henhaobio
Sign in

bioRxivMolecular BiologyPreprint

Anisomycin-induced ribosome collisions follow a dose-dependent positional rule, distinct from the UV-induced collision landscape

Sébastien ChamoisiD, Julien SoudetiD, Virginie RicciiD, Félix NaefiD, Cédric GobetiD, David GatfieldiD

PDFDOICodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

Ribosome collisions trigger stress pathways such as the ribotoxic stress response, yet how different ribosome-impairing stimuli shape where collisions form on mRNAs remains unclear. Here, we combine monosome- and disome-resolved ribosome profiling with proteomics of collided-ribosome fractions to compare anisomycin- and UV-induced ribosome impairment in mouse embryonic stem cells. Low-dose anisomycin, a widely used collision-inducing paradigm, does not generate collisions dispersed along transcripts but a post-initiation collision zone, with stalled monosomes and disomes accumulating within the first ~70 nucleotides of coding sequences across most translated mRNAs. High-dose anisomycin, commonly considered collision-poor, produces the same pattern shifted closer to the start codon and reduced in amplitude. These observations support a single dose-dependent positional rule, in which drug occupancy determines how far ribosomes progress before arrest and ongoing initiation supplies trailing ribosomes. In contrast, 30 min after UV irradiation, collisions are detectable biochemically but appear more dispersed, with few recurrent hotspots and preferential accumulation on highly translated mRNAs. Collided-ribosome fractions share enrichment of the collision sensor EDF1 across conditions, whereas recruitment of ZNF598 and initiation factors is most evident after UV. Our results refine the interpretation of anisomycin-based collision paradigms and show that collision landscapes depend on the nature of the stimulus.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (28 Oct 2026) · CC BY 4.0

Low- and high-dose anisomycin confine ribosome collisions to a dose-shifted zone just past the start codon, whereas 30 min after UV, collisions appear more dispersed and favor highly translated mRNAs.

Low-dose elongation inhibitors are a standard way to induce ribosome collisions and ribotoxic stress signalling, but the assumed model (scattered stalls at low dose, no collisions at high dose) had not been tested transcriptome-wide in mammalian cells. Disome-resolved ribosome profiling plus collided-ribosome proteomics in mouse embryonic stem cells (mESCs) map where drug- and UV-induced collisions actually form and which factors they recruit.

Key results

  1. ShownLow-dose anisomycin (0.1 mg/L, 15 min) does not spread collisions along coding sequences. Stalled monosomes and collided disomes instead build up in a post-initiation zone shared by most translated mRNAs.

    Metagene peaks at ∼30 nt (monosome) and ∼63 nt (disome, leading-ribosome A-site) downstream of the start codon. Cumulative profiles confine the treatment gain to early CDS, with no downstream disome rise vs DMSO. Per-transcript fold-change heat maps of the 5,000 most disome-covered transcripts show the same positional increase in most rows. · Low- and high-dose anisomycin produce a…; ANS produces a uniform collision zone across…

  2. ShownHigh-dose anisomycin (100 mg/L), usually considered collision-poor, gives the same post-initiation pattern. It is shifted toward the start codon, has lower collision amplitude, and adds a loss of monosome signal at the start codon itself.

    Monosome/disome peaks at ∼24/∼48 nt vs ∼30/∼63 nt under LD ANS, with lower disome amplitude. Phospho-JNK is reduced but detectable, and MNase-resistant disomes are lower than under LD ANS. HD ANS called 1,354 up/950 down monosome sites (332 genes) but fewer disome sites (658 up/678 down; 257 genes). · Establishing anisomycin and UV conditions that…; Low- and high-dose anisomycin produce a…; ANS produces a uniform collision zone across…; Site-resolved analyses reveal abundant…

  3. ShownAt 30 min after UVB, disome gains are uneven: they concentrate on highly translated mRNAs, enriched for ribosomal-protein and translation-factor transcripts. Collision propensity follows ribosome loading rather than the 5′ TOP motif.

    k-means clustering of UV disome fold-change profiles split the 5,000 transcripts into cluster 1 (n = 671; post-initiation rise plus downstream run-off), cluster 2 (n = 1,342; broad gain) and cluster 3 (n = 2,961; little change). ANS disome peaks also fall from cluster 1 to 3. 85 TOP mRNAs and matched non-TOP mRNAs show very similar disome profiles. · ANS produces a uniform collision zone across…

  4. ShownUV produces few recurrent stalling or collision sites and leaves codon dwell times intact. Anisomycin yields many positionally defined sites but no codon-specific pausing; its only reproducible codon effect is shorter A-site dwell on glycine codons.

    UV: 101 up/4 down monosome sites (87 genes) and 118 up/56 down disome sites (96 genes), vs LD ANS 1,166 up/142 down monosome (413 genes) and 886 up/1,240 down disome sites (408 genes). Ribo-DT UV vs control monosome dwell times r = 0.85-0.89, with no significant codon. All four glycine codons had reduced A-site dwell under both ANS doses. · Codon dwell times (Ribo-DT); Site-resolved analyses reveal abundant…

  5. SuggestedDisome fractions show EDF1 enrichment under all three conditions. ZNF598, downstream RQC factors and translation initiation factors co-sediment most clearly after UV.

