High-throughput competitive binding assay for targeting RNA tertiary structures with small molecules: application to pseudoknots and G-quadruplexes

Abstract

There is an indisputable need for new screening methodologies to identify small molecules that target RNA tertiary structures, such as pseudoknots or G-quadruplexes. Here, we present a high-throughput competitive binding antisense assay designed to identify ligands for complex RNA tertiary structures. In this assay, initially customized for the bacterial PreQ1-I riboswitch pseudoknot, ligands compete with a rationally designed quencher-labelled antisense oligonucleotide for binding to a fluorophore-labelled riboswitch. The method is validated for four PreQ1-I riboswitches, using the natural riboswitch ligand PreQ1 and various analogues. A commercial RNA-focused library consisting of ∼15 000 compounds was then screened against the Fusobacterium nucleatum riboswitch, leading to the identification of a promising hit, 4494, which showed competitive binding activity to all PreQ1-I riboswitches and was able to inhibit translation of a riboswitch-regulated reporter gene. Although resynthesis of 4494 revealed that its activity originated from an ∼1% guanine contamination, this result underscores the assay's exceptional sensitivity. To demonstrate its versatility, the assay was tailored for a SARS-CoV-2 G-quadruplex structure and validated with several known G-quadruplex ligands. This work shows that the competitive binding antisense assay is a powerful addition to the RNA-targeting toolbox, facilitating the discovery of ligands for diverse RNA tertiary structures.

Graphical Abstract

Figure 1.

(A) 2D structure of the Fusobacterium nucleatum (Fnu) PreQ₁-I riboswitch in metabolite-free (left) or metabolite-bound (right) conformation. Putative interactions of PreQ₁ with several bases of the riboswitch are shown as dotted lines. (B) 2D structure of the SARS-CoV-2 G-quadruplex GQ-3467. Guanine tetrads are indicated as purple parallel planes. (C) Guanine tetrad molecular structure. The tetrad is stabilized by Hoogsteen hydrogen bonding and a potassium (K⁺) ion. All in all, in this study we present a CB ASsay using antisense RNAs applicable to complex and clinically relevant RNA structures. We initially focused on the therapeutically relevant PreQ₁-I riboswitch pseudoknot, for which specific ligands have been reported, and established a robust workflow suitable for HTS. To demonstrate the assay’s broader applicability, we adapted the methodology to the GQ-3467 structure from SARS-Cov2 and validated it using a panel of well-known GQ ligands. Together, these results provide a detailed framework for the development of CB ASsays targeting diverse RNA tertiary structures, thereby expanding the chemical biology toolbox for RNA-targeted drug discovery.

Figure 2.

Schematic overview of the CB ASsay for pseudoknot (PK) structures. A Cy5-PK is incubated with a potential ligand, after which a DQ-ASO is added. If a small-molecule ligand binds to the PK, the PK is stabilized and hybridization with the DQ-ASO is prevented. In contrast, if the ligand does not bind to the Cy5-PK, the DQ-ASO hybridizes with the Cy5-PK and quenches Cy5 fluorescence signal.

Figure 3.

CB ASsay for the Fnu PreQ₁-I riboswitch. (A) Schematic representation of the equilibrium between the Cy5-labelled Fnu PreQ₁-I riboswitch and IBRQ-ASO with and without ligand binding. The essential nucleotide C17 is encircled. (B) Base pairing of PreQ₁ to C17 (WT) or C17U (Mutant). (C) Molecular structures of PreQ_1, guanine, adenine, 7-carboxy-7-deazaguanine (7c7dag), 7-deazaguanine (7dag), and 2,6-diaminopurine (2, 6dap). CB ASsay for the Fnu WT (D) or Fnu C17U mutant (E) PK with PreQ₁ and its analogues. Dotted lines denote the minimum and maximum Cy5 signal as determined by the positive and negative controls. All measurements were performed in duplicate.

Figure 4.

CB ASsay of PreQ₁ and its analogues for the Tte, Bsu, and Efa WT PK. Dotted lines denote the minimum and maximum Cy5 signal as determined by the positive and negative controls. All measurements were performed in duplicate.

Figure 5.

Hit validation of identified hits. (A) Molecular structure of identified hits. CB ASsay of identified hits for the Fnu WT (B), Fnu mutant PreQ₁-I riboswitch (C), Fnu WT with 200 molar equivalents of yeast tRNA (D), the WT Tte (E), Bsu (F), and Efa (G) PreQ₁-I riboswitches. Dotted lines denote the minimum and maximum Cy5 signal as determined by the positive and negative controls. All measurements were performed in duplicate.

Figure 6.

Schematic overview of the CB ASsay for GQ structures. A dark quencher (DQ)-labelled GQ (DQ-GQ) is incubated with a potential ligand, after which a Cy5-ASO is added. If a small-molecule ligand binds to the GQ, the GQ is stabilized and hybridization with the Cy5-ASO is prevented. In contrast, if the ligand does not bind to the DQ-GQ, the Cy5-ASO hybridizes with the DQ-GQ and quenches Cy5 fluorescence signal.

Figure 7.

CB ASsay for the GQ at position 3467 (GQ-3467) in the SARS-CoV-2 genome. (A) Schematic representation of the equilibrium between the IBRQ DQ-labelled GQ-3467 and the Cy5-labelled ASO with and without ligand binding. Guanine tetrads are indicated in purple parallel planes. (B) Molecular structures of GQ-ligands PhenDC3, TMPyP4, TrisQ, PDS, and BRACO19. (C) CB ASsay of GQ-3467 with the general GQ ligands. Dotted line denotes the maximum Cy5 signal as determined by the positive control. All measurements were performed in duplicate. (D) Quenching analysis of the general GQ ligands in the presence of 50 nM Cy5-labelled ASO.