Ribosome Quality Control mitigates the cytotoxicity of ribosome collisions induced by 5-Fluorouracil

Abstract

Ribosome quality control (RQC) resolves collided ribosomes, thus preventing their cytotoxic effects. The chemotherapeutic agent 5-Fluorouracil (5FU) is best known for its misincorporation into DNA and inhibition of thymidylate synthase. However, while a major determinant of 5FU’s anticancer activity is its misincorporation into RNAs, the mechanisms by which cancer cells overcome the RNA-dependent 5FU toxicity remain ill-defined. Here, we report a role for RQC in mitigating the cytotoxic effects of 5FU. We show that 5FU treatment results in rapid induction of the mTOR signalling pathway, enhanced rate of mRNA translation initiation, and increased ribosome collisions. Consistently, a defective RQC exacerbates the 5FU-induced cell death, which is mitigated by blocking mTOR pathway or mRNA translation initiation. Furthermore, 5FU treatment enhances the expression of the key RQC factors ZNF598 and GIGYF2 via an mTOR-dependent post-translational mechanism. This adaptation likely mitigates the cytotoxic consequences of increased ribosome collisions upon 5FU treatment.

Graphical Abstract

Graphical Abstract.

Introduction

The optimal use of many anti-cancer treatments is impeded by an insufficient understanding of their mechanisms of action. 5-Fluorouracil (5FU) is the most commonly used chemotherapeutic agent and backbone of standard-of-care chemotherapy regimens for several types of solid tumours, including colorectal, pancreatic, breast, and head and neck cancers. 5FU is a pyrimidine analogue that, once in the cell, is converted to three active metabolites: (i) FdUMP, which inhibits the nucleotide synthetic enzyme thymidylate synthase (TS); (ii) FdUTP, which mis-incorporates into DNA; and iii) FUTP that mis-incorporates into RNAs (1). Misincorporation of FUTP into RNAs is a major component of the anticancer activity of 5FU, as RNAs accumulate ∼3000–15000× more 5FU metabolites than DNA (1–7). This results in the production of abnormal RNAs (e.g. through altered splicing) (2), aberrant post-transcriptional modifications (e.g. pseudo-uridylation) (7), alteration of ribosome biogenesis or functions (8) and aberrant translation-dependent events (e.g. stop-codon readthrough) (4,8). However, the underlying mechanistic details of cellular response to the RNA-dependent 5FU cytotoxicity remains ill-defined (9).

The eukaryotic mRNA translation process consists of three main stages: initiation, elongation, and ‘termination and recycling of the ribosomes’. Initiation is facilitated by a 5′ m7GpppN cap structure and the aid of the eukaryotic initiation factor 4F (eIF4F) complex, which consists of the cap-binding subunit eIF4E, RNA helicase eIF4A, and the large subunit eIF4G. eIF4F recruits the 43S pre-initiation complex (PIC) via eIF3. This results in the formation of the 48S PIC that scans the 5′ UTR until a start codon is recognised (10). Upon detection of the start codon by the PIC, the 60S ribosomal subunit joins to form the 80S ribosome and initiates the ‘elongation phase’, which continues until the ribosome arrives at a stop codon, whereupon translation termination can occur (11). Initiation is the main stage of regulation of translation, which is achieved via several mechanisms [reviewed in (12)], including abrogation of the interaction between eIF4E and eIF4G by the family of small eIF4E-binding proteins (4E-BPs). Affinity of 4E-BPs for eIF4E is regulated by phosphorylation of 4E-BPs on several residues by the mechanistic Target of Rapamycin Complex 1 (mTORC1) that controls the rate of translation initiation and other key metabolic processes in response to external and internal stimuli, such as nutrient availability (13).

Beyond the imperative to regulate translation rate during initiation (10), the synthesis of full-length, correctly folded proteins is vital for maintaining homeostasis and preventing diseases. Translating ribosomes can be stalled by obstacles such as the presence of damaged (e.g. oxidised) or modified nucleotides and premature poly(A) sequence within the coding region (14–19). Notably, efficiently translated mRNAs are more prone to ribosome collisions (20–22), due to the higher chance of presence of trailing ribosomes that collide into a stalled ribosome. Unresolved stalled ribosomes could have deleterious consequences due to protein aggregation or release of truncated proteins. The ribosome quality control (RQC) mechanism detects collided ribosomes and removes the aberrant mRNAs and nascent peptides, thus preventing their cytotoxic consequences (23,24). Conversely, an overwhelmed (e.g. due to excessive mRNA damage) or dysfunctional RQC results in cell death (25).

Significant progress has been made in understanding the molecular mechanisms of RQC (26), including identification of crucial RQC factors, such as the RNA-binding and E3 ubiquitin ligase protein ZNF598, which triggers RQC by promoting mono-ubiquitination of the small ribosomal subunits (19,27,28). ZNF598, also recruits the cap-binding protein and translational repressor 4EHP via GIGYF2 and thereby prevents further translation initiation on the aberrant mRNA (29). Alternatively, EDF1 can act as a sensor of ribosome collisions and recruit GIGYF2-4EHP complex at collided ribosomes, independent of ZNF598 (30). The ASCC complex containing ASCC3, ASCC2, and TRIP4, participates in the disassembly of collided ribosomes. Subsequently, PELO and ABCE1 split the ribosomal subunits, and ANKZF1/VMS1, in cooperation with Arb1, release the nascent peptide chain from 60S complexes. In parallel, the Ltn1/VCP/NEMF complex promotes the proteasomal degradation of the nascent peptide (26). Through this process, RQC monitors the integrity of mRNA translation and contributes to the maintenance of cellular homeostasis (31,32). Conversely, defects in RQC, for instance due to depletion of Ltn1 or NEMF, lead to major pathological consequences such as neurological disorders resulting from accumulation of toxic protein aggregates (15,24,33,34). Nevertheless, while the causal implication of dysregulation of the general mRNA translation machinery in tumorigenesis and therapy resistance is well-documented (12,35), little is known about the role of RQC in cellular responses to anti-cancer treatments that impact RNA metabolism.

Here, we demonstrate that, contrary to previous assumptions, shortly after 5FU treatment, a major reprogramming of the cellular mRNA translation machinery occurs. This includes activation of the mTORC1 signalling pathway, an elevated rate of translation initiation, and increased ribosome collisions. Crucially, we show that RQC mitigates against 5FU-induced cell death by resolving the collided ribosomes. Conversely, these collided ribosomes accumulate in the RQC-deficient cells, resulting in enhanced 5FU-induced cell death. We also demonstrate that, inhibition of mTOR or repression of cap-dependent mRNA translation initiation reverses the sensitisation to 5FU seen in RQC-deficient cells. Furthermore, our data suggest the presence of a hitherto unknown intrinsic cellular mechanism of post-translational upregulation of expression of the key RQC factors ZNF598 and GIGYF2 upon mTOR activation. Thus, 5FU treatment leads to an mTOR-dependent upregulation of these RQC factors, which further bolsters the cellular response to the 5FU-induced ribosome collisions and mitigates against 5FU-induced cytotoxicity.

Materials and methods

Cell lines and culture conditions

HCT116 human colorectal cancer and SUIT-2 human pancreatic ductal adenocarcinoma cells were grown in McCoy's 5A medium (Gibco, Cat. # 26600080) supplemented with 10% v/v dialysed foetal bovine serum (FBS) (Sigma, Cat. # F0392), 1 mM sodium pyruvate (Gibco, Cat. # 11360039), 100 U/ml penicillin, 100 μg/ml streptomycin (Gibco, Cat. # 15070063). HEK293 (Thermo Fisher Scientific) were cultured in DMEM (Dulbecco's modified Eagle's medium; Gibco, Cat. # 41965039) supplemented with 10% FBS (Gibco, Cat. # 10270106) and 100 U/ml penicillin and 100 μg/ml streptomycin (Gibco, Cat. # 15070063). All cells were maintained at 37°C, in a humidified atmosphere with 5% CO₂ and regularly tested for presence of mycoplasma contamination using mycoplasma detection kit (abm, Cat. # G238).

