Interactome Rewiring Following Pharmacological Targeting of BET Bromodomains

Abstract

Targeting bromodomains (BRDs) of the bromo-and-extra-terminal (BET) family offers opportunities for therapeutic intervention in cancer and other diseases. Here, we profile the interactomes of BRD2, BRD3, BRD4, and BRDT following treatment with the pan-BET BRD inhibitor JQ1, revealing broad rewiring of the interaction landscape, with three distinct classes of behavior for the 603 unique interactors identified. A group of proteins associate in a JQ1-sensitive manner with BET BRDs through canonical and new binding modes, while two classes of extra-terminal (ET)-domain binding motifs mediate acetylation-independent interactions. Last, we identify an unexpected increase in several interactions following JQ1 treatment that define negative functions for BRD3 in the regulation of rRNA synthesis and potentially RNAPII-dependent gene expression that result in decreased cell proliferation. Together, our data highlight the contributions of BET protein modules to their interactomes allowing for a better understanding of pharmacological rewiring in response to JQ1.

Keywords: AP-MS; BET; JQ1; KacY; bromodomain; nucleolus; protein crystallography; proteomic network; rRNA; rewiring.

Graphical abstract

Figure 1

BET Proteins Are Molecular Scaffolds Interacting with Distinct Proteins (A) Modular organization of BET proteins (domain boundaries in amino acids). (B) BETs scaffold transcriptional regulators to acetylated histones. Inset: JQ1 competes with Kac-containing peptides for BRD association. (C) Overview of experimental setup used to quantify the BET interaction network upon JQ1 treatment. (D) Heatmap of BET high-confidence interaction partners identified by AP-MS in the JQ1 time course. See also Figure S1 and Tables S1 and S2.

Figure 2

Pharmacological BRD Inhibition Modulates the BET Interactome (A) Network view of selected protein complexes or families (groups I–XII; names in B, details in Table S2D) associated with BETs. Node size displays the relative abundance in cells untreated or treated with JQ1 for 1 hr. (B) Relative spectral count contributions of individual BET proteins to selected groups or complexes. (C) Dot plots of selected interaction partners associated with individual BETs after JQ1 treatment. See also Figure S1 and Tables S1 and S2.

Figure 3

BET BRDs Initiate Interactions with Non-histone Kac-XX-Kac Peptides (A) Peptide SPOT validation of histone-like peptides containing a Kac-XX-Kac motif. The heatmap shows binding intensities against the first (BD1) and second (BD2) BRDs of BRD4. Peptides exhibiting strong (≥75% of maximum) intensity toward one domain, with a ≥2-fold lower intensity toward the other domain are highlighted. (B) Unique peptides containing K-XX-K motifs found in the human proteome. The inset highlights the binding results from (A) toward BRD4 BRDs. (C) Peptide LOGOs derived from very strong (≥85% of maximum intensity) binding in the SPOT arrays shown in (A). (D) Crystal structure of BRD4/BD1 bound to an E2F1 di-Kac peptide (K117ac-XX-K120ac motif) or the previously published histone H4 K5ac/K8ac peptide (PDB: 3UVW). (E) Structural overlay of BRD4/BD1 complexes with Kac-GX-Kac-bearing peptides shown in cartoon, highlighting the topology of the BRD cavity with respect to the conserved asparagine (N140) and the bulky tryptophan of the WPF shelf (W81). See also Figure S2 and Tables S1, S3, S4, and S5.

Figure 4

Different Modes of BET BRD Recognition of Kac (A–D) Crystal structures of BRD4/BD1 bound to histone H4 (PDB: 3UVW) and indicated peptides derived from SIRT7 (A), di-Kac H3 (B), BAZ1B (C), and SRPK1 (D). The peptide 2Fc-Fo maps contoured at 2σ are shown in the insets. (E) Sedimentation velocity experiments of BET BRDs individually (BD1, BD2), in equimolar mixtures (1:1) or tandem constructs (BD1:2) demonstrating lack of self- or hetero-association. (F) Ab initio shapes of BET tandem BRD constructs restored from SAXS data; mean distances (n = 100) are shown next to the models, in agreement with the hydrodynamic shape calculated using prolate ellipsoid models in (E). See also Figure S3 and Tables S1, S3, S4, and S5.

