DAXX adds a de novo H3.3K9me3 deposition pathway to the histone chaperone network

Abstract

A multitude of histone chaperones are required to support histones from their biosynthesis until DNA deposition. They cooperate through the formation of histone co-chaperone complexes, but the crosstalk between nucleosome assembly pathways remains enigmatic. Using exploratory interactomics, we define the interplay between human histone H3-H4 chaperones in the histone chaperone network. We identify previously uncharacterized histone-dependent complexes and predict the structure of the ASF1 and SPT2 co-chaperone complex, expanding the role of ASF1 in histone dynamics. We show that DAXX provides a unique functionality to the histone chaperone network, recruiting histone methyltransferases to promote H3K9me3 catalysis on new histone H3.3-H4 prior to deposition onto DNA. Hereby, DAXX provides a molecular mechanism for de novo H3K9me3 deposition and heterochromatin assembly. Collectively, our findings provide a framework for understanding how cells orchestrate histone supply and employ targeted deposition of modified histones to underpin specialized chromatin states.

Keywords: ASF1; DAXX; HJURP; NASP; epigenetic; gene silencing; heterochromatin; histone chaperone; nucleosome assembly; protein network; proteomics.

Graphical abstract

Figure 1

Histone chaperones directly interact with diverse cellular processes (A) The triple SILAC IP-MS strategy for mapping the histone chaperone network. (B) Network analysis of histone-independent interactors identified for ASF1a, ASF1b, NASP, DAXX, HJURP, and DNAJC9. Nodes and edges are colored based on their identification in the different pull-downs. (C) Upset plot showing the overlap of histone-independent interactomes, colored as in (B). (D) Clustering of histone-independent interactors using functional associations annotated in the STRING database and the MCL algorithm. Protein-protein functional associations are shown with a black line according to the string database, protein nodes colored as in (B). See also Figure S2A. (B–D) Proteins are referred to by human UniProt protein identification code. Data generated from n = 4 biological replicates. See also Table S1 and Figures S1 and S2.

Figure 2

Histone chaperones collaborate through their histone-dependent associations (A) Network analysis of the histone-dependent interactors identified for ASF1a, ASF1b, NASP, DAXX, HJURP, DNAJC9, MCM2, and TONSL. Nodes and edges are colored based on their identifications in the different pull-downs. (B) Upset plot showing the overlaps between histone-dependent interactomes. Protein nodes, edges are colored based on their identifications in the different pull-downs, colored as in (A). (C) Clustering of histone-independent interactors using functional associations annotated in the STRING database and the MCL algorithm. Protein-protein functional associations are shown with a black line according to the string database, protein nodes colored as in (A). (A–C) Proteins are referred to by human UniProt protein identification code. Data generated from n = 4 biological replicates. See also Table S1 and Figure S1.

Figure 3

Validation of previously uncharacterized histone-dependent interactors (A) Bubble plot showing histone-dependent interactors across triple SILAC IP-MS interactomes. Proteins are referred to by human UniProt protein identification code. Data generated from n = 4 biological replicates. Colors represent Log2 SILAC ratios (HBM/WT), and radii represent p values. See also Table S1. (B) Pull-downs of Strep-HA-tagged histone chaperones WT, HBM, B domain mutant (BDM), or ATRX-binding mutant (ABM) compared with control purifications (−) from soluble cell extracts probed by western blot. Representative of n = 2 biological replicates. (C) “Local” AlphaFold prediction of SPT2 (yellow) and ASF1A (magenta) histone-binding domains bound to H3.1–H4 (red and cyan) depicting the high-confidence regions of the full-length AlphaFold prediction. (D) Predicted alignment error (PAE) plot showing confidence of residue contacts in the full-length SPT2–H3.1–H4–ASF1A “global” AlphaFold prediction. Red dashed lines indicate high-confidence interactions between protein chains in the “local” prediction shown (left). See also Figures 3A–3C. (E) Alignment of local AlphaFold prediction of SPT2–H3.1–H4–ASF1A (colored as C) to the crystal structure of SPT2–(H3.2–H4)₂ (white, PDB: 5BS7, with H3.2–H4 omitted for clarity). See also Figures S3D and S3E.

Figure 4

Histone chaperone perturbation demonstrates their functional connectivity (A) Clustering analysis showing Euclidean distances between median normalized LFQ intensities (LFQ_M.N.) for proteins identified in H3.1 and H3.3 pull-downs by MS. (B) Bubble plot showing the changes in abundance of proteins identified in histone H3.1 and H3.3 pull-downs from extracts siRNA depleted of the chaperones ASF1A, ASF1B, NASP, and DAXX compared with control conditions (siCTRL). Colors represent Log2 ratios of median normalized LFQ intensities (siRNA/siCTRL), and radii represent p values. (A and B) Representative of n = 5 biological replicates, with median normalized LFQ intensities quantified on a protein or ^∗peptide level by MS. Proteins are referred to by human UniProt protein identification code. See also Table S1. (C) Western blot of soluble extracts from cells expressing H3.1-FlagHA (left) and H3.3-FlagHA (right) siRNA depleted for ASF1A, ASF1B, DAXX, or NASP and compared with control knockdowns siCTRL. Representative of n = 5 biological replicates. See also Figure S4.