Immunofluorescence

HCT116 cells were seeded onto poly-l-lysine coated cover slips in 6 well plates. Cells were seeded at a density of 1.8 × 10⁵ and allowed to adhere for 24 h. Cells were then treated with 2.5 μM 5FU (Merck, Cat. # 343922) or vehicle for the indicated times. After treatment, the cells were washed in PBS and fixed in pre-chilled 70% ethanol for 30 min at −20°C. For RNase digestion, fixed cells were incubated with 400 U of RNase I (Thermo Fisher Scientific, Cat. # EN0601) in PBS for 1 h at 37°C, followed by washes with PBS post-digestion. Prior to 5FU detection, depurination of the cells was performed with 2 M hydrochloric acid for 30 min at room temperature followed by 3 washing with PBS. The cells were then blocked with a bovine serum albumin/normal goat serum buffer containing 0.5% Triton X-100 for 1 h. Subsequently, cells were further blocked with a streptavidin/biotin blocking kit (Vector Laboratories, Cat. # SP2002). Next, 5FU uptake into cells was captured using an anti-BrdU antibody (BU-33; Merck) at 1:150 dilution in 1% BSA/PBS. After 1 h incubation at room temperature with the primary antibody, 5FU was detected using an anti-mouse biotinylated secondary antibody (Thermo Fisher Scientific, Cat. #A10519) at 1:200 dilution in 1% BSA/PBS for 1 h at room temperature followed by incubation with streptavidin-conjugated Alexa Fluor™ 488 (Thermo Fisher Scientific, Cat. # S11223) at 1:1000 dilution in PBS for 1 h at room temperature. Cells were then counterstained with DAPI, and coverslips mounted onto microscopy slides. Images were taken using confocal microscopy (Olympus Spinning Disk) at 40x magnification. 5FU fluorescence was analysed with ImageJ software (NIH).

Generation of ZNF598-KO HCT116 and SUIT-2 cells

The pSpCas9(BB)-2A-GFP (Addgene, Cat. 48138) plasmid linearised using BbsI (Thermo Fisher Scientific, Cat. #ER1011) and the oligodeoxynucleotides encoding guide RNA (5′-ACCGCTGCTCTACCAAGATG) for targeting the coding region of ZNF598 gene were ligated. The ligation reaction mix was transformed into E. coli Dh5∝ strain and after transformation, the guide sequence containing pSpCas9(BB)2A-GFP plasmids were isolated and verified by Sanger sequencing using the U6 promoter forward primer. To generate the knockout cells, 250000 cells were plated into 12-well plates and transfected with the guide sequence containing pSpCas9(BB)-2A-GFP plasmid using Lipofectamine 3000 (Thermo Scientific, Cat. # 11668019). 24 h after transfection, GFP positive cells were sorted by FACS into 96-well plates and cultivated until colonies were obtained. Clonal cell lines were analysed by western blot for the absence of the protein. Clones showing lack of ZNF598 protein expression were verified by Sanger sequencing using ZNF598 specific primer pairs: GCAGCTCCTGAGCGGGGAG (forward) and CCGGACCTCAGTGAGGCAAAGT (reverse).

Lentivirus shRNA packaging

Lentivirus pseudovirions were produced by transfecting HEK293T cells using Lipofectamine 2000 (Thermo Fisher Scientific, Cat. # 11668027) and 1.25 μg shRNA plasmid along with equivalent amount of psPAX2 (Addgene, plasmid 12260) and pMD2.G (Addgene, plasmid 12259) in 3:1 ratio in 6 well plates. 48 h post-transfection, media was collected from which pseudovirions were purified by filtration (Filtropur S, 0.2 μm, Sarstedt, Cat. # 83.1826.001) after a brief centrifugation (1500 × g, 5 min). The following shRNAs were used in this study: non-targeting shRNA controls (Sigma, Cat. # SHC002), shZNF598#1 (Sigma, Cat. # TRCN0000073158), shZNF598#2 (Sigma, Cat. # TRCN0000073162), shZNF598#3 (Sigma, Cat. # TRCN0000222610).

RNA extraction and quantitative RT-PCR

Total RNA was isolated using TRIzol reagent (Thermo Fisher Scientific, Cat. # 15596026) as per the manufacturer's protocol. 1 μg purified total RNA was treated with DNase I (Thermo Fisher Scientific, Cat. # EN0521) prior to reverse transcription using SuperScript™ III Reverse Transcriptase (Thermo Fisher Scientific, Cat. # 18080085) and random hexamers. The following primers were used for PCR reactions: ZNF598-Forward: AAAGGTGTACGCATTGTACAGG, ZNF598-Reverse: CTCCAGGTCCCCGAAGAG, GIGYF2-Forward: TCTGTGGGTCAGGAATTTGG, GIGYF2-Reverse: GACATCTGACCACAACCAAAGA, 28S rRNA-Forward: CGATGTCGGCTCTTCCTATC, 28S rRNA-Reverse: TCCTCAGCCAAGCACATACA. Quantitative PCR was performed using LightCycler® 480 Instrument II (Roche, Cat. # 05015278001) using the LightCycler 480 SYBR Green I Master mix (Roche, Cat. # 04887352001), according to the manufacturer's protocol.

Western blotting

Cells were washed with pre-chilled PBS and detached from the plates using plastic scrapers. After centrifugation, cell pellets were lysed in RIPA buffer (50 mM Tris–HCl pH 7.4, 150 mM NaCl, 2 mM EDTA, 1% NP-40, 0.1% SDS) supplemented with 1 mM sodium orthovanadate (Na₃VO₄) and protease inhibitors (Roche, Cat. # 11836170001). Total protein concentration was measured by Bradford protein assay (Bio-Rad, Cat. # 5000006) and 40 μg of total protein was mixed with 5x loading buffer (0.25% Bromophenol blue, 0.5 M DTT, 50% glycerol, 10% SDS, and 0.25 M Tris-Cl pH 6.8). Samples were boiled for 3 min, followed by incubation on ice for 10 min and loaded on SDS-PAGE gel and wet transfer onto PVDF membrane (Merck, Cat. # IPFL00010). Thereafter, membranes were blocked with 5% BSA (Thermo Fisher Scientific, Cat. # A9647) at room temperature for 1 h and incubated with the indicated primary antibodies overnight at 4°C and secondary antibody for 2 h at room temperature. All blots were scanned, and images were captured using the Odyssey system (LI-COR, Cat. # ODY-1540). When necessary, brightness and contrast adjustments were applied evenly across the entire image. Uncropped versions of all western blot images are presented in Supplementary Figures S6–S18. For quantification of the total or phospho-proteins, the intensity of each band was measured with ImageJ software (NIH) and normalised against the total protein (in case of phospho-proteins) or the loading control (β-actin) after subtracting immunoblot background intensity.

The following antibodies were used in this study: rabbit anti-phospho-RPS6 (S240/244) (Cell Signalling, Cat. # 5364), rabbit anti-RPS6 (Cell Signalling, Cat. # 2217), rabbit anti-phospho-4E-BP1 (Cell Signalling, Cat. # 2855), rabbit anti-4E-BP1 (Cell Signalling, Cat. # 9644), rabbit anti-phospho-eIF2α (Abcam, Cat. # ab32157), rabbit anti-eIF2α (Cell Signalling, Cat. # 5324), rabbit anti-phospho-p38 MAPK (Cell Signalling, Cat. # 9211), rabbit anti-p38 MAPK (Cell Signalling, Cat. # 9212), rabbit anti-phospho-eIF4E (Abcam, Cat. # ab76256), mouse anti-eIF4E (BD Biosciences, Cat. # 610270), rabbit anti-eS10 (Abcam, Cat. # ab151550), rabbit anti-RACK1 (Cell Signalling, Cat. # 4716), rabbit anti-ZNF598 (Invitrogen, Cat. # 703601), rabbit anti-GIGYF2 (Proteintech, Cat. # 24790–1-AP), mouse anti-puromycin (Merck, Cat. # MABE343), rabbit anti-PARP (Cell Signalling, Cat. # 9542), rabbit anti-EDF1 (Abcam, Cat. # ab174651), rabbit anti-ASCC3 (Proteintech, Cat. # 17627–1-AP), mouse anti-β-Tubulin (Sigma, Cat. # T4026), rabbit anti-GAPDH (Proteintech, Cat. # 10494-1-AP), mouse anti-β-actin (Sigma, Cat. # A5441), rabbit anti-Vinculin (Cell Signalling, Cat. # 13901S), IRDye® 680RD donkey anti-mouse IgG (LI-COR, Cat. # 926-68072), and IRDye® 800CW donkey anti-rabbit IgG (LI-COR, Cat. # 926-32213).

