Fig 2: Proteomic identification of collision-specific ribosome interacting factors.(A) Plot of proteins identified by quantitative mass spectrometry in the polysome fraction of cytosol separated by sucrose gradient centrifugation. The cells were either left untreated, treated with low-dose emetine (1.8 µM) to induce collisions (low eme), or treated with high-dose emetine (360 µM) to freeze ribosomes in place (high eme). Pairwise comparisons are plotted such that factors specifically associating with collided ribosomes (i.e., only in low-dose emetine samples) should fall in the upper-right quadrant. Ribosomal proteins are indicated in orange. The position of 3 standard deviations (SD) from the mean along each axis is indicated with a dashed line. Proteins that deviate 3 SD or more along both axes are labelled (in blue and red). (B) Cells knocked out for ZNF598 (∆ZNF598) were treated with low or high dose emetine as in panel A and the polysome fractions were analysed by mass spectrometry. These data (x-axis) are plotted relative to the same comparison in wild type cells (y-axis). The position of 3 SD along each axis is indicated with a dashed line, and the data points for the same proteins from panel A are labelled. Note that only EDF1 and GIGYF2 fall into the upper right quadrant in both plots. As expected, ZNF598 was not detected in ∆ZNF598 cells. (C, D) Sucrose gradient fractionation and immunoblots of cytosolic lysates prepared from untreated cells versus low-dose emetine treated cells. WT cells were analysed in panel C and ∆ZNF598 cells in panel D. RACK1 is a 40S-ribosomal protein. The positions of fractions containing mono- and polysomes are indicated. Source data are provided for the mass spectrometry analysis.Figure 1—source data 1.Proteins identified by mass spectrometry of polysomes.

Fig 3: Effect of EDF1 on frameshifting at stall sites.(A) Schematic representation of the reporter constructs used to analyse frameshifting. RFP was directly appended to each of the three tested XBP1 reporters (see amino acid sequences below) in three different reading frames. Note that S255A mutation exacerbates, whereas W256A reduces ribosomal stalling, relative to the wild type XBP1 sequence. (B) In-frame versions of the constructs from panel A were co-transfected with BFP-encoding plasmid into Flp-In T-REx 293 cells and analysed by flow cytometry after 20 hr of transgene expression. At least 20,000 transfected (BFP-positive) cells were analysed here and in all the subsequent experiments involving frameshifting reporters. The RFP:GFP ratio, an indicator of readthrough at the XBP1 sequence, is plotted as a histogram for each reporter. (C) The same in-frame constructs as in panel B were analysed in WT and ZNF598 KO cells. Note that the RFP:GFP ratio is elevated in ∆ZNF598 cells for WT XBP1 and XBP1(S255A), but not for XBP1(W256A). (D) XBP1(S255A) (left) or XBP1(W256A) (right) was transfected into cells pre-treated for 72 hr with non-targeting siRNA (control) or EDF1-targeting siRNA and analysed by flow cytometry after 20 hr of reporter expression. Expression of GFP (bottom panels) and RFP (top panels) was plotted as histograms. Gray shaded traces represent cells positively transfected with in-frame (+0) constructs, blue traces represent −1 constructs and red traces represent +1 constructs. Background fluorescence observed in non-transfected cells is indicated by the unshaded thin gray traces. The amount of RFP signal observed for out-of-frame constructs (+1,–1), represented as a percentage of the in-frame signal, is indicated above each graph. Note that GFP expression is not influenced by the RFP reading frame.

Fig 4: Analysis of interactions between EDF1, ZNF598, and GIGYF2.(A) WT HEK293 cells were transfected with constructs encoding indicated proteins (Z- ZNF598-3xFLAG; G - 4EHP-P2A-TST-GIGYF2; E - EDF1-GFP) and allowed to express for 20 hr. Cytosol was depleted of ribosomes by sedimentation and subjected to immunoprecipitation (IP) of 3xFLAG-ZNF598 (left panel) or EDF1-GFP (right panel). When indicated, transfected cells were additionally pre-treated with 1.8 μM emetine for 30 min prior to lysis. Total lysate as well as eluates after IP were analysed by western blotting. The stained blot verifies that the immunoprecipitated proteins were recovered. The IP lanes for each antigen (and the stained blot) are from the same gel and exposure as the lysate lanes, with intervening lanes removed digitally. The relative amounts of lysate and IP samples are indicated on the figure. (B) Cells lacking EDF1 (∆EDF1) or GIGYF2 (∆GIGYF2) were transfected with indicated plasmids and analysed as in panel A. (C) WT (top panel) or ∆EDF1 (bottom panel) cells were transfected with ZNF598-3xFLAG and 4EHP-P2A-TST-GIGYF2 encoding plasmids. After 20 hr of expression and, when indicated, pre-treatment with 1.8 μM emetine, cell lysates were fractionated on a sucrose gradient. Ribosome free fractions 1–4 and heavy polysome fractions 7–10 were pooled and each was analysed by FLAG-IP and western blotting. (D) WT or ∆EDF1 cells were transfected with indicated plasmids and IP was performed as in panel B. Note that EDF1 tagged with 3xFLAG does not recover GIGYF2 (lane 2), unlike ZNF598 containing the same tag (lanes 4 and 6). (E) ZNF598-GIGYF2-4EHP form a complex independent of the ribosome. This complex does not include or require EDF1. However, both EDF1 and the ZNF598-GIGYF2-4EHP engage collided ribosomes. Without EDF1, the ZNF598-GIGYF2-4EHP complex is less stably bound to collided ribosomes.

Fig 5: EDF1 facilitates GIGYF2 function at collided ribosomes.(A) The indicated cell lines bearing stably integrated translation reporters were treated with the indicated siRNAs for 72 hr. After additional 20 hr of doxycycline-induced reporter expression, cells were collected and analysed by flow cytometry. Histograms representing GFP expression were plotted on a logarithmic scale. (B) WT cells (left panel) and ∆EDF1 cells (right panel) were treated with low dose emetine (1.8 µM) for 1 hr and their cytosolic extracts were analysed by sucrose gradient centrifugation followed by western blotting. The profiles of GIGYF2 across the gradient, quantified by densitometry, are depicted below the blots. (C) Radiolabelled and immunopurified WT EDF1 or R85G mutant EDF1 produced by in vitro translation was added to separate non-labelled in vitro translation reactions in which ribosomes on native globin mRNA were stalled and collided at the stop codon with eRF1AAQ. The samples were then separated by sucrose gradient centrifugation and EDF1 detected by autoradiography. Note that almost 100% of the WT EDF1 comigrates with collided ribosomes, whereas only ~40% of the R85G mutant is found in polysomal fractions with remainder at the top of the gradient. (D) ∆EDF1 cells with a stably integrated stalling reporter were transfected with a control plasmid (expressing BFP; shaded gray), WT EDF1 (green traces) and R85G EDF1(blue traces) and expression of transgenes was allowed for 24 hr. After additional 20 hr of doxycycline-induced expression of the stalling reporter, GFP expression was analysed by flow cytometry. Note that only WT, but not the R85G mutant reduces the expression of GFP to the level comparable to WT cells.