The Methyltransferase Activity of Clr4Suv39h Triggers RNAi Independently of Histone H3K9 Methylation  Erica L. Gerace, Mario Halic, Danesh Moazed  Molecular Cell  Volume 39, Issue 3, Pages 360-372 (August 2010) DOI: 10.1016/j.molcel.2010.07.017 Copyright © 2010 Elsevier Inc. Terms and Conditions

Molecular Cell 2010 39, 360-372DOI: (10.1016/j.molcel.2010.07.017) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 1 Rik1 Interacts with RITS and RDRC in a Dcr1-Dependent Manner (A) Diagram showing the RITS, RDRC, and CLRC complex and the interactions explored by co-IP experiments. (B) Western blots showing that Rik1-myc coprecipitates with Tas3-TAP and Rdp1-TAP. (C and D) The interaction of Rik1-myc with Tas3-TAP and Rik1-TAP with Flag-Ago1 is greatly diminished in dcr1Δ cells. (E) Western blot of coimmunoprecipitation experiments showing that RNase A does not affect the interaction of Rik1-myc with Tas3-TAP. Agarose gel (bottom) stained for RNA extracted from lysates after immunoprecipitation showing effective RNA degradation by RNase A. (F) The addition of ethidium bromide does not affect the coprecipitation of Rik1-myc with Tas3-TAP. Molecular Cell 2010 39, 360-372DOI: (10.1016/j.molcel.2010.07.017) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 2 The Interaction of the Rik1 Subunit of CLRC with the Tas3 Subunit of RITS Depends on Clr4 and Clr4 Activity (A) Western blots of immunoprecipitations showing that the interaction of Tas3-TAP with Rik1-myc is greatly reduced in clr4Δ cells and to a greater extent than in dcr1Δ cells. (B) Wild-type Flag-Clr4, but not catalytically inactive Flag-Clr4 H410D/C412A SET domain mutant, rescues the loss of Rik1-myc and Tas3-TAP interaction in clr4Δ cells, indicating that the methyltransferase activity of Clr4 mediates complex formation. Wild-type and mutant Clr4 proteins were expressed at similar levels (bottom). Note that the blot in (B) shows a shorter exposure than the blot in panel (A). Molecular Cell 2010 39, 360-372DOI: (10.1016/j.molcel.2010.07.017) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 3 Centromeric dg siRNAs Are Present at Near Wild-Type Levels in H3K9 Mutant Cells (A) Detection of Argonaute-associated centromeric small RNAs by splinted ligation in wild-type, clr4Δ, H3K9A, H3K9R, and clr4Δ cells carrying a reintegrated Flag-clr4 allele. dg siRNAs are present in both H3K9R and H3K9A cells but accumulate to near wild-type levels in H3K9R cells, indicating that siRNA amplification can occur independently of H3K9 methylation. In contrast, dh siRNA levels in H3K9 mutant cells are dramatically reduced, similar to the levels observed in clr4Δ cells, indicating that H3K9 methylation is required for dh siRNA generation. The relative amount of input RNA in each reaction is indicated above each panel, and relative siRNA levels are indicated below each panel. (B) dg and dh siRNA levels in cells expressing the clr4 H410D/C412A SET domain mutant are as low as clr4Δ cells. Molecular Cell 2010 39, 360-372DOI: (10.1016/j.molcel.2010.07.017) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 4 Tethering Rik1 to ura4 Transcripts Initiates RNAi-Dependent Silencing (A) Schematic of the λN tethering system with five BoxB repeats inserted in the 3′UTR of the endogenous ura4+ locus. Upon transcription, the BoxB sequences form hairpin structures that are bound by the λN peptide of the fusion protein. (B) Tethering Rik1-λN to ura4+-5BoxB transcripts initiates silencing of the ura4+ locus as indicated by growth on 5-FOA medium to a similar extent as that previously reported for Tas3-λN. N/S, nonselective medium. (C) Northern blot analysis of total RNA extracted from the indicated strains shows relative expression of ura4 mRNA. A wild-type strain and one carrying ura4+ inserted at the centromeric imr serve as controls. The blot was reprobed for act1 mRNA as a loading control. (D) Mutating RNAi components leads to a loss of silencing, as observed by loss of growth on 5-FOA medium. (E) dcr1Δ and ago1Δ strains were transformed with an empty plasmid or one expressing the corresponding wild-type or catalytically inactive enzyme. dcr1 D937A is a mutation of the RNase III catalytic domain, and ago1 D580A is a mutation of the DDH catalytic motif. As indicated by growth on 5-FOA, only expression of the wild-type enzyme rescues the loss of silencing phenotypes. Molecular Cell 2010 39, 360-372DOI: (10.1016/j.molcel.2010.07.017) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 5 Silencing Initiated by Rik1 Tethering is Heterochromatin Independent (A) Silencing assays showing that mutations in CLRC subunits or heterochromatin factors swi6+ and sir2+ cause only slight defects in Rik1-λN-induced silencing of ura4+-5BoxB. Growth on 5-FOA medium indicating ura4+-5BoxB silencing. (B) Chromatin immunoprecipitation experiments showing that Rik1-λN silencing does not promote H3K9 methylation