Chp1-Tas3 Interaction Is Required To Recruit RITS to Fission Yeast Centromeres and for Maintenance of Centromeric Heterochromatin^{▿ †}

Media and chemicals.

Fission yeast were maintained on rich medium (YES) or on pombe minimal medium with glutamate (PMG) with appropriate supplements if nutritional selection was necessary (31). All chemicals were obtained from Sigma unless otherwise indicated.

Strain generation.

Strains used in this study are listed in Table 1. The chp1Δ his3⁺, ago1Δ ura4⁺, clr4Δ LEU2, clr4Δ kan^R, rik1Δ LEU2, tas3Δ ura4⁺, dcr1Δ kan^R, and tas3Δ kan^R null alleles and the chp1-myc₆⁺-, tas3-TAP⁺-, and Flag₃-ago1⁺-tagged alleles have been described previously (36-39). The tas3_Δ10-24-TAP allele was generated in a two-step process. Nucleotides 27 to 69 of the tas3-TAP open reading frame were replaced with the ura4⁺ marker gene by transforming Tas3-TAP cells by electroporation with a PCR product containing ura4⁺ flanked by ∼80 bases of homology to sequences flanking nucleotides 27 to 69 of tas3. ura4⁺ transformants were selected by growth on medium lacking uracil, and the proper integration of ura4⁺ into tas3-TAP was checked by PCR. The removal of ura4⁺ and the generation of tas3_Δ10-24-TAP were achieved by retransformation with a chimeric PCR product and selection for growth of cells on medium containing 5-fluoro-orotic acid (5-FOA). PCR and DNA sequencing were used to confirm the generation of this allele. The non-epitope-tagged tas3_Δ10-24 allele was generated similarly, using a wild-type strain (PY 42) for the targeting of tas3⁺. The rdp1-Flag₃ allele was generated by the transformation of PY 42 cells with a chimeric PCR product amplified from JP-931 using primers with homology to sequences flanking the stop codon of Rdp1 and that amplify the Flag₃ and Kan^r sequences from JP-931.

DNA constructs.

Yeast two-hybrid vectors were generated as described previously (39) by PCR amplification of domains of the Tas3 and Chp1 open reading frames and by subcloning them into pGBKT7 (Chp1) and pGADT7 (Tas3) vectors (Clontech). The vector for the reintegration of clr4⁺ (JP-1084) was previously described and allows the reintegration of a genomic clone of clr4⁺ (under the control of its endogenous regulatory sequences) into clr4 null cells (37). JP-931 (for C-terminal Flag-tagging Rdp1) was derived from pFA6a-GFP(S65T)-KanMX-6 (3) after the release of green fluorescent protein (GFP) sequences by digestion with PacI and AscI and their replacement with annealed oligonucleotides that bear PacI- and AscI-compatible ends and that encode a 3× Flag epitope sequence (M2) followed by a stop codon.

Protein interaction assays. (i) Yeast two-hybrid assay.

Budding yeast manipulations were performed as previously described (39). The interaction of proteins resulted in the activation of three integrated GAL4-dependent reporters and was detected by the growth of yeast on selective medium and the production of blue coloration due to the metabolism of X-α-galactose (X-α-Gal) present in the medium. Protein truncations were designed to minimize structural disruptions based on information from secondary-structure predictions. The expression of all constructs was confirmed by Western blot analysis of cell extracts using antibodies directed against Gal4 activation domain (GAD) fusions with Tas3 (anti-hemagglutinin; 12CA5) or Gal4 DNA binding domain (GBD) fusions with Chp1 (anti-myc; 9E10).

(ii) Fission yeast.

Cell extracts were prepared by grinding cells in liquid N₂ in a pestle and mortar as previously described but using 2.4 × 10⁹ cells per ml of extraction buffer for each immunoprecipitation (39). Western blots were probed with M2 anti-Flag antibody (Sigma), anti-myc antibody (9E10; Roche), or anti-tandem affinity purification tag (TAP) antibody (rabbit anti-immunoglobulin G [IgG] horseradish peroxidase; Sigma).

