Domain-Selective BET Ligands Yield Next-Generation Synthetic Genome Readers/Regulators with Nonidentical Cellular Functions

Abstract

SynTEF1, a prototype synthetic genome reader/regulator (SynGR), was designed to target GAA triplet repeats and restore the expression of frataxin (FXN) in Friedreich's ataxia patients. It achieves this complex task by recruiting BRD4, via a pan-BET ligand (JQ1), to the GAA repeats by using a sequence-selective DNA-binding polyamide. When bound to specific genomic loci in this way, JQ1 functions as a chemical prosthetic for acetyl-lysine residues that are natural targets of the two tandem bromodomains (BD1 and BD2) in bromo- and extra-terminal domain (BET) proteins. As next-generation BET ligands were disclosed, we tested a select set with improved physicochemical, pharmacological, and bromodomain-selective properties as substitutes for JQ1 in the SynGR design. Here, we report two unexpected findings: (1) SynGRs bearing pan-BET or BD2-selective ligands license transcription at the FXN locus, whereas those bearing BD1-selective ligands do not, and (2) rather than being neutral or inhibitory, an untethered BD1-selective ligand (GSK778) substantively enhances the activity of all active SynGRs. The failure of BD1-selective SynGRs to recruit BRD4/BET proteins suggests that rather than functioning as "epigenetic/chromatin mimics," active SynGRs mimic the functions of natural transcription factors in engaging BET proteins through BD2 binding. Moreover, the enhanced activity of SynGRs upon cotreatment with the BD1-selective ligand suggests that natural transcription factors compete for a limited pool of nonchromatin-bound BET proteins, and blocking BD1 directs pan-BET ligands to more effectively engage BD2. Taken together, SynGRs as chemical probes provide unique insights into the molecular recognition principles utilized by natural factors to precisely regulate gene expression, and they guide the design of more sophisticated synthetic gene regulators with greater therapeutic potential.

Figure 1.. Design of next-generation SynGRs for disease-causing GAA repeat expansions within the FXN locus.

(A) A model of the cascade of interactions and reactions initiated by SynGR1 (Syn-TEF1) to restore transcription at the frataxin (FXN) locus in Friedreich’s ataxia cells. Pol II, RNA polymerase. (B) Schematic of the BET family member BRD4 protein and co-crystal structures of BD1 (bronze) and BD2 (silver) domains each bound to JQ1 (PDB: 3MXF, 3ONI). ET, extra-terminal domain; CTD, C-terminal domain. (C) Tanimoto similarity scores of BET ligands used in the construction of SynGRs. (D-K) Chemical and co-crystal structures of BET ligands (green sticks) JQ1 (D), OTX015 (E), I-BET762 (F), I-BET726 (G), SJ432 (H), PLX51107 (I), I-BET151 (J), and GSK778 (K) in complex with either BRD2-BD2 (H) or BRD4-BD1 (all others). PDB: 3MXF, 5WMD, 3P5O, 4BJX, 6DDJ, 5WMG, 3ZYU, and 6SWN, respectively. Magenta arrows indicate linker attachment site for coupling to PA1.

Figure 2.. Binding of PA1 and SynGRs to DNA and bromodomains in vitro.

Structures of PA1-based Syn-GRs and representative SPR sensorgrams for PA1 alone and SynGR1 to SynGR8 measured in the presence of hairpin DNA (hDNA1) with the PA1 target sequence. For each compound (PA1 or SynGR), the sensorgrams represent 3, 6.25, 12.5, 25, 50, 100 and 200 nM (bottom to top) of compound. Solid black lines are best-fit values for the global kinetic fitting of the results with a single-site function. BRD4 K_D values are taken from the literature and represent the unconjugated BET ligand; for OTX015 and I-BET726, binding affinities to individual bromodomains are not available.

Figure 3.. Licensing of FXN transcription by SynGRs in FA cells.

(A) Relative expression of FXN mRNA in GM15850 cells following treatment for 24 hours with 0.1, 0.3 and 1 μM of the indicated SynGR, normalized to 0.1% DMSO solvent control. Error bars are SD of 3 replicates. (B, C) Correlation of BET ligand binding affinity to individual bromodomains BD1 (B) and BD2 (C) with SynGR activity at the FXN locus, following treatment with 1 μM SynGR for 18 h. (D) Relative expression of FXN mRNA in GM15850 cells following treatment at increasing time intervals up to 48 hours with 1 μM of the indicated SynGR, normalized to 0.1% DMSO solvent control. Error bars are SD of 3 replicates. (E) ChIP-qPCR for BRD4, normalized to percent input, at amplicons across the FXN gene in GM15850 cells following treatment for 24h with 1 μM SynGR1, SynGR5, or SynGR8, or 0.1% DMSO. Error bars are SD of 2 replicates. (F) Next generation sequencing of genomic DNA enriched from GM15850 cells treated for 24h with 1 μM SynGR1, SynGR5, or SynGR8, or 0.1% DMSO.

Figure 4.. Co-treatment of SynGRs with BET-binding ligands.

(A) Relative expression of FXN mRNA, normalized to SynGR1 treatment alone, after treatment of GM15850 cells with SynGR1 (1 μM) for 1 h followed by BET-binding ligands JQ1, SJ432, or GSK778 at increasing concentrations for 23 h. Error bars are SD of 3 replicates. (B) Relative expression of FXN mRNA, normalized to 0.1% DMSO control, after treatment of GM15850 cells with each SynGR (1 μM) alone for 24 h (white bars) or each SynGR for 1 h followed by BD2-binding ligand SJ432 (0.5 μM, cyan bars) or BD1-binding ligand GSK778 (0.5 μM, purple bars) for 23 h. Error bars are SD of 3 replicates. (C) ChIP-qPCR for BRD4, normalized to percent input, at amplicons across the FXN gene in GM15850 cells following treatment for 24h with 1 μM of SynGR1 (white bars) or for 1 h with 1 μM of SynGR1 followed by 23 h with 0.5 μM of JQ1, SJ432, or GSK778 (colored bars). Error bars are SD of 2 replicates.

Figure 5.. Models of the BET protein interactions with acetylated client proteins.

(A) Interaction between acetylated chromatin and BD1 of a BET protein. (B) Interaction between acetylated transcription factor and BD2 of a BET protein. (C) BET protein recruitment by SynGRs occurs via binding BD2, indicating that SynGRs behave as synthetic transcription factors, not mimics of epigenetic marks.

Scheme 1.

Synthesis of SynGR1, SynGR3, SynGR4 and SynGR6

Scheme 2.

Synthesis of SynGR2, SynGR5, SynGR7 and SynGR8