Figure 5

Contributions of the ET Domain to the BET Interactome (A) Overlap of full-length (FL)-BRD4 interactors (within ∼ ± 2 LFC in spectral count ratio following JQ1 treatment) and BRD4/ET domain highlighting 12 common proteins. (B) Recovery of FLAG-tagged BRD9 from pull-downs with indicated recombinant BRD4 domains. (C) Identification of the BRD9 binding site mediating interactions with BRD4/ET: (i) recovery of MBP-tagged BRD9 fragments with recombinant BRD4/ET domain; (ii) peptide SPOT array of BRD^91–99 blotted against the BRD4/ET domain; (iii) SPOT alanine-scanning of BRD9^20–38 against BRD4/ET. (D) Schematic of BET ET-motif discovery employing AP-MS, SPOT arrays, and alanine scanning. Refined LOGO motifs are shown on the right. The bar charts on the top of each LOGO represent the relative residue contribution to the overall peptide binding following SPOT-ALA scanned array quantifications. (E) Assessment of the behavior in SPOT assays of the indicated motif classes upon polarity reversal of the ET surface (BRD4/ET wt vs mut). (F) Cellular validation of ET-specific interactions using LacO/LacR chromatin immobilization. U-2 OS cells with a stably integrated LacO array were transfected with FL-mCherry-BRD4 (WT or ΔET) and FL-GFP-BRD9 (WT or mutant). See also Figures S4 and S5 and Tables S1, S2, and S3.

Figure 6

BRD Inhibition Modulates BRD3 Localization (A) Still images of indicated GFP-tagged BET constructs in live U-2 OS cells. (B) Average BRD3 WT or (BD1:2)^mut ChIP-seq read counts plotted over genes. TSS, transcription start site; TES, transcription end site. Inset: binding sites detected with each construct. (C) Genome browser tracks showing BRD3 occupancy across the MYC and CCND2 gene loci. y axis: normalized read counts in reads per million per basepair. (D) Schematic representation of a single human rDNA repeat relative to the transcription start site (TSS) of the rDNA repeat (x axis; based on GenBank U13369; SP: spacer promoter; UCE, upstream control element; IGS, intergenic spacer; CP, core promoter). y axis: normalized read counts of BRD3 WT, (BD1:2)^mut, and TCOF1 (from HeLa cells; Calo et al., 2018). (E) GFP-BRD3 ChIP-qPCR to rDNA H0, H1, and H27 (see D) with and without JQ1 for 1 hr. x axis: signal fold enrichment against rabbit IgG isotype control purifications. Data represent the mean ± SEM (n = 3) of two biological replicates. p values were calculated using Student’s t test and are represented by ^∗∗∗p < 0.001; ^∗∗p < 0.01; ns, not significant. See also Figure S6 and Tables S1 and S2.

Figure 7

BRD3 Impacts rRNA Production and Cell Proliferation (A) U-2 OS cells were treated with JQ1 or DMSO for 1 hr and then fed 5-EU for 1 hr prior to staining for BRD3, fibrillarin (FIB), and click chemistry to 5-EU-labeled RNA. (B) Quantitative immunofluorescence of U-2 OS cells treated with various concentrations of tetracycline (to titrate BRD3 levels) or JQ1. 5-EU signal overlapping with fibrillarin signal (i.e., nucleolar RNA) was quantified for >400 cells for each experimental condition. ^∗∗∗p value <0.001, by two-tailed Student’s t test. (C) Cell proliferation assay for U-2 OS treated with tetracycline over 6 days (n = 3). (D) Cell proliferation assay for U-2 OS cells induced with 1 μg/mL tetracycline over 6 days (n = 3). (E) Violin plots of CERES scores showing gene depletion effects from CRISPR-Cas9 loss-of-function screens in 342 cancer cell lines (Meyers et al., 2017). (F) Model of BET protein recruitment functions and the impact of BRD3 overexpression. In proliferating cells, BETs bind to acetylated proteins, including histones through their tandem BRDs, and recruit to chromatin additional transcriptional regulators through other modular domains. High levels of BRD3 antagonize this by decreasing the levels of other BETs, and competing for binding to common loci, in addition to reducing rRNA levels, with a net result of decreasing proliferation. See also Figure S7 and Tables S1 and S2.