Real-time PCR. (i) Transcript analysis.

cDNA was generated by the random priming of total cellular RNA, which was prepared as previously described (37). The detection of transcripts by real-time PCR used primers targeting the dh region of the centromere (JPO-769 and JPO-770) that are predicted to amplify from six copies of the repeat within the genome of the h⁻ strains used for these analyses. Primers for alcohol dehydrogenase (adh1) were used as the euchromatic normalization control (JPO-793 and JPO-794). Centromeric transcripts from dg (PCR amplicon B) were measured using JPO-986 (GATACTGATAATATTGAGATCCACAGCAC) and JPO-987 (GCGATGCCAAACAACAATATTG). Telomeric transcripts were detected using primers targeting the telomeric sequence from within the helicase domain of tlh1⁺ and tlh2⁺ (JPO-816, CGTTTTTGATACCGGCGC; JPO-819, TTGCCGTAACGACATCATGG) and do not amplify from centromere homologous sequences present elsewhere in tlh1 and tlh2.

Real-time PCR was performed from duplicate cDNA samples, and cDNA was generated from two or three biological duplicates. The comparative threshold cycle, or ΔΔC_T, method was used for the analysis of transcript levels (37). This method relies on the use of primer sets that amplify with 100% efficiency, such that the titration of template material in PCRs yields perfect standard curves (with a slope of −3.3 ± 0.1 and an R² coefficient of determination of 0.99) when the C_T is plotted against the concentration of the template. The linear range of amplification for each set of primers was verified, and experiments were performed within this range. The use of such primer sets allows for the comparative analysis of the C_Ts for samples of interest relative to the C_Ts of wild-type controls. The averages of these experiments are shown, with error bars representing the standard errors of the means.

(ii) ChIP.

Chromatin immunoprecipitation (ChIP) experiments were performed and analyzed as described previously (37) using primers from the dh region of the centromere (JPO-769 and JPO-770), the dg region of the centromere (JPO-986 and JPO-987), adh1 (JPO-793 and JPO-794), and telomeres (JPO-816 and JPO-819). Experiments were repeated at least three times, and the averages from these experiments (with standard errors of the means) are shown. A second fixation step was used for anti-Rdp1-Flag ChIPs (see Fig. 4E), with the fixation of cells for 30 min in dimethyl adipimidate (5 mM in phosphate-buffered saline; Pierce) after the cells had been washed following formaldehyde fixation. Fixation was stopped by the addition of glycine.

siRNA analyses were performed as described previously (37) for dh siRNAs. dg siRNAs were detected by hybridization with a probe generated by amplification with JPO-566 (TGGTAATACGTACTAGCTCTCG) and JPO-567 (AACTAATTCATGGTGATTGAT).

Cell growth assays.

Cells were grown in YES at 25°C to 5 × 10⁶ cells/ml, washed in PMG with no supplements, and spotted onto PMG complete, PMG-uracil, or PMG supplemented with 2 mg 5-FOA/ml. Serial (1:5) dilutions of fission yeast were made, and a total of 1.2 × 10⁴ cells were contained in the first spot. Plates were grown for 5 days at 25°C. To assess the maintenance of silencing at the mat2/mat3 locus, serial (1:5) dilutions of cells were grown overnight in YES medium at 25°C to 5 × 10⁶ cells/ml and washed in PMG with no supplements, spotted onto PMG medium containing reduced adenine or no adenine, and incubated for 5 days at 25°C. A total of 1.2 × 10⁴ cells were contained in the first spot.

Immunofluorescence.

Immunofluorescence and cytological chromosome missegregation assays were performed as described previously (39). Anti-tubulin antibody (a kind gift of K. Gull) was used at a 1/30 dilution and was revealed with fluorescein isothiocyanate-conjugated anti-mouse antibody (Jackson Immunologicals) diluted to 1:100. 4′,6′-Diamidino-2-phenylindole (DAPI) was used to identify nuclei.

Defining the interaction domain between Tas3 and Chp1.

Our previous studies indicated that a Tas3-Chp1 subcomplex was sufficient for the maintenance of centromeric heterochromatin (37). To test the importance of Tas3's association with Chp1 for the recruitment of RITS to heterochromatin, we sought to generate a mutant of Tas3 that specifically abrogated its interaction with Chp1. To identify the minimal sequences of Tas3 required for interaction with Chp1, truncations of Tas3 were tested for their interactions with full-length Chp1 in a GAL4 reporter two-hybrid assay in budding yeast. These analyses identified residues 9 to 83 of Tas3 as necessary and sufficient for interaction with Chp1 and are summarized in Fig. 1A and Fig. S1 in the supplemental material.