Figure 5

DAXX escorts new H3.3 K9 methylated histones prior to deposition (A) Strategy for profiling the peptides and PTMs of histones H3 and H4 from soluble sNASP and DAXX purifications. (B) Quantification of H3.1/2 and H3.3 peptides associated with sNASP and DAXX. (C–E) Quantification of PTMs on (C) H4 peptides 20–23, (D) H4 peptides 4–17, and (E) H3 peptides 9–17 associated with sNASP and DAXX. (B–E) Percentages are relative to the total intensity of related peptides and averaged across n = 4 biological replicates of sNASP and DAXX purifications. PTMs were normalized using heavy peptides standards. Error bars represent SD, p-values represent unpaired two-sided t tests. See also Figures S5 and S7 and Table S2. (F) Pull-downs of Strep-HA-tagged DAXX WT, HBM, or ABM compared with control purifications (−) from soluble cell extracts probed by western blot for histone modifications and compared with histone PTM levels on chromatin by serial dilution. Representative of n = 2 biological replicates. (G) (Left) Experimental design. Strep-HA-DAXX expressing cells were pulsed with heavy SILAC medium for 24 h. (Right) Proportion of new and old histones on chromatin and DAXX soluble complexes across n = 6 biological replicates. Error bars represent SD. See also Table S1. (H) Western blot of soluble and chromatin extracts from cells treated with cycloheximide (CHX, 15–30 mins) and compared with DMSO control. Representative of n = 2 biological replicates.

Figure 6

DAXX promotes the catalysis of H3.3 K9me3 in collaboration with SETDB1 (A) Quantification of PTMs on H3 peptides 9–17 associated with DAXX upon SETDB1, SUV39H1/H2, or control siRNA depletions averaged across n = 4 (siSETDB1 and siSUV39H1/H2) and n = 3 (siCTRL) biological replicates. See also Figures S6A and S7. (B) In vitro histone methyltransferase assay with recombinant proteins analyzed by western blot with average quantification of H3K9me3 representative of n = 4 independent experiments. H3K9me3 band intensities were normalized on corresponding H4 intensities. Error bars represent SD, p-values represent unpaired two-sided t tests. (C) LFQ quantification of K9 methylation on H3 peptides 9–17 acetylated on K14 detected in H3.3 (left) and H3.1(right) IP-MS. Mean values are indicated representative of n = 5 biological replicates, p-values represent unpaired two-sided t tests. Data from Figure 4 reanalyzed for PTMs, see STAR Methods. (D) Pull-downs of Strep-HA-tagged DAXX from soluble cell extracts expressing Strep-HA-DAXX WT, HBM, or SIMΔ mutants compared with control purifications probed by western blot. Images shown are representative of n = 2 biological replicates. See also Figure S6B. (E) Bubble plot showing the changes in abundance for factors associated with DAXX HBM, ABM, or SIMΔ mutants compared with WT DAXX. Proteins are referred to by human UniProt protein identification code. SETDB1-linked factors are indicated in magenta. Data generated from n = 4 biological replicates. DAXX WT vs. SIMΔ dataset was bait normalized to account for the lower expression level of DAXX SIMΔ mutant. Colors represent Log2 SILAC ratios (Mut/WT), and radii represent p values. See also Table S1 and Figures S6E and S6F. (F) Quantification of PTMs on H3 peptides 9–17 associated with DAXX WT or SIMΔ mutant averaged across n = 4 biological replicates. PTMs were normalized using heavy peptides standards. (G) Quantification of PTMs on H3 peptides 9–17 associated with DAXX WT or ABM mutant averaged across n = 3 biological replicates. (A, F, and G) Percentages are relative to the total intensity of related peptides and averaged across biological replicates of DAXX purifications. Error bars represent SD. PTMs were normalized using heavy peptides standards, p-values represent unpaired two-sided t tests. See also Table S2.

Figure 7

DAXX adds a de novo H3.3K9me3 deposition pathway to the histone chaperone network During histone supply ASF1 handles H3.1/2/3–H4 dimers and forms several histone-dependent co-chaperone complexes with other histone chaperones, notably ASF1 channels H3 variants toward distinct deposition complexes on chromatin. We identified a new ASF1-centered histone supply pathway to SPT2 that is H3 variant independent, as well as an upstream role for ASF1 in delivering H3.3 histones to DAXX, and an H3.1 variant specificity in TONSL. We found DAXX facilitates the catalysis of H3.3K9me3 through SETDB1 and SUV39H1 methyltransferase recruitment prior to histone deposition on chromatin. DAXX-bound H3.3–H4 recruits SETDB1 and SUV39H1, and the interaction of DAXX with SETDB1 is additionally dependent on SUMOylation. Other factors involved in heterochromatin establishment are differentially dependent on ATRX (e.g., the ChAHP complex) and SUMOylation (e.g., SMCHD1-LRIF1), and we speculate that this represents alternative pathways for H3K9me3 deposition supporting de novo heterochromatin silencing at distinct genomic locations.