and Moazed, 2007). However, the mechanism that limits siRNAs to the recruitment of heterochromatin factors (such as the ClrC complex) only in cis remains poorly understood. siRNAs are thought to facilitate proper positioning of Ago1 for slicing target transcripts, thereby stimulating RDRC synthesis of double-stranded RNAs (dsRNAs) (reviewed in Buhler and Moazed, 2007; Grewal and Jia, 2007). It is possible that even though heterochromatin assembly is stimulated by interactions between ClrC and siRNA-bound RITS, other aspects of heterochromatic transcription that are not exclusively dependent on the RNAi machinery may also help mediate ClrC recruitment. Consistent with this view, ClrC components are not entirely depleted from heterochromatic repeats in cells lacking the RNAi machinery (Zhang et al., 2008). Possible mediators of ClrC recruitment are the dsRNAs that arise either from transcription of heterochromatic repeats or by the activity of RNA-dependent RNA polymerases (Figure 1). This scenario of ClrC targeting by dsRNAs could explain a number of puzzling observations. For instance, when compromised heterochromatin (for example, in cells lacking histone deacetylases) is rendered permissive for transcription, ClrC subunits display increased binding at the transcribed centromeric repeat sequences that requires the presence of the RNAi machinery (Zhang et al., 2008). Also, the appearance of dsRNAs at certain convergent gene loci can result in the transient formation of heterochromatin (Gullerova and Proudfoot, 2008).

Figure 1. Transcription and Heterochromatin Formation.

Heterochromatin assembly in fission yeast requires coordinated function of histone-modifying enzymes (ClrC and histone deacetylases), HP1 proteins (Chp2 and Swi6), and the RNA interference (RNAi) machinery. RNAi factors include the Dicer enzyme (Dcr1), the RNA-induced transcriptional silencing (RITS) complex, and the RNA-dependent RNA polymerase complex (RDRC) that process centromeric repeat transcripts into siRNAs.

(Top) During S phase of the cell cycle, the relatively open heterochromatic structure permits heightened RNA Pol II activity at centromeric repeats. This, in turn, stimulates the recruitment of heterochromatin-assembly factors such as the ClrC subunit Rik1 and the RITS subunit Argonaute 1 (Ago1), as well as histone H3 lysine 36 methylation by the Set2 methyltransferase implicated in the recruitment of the histone deacetylase (HDAC) silencing complexes such as Clr6 (Chen et al., 2008). Interaction between ClrC and RITS stabilizes their binding to chromatin and facilitates the processing of centromeric repeat RNAs to siRNAs (Zhang et al., 2008). Recruitment of ClrC may also be mediated by downstream siRNA products such as double-stranded RNAs. Methylation of lysine 9 on histone H3 (H3K9me) by the Clr4 subunit of ClrC not only recruits HP1 proteins but also establishes a positive feedback loop by stabilizing the chromatin association of ClrC (via Clr4 chromodomain) and RNAi components such as RITS (via Chp1 chromodomain).

(Bottom) In G2 phase, HP1 proteins bound to H3K9me not only recruit silencing factors such as the HDAC complex SHREC but also an antisilencing factor Epe1 that promotes Pol II transcription. Spreading of HP1 proteins and H3K9me from the original nucleation sites allows heterochromatin to serve as a versatile recruiting platform for factors involved in diverse chromosomal processes. These include RNAi machinery that mediates posttranscriptional silencing in cis (cis-PTGS), HDACs involved in transcriptional gene silencing (TGS), and factors that are essential for genome stability.

Although heterochromatin and bidirectional transcription, which can potentially result in the production of dsRNAs, have both been documented in all known eukaryotes, RNAi is entirely absent in some eukaryotic lineages like Saccharomyces. It is conceivable that in systems either lacking or not using siRNAs, dsRNAs could be the transcriptional “red flag” that triggers the hijacking of the transcriptional machinery for the assembly of heterochromatin.

