Structural insights into the human NuA4/TIP60 acetyltransferase and chromatin remodeling complex

Abstract

The human nucleosome acetyltransferase of histone H4 (NuA4)/Tat-interactive protein, 60 kilodalton (TIP60) coactivator complex, a fusion of the yeast switch/sucrose nonfermentable related 1 (SWR1) and NuA4 complexes, both incorporates the histone variant H2A.Z into nucleosomes and acetylates histones H4, H2A, and H2A.Z to regulate gene expression and maintain genome stability. Our cryo-electron microscopy studies show that, within the NuA4/TIP60 complex, the E1A binding protein P400 (EP400) subunit serves as a scaffold holding the different functional modules in specific positions, creating a distinct arrangement of the actin-related protein (ARP) module. EP400 interacts with the transformation/transcription domain-associated protein (TRRAP) subunit by using a footprint that overlaps with that of the Spt-Ada-Gcn5 acetyltransferase (SAGA) complex, preventing the formation of a hybrid complex. Loss of the TRRAP subunit leads to mislocalization of NuA4/TIP60, resulting in the redistribution of H2A.Z and its acetylation across the genome, emphasizing the dual functionality of NuA4/TIP60 as a single macromolecular assembly.

Fig. 1.. Overall structure of the human NuA4/TIP60 complex.

(A) Domain architecture of individual subunits within the TIP60 complex that were visualized in this study, with regions modeled shown in color and unmodeled regions shown in white. Residue numbers at boundary regions are indicated. The visualized subunits are associated into three main modules, indicated by the dashed boxes. The TRRAP module comprises the TRRAP subunit and the SANT and HD domains of EP400. The P400 subcomplex consists of two modules: ARP and BASE. The ARP module includes EPC1, DMAP1, BAF53a, and ACTB as well as the pre-HSA, HSA, and PMD domains of P400. The BASE module consists of three copies each of RUVBL1 and RUVBL2 as well as one each of VPS72, BAF53a, H2AZ, H2B, and the Insertion and Motor domains of EP400. The same color code for the subunits is used throughout the figures. (B) Silver-stained gel of native TIP60 complex purified from K562 cells expressing EPC1–3xFlag-2xStrep (tandem affinity purification from nuclear extracts). Subunits are identified on the right. (C) Histone exchange assay carried out with purified native TIP60 by using recombinant dinucleosomes on beads and added H2A.Z-H2B dimers. Efficient H2A.Z incorporation was detected by immunoblotting in the presence of the complex, ATP, and acetyl-CoA when indicated. H3 and H2B signals served as controls to show the amount of nucleosomes present in the reactions. (D) The HAT assay was performed on nucleosomes BY using native TIP60 in the presence of Acetyl-CoA and ATP when indicated. HAT activity was assessed with a H4K5ac antibody, with the H3 signal serving as loading control. (E) Front view of the cryo-EM map of endogenous TIP60, with the subunits colored as depicted in (A). Structures of the P400 subcomplex (a composite map of the ARP module, BASE module, and VPS72-H2AZ-H2B) and the TRRAP module. (F) Localization of the different regions of EP400 within the TIP60 complex. EP400 and EPC1 are depicted in hot pink and yellow, respectively, whereas the remaining subunits are shown in gray. The TINTIN and HAT modules are flexibly attached to the rest of the complex and were not visible in our reconstruction (except for part of EPC1 that connects the HAT module to the rest of the complex). Light gray cartoons are used as a schematic indication of each module and provide an estimated localization rather than a precise structural depiction (same for Fig. 2A).

Fig. 2.. Structure of the TIP60 ARP module.

(A) (Left) The ARP module (turquoise) is connected to all other modules within the TIP60 complex. (right) Front and back views of the ARP module are shown in ribbon diagram with subunits colored as in Fig. 1A. (B) A positively charged segment of EPC1 (shown as a yellow ribbon) inserts into a negatively charged cavity within the ARP module (rendered as a surface with electrostatics), tethering the HAT module to the rest of the complex. Four positively charged residues (R457, R458, R461, and R464) involved in the interaction are depicted in stick representation and labeled. (C to E) Detailed views of the interactions between (C) the pre-HSA domain of EP400 and DMAP1, (D) the pre-HSA domain and the PMD of EP400, and (E) the SANT domain of DMAP1 and BAF53a. The residues participating in the interactions are shown in stick representation and labeled. Model building was based on the cryo-EM map of ARP module (fig. S3A). (F) Biochemical data showing that the N-terminal amino acids up to R464 of EPC1 are sufficient to tether the HAT module to the rest of the TIP60 complex. Cells expressing EPC1(1–464)-3xFlag-2xStrep from the AAVS1 safe harbor were used for tandem affinity purification, analyzed by immunoblotting, and compared with a mock cell line processed in parallel. See fig. S10 for silver-stained gel and MS analysis. (G) Biochemical data showing the critical role of the DMAP1 SANT domain and key amino acid D190 in the association of this protein with the TIP60 and SRCAP complexes. Cells expressing either WT DMAP1–3xFlag-2xStrep from the AAVS1 safe harbor, mutant of amino acid 190 (D190A/L191A/K192A), or a mutant form lacking the SANT domain (amino acids 149 to 199; ΔSANT) were used for tandem affinity purification, analyzed by immunoblotting, and compared to a mock cell line processed in parallel. See fig. S13 for silver-stained gel and MS analysis. (H) Comparison of domain architecture for the human TIP60 and the yeast NuA4 complexes, aligned on their ARP domains. The position of the Tra1 subunit in NuA4 (the homolog of human TRRAP) overlaps with that of the BASE module in TIP60. Single-letter abbreviations for the amino acid residues referenced throughout this paper are as follows: R, Arg; H, His; D, Asp; K, Lys; Y, Tyr; I, Ile; L, Leu; F, Phe; E, Glu; V, Val; A, Ala; Q, Gln; P, Pro.

