Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2016 Jun 23;534(7608):558-61.
doi: 10.1038/nature17978. Epub 2016 Jun 15.

Rocaglates Convert DEAD-box Protein eIF4A Into a Sequence-Selective Translational Repressor

Affiliations
Free PMC article

Rocaglates Convert DEAD-box Protein eIF4A Into a Sequence-Selective Translational Repressor

Shintaro Iwasaki et al. Nature. .
Free PMC article

Abstract

Rocaglamide A (RocA) typifies a class of protein synthesis inhibitors that selectively kill aneuploid tumour cells and repress translation of specific messenger RNAs. RocA targets eukaryotic initiation factor 4A (eIF4A), an ATP-dependent DEAD-box RNA helicase; its messenger RNA selectivity is proposed to reflect highly structured 5' untranslated regions that depend strongly on eIF4A-mediated unwinding. However, rocaglate treatment may not phenocopy the loss of eIF4A activity, as these drugs actually increase the affinity between eIF4A and RNA. Here we show that secondary structure in 5' untranslated regions is only a minor determinant for RocA selectivity and that RocA does not repress translation by reducing eIF4A availability. Rather, in vitro and in cells, RocA specifically clamps eIF4A onto polypurine sequences in an ATP-independent manner. This artificially clamped eIF4A blocks 43S scanning, leading to premature, upstream translation initiation and reducing protein expression from transcripts bearing the RocA-eIF4A target sequence. In elucidating the mechanism of selective translation repression by this lead anti-cancer compound, we provide an example of a drug stabilizing sequence-selective RNA-protein interactions.