To further define the Tas3-Chp1 interface, small internal deletions removing regions of predicted secondary structure were made within the Chp1 interaction domain of Tas3, and the effect of these mutations on Tas3's interaction with Chp1 was assessed by a two-hybrid assay (Fig. 1B). A mutant of Tas3 lacking the first predicted alpha helix (residues 10 to 24) failed to interact with Chp1. This loss of interaction was not caused by the instability of the mutant protein, as its steady-state expression level was similar to that of the wild type (Fig. 1C).

The Tas3_Δ10-24 mutant cannot associate with Chp1 in fission yeast.

Tas3 is fully functional when the genomic locus is tagged at the C terminus with 13 myc epitopes, but the Tas3-13xmyc protein is unstable in the absence of Chp1 (39). A C-terminal fusion of tas3⁺ to the TAP tag (Tas3-TAP) is similarly fully competent for centromere function but is stably expressed in the absence of Chp1 (Fig. 2A, Table 2). As antibodies against the endogenous Tas3 protein are not yet available, the genomic tas3-TAP allele was used to investigate the role of the Chp1-Tas3 interaction in the function of the RITS complex.

The deletion of residues 10 to 24 from the genomic tas3-TAP allele (tas3_Δ10-24-TAP) yielded a mutant protein that was expressed at levels similar to that of Tas3-TAP (Fig. 2A). To determine if the loss of residues 10 to 24 from Tas3-TAP blocked binding to Chp1, the wild-type and mutant tas3-TAP alleles were introduced into strains bearing the chp1-6xmyc allele. Immunoprecipitation experiments demonstrated that Tas3-TAP efficiently precipitated Chp1-6xmyc, whereas Tas3_Δ10-24-TAP did not coimmunoprecipitate Chp1-6xmyc (Fig. 2B). Thus, the removal of the first predicted alpha helix from Tas3 (residues 10 to 24) prevented Tas3's interaction with Chp1 in fission yeast.

Tas3_Δ10-24 associates with Ago1 independently of Chp1.

We recently demonstrated that Tas3 binds Ago1 through a conserved GW-rich domain localized between residues 200 and 310 of Tas3 (37). To address whether the removal of residues 10 to 24 of Tas3 prevented its association with Ago1, a fully functional 3xFlag-ago1 allele was introduced into strains expressing wild-type or mutant Tas3-TAP and Chp1-6xmyc, and the association between these Tas3-TAP proteins and Ago1 was tested by immunoprecipitation. Interestingly, the Tas3_Δ10-24-TAP protein efficiently immunoprecipitated 3xFlag-Ago1, suggesting that Tas3 still binds to Ago1 even when the binding to Chp1 is abolished (Fig. 2C). These experiments suggest that Ago1 and Chp1 do not physically interact but that their association is dependent on Tas3. This was further tested by a two-hybrid assay, but we failed to see an association between Chp1 and Ago1 (see Fig. S2 in the supplemental material). This may be a limitation of the two-hybrid technique, since we also were unable to detect an association between Tas3 and Ago1 in two-hybrid assays (see Fig. S2 in the supplemental material). Thus, Tas3 can serve as an adapter protein or scaffold to coordinately bind to both Ago1 and Chp1 in fission yeast to facilitate RITS function.

Loss of interaction between Chp1 and Tas3-Ago1 perturbs centromeric heterochromatin.

To assess whether the loss of the Chp1 association with the Tas3-Ago1 subcomplex affected the maintenance of centromeric heterochromatin, we first monitored the expression of a centromeric transgene (cen::ura4⁺) in comparative growth assays. The cen::ura4⁺ transgene is located within the dg region of the outer repeats of the centromere (Fig. 3A) at a site known to be efficiently silenced in wild-type cells by both transcriptional and posttranscriptional silencing mechanisms [strain FY 648 in reference 2; otr1R(dg-glu)SphI::ura4⁺] (7, 8, 19). Cells bearing the tas3-TAP allele silenced cen::ura4⁺ expression, allowing growth on media containing 5-FOA, a drug that is toxic to cells expressing ura4⁺. In contrast, cells bearing the tas3_Δ10-24-TAP allele, like those with a deletion of tas3, failed to grow on 5-FOA, indicating that cen::ura4⁺ is expressed and that centromeric heterochromatin is disrupted when the Tas3-Chp1 interaction is abrogated (Fig. 3B).