Fig. 3.. Architecture of the P400 subcomplex and comparison with other INO80 family remodelers.

(A) SDS-PAGE gel of the reconstituted P400 subcomplex, with subunits indicated on the right. (B) Side view of the cryo-EM reconstruction of reconstituted P400 subcomplex (overall resolution, 2.6 Å). The connection region between the ARP and BASE modules is highlighted by the red dash box. (C) Zoomed-in view of the interaction region between the ARP and BASE modules. (Top) Ribbon diagram with residues involved in salt bridges and hydrogen bonds between EP400 (hot pink), VPS72 (cyan), and DMAP1 (dark blue), explicitly depicted. (Bottom) Hydrophobic interactions of EP400 (hot pink ribbon) with VPS72 and DMAP1, which are shown as a colored surface depicting hydrophobicity (green, hydrophilic; white, hydrophobic). (D) Side-by-side structural comparison between the P400 subcomplex described in this work (left), the SWR1-NCP complex (middle; PDB: 6GEN) and the SRCAP-NCP complex (right; PDB: 8X19). The motors of these three chromatin remodelers (hot pink) are positioned to be approximately aligned with each other [and to (B)]. Only the ARP module (turquoise), Arp6 (army green), AAA-ATPase hexamer (gray), motor and HSA domains of the remodeler (hot pink), BAF53a (brunt orange), VPS72 C terminus (cyan), ZNHIT (light purple), and ACTB (tan) are shown for clarity. The nucleosome engaged by SWR1 and SRCAP is shown for reference in gray ribbon.

Fig. 4.. The SANT and HD domains of EP400 tether the TRRAP subunit to the rest of the complex.

(A) Cryo-EM density map for the TRRAP module. The HEAT, FAT, and PIKK-like domains of TRRAP are depicted in light gray, steel blue, and dark gray, respectively, and EP400 segments are shown in hot pink. (B) Structural comparison of the TRRAP module of TIP60 (left) with human SAGA (right; PDB: 7KTR), showing TRRAP in gray and interacting elements from either complex in hot pink. The overlapping binding surfaces are highlighted by orange and aqua dashed boxes. (C) Detailed view of the interaction between the SANT domain of EP400 and TRRAP. TRRAP is shown as an electrostatic surface, with three distinct charged patches involved in EP400 interaction marked by dashed ovals. Key amino acids in the SANT domain participating in the interaction are shown in stick representation and labeled. (D) Close-up view of the interactions between the HD domain of EP400 and TRRAP. The surface representation of TRRAP shows hydrophilic regions depicted in green, and hydrophobic, in white. Residues of the HD domain of EP400 involved in hydrophobic interactions are displayed in stick representation and labeled. (E) Biochemical data showing the critical role of the SANT and HD domains of EP400 in the association of the TRRAP module within the rest of the TIP60 complex. Cells expressing either WT EP400–3xFlag-2xStrep from the AAVS1 safe harbor or a mutant form lacking the SANT and HD domains (amino acids 2361 to 2530) were extracted followed by anti-Flag immunoprecipitation and Flag peptide elution and analysis by immunoblotting.

Fig. 5.. TIP60-specific function of TRRAP in vivo.

(A) TSS-proximal H2A.Zac signal defined as the peaks located near an annotated TSS (TSS ± 2 Kb) and represented by a heatmap (left; peak center ± 5 Kb) and a box plot (right), which show regions with significant decrease (Down), significant increase (Up), or no significant change (Unchanged) in H2A.Zac signal. (B) Enhancer-proximal H2A.Zac signal defined as the peaks located near an annotated enhancer (enhancer ± 2 Kb) and represented by a heatmap (left; peak center ± 5 Kb) and a box plot (right), showing regions with significant decrease, significant increase, or no significant change in H2A.Zac signal. (C) Box plot representation of differential gene expression based on RNA-seq in the DTRRAP mutant versus the WT condition, showing normalized transcript counts on genes that are significantly down-regulated, significantly up-regulated, or with no significant change. (D) Heatmap and box plot representation of H2A.Zac ChIP-seq signal for peaks that are located near the TSS (± 2 Kb) of genes identified in (C). (E) Genome browser view of normalized read counts of RNA-seq, H2A.Zac ChIP-seq, and H2A.Z ChIP-seq over the gene CCND1. (F) Genome browser view of normalized read counts of RNA-seq, H2A.Zac ChIP-seq, and H2A.Z ChIP-seq over the gene DACH1. (G) ChIP-qPCR of FLAG-tagged EP400 showing occupancy at the promoters of DACH1 and CCND1 in ΔTRRAP mutant versus WT cells [error bars are standard errors from three biological replicates (n = 3); *P < 0.05; **P < 0.01]. ChIP-seq data are merged from two biological replicates (n = 2). RNA-seq data are from three biological replicates (n = 3). A two-sided Mann-Whitney U rank sum test was performed on all the box plots, and their P values are indicated in table S3.