Set2 is a nucleosomal histone H3-selective methyltransferase that mediates transcriptional repression

Abstract

Recent studies of histone methylation have yielded fundamental new insights pertaining to the role of this modification in gene activation as well as in gene silencing. While a number of methylation sites are known to occur on histones, only limited information exists regarding the relevant enzymes that mediate these methylation events. We thus sought to identify native histone methyltransferase (HMT) activities from Saccharomyces cerevisiae. Here, we describe the biochemical purification and characterization of Set2, a novel HMT that is site-specific for lysine 36 (Lys36) of the H3 tail. Using an antiserum directed against Lys36 methylation in H3, we show that Set2, via its SET domain, is responsible for methylation at this site in vivo. Tethering of Set2 to a heterologous promoter reveals that Set2 represses transcription, and part of this repression is mediated through the HMT activity of the SET domain. These results suggest that Set2 and methylation at H3 Lys36 play a role in the repression of gene transcription.

FIG. 1.

Partial purification of two nucleosomal H3 HMT activities from S. cerevisiae. (A) Schematic representation of the chromatographic procedure used to partially purify two nucleosome-specific HMT activities from S. cerevisiae whole-cell extracts (see text for details). (B) Partially purified extracts fractionated by a Mini S column were assayed for HMT activity using chicken core histones or nucleosomes as the substrate. Incorporation of ³H-AdoMet was measured using the filter-binding assay, and identified peaks of activity are labeled. (C) A portion of the HMT assays from the above Mini S reactions involving nucleosomes were resolved on an SDS-15% PAGE gel and examined by Coomassie staining (lower) and fluorography (upper).

FIG.2.

Set2 is identified as a component of the first peak of nucleosomal HMT activity. (A) Peak 1 HMT activity from the Mini S column was fractionated using a Superose 12 size exclusion column, and collected fractions were analyzed for nucleosome-specific HMT activity by the filter-binding assay. Size estimation of peak 1 activity was determined to be ≈175 kDa; calibration sizes are shown. (B) Silver stain gel of the Superose 12 fractions containing HMT peak 1 activity. Candidate bands which correlated well with the peak of HMT activity were excised, and the proteins were identified by mass spectrometry. Asterisks surround the candidate band that was identified as Set2. (C) A schematic representation of the domain structure of Set2. The SET domain and flanking cysteine-rich regions (C) are shown. These flanking Cys-rich regions are also denoted as the pre- and post-SET domains, and we note that the pre-SET domain of Set2 is noncanonical (see reference for alignment). A putative proline-rich-binding motif (WW motif) is also shown. Peptide fragments identified by mass spectrometry and their relative locations in Set2 are shown.

FIG. 3.

Peak 1 HMT activity in S. cerevisiae is dependent on Set2. Whole-cell extracts prepared from S. cerevisiae strains which were either wild type, set2Δ, or set2Δ containing a Set2-Flag expression construct were bound and eluted from nickel-agarose, passed through a Mono Q column, and then fractionated on a Mono S column as outlined in Fig. 1. Fractions were assayed for nucleosome-specific HMT activity, and reaction products were resolved on SDS-15% PAGE gels followed by Coomassie staining (lower) and fluorography (upper). Western blot analysis of the expression and location of Set2-Flag on the Mono S column is shown. Asterisks indicate partial H3 breakdown products that are typically observed; P1 and P2 indicate peak 1 and peak 2, respectively.

FIG. 4.

Set2 methylation activity is dependent upon the SET domain. (A) Schematic representation of the domain structure of the recombinant Set2 protein expressed in bacteria. Substitutions made of highly conserved amino acids in the SET domain are shown. (B) Coomassie and Western blot analysis of affinity-purified recombinant Set2-Flag and Set2-Flag SET domain mutants. (C) Affinity-purified recombinant Set2-Flag and Set2-Flag SET domain mutants were incubated in the presence or absence of either chicken core histones (Free hist.) or nucleosomes (Nuc.) along with ³H-AdoMet. Reaction products were analyzed by the filter-binding assay. (D) RP-HPLC-purified histones isolated from HMT reactions involving recombinant Set2-Flag and nucleosomes were resolved on an SDS-15% PAGE gel and examined by Coomassie staining (lower panel) and fluorography (upper panel). Asterisks indicate partial H3 breakdown products that were observed.