Conflict of interest statement

The authors declare no competing financial interests

Figures

Extended Data Figure 1
Extended Data Figure 1. RocA represses translation, targeting to eIF4A
(a) Polysome profiling experiments with RocA and PP242 treatments. RocA disrupts polysomes dose-dependently. (b) Western blot of phospho-eIF2α and phospho-4EBP shows that effect of RocA is independent of known translation control targeting to eIFs. Phosphorylation of eIF2α and dephosphorylation of 4EBP were induced by Thapsigargin and PP242, respectively. (c and d) Luciferase reporter assay possessing PTGES3 5′ UTR (Figure 1c) with exogenous expression of WT or RocA resistant eIF4A mutants (c) and western blot of endogenous and exogenous eIF4A (d). eIF4A is the main molecular target of RocA. Data represent mean and S.D. (n = 3). (e and f) Correlation of sum of the footprint reads to 13 mitochondrial mRNAs among different conditions (e) and correlation of sum of the footprint reads from cytoplasmic ribosomes to each transcript between biological replicates (f). r is Pearson’s correlation. P value is calculated by Student′s t-test. (g and h) Tile plot of codon periodicity along length of mitochondria footprints (g, left) and mitochondria footprint length distribution (g, right) and codon periodicities of 31 nt mitochondrial footprints among different conditions (h). Footprints with 31-nt length showed most homogenous codon periodicity and this periodicity was retained with RocA treatment, showing that mitochondrial ribosome translates even in high doses of RocA.
Extended Data Figure 2
Extended Data Figure 2. Extended Data Figure 2. RocA represses translation without mRNA degradation
(a) Metabolic labeling of nascent peptides with OP-puro. The OP-puro incorporated nascent peptides were visualized by Click reaction with Alexa488-azide (middle) and quantified (right). Data represent mean and S.D. (n = 3). (b) Correlation of translation -fold change among different concentrations of RocA treatments. ρ: Spearman’s rank correlation coefficient. (c) MA plot of mean footprint reads between 0.03 μM RocA treatment and non-treatment normalized to library sizes to footprints -fold change by 0.03 µM RocA treatment (left) and the correlation of translation -fold change between 0.03 and 3 µM of RocA treatments (right), highlighting high-sensitivity mRNAs at 0.03 µM RocA treatment. ρ: Spearman’s rank correlation coefficient. (d) Scatter plots of footprints -fold change normalized to mitochondrial footprints and mRNA -fold change by RocA treatments. RocA represses translation without significatnt mRNA change. (e) qPCR from the samples of Figure 1c. Data represent mean and S.D. (n = 3).
Extended Data Figure 3
Extended Data Figure 3. Secondary structure in 5′ UTR is not strong determinant of RocA sensitivity
(a) Cumulative fractions along length of 5′ UTR, minimum ΔG among all 30-mer windows along a 5′ UTR, ΔG in cap-proximal region (30 nt) of 5′ UTR, and Gini difference are plotted to total, RocA high-sensitivity, and RocA low-sensitivity mRNAs. Significance is calculated by Mann-Whitney U test. (b) Cumulative fractions along translation -fold change by RocA are plotted to total mRNAs and mRNAs with predicted G-quadruplexes in 5′ UTRs. Significance is calculated by Mann-Whitney U test. The impact of presence of G-quadruplex in 5′ UTR is modest in RocA sensitivity. (c) The 5′ UTRs with G-quadruplexes and randomized control sequence were fused to Renilla luciferase and these reporter mRNAs were transfected prior to treatment with RocA as indicated. Data represent mean and S.D. (n = 3). G-quadruplex does not show the prominent RocA sensitivity.
Extended Data Figure 4
Extended Data Figure 4. Characterization of translational inhibition by Hippuristanol and PP242
