Sir2 is required for Clr4 to initiate centromeric heterochromatin assembly in fission yeast

Abstract

Heterochromatin assembly in fission yeast depends on the Clr4 histone methyltransferase, which targets H3K9. We show that the histone deacetylase Sir2 is required for Clr4 activity at telomeres, but acts redundantly with Clr3 histone deacetylase to maintain centromeric heterochromatin. However, Sir2 is critical for Clr4 function during de novo centromeric heterochromatin assembly. We identified new targets of Sir2 and tested if their deacetylation is necessary for Clr4-mediated heterochromatin establishment. Sir2 preferentially deacetylates H4K16Ac and H3K4Ac, but mutation of these residues to mimic acetylation did not prevent Clr4-mediated heterochromatin establishment. Sir2 also deacetylates H3K9Ac and H3K14Ac. Strains bearing H3K9 or H3K14 mutations exhibit heterochromatin defects. H3K9 mutation blocks Clr4 function, but why H3K14 mutation impacts heterochromatin was not known. Here, we demonstrate that recruitment of Clr4 to centromeres is blocked by mutation of H3K14. We suggest that Sir2 deacetylates H3K14 to target Clr4 to centromeres. Further, we demonstrate that Sir2 is critical for de novo accumulation of H3K9me2 in RNAi-deficient cells. These analyses place Sir2 and H3K14 deacetylation upstream of Clr4 recruitment during heterochromatin assembly.

Figure 1

The presumed catalytic mutant, sir2N247A, causes defects in the maintenance of subtelomeric heterochromatin. (A) Alignment of fission yeast Sir2 with nearest homologues from budding yeast and man. The fission yeast sir2N247A mutant is analogous to the budding yeast Sir2pN345A, which lacks catalytic activity. Position of the N247A substitution is indicated. (B) The predicted catalytic mutant Sir2N247A–TAP is stably expressed. Western blot of extracts from Sir2–TAP and sir2N247A–TAP expressing cells, probed for TAP with tubulin as loading control. (C) Subtelomeric transcripts accumulate in sir2Δ and sir2N247A mutant cells. mRNA transcripts from subtelomeric tlh genes were quantified by quantitative real-time PCR (qRT–PCR) amplification of cDNA, and normalized to transcript levels of the adh1⁺ (alcohol dehydrogenase) control. Graphic data present the average of two distinct experimental replicates and error bars depict s.e.m. (D) H3K9 acetylation is increased and (E) H3K9me2 methylation is abolished at subtelomeres in sir2 deletion and sir2N247A mutants. qRT–PCR analysis of ChIP experiments monitoring relative enrichment of tlh sequences over adh1⁺ control euchromatic locus in immunoprecipitated samples. Graphs represent average of data obtained from two biological samples, with error bars depicting s.e.m.

Figure 2

Sir2 functions redundantly with Clr3 for centromeric heterochromatin maintenance. (A) Deletion of sir2 or sir2N247A does not impact centromeric transgene silencing. Serial dilutions of WT, clr4Δ, sir2Δ, and sir2N247A yeast strains bearing the cen::ura4⁺ transgene were assayed for growth on nonselective (complete) medium, as well as selective medium lacking uracil (−URA), and complete medium additionally containing +FOA. (B) Centromeric transcripts do not appreciably accumulate in sir2 mutants. Centromeric dh, dg, and ura4⁺mRNA transcripts were quantified by quantitative real-time PCR amplification of cDNA, and normalized to transcript levels of the adh1⁺ control. Graphic data present the average of two distinct experimental replicates and error bars depict s.e.m. (C) siRNA production is unaffected in sir2 mutants. Centromeric dh siRNA production was evaluated by northern blot relative to snoRNA controls in WT, clr4Δ, and sir2Δ mutants. (D) H3K9 acetylation at centromeres is increased in sir2 mutants, while H3K9 dimethylation, (E) is unaffected. ChIP experiments were performed using antibodies against the acetylated or dimethylated H3K9 epitope. Specific enrichment of immunoprecipitated centromeric dh sequences was determined relative to the adh1⁺ euchromatic locus by real-time PCR. Graphical data present the average of two biological replicates, with error bars depicting the s.e.m. (F) sir2Δ and (G) sir2N247A mutants function redundantly with Clr3 to maintain centromeric transgene silencing. Serial dilutions of yeast strains bearing the cen::ura4⁺ transgene were assayed for growth on nonselective (complete) medium, as well as selective medium lacking uracil (−URA), and complete medium additionally containing +FOA. (H) Sir2 and Clr3 function redundantly to maintain centromeric heterochromatin. Swi6 recruitment to centromeres is lost in sir2 clr3 compound mutants. ChIP experiments were conducted using antibodies specific for Swi6, and enrichment of centromeric dh sequences relative to adh1⁺ was quantified by real-time PCR and normalized to input DNA. Data represent the average of two biological replicates, with error bars depicting the s.e.m.

