Refined read‐out: The hUHRF1 Tandem‐Tudor domain prefers binding to histone H3 tails containing K4me1 in the context of H3K9me2/3

Abstract UHRF1 is an essential chromatin protein required for DNA methylation maintenance, mammalian development, and gene regulation. We investigated the Tandem‐Tudor domain (TTD) of human UHRF1 that is known to bind H3K9me2/3 histones and is a major driver of UHRF1 localization in cells. We verified binding to H3K9me2/3 but unexpectedly discovered stronger binding to H3 peptides and mononucleosomes containing K9me2/3 with additional K4me1. We investigated the combined binding of TTD to H3K4me1‐K9me2/3 versus H3K9me2/3 alone, engineered mutants with specific and differential changes of binding, and discovered a novel read‐out mechanism for H3K4me1 in an H3K9me2/3 context that is based on the interaction of R207 with the H3K4me1 methyl group and on counting the H‐bond capacity of H3K4. Individual TTD mutants showed up to a 10,000‐fold preference for the double‐modified peptides, suggesting that after a conformational change, WT TTD could exhibit similar effects. The frequent appearance of H3K4me1‐K9me2 regions in human chromatin demonstrated in our TTD chromatin pull‐down and ChIP‐western blot data suggests that it has specific biological roles. Chromatin pull‐down of TTD from HepG2 cells and full‐length murine UHRF1 ChIP‐seq data correlate with H3K4me1 profiles indicating that the H3K4me1‐K9me2/3 interaction of TTD influences chromatin binding of full‐length UHRF1. We demonstrate the H3K4me1‐K9me2/3 specific binding of UHRF1‐TTD to enhancers and promoters of cell‐type‐specific genes at the flanks of cell‐type‐specific transcription factor binding sites, and provided evidence supporting an H3K4me1‐K9me2/3 dependent and TTD mediated downregulation of these genes by UHRF1. All these findings illustrate the important physiological function of UHRF1‐TTD binding to H3K4me1‐K9me2/3 double marks in a cellular context.


Supplemental Figure 2. Influence of arginine methylation on binding of H3K9me2/3 peptides by TTD.
A Structure of the UHRF1 TTD-LIG1 peptide complex ( 1 .PDB: 5YY9).To investigate the potential influence of R-methylation on TTD peptide binding, we examined crystallographic data.TTD is shown in tan as surface.The peptide is shown in orange.R121 corresponding to H3K4 and K126me3 corresponding to H3K9me3 are shown in blue and pink.They approach the protein in two distinct deep binding pockets.P119 corresponding to H3R2 and R125 corresponding to H3R8 are shown in yellow.These residues point into the solvent (P119) or are placed on the surface of the domain (R125) without an apparent opportunity for preferential Rme2 binding.B As shown in panel (A), the H3R8 equivalent residue makes a surface contact to the protein, hence the effect of its methylation has been studied in more detail.TTD binding to H3 peptides with K9me2 on Celluspots peptide arrays was inhibited by both structural isomers of R8me2 (symmetric and asymmetric).Unfortunately, no H3R2me-K9me2/3 double mark peptides are available on CelluSpots arrays.C hUHRF1-TTD binding to the H3R8me2s-K9me3 peptide was studied in equilibrium peptide binding titrations analysed by the fluorescence anisotropy (FA) change and shown to have a KD of 2400 nM, while the KD of binding to H3K9me3 is 680 nM.This result inidcates that H3-tail binding is inhibited in presence of R8me2s.

Supplemental Figure 3. Controls and exemplary primary data of the hUHRF1-TTD peptide binding experiments shown in Figure 2 and Table 1.
A TTD wild-type (WT) binds H3K4me1-K9me2/3 peptides more strongly than H3K9me2/3 alone in equilibrium peptide binding titrations analysed by the fluorescence anisotropy (FA) change.H3K9me2 spans extremely wide regions of the genome.Chromosome arm-wide exemplary browser views showing megabase-wide regions of H3K9me2 enrichment.Data from a histone mark established as very broad (H3Κ9me3) show sharper and more defined regions of enrichment.HepG2 H3Κ9me3 data were taken from public datasets 2 .All tracks in RPKM, yaxes start from 0. All coordinates in hg38, gene annotation from RefSeq.

Supplemental Figure 12. Additional data related to promoter and enhancer binding in HepG2 cells.
A Genes robustly down-regulated in HepG2 vs. liver tissue (same genes as in Figure 5D) are HepG2 and liver specific, according to gene set enrichment analysis (GSEA) on Enrichr (maayanlab.cloud/Enrichr) 3 .
B TTD peaks occur more frequently on enhancers, especially distal enhancers as annotated by ChIP-Enrich (chip-enrich.med.umich.edu).
D Genes with TTD peaks on their enhancers (found in HepG2, same genes as in Figure 6C) are HepG2 specific, according to GSEA on Enrichr.E Exemplary browser views showing hUHRF1-TTD flanking peaks of selected cell-type specific transcription factors, ARID5B 5 , FOXA1 aka HNF3α 6 , HNF4A 6 , and DNase 7 in HepG2 cells.7.

Supplemental Figure 13. Additional data related to full-length mUHRF1 colocalization with H3K4me1 and gene regulation by UHRF1 in HCT116 cells shown in Figure
A Additional data referring to the mUHRF1 and H3K4me1 ChIP-seq shown in Figure 7B.Plot of the average mUHRF1 and H3K4me3 ChIP-seq signals in 2 kb bins genome-wide and Pearson´s correlation (r).Comparison with Figure 4G shows that mUHRF1 ChIP-seq correlates with H3K4me1 better than H3K4me3 in E14 mESC.7C-E.B Counts of differentially regulated genes (DRGs) and corresponding enhancer regions using FC ≥ |1.5| from microarrays using UHRF1 KD treated HCT116 8 .HCT116 cells with wildtype rescued cells over mock (WT) microarray had less DRGs than the Y188A rescued over mock (TTD*).