FIG. 5.

Methylation of H3 at lysine 36 is mediated by Set2. (A) RP-HPLC-purified H3 isolated from Set2 reactions involving ³H-AdoMet and chicken nucleosomes was digested with endoproteinase Glu-C, and the resulting peptide fragments were separated by RP-HPLC. A labeled H3 fragment, determined to be amino acids 1 to 50 by mass spectrometry (data not shown), was further digested with the endoproteinase Arg-C followed by RP-HPLC fractionation. From this digestion, a ³H-methylated peptide fragment identified as being amino acids 27 to 38 in H3 by mass spectrometry (data not shown) was subjected to N-terminal automated sequencing, and ³H radioactivity eluted from each cycle was counted. Amino acids identified at each cycle of microsequencing are listed; numbers correspond to the known positions of H3 residues. (B) Recombinant H3 or chicken nucleosomes reacted in the presence or absence of purified recombinant Set2 along with AdoMet were immunoblotted with the α-Me(Lys36)H3 antiserum (upper panel). Identical reactions were performed and used in peptide competition assays in which the α-Me(Lys36)H3 antiserum was incubated for 2 h at 25°C with the various peptides shown (peptides were at 10 μg/ml). Parallel reactions were performed and examined by Coomassie staining to monitor loading (lower panel). Asterisk indicates the partial H3 breakdown product observed. (C) Set2 mediates H3 Lys36 methylation in vivo. Whole-cell extracts from wild-type S. cerevisiae strains or strains with a deletion of the SET domain were probed with either the α-Me(Lys36)H3 antiserum, the α-Me(Lys4)H3 antiserum, or the α-acetyl-H4 antiserum (used for a loading control). Details of the S. cerevisiae strains used are as follows (from left to right as shown in the panel): WT (MBY1198), set1Δ (YHR119w), WT (BY4742), set2Δ (YJL168c), set3Δ (YKR029c), set4Δ (YJL105w), set5Δ (YHR207c), set6Δ (YPL165c), and set7Δ (YDR257c). MBY1198 and BY4742 are the isogenic wild-type strains for the set1Δ and set2Δ to set7Δ strains, respectively. (D) The SET domain of Set2 is essential for mediating H3 Lys36 methylation in vivo. Whole-cell extracts from wild-type, set2Δ, and set2Δ strains with either the vector control plasmid (Vector), wild-type Set2 expression plasmid (Set2-Flag), or either of the SET domain mutants [Set2 (R195G) and Set2 (201A)] were probed with the α-Me(Lys36)H3 antiserum. For a loading control, these extracts were probed with the α-acetyl-H4-specific antiserum. Expression of wild-type or mutant Set2 proteins was monitored using the anti-Flag monoclonal antiserum.

FIG. 6.

SET domain point mutants in Set2 are partially defective in transcriptional repression when tethered to a heterologous promoter. (A) The wild-type S. cerevisiae strain YMH171 was transformed with either LexA vector-only, Rpd3-LexA, Set2(WT)-LexA, Set2(R195G)-LexA, or Set2(C201A)-LexA along with the CYC1-lexA_op-lacZ reporter plasmid (pCK30), and β-galactosidase assays were performed. Three independent transformants of each strain were assayed in duplicate, and the results are reported as the fold repression relative to the activity of the LexA vector alone. Asterisks indicate that the repression mediated by the two SET domain mutants of Set2 was significantly less (P ≤ 0.05) than that mediated by wild-type Set2-LexA. (B) To monitor expression of the LexA constructs, extracts from the above experiment were immunoblotted with an anti-LexA polyclonal antiserum. Expression of the vector-only LexA DNA-binding domain (which is 24.5 kDa) was omitted from the figure.