Cancer-associated polybromo-1 bromodomain 4 missense variants variably impact bromodomain ligand binding and cell growth suppression

Abstract

The polybromo, brahma-related gene 1-associated factors (PBAF) chromatin remodeling complex subunit polybromo-1 (PBRM1) contains six bromodomains that recognize and bind acetylated lysine residues on histone tails and other nuclear proteins. PBRM1 bromodomains thus provide a link between epigenetic posttranslational modifications and PBAF modulation of chromatin accessibility and transcription. As a putative tumor suppressor in several cancers, PBRM1 protein expression is often abrogated by truncations and deletions. However, ∼33% of PBRM1 mutations in cancer are missense and cluster within its bromodomains. Such mutations may generate full-length PBRM1 variant proteins with undetermined structural and functional characteristics. Here, we employed computational, biophysical, and cellular assays to interrogate the effects of PBRM1 bromodomain missense variants on bromodomain stability and function. Since mutations in the fourth bromodomain of PBRM1 (PBRM1-BD4) comprise nearly 20% of all cancer-associated PBRM1 missense mutations, we focused our analysis on PBRM1-BD4 missense protein variants. Selecting 16 potentially deleterious PBRM1-BD4 missense protein variants for further study based on high residue mutational frequency and/or conservation, we show that cancer-associated PBRM1-BD4 missense variants exhibit varied bromodomain stability and ability to bind acetylated histones. Our results demonstrate the effectiveness of identifying the unique impacts of individual PBRM1-BD4 missense variants on protein structure and function, based on affected residue location within the bromodomain. This knowledge provides a foundation for drawing correlations between specific cancer-associated PBRM1 missense variants and distinct alterations in PBRM1 function, informing future cancer personalized medicine approaches.

Keywords: bromodomains; cancer biology; cancer mutations; chromatin remodeling; epigenetics; function; genomics; histone acetylation; protein stability; site-directed mutagenesis; structure.

Figure 1

PBRM1 incurs frequent missense mutations in the context of cancer.A, proportion of cancer-associated PBRM1 variants by mutation type, as annotated in the COSMIC (Catalog of Somatic Mutations in Cancer) database. B, percentage of cancer-associated missense mutations of PBRM1 by functional domain, as annotated in the COSMIC database. C, cancer-associated missense mutations of PBRM1-BD4 per residue across the entire peptide sequence, as annotated in the COSMIC database. Bromodomains (BD; blue), bromo-adjacent homology (BAH; red), and high-mobility group (HMG; green) domains are denoted, with domain boundaries determined from Pfam annotations ± 15 aa residues. D, structure-based sequence alignment of PBRM1 bromodomains, with the position of the four bromodomain ⍺-helices (blue) shown above. BD4 and the residues studied herein are highlighted within the sequence alignment. The heat map demonstrates the conservation level per residue across the six PBRM1 bromodomains, where higher conservation is indicated by reds and lower conservation is indicated by blues. E, Rosetta flexible peptide docking of an H3K14ac peptide (gray) from PBRM1-BD2 (PDB ID 2KTB) (79) to PBRM1-BD4 (PDB ID 3TLP) (9); mutated residues are represented as spheres and color-coded by the number of unique missense variants per residue examined in this study. H3K14ac, lysine-14 acetylation on histone H3; PBRM1, polybromo-1.

Figure 2

Cancer-associated PBRM1-BD4 missense variants primarily exhibit decreased protein stability with intact secondary and tertiary structure integrity.A, PBRM1-BD4 missense variant T_m determined by SYPRO Orange thermal shift assay (controls shown in light gray, cancer-associated PBRM1-BD4 missense variants in dark gray); ΔT_m of PBRM1-BD4 variants compared to WT is also demonstrated (negative ΔT_m denoted in blue, positive ΔT_m in red), where error bars represent SD; n = 9 for PBRM1-BD4 WT and all missense variants except N601K, where n = 6. B, heat map indicates a change in Gibbs free energy (ΔΔG) of PBRM1-BD4 missense variants compared to WT estimated by Rosetta modeling software (more divergent values shown in blue, less divergent in red). C, correlation of SYPRO Orange thermal shift assay and ΔΔG datasets, where horizontal error bars represent SD of protein melting temperatures determined by the SYPRO Orange thermal shift assay. D, CD spectrum of PBRM1-BD4 missense variant R540T. E, CD spectrum of PBRM1-BD4 missense variant R576P. F, ¹H-NMR spectra of PBRM1-BD4 WT and cancer-associated PBRM1-BD4 missense variants. The gray highlighted regions correspond to the spectral regions (backbone amide proton ∼6.5–9.5 ppm; saturated alkane methyl proton ∼0–1.25 ppm) used to assess variant tertiary structural integrity (56, 57). BD4, fourth bromodomain; CD, circular dichroism; ¹H-NMR, one-dimensional proton NMR spectroscopy; PBRM1, polybromo-1; T_m, protein melting temperature.