The disruption of the silencing of the centromeric transgene is an indirect readout of a failure in the maintenance of centromeric heterochromatin. This may be caused by an accumulation of centromeric transcripts following the failure of the Ago1-mediated destruction of the transcripts or of their Dicer-mediated processing into siRNAs. Alternatively, the disruption of the transcription of centromeric sequences can perturb the formation of centromeric heterochromatin without the accumulation of centromeric transcripts (14). To distinguish between these possibilities, quantitative real-time PCR was performed on cDNA derived from strains lacking centromeric transgenes to monitor the levels of endogenous centromeric transcripts generated from the dh region of the centromere (using primers that amplify site A) (Fig. 3A). Silencing in this region of the centromere is attributed to cotranscriptional gene-silencing processes, with Ago1 mediating the destruction of centromeric transcripts (8, 19). The level of centromeric transcripts in wild-type cells relative to that of adh1 expression was normalized to 1, and as expected we saw an accumulation of high levels of centromeric transcripts in cells lacking the Clr4 histone H3 K9 methyltransferase activity (∼50-fold increase), with an intermediate accumulation (∼15- to 30-fold) in cells lacking either Tas3 or Chp1 (Fig. 3C). Cells bearing tas3_Δ10-24-TAP accumulated transcripts to about 30-fold the levels found in tas3-TAP cells, which is consistent with the disruption of the Tas3-Chp1 interaction causing a defect in the processing of centromeric transcripts but not in their production. We also analyzed transcript levels at the dg region of the centromere (Fig. 3A), in which silencing is attributable to both transcriptional gene-silencing and cotranscriptional gene-silencing processes (8, 19). Transcripts from the dg region accumulated to 60- to 90-fold higher levels in clr4Δ, ago1Δ, chp1Δ, and tas3Δ cells than in wild-type cells and were 80-fold higher in tas3_Δ10-24-TAP cells than in the wild type (Fig. 3D).

Disruption of Chp1-Tas3 association causes a defect in siRNA production.

To determine whether the accumulation of centromeric transcripts was attributable to a defect in the Dicer-mediated processing of these transcripts, small RNAs were isolated from tas3_Δ10-24-TAP cells and assessed for the presence of centromeric siRNAs. siRNAs from both the dh and dg regions of the centromere were detected in isolates from Tas3-TAP cells but not from Tas3_Δ10-24-TAP cells (Fig. 3E and F), indicating that the loss of the Chp1-Tas3 interaction causes a defect in the processing of centromeric transcripts into siRNAs.

Disruption of Chp1's association with Tas3-Ago1 prevents both RITS and RDRC association with centromeres and causes a reduction in centromeric H3K9Me2.

The accumulation of centromeric transcripts and the loss of siRNAs indicate a failure in the recruitment of RITS and RDRC to centromeres. ChIP assays demonstrated that Ago1 efficiently interacted with the dh region of centromeres in tas3-TAP cells but did not do so in cells expressing tas3_Δ10-24-TAP, suggesting that the Chp1-Tas3 interaction is required to recruit the Tas3-Ago1 subcomplex to the dh region of centromeres (Fig. 4A). Similarly, ChIP with IgG Sepharose beads to purify TAP-tagged proteins showed that Tas3 efficiently bound to dh centromeric sequences in Tas3-TAP but not Tas3_Δ10-24-TAP cells (Fig. 4B). Centromeric sequences also were efficiently immunoprecipitated by antibodies directed against Chp1 from cells expressing tas3-TAP but not from cells expressing tas3_Δ10-24-TAP or strains deleted for tas3 or clr4 (Fig. 4C), demonstrating that the loss of Tas3-Chp1 interaction results in the loss of Chp1 association with the dh region of centromeres. One possible cause for the loss of Chp1 association is reduced levels of H3K9Me2 (38). Consistently with this possibility, ChIP clearly demonstrated that tas3_Δ10-24-TAP mutants showed reduced H3K9Me2 levels at dh centromeric sequences compared to those of tas3-TAP cells (Fig. 4D). In contrast, only a minor reduction in centromeric H3K9Me2 levels was seen at centromeric dg repeat sequences in ago1Δ, tas3Δ, chp1Δ, and tas3_Δ10-24-TAP mutants (Fig. 4E).