Polysome profiling

Cells were maintained at maximum 60–70% confluency. After pre-treatment with Cycloheximide (100  μg/ml; Sigma, Cat. # 01810) for 5  min, cells were collected by centrifugation at 4°C for 5  min and lysed in 500  μl hypotonic buffer containing 5 mM Tris–HCl, pH 7.5, 2.5  mM MgCl₂, 1.5  mM KCl, complete EDTA-free protease inhibitor cocktail (Roche, Cat. # 04693159001), 100  μg/ml Cycloheximide, 2  mM DTT, 200  U/ml RiboLock RNase Inhibitor (Thermo Fisher Scientific, Cat # EO0382), 0.5% v/w Triton X-100, and 0.5% v/w sodium deoxycholate. The lysates were cleared by centrifugation at 20000 × g for 5  min at 4°C. 400 μg of RNA, measured by NanoDrop 2000 (Thermo Fisher Scientific), were loaded onto 10%–50% sucrose gradients. The samples were sedimented by velocity centrifugation at 36000 × rpm for 2 h at 4°C using SW40Ti rotor in Optima L-80XP ultracentrifuge (Beckman Coulter). Absorbance at 254 nm was measured from lower to higher sucrose gradients using an ISCO gradient fractionation system and the optical density at 254 nm was continuously recorded with a Foxy JR Fractionator (Teledyne ISCO).

Detection of ribosome collisions by micrococcal nuclease treatment

HCT116 cells were treated with 5FU for the indicated times. Cells were pre-treated with Cycloheximide (100  μg/ml) for 5  min before harvesting at maximum confluency of 60–70%, collected by centrifugation at 4°C for 5 min and lysed in 500 μl hypotonic buffer containing 5  mM Tris–HCl, pH 7.5, 2.5  mM MgCl₂, 1.5  mM KCl, complete EDTA-free protease inhibitor cocktail, 100  μg/ml Cycloheximide, 2  mM DTT, 200  U/ml RiboLock RNase Inhibitor, 0.5% v/w Triton X-100 and 0.5% v/w sodium deoxycholate. Indicated cells were pre-treated with 15 nM Torin-1 (Sigma, Cat. # 475991) or 25 μM 4EGI-1 (Selleck Chemicals, Cat. # S7369) for 24 h before incubation with 2.5 μM of 5FU for the indicated times. The lysates were cleared by centrifugation at 20000 × g for 5  min at 4°C. CaCl₂ was added to the lysate to a final concentration of 1 mM, and 500 μg of the lysate was digested with 1000 U micrococcal nuclease (NEB, Cat. # M0247S) for 30 min at 22°C. Digestion by micrococcal nuclease was stopped by adding 2 mM EGTA. Afterwards, RNA-equivalent amounts of lysate (were resolved on 10–50% sucrose gradients and absorbance at 254 nm was measured from lower to higher sucrose gradients.

Cell growth assay measurement

HCT116 or SUIT-2 cells were seeded at 25000 cells per well in 6-well plates in complete medium for the indicated times. Cells were trypsinised and stained with Trypan Blue for 2 min and counted under the microscope using Haemocytometer. The process was repeated for 5 days to monitor the comparative cell growth.

Colony formation assay

HCT116 and SUIT-2 cells were seeded at 400 cells per well in 6-well plates. To study differential colony formation after drug treatment, cells were treated with respective doses of 5FU and maintained at 37°C and 5% CO₂. After 7 days, fresh media containing respective doses of 5FU was replenished. After 12 days, cells were rinsed briefly with PBS before staining with colony fixation-staining solution (crystal violet 0.4% w/v in 95% Methanol) for 5 min, followed by two brief washes in ice cold water and left to dry overnight. Afterwards, the number of colonies (diameters > 200 μm) was counted using the Oxford Optronix GelCount™ system (v1.1.2.0; Oxford Optronix).

Dose response assays and IC50 measurement

HCT116 and SUIT-2 cells were seeded at 2000 cells per well in 96-well plates in complete media containing dialyzed FBS. The following day fresh media containing serial dilutions of 5FU (Merck, Cat. # 343922), FUDR (Sigma, Cat. # F0503), FUTP (Biorbyt, Cat. # orb64970) or Gemcitabine (Sigma, Cat. # G6423) were added. 72 h later cell viability was measured using CellTiter-Glo® Luminescent Cell Viability Assay reagent (Promega, Cat. # G7572) according to the manufacturer's protocol. The comparative luminescence was measured on the Synergy Microplate reader (Biotek, Synergy HTX).

Surface sensing of translation (SUnSET) assay

HCT116 and SUIT-2 cells were seeded at 2.5 × 10⁵ cells per well in 6-well plates and treated with 5FU (2.5 μM) for indicated times. Puromycin dihydrochloride (Sigma, Cat. # P8833) was added to the media at 2 μg/ml final concentration for 20 min prior to cell lysis with RIPA buffer. A group pre-treated with 100 μg/ml of cycloheximide for 5 min was used as an additional negative control. 10 μg of cell lysates were loaded in a 10% SDS-PAGE gel, followed by western blotting, and probing with anti-Puromycin antibody. The total signal density was quantified using ImageJ software (NIH).

Pulse labelling with the methionine analogue l-homopropargylglycine

Cells were seeded at 1000 cells per well in 96-well plates and treated with 2.5 μM of 5FU for the indicated time, followed by incubation in methionine-free medium containing the methionine analogue l-homopropargylglycine (HPG) for 30 min. Cells were next fixed with 4% paraformaldehyde for 15 min and permeabilised using 0.5% Triton X-100 solution. The differential level of nascent protein synthesis was quantified using Click-iT™ HPG Alexa Fluor™ 488 Protein Synthesis Assay Kit (Invitrogen, Cat. # C10428) according to the manufacturer's protocol on a BMG FLUOstar Omega Microplate Reader. The signal was normalised against the unstained samples for each set of experiments.

Propidium iodide staining and flow cytometry

Analysis of apoptosis by propidium iodide staining and flow cytometry was performed as described previously (36). Briefly, cells were seeded at 1 × 10⁵ cells per well in 6-well plates and treated with 5FU (2 μM) for indicated times. The cells were scraped and centrifuged at 200 × g for 5 min at room temperature and the pellet was resuspended in 1× PBS. The cells were fixed by resuspending in 70% (v/v) cold ethanol and incubated on ice for 30 min. Fixed cells were washed with 1× PBS and resuspended in DNA staining solution (20 μg/ml of Propidium Iodide (Sigma, Cat. # P4864) and 200 μg/ml of RNase A (Thermo Fisher Scientific, Cat. # EN0531) in PBS for 30 min at room temperature in the dark. The stained cells were analysed by BD LSR II flow cytometry machine (BD Biosciences) as per manufacturers protocol. The percentage of late apoptotic hypodiploid DNA peaks was measured using BD FACSDiva™ software.

Statistical analyses

Statistical analyses were performed using Prism 6 (GraphPad). Error bars represent standard deviation (SD) from the mean of at least three independent replicates and individual data points are depicted in bar graphs. Statistical significance was set a priori at 0.05. Only one observation per sample was collected.

Results

5FU treatment enhances global mRNA translation

Recent studies suggest that 5FU treatment results in downregulation of mRNA translation (4,37). However, these studies predominantly focussed on the impact on mRNA translation following 5FU exposures of longer than 24 h. We postulated that such lengthy exposure to 5FU might confound a dissection of the impact of 5-FU upon translation due to the pleiotropic effects of 5FU, including stress induced by DNA damage upon prolonged 5FU treatment. Consequently, these analyses may not accurately reflect the effect of 5FU on mRNA translation.

