Precise small-molecule recognition of a toxic CUG RNA repeat expansion

Abstract

Excluding the ribosome and riboswitches, developing small molecules that selectively target RNA is a longstanding problem in chemical biology. A typical cellular RNA is difficult to target because it has little tertiary, but abundant secondary structure. We designed allele-selective compounds that target such an RNA, the toxic noncoding repeat expansion (r(CUG)^exp) that causes myotonic dystrophy type 1 (DM1). We developed several strategies to generate allele-selective small molecules, including non-covalent binding, covalent binding, cleavage and on-site probe synthesis. Covalent binding and cleavage enabled target profiling in cells derived from individuals with DM1, showing precise recognition of r(CUG)^exp. In the on-site probe synthesis approach, small molecules bound adjacent sites in r(CUG)^exp and reacted to afford picomolar inhibitors via a proximity-based click reaction only in DM1-affected cells. We expanded this approach to image r(CUG)^exp in its natural context.

Figure 1

DM1 is caused by a toxic gain of function by r(CUG)^exp. We designed small molecules that ameliorate disease-associated phenotypes and that can be used to assess target selectivity in DM1 cells. (a) Structure of 1, a non-covalent binding compound. Multivalent small molecules are represented by purple spheres (RNA-binding modules) connected by a line (N-methyl peptide scaffold). Also shown is the secondary structure of r(CUG)^exp and binding of MBNL1, which causes disease. Release of MBNL1 by small molecules improved DM1-associated defects. (b) 1 rescued MBNL1 exon 5 splicing defects in DM1-affected cells (n = 6, 6 biological replicates, 2 replicate experiments). (c) 1 reduced the number of r(CUG)^exp nuclear foci in DM1 cells (n = 100 cells, 5 biological replicates, 2 replicate experiments). (d) Representative images from RNA-FISH experiments to assess formation of nuclear foci. Scale bars represent 5 μM. Data represent mean values ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001, as determined by a two-tailed Student t test.

Figure 2

(a) Structure of 2; purple spheres indicate the RNA-binding modules, the blue triangle is the biotin purification tag and the gray X is the cross-linking module chlorambucil. (b) Treatment of 2 in DM1-affected cells rescued MBNL1 pre-mRNA splicing (n = 6, 6 biological replicates, 2 replicate experiments). (c) 2 reduced the number of nuclear foci in DM1 cells (n = 100 cells, 5 biological replicates, 2 replicate experiments). (d) 2 reacted selectively with r(CUG)^exp-containing DMPK mRNA in DM1-affected cells. SUPT20HL1, SORCS2 and SCUBE2 mRNAs were not detected in the pulled-down fraction. Enrichment in wild-type DMPK was not observed in non-DM1 cells. Enrichment was calculated by normalizing to the least-abundant mRNA in each cell type (six biological replicates each with three technical replicates, two replicate experiments). (e) Chem-CLIP-Map defined the sites of reaction of 2 in DMPK mRNA in DM1-affected cells. RNase H digestion of the DMPK mRNA followed by capture and RT-qPCR revealed that the molecule reacted with the 3′ UTR fragment that contained r(CUG)^exp. Primer set locations are indicated with A, B, C and D (n = 6, 6 biological replicates each with 3 technical replicates, 2 replicate experiments). Data represent mean values ± s.e.m. *P < 0.05, **P < 0.01, as determined by a two-tailed Student t test.

Figure 3

(a) Structure of 3; purple spheres indicate the RNA-binding modules and the orange ‘Pac-Man’ shape represents the bleomycin-cleaving module. (b) Evaluation of the cleavage-based approach in DM1-affected cells revealed that 3 cleaved r(CUG)^exp DMPK mRNA. A competitive experiment with 3 and excess 1 reduced the amount DMPK cleavage, indicative of competitive binding. The orange Pac-Man represents bleomycin (n = 6, 6 biological replicates each with 3 technical replicates, 2 replicate experiments). Relative abundance was determined by normalizing to GAPDH. (c) qPCR analysis of short r(CUG)-repeat-containing mRNAs revealed that 3 selectively cleaved r(CUG)^exp found in DM1 DMPK (n = 6, 6 biological replicates each with 3 technical replicates, 2 replicate experiments). (d) 3 rescued MBNL1 pre-mRNA splicing defects in DM1 cells (n = 6, 6 biological replicates, 2 replicate experiments). Data represent mean values ± s.e.m. *P < 0.05, **P < 0.01, as determined by a two-tailed Student t test.

Figure 4

(a) Structure of optimal click compound, 6, and scheme of r(CUG)^exp-catalyzed click chemistry between azides and alkynes. (b) 4 and 5 only reacted in the presence of r(CUG)₁₂ (n = 6, 6 biological replicates, 2 replicate experiments). (c) Oligomerization between 4 and 7 was only observed in DM1-affected cells; thus, r(CUG)^exp was the catalyst (n = 6, 6 biological replicates, 2 replicate experiments). (d) ChemReactBIP identified cellular targets of 8. The only mRNA significantly enriched was DMPK containing r(CUG)^exp (n = 6, 6 biological replicates each with 3 technical replicates, 2 replicate experiments). (e) 6 rescued the MBNL1 pre-mRNA splicing defect at picomolar concentrations in DM1 cells (n = 9, 9 biological replicates, 3 replicate experiments). Data represent mean values ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001, as determined by a two-tailed Student t test.

Figure 5

(a) Structures of FRET sensors. r(CUG)^exp catalyzed the synthesis of a multivalent compound with FRET donors and acceptors in close proximity. (b) Enhanced FRET was observed when compounds were incubated with r(CUG)₁₂, but not with the fully base-paired control RNA r(GC)₂₀ (n = 10, 10 biological replicates, 2 replicate experiments). (c) Fluorescence lifetime measurements in DM1-affected cells revealed that FRET occurred only when both sensors could react via click chemistry (n = 30 cells, 4 biological replicates, 2 replicate experiments). (d) 6 rescued nucleocytoplasmic transport of a luciferase transcript harboring r(CUG)^exp in its 3′ UTR (n = 10, 10 biological replicates, 2 replicate experiments). (e) 6 did not change luciferase mRNA levels, indicating that the enhancement in luminescence is due to improvement of DM1-associated nucleocytoplasmic transport defects (n = 3, 3 biological replicates each with 3 technical replicates, 1 replicate experiment). Data represent mean values ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001, as determined by a two-tailed Student t test.