We tested whether the loss of RITS from centromeres disrupted the recruitment of RDRC. A fully functional C-terminal Flag₃-tagged Rdp1 allele (see Table S1 in the supplemental material) was introduced into cells expressing tas3-TAP or tas3_Δ10-24-TAP. Clr4 is required for the association of RDRC and RITS with chromatin (32, 48); thus, we also introduced tas3-TAP and rdp1-Flag alleles into a clr4 null background to serve as a negative control. ChIP performed with anti-Flag antibodies demonstrated that Rdp1-Flag associated with centromeric dh sequences in the tas3-TAP clr4⁺ background but did not do so in clr4 null cells or in the tas3_Δ10-24 TAP clr4⁺ background (Fig. 4F). Thus, RDRC is not recruited to centromeres in cells expressing tas3_Δ10-24-TAP.

Loss of association between Chp1 and Tas3-Ago1 causes genomic instability.

The disruption of centromeric heterochromatin usually results in defective cohesion between sister chromatids and elevated rates of chromosome missegregation. We therefore monitored the efficiency of chromosome segregation in anaphase cells of the tas3_Δ10-24-TAP mutant by immunofluorescence. Whereas cells expressing tas3-TAP rarely gave rise to chromosome missegregation events (<1%), cells bearing tas3_Δ10-24-TAP exhibited high levels of chromosome missegregation (26%), similarly to what was seen for tas3Δ or chp1Δ null cells (Table 2).

Disruption of Tas3-Chp1 association does not impact heterochromatin maintenance but perturbs its establishment at the mating-type locus.

Components of RITS also localize to other sites of heterochromatin in fission yeast, such as the telomeres and mating-type locus (10, 39, 42). Interestingly, Tas3 and Chp1 can associate with these loci independently of the RNAi pathway (39). The loss of RNAi components or Chp1 has little effect on the maintenance of silencing of transgenes inserted at the mating-type locus (17, 39, 42, 47). To assess if the loss of Chp1-Tas3 interaction disrupted mat2/mat3 silencing, the tas3_Δ10-24-TAP mutation was introduced into cells bearing an ade6⁺ transgene inserted at mat3 (46) (Fig. 5A). This transgene is efficiently silenced in wild-type cells, but this repression is alleviated upon the deletion of clr4. Similarly to our results with chp1 null and ago1 null cells, no alleviation of silencing of the transgene was observed in tas3 null cells or in cells bearing tas3_Δ10-24-TAP. Thus, silencing at the mating-type locus appears to be independent of RITS components and of the RNAi pathway, most likely because redundant RNAi-independent pathways can maintain heterochromatin at mat2/mat3 (17, 20, 24, 46).

Chp1 is important for the establishment of heterochromatin at the mating-type locus (42). To test whether Chp1's role in the establishment of heterochromatin at mat2/mat3 relies on its association with Tas3, we asked if tas3_Δ10-24-TAP expression blocked the establishment of heterochromatin at mat2/mat3 after the removal and reintroduction of clr4⁺ (Fig. 5B). Cells lacking clr4 express mat3M::ade6 and grow on medium lacking adenine. On the reintegration of clr4⁺ (under its native regulatory sequences) into clr4 null tas3⁺-TAP cells, silencing was efficiently reestablished, with the repression of ade6⁺ expression, the red coloration of colonies, and the inhibition of growth on medium lacking adenine. In contrast, silencing was not reestablished in tas3_Δ10-24-TAP cells, and cells grew on medium lacking adenine in a manner similar to that of the clr4 null cells. Thus, the association between Chp1 and Tas3 is required for the efficient establishment of heterochromatin at the mating-type locus.

Loss of Tas3 or Chp1, but not disruption of the RNAi pathway, promotes accumulation of telomeric transcripts.

Heterochromatin also assembles in the subtelomeric regions of fission yeast chromosomes, where it extends for ∼50 kb (22). Silencing in these regions depends on Clr4, Swi6, and Taz1, and the recruitment of Clr4 to these sequences is mediated at least in part by Taz1 binding to telomeric repeats at the end of the chromosome and via RITS associating with centromere homologous sequences located within the open reading frame of subtelomeric helicase genes (12, 22, 29, 34). Similarly to the mating-type locus, the maintenance of silencing at telomeres depends on redundant RNAi-dependent and RNAi-independent mechanisms; thus, the removal of genes encoding RNAi components has little impact on the maintenance of the silencing of telomeric transcripts or of reporters inserted within subtelomeric loci (7, 17, 18, 22).