at the ura4+-5BoxB locus. fbp1+ or act1+ serve as euchromatic controls. For comparison, tethering Tas3-λN leads to H3K9 methylation at the ura4+-5BoxB locus and serves as a positive control. (C) Quantification of H3K9 dimethylation (H3K9 diMe) for the ChIP experiment in (B). Tas3-λN, but not Rik1-λN, tethering leads to a 4- to 5-fold enrichment of H3K9 diMe over control loci. Schematic (bottom) shows locations probed by primer sets indicated by boxes. Error bars represent SD for three independent experiments. (D) The genomic reintegration of a chp1 allele lacking its chromodomain restores silencing in rik1-λN chp1Δ cells as indicated by growth on 5-FOA medium. The same genomic reintegration into tas3-λN chp1Δ cannot restore silencing. (E) The genomic reintegration of wild-type clr4+ into either rik1-λN clr4Δ or tas3-λN clr4Δ restored silencing to wild-type levels, as indicated by growth on 5-FOA medium. However, whereas the catalytically inactive clr4 H410D/C412A mutant does not rescue Tas3-λN silencing, it does rescue Rik1-λN silencing to the levels observed in clr4Δ cells shown in (A). Molecular Cell 2010 39, 360-372DOI: (10.1016/j.molcel.2010.07.017) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 6 The CPSF_A Domain of Rik1 Is Required for H3K9 Methylation and Silencing at a Step Downstream of siRNA Amplification (A) Schematic diagram showing the location of the Rik1 CPSF_A domain. Rik1ΔC lacks the entire C terminus, amino acids 706 to 1040, of Rik1. Both wild-type and truncated Rik1 contain a C-terminal TAP tag. (B) Western blot of whole-cell lysate from untagged Rik1-TAP and Rik1ΔC-TAP strains showing that Rik1 and Rik1ΔC protein are expressed at similar levels. The blot was stripped and reprobed for β-actin as a loading control. (C) Western blots showing that Flag-Ago1 immunoprecipitates with Rik1ΔC -TAP as efficiently as with full-length Rik1-TAP. (D) Chromatin immunoprecipitation experiments showing that the Rik1 C terminus is required for H3K9 methylation at the centromeric dg repeats, as there is a loss of H3K9me in the rik1ΔC-TAP cells. Fold enrichments were calculated after normalization to fbp1+ and relative to the rik1Δ strain, which was given the value of 1.0. (E) Quantitative RT-PCR for centromeric dg or dh transcripts shows that these transcripts are similarly derepressed in rik1ΔC-TAP, clr4Δ, and rik1Δ cells. Fold increase is calculated relative to the tdh1 control. The error bars represent standard deviations for three independent experiments. (F and G) Detection of Argonaute-associated small RNAs by splinted ligation. dg siRNA levels are reduced in rik1ΔC-TAP compared to wild-type, but not to the same extent as in rik1Δ cells, which contain higher levels of siRNAs than clr4Δ cells. In contrast, in rik1Δ and rik1ΔC-TAP cells, dh siRNA levels are reduced to the low levels observed in dcr1Δ and clr4Δ cells. The relative amount of input RNA in each reaction is indicated above each panel, and relative siRNA levels are indicated below each panel. Bottom panel in (F) shows Flag-Ago1 stained by colloidal Coomassie after purification and prior to small RNA extraction. The amount of RNA added to each splinted ligation reaction is normalized based on relative Flag-Ago1 protein level. (H) Tethering Rik1ΔC-λN to the ura4-5BoxB transcripts leads to silencing of ura4+ to the same extent as Rik1-λN tethering, as indicated by growth on 5-FOA medium. Molecular Cell 2010 39, 360-372DOI: (10.1016/j.molcel.2010.07.017) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 7 Two-Step Model for TriComplex Assembly and siRNA Amplification (A) The assembly of the RITS-RDRC-CLRC TriComplex on cen-dg transcripts. TriComplex assembly depends on Clr4 methyltransferase activity, likely through the methylation of a TriComplex subunit or a histone lysine other than H3K9, and siRNA generation. The methylation event may create a binding site for one of the TriComplex chromodomain proteins, Chp1 and/or Clr4. (B) After TriComplex-mediated siRNA amplification, siRNA-programmed RITS complexes target nascent centromeric dg transcripts and direct Clr4-mediated H3 methylation. Unlike the dg repeats, siRNA amplification at dh repeats (data not shown) requires H3K9 methylation. The C terminus of Rik1 is specifically required for RNAi-mediated transcriptional gene silencing and helps stabilize CLRC-RNAi interactions, possibly by binding to dsRNA produced by RDRC. (C) Tethering Rik1-λN to ura4+-5BoxB transcripts induces posttranscriptional gene silencing of ura4+-5BoxB. Rik1 can efficiently recruit RNAi, leading to transcript degradation without inducing H3K9 methylation at the ura4+-5BoxB locus, and bypasses the requirement for the other subunits of CLRC, Clr4 activity, and the Rik1 C-terminal CPSF_A domain. Molecular Cell 2010 39, 360-372DOI: (10.1016/j.molcel.2010.07.017) Copyright © 2010 Elsevier Inc. Terms and Conditions