Using an antibody able to detect 5-Fluorouridine incorporated into nucleic acids (38,39), we assessed the covalent immobilisation of 5FU metabolites within single cells. We observed a rapid cytoplasmic 5FU localisation within 30 min of treatment, which remained at similar levels up to 24 h (Figure 1A) and was very sensitive to RNase digestion (Supplementary Figure S1A). These data emphasise that 5FU has a substantial and early impact on RNA metabolism and underscore the importance of examining the impact of 5FU on mRNA translation during these early time points. Therefore, we sought to investigate the changes in the mRNA translation profile upon short (<12 h) and long-term (>24 h) treatments of HCT116 (colorectal cancer) and SUIT-2 (pancreatic adenocarcinoma) cells with 5FU. Surface Sensing of Translation (SUnSET) assay, a nonradioactive puromycin/antibody-based tool for quantification of protein synthesis (40), revealed that while prolonged (48 h) 5FU treatment slightly reduced the rate of new protein synthesis in HCT116 and SUIT-2 cells, shorter 5FU treatment (<12 h) surprisingly resulted in increased rate of protein synthesis (Figure 1B and Supplementary Figure S1B). The rapid enhancement in the rate of protein synthesis upon 5FU treatment was corroborated by a pulse-labelling assay with the methionine analogue L-homopropargylglycine (41) (Figure 1C and Supplementary Figure S1C). Furthermore, a polysome profiling assay to analyse the association (loading) of mRNAs with ribosomes revealed a shift from the untranslated or weakly translated sub-polysome region to polysomes within 8 h after 5FU treatment (Figure 1D), which was reversed at 48 h post-treatment (Supplementary Figure S1D). This indicates that mRNA-ribosomes association is enhanced within hours after 5FU treatment.

Figure 1.

5FU treatment enhances global mRNA translation. (A) Left: HCT116 Cells were treated with 5FU (2.5 μM) for 0.5, 4 and 24 h for immunofluorescence analysis of 5FU incorporation. Cells were stained for 5-FU (green) using anti-BrdU antibody and counterstained with the nucleic acid dye DAPI (blue). An untreated control (0 h) was included, which displays low level of background. Right: Graph showing 5FU fluorescence per cell (29 cells for the 0 h and 50 cells were quantified in the 0.5, 4 and 24 h groups) relative to 0 h control. **** P< 0.0001, one way ANOVA with Dunnett's multiple comparisons test. Scale bar = 45 μm. (B) Quantification of new protein synthesis by Surface Sensing of Translation (SUnSET) assay in HCT116 cells treated with 5FU (2.5 μM) for the indicated times. Left: Representatives immunoblot analysis of lysates probed with the indicated antibodies. Right: The bar graph represents the relative change in puromycin/β-actin signal density in each group, measured by ImageJ. Data are presented as mean ± SD; n = 4 independent replicates; **P< 0.01, two-tailed Student's t-test. (C) Quantification of new protein synthesis by pulse labelling with the methionine analogue L-homopropargylglycine (HPG). HCT116 cells were treated with 5FU (2.5 μM) for the indicated times. The signal was normalised against the unstained samples for each group. Data are presented as mean ± SD; n = 3 independent replicates; *P< 0.05, two-tailed Student's t-test. (D) Analysis of general mRNA–ribosome association in HCT116 cells. Left. Cells were treated with 5FU (2.5 μM) for the indicated times before harvesting and analysis using polysome profiling assay. A shift from the sub-polysome to polysomes fractions indicates enhanced mRNAs-ribosomes association. Right. Quantification of the polysome/sub-polysome ratio in control and 5FU treated cells. Data are presented as mean ± SD; n = 3 independent replicates; **P< 0.01, two-tailed Student's t-test. (E) Western blot analysis of effects of 5FU treatment (2.5 μM) for the indicated times on the signalling pathways regulating mRNA translation machinery in HCT116 cells. To avoid excessive non-specific signals due to repeated re-probing, identical samples were run on separate gels. β-Actin was used as loading control for each membrane. Quantification of the pS6 and p4E-BP1 expression is presented in Supplementary Figure S1E and F. Also see the relatedSupplementary Figure S1.

We next investigated the impacts of 5FU treatment on core mechanisms of regulation of mRNA translation initiation. 5FU treatment markedly and rapidly (<1 h) induced phosphorylation of RPS6 (S240/244) and 4E-BP1 (T37/46), key markers of mTORC1 activity, which decreased with longer (>12 h) treatment (Figure 1E and Supplementary Figure S1E–G). Notably, 5FU treatment did not activate the Integrated Stress Response (ISR) pathway (indicated by eIF2α phosphorylation on S51 residue) or the MNK-regulated eIF4E phosphorylation on S209 (Figure 1E and Supplementary Figure S1G). Altogether, these data suggest that 5FU-treatment rapidly enhances global mRNA translation via activation of the mTORC1 pathway.

RQC resolves the 5FU-induced ribosome collisions

We hypothesised that the enhanced rate of mRNA translation (Figure 1), combined with the known impacts of 5FU on production of abnormal RNAs (2), post-transcriptional modifications (7), alteration of ribosome functions (8), and aberrant translation events (e.g. stop-codon readthrough (4,8)) could lead to ribosome collisions and triggering of the RQC mechanism. We observed an increased mono-ubiquitination of ribosomal small subunit protein eS10, a marker of ribosome collision and activation of RQC (19), upon 5FU treatment (Figure 2A). The RNA-binding protein and E3 ubiquitin ligase ZNF598 plays a key role in detection of ribosome collisions and triggering RQC, and its deletion abrogates RQC (19,27,28). To investigate the role of RQC in the cellular response to 5FU, we generated ZNF598-knockout (KO) HCT116 and SUIT-2 cells with CRISPR-Cas9 (Supplementary Figure S2A). We observed that unlike in the parental cells, where 5-FU treatment triggered eS10 mono-ubiquitination, this effect was absent in ZNF598-KO cells (Figure 2B and Supplementary Figure S2B). Considering the observed increased general mRNA translation and protein synthesis in the first few hours post 5FU treatment (Figure 1B–D), we next assessed the impact of ZNF598-KO on 5FU-induced mRNA translation. Polysome profiling revealed that, similar to the parental cells (Figure 1D), mRNA-ribosome association in ZNF598-KO HCT116 cells was increased at 6 h post 5FU treatment (Figure 2C). However, pulse-labelling with the methionine analogue L-homopropargylglycine for analysis of nascent protein synthesis showed that compared with the parental cells, ZNF598-KO diminished the 5FU-induced protein synthesis (Figure 2D). We hypothesised that this discrepancy in 5FU-induced mRNA-ribosome association (detected via polysome profiling) versus protein synthesis (observed in pulse labelling) in ZNF598-KO cells is due to the inability of the RQC-deficient cells to resolve 5FU-induced ribosome collisions. This deficiency results in accumulation of non-productive stalled ribosomes in polysomes and reduction of protein synthesis in ZNF598-KO cells. To test this, we assessed the occurrence of ribosome collisions by Micrococcal Nuclease (MNase) digestion assay in 5FU-treated parental and ZNF598-KO cells. MNase degrades the mRNA between ribosomes, thereby releasing 80S ribosomes. However, when ribosome collisions occur, the inter-ribosomal mRNA of the collided ribosomes (disomes and trisomes) is protected from MNase digestion (42). We observed that while the untreated ZNF598-KO cells accumulate only a small amount of disomes compared with their parental counterparts (Figure 2E and Supplementary Figure S2C, D), 5FU treatment results in a clear accumulation of disome and trisome peaks in ZNF598-KO but not in parental cells (Figure 2F and Supplementary Figure S2E, F). These data further indicate an enhanced rate of ribosome collisions upon 5FU treatment that are readily resolved by RQC mechanism in parental cells, whereas RQC-deficient ZNF598-KO cells are impaired in their ability to resolve the accumulated collided ribosomes.

Figure 2.

