|Home | About | Journals | Submit | Contact Us | Français|
RNA interference (RNAi) is critical for the assembly of heterochromatin at fission yeast centromeres. Central to this process is the RNA-induced Initiation of Transcriptional gene Silencing (RITS) complex, which physically anchors small non-coding RNAs to chromatin. RITS includes Ago1, the chromodomain protein Chp1, and Tas3, which bridges between Chp1 and Ago1. Chp1 is a large protein with, apart from its chromodomain, no recognizable domains. Here we describe how the structured C-terminal half of Chp1 binds the Tas3 N-terminal domain, revealing Chp1's tight embrace of Tas3. The structure also reveals a PIN domain at the C-terminal tip of Chp1 that controls subtelomeric transcripts through a post-transcriptional mechanism. We suggest that the Chp1-Tas3 complex provides a solid and versatile platform to recruit both RNAi-dependent and RNAi-independent gene-silencing pathways for locus-specific regulation of heterochromatin.
Tightly controlled establishment and maintenance of heterochromatin is crucial for proper chromosome segregation, transcriptional control and many other chromosome associated processes. In the fission yeast Schizosaccharomyces pombe, centromeres, telomeres, and the mating type locus are heterochromatic. These regions are governed by a complex system that relies on RNA-guided processes as well as DNA-specific binding events and modification of chromatin1,2. Heterochromatin formation and gene silencing at the centromere is highly dependent on transcription of the pericentromeric dh and dg repeats by RNA polymerase II (Pol II) and the subsequent recruitment of the RNAi machinery. The RNAi system associated with heterochromatin consists of the ribonuclease III (RNase III)–like enzyme Dicer (Dcr1) generating small RNA molecules of 20-30 nucleotides (nt) in length from centromeric transcripts that are amplified by an RNA-dependent RNA polymerase Rdp13. These small RNA products are loaded into the Argonaute family protein Ago1. Processing of transcripts at the centromere by the RNAi machinery targets them for degradation through post-transcriptional gene silencing mechanisms that operate in cis (cis-PTGS)4-6. RNAi is also cruicial for the recruitment of chromatin factors by contributing to the nucleation of the histone 3 lysine-9 (H3K9) specific Clr4 histone methyltransferase and its associated complex (ClrC)7-9.By establishing H3K9 methylation, Clr4 provides the platform for the assembly of a repressive chromatin structure consisting of the HP1 homologous chromodomain proteins Chp2 and Swi6, histone deacetylases, chromatin remodeler and histone chaperone complexes10. At the centromere there are thus two self-enforcing systems acting in parallel, on one hand through transcriptional gene silencing (TGS) and on the other through cis-PTGS.
At the nexus of these two systems the RNA-induced initiation of Transcriptional Gene Silencing (RITS) complex has been identified as a central player, linking the RNAi pathway with heterochromatin formation11,12. RITS contains the chromodomain protein Chp1, the GW protein Tas3 and the argonaute protein Ago1 in complex with a small RNA. Chp1 has been shown to bind the heterochromatic histone H3K9 methyl mark with high affinity13, while Ago1 targets RNA transcripts via its bound siRNA14,15. The cooperation of these two activities mediated through assembly of the RITS complex is essential for the establishment and maintenance of heterochromatin.Tas3 provides the backbone of the RITS complex, binding to Chp1 through its N-terminal domain (NTD), to Ago1 by means of its GW domain, and features a multimerization domain at its C terminus16-20. RITS function is critical for heterochromatin formation because of its crucial role in recruitment of the RNA-dependent RNA Polymerase complex (RDRC) for amplification of siRNA precursor transcripts, as well as in the recruitment of the Clr4 methyltransferase Complex (ClrC) for the deposition of methyl marks7-9,21. So far, the molecular details of these recruitment processes remain unclear.Chp1 is a protein of 960 amino acid residues with the chromodomain located at its N terminus22,23. Its C terminus has been shown to be required for the interaction with Tas316. In order to better understand the structural and functional role of the RITS complex, we have identified the folded core of the Chp1-Tas3 sub-complex and determined its structure through X-ray crystallography. The structure reveals Chp1's tight embrace of Tas3, and the unexpected presence of a C-terminal PIN domain in Chp1. We show that the PIN domain contributes significantly to silencing of subtelomeric transcripts through post-transcriptional gene silencing in a mechanism independent of RNAi.
