DMDA-PatA mediates RNA sequence-selective translation repression by anchoring eIF4A and DDX3 to GNG motifs

Abstract

Small-molecule compounds that elicit mRNA-selective translation repression have attracted interest due to their potential for expansion of druggable space. However, only a limited number of examples have been reported to date. Here, we show that desmethyl desamino pateamine A (DMDA-PatA) represses translation in an mRNA-selective manner by clamping eIF4A, a DEAD-box RNA-binding protein, onto GNG motifs. By systematically comparing multiple eIF4A inhibitors by ribosome profiling, we found that DMDA-PatA has unique mRNA selectivity for translation repression. Unbiased Bind-n-Seq reveals that DMDA-PatA-targeted eIF4A exhibits a preference for GNG motifs in an ATP-independent manner. This unusual RNA binding sterically hinders scanning by 40S ribosomes. A combination of classical molecular dynamics simulations and quantum chemical calculations, and the subsequent development of an inactive DMDA-PatA derivative reveals that the positive charge of the tertiary amine on the trienyl arm induces G selectivity. Moreover, we identified that DDX3, another DEAD-box protein, is an alternative DMDA-PatA target with the same effects on eIF4A. Our results provide an example of the sequence-selective anchoring of RNA-binding proteins and the mRNA-selective inhibition of protein synthesis by small-molecule compounds.

Fig. 1. Comparative analysis of translation changes induced by eIF4A-targeting compounds in cells.

a Schematic of ribosome profiling experiments. The chemical structures of the eIF4A-targeting compounds used in the experiments are shown. b Principal component (PC) analysis of the translation changes analyzed by ribosome profiling under the indicated conditions. c Spearman’s correlation coefficients (ρ, two-tailed) for translation changes induced by drug treatments. The color scales for ρ are shown. d MA (M, log ratio; A, mean average) plot of the translation fold change with 0.1 μM DMDA-PatA treatment. Low-sensitivity mRNAs (FDR ≤ 0.01 and log₂-fold change ≥1 from the mean) and high-sensitivity mRNAs (FDR ≤ 0.01 and log₂-fold change ≤ −1 from the mean) are highlighted. e Schematic of the RNA pulldown-Seq experiments. mRNAs associated with SBP-tagged eIF4A1 in the cells were isolated and subjected to deep sequencing. f Cumulative distribution of the mRNA fold change in RNA pulldown-Seq data for SBP-tagged eIF4A1 upon 0.01 μM DMDA-PatA treatment. DMDA-PatA low-sensitivity and high-sensitivity mRNAs (defined in d) were compared to total mRNAs. The significance was calculated by the Mann‒Whitney U test (two-tailed). Source data are provided as a Source Data file.

Fig. 2. DMDA-PatA provides GNG motif preference on eIF4A1.

a Schematic of RNA Bind-n-Seq experiments. Randomized RNAs associated with recombinant SBP-tagged eIF4A1 in vitro were isolated and subjected to deep sequencing. b Comparison of the enrichment of 4-mer motifs with DMSO and those with DMDA-PatA. AMP-PNP and 15 pmol of recombinant eIF4A1 were included in the reaction. Motifs containing GNG are highlighted. c Box plots for motif enrichment in RNA Bind-n-Seq (with AMP-PNP) on eIF4A1 with DMSO or DMDA-PatA in the indicated 4-mer species. d Motif enrichment Z score along the titrated recombinant eIF4A1 in RNA Bind-n-Seq (with AMP-PNP) for the indicated 4-mer species with or without DMDA-PatA. The median (line) and upper/lower quartiles (shade) are shown. e, h Fluorescence polarization assay for FAM-labeled RNAs along the titrated recombinant eIF4A1 (wild type or Phe163Leu-Ile199Met mutant) with ADP and Pi. The indicated RNA sequences at 10 nM were used with or without 50 μM DMDA-PatA. The data are presented as the mean (point) and s.d. (error) for replicates (n = 3). f Rank plot for 4-mer motifs enriched in RNA Bind-n-Seq (with ADP and Pi) on eIF4A1 in the presence of DMDA-PatA. Motifs containing GNG are highlighted. g Box plots for motif enrichment in RNA Bind-n-Seq (with ADP and Pi) on eIF4A1 with DMDA-PatA in the indicated 4-mer species. In the box plots, the medians (centerlines), upper/lower quartiles (box limits), and 1.5× interquartile ranges (whiskers) are shown. The significance was calculated by the Mann‒Whitney U test (two-tailed) for all motifs (n = 256), GNG motifs (n = 31), GAG motifs (n = 8), GUG motifs (n = 8), GGG motifs (n = 8), and GCG motifs (n = 8) (c, g). Source data are provided as a Source Data file.