Figure 3

Sir2 is required for de novo silencing and establishment of Clr4-dependent H3K9 methylation at centromeres. (A) Schematic of assay used to evaluate the requirement for sir2⁺ in promoting de novo clr4⁺-dependent silencing activity at centromeres. (B) Sir2 is required for de novo centromeric silencing upon reintroduction of the clr4⁺ methyltransferase. Serial dilutions of yeast bearing the cen::ura4⁺ transgene were assayed for growth on complete medium, as well medium lacking uracil (−URA), and complete medium containing +FOA. (C) Reintroduction of the clr4⁺ methyltransferase in sir2Δ clr4Δ cells reveals defective establishment of transcriptional silencing at centromeres. Centromeric dh, dg, and ura4⁺ mRNA transcripts were quantified by quantitative real-time PCR amplification of random primed cDNA and normalized to transcript levels for the euchromatic adh1⁺ control. Experimental data present the average of two experimental replicates. (D) siRNA production occurs in cells defective for de novo centromeric silencing. Centromeric dh siRNA production was evaluated by northern blot relative to snoRNA controls. (E–H) Changes in chromatin structure underlie transcriptional silencing defects in cells defective for de novo centromeric silencing. ChIP experiments were performed using antibodies specific for dimethylated (E) or acetylated H3K9 (F), acetylated H3K14 (G) and Chp1 (H), and data represent the average of two experimental replicates, including a total of four biological replicates from two distinct sir2Δ clr4Δ to clr4⁺ reintegrant strains.

Figure 4

The sir2N247A mutant exhibits defects in de novo assembly of heterochromatin at centromeres. (A) sir2N247A cells show some defect in establishment of centromeric silencing. Serial dilution assays were performed to monitor silencing of cen::ura4 transgene. (B) De novo silencing defects observed in yeast expressing sir2N47A reflect defective transcriptional silencing at centromeres. Centromeric dh, dg, and ura4⁺ mRNA transcripts were quantified by quantitative real-time PCR and normalized to levels of the euchromatic adh1⁺control. Data represent the average of two experimental replicates, including a total of four biological replicates from two distinct sir2N247A-his3 clr4Δ to clr4⁺ reintegrant strains. (C) sir2N247A mutation does not disrupt siRNA production during maintenance or establishment. Centromeric dh siRNA production was evaluated by northern blot relative to that of a snoRNA control in each of the cell backgrounds indicated.

