Structural and mechanistic insights into ATRX-dependent and -independent functions of the histone chaperone DAXX

Abstract

The ATRX-DAXX histone chaperone complex incorporates the histone variant H3.3 at heterochromatic regions in a replication-independent manner. Here, we present a high-resolution x-ray crystal structure of an interaction surface between ATRX and DAXX. We use single amino acid substitutions in DAXX that abrogate formation of the complex to explore ATRX-dependent and ATRX-independent functions of DAXX. We find that the repression of specific murine endogenous retroviruses is dependent on DAXX, but not on ATRX. In support, we reveal the existence of two biochemically distinct DAXX-containing complexes: the ATRX-DAXX complex involved in gene repression and telomere chromatin structure, and a DAXX-SETDB1-KAP1-HDAC1 complex that represses endogenous retroviruses independently of ATRX and H3.3 incorporation into chromatin. We find that histone H3.3 stabilizes DAXX protein levels and can affect DAXX-regulated gene expression without incorporation into nucleosomes. Our study demonstrates a nucleosome-independent function for the H3.3 histone variant.

Fig. 1

Structure and analyses of an ATRX–DAXX interaction surface. a Domain structures of human DAXX (hDAXX) and hATRX. ATRX interacts with the N-terminal four-helix bundle (4HB) in DAXX via a central region. b Alignment showing residues of the highly conserved DAXX-interacting segment of hATRX (DAXX-binding motif, DBM) within the indicated vertebrate species. c Crystal structure showing the intermolecular interactions within the artificial trimer of the hDAXX 4HB–hATRX DBM fusion protein with each monomer highlighted in a different color (left). Ribbon view showing the α2 and α4 helices of hDAXX 4HB bound to the hATRX DBM (right). d–f Details of the interactions between hDAXX 4HB and hATRX DBM. An overview is shown in d. e and f Show zoomed-in views on the contacts between F87 in hDAXX 4HB and I1280 in hATRX DBM, as well as between Y124 in hDAXX 4HB and K1273 in hATRX DBM, respectively. g, h ITC results validating the observed interactions of DAXX 4HB and ATRX DBM in d. Dissociation constants (K _d) of binding are shown. i Pulldown of full-length ATRX from murine embryonic fibroblast (MEF) nuclear extract using GST-tagged mouse DAXX (mDAXX) 4HB (WT and Y130A mutant) as bait. Immunoblot for ATRX and Coomassie stain of GST-tagged proteins in the eluate. j Pulldown of full-length DAXX from MEF nuclear extract using GST-tagged mATRX DBM (WT and L1255A) as bait. Immunoblot for DAXX and Coomassie stain of GST-tagged proteins in the elution are shown

Fig. 2

Analysis of DAXX-dependent and ATRX-dependent gene and retroelement expression in mESCs. a Venn diagram showing the overlap of genes differentially expressed in Daxx knockout (KO) and Atrx KO mESCs derived from publicly available RNA-Seq data (GEO accession no.: GSE73881). The p-value was determined by using hypergeometric distribution. b Validation of selected overlapping differentially expressed genes by RT-qPCR in Atrx ^fl and Atrx ^del mESCs, as well as Daxx ^−/− mESCs, rescued with different Daxx transgenes, as indicated. Data are presented as relative expression and were normalized to Actb. Shown are results from at least three independent experiments (error bars depict S.E.M.). c–f Expression of IAP, MusD and LINE-1 elements determined by RT-qPCR in c Daxx ^−/− mESCs, rescued with Daxx transgenes (WT, F93A, Y130A), in d Atrx ^fl and Atrx ^del mESCs, in e Setdb1 ^fl/fl and 4-hydroxy tamoxifen-(4OHT)treated Setdb1 ^fl/fl mESCs, and in f Kap1 ^fl/fl and Kap1 ^fl/fl +4OHT mESCs. Data are presented as relative expression and were normalized to Actb or Gapdh. Shown are results from at least three independent experiments (error bars depict S.E.M.). Immunoblots to confirm transgene expression or absence of target protein are shown adjacent to each panel. g, h Anti-HA ChIP-qPCR for indicated genomic regions in mESCs with stable H3.3-HA knock-in. g Shows Daxx ^−/− mESCs, rescued with Daxx transgenes (WT, Y130A); h shows Atrx ^fl and Atrx ^del mESCs. Data are presented as per cent input (error bars depict S.E.M.). Shown are the results of two independent experiments. Asterisks in b through h denote significance by Student’s t-test: *p < 0.05, **p < 0.01, ***p < 0.001, n.s. not significant

Fig. 3

DAXX associates with SETDB1–KAP1 co-repressor proteins independently of ATRX. a HeLa nuclear extract containing FLAG-hDAXX was subjected to anti-FLAG IP. Eluates were analyzed by immunoblot for DAXX and ATRX (top). Elutions 2–6 were pooled and subjected to anion exchange chromatography (Mono Q) with a gradient of 150 mM to 1 M KCl (bottom). Shown are immunoblots of Mono Q fractions with antibodies for ATRX, DAXX, SETDB1, and histones H3 and H4. b 293T cells transfected with no construct (control) or FLAG-hSETDB1 were subjected to anti-FLAG IP. c 293T cells transfected with no construct (control) or FLAG-hDAXX were subjected to anti-FLAG IP. Inputs and eluates in b and c were analyzed for SETDB1, DAXX, ATRX, and KAP1 by immunoblot. Ponceau stains show the exogenous FLAG-tagged proteins present in the elution samples. d Immunoblots of input and eluates from FLAG-tagged mDAXX (WT, F93A, and Y130A) and control (no FLAG-tagged transgene) immunoprecipitation from MEF nuclear extract