ZNF598 resolves 5FU induced collided ribosomes. (A) Western blot analysis of mono-ubiquitination of eS10 upon 5FU treatment (2.5 μM) for the indicated time in HCT116 cells. β-actin was used as a loading control. Treatment with Anisomycin (5 μg/ml) for 1 h was used as a positive control. (B) Left. Western blot analysis of mono-ubiquitination of eS10 in parental and ZNF598-KO HCT116 cells upon 5FU treatment (2.5 μM) for 1 h. β-actin was used as loading control. Right: Densitometric quantitation of the mono-ubiquitinated eS10/total eS10. Data are normalised against the respective control for each replicate and presented as mean ± SD; n = 4 independent replicates; *P< 0.05, two-tailed Student's t-test. (C) Left: Polysome profiling analysis of general mRNA–ribosome association in in ZNF598-KO HCT116 cells treated with 5FU (2.5 μM) for 6 h or vehicle. Right: Quantification of the polysome/sub-polysome ratio in vehicle and 5FU treated cells. Data are normalised against the respective control for each replicate and presented as mean ± SD; n = 4 independent replicates; *P< 0.05, two-tailed Student's t-test. (D) Quantification of new protein synthesis by pulse labelling with the methionine analogue l-homopropargylglycine in parental and ZNF598-KO HCT116 cells treated with 5FU (2.5 μM) for 4 h. Data are normalised against the respective control for each replicate and presented as mean ± SD; n = 3 independent replicates; **P< 0.01, two-tailed Student's t-test. (E) Left: Assessment of ribosome collisions in parental and ZNF598-KO HCT116 cells by polysome profiling following micrococcal nuclease (MNase) digestion of cell lysates. Right: Quantification of the disome/monosome ratio in each condition. Data are normalised against the respective control for each replicate and presented as mean ± SD; n = 3 independent replicates; ns = non-significant; two-tailed Student's t-test. (F) Left: MNase digestion in parental and ZNF598-KO HCT116 cells treated with 5FU (2.5 μM) for 12 h. Right: Quantification of the disome/monosome ratio in each condition. Data are normalised against the respective control for each replicate and presented as mean ± SD; n = 3 independent replicates; **P< 0.01, two-tailed Student's t-test. Also see the relatedSupplementary Figure S2 .

ZNF598 deficiency enhances the sensitivity of cancer cells to 5FU

Based on our findings, we hypothesised that the ability of the cells to resolve 5FU-induced collided ribosomes by RQC may mitigate against 5FU-induced lethality in cancer cells. Our initial characterisation of the ZNF598-KO HCT116 and SUIT-2 cells revealed that ZNF598-KO resulted in a slightly reduced growth rate (Figure 3A and Supplementary Figure S3A). Importantly, ZNF598-KO markedly increased the 5FU sensitivity of HCT116 (IC₅₀= 0.73 μM and 0.19 μM in parental and ZNF598-KO, respectively; Figure 3B) and SUIT-2 cells (IC₅₀= 1.36 μM and 0.23 μM in parental and ZNF598-KO respectively; Supplementary Figure S3B). Similar results were observed upon depletion of ZNF598 expression by three different shRNAs in HCT116 cells (Supplementary Figure S3C). Enhanced 5FU-induced cell death upon ZNF598-KO was also corroborated by impaired clonogenic survival (Figure 3C and Supplementary Figure S3D). Moreover, these enhanced growth inhibitory effects correlated with elevated 5FU-induced apoptosis, assessed by PARP cleavage (Figure 3D & Supplementary Figure S3E) and accumulation of sub-G1 population in propidium iodide (PI) stained ZNF598-KO cells (Supplementary Figure S3F). Previous reports suggested that increased levels of unresolved collided ribosomes in ZNF598-KO cells RQC leads to activation of the Ribotoxic Stress Response (RSR) and apoptosis via ZAKα/p38 pathway (25). We noted that while 5FU treatment did not induce p38 phosphorylation (a marker of RSR activation) in parental cells, RQC-deficient ZNF598-KO cells exhibited distinctively increased p38 phosphorylation upon 5FU treatment (Supplementary Figure S3G), which may at least partially explain the mechanism of enhanced 5FU-induced toxicity downstream of the accumulated collided ribosomes in ZNF598-KO cells. These data strongly suggest a cytoprotective role for ZNF598/RQC in cellular response to 5FU, creating a dependency on RQC to mitigate the cytotoxic impacts of 5FU.

Figure 3.

ZNF598 deficiency enhances the sensitivity of cancer cells to 5FU. (A) Cell growth assay with parental and ZNF598-KO HCT116 cells after the indicated times. Data are shown as mean ± SD; n = 3 independent replicates; *P< 0.05, two-tailed Student's t-test. (B) Dose-response assay for measurement of sensitivity of parental and ZNF598-KO HCT116 cells to 5FU. Cell viability was measured 72 h post-treatment using CellTiter-Glo® luminescent cell viability assay. (C) Left: Colony formation assay with parental and ZNF598-KO HCT116 cells 12 days post-seeding. Cells were treated with the indicated concentration of 5FU. Right: Quantification of number of colonies with a diameter >200 μm in each well. Data are presented as mean ± SD; n = 3 independent replicates; *P< 0.05; **P< 0.01; ***P< 0.001, two-tailed Student's t-test. (D) Western blot analysis of expression of cleaved PARP (c-PARP) using lysates derived from parental and ZNF598-KO HCT116 cells treated with the indicated doses of 5FU for 72 h. β-actin was used as loading control. (E–G) Dose-response assay for measurement of sensitivity of parental and ZNF598-KO HCT116 cells to (E) FUDR, the precursor of the DNA-incorporating 5FU metabolite, (F) RNA-incorporating 5FU metabolite FUTP, and (G) Gemcitabine. 72 h post-treatment, cell viability was measured using CellTiter-Glo® luminescent cell viability assay. Also see the relatedSupplementary Figure S3.

As noted above, 5FU-derived metabolites can exert their cytotoxic effects via three potential mechanisms: the misincorporation of FUTP into RNAs; the misincorporation of FdUTP into DNA; or via FdUMP, which inhibits thymidylate synthase. To test a role for the latter two pathways in the increased sensitivity of ZNF598-KO cells to 5FU, we compared the effects of ZNF598-KO on cell viability in response to treatment with FUDR, a precursor of FdUMP and FdUTP. ZNF598-KO resulted in a ∼2-fold increased sensitivity to FUDR (Figure 3E and Supplementary Figure S3H). However, this effect could, at least partly, be due to the metabolism of FUDR that, besides generation of FdUMP and FdUTP, can also give rise to FUTP. In comparison, ZNF598-KO cells were even more robustly (>3-fold) sensitised to FUTP, which only misincorporates into RNAs (Figure 3F and Supplementary Figure S3I). Notably, ZNF598-depletion did not have an impact on sensitivity to Gemcitabine (Figure 3G & Supplementary Figure S3J), another nucleic-acid-incorporating chemotherapeutic reagent (43). These data indicate that the RQC machinery has a selective role in mitigating the cellular response to 5FU-induced toxicity, which can be uncoupled from its incorporation into DNA or thymidylate synthase inhibition and the mechanisms of action of other RNA-incorporating antimetabolites.

Inhibition of translation initiation reduces 5FU-induced ribosome collision and cellular dependency on RQC

We further sought to investigate the underlying mechanism of 5FU-induced ribosomal collision. Our earlier observations (Figure 1B-E) revealed increased mRNA translation initiation and mTORC1 activity upon 5FU treatment. It has been previously reported that increased translation initiation could exacerbate ribosome collisions as it leads to an increased presence of trailing ribosomes that collide with a stalled ribosome (21,44,45). Considering the key role of mTORC1 in regulation of translation initiation (12), we investigated the impact of blocking mTOR activity on 5FU-induced ribosome collisions. Treatment with the mTOR inhibitor Torin-1 decreased eS10 ubiquitination in the parental HCT116 cells in response to 5FU treatment (Figure 4A) and reduced 5FU-induced accumulation of disomes and trisomes in ZNF598-KO cells (Figure 4B). We reasoned that if 5FU-induced ribosome stalling contributes to cytotoxicity, the mTOR-dependent reduction in ribosome stalling would alter cell fate. Indeed, we observed that mTOR inhibition reverses the sensitisation to 5FU seen in ZNF598-KO cells (Figure 4C). Importantly, mTOR has multiple downstream effectors, which regulate several processes other than mRNA translation (13). Thus, we sought to use orthogonal means to verify that our observation of reduced 5FU sensitivity in the RQC-deficient cells upon mTOR inhibition is due to repression of translation initiation. We used the eIF4E/eIF4G interaction inhibitor 1 (4EGI-1), which blocks cap-dependent translation initiation by dissociation of the eIF4F complex (46). Corroborating our findings with mTOR inhibition, 4EGI-1 treatment also significantly reduced the accumulation of collided ribosomes (disomes and trisomes) in 5FU-treated ZNF598-KO cells (Figure 4D) and alleviated the enhanced 5FU sensitivity of the ZNF598-KO cells compared with the parental cells (Figure 4E).

