Fig 1.
(A) Radial fingerprint projection over the PC1 and PC2 subspace. Inactive (red), inconclusive (orange), and active molecules (blue) are scattered over the subspace. White dots represent previously reported bioactive molecules [13,14,26]. (B) The observed frequencies of the main physicochemical property distribution characteristics in the chemical dataset: molecular weight (MW), hydrogen bond acceptors (HBAs), hydrogen bond donors (HBDs), formal charge, topological surface area (TPSA), and xlogP.
Fig 2.
(A) Distribution of the principal physicochemical properties in our in-house chemical library containing more than 300 compounds: molecular weight (MW), hydrogen bond acceptors (HBAs), hydrogen bond donors (HBDs), formal charge, topological surface area (TPSA), and xlogP. (B) The most commonly observed fragments were ordered by their activity-adjusted frequency values (see Methods for a detailed description).
Fig 3.
The molecular structure of the compounds we considered, based on the pyrido[2,3-d]pyrimidine scaffold (compound family 1).
This family is divided into dimers (1–1 to 1–3) and monomers (1–4 to 1–14).
Fig 4.
The molecular structure of pentamidine like compounds we considered (compound family 2).
Fig 5.
(A) Schematic representation of the r(CUG)3 model system. (B) Druggability analysis description of the S1 and S2 model systems. RNA is shown as a surface representation with C•G and G•C pairs in white and U-U pairs in green; three druggability solutions were obtained per system. Each druggable region, or hotspot, is represented by a colored sphere (red to blue, from lowest to highest binding-energy, respectively). Notice that the druggable sites in model S1 are distributed along the major groove but are mainly located in the U-U pairs. Model S2 had a stacking interaction pattern caused by the imidazole fragment, which stacks via one U-U pair and forms an H-bond with the O4 atom of U14 (3.0 Å). Some features can also be observed along the minor groove.
Fig 6.
(A) LigandRNA score reported for compound families 1 and 2. Pentamidine and Hoechst 33258 reference energies are included as red and grey lines, respectively. (B) Superposition of a pyrido[2,3-d]pyrimidine fragment and the binding hotspots (model B, solution 2) predicted with the druggability analysis. (C) Hypothesized interaction between a pyrido[2,3-d]pyrimidine subunit and a uracil residue.
Fig 7.
In vitro assays to evaluate the CUG-binding potential of compounds 1–3 and 2–5.
Fluorescence polarization assays with the indicated concentrations of (A) four substituted pyrido[2,3-d]pyrimidines (compound family 1) and (B) four pentamidine-like compounds (compound family 2). Data were normalized to 100 μM pentamidine as a positive control (marked as a red line). Steady-state fluorescence emission profile for CUGexp-TO (red line) and changes in the emission intensity after addition of increasing amounts of (C) 1–3 and (D) 2–5 recorded in buffered NaCAC 50mM aqueous solution (pH = 7.4) at 298.1 ± 0.1 K. [TO] = 0.5 μM, [RNA] = 0.25 μM, [Ligand] = 0.1 mM.λexc = 485 nm. (E) Ligand-CUGexp binding constants determined by displacement assays with TO. aValues in parenthesis are standard deviations in the last significant figure.
Fig 8.
Quantification of the number of foci present after treatment with compounds 1–3 and 2–5.
(A) Quantification of foci per cell from the analysis of a minimum of 1000 DM1 fibroblast nuclei in a range of four log dilutions starting at 40 μM for 2–5 and 100 μM for 1–3. The results were normalized to DM1 cells treated with 1% DMSO (marked as a red line). (B) Representative fluorescence in situ hybridization (FISH) images showing foci in DM1 fibroblasts after 48 h treatment with a range of four log dilutions starting at 40 μM for 2–5 and 100 μM for 1–3. 1% DMSO and 4 μM ChA3 were used as a negative and positive control, respectively. Nuclei were stained with Hoechst 33342 (blue) and rCUGexp RNA foci were detected with a Cy3-labelled probe (red). Images included 2x enlargement of selected nucleus.
Fig 9.
Subcellular distribution of MBNL1 and in vivo evaluation of therapeutic potential.
(A) Confocal images of co-localization of MBNL1 (green) and foci (red) in DM1 myoblasts. Cells were treated with compounds for 48 h; nuclei were counter-stained with DAPI (blue). 1% DMSO-treated DM1 myoblasts and normal myoblasts were used as controls. Images include 2x enlargement of selected nucleus. (B) Climbing assay performed on 30 adult male flies fed for five days with the indicated compound. Climbing velocity was significantly higher in flies receiving 40 μM compound 2–5 or 100 μM compound 1–3 compared to DM1-model flies receiving only 1% DMSO solvent carrier. The average climbing velocity for wild type flies is marked as a red line. Statistics were calculated using the Student's t-test (**p < 0.01, ***p < 0.001).