(a) Polysome profiling experiments with Hipp treatments. Hipp disrupts polysomes dose-dependently. (b) Histograms of number of transcripts along footprints -fold change with 0.01 and 1 μM Hipp treatment compared to non-treatment, normalized to mitochondrial footprints. Median -fold change is shown. Bin width is 0.1. (c) MA plot of mean footprint reads between 1 μM Hipp treatment and non-treatment normalized to library sizes to translation -fold change by 1 μM Hipp treatment, highlighting high-sensitivity and low-sensitivity mRNAs. (d) Cumulative fractions along length of 5′ UTR, minimum ΔG among all 30-mer windows along a 5′ UTR, ΔG in cap-proximal region (30 nt) of 5′ UTR, and Gini difference are plotted to total, Hipp high-sensitivity, and Hipp low-sensitivity mRNAs. Significance is calculated by Mann-Whitney U test. (e) Translation -fold changes by RocA and Hipp are modestly correlated. ρ: Spearman’s rank correlation coefficient. (f) MA plot of mean footprint reads between 2.5 μM PP242 treatment and non-treatment normalized to library sizes to translation -fold change by PP242 treatment, highlighting PP242 target mRNAs. (g) Cumulative distributions of translation -fold change caused by RocA and Hipp treatment are plotted for total and PP242-target mRNAs. Significance is calculated by Mann-Whitney U test.
Extended Data Figure 5
Extended Data Figure 5. Purification of SBP-tagged eIF4A and co-purified RNA from HEK 293 cells
(a) Western blot of exogenous SBP-eIF4A and endogenous eIF4A in tetracycline-inducible stable cell line. Expression of physiological levels of the tagged allele attenuated endogenous eIF4A expression but preserved overall eIF4A levels, likely reflecting the same feedback loop previously reported between eIF4AI and eIF4AII. (b) CBB staining of purified SBP-eIF4A and SYBR Gold staining of purified RNA bound to SBP-eIF4A with or without Micrococcal Nuclease (MNase). (c) Correlation of sum of the mRNA fragment reads of each transcript between biological replicates of RIP-seq. r is Pearson’s correlation coefficient. P value is calculated by Student′s t-test. (d) Histogram of the number of transcripts along RNA/eIF4A interaction -fold change by RIP-Seq when cells are treated with 0.03 or 0.3 µM RocA normalized to spiked-in RNA. Data present the same mRNAs analyzed in Figure 1a. Median -fold change is shown. Bin width is 0.1. (e) Correlation of RIP -fold change between different concentration of RocA treatments. ρ: Spearman’s rank correlation coefficient. (f) Correlation of translation -fold change to RIP -fold change with the same concentration of RocA treatment. ρ: Spearman’s rank correlation.
Extended Data Figure 6
Extended Data Figure 6. Motif enrichment by Bind-n-Seq
(a) Nucleotide composition in each length of reads in input RNAs for Bind-n-Seq. Input RNAs are random in entire read length. (b) Length distribution of reads from Bind-n-Seq. RNAs bound to eIF4A showed longer length distribution, indicating that eIF4A has preference for longer RNAs. (c) Correlations of 4-mer motif enrichment in Bind-n-Seq by 0.03 μM RocA treatment to that by 0.3 μM RocA treatment. (d) Correlations between 5-mer and 6-mer motif enrichment in Bind-n-Seq by 0.03 μM RocA treatment and motif prediction of 0.03 μM RocA effect in RIP-Seq. ρ: Spearman’s rank correlation. (e) Highest-scoring 5-mer and 6-mer motifs in Bind-n-Seq and RIP-Seq. (f) Cumulative fractions along number of 4-mer motifs (Figure 2b) in 5′ UTR are plotted to total, RocA high-sensitivity, and RocA low-sensitivity mRNAs. Significance is calculated by Mann-Whitney U test. (g) Correlations of Bind-n-Seq motif enrichment (5-mer) by eIF4A to that by 0.03 μM RocA treatment. The motifs appeared in RNAs used in Extended Data figure 8 are highlighted. (h) Correlation of Bind-n-Seq motif enrichment (5-mer) by eIF4A to motif prediction of Hipp effect in translation change, which is define as Spearman’s correlation of motif number in 5′ UTR to translation -fold change by Hipp. mRNAs with high affinity motif to eIF4A in 5′ UTR are resistant to Hipp treatment. (i) The correlation between enriched motifs of replicates in Bind-n-Seq with ADP + Pi. ρ: Spearman’s rank correlation.
Extended Data Figure 7
Extended Data Figure 7. Characterization of iCLIP data
(a) CBB staining of purified SBP-eIF4A protein in iCLIP procedure. (b) Visualization of RNA-crosslinked with SBP-eIF4A and unknown proteins by 32P labeling of RNA. We avoided the contamination of RNAs cross-linked to the additional, co-purifying, unknown proteins. (c) Distribution of read length in iCLIP libraries. Avoidance of contaminating RNAs restricted us to short RNAs, which likely correspond to the region of RNA physically protected by eIF4A binding, or footprint (d) Nucleotide bias along the reads in iCLIP libraries. The crosslinking bias for U may underestimate the preference for polypurine motifs. (e) Correlations of iCLIP motif enrichment (4-mer) by different RocA concentrations. (f) Correlations of iCLIP motif enrichment (4-mer) by 3 μM RocA and motif prediction of 0.03 μM RocA effect in RIP-Seq. The motifs shown in Figure 3b are highlighted. ρ: Spearman’s rank correlation.