Figure 3

Cancer-associated PBRM1-BD4 missense variants exhibit variable ligand binding capacity.A, heat map indicates PBRM1-BD4 variant (0.1–10 μM) binding of H3K14ac peptide (50 nM) compared to WT in AlphaScreen assays (n = 3). B, AlphaScreen titrations of PBRM1-BD4 P556S, M586I, and R576L (0.001–2 μM) against a biotinylated histone H3K14ac (50 nM) peptide (n = 3), where error bars represent SEM. C, EMSA of PBRM1-BD4 WT (0–50 μM) binding to 150 nM Widom 601 DNA (representative of n = 2). D, EMSA of PBRM1-BD4 variants (20 μM) binding to 150 nM Widom 601 DNA (representative of n = 2). Alpha, amplified luminescent proximity homogeneous assay; BD4, fourth bromodomain; EMSA, electrophoretic mobility shift assay; H3K14ac, lysine-14 acetylation on histone H3; PBRM1, polybromo-1.

Figure 4

Structural insights into the functional effects of cancer-associated PBRM1-BD4 missense variants.A, conserved residue Y580 stabilizes the loop-helix fold between the AB loop and the adjacent α_B helix. B, frequently mutated residue R576 helps maintain the structural integrity of the α_Z helix. C, frequently mutated residues M523 and M586 in the α_Z and α_B helices contribute to the stability of the PBRM1-BD4 α-helical core. D, conserved residue R540 contributes to the histone Kac binding pocket and adjacent α_C helix stability. BD4, fourth bromodomain; PBRM1, polybromo-1.

Figure 5

Cancer-associated PBRM1-BD4 missense variants exhibit impaired stability and acetylated histone binding.A, immunoblot demonstrates equivalent expression of full-length PBRM1-BD4 missense variants in a Caki-2 tetracycline-inducible system. B, coimmunoprecipitation with BRG1 and V5-tagged PBRM1. C, immunoblots of V5-tagged PBRM1-BD4 WT and PBRM1-BD4 missense variants and beta-actin from Caki-2 cells treated with 100 μg/ml cycloheximide for 0, 2, 6, 10, or 24 h. D, immunoblot densitometry quantitation of PBRM1-BD4 WT and PBRM1-BD4 missense variants at 2 h of cycloheximide treatment. Significance was calculated using an unpaired Student’s t test, where ∗p < 0.05, and error bars represent the SEM. E, correlation of PBRM1-BD4 WT and PBRM1-BD4 missense variant protein stability as assessed by the cellular cycloheximide chase assay at 2 h and the biophysical SYPRO Orange thermal shift assay, where horizontal error bars represent SD of immunoblot densitometry quantitation at 2 h of cycloheximide treatment and vertical error bars represent SD of protein T_m values determined by the SYPRO Orange thermal shift assay. F, acetylated histone H3 peptide pulldown (n = 2) by PBRM1 WT and PBRM1-BD4 missense variants as measured by fold enrichment of H3K14,18,23,27ac(1–30) over input. Significance was calculated using an unpaired Student’s t test where ∗p < 0.05 and error bars represent SEM. BD4, fourth bromodomain; BRG1, brahma-related gene 1; PBRM1, polybromo-1; T_m, protein melting temperature.

Figure 6

Cancer-associated PBRM1-BD4 missense variants exhibit impaired cancer cell growth suppression and PBRM1 target gene regulation.A, workflow of the FACS cell proliferation competition assay. B, Caki-2 PBRM1 WT and PBRM1-BD4 missense variant cell growth over 22 days. Significant differences between cell lines were calculated using a two-way ANOVA (mixed model) with Tukey post hoc analysis where ∗p < 0.05 and ∗∗p < 0.01, and colored wedges around trendlines represent SD (n = 2 for vector, n = 4 for WT, n = 3 for missense variants). C, RT-qPCR (n = 3) of select PBRM1 target and nontarget genes. Significance was calculated using an unpaired Student’s t test where ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001 and error bars represent the SEM. BD4, fourth bromodomain; FACS, fluorescence-activated cell sorting; PBRM1, polybromo-1; RT-qPCR, quantitative real-time PCR.