Girolline is a sequence context-selective modulator of eIF5A activity

Abstract

Natural products have a long history of providing probes into protein biosynthesis, with many of these compounds serving as therapeutics. The marine natural product girolline has been described as an inhibitor of protein synthesis. Its precise mechanism of action, however, has remained unknown. The data we present here suggests that girolline is a sequence-selective modulator of translation factor eIF5A. Girolline interferes with ribosome-eIF5A interaction and induces ribosome stalling where translational progress is impeded, including on AAA-encoded lysine. Our data furthermore indicate that eIF5A plays a physiological role in ribosome-associated quality control and in maintaining the efficiency of translational progress. Girolline helped to deepen our understanding of the interplay between protein production and quality control in a physiological setting and offers a potent chemical tool to selectively modulate gene expression.

Fig. 1. Giro stalls elongating ribosomes.

a Molecular Structure of Giro b Measurement of translation by metabolic labeling with O-propargyl puromycin (OP-puro) in the presence of increasing doses of Giro. The experiment was conducted in triplicate with background deduction. Error bars denote the standard deviation of the mean. The data shown is representative of three independently conducted experiments. c In vitro polysome profiling of [³²P]-labeled globin mRNA in rabbit reticulocyte lysate in the presence of DMSO (gray), the non-hydrolyzable GTP analog GMP-PNP (blue) or Giro (orange). Reactions were ultracentrifuged through a sucrose gradient and fractionated. The RNA content of each fraction was measured by scintillation counting. Ribosomal populations are marked in the graph. Data is representative of three independently conducted experiments. d Metagene analysis of averaged ribosome profiling reads around the translation start site for both monosomes and disomes. Overlay of 1 µM (yellow) or 10 µM (orange) Giro-treated samples with DMSO control (gray). Reads were normalized by total read count and their 5′ ends were mapped from −150 to 300 nucleotides from the start codon. e Metagene analysis of average ribosome profiling reads mapped from − 300 to 300 nucleotides from the translation stop site. Since the ribosomal A-site is displaced by ~ 15 nucleotides (monosomes) or ~ 45 nucleotides (disomes) from the 5′ end of the read, the peaks appear shifted from the stop codon (indicated by arrows on graphs for 1 µM Giro treatment). f Monosome polarity shift analysis in the presence of 1 µM and 10 µM Giro. The distribution of ribosome density was evaluated from 5′ to 3′ for all open reading frames. A negative polarity score indicates a higher ribosome density towards the translation start site. g Polarity shift analysis for disome data. Source data are provided within the Source Data file.

Fig. 2. Sequence-selective stalling.

a Evaluation of increased E, P, or A-site occupancy of disome fractions treated with 1 µM or 10 µM Giro. Occupancy scores for each amino acid in the presence of Giro (y-axis) were plotted against DMSO control (x-axis). Points above the green line indicate overrepresented amino acids in the presence of Giro. b Amino acid enrichment analysis using the k-mer probability logo (kpLogo) web tool, which also considers positional interdependencies. Letters above the line indicate the most significantly enriched amino acids in each position. Red numbers indicate a calculated p-value below 0.01 using a one-sided two-sample Student t test. c Further enrichment analysis, focusing on dipeptides occupying E and P-site or P and A-site together. Dipeptide combinations above the green line are overrepresented. Enriched Lys in the E-site or enriched Pro in the A and P-sites are highlighted in orange. d Monosome and Disome footprints mapped across the COX6C ORF for DMSO control (gray), 1 µM Giro (yellow), and 10 µM Giro (orange). The main stall site under Giro treatment is marked with an arrow in the map for 10 µM Giro treatment with the nucleotide and amino acid sequence in E, P, and A-site indicated. e Codon-wise enrichment analysis of disome data for the ribosomal E-site in 10 µM Giro treated cells. The two Lys codons AAA and AAG are outlined in orange. AAG accounts for about 58% of lysines in the human genome, AAA for ~ 42%. f Di-codon enrichment analysis for E and P-site. Enriched dipeptides Lys-Pro (from codons AAA-CCU and AAA-CCG) and Lys-Phe (AAA-UUU) were entered in orange. Source data are provided within the Source Data file.