Extended Data Figure 8
Extended Data Figure 8. eIF4A/RNA affinity measured by fluorescence polarization
(a) CBB staining of recombinant proteins used in this study. (b) Summary of Kd between RNA and eIF4A among the conditions assayed. (c, e-g, i) Direct measurement of the eIF4A/RNA affinity by fluorescence polarization for eIF4A WT, eIF4A (VX4GKT), or eIF4A (D296A-T298K) and 5′ FAM-labeled RNAs in the presence or absence of RocA. Data represent mean and S.D. (n = 3). (d) ATP crosslinking assay with eIF4A WT and eIF4A (VX4GKT). (h) Pulldown assay with His-MBP-eIF4A expressed in E. coli and eIF4E/G in RRL.
Extended Data Figure 9
Extended Data Figure 9. Characterization of toeprinting assay
(a) Diagram of the reporters used in this study. (b, and c) In vitro translation in RRL with mRNAs containing seven polypurine motif (AGAGAG) insertions (b) and qPCR from the samples (c). (d) Dideoxy terminated sequencing of RNA by reverse transcription verified the toeprinting product length terminated by 48S ribosomes. (e) Ribosome toeprinting assay performed in RRL in the presence of m7-GTP in the presence or absence of 3 μM RocA treatment. (f) Toeprinting assay using 10 μM recombinant eIF4A in the presence or absence of 10 μM RocA treatment. (g) Toeprinting assay (top) and RNase I footprinting assay (bottom) using 10 μM recombinant eIF4A with mRNA containing one AGAGAG motif at the middle in the presence or absence of 10 μM RocA treatment. (h and i) Toeprinting assay using 10 μM recombinant eIF4A (VX4GKT) or (D296A-T298K) with mRNA containing seven AGAGAG motifs in the presence or absence of 10 μM RocA treatment. (j) Pre-formation of the complex with RocA and eIF4A (VX4GKT) or (D296A-T298K) on the mRNA bearing seven polypurine motifs represses the translation from the mRNA in RRL. (k) Basal translation level from mRNA containing seven AGAGAG with the supplementation of recombinant eIF4A. (l) In vitro translation in RRL with mRNAs with single polypurine motif (AGAGAG) insertion at the different positions in 5′ UTR (m) Basal translation level from mRNAs bearing PV IRES and PV IRES with three AGAGAG. In b-c and h-j, data represent mean and S.D. (n = 3).
Extended Data Figure 10
Extended Data Figure 10. 5′ UTR footprints accumulated in RocA treatments are come from uORFs
(a) The distributions of specific footprint length, which is a hallmark of 80S ribosomes, from CDS and 5′ UTR are indistinguishable. (b) The change in ribosome footprint counts for 5′ UTRs and CDSs when cells are treated with 3 μM RocA or 1 μM Hipp compared to non-treatment, normalized to mitochondrial footprints. Median -fold change is shown. Bin width is 0.1. Analysis is restricted to mRNAs bearing footprints in the 5′ UTR in the non-treatment condition. (c) Meta-gene analysis of low-sensitivity transcripts to RocA. Reads are normalized to the sum of mitochondrial footprints reads. (d) The illustration of the definition of uORF translation intensity. (e) Transcripts sensitive to RocA contain more active uORFs, as measured by cumulative distributions of the uORF translation intensity (c). Significance is calculated by Mann-Whitney U test. (f) The summary of deep sequencing based approaches used in this study and corresponding figures.
Figure 1
Figure 1. RNA sequence selectivity is imparted upon eIF4A by RocA causing selective translation repression
(a) Histogram of the number of transcripts along translation -fold change by ribosome profiling when cells are treated with 0.03, 0.3, or 3 μM RocA, normalized to the number of mitochondrial footprints. Median -fold change is shown. Bin width is 0.1. (b) MA plot of mean footprint reads between 3 μM RocA treatment and non-treatment normalized to library sizes versus translation -fold change by 3 μM RocA treatment, highlighting high-sensitivity and low-sensitivity mRNAs. (c) The 5′ UTRs of indicated genes were fused to Renilla luciferase and these reporter mRNAs were transfected prior to treatment with RocA as indicated. Data represent mean and standard deviation (S.D.) (n = 3). (d) Correlation of translation -fold change to RIP -fold change with RocA treatment. ρ: Spearman’s rank correlation.