The complex of full length Chp1 with Tas3 (1-309), deleting the aggregation prone Tas3 C-terminal α-helical motif was obtained by co-expression of the two subunits in the Multibac baculovirus expression system24. In this heterologous expression system, Chp1 and Tas3 behave as an obligatory heterodimer, since Tas3 stability is entirely dependent on co-expression with Chp1. Chp1 can be obtained individually as a OneStrep-SumoStar (OSS) fusion protein, but aggregates upon removal of the tag. Chp1 and Tas3 are both predicted to contain highly flexible regions (Supplementary Figure 1a). We therefore used limited proteolysis on the purified complex in order to identify stably folded domains. These experiments revealed that the Tas3 N-terminal domain (NTD) and the C-terminal half of Chp1 (residues 504-960) form the stable core of the complex (Figure 1a and Supplementary Figure 1b). The boundaries determined for Chp1 agree well with the region identified to interact with Tas316. The C-terminal boundary of the Tas3 fragment could not be determined accurately due to heterogeneity in the proteolytic fragments, therefore the fragment (residues 9-83) previously shown to be sufficient in maintaining Chp1-Tas3 interaction was used19. The Chp1(504-960)-Tas3(9-83) core was expressed with an OSS tag on the N terminus of either Tas3 or Chp1. A mass of 93 kDa was measured by multiangle light scattering (MALS) for the OSS-Chp1 tagged complex, corresponding closely to the 87.7 kDa expected for a 1:1 stoichiometry. The OSS-Tas3 tagged construct (Supplementary Figure 2) yielded well diffracting crystals that were used to solve the structure by the Single Anomalous Dispersion method (see Supplementary Methods). In the crystal structure we observe two molecules in the asymmetric unit with a critical crystal contact mediated by the OSS tag (Supplementary Figure 3).
The structure of Chp1 and Tas3 reveals a tight, elongated complex, resembling the hull of a ship (Figure 1b), with dimensions 116 Å × 44 Å × 36 Å. The complex can be divided into four subdomains: three distinctive mixed α/β domains in Chp1, and an α-helical bundle in Tas3, which is completely engulfed by the first two domains of Chp1 and the linker between them. The two molecules in the asymmetric unit are very similar (overall RMSD of 1.25 Å), except for a difference along their long axis that is manifested in a bend of 7 degrees distributed across the linker domain residues 666-672 of Chp1.Structural similarity searches for individual domains identified related structures for each (Supplementary Figure 4). However, the overall arrangement and protein-protein interface observed here represents a novel architecture. Domain I of Chp1 (aa 515-651) is a β-barrel domain, and shows strong structural similarity to the SPOC domain of the transcriptional corepressor SHARP 25. The Chp1 SPOC domain binds the first α-helix (α1) of Tas3 (residues 9-17),embedding it in a deep hydrophobic groove formed by the Chp1 SPOC domain and the first α-helix of the Chp1 linker domain (aa 654-661) (Supplementary Figure 5). The Tas3 residues 18-23 following the α1 helix bind to the Chp1 SPOC domain in an extended conformation, forming an interface reminiscent of the one used by the SPOC domain of SHARP for interaction with SMRT/NCoR.
The Chp1 SPOC domain is connected to the second Chp1 domain (aa 684-808) by a 33 amino-acid linker that forms two Tas3 flanking helices and bridges across the Tas3 N-terminal domain (NTD). The α2 and α3 helices of Tas3 fold as a helical hairpin that is “crosslinked” at its base by α1 and surrounded by four shorter α-helices of Chp1, comprised by the α6 and α7-helices of the linker domain, the α3-helix of the SPOC domain and α8 of Domain II (Supplementary Figure 6). It is interesting to note that GW182 proteins, which share homology with Tas3 in their GW motifs, are predicted to have a coiled-coil domain at the N terminus, although it has not been linked to function. It is also notable that the silencing protein Sir4 in S. cerevisiae forms a coiled coil that interacts with the C-terminal half of Sir326. Domain II of Chp1 is a member of the Rossmann fold family, resembling eukaryotic phosphatases and receiver domains of bacterial phosphotransferase signaling systems. In the complex observed here, one side of this domain, which includes the putative active site region interacts extensively with Tas3 α2 and α3.