Transcription, Histone Deacetylases, and Chromatin Organization

The universal role of histone deacetylases (HDACs) in establishing a closed heterochromatic structure has long been documented in many systems. However, recent findings in fission yeast reveal a dynamic interplay between HDACs and heterochromatin-linked transcription. The functions of class I HDACs (S. cerevisiae Rpd3; S. pombe Clr6) provide one such example. Rpd3/Clr6 exists in two distinct complexes. One of these complexes (S. cerevisiae Rpd3L; S. pombe complex I) targets gene promoters. In contrast, the other complex (S. cerevisiae Rpd3S; S. pombe complex II) targets gene bodies, partly through the recognition of Set2-dependent histone H3 lysine 36 methylation (H3K36me) that is associated with RNA Pol II transcript elongation (Li et al., 2007). In fission yeast, Clr6 complexes have analogous functions in the regulation of bidirectional transcription at heterochromatic repeats. Complex I preferentially targets the “reverse” strand of the bidirectional transcription (Nicolas et al., 2007), driven by known promoter elements (Djupedal et al., 2005), whereas complex II suppresses the transcription of the “forward” strand (Nicolas et al., 2007). The “reverse” and “forward” strands of the heterochromatic repeat transcript are equivalent to the sense and antisense strands, respectively, of a typical RNA Pol II gene. Intriguingly, RNA Pol II transcription is also restrained by the action of another class of HDACs, the class II HDAC Clr3. Biochemical characterization of Clr3 reveals that it is a part of a multisubunit SHREC complex containing a SNF2 chromatin-remodeling factor homolog called Mit1 (Sugiyama et al., 2007). Clr3 coordinates its HDAC activity with Mit1 ATPase activity to properly position nucleosomes and create repressive chromatin structures essential for heterochromatic transcriptional silencing. Whether transcription-coupled processes contribute directly to SHREC localization is not known. However, the targeting of SHREC to centromeric repeats requires an H3K9me-HP1 platform that is established by transcription and RNAi machineries (Sugiyama et al. 2007; Grewal and Jia, 2007). Moreover, HDACs can also be recruited by DNA-binding factors such as centromere protein B (CENP-B) to form repressive chromatin structures at transposons and other loci. In both instances, HDACs assemble repressive chromatin, rendering it not only transcriptionally silent but also resistant to recombination, thereby promoting genome stability.

Cell Cycle Regulation of Transcription and Heterochromatin

Heterochromatic sequences are transcribed, and yet heterochromatic structure can exert a repressive effect on transcription by prohibiting RNA Pol II access (reviewed in Grewal and Jia, 2007). How does RNA Pol II gain access to the underlying DNA shielded by heterochromatin? Evidence suggests that in fission yeast, Swi6/HP1 protein not only mediates targeting of HDAC silencing factors but also recruits a JmjC domain-containing antisilencing factor called Epe1 that facilitates transcription of heterochromatic repeats (reviewed in Grewal and Jia, 2007). Moreover, recent studies have suggested that this apparent paradox could be further resolved by a mechanism involving differential temporal modulation of RNA Pol II accessibility during the cell cycle. The epigenetic landscape of heterochromatin is dynamically controlled throughout the cell cycle (Chen et al., 2008; Kloc et al., 2008), leading to the recruitment of disparate factors that confer transcriptional silencing. RNA Pol II occupancy at heterochromatin is restricted for most of the cell cycle (Chen et al., 2008). During the G2 phase, which occupies a large portion of the cell cycle, HP1 family proteins Swi6 and Chp2 aid in the recruitment and spreading of chromatin modifiers such as SHREC and mediate the assembly of repressive chromatin refractory to RNA Pol II transcription (Grewal and Jia, 2007; Sugiyama et al., 2007). During mitosis, however, serine 10 of histone H3 is phosphorylated (H3S10P), resulting in heterochromatin remodeling by interfering with the binding of HP1 proteins to H3K9me. However, the H3S10P-mediated decrease in the chromatin association of HP1 proteins during mitosis does not result in loss of heterochromatic silencing. Instead, the decrease in chromatin-bound HP1 proteins correlates with the recruitment of the condensin complex (essential for chromosome condensation) to heterochromatin to repress centromeric transcription (Chen et al., 2008).

There exists a short “window of opportunity” during S phase in which heterochromatin is relatively accessible to RNA Pol II for transcription of the underlying repeat sequences (Chen et al., 2008; Kloc et al., 2008) (Figure 1). Evidence suggests that this increased S phase transcription coincides temporally with the recruitment of factors involved in heterochromatin assembly at repeat elements (Chen et al., 2008). The upregulation of heterochromatin transcripts occurs preferentially on the “antisense” (forward) strand. Antisense transcripts may generate signals for the docking of heterochromatin assembly factors to repeat nascent transcripts. Indeed, Rik1 (a ClrC subunit) and Ago1 (a RITS complex subunit) are preferentially enriched at heterochromatic repeats during S phase (Chen et al., 2008) (Figure 1). Moreover, elongating RNA Pol II also targets the H3K36me modification associated with enhanced recruitment of HDAC silencing activities to heterochromatic repeats (Chen et al., 2008). Preferential S phase binding of these factors may facilitate heterochromatin reconstitution during G2 when cis-acting posttranscriptional gene silencing (cis-PTGS) by RNAi and transcriptional silencing (TGS) by HDAC activities associated with HP1 proteins (Chp2 and Swi6) coordinately silence heterochromatic sequences (Grewal and Jia, 2007; Sugiyama et al., 2007).