Figure 4.

Inhibition of translation initiation reduces the 5FU-induced ribosome collisions and cell death in ZNF598-deficient cells. (A) Top: Western blot analysis of mono-ubiquitination of eS10 in lysates derived from HCT116 cells pre-treated with Torin-1 (15 nM) or vehicle for 24 h, followed by treatment with 5FU (2.5 μM) for 1 h. β-actin was used as loading control. Bottom: Densitometric quantitation of the mono-ubiquitinated eS10/total eS10. Data are normalised against the respective control for each replicate and presented as mean ± SD; n = 3 independent replicates; *P< 0.05, two-tailed Student's t-test. (B) Left: MNase digestion in ZNF598-KO HCT116 cells pre-treated with Torin-1 (15 nM) or vehicle for 24 h, followed by treatment with 5FU (2.5 μM) for 12 h. Right: Quantitation of the disome/monosome ratios. Data are normalised against the respective control for each replicate and presented as mean ± SD; n = 3 independent replicates; **P< 0.01, two-tailed Student's t-test. (C) Dose-response assay for measurement of 5FU sensitivity of parental and ZNF598-KO HCT116 cells in the presence of Torin-1(15 nM). 72 h post-treatment cell viability was measured using CellTiter-Glo® luminescent cell viability assay. (D) Left. MNase digestion in ZNF598-KO HCT116 cells pre-treated with 4EGI-1 (25 μM) or vehicle for 24 h, followed by treatment with 5FU (2.5 μM) for 12 h. Right. Quantification of the disome/monosome ratio. Data are normalised against the respective control for each replicate and presented as mean ± SD; n = 3 independent replicates; *P< 0.05, two-tailed Student's t-test. (E) Dose-response assay for measurement of 5FU sensitivity of parental and ZNF598-KO HCT116 cells in the presence of 4EGI-1 (25 μM). 72 h post-treatment cell viability was measured using CellTiter-Glo® luminescent cell viability assay.

Altogether, these data demonstrate that the enhanced rate of mRNA translation initiation upon 5FU treatment contributes to the increased ribosome collisions in 5FU-treated cells and the elevated 5FU-induced cell death in RQC-deficient cells. Thus, blocking mTOR activity or translation initiation reduces the risk of ribosome collisions in 5FU-treated cells and thereby alleviates the requirement for a functional RQC to prevent the cytotoxic consequences of collided ribosomes.

mTOR-mediated upregulation of key RQC factors ZNF598 and GIGYF2 by 5FU

Considering the cyto-protective role of ZNF598/RQC against 5FU-treatment, we next sought to investigate the impact of 5FU on expression of ZNF598 and several other factors involved in detection of ribosome collision and triggering RQC, namely EDF1, ASCC3, RACK1 and GIGYF2 (26). 5FU treatment increased the expression of ZNF598 and GIGYF2, a key factor in ZNF598- and EDF1-mediated translational repression of the mRNAs with collided ribosomes (29,30) and to a lesser extent ASCC3, but not RACK1 or EDF1 (Figure 5A and Supplementary Figure S4A-C). Further inspection revealed a remarkably rapid increase in expression of ZNF598 and GIGYF2 within 30 min of 5FU treatment (Figure 5B and Supplementary Figure S4D, E). RT-qPCR measurement found no significant change in ZNF598 or GIGYF2 mRNA levels within 12 h post 5FU treatment (Figure 5C and Supplementary Figure S4F), which suggests involvement of a post-transcriptional mechanism of regulation. To further dissect this mechanism, we assessed the role of translational regulation in increased expression of ZNF598 and GIGYF2. We pre-treated the cells with the translation inhibitor Cycloheximide (CHX), followed by treatment with 5FU. Whereas abundance of ZNF598, GIGYF2 and EDF1 was decreased in vehicle treated cells in the presence of CHX (Supplementary Figure S4G), 5FU-treatment increased the accumulation of ZNF598 and GIGYF2, even in presence of CHX, while EDF1 abundance remained unchanged (Figure 5D and Supplementary Figure S4H, I). These data suggest the presence of a post-translational mechanism of regulation that rapidly increases the abundance of the key RQC factors ZNF598 and GIGYF2 in response to 5FU treatment. Importantly, previous studies with ribosome stalling reporters and eS10 mono-ubiquitination suggested that the amount of endogenous ZNF598 is a limiting factor during RQC; consequently, ZNF598 overexpression significantly increased the rate of RQC (19,47). Therefore, the augmented expression of this key rate-limiting enzyme could further enhance the ability of the cell to cope with the increased rate of ribosome stalling/collisions upon 5FU treatment.

Figure 5.

mTOR-dependent upregulation of expression of ZNF598 and GIGYF2 by 5FU-derived metabolites and UTP. (A) Western blot analysis of expression of indicated proteins in lysates derived from HCT116 cells treated with 5FU (2.5 μM) for the indicated times. Quantification of the ZNF598 and GIGYF2 protein expression is presented in Supplementary Figure S4A and B. (B) Western blot analysis of expression of the indicated proteins in lysates derived from HCT116 cells treated with 5FU (2.5 μM) for the indicated times. Quantification of the ZNF598 and GIGYF2 protein expression is presented in Supplementary Figure S4D and E. (C) Quantitative RT-PCR analysis of expression of ZNF598 and GIGYF2 mRNAs in HCT116 cells upon 5FU treatment (2.5 μM) for the indicated times. Data are presented as mean ± SD; n = 3 independent replicates; ns = non-significant; two-tailed Student's t-test. (D) Western blot analysis of expression of the indicated proteins in lysates derived from HCT116 cells pre-treated with 5FU (2.5 μM) for 1 h followed by incubation for the indicated times in the presence or absence of 100 μg/ml Cycloheximide (CHX). Quantification of the ZNF598 and GIGYF2 protein expression is presented in Supplementary Figure S4H and I. (E) Western blot analysis of expression of the indicated proteins in lysates derived from HCT116 cells pre-treated with Torin-1 (15 nM) or vehicle for 24 h, followed by treatment with 5FU (2.5 μM) for the indicated times. Quantification of the ZNF598 and GIGYF2 protein expression is presented in Supplementary Figure S5C and D. (F, G) Western blot analysis of lysates derived from HCT116 cells treated with (F) FUTP (1 μM) or (G) UTP (1 μM) for the indicated times. Quantification of the ZNF598 and GIGYF2 protein expression is presented in Supplementary Figure S5E and F (FUTP) and Supplementary Figure S5I and J (UTP). (H) Western blot analysis of lysates derived from HCT116 cells treated with the small molecule mTOR activator MHY1485 (200 nM) for the indicated times. Quantification of the ZNF598 and GIGYF2 protein expression is presented in Supplementary Figure S5K and L. (I) Top. Western blot analysis of lysates derived from HCT116 cells treated with the small molecule mTOR activator MHY1485 (200 nM) for the indicated times. Bottom. Densitometric quantitation of the mono-ubiquitinated eS10 / total eS10. Data are normalised against the respective control for each replicate and presented as mean ± SD; n = 3 independent replicates; ns = non-significant; **P< 0.01, two-tailed Student's t-test. Treatment with Anisomycin (5 μg/ml for 1 h) was used as a positive control. Also see the relatedSupplementary Figures S4 and S5.