Fig. 3. GNG motif-selective clamping of eIF4A causes the repression of translation initiation.

a Rank plot for motif prediction by ribosome profiling under 0.1 μM DMDA-PatA treatment. Spearman’s correlation coefficients (ρ, two-tailed) between the number of 4-mer motifs found in the 5′ UTRs and translation changes of the mRNAs were calculated. Motifs containing GNG are highlighted. b Rank plot for motif prediction by RNA pulldown-Seq under 0.01 μM DMDA-PatA treatment. Spearman’s correlation coefficients (ρ, two-tailed) between the number of 4-mer motifs found in 5′ UTRs and mRNA changes on SBP-tagged eIF4A1 were calculated. Motifs containing GNG are highlighted. c Schematic of reporter mRNAs with 7× NGNGNG motifs and the control CAA repeats (left). These mRNAs were subjected to in vitro translation with RRL and titration with DMDA-PatA (right). The data are presented as the mean (point) and s.d. (error) for replicates (n = 3). d, f Toeprinting assay to probe the 48S ribosomes assembled on the start codons in the indicated reporter mRNAs with or without 10 μM DMDA-PatA. cDNA synthesized with FAM-labeled reverse transcription primers was analyzed by capillary electrophoresis. A magnified view of the results for reporter mRNA with 7× AGAGAG motifs (the area defined by the dashed line in d) is shown in f. AU, arbitrary unit. e Relationships between translational repression observed during in vitro translation (at 3 μM DMDA-PatA) (c) and the reduction in 48S formation (Supplementary Fig. 3g) for the indicated reporter mRNAs. The data are presented as the mean (point) and s.d. (error) for replicates (n = 3). The regression line (dashed line) is shown. g Toeprinting assay with the recombinant eIF4A1 protein on the reporter mRNA with 7× AGAGAG motifs with or without 10 μM DMDA-PatA. cDNA synthesized with FAM-labeled reverse transcription primers was analyzed by capillary electrophoresis. A magnified view of the results is shown at the bottom. h In vitro translation of reporter mRNAs (with 7× AGAGAG motifs or CAA repeats, both at 90.9 nM) preincubated with recombinant eIF4A1 and DMDA-PatA. A size exclusion column was used to eliminate free DMDA-PatA. The data are presented as the mean (bar) and s.d. (error) for replicates (point, n = 3). The significance was calculated by Student’s t test (two-tailed). Source data are provided as a Source Data file.

Fig. 4. MD simulations and FMO calculations elucidated the energetic impact of RNA sequences on the association with DMDA-PatA.

a Overlaid simulated structures (n = 10) of polypurine RNA•DMDA-PatA•eIF4A1 complexes analyzed by MD simulation. The complexes with RNAs possessing the indicated substitutions were investigated. b The interaction energy between DMDA-PatA and the indicated bases along the investigated complexes through FMO calculations. The data are presented as the mean (bar) for 10 simulated complexes (point). c The difference in the interaction energy between DMDA-PatA and the indicated bases from the original sequence (₆GAGA₉). See b for the source data. The color scale is shown. d The difference in the electrostatic interaction energy between DMDA-PatA and the indicated bases from the original sequence (₆GAGA₉). See Supplementary Fig. 4c for the source data. The color scale is shown. e, f Representative structures obtained by MD simulations (at 100 ns) with ₆GAGA₉ (original sequence) or ₆AGAG₉. The net charge (e) on the indicated groups and the distances are shown (mean ± s.d. for 10 simulated structures). Source data are provided as a Source Data file.