We tested whether telomeric transcripts accumulated in the tas3_Δ10-24-TAP mutant by monitoring transcripts of subtelomeric tlh genes using primers that recognize the helicase domain. tlh transcripts accumulated in chp1 null strains as seen previously (18) and in cells deficient for tas3 (Fig. 6A) but did not accumulate in either the tas3_Δ10-24-TAP mutant or in ago1 null cells. Taken together, these results suggest that RITS plays little role in the maintenance of the silencing of subtelomeric helicase genes but that Tas3 and Chp1 contribute, independently of each other, to the suppression of transcript accumulation from subtelomeric genes. If Tas3 and Chp1 affect distinct pathways, cells that are mutant for both tas3 and chp1 may further accumulate subtelomeric transcripts. Surprisingly, however, tas3Δ chp1Δ double mutants showed no further accumulation of tlh transcripts (Fig. 6B), suggesting that Tas3 and Chp1 act through a common or redundant pathway. We also analyzed telomere length regulation in these strains (see Fig. S3 in the supplemental material), but we found that of the strains tested (including tas3 and chp1 nulls and the double null), only the clr4 deletion had an impact on telomere repeat length, with a slight lengthening of the average size of the telomeric repeats compared to that of wild-type cells.

We performed ChIP to assess Chp1's association with telomeres in different mutant backgrounds (Fig. 6C). The telomeric association of Chp1 was somewhat reduced in cells bearing TAP-tagged alleles of Tas3, suggesting that the epitope tag on Tas3 interferes (directly or indirectly) with Chp1 recruitment, although no such defect is seen at centromeres (Fig. 4C). However, given this caveat, there was no further reduction of Chp1 association with telomeres in tas3_Δ10-24-TAP cells compared to that of tas3-TAP cells. In addition, Chp1's association with telomeres was maintained in tas3 null cells and, as noted previously (39), was increased in ago1 null cells compared to that of tas3-TAP cells. Not surprisingly, since Chp1 binds H3K9Me2, ChIP revealed that telomeric H3K9Me2 is maintained in tas3_Δ10-24-TAP mutants at levels similar to those of tas3-TAP cells (Fig. 6D). Thus, the maintenance of telomeric heterochromatin in tas3_Δ10-24-TAP cells may be due to H3K9Me2 recruited via RNAi-independent mechanisms. This contrasts with the situation for centromeric dh sequences, for which the H3K9Me2 levels and Chp1 association are reduced in cells expressing tas3_Δ10-24 TAP.

We also analyzed the association of Tas3 with telomeres (Fig. 6E). Whereas Tas3-TAP associated with tlh sequences, Tas3_Δ10-24-TAP did not. Consistently with this, the Tas3-TAP association with tlh was wholly dependent on Chp1. These results provide an explanation for our failure to detect further accumulation of tlh transcripts on combining tas3 and chp1 null mutations (Fig. 6B), since Tas3 is not associated with telomeres in chp1 null cells. However, it is difficult to explain why telomeric transcripts do not accumulate in tas3_Δ10-24-TAP cells, since Tas3_Δ10-24-TAP does not associate with telomeres, especially given that tlh transcripts accumulate in cells lacking tas3⁺. This difference in phenotype could result from our comparison of strains lacking Tas3 altogether (tas3 null) to strains in which the Tas3_Δ10-24-TAP protein is expressed but not localized at telomeres.

We next tested whether the establishment of the silencing of tlh genes was perturbed by the tas3_Δ10-24-TAP mutation. Chp1 is required for the efficient establishment of telomeric heterochromatin (42), but the use of the tas3_Δ10-24-TAP mutant allows us to discern if this requirement comes from Chp1's role in RITS or if Chp1 plays an RITS-independent role in the establishment of telomeric heterochromatin, which would be similar to its RITS-independent role in the silencing of tlh transcripts (Fig. 6A). However, we found that the TAP tag on tas3⁺ prevented the efficient establishment of telomeric heterochromatin in clr4 null cells following the reintroduction of clr4⁺. We regenerated the tas3_Δ10-24 allele in a non-epitope-tagged tas3⁺ locus and asked whether tlh transcripts were suppressed following the removal and reintroduction of clr4⁺. The loss of association between Tas3 and Chp1 prevented the efficient silencing of tlh transcripts following the reintroduction of clr4⁺ (Fig. 6F).