Notably, we observed that the upregulation of ZNF598 and GIGYF2 expression upon 5FU treatment follows the rapid activation of mTORC1 signalling (Figure 5B and Supplementary Figure S5A). Therefore, we assessed the possible role of the mTOR pathway in this process. Treatment with mTOR inhibitor Torin-1, resulted in a dose-dependent reduction in ZNF598 and GIGYF2 expression (Supplementary Figure S5B). Furthermore, pre-treatment with Torin-1 abrogated the 5FU-induced increase in ZNF598 and GIGYF2 expression (Figure 5E, Supplementary Figure S5C, D). In addition, similar to 5FU, treatment with FUTP (Figure 5F & Supplementary Figure S5E, F), as well as FUDR (Supplementary Figure S5G), also induced activation of the mTORC1 pathway and upregulation of ZNF598 and GIGYF2. This suggests that activation of mTOR pathway and upregulation of ZNF598 and GIGYF2 expression is a shared feature of 5FU-derived metabolites and can be uncoupled from the ability of these metabolites to incorporate into RNA and DNA, or to inhibit thymidylate synthase. Importantly, we also observed that, as was reported previously (48), treatment with non-fluorinated UTP also rapidly increased mTORC1 activity and expression of ZNF598 and GIGYF2 (Figure 5G and Supplementary Figure S5H–J). This indicates the presence of a shared mechanism of upregulation of mTOR activity among uridine derivatives, regardless of their fluorination status. We next examined if upregulation of ZNF598 and GIGYF2 is restricted to the conditions where mTOR is activated by uridine derivatives. We observed that chemical activation of mTOR with MHY1485 (49) in the absence of exogenous uridine-derived metabolites also increased ZNF598 and GIGYF2 expression (Figure 5H and Supplementary Figure S5K, L) and mono-ubiquitination of eS10 (Figure I). This indicates that the MHY1485-treated cells (i.e. cells in which mTOR activity is stimulated) are more capable of triggering RQC and therefore dealing with the stalled/collided ribosomes.

Taken together, these data suggest the presence of an intrinsic mechanism of regulation of ZNF598 and GIGYF2 following mTOR activation that serves to avert the potential accumulation of collided ribosomes during mRNA translation activation. In the context of cancers such as CRC, wherein mRNA translation rate is elevated (50,51), this intrinsic cellular mechanism could alleviate the cytotoxic repercussions of mTORC1 activation during 5FU treatment (Figure 6A). Notably, using publicly available datasets for protein expression in normal and cancerous tissues (52), we observed increased expression of ZNF598, GIGYF2 and ASCC3 proteins in CRC patient samples (Figure 6B–D).

Figure 6.

mTOR couples the elevated mRNA translation initiation and increased rate of ribosome collisions with enhanced expression of RQC factors in 5FU-treated cells. (A) Proposed model for mechanism of 5FU-induced enhanced ribosome collisions and modulation of RQC activity. 5FU rapidly enhances mTOR activity by an unknown mechanism, leading to increased translation initiation via phosphorylation of factors such as 4E-BPs. Increased translation initiation elevates the risk of ribosome collisions. Fluorinated Uridines may also increase the risk of ribosome stalling through direct incorporation into the mRNA ORF or non-coding RNAs such as rRNAs, thereby further increasing the chances of ribosome collision. In parallel, activation of mTOR by 5FU also leads to stabilisation of ZNF598 and GIGYF2, which further bolsters RQC mechanism and contributes to mitigating the cytotoxic consequences of 5FU-induced ribosome stalling and collisions. Image was generated with BioRender.com. (B-D) Protein expression levels of (B) ZNF598, (C) GIGYF2, and (D) ASCC3 in normal colon and colon adenocarcinoma samples. Data are extracted from The National Cancer Institute's Clinical Proteomic Tumor Analysis Consortium (CPTAC) (52) using UALCAN data analysis platform. Z-values (number of standard deviations away from the mean) as well as Pvalues are annotated for each protein.

Discussion

Driven by the misconception that the transient nature of RNAs renders damage to them therapeutically inconsequential, the potential role of mechanisms such as RQC in the response of cancer cells to chemotherapy has been largely overlooked. Herein, we described a hitherto unknown mechanism of RNA-dependent 5FU toxicity, a drug which as noted above, incorporates into RNAs considerably more efficiently than it does into DNA.

Our study reveals a critical role for the mRNA translation machinery, mTOR signalling pathway, and RQC, a homeostatic mechanism of monitoring mRNAs to ensure translation integrity, in shaping the cellular response to 5FU. Recent studies suggest that 5FU treatment results in downregulation of mRNA translation (4,37), although upregulated translation of a subset of mRNAs associated with cell survival has also been observed after 24 hours of 5FU treatment (8). We provide unequivocal data which show that exposure to 5FU rapidly activates the mRNA translation machinery. This effect, previously obscured due to a tendency of the field to focus on the outcomes of 5FU cytotoxicity in long-term treatments, underscores the nuanced and dynamic influence of 5FU on mRNA translation. Importantly, we also discovered that 5FU treatment elevated the rate of ribosome collisions. Furthermore, we demonstrate that 5FU and its metabolites, as well as the non-fluorinated UTP, activate the mTOR pathway, which orchestrates the upregulation of key RQC factors ZNF598 and GIGYF2, along with the elevated rate of translation initiation and ribosome collisions. The RQC mechanism, activated in response to 5FU, efficiently resolves collided ribosomes and acts as a protective mechanism that abrogates the long-term 5FU-induced toxicity. Conversely, a defective RQC leads to accumulation of 5FU-induced collided ribosomes and enhanced 5FU-induced cell death. This study is the first indication that ribosome collisions can be induced by a widely employed chemotherapeutic agent and that the RQC pathway determines the efficacy of drug-induced cell death. This identifies 2 million people annually, where drug treatment-related modulation of RQC occurs, with an obvious potential for combination therapy.

We provide robust evidence that 5FU treatment causes ribosome collisions. However, the exact mechanism by which 5FU induces ribosome stalling is still not clear. Our data revealed that through dysregulation of key signalling pathways, namely activation of mTORC1, 5FU treatment leads to enhanced mRNA translation initiation. This can pose an increased risk of ribosome collisions by elevating the ribosomal traffic on faulty mRNAs (e.g. aberrantly spliced), combined with aberrant translation events (e.g. stop-codon readthrough), which are induced by 5FU (2,4,8) and could potentially lead to ribosome stalling (29). Interestingly, a recent report demonstrated that treatment with DNA damaging agents such as Etoposides leads to the transfer RNase SLFN11-mediated cleavage and reduced abundance of tRNA^UUA, leading to ribosome stalling and triggering of the ZAKα-mediated RSR and cell death (53). A similar mechanism following 5-FU mediated DNA damage may also contribute to ribosome stalling and the triggering of RQC. Furthermore, it is plausible that direct incorporation of 5FU metabolite FUTP in the open-reading frame (ORF) of the mRNAs or 5FU-induced changes in nucleotide modifications within the ORF could also impede ribosomes and result in stalling. While the ability of fluorinated uridines to directly impede ribosomes is not known, other types of modified uridine within the ORF including pseudouridine (16) and 5-methyluridine (54) were shown to stall ribosomes and trigger RQC (17). Incorporation of 5FU-derived metabolites into non-coding RNAs, such as ribosomal RNAs (rRNAs) or transfer RNAs (tRNAs) may also contribute to the 5FU-induced ribosome stalling. Recent studies demonstrated the significant effects of incorporation of 5FU-derived metabolites into rRNAs on ribosome biogenesis and function (5,8). In parallel, 5FU also alters tRNA modifications, stability, and function (7,55,56), events which have, in other contexts, been linked to ribosome frameshifting and stalling (57,58). While further studies are required to deduce the exact mechanism by which 5FU induces ribosome stalling, our data clearly demonstrates that the enhanced rate of translation initiation induced by 5FU further exacerbates the rate of ribosome collisions. Consistent with this notion, blocking translation initiation with 4EGI-1 or Torin-1 markedly decreased the rate of ribosome collisions, presumably by lowering the ribosomal traffic and reducing the risk of collisions with stalled ribosomes. Previous studies also demonstrated that blocking mTORC1-mediated translation attenuates the Pelo knockout RQC-deficiency phenotypes in mouse epidermal stem cells (31), while 4EGI-1 treatment reduced ribosome collisions and ribotoxic stress associated with Huntington's disease (59).