The key function of the Tas3-NTD in complex formation observed here is in excellent agreement with previous work showing that deletion of Tas3 residues 10-24 including helix α1, abolishes Chp1-Tas3 complex formation, leading to extensive loss of heterochromatic silencing19. In addition, two-hybrid mapping revealed that residues encompassing the SPOC domain of Chp1 (aa 505-661) suffice for interaction with Tas3 (Supplementary Figure 7). The interaction between Domain II and Tas3 contributes 42% of the buried surface area of the complex. However, we could not verify the interaction between Domain II and Tas3 in yeast two-hybrid assays due to instability of the Domain II fusion protein (Supplementary Figure 7).
The tight embrace of Tas3 by the SPOC domain, the linker, and Domain II make up the interaction surface between Chp1 and Tas3, resulting in a substantial buried surface area of 4890 Å2. PISA analysis yields a ΔGsolv of -39 kcalmol-1, indicative of a strong hydrophobic interaction27, comprising of 59 residues of Tas3 and 65 residues of Chp1 with 17 hydrogen bonds and 4 salt bridges (Supplementary Figure 5,6). Buried within the relatively flat interface between Domain II and Tas3 α3 we observe a cavity containing two strong spherical peaks of electron density coordinated by Arg69 of Tas3.These were interpreted as chloride ions based on the chemical environment, the fit to the electron density, and buffer ingredients. There is a possibility that the chloride ions substitute for parts of a functionally important small molecule ligand that might occupy this cavity under different conditions. This hypothesis was tested by introducing Tas3 mutations R69A and C68A into S. pombe. However no discernible effect on transcript levels at centromeres could be observed in these mutants (Supplementary Figure 8).
The extensive interface formed between Tas3 and the SPOC domain, linker and Domain II of Chp1 suggests that the complex is an obligate heterodimer. This is supported by the observation that Tas3's stability in the absence of Chp1 is highly dependent on a C-terminal TAP tag16,19. Since no protein stability issues have been observed for Tas3 in the absence of Ago1, neither in vivo nor in vitro, the tight interface between Chp1 and Tas3 lends support to the model proposing that Chp1 and Tas3 form a stable and possibly polymeric20 platform, with Ago1 being a more loosely associated component of the complex16.
Domain III of Chp1 (aa 809-960) is independent of the Chp1-Tas3 interface, and shares a strong structural similarity with PIN domains (Figure 2a and Supplementary Figure 4). PIN domains are abundant throughout all organisms. They function as nucleases or as protein-protein interaction modules28-30. Specifically, the Chp1 PIN domain is most similar to the PIN domains of SMG5/EST1B and SMG6/EST1A31, proteins involved in the nonsense mediated mRNA decay (NMD) pathway32-34 and, interestingly, in telomere maintenance in metazoan organisms35,36. The PIN domain of human SMG5 has also been implicated in recruitment of the phosphatase PP2A to UPF137,38. Furthermore, it shares similarity with the PIN domain of the exosome subunit Dis3/Rrp44, which apart from its endonuclease function is also critical for association with the exosome core30,39.