Figure 2
Figure 2. RNA Bind-n-Seq and iCLIP reveal that RocA preferentially increases the affinity between eIF4A and polypurine motif
(a) Correlations between 4-mer motif enrichment in Bind-n-Seq by 0.03 µM RocA treatment and motif prediction of 0.03 µM RocA effect in RIP-Seq. ρ: Spearman’s rank correlation. (b) Highest-scoring elements in Bind-n-Seq and RIP-Seq. (c) The change in mRNA binding for mRNAs with or without the enriched 4-mer motif (b) in their 5′ UTRs is shown as the RIP -fold change by RocA normalized to spike-in RNA. Significance is calculated by Mann-Whitney U test. (d) Enrichment of 4-mer motifs (b) in iCLIP by RocA treatment relative to control DMSO treatment. (e) The frequency of the 4-mer motif (b) in the 5′ UTR predicts whether a mRNA is high- or low-sensitivity, based on the difference in cumulative distributions of motifs in the 5′ UTR. Significance is calculated by Mann-Whitney U test. (f) Reporter assay in HEK 293 cells with a CAA-repeat 5′ UTR containing seven polypurine motif (AGAGAG) insertions (Extended Data Figure 9a). Data represent mean and S.D. (n = 3).
Figure 3
Figure 3. RocA clamps eIF4A on polypurine motif even after ATP hydrolysis
(a, b) Direct measurement of the eIF4A/RNA affinity by fluorescence polarization for eIF4A and 5′ FAM-labeled RNAs in the presence or absence of RocA. Data represent mean and S.D. (n = 3). (c) Motif enrichments along entire 4-mer motifs in Bind-n-Seq with ADP + Pi and highest-scoring elements (inset). (d) Competition assay with unlabeled RNA. Data represent mean (n = 3). (e) Ribosome toeprinting assay performed in RRL in the presence of GMP-PNP in the presence or absence of 3 μM RocA treatment. (f) Relative RNase I cleavage protected by eIF4A/RocA complex on mRNA containg one AGAGAG at the middle in footprinting assay. See the original data in Extended Data Figure 9f.
Figure 4
Figure 4. eIF4A/RocA complexes on polypurine motifs block scanning of pre-initiation complex, inducing uORF translation
(a) Pre-formation of the complex with RocA and eIF4A on the mRNA bearing seven polypurine motifs represses the translation from the mRNA in RRL. (b) The supplementation of recombinant eIF4A protein to RRL in vitro transaltion reaction with 10 μM Hipp or 3 μM RocA. (c) In vitro translation in RRL with mRNAs with native PV IRES and that with three polypurine motifs (Extended Data Figure 9a). (d) Meta-gene analysis of high-sensitivity transcripts to RocA. Reads are normalized to the sum of mitochondrial footprints reads. Histogram of the position of the first polypurine motif (6-mer) after uORF initiation codon (inset). P value is calculated by Fisher’s exact test. Bin width is 12 nt. (e) Western blot of SBP translated from uORF and downstream major ORF in RRL with 0.03 μM RocA treatment. Quantification of bands normalized to long form with DMSO treatment is shown. For gel source data, see Supplementary Fig. 1. (f) Schematic representation of RocA-mediated translation control. RocA clamps eIF4A onto mRNA by selective affinity enhancement for a polypurine motif in eIF4F-, cap-, and ATP-independent manners, which then blocks scanning of pre-initiation complex, introducing premature translation from uORF and inhibiting downstream ORF translation. In b and c, data represent mean and S.D. (n = 3).

Similar articles

See all similar articles

Cited by 62 articles

See all "Cited by" articles

References

    1. Cencic R, et al. Antitumor activity and mechanism of action of the cyclopenta[b]benzofuran, silvestrol. PLoS One. 2009;4:e5223. - PMC - PubMed
    1. Bordeleau ME, et al. Therapeutic suppression of translation initiation modulates chemosensitivity in a mouse lymphoma model. J Clin Invest. 2008;118:2651–2660. - PMC - PubMed
    1. Liu T, et al. Synthetic silvestrol analogues as potent and selective protein synthesis inhibitors. J Med Chem. 2012;55:8859–8878. - PubMed
    1. Santagata S, et al. Tight coordination of protein translation and HSF1 activation supports the anabolic malignant state. Science. 2013;341:1238303. - PMC - PubMed
    1. Wolfe AL, et al. RNA G-quadruplexes cause eIF4A-dependent oncogene translation in cancer. Nature. 2014;513:65–70. - PMC - PubMed

Publication types

MeSH terms

Feedback