    Input- and ribosomal-protein-normalised proteomics (n = 3; two-sided t-tests, nominal p < 0.05, no FDR): EDF1 up in disomes under LD ANS, HD ANS and UV; ZNF598 enriched only in UV disomes; LTN1, NEMF, ANKZF1 and most eIF3 subunits most consistent under UV; ABCF1 depleted in all conditions. · MS data normalisation and post-processing; Collided-ribosome fractions show shared and…

Methods, data and reproducibility

mESCs were treated with anisomycin (0.1 or 100 mg/L, 15 min) or UVB (0.05 J/cm^2, 30 min recovery). Monosome and disome footprints were sequenced with matched RNA-seq, and nuclease-resistant gradient fractions were profiled by data-independent acquisition mass spectrometry. Analyses used metagenes, per-codon Fisher-test pause sites and Ribo-DT dwell times.

  • Monosome and disome Ribo-seq, LD ANS (0.1 mg/L) vs DMSO, batch 1; one treated monosome and one control disome library omitted for low read numbers · mouse (male 129 × C57BL/6 mESCs) · n = 3 control and 3 treated biological replicates · GSE344663
  • Monosome and disome Ribo-seq, HD ANS (100 mg/L) and UV (30 min post-irradiation) vs matched controls, batch 2 · mouse (male 129 × C57BL/6 mESCs) · n = 4 treated and 4 control biological replicates · GSE344663
  • Total RNA-seq (Ribo-Zero Plus) from aliquots of the same lysates used for footprinting, all treatments and controls · mouse · n = n=3 per arm (LD ANS); n=4 per arm (HD ANS, UV) · GSE344668
  • DIA-PASEF proteomics of monosome and disome fractions from MNase-digested sucrose gradients plus input lysates, LD ANS, HD ANS and UV vs controls · mouse · n = N=3 per condition · PXD077189 (ProteomeXchange/PRIDE)
  • Polysome profiles of MNase-digested and undigested lysates for each treatment and control · mouse · n = N=3 independent experiments
  • Western blots of phospho-JNK across an ANS titration (0.001 to 100 mg/L) and a UV recovery time course (5 to 120 min), plus phospho-eIF2α under LD ANS, HD ANS and UV · mouse
  • Analysis subset: transcripts with highest control disome coverage across the CDS (Supplementary Table S2) · mouse · n = 5000 transcripts
  • UniProt mouse reference proteome (February 6th, 2025) plus contaminant database for DIA search · mouse · n = 54’747 sequences
Methods & tools
Spectronaut 20.4 (Pulsar engine), MaxLFQ, STAR v2.7.11b, TrimGalore v0.6.10, UMI-tools, riboWaltz (P-site offsets), GenomicAlignments summarizeOverlaps, DESeq2 v1.46.0, Ribo-DT (codon dwell times), deepTools, Snakemake, ShinyGO
Reagents & environment
STAR v2.7.11b (2.7.0f for Ribo-DT), TrimGalore v0.6.10, fastx_toolkit v0.0.14, UMI-tools v1.0.0git, DESeq2 v1.46.0, Spectronaut 20.4; mouse GRCm39 Ensembl release 111; mapping wrapped in a Snakemake workflow
Reproducibility gaps
GEO and PRIDE deposits are listed with redacted reviewer access tokens, which suggests the data were not yet public when the preprint was posted.
Limitations (6)
  • Ribosome states were preserved with different elongation inhibitors (cycloheximide for profiling, anisomycin for gradients), so the two readouts may capture partly different collided-ribosome populations. · stated by the authors
  • Footprints were selected in fixed length windows with fixed A-site offsets, and anisomycin alters nuclease trimming. As a result, peak positions and the dose-dependent shift are only approximate, although the zone itself held up under these choices. · stated by the authors
  • Signals are relative: profiling is library-normalised, and proteomics reports co-sedimentation rather than direct binding (n = 3, nominal p-values without FDR). Site and dwell-time detection both scale with coverage, so magnitudes cannot be compared across conditions. · stated by the authors
  • Only one cell type, a single 15-min anisomycin exposure and one 30-min post-UV time point were profiled. ZAKα dependence, UV photolesions and ribosomal ubiquitylation were not measured. · stated by the authors
  • Highly covered transcripts are easier to detect in transcript- and site-level analyses, so the apparent collision preference for highly translated mRNAs may partly reflect detection sensitivity. · stated by the authors
  • The initiation-supply model, including its use to explain sparse UV collisions via the integrated stress response, is inferred from two doses. Initiation was never inhibited or titrated to test it directly. · noted in this summary

Read this if you use anisomycin or UV to trigger ribosome collisions or ribotoxic stress and need to know where on mRNAs collisions actually form and which surveillance factors they recruit.

Citations
0

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISLI '26Fedry '23Han '20Marks '26Garshott '21Alcalde '26
71% semantic matchMolecular Biology
J LI, Timo Denk, Matthias Thoms+3 more·bioRxiv·Sep 10, 2026