SETDB1 activity is globally directed by H3K14 acetylation via its Triple Tudor Domain

Abstract

SETDB1 (SET domain bifurcated histone lysine methyltransferase 1) is a major protein lysine methyltransferase trimethylating lysine 9 on histone H3 (H3K9) which is involved in heterochromatin formation and silencing of repeat elements (REs). It contains a unique Triple Tudor Domain (3TD), which specifically binds the dual modification of H3K14ac in the presence of H3K9me1/2/3. Here, we explored the role of the 3TD H3-tail interaction for the H3K9 methylation activity of SETDB1. We generated a binding reduced 3TD mutant and demonstrate in biochemical methylation assays on peptides and recombinant nucleosomes containing H3K14ac and H3K14ac analogs, respectively, that H3K14 acetylation is crucial for the 3TD mediated recruitment of SETDB1. We also observe this effect in cells where SETDB1 binding and activity is globally correlated with H3K14ac, and knockout of the H3K14 acetyltransferase HBO1 causes a drastic reduction in H3K9me3 levels at SETDB1 dependent sites. Regions with DNA hypomethylation after SETDB1 knockout also show an enrichment in SETDB1-dependent H3K9me3 and H3K14ac. Further analyses revealed that 3TD is particularly important at specific target regions like L1M REs, where H3K9me3 cannot be efficiently reconstituted by the 3TD mutant of SETDB1. In summary, our data demonstrate that the H3K9me3 and H3K14ac are not antagonistic marks but rather the presence of H3K14ac is required for SETDB1 recruitment via 3TD binding to H3K9me1/2/3-K14ac regions and establishment of H3K9me3.

Graphical Abstract

Figure 1.

Specificity analysis of human SETDB1. (A) Schematic representation of SETDB1 domains. The mutations used in this study are indicated (based on Uniprot Q15047 and (23)). (B) Substrate sequence specificity analysis of full-length SETDB1 (see also Supplementary Figure S1B) using SPOT peptide array methylation of H3 (1–15) peptides. The array contains a systematic collection of single mutants of position 1–15 of the H3K9 target sequence in which each amino acid residue was exchanged against 18 natural amino acids. The peptide array was methylated by SETDB1 in methylation buffer containing radioactively labeled AdoMet and the transfer of methyl group was detected by autoradiography. For a corresponding substrate specificity analysis of the SETDB1 F332A mutant see Supplementary Figure S2. (C) Analysis of two independent peptide array methylation experiments, where the signals were quantified, averaged, normalized and activities indicated in grayscale (see also Supplementary Figure S1C). (D) Discrimination factor analysis showing the sequence substrate specificity of SETDB1 in methylating the H3K9 peptide.

Figure 2.

Peptide and recombinant nucleosome methylation by SETDB1. H3 peptides (A), recombinant nucleosomes (B) and mixed peptides (C) containing different modifications were incubated with full-length WT SETDB1 or its 3TD mutant F332A in methylation buffer containing radioactively labeled AdoMet, and the transfer of methyl group was detected by autoradiography. The data are represented as means ± SEM of at least four (panel A) or three (panels B and C) independent experiments. See also Supplementary Figures 1B, 3 and 4; H3K9me3: , H3K14ac: .

Figure 3.

SETDB1 is recruited to H3K14ac containing target loci. (A) Scheme of the cell HCT116 derived lines generated in this study by SETDB1 KO followed by reconstitution with SETDB1 WT or mutants and by HBO1 KO. (B and C) H3K9me3 (B) and H3K14ac (C) ChIP-qPCR at known SETDB1 target regions (51,52) showing changes in the histone modification as indicated upon SETDB1 KO and reconstitution with WT SETDB1 of its catalytically inactive mutant H1224K or the 3TD mutant F332A. Moreover, HBO1 KO cells were investigated. ChIP was performed on mononucleosomes isolated from two individual biological replicates from each cell line (represented as dots). Averages of the individual measurements are represented as bars.

Figure 4.

SETDB1 is dependent on H3K14ac for efficient methylation of H3K9. (A) Heatmaps were generated using all H3K9me3 peaks from the different cell lines for H3K9me3 and H3K14ac ChIP-seq as indicated and were plotted as regions with 3 kb flanking either side. K-means clustering was based on the differential H3K9me3 signals, either as SETDB1 independent or SETDB1 dependent. Sorting was by decreasing H3K9me3 intensity. Datasets form the literature of KAP1 in HCT116 cells (GSM1866695) (57) and SETDB1 in HEK293 cells (GSM5331059) (58). See also Supplementary Figures S6 and S7C. (B) IGV browser views of ChIP-seq profiles at exemplary SETDB1 dependent regions. (C) Enrichment and depletion of SETDB1 dependent and independent H3K9me3 peaks in different chromatin regions. In the upper panel, occurrence of peaks in chromatin types (ChromHMM) of HCT116 cells was determined and compared with randomized peaks to calculate observed/expected ratios. In the lower panel, occurrence of peaks in RE annotated by RepeatMasker was analyzed in the same way. ZNF/Rpts: Zinc-finger genes and RE, Het: Heterochromatin, ReprPC: repressed PolyComb, ReprPCWk: Weak repressed PolyComb, Quies: Quiescent. (D) Averaged H3K14ac ChIP-seq signal from panel (A) in SETDB1 dependent and independent regions in HCT116 and SETDB1 KO. (E) Heatmap of H3K9me3 ChIP-seq in SETDB1 KO cells after rescue with WT or F332A SETDB1. SETDB1-dependent H3K9me3 regions were sorted by decreasing intensity of SETDB1 WT/F332A, which was then applied for sorting KAP1 and SETDB1 tracks.

Figure 5.

3TD of SETDB1 is important for the recruitment and efficient methylation in the L1M RE. (A) Enrichment and depletion of H3K9me3 peaks of different cell lines in RE. RE families and subfamilies were annotated by RepeatMasker. (B) Enrichment of H3K9me3 peaks in SETDB1 WT and F332A mutant cells for individual RE classes. The log2-fold change of H3K9me3 peak enrichment in SETDB1 WT over F332A mutant is plotted against the corresponding P-value. Red dots represent individual RE with stronger methylation in SETDB1 WT. Some of the L1M elements which are only methylated in SETDB1 WT are annotated. Green dots represent RE with stronger methylation in the F332A mutant cell line showing enrichment of SVA and Alu elements. (C and D) Heatmap H3K9me3 and K14ac signals on SETDB1 specific L1M RE (C) or SVA (SINE-VNTR-Alus) retrotransposons (D) sorted based on decreasing RE region length. (E and F) Example browser views of H3K9me3 and H3K14ac tracks on representative L1M (E) and SVA (F) loci.

Figure 6.

H3K9me1/2/3-K14ac mediated efficient methylation of H3K9 by SETDB1. Schematic representation of the SETDB1 dependent trimethylation of H3K9 where 3TD is bound to H3K14ac, which is installed by HBO1, combined with H3K9me1/2/3 in one histone tail which then positions the SET domain of SETDB1 to methylate the other H3K9 of the same nucleosome. This is followed by the removal of the H3K14ac by co-recruited HDAC activity resulting in the formation of heterochromatin.