Figure 5

Sir2, but not the catalytic point mutant sir2N247, is catalytically active in vitro, and exhibits preference for deacetylation of H3K4Ac and H4K16Ac peptides. (A) Affinity purified recombinant GST–Sir2 and GST–sir2N247A are stable proteins. Equal amounts (∼1 μg) of GST, GST–Sir2, and GST–sir2N247A were resolved by SDS–PAGE and stained using Coomassie brilliant blue. (B) GST–Sir2, but not GST–sir2N247A, is catalytically active and sensitive to inhibition by nicotinamide. Catalytic activity of GST fusion proteins was evaluated by fluorogenic deacetylation assay. GST–Sir2 activity was also evaluated in the presence 2 mM nicotinamide, a sirtuin inhibitor. (C, D) GST–Sir2, but not GST–sir2N247A, promotes ³²P NAD⁺ cofactor hydrolysis in the presence of calf thymus histones. (C) The ability of GST fusions to promote hydrolysis of ³²P NAD⁺ was appraised in the presence of acetylated calf thymus histones by TLC and autoradiography. GST–Sir2 activity was also evaluated in the presence 5 mM nicotinamide. (D) Graphical representation of data from panel C, one of the two experimental replicates, following analysis by quantitative densitometry. (E) Sir2 exhibits preference for deacetylation of H3KAc and H4K16Ac peptides, but also deacetylates H3K9Ac and H3K14Ac peptides. Evaluation of ³²P NAD⁺ hydrolysis products was performed following the coupled NAD⁺ hydrolysis/lysine deacetylation reaction, in the presence of differentially acetylated or unacetylated peptides derived from the N-terminal 19 amino acids of Schizosaccharomyces pombe histones H3 or H4. TLC autoradiographs were evaluated by computational densitometry. Graphs represent the mean of two experimental replicates, for which representative autoradiographs are presented in Supplementary Figure S2A–C.

Figure 6

H3K4 and H4K16 do not contribute to de novo centromeric silencing. (A, B) The integrity of H3K9 and H3K14, but not H3K4 or H4K16, is critical to maintenance of centromeric silencing. Accumulation of centromeric transcripts from the dh (A) and dg (B) repeats was evaluated by quantitative real-time PCR of cDNA derived from yeast expressing sole copies of histone H3 and histone H4 with the mutations indicated. Euchromatic act1⁺ transcripts serve as a normalization control. Neither H3K4 (C, D) nor H4K16 (E, F) are critical for de novo centromeric silencing mediated by clr4⁺. Experimental data present the average of two experimental replicates, including a total of four biological replicates from two distinct H3K4A clr4Δ to clr4⁺ and H4K16G clr4Δ to clr4⁺ reintegrant strains. (G) Flag–Clr4 is equivalently expressed in cells bearing mutations in histone H3. Western analysis of clr4Δ cells bearing different histone H3 mutations and transformed with episomal Flag–Clr4 expression vector demonstrates equivalent Flag–Clr4 expression compared with tubulin loading control. (H) Flag–Clr4 is efficiently recruited to centromeres, but fails to localize in H3K9A and H3K14A mutant cells. Q–PCR analysis of ChIP for Flag–Clr4 recruitment to centromeric dh sequences compared with adh1⁺ euchromatic control. Data averaged from five experimental replicates, with s.e.m. shown.

Figure 7

De novo H3K9 methylation and Swi6 recruitment in RNAi-deficient backgrounds is dependent on Sir2. (A) Sir2 is required for centromeric H3K9 methylation upon clr4⁺ expression in RNAi-defective cells. Quantitative real-time PCR analysis of ChIP experiments performed with anti-H3K9me2 antibody, monitoring relative enrichment of centromeric dg and dh sequences over adh1⁺ euchromatic locus following episomal clr4⁺ expression in clr4Δ, clr4Δdcr1Δ, and clr4Δdcr1Δsir2Δ strains. Results are s.e.m. for duplicate independent ChIP experiments, with inclusion of duplicate data for two biological replicates for clr4Δdcr1Δsir2Δ strains. (B) Sir2 is required for centromeric Swi6 recruitment upon clr4⁺ expression in RNAi-defective cells. Swi6 ChIP experiments performed on strains outlined in (A), assessing relative enrichment at centromeric dg and dh sequences relative to adh1⁺ control.