SILCS-RNA: Toward a Structure-Based Drug Design Approach for Targeting RNAs with Small Molecules

Abstract

RNA molecules can act as potential drug targets in different diseases, as their dysregulated expression or misfolding can alter various cellular processes. Noncoding RNAs account for ∼70% of the human genome, and these molecules can have complex tertiary structures that present a great opportunity for targeting by small molecules. In the present study, the site identification by ligand competitive saturation (SILCS) computational approach is extended to target RNA, termed SILCS-RNA. Extensions to the method include an enhanced oscillating excess chemical potential protocol for the grand canonical Monte Carlo calculations and individual simulations of the neutral and charged solutes from which the SILCS functional group affinity maps (FragMaps) are calculated for subsequent binding site identification and docking calculations. The method is developed and evaluated against seven RNA targets and their reported small molecule ligands. SILCS-RNA provides a detailed characterization of the functional group affinity pattern in the small molecule binding sites, recapitulating the types of functional groups present in the ligands. The developed method is also shown to be useful for identification of new potential binding sites and identifying ligand moieties that contribute to binding, granular information that can facilitate ligand design. However, limitations in the method are evident including the ability to map the regions of binding sites occupied by ligand phosphate moieties and to fully account for the wide range of conformational heterogeneity in RNA associated with binding of different small molecules, emphasizing inherent challenges associated with applying computer-aided drug design methods to RNA. While limitations are present, the current study indicates how the SILCS-RNA approach may enhance drug discovery efforts targeting RNAs with small molecules.

Figure 1.

Process flow diagram for the SILCS-RNA protocol. Three-stage GCMC refilling is applied to the neutral (N) solute SILCS simulations, and two-stage filling is applied to the charged (C) solute SILCS simulations.

Figure 2.

Neutral and charged solutes used in the SILCS-RNA approach. On the right, a description of different atoms from the solutes and the major functional groups represented by the respective atoms as used to define the FragMaps. The MAMC and ACEO maps are not used during the SILCS-MC docking to avoid overcounting. MAMC labeled atoms actually use the MAMN maps in specific cases with connected N+ atoms classified as NCLA (e.g., tpp001). ACEO maps are only used for visualization.

Figure 3.

Conformational diversity on RNAs based on the available experimental crystal structures (Table 1). Gray spheres represent the experimentally identified ligands on each RNA. (A) FMN riboswitch structures aligned to PDB 5KX9. (B) TPP riboswitch structures aligned to PDB 2GDI. (C) THF riboswitch structures aligned to PDB 4LVV. (D) dG riboswitch structures aligned to PDB 3SKI. € HIV-1 TAR RNA structures aligned to PDB 6CMN. (F) HCV IRES IIa structures aligned to PDB 3TZR. (G) IAV promoter structures aligned to PDB 2LWK.

Figure 4.

Average number of MAMY fragments present in the SILCS simulations as a function of the GCMC-MD cycles for (A) the FMN riboswitch and (B) the HCV-IRES-IIa RNA under different ion-atmosphere conditions. The error bars represent the standard deviation across 10 independent runs.

Figure 5.

SILCS FragMaps for (A) FMN riboswitch (B) HCV-IRES-IIa RNA. The FragMaps are shown as colored mesh: generic apolar (light green, GFE level −0.5 kcal/mol), generic donor (blue, GFE level −0.5 kcal/mol), generic acceptor (red, GFE level −0.5 kcal/mol), methanol oxygen (light blue, GFE level −0.5 kcal/mol) and methylammonium ion (cyan, GFE level −2.0 kcal/mol). The sand-colored transparent surface represents the SILCS exclusion map. The flavin mononucleotide ligand (fmn001) and ISIS-11-b ligand (hcv001) are shown in the binding sites of the respective RNAs (yellow sticks) identified with red dotted circles.

Figure 6.

RNA intercalation site SILCS FragMaps for (I) TPP riboswitch and (II) HCV-IRES-IIa RNA. (A) Generic apolar maps in light green mesh, (B) generic donor maps in blue mesh, (C) generic acceptor maps in red mesh, (D) alcohol (MEOO) maps in light blue mesh, and (E) positive methylammonium nitrogen (MAMN) maps in cyan mesh with negative acetate oxygen (ACEO) maps in orange mesh.

Figure 7.

RNA structure pocket site SILCS FragMaps for (I) FMN riboswitch, (II) GUA riboswitch, and (III/IV) THF riboswitch binding sites B1 and B2, respectively. (A) Generic apolar maps in light green mesh, (B) generic donor maps in blue mesh, (C) generic acceptor maps in red mesh, (D) alcohol (MEOO) maps in light blue mesh, and (E) positive methylammonium nitrogen (MAMN) maps in cyan mesh with negative acetate oxygen (ACEO) maps in orange mesh.

Figure 8.

RNA minor and major groove site SILCS FragMaps for the HIV1-TAR sites (I) B1 with ligand rbt203(tar008), (II) B2 with ligand neomycin-B(tar010), (III) B3 with ligand mv2003(tar012) and (IV) for the IAV RNA promotor. (A) Generic apolar maps in light green mesh, (B) generic donor maps in blue mesh, (C) generic acceptor maps in red mesh, (D) alcohol (MEOO) maps in light blue mesh, and (E) positive methylammonium nitrogen (MAMN) maps in cyan mesh with negative acetate oxygen (ACEO) maps in orange mesh. RNA molecules are shown in cartoon form and ligand molecules are shown as sticks.

Figure 9.

Correlation between experimental ΔG_bind and LGFE scores predicted using SILCS-MC calculations with generic ACS (GAS21) for all of the RNA targets.

Figure 10:

Atomic and group GFE contributions to the total LGFE for benzimidazole derivatives shown to bind HCV-IRES-IIa RNA. (A) hcv007, (B) hcv003, (C) hcv002. The FragMaps in the top panels show generic apolar (light green mesh), generic donor (blue mesh), and generic acceptor (red mesh) maps at GFE level −1.2 kcal/mol. The ligand atoms carbon, nitrogen, oxygen, and hydrogen are shown as yellow, blue, red, and white colored spheres, respectively. The HCV RNA is shown in transparent cartoon form. SILCS-MC docking was performed with the GAS21 using a 2 Å radius. Values in parentheses refer to experimental binding affinities.

Figure 11.

SILCS-Hotspots analysis for each RNA with the ASTEX mini fragment database. RNA is shown in cartoon form, the ligands from experimental structures are shown as sticks, and the hotspot locations are shown as spheres. The color scheme for the hotspots corresponds to a red–white–blue spectrum with red being highly favorable and blue being less favorable. The dashed red circles highlight the hotspots in the space around the known binding sites for the respective RNAs. The hotspots (HSs) selected for docking of FDA approved compounds are pointed out with hotspots in known binding sites highlighted with yellow.