Fig. 3. Giro interferes with ribosome-eIF5A interaction.

a Affinity pulldown. Flag-tagged eIF5A was used to pull down ribosomes in the presence or absence of Giro. Hypusine-deficient mutant eIF5A K50A was used as a negative control. Hypusinating enzymes DOHH and DHS were co-expressed with Flag-tagged wild-type eIF5A. Ribosome pulldown was measured by Western blotting (IP-WB) against ribosomal proteins RPS6, RPL36A and RPL3. Expression of hypusinating enzymes DOHH and DHS was confirmed by Western blotting on the whole cell lysate input fraction against the V5-tag on a separate membrane (Input-WB). Expression of the K50A mutant appeared stronger than wildtype for unknown reasons. b Flag-pulldown of ribosomes at increasing Giro concentrations. c In vivo polysome profiling. Cells were treated with DMSO control or 50 µM Giro for 1 h, and lysates were centrifuged through a sucrose gradient. RNA content was monitored by OD₂₅₄. Key ribosomal populations are labeled in the DMSO control sample. Fractions were collected for Western blotting and probed for the presence of eIF5A and ribosomal protein RPS6. Fraction numbers are indicated below. Western blotting data is representative of at least two independently conducted experiments. d Disome footprints of 10 µM Giro treated (orange) or eIF5A-KD (light green) cells and their respective controls mapped across the ORFs of ATP5B, COX7A, and NDUFC1. The main disome peak is marked by an arrow with the mRNA and amino acid sequences across ribosomal E, P, and A-sites indicated. e Metagene analysis of significant pause sites. Genes with significant Giro-induced pause sites (1 standard deviation above the mean for reads in each gene) were probed for the location of pause sites under eIF5A depletion with their normalized reads plotted within − 30 and 30 nucleotides of the Giro-induced stall site. f Disome footprints as in d where the Giro-induced stall site occurred 5′of the eIF5A-KD stall and contained AAA-encoded Lys in the E-site. Source data are provided within the Source Data file.

Fig. 4. Sequence-selective stalling on FACS reporters.

a FACS reporter used in this study. The vector produces three peptides from one transcript: EGFP as internal control, linker, and RFP as experimental readout. The linker sequence was varied for each set of experiments. Cells were transiently transfected with the reporter vector before overnight treatment with DMSO control or 1 µM Giro b Control readout for the empty linker (XXX)₀. Graphs indicate the RFP signal plotted against EGFP and the histograms of EGFP and RFP signal respectively. The left peaks in each histogram indicate non-fluorescing cells. c FACS readout for a vector containing a linker of 20 consecutive AAA-Lys codons (AAA)₂₀ in the presence of DMSO or 1 µM Giro. Arrows in the Giro sample highlight cell populations with low levels of RFP production. d FACS readout as above but for vector containing 20 consecutive AAG-Lys codons (AAG)₂₀. Data for each condition are representative of at least 3 independently carried out experiments. Gating information is available in Supplementary Fig. 8.

Fig. 5. eIF5A is required for unimpeded AAA translation.

a, b Overlay of EGFP and RFP histograms for a non-targeting siRNA pool a and eIF5A-KD b treated with 1 µM Giro on the (AAA)₂₀ reporter. a Histograms of EGFP and RFP expression in control RNAi pool cells upon 1 µM Giro (orange) or DMSO control (black) treatment (c.f. Figure 4c). b Histograms of EGFP and RFP expression in cells with depleted eIF5A in the presence or absence of 1 µM Giro. Arrows highlight the peak of RFP expression in cells treated with DMSO control for cells transfected with non-targeting siRNA a or eIF5A-KD b. c, d EGFP and RFP histograms for non-targeting RNAi pool c or eIF5A-KD d in the presence or absence of 1 µM Giro on the (AAG)₂₀ reporter. Data is representative of at least three independently conducted experiments. Gating information and histograms are available in Supplementary Fig. 8.

Fig. 6. Giro induces RQC.

a Adjusted reporter construct for Western blotting. The N-terminal P2A site was deleted, and a stop codon was added to the end of the linker. b Western blot against EGFP for cells depleted in eIF5A and key factors in ribosome quality control, each treated with DMSO or 1 µM Giro and transfected with (AAA)₂₀-reporter. Arrows indicate full-length protein of ~ 48 kDa and truncated products c western blot against EGFP for cells depleted in eIF5A or RQC factors and expressing a (GAA)₂₀ -reporter. Data is representative of three independently conducted experiments. Source data are provided within the Source Data file.

Fig. 7. Model of Giro- and eIF5A-mediated translational modulation.

a On sequences slowing translation, including but not limited to poly-Pro, eIF5A is required to expedite translation elongation. b Slowed translation enables Giro binding to the ribosome, preventing eIF5A from aiding peptidyl transfer, thereby turning ribosomal slowing into ribosomal stalling and subsequent collision with the following ribosome. This prematurely activates the ribosome-associated quality control pathway (RQC), which splits the ribosomal subunits apart and initiates the destruction of the nascent peptide.