Fig. 4

H3.3 stabilizes DAXX protein and affects ERV repression independently of its deposition. a Immunoblot for DAXX, ATRX, HIRA, and H3.3 in whole-cell lysates from the noted mESC cell lines, and loading controls anti-Tubulin and anti-H3. b Immunoblot of the indicated Daxx transgenes introduced into Daxx ^−/− mESCs, levels of HA-tagged endogenous H3.3B, and anti-Tubulin and anti-H3 loading controls. c Expression of IAP, MusD, and LINE-1 elements determined by RT-qPCR in Daxx ^−/− mESCs, rescued with Daxx transgenes (WT, L363A–L367A). Data are shown as relative expression and were normalized to Gapdh or Actb. Shown are results from four independent experiments (error bars depict S.E.M.). d Sequence context and location of amino acids L126 and I130 (highlighted in magenta) in H3.3 in a co-crystal structure in complex with DAXX and H4 (PDB ID: 4H9N). e Immunoblot of DAXX, ATRX, and HIRA in H3.3 KO mESCs (control) and cells stably transduced with the indicated H3.3-HA-FLAG transgenes. Immunoblots for DAXX, HIRA, ATRX, Tubulin, HA, H3.3, and H3 are shown. a and e were split from the same original image. The full blot is shown in Supplementary Fig. 10a. The H3.3 antibody was raised against the ‘Ala87-Ala88-Ile89-Gly-90’ motif in H3.3 and does therefore not recognize H3.3 G90M. f Expression of IAP and MusD elements determined by RT-qPCR in H3.3 KO mESCs and cells transduced with the indicated H3.3 transgenes. Data are shown as relative expression and were normalized to Actb from two independent experiments (error bars depict S.E.M.). g Anti-HA ChIP-qPCR at the indicated genomic regions for different H3.3-HA-FLAG transgenes in the background of H3.3 KO mESCs. Data are presented as per cent input (error bars depict S.E.M.). Shown are the results of two independent experiments. Asterisks in c, f and g denote statistical significance as obtained by Student’s t-test: *p < 0.05, **p < 0.01, ***p < 0.001, n.s. not significant

Fig. 5

Analyses of DAXX-dependent and histone H3.3-dependent gene expression in mESCs. a Venn diagram depicting the intersection of differentially expressed genes (greater than twofold change, FDR > 5%) in DAXX WT and H3.3 WT/L16A-I130A compared to the respective KO cell line. Duplicate biological replicates for each cell line were analyzed in the RNA-Seq experiment. The p-values were determined by using hypergeometric distribution. b Heatmap showing gene expression pattern and hierarchical clustering of the indicated mESC lines with respect to genes differentially expressed (greater than twofold change) between mESC Daxx ^−/− and mESC Daxx ^−/−;WT. Biological replicates (columns) are ordered according to hierarchical clustering of differentially expressed genes. The vertical listing of genes was based on descending Z scores for a Daxx ^−/− RNAseq replicate sample. c, d qRT-PCR validation of the RNA-Seq experiment for the indicated genes in Daxx ^−/− mESCs rescued with WT DAXX (c) or in H3.3 KO mESCs rescued with the indicated H3.3 transgenes (d). Data are shown as relative expression and were normalized to Actb. Shown are results from four independent experiments (error bars depict S.E.M.). Asterisks denote statistical significance as obtained by Student’s t-test: *p < 0.05, **p < 0.01, ***p < 0.001, n.s. not significant

Fig. 6

HDAC activity contributes to DAXX-mediated ERV repression. a Immunoblot for DAXX and HDAC1 in input and eluate from immunoprecipitation of FLAG-mDAXX (WT, F93A, and Y130A) and control (no FLAG-tagged transgene) in Daxx ^−/− MEFs. b Expression of IAP and MusD elements determined by RT-qPCR Daxx ^−/− and Daxx ^−/−;WT treated with DMSO control or 5 nM trichostatin A (TSA) for 48 h. Data are presented as relative expression and were normalized to Actb. Shown are results from two independent experiments (error bars depict S.E.M.). Asterisks denote statistical significance as obtained by Student’s t-test: *p < 0.05, **p < 0.01, ***p < 0.001, n.s. not significant. c 293T cells transfected with no construct (control) or FLAG-hDAXX were subjected to anti-FLAG IP. d 293T cells transfected with no construct (control) or FLAG-hSETDB1 were subjected to anti-FLAG IP. Inputs and elutions in c and d were analyzed for SETDB1, DAXX, ATRX, HDAC1, and KAP1 by immunoblot. Ponceau stains show the overexpressed FLAG-tagged proteins present in the elution samples. e Immunoprecipitation for endogenous HDAC1 alongside IgG control was performed in Daxx ^−/− and Daxx ^−/−;WT MEF. Shown are immunoblots for HDAC1 and KAP1 in input and eluates. f At least two biochemically distinct DAXX–H3.3-containing complexes exist in mammalian cell: One complex contains DAXX, ATRX, and H3.3–H4, and deposits H3.3-containing nucleosomes at telomeres, and is involved in regulating pericentric repeats and imprinted genes. The second complex contains DAXX, H3.3–H4, SETDB1, and KAP1 proteins and functions to repress ERVs. The DAXX–SETDB1–KAP1 complex may have evolved other functions in addition to repression of ERVs