B-D Additional data related to Figure
C TTD* up-regulated genes are enriched in H3K9me2 on their FANTOM5 enhancers 9 .Shown are the differentially regulated genes (DRGs) from the wild-type (WT) over mock non-responsive genes, sorted according to their status in Y188A mutant over mock (TTD*).The mean H3K9me2 signal of each group was plotted from HCT116 ChIP-seq data 10 .Central line is median, box borders are 25th to 75th percentile, and whiskers 5th to 95th.p values are from one-way ANOVA with Bonferroni correction.non-r.non-responding.D TTD* up-regulated genes are enriched in H3K4me1-K9me2 on their FANTOM5 enhancers 9 .Shown are the differentially regulated genes (DRGs) from all genes sorted according to their status in Y188A mutant over mock (TTD*).The mean H3K4me1 and H3K9me2 signal of each group was plotted from HCT116 ChIP-seq data 10,11

CSupplemental Figure 4 .Supplemental Figure 5 .ABCSupplemental Figure 7 .Supplemental Figure 9 .
Control data showing that TTD WT binds weakly to the H3K4me1 peptide.Representative data showing peptide binding of TTD M224A.D Representative data showing peptide binding of TTD D142A.E Representative data showing peptide binding of TTD D142E.Data points are average fraction bound (Θ) of n ≥2 independent experiments, error bars are 0.95 confidence intervals (CI).KD are the mean of n ≥2 independent fits, errors are 0.95 confidence intervals (CI).Antibody validation data related to Figure 3B and Supplemental Figure 5A.Lot-specific validation of antibody specificity using CelluSpots arrays, performed in the same time frame as the corresponding western blot experiments.All used antibodies are specific for the correct H3 PTM.A Antibody validations for the CIDOP experiments shown in Supplemental Figure 5A.The α-K4me1 was tested under ChIP conditions and the α-K9me2 under western blot conditions to validate the antibodies' specificity in sequential ChIP-western blot.Catalogue and lot numbers are on the top left of the corresponding array.The target epitope is indicated in the box above the array and peptides with the epitope are marked with a circle of the same colour.B Antibody validations for the CIDOP experiments shown in Figure 3B.Experiments were conducted under western blot conditions (Supplemental Table 3).Additional data related to Figure 3B.Detection of the occurrence of the H3K4me1-K9me2 double mark by H3K4me1 ChIP followed by H3K9me2 western blot.Primary data of western blots from 5 biological replicates of the α-K4me1 ChIP experiments.Quantification of western blots shown in panel A from n = 5 biological replicates of α-K4me1 ChIP experiments.The bars show the mean and error bars 0.95 confidence intervals.Primary data of western blots from 3 additional biological replicates of the CIDOP experiments shown in Figure 3B.D Quantification of western blots shown in panel C from n ≥ 3 biological replicates of CIDOP experiments.The bars show the mean and error bars 0.95 confidence intervals (CI).Significance levels were assigned as follows: n.s.p > 0.05, *p ≤ 0.05, **p ≤ 0.01.Supplemental Figure 6.Validation of CIDOP-seq data shown in Figure 3C by CIDOP-qPCR.A TTD CIDOP-qPCR and α-H3K9me2 ChIP-qPCR data showed enrichment at a H3K9me2 reporter region (A) and depletion at a H3K4me3 reporter region (B) for two independent biological replicates.Control experiments with binding deficient TTD D142A mutant and IgG demonstrated the specificity of the assay.Control ChIP with α-H3K9me2 and CIDOPs with MPP8-CD and TAF3-PHD verified the assayed amplicons.For all regions n = 2, the bars represent the mean and error bars 0.95 confidence intervals (CI).B Coordinates and amplicon size of H3K9me2 and H3K4me3 reporter regions used in the qPCR assays.C CIDOP-seq and ChIP-seq data for the reporter loci used in qPCR assays.The amplicon borders are annotated with vertical bars in the middle of the window.All tracks in Reads Per Kilobase of transcript, per Million mapped reads (RPKM), y-axes start from 0. Coordinates in hg38.Additional browser views of the CIDOP-seq and ChIP-seq data shown in Figure 3C demonstrating reproducibility of experimental replicates.Data from hUHRF1-TTD CIDOP-seq and α-H3K9me2 ChIP-seq in two biological replicates showed good reproduction and were pooled.Representative browser views.All tracks in RPKM, y-axes start from 0. Coordinates in hg38.Supplemental Figure 8.Additional browser views of the CIDOP-seq and ChIP-seq data shown in Figure 3C demonstrating TTD CIDOP signal enrichment in regions with both H3K4me1 and H3K9me2/3.Additional representative data showing that hUHRF1-TTD CIDOP-seq is enriched in regions with both H3K4me1 and H3K9me2/3.All tracks in RPKM, y-axes start from 0. Coordinates in hg38.Additional browser views of the ChIP-seq data shown in Figure 3E demonstrating the broad genomic localisation of H3K9me2 and H3K9me3.

Table 5 . Oligonucleotides used for qPCR assays in this study.
. Central line is median, box borders are 25th to 75th percentile, and whiskers 5th to 95th.p values are from one-way ANOVA with Bonferroni correction.non-r.non-responding.

Table 6 . NGS public datasets used in this study.
Published ChIP-seq data were downloaded as fastq files or pre-processed ENCODE data (DNase, WGBS).Microarray data were downloaded as pre-processed txt files containing the Lowess normalized log2(FC) ratio.