Chp1 is important for both the establishment and maintenance of heterochromatin in fission yeast16,22,23,40,41.In wild type fission yeast, the assembly of centromeric repeats and of subtelomeric genes into silenced domains occurs via both transcriptional and post-transcriptional gene silencing5,6,42. Cells lacking RITS components accumulate transcripts from pericentromeric repeats3,11,23. At telomeres, silencing of the subtelomeric telomere-linked helicase (tlh)genes is derepressed following deletion of Chp1 or Tas3, but not of other RNAi components, suggesting that Chp1 and Tas3 have additional roles in silencing tlh transcripts outside the context of RITS19,40. In order to investigate the function of the PIN domain of Chp1, DNA sequences encoding residues 809-960 encompassing the PIN domain were deleted from the genomic chp1 allele in fission yeast to generate chp1ΔC. Real time PCR analyses of transcripts derived from the dg or dh outer repeats of the centromere (Figure 3a) demonstrated that whereas cells lacking Chp1 displayed strong accumulation of transcripts, centromeric heterochromatin was unaffected by the loss of the PIN domain in chp1ΔC strains (Figure 3b,c,d Supplementary Figure 9). In marked contrast, subtelomeric tlh transcripts (Figure 3a) accumulated in chp1ΔC strains (Figure 3b,e Supplementary Figure 9). These data suggest that the PIN domain of Chp1 contributes to the RNAi independent function observed for Chp1 in the regulation of subtelomeric chromatin. Next we asked whether the PIN domain was involved in chromatin-mediated transcriptional gene silencing or post-transcriptional regulation of subtelomeric sequences. If the PIN domain affects transcriptional regulation, RNA polymerase II (Pol II) occupancy might be expected to increase on subtelomeric sequences chp1ΔC strains. ChIP with anti-Pol II antibodies revealed that Pol II occupancy on tlh sequences was relatively unaffected by loss of chp1+. In contrast, at centromeres, Pol II occupancy increased in strains lacking chp1+, but not in the chp1ΔC background (Figure 4). These data suggest that Chp1's PIN domain is required for post-transcriptional processing of subtelomeric transcripts.
As mentioned, PIN domains are often associated with nuclease activity. We wondered whether the defect in tlh transcript regulation could be attributed to a possible nucleolytic activity of the PIN domain. Structure based alignment (Figure 2c) indicates that the Chp1 PIN domain has only a subset of the conserved aspartate residues critical for nuclease activity attributed to some PIN domains. Nonetheless, Chp1 still possesses a number of amino acids in the active site that might be able to bind the Mg2+ ions required for nuclease function (Figure 2b). We therefore performed nuclease assays with the construct used for crystallization, but did not detect any discernible nuclease activity (data not shown), as is the case for SMG531.
To further test a possible nuclease activity in vivo, a presumed catalytically dead mutation (D904A) was engineered into the PIN domain of chp1+ in an otherwise wild-type strain. In contrast to chp1ΔC, this mutant showed no accumulation of subtelomeric transcripts (Figure 3e, Supplementary Figure 9), consistent with our observation that the PIN domain does not possess nucleolytic activity in vitro.
A prominent positively charged patch is nestled between domain II and the PIN domain of Chp1 (Figure 1c and Supplementary Figure 10). In order to gauge its functional significance, we mutated two PIN-domain arginines, Arg923 and Arg924, which are part of this basic patch and are conserved (see below), to alanine. Interestingly, mutation of both arginines caused a similar defect in tlh transcript regulation as deletion of Chp1's entire PIN domain (Figure 3b,f,g and Supplementary Figure 9).
A PSI-Blast search using the sequence corresponding to the Chp1 fragment observed in the crystal structure identified a protein family particular to fungi (Figure 5a). Based on the sequence alignment to Chp1, the members of this family share common features, such as a chromodomain at the N terminus and a C-terminal PIN domain, separated by a distance of around 700-1000 amino acids. Within the PIN domain we observe strong conservation of the functionally important “basic patch” arginines Arg923 and Arg924. Interestingly, the Tas3 interacting residues of Chp1 are not well conserved (Figure 5b), and we were unable to identify Tas3 homologs outside the Schizosaccharomyces clade. It remains to be tested whether the RITS complex is conserved in fungi.
The RITS complex plays a central role in the formation of heterochromatin.It acts as a switch that integrates the sequence specificity of the RNAi effector protein Argonaute with sensing the H3K9 methyl marks on chromatin through its chromodomain protein Chp113. Once recruited to a specific location, it serves as a platform for binding of ClrC and RDRP, which lead to robust heterochromatin formation and gene silencing. The structure of Chp1 bound to Tas3 presented here shows that these two RITS subunits interact tightly to form a solid bridge between chromatin substrate and the RNAi machinery. This structure together with the structures of the Chp1 chromodomain13 - the attachment point to chromatin, and the α-helical motif on the C terminus of Tas320 - a multimerization module, most likely represent most, if not all, of the structured domains of the Chp1-Tas3 complex. Long, unstructured regions connect these domains, allowing the Chp1-Tas3 backbone to span great distances between attachment sites, thereby possibly crosslinking neighboring strands of chromatin fibers. Combining all the available information suggests cooperation between Ago1 and the Chp1-Tas3 complex in spreading on chromatin, thereby setting up a robust, polymeric platform that supports the enforcement of heterochromatin (Figure 6).