RQC is a translation-dependent mechanism of quality control and more actively translated mRNAs are more prone to ribosome collisions and triggering RQC (20–22), due to the higher chance of presence of trailing ribosomes that collide into a stalled ribosome. Metabolic stress and oxidative agents such as nitric oxide and reactive oxygen species, which are common in cancer cells and directly damage ribonucleotide bases, impede the transition of translating ribosomes and cause ribosome stalling (14,15,60,61). Furthermore, following oncogenic transformation, overall translation activity is often elevated, partly due to mutation and activation of signalling pathways such as PI3K/mTOR (62), which stimulate mRNA translation (12). Consequently, cancers such as CRC that are reliant on elevated mRNA translation for tumour growth (50,51) may exhibit dependency on intact RQC to mitigate the detrimental impacts of ribosome collisions. Our data demonstrate the synergistic enhancement of 5FU toxicity upon depletion of ZNF598 in two cell lines derived from two cancers treated with 5FU-based chemotherapy (colorectal cancer and pancreatic ductal adenocarcinoma). This provides a promising opportunity for enhancement of the 5FU anticancer efficacy by targeting the RQC machinery. However, further studies are required to fully characterise the impacts of genetics or pharmacological inhibition of various RQC factors on the response of cancer as well as non-cancer cells to 5FU-based standard of care treatments. Importantly, our data also indicates that repression of mTOR activity and mRNA translation may render RQC-deficient cells less susceptible to the 5FU-derived ribosome collisions and cell death. While the precise impact of mTOR inhibition on 5FU efficacy is disputed (5,63,64), our findings provides important insight into the potential cancer subtypes (i.e. high mTOR activity and mRNA translation) that may further benefit from a putative treatment that combines 5FU with RQC inhibition.

Significant progress has been made in our understanding of the mechanism of RQC (26), its important role in maintenance of homeostasis (31,32) and triggering global reprogramming of gene expression in response to stress (15,25,32,60,65). Yet, important gaps exist in our understanding of how the RQC mechanism itself is regulated in response to environmental stimuli that impact cellular metabolism and mRNA translation, including the antimetabolites such as 5FU. Our data indicate the presence of a hitherto unknown cellular mechanism that increases expression of key RQC factors ZNF598 and GIGYF2 upon 5FU treatment and activation of mTOR signalling pathway. While the exact mechanism by which mTOR regulates the expression of ZNF598 and GIGYF2 is not completely understood, our data indicate a post-translational mechanism of regulation of protein stability (Figure 5). Notably, a recent study suggested that mitochondrial stress, which is triggered upon 5FU treatment in vitro and in vivo (66–68), leads to increased ZNF598 stability (69). However, it is not clear how mitochondrial stress and/or mTOR activation could lead to ZNF598 or GIGYF2 deubiquitination and stabilisation upon 5FU treatment. Potential mechanisms include deactivation of an E3 ligase (plausibly the E3 activity of ZNF598 itself) or activation of a deubiquitinating (DUB) enzyme that leads to deubiquitylation and stabilisation of these proteins. Interestingly, depletion of USP9X—a DUB which reportedly enhances RQC by deubiquitylation and stabilisation of ZNF598 (70) - has been linked to increased sensitivity of CRC cells to 5FU (71,72).

The mTOR pathway functions as a metabolic rheostat that integrates cellular response to various intracellular and extracellular signals (e.g. nutrients) (73). Our data clearly demonstrate a robust and rapid, albeit temporal, activation of mTOR pathway upon treatment with 5FU and its metabolites, as well as the non-fluorinated UTP. Although the important role of mTOR in sensing the intracellular levels of purines has been demonstrated (74), the available information on the mechanisms of monitoring the cellular pool of pyrimidines is scant. Thus, the mechanism by which 5FU and UTP lead to the activation of mTOR pathway remains to be understood. Interestingly, previous studies suggested that ribosome stalling (31,75) as well as impaired ribosomal biogenesis (76) also activate mTORC1 signalling and stimulate translation initiation. It is therefore plausible that the impact of 5FU on ribosome biogenesis (5) and the induction of ribosome stalling, as discussed above, may at least partially explain the rapid activation of mTOR signalling pathway by 5FU.

mTOR kinase forms two distinct protein complexes, mTORC1 and mTORC2, the former of which plays a pivotal role in regulation of mRNA translation. This is achieved via phosphorylation and regulation of activity of a group of translation factors and RNA-binding proteins, including 4E-BPs, S6Ks, and LARP1 (12). Active mTORC1 particularly favours translation of the highly abundant and efficiently translated nuclear encoded mitochondrial proteins (77) and terminal oligopyrimidine (TOP) mRNAs (78,79). As noted above, actively translated mRNAs are more prone to ribosome collisions and triggering RQC (20–22). We propose that the swift stabilisation of ZNF598 and GIGYF2 proteins following mTOR activation serves as a mechanism to avert the potential accumulation of collided ribosomes during mRNA translation upregulation. In the context of cancers such as CRC wherein mTOR activity is frequently upregulated (80), this intrinsic cellular mechanism likely aids cancer cells to alleviate the cytotoxic repercussions of mTORC1 activation during 5FU treatment. Our examination of publicly available datasets revealed heightened expression of ZNF598, GIGYF2, and ASCC3 in CRC in comparison to non-tumour tissues. Further investigations are essential to elucidate the impact of mTOR activation, such as mutations in components of the PI3K/mTOR pathway, on the increased expression of these factors and their role in tumour maintenance.

Altogether, we demonstrate that contrary to previous assumptions, 5FU treatment leads to rapid activation of mRNA translation as well as induction of ribosome collisions. We show the critical role of the RQC mechanism in resolving the 5FU-induced ribosome collisions and mitigating their deleterious consequences, including a profound impact on cellular viability and clonogenic survival in response to this keystone chemotherapeutic. We showed evidence that 5FU treatment of ZNF598-KO cells leads to activation of RSR, which may at least partially explain the mechanism of enhanced toxicity in the 5FU-treated RQC-deficient cells. However, the unresolved collided ribosomes in 5FU treated ZNF598-KO cells could also lead to cell death via other potential downstream effectors, such as ISR (81) or cGAS-STING pathway (82). Deducing the precise mechanism of enhanced cell death caused by accumulation of the collided ribosomes upon 5FU treatment of the RQC-deficient cells will require further studies. This report presents compelling evidence pointing to the existence of a previously unrecognised mechanism of coupling mTOR-mediated activation of mRNA translation to the increased expression of key RQC factors. This mechanism may hold broad significance across various biological and pathological contexts involving mTOR-mediated upregulation of mRNA translation, including development, differentiation, and tumorigenesis. Besides providing fundamental insights into the mechanisms of regulation of RQC activity, this work elucidates a mechanism of cellular response to the RNA-dependent toxicity of 5FU, the mainstay component of the chemotherapeutic treatments for several common and highly lethal cancers. Thus, our findings may lead to potentially significant improvement in utility of 5FU-based anti-cancer treatments by leveraging the cancer cell dependency on RQC.

Data availability

All data are incorporated into the article and its online supplementary material.

Supplementary data

Supplementary Data are available at NAR Online.

Funding

Academy of Medical Sciences Springboard Award [SBF007\100026]; Royal Society Research Grant [RGS\R1\221075]; Patrick Johnston Research Fellowship (to S.M.J.); Royal Society Research Grant [RGS\R2\222149]; Division of Cancer Sciences, University of Manchester (to J.R.P.K.); T.M. is supported by a PhD studentship from Brainwaves Northern Ireland; O.O is supported by a fellowship from the Scientific and Technological Research Council of Turkey (TUBITAK); Z.V.S. is supported by a PhD studentship from The Christie Charity. Funding for open access charge: Queen's University Belfast is part of the Read & Publish agreement and will cover the cost of publication.

Conflict of interest statement. None declared.