Small Molecule Recognition and Tools to Study Modulation of r(CGG)(exp) in Fragile X-Associated Tremor Ataxia Syndrome

Abstract

RNA transcripts containing expanded nucleotide repeats cause many incurable diseases via various mechanisms. One such disorder, fragile X-associated tremor ataxia syndrome (FXTAS), is caused by a noncoding r(CGG) repeat expansion (r(CGG)(exp)) that (i) sequesters proteins involved in RNA metabolism in nuclear foci, causing dysregulation of alternative pre-mRNA splicing, and (ii) undergoes repeat associated non-ATG translation (RANT), which produces toxic homopolymeric proteins without using a start codon. Here, we describe the design of two small molecules that inhibit both modes of toxicity and the implementation of various tools to study perturbation of these cellular events. Competitive Chemical Cross Linking and Isolation by Pull Down (C-Chem-CLIP) established that compounds bind r(CGG)(exp) and defined small molecule occupancy of r(CGG)(exp) in cells, the first approach to do so. Using an RNA GFP mimic, r(CGG)(exp)-Spinach2, we observe that our optimal designed compound binds r(CGG)(exp) and affects RNA localization by disrupting preformed RNA foci. These events correlate with an improvement of pre-mRNA splicing defects caused by RNA gain of function. In addition, the compounds reduced levels of toxic homopolymeric proteins formed via RANT. Polysome profiling studies showed that small molecules decreased loading of polysomes onto r(CGG)(exp), explaining decreased translation.

Figure 1

Modes of toxicity associated with r(CGG)^exp, the causative agent of FXTAS. (A) The repeating RNA folds into a hairpin structure that binds and sequesters proteins that regulate RNA processing events. (B) Repeating transcripts are translated without a start codon in a mechanism called repeat associated non-ATG translation (RANT), producing toxic polymeric proteins that contribute to pathogenesis. (C) Small molecules that bind to r(CGG)^exp free bound proteins, improve defects in RNA processing, and inhibit production of RAN, but not canonical, translation products.

Figure 2

Structures of designer compounds and validation of their binding to and occupancy of r(CGG)^exp. (A) Structure of 1a. (B) Modularly assembled compounds thereof, 2HE-nNMe. (C) Structure of 1a-CA-Biotin used as the Chem-CLIP probe, allowing pull-down or r(CGG)₈₈-GFP transcripts in cells. (D) Results of Chem-CLIP. A ∼5-fold enrichment of r(CGG)₈₈-GFP was observed in the pulled down fractions from cells treated with 1a-CA-Biotin. (E) Results of C-Chem-CLIP. Co-treatment of cells with 1a or 2HE-5NMe and 1a-CA-Biotin (2.5 μM) inhibited the probe from binding to the RNA target. Target occupancy of 1a and 2HE-5NMe compounds for r(CGG)₈₈-GFP in cells was determined by measuring the IC₅₀ for inhibition of the 1a-CA-Biotin reaction with r(CGG)₈₈-GFP. ****p < 0.0001, as determined by a Student's t test.

Figure 3

1a and 2HE-5NMe shown to increase the thermal stability of r(CGG)₁₂, which has five 1 × 1 nucleotide GG internal loops. Optical melting experiments were completed with 1 μM r(CGG)₁₂ and the indicated concentrations of compound. (A) Representative plots of optical melting experiments ([compound] = 5 μM). (B) Plot of the melting temperature of r(CGG)₁₂ as a function of 2HE-5NMe or 1a concentration. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 as determined by a Student t test.

Figure 4

2HE-5NMe shown to improve FXTAS-associated pre-mRNA splicing defects in a cellular model. Top, representative gel image of the alternative pre-mRNA splicing of the SMN2 mini-gene in the presence and absence of r(CGG)^exp upon treatment with 2HE-5NMe or 1a. Bottom, plot of the quantification of SMN2 exon 7 inclusion upon compound treatment. *p < 0.05 and ***p < 0.001, as determined by a Student t test.

Figure 5

Effect of 2HE-5NMe on r(CGG)^exp-containing nuclear foci. Previously, it was shown that 1a inhibits foci formation but cannot disrupt pre-existing foci.(26) (A) Top left, inhibition of nuclear foci formation by 2HE-5NMe. Bottom left, disruption of nuclear foci formation by 2HE-5NMe. Right, quantification of nuclear foci in 2HE-5NMe-treated cells. (B and C) Time course on the effects of 2HE-5NMe (50 μM) on foci formation and improvement of alternative pre-mRNA splicing defects. (B) Representative images of nuclear foci at different times. Scale bar, 10 μm. (C) Representative gel images of SMN2 mini-gene splicing products at different times (left) and their quantification (right). *p < 0.05, **p < 0.001, and ***p < 0.0001 as determined by a Student t test.

Figure 6

Inhibition of RANT and accumulation of RANT products in nuclear inclusions. (A) 1a and 2HE-5NMe inhibit RANT of r(CGG)₈₈-GFP. 1a (left) and 2HE-5NMe (right) inhibit RANT of r(CGG)^exp but not canonical translation of the downstream ORF, which encodes GFP, as determined by Western blotting. P values were determined by a Student t test. (B) 1a decreases FMRpolyG-GFP accumulation and aggregation. Left, representative images from COS7 cells expressing (CGG)₉₀-GFP for 48 h in the presence of vehicle (DMSO) or compound 1a at the indicated concentration. Aggregates of FMRpolyG-GFP fusion protein accumulate in these cells and are reduced in the presence of 1a. Right, percent of FMRpolyG-GFP inclusions 24 (blue) and 48 (red) h post treatment with vehicle or 1a. Data were pooled from three independent experiments. Drug treatment was associated with a significant reduction in inclusion number. P values were determined by Fisher's exact test. *p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001.

Figure 7

Polysome profiling to study ribosome loading onto r(CGG)^exp in the presence and absence of compound. (A) Sucrose gradients for untreated, 1a- or 1a-CA-Biotin-treated cells. (B) Distribution of r(CGG)₈₈-(No ATG)-GFP. (C) Distribution of β-actin mRNA. (D) Quantification of the results shown in B and C.