In addition to the role of the RITS complex in both RNAi and Clr4-mediated repression of transcription, the identification of a PIN domain in Chp1 provides important clues for the unique function of the Chp1-Tas3 complex in RNAi-independent silencing of telomeric transcripts. Although the PIN domain of Chp1 itself does not appear to possess endonuclease activity, the elevation of transcripts without change in Pol II occupancy in Chp1 mutants is clear evidence for an important role of Chp1 and Tas3 in post-transcriptional gene silencing (PTGS) at the subtelomere, much of it mediated by the PIN domain. The PIN domain may be involved in recruiting nuclease activities to subtelomeric regions, much like the PIN domain of the exosome subunit Rrp44/Dis3 was demonstrated to tether Rrp44 to the rest of the core exosome30,39. In addition, it has been previously reported that subtelomeric tlh transcripts are regulated by the S. pombe TRAMP complex and the exosome6. It will be interesting to determine whether the PIN domain of Chp1 is similarly involved in recruitment of the exosome to subtelomeric regions for processing of tlh transcripts. Since mutation of the basic region at the neck of the PIN domain yields the same loss of subtelomeric transcript regulation, it is possible that the basic region is mediating protein-protein interactions. Or it might be involved in nucleic acid binding such as binding of the tlh transcripts themselves, thus presenting substrates to an RNA degradation pathway, the exosome, for example, that is recruited through the PIN domain. Given that Chp1 localizes to all sites of heterochromatin, and that the exosome is also involved in regulation of centromeric transcripts6, the question still remains as to why Chp1's PIN domain plays no role in the regulation of centromeric transcripts. One possibility is that Ago1 and the RNAi pathway dominate PTGS at the centromere, thereby masking the activity of the PIN domain. The picture thus emerging suggests that the various modules of the RITS complex play context-specific roles (Figure 6). It would be of great interest to elucidate how the various functions of RITS are specifically regulated.
It is curious that fungal EST1 proteins, in contrast to their metazoan counterparts, do not feature a PIN domain. Since we present evidence for a subtelomeric silencing role of Chp1's PIN domain, there exists a possibility that Chp1 provides a functionality that has been incorporated into the telomeric EST1 proteins in higher organisms. It is intriguing that proteins bearing both a chromodomain and a PIN domain are conserved only within fungi, and that tlh genes (as their name implies) are restricted to subtelomeric domains in fungi43. Future experiments will address whether the physical link between RNAi and heterochromatin observed in the Chp1-Tas3 structure, as well as the RNAi-independent role of Chp1's PIN domain in processing tlh transcripts is conserved amongst organisms of the fungal kingdom.
We thank Chandni Ashar and Aaron Kosinski for technical support and Olivia George and Brandon Lowe for assistance, A. Héroux for help at the National Synchrotron Light Source, which is supported by Department of Energy, Office of Basic Energy Sciences. T.S. was supported by a Human Frontiers in Research Program fellowship. This work was supported by NIH grant R01GM084045, NCI Cancer Center support grant P30CA021765, the American Lebanese Syrian Associated Charities (to J.F.P.) and the Louis Morin Charitable Trust (to L.J). L.J. is an investigator of the Howard Hughes Medical Institute. We thank G.Thon, T.Nakamura, R. Martienssen and M. Zaratiegui for helpful discussions, and G. Thon, and R. Allshire for strains, and K. Gull for anti-TAT1 antibody.
Accession codes: The atomic coordinates and structure factors for the Chp1-Tas3 complex have been deposited in the Protein Data Bank with accession number 3TIX.
Author contributions: T.S. designed, conducted and interpreted the in vitro, structural studies and bioinformatics analyses and wrote the paper. G.J. generated yeast strains and performed transcript analyses and westerns. S.S. generated yeast strains, performed transcript analyses, westerns and ChIP experiments. J.F.P generated yeast strains, designed and interpreted experiments and wrote the paper. L.J. designed and interpreted experiments and wrote the paper.