Following our demonstration that the TAP tag on Tas3 precluded the establishment of heterochromatin at telomeres, we questioned whether the TAP tag influences the stability of the Tas3 protein. In particular, we were interested in determining whether Tas3 normally is unstable in cells lacking Chp1, as we have reported previously for the myc-tagged Tas3 protein (39). If so, we expect that the Tas3_Δ10-24 mutation will result in an unstable protein, since the association between Tas3 and Chp1 is lost in this mutant background. Unfortunately, there are no antibodies available against Tas3, so we could not directly assess this question. Instead, we asked whether telomeric transcript regulation differs in strains expressing the untagged Tas3 _Δ10-24 protein compared to that of strains expressing the TAP-tagged protein. We found that in strains expressing the untagged Tas3_Δ10-24 protein, telomeric transcripts accumulated to levels similar to that of tas3 null cells (Fig. 6G). This result suggests, therefore, that the presence of the TAP tag on the Tas3_Δ10-24 protein alters the behavior of this mutant protein at telomeres.

Given the different results that we obtained for telomeric transcript regulation in comparisons of TAP-tagged to untagged Tas3_Δ10-24 strains, we reanalyzed the centromeric dh transcript accumulation in untagged tas3_Δ10-24 cells and found that centromeric transcripts accumulated to levels similar to those of tas3 null cells and tas3_Δ10-24-TAP cells (Fig. 6H).

Chp1-Tas3 interaction is important for centromeric heterochromatin integrity.

The data that we present clearly demonstrate that Tas3 is a critical component of RITS and that the association between Tas3 and Chp1 is essential for the maintenance of centromeric heterochromatin. Tas3 binds to Chp1 through its N-terminal domain, a region that is distinct from the GW-rich region of Tas3 that is important for association with Ago1 (37). Indeed, a mutant Tas3 (lacking residues 10 to 24) that is unable to associate with Chp1 retains the association with Ago1. This Tas3_Δ10-24-TAP mutant cannot support the recruitment of RITS or RDRC to centromeres. Since Dcr1 physically associates with RDRC (11), Dcr1 recruitment to centromeres likely also is lost in the Tas3_Δ10-24-TAP cells. These mutant cells accumulate centromeric transcripts, fail to generate centromeric siRNAs, and cannot recruit the high levels of H3K9-methyltransferase activity that are required at centromeres to maintain centromeric heterochromatin.

Our demonstration that the separation of Chp1 from the Tas3-Ago1 subcomplex disrupts centromeric heterochromatin contrasts strongly with the lack of a centromeric phenotype when Ago1 is separated from the Chp1-Tas3 subcomplex (37). In these cells, the levels of H3K9Me2 at centromeres are maintained, allowing the recruitment of Chp1-Tas3, and siRNAs are produced, promoting the recruitment of Ago1 independently of Tas3-Chp1. Thus, our data strongly support the idea that the recruitment of RITS to centromeres relies on Chp1-mediated targeting to chromatin. The recruitment of Chp1 to chromatin depends on the integrity of the chromodomain (35, 39). The chromodomain of Chp1 can bind H3K9Me3 peptides in vitro (38, 42), and the pattern of localization of Chp1 overlaps with that of H3K9Me2 and is dependent on the activity of the Clr4 H3K9Me2 transferase (10, 36, 39, 42). Cells deficient in the RNAi pathway show a reduction, but not a complete loss, of H3K9Me2 at centromeres (42, 49), demonstrating that, as was the case for telomeres and the mating-type locus, redundant RNAi-dependent and RNAi-independent mechanisms promote the recruitment of Clr4 to centromeres (Fig. 7). For heterochromatin establishment, we suggest that RNAi-independent H3K9Me2 initially recruits Chp1 and RITS to centromeres (37), which in turn recruit RDRC and Dcr1 activities (11, 32). RDRC then generates dsRNA from centromeric transcripts, which is cleaved by Dcr1 to yield siRNAs that ultimately are incorporated into RITS and may function together with Ago1 to suppress the accumulation of nascent transcripts through sequence-directed cleavage activity (9, 19). The association of RITS and RDRC with centromeres facilitates the recruitment (through an as-yet-unidentified mechanism) of high levels of Clr4 activity, resulting in the further recruitment of Chp1 and RITS in an RNAi-dependent positive feedback loop (45), which serves to amplify the H3K9Me2 signal to support the maintenance of centromeric heterochromatin. The loss of the binding of Chp1 to Tas3-Ago1 is sufficient to break this feedback loop, resulting in the loss of recruitment of Tas3-Ago1 to centromeres, the loss of the association of RDRC, a failure to generate siRNAs, reduced levels of H3K9Me2 at centromeres, and, thus, the loss of centromeric recruitment of Chp1.