Coupling of S phase transcription with increased heterochromatin assembly activities may also facilitate the maintenance of the heterochromatin state. This could be achieved by similarly promoting the recruitment of heterochromatin assembly factors such as components associated with the ClrC, HDAC, and RITS complexes (Chen et al., 2008). However, this process of transcription-coupled establishment of silenced chromatin is most likely facilitated by parental lysine 9-methylated histone H3s that are believed to be randomly segregated to daughter chromatids during chromo-some replication. H3K9 methylation not only stabilizes the chromatin association of RNAi factors recruited by RNA Pol II transcription, but also engages ClrC (through the Clr4 chromodomain), which could methylate adjacent newly assembled histones. This thus promotes the re-establishment of parental histone modification patterns and heterochromatin assembly (Zhang et al., 2008). Heterochromatin factors have also been reported to associate with histone chaperones such as CAF1 (Quivy et al., 2004). These proteins, which function during S phase, may act in concert with transcription- and H3K9me-dependent mechanisms to reassemble heterochromatin. The combinatorial effects of various heterochromatin stabilization pathways during S phase could enhance the rapid reconstitution of heterochromatin, thereby contributing to the stable inheritance of the heterochromatic state.

Work performed in mammalian cells showed that S phase transcription of heterochromatic repeats initiates before replication of the heterochromatic sequences (Lu and Gilbert, 2007). Transcription is induced in response to cell proliferation and may involve H3S10P or dissociation of HP1 proteins triggered by specific signals during the cell cycle. Whether S phase transcription facilitates heterochromatin assembly in a manner similar to the process of transcription-coupled establishment of silenced chromatin in fission yeast is not known. It is possible that long noncoding RNAs mediate targeting of heterochromatin complexes during S phase in a manner analogous to that proposed for Air or Xist/Tsix RNAs for heterochromatin formation. Alternatively, these transcripts could mediate generation of dsRNAs that in turn provide signals for heterochromatin assembly, as discussed above.

Alternative Roles of RNAi-Directed Heterochromatin

As more knowledge is gained about the molecular constituents and assembly mechanisms of heterochromatin, a growing body of evidence indicates that many characteristics thought to be unique to constitutive heterochromatin are found outside of the heterochromatin domain. For example, certain euchromatic genes are decorated with classical heterochromatic markers such as HP1 and methylated H3K9 that appear to correlate with gene activation rather than gene repression (reviewed in Grewal and Jia, 2007). Intriguingly, a new role for heterochromatin components was recently reported in fission yeast in which bidirectional transcription during the G1 phase of the cell cycle at certain convergent genes has been suggested to induce RNAi-mediated local assembly of heterochromatin structures (Gullerova and Proudfoot, 2008). The presence of heterochromatin at these sites provides a recruiting platform for cohesin, the protein complex responsible for sister chromatid cohesion during cell division. Once localized, cohesin is believed to promote transcription termination between convergent genes during the G2 phase of the cell cycle, thereby preventing further dsRNA synthesis and disassembly of the heterochromatin. Delocalization of cohesin during M phase permits reexpression of dsRNAs and reconstitution of heterochromatin in the subsequent G1 phase of the cell cycle (Gullerova and Proud-foot, 2008). Future studies are needed to attain full understanding of the functions of RNAi and cohesin at these sites and to explore the possible role of transient heterochromatin assembly in genome stability.

Perhaps no example better illustrates the versatility and power of the RNAi-mediated heterochromatin assembly platform in other cellular processes than the process of selective DNA elimination in the ciliate Tetrahymena thermophila. The single-celled Tetrahymena contains two nuclei called the micronucleus and the macronucleus. During mating (conjugation), the differentiation of the new macronucleus from a micronuleus is accompanied by elimination of DNA elements known as internal eliminated segment sequences. Remarkably, this DNA elimination process requires the RNAi machinery and small RNAs that specifically target heterochromatin marks such as H3K9me and chromodomain proteins to the internal eliminated segment sequences (reviewed in Mochizuki and Gorovsky, 2004). Although the exact role of heterochromatin factors in DNA elimination remains to be elucidated, it is tempting to speculate that as in fission yeast, transient formation of heterochromatin platform, instead of recruiting cohesin, facilitates the recruitment of DNA nucleases such as recombinases to trigger DNA elimination.

Transcription in eukaryotic genomes was once thought to be confined to euchromatin. However, our view of heterochromatin has been redefined with the discovery of the roles for the transcription apparatus and its transcripts as heterochromatin scaffolds and as substrates for the generation of small RNAs and dsRNAs that help to form transcriptionally silenced regions. Moreover, recent revelations that most of the genome is transcribed suggest a much broader role for transcription. Given that the epigenetic state of a genetic locus is largely influenced by its transcriptional state, it is possible that a mechanism analogous to the transcription-coupled establishment of silenced chromatin in fission yeast may not only be responsible for the faithful inheritance of heterochromatin but may mediate the general maintenance of chromatin states.