Fig. 5. The tertiary amine on the trienyl arm of DMDA-PatA confers GNG selectivity on eIF4A1.

a Chemical structure of iPr-DMDA-PatA. b Box plots for motif enrichment in RNA Bind-n-Seq (with ADP and Pi) on eIF4A1 with DMDA-PatA or iPr-DMDA-PatA in the indicated 4-mer species. c Fluorescence polarization assay for FAM-labeled RNAs along the titrated recombinant eIF4A1 with ADP and Pi. The indicated RNA sequences at 10 nM were used with or without 50 μM iPr-DMDA-PatA. The data are presented as the mean (point) and s.d. (error) for replicates (n = 3). d Comparison of the enrichment of 4-mer motifs with DMDA-PatA and those with iPr-DMDA-PatA. ADP and Pi were included in the reaction. Motifs containing GNG are highlighted. The regression line (dashed line) for non-GNG motifs (gray points) is shown. ρ, Spearman’s correlation coefficient (two-tailed). e The mRNAs shown in Fig. 3c were subjected to in vitro translation with RRL with the titration of iPr-DMDA-PatA. The data are presented as the mean (point) and s.d. (error) for replicates (n = 3). f The relative amount of OP-Puro incorporation into the nascent peptide in HEK293 cells was analyzed with the titrated DMDA-PatA or iPr-DMDA-PatA. The data are presented as the mean (point) and s.d. (error) for replicates (n = 3). g The relative cell viability was analyzed with titrated DMDA-PatA and iPr-DMDA-PatA. The data are presented as the mean (point) and s.d. (error) for replicates (n = 3). In the box plots, the medians (centerlines), upper/lower quartiles (box limits), and 1.5× interquartile ranges (whiskers) are shown. The significance was calculated by the Mann‒Whitney U test (two-tailed) for all motifs (n = 256), GNG motifs (n = 31), GAG motifs (n = 8), GUG motifs (n = 8), GGG motifs (n = 8), and GCG motifs (n = 8) (b). Source data are provided as a Source Data file.

Fig. 6. DMDA-PatA-mediated DDX3X clamping on the GNG motif occurs in an ATP-independent manner.

a, e Fluorescence polarization assay for FAM-labeled RNAs along the titrated recombinant DDX3X (helicase core) (wild type, Gln360Leu mutant, or Gln360Pro mutant) with ADP and Pi. The indicated RNA sequences at 10 nM were used with or without 50 μM DMDA-PatA. The data are presented as the mean (point) and s.d. (error) for replicates (n = 3). b Rank plot for 4-mer motifs enriched in RNA Bind-n-Seq (with ADP and Pi) on DDX3X (helicase core) in the presence of DMDA-PatA. Motifs containing GNG are highlighted. c Box plots for motif enrichment in RNA Bind-n-Seq (with ADP and Pi) on DDX3X (helicase core) with DMDA-PatA in the indicated 4-mer species. d Box plots for motif enrichment in RNA Bind-n-Seq (with ADP and Pi) on DDX3X (helicase core) with DMDA-PatA or iPr-DMDA-PatA in the indicated 4-mer species. f Relative cell viability under the indicated conditions after treatment with 0.1 μM DMDA-PatA for 48 h. The data are presented as the mean (bar) and s.d. (error) of replicates (point, n = 3). g Schematic of DMDA-PatA-mediated repression of mRNA-selective translation. Clamping of eIF4A1/2 and/or DDX3X to the GNG motif in the 5′ UTR provides steric hindrance for ribosome scanning. In the box plots, the medians (centerlines), upper/lower quartiles (box limits), and 1.5× interquartile ranges (whiskers) are shown. The significance was calculated by the Mann‒Whitney U test (two-tailed) for all motifs (n = 256), GNG motifs (n = 31), GAG motifs (n = 8), GUG motifs (n = 8), GGG motifs (n = 8), and GCG motifs (n = 8) (c, d) and by the Tukey‒Kramer test (two-tailed) (f). Source data are provided as a Source Data file.