At telomeres and the mating-type locus, heterochromatin assembly relies on both RNAi-dependent and RNAi-independent mechanisms (17, 18, 20, 22, 24, 39). However, at these loci the loss of the RNAi pathway has little impact on H3K9Me2 levels, whereas at centromeres the loss of the RNAi-dependent positive feedback loop reduces H3K9Me2 levels below the critical threshold that can support the maintenance of heterochromatin. The RNAi-dependent positive feedback loop therefore is more critical to the maintenance of heterochromatin at centromeres than at other loci, suggesting that the RNAi-independent recruitment of Clr4 to centromeres is inefficient or that heterochromatin-destabilizing activities (such as the activity of transcriptional promoters) are more pronounced at centromeres.

Role of Tas3 and Chp1 at mat2/mat3 and subtelomeres.

Our studies of cells bearing Tas3_Δ10-24 have illuminated differences in the regulation of distinct heterochromatic loci. We demonstrate that at the mating-type locus, the Chp1-Tas3 association is not required for the maintenance of heterochromatin but is required for its establishment. While it was known that Chp1 plays a key role in the establishment of heterochromatin at the mating-type locus (42), through studying heterochromatin establishment in the Tas3_Δ10-24-TAP mutant we have demonstrated that it is likely the disruption of RITS, and not the loss of an RITS-independent Chp1 function, that impacts heterochromatin establishment. At telomeres, we see yet a different pattern of regulation. Here, we show that although the silencing of transcripts from the subtelomeric helicase genes is maintained following the disruption of RITS function in ago1-deficient cells, tlh transcripts accumulate upon the deletion of chp1, as seen previously (18), in tas3 null cells and in the tas3_Δ10-24 mutant. These results suggest that Tas3 and Chp1 function in a subcomplex that is independent of Ago1 (and therefore of RITS) to regulate tlh transcript levels. We saw no further accumulation of telomeric transcripts in tas3 chp1 double null cells (Fig. 6B) compared to that of chp1 null cells, which can be explained either by the epistasis of the two gene products, which is consistent with their association, or because Tas3 is unstable or not able to associate with telomeres in chp1-deficient cells (Fig. 6E). We note, however, that tlh transcripts always accumulate to a greater extent in chp1 null cells than in tas3 null cells, suggesting that Chp1 plays roles at telomeres independently of its association with Tas3. In contrast, Tas3 localization to telomeres is absolutely dependent on Chp1.

Interestingly, it has previously been reported that the deletion of chp1 does not prevent the silencing of telomeric reporter genes and does not affect the levels of H3K9Me2 or Swi6 association with the helicase genes (10, 22, 47). In addition, only transcripts from the forward (or sense) strand accumulate in chp1 null cells rather than transcripts from both strands, as is seen in clr4 null backgrounds (18). Thus, it is possible that the accumulation of tlh transcripts in tas3- and chp1-deficient cells and in the tas3_Δ10-24 mutant is not caused by a disruption of subtelomeric heterochromatin but by a defect in the processing of tlh transcripts, similarly to that reported recently for cells mutant in cid14 or components of the exosome complex (7), in which the Swi6 and Chp1 association with tlh genes is unperturbed even though tlh transcripts accumulate. Thus, Tas3 and Chp1 may be important for the proper recruitment of RNAi-independent exosome activities to modulate the posttranscriptional degradation of subtelomeric tlh gene transcripts. Alternatively, tlh transcript levels may simply accumulate following the increased transcription of these genes in tas3 or chp1 mutant backgrounds despite the maintenance of heterochromatin at this locus.