Cellular responses to halofuginone reveal a vulnerability of the GCN2 branch of the integrated stress response
2022; Springer Nature; Volume: 41; Issue: 11 Linguagem: Inglês
10.15252/embj.2021109985
ISSN1460-2075
AutoresAleksandra P. Pitera, María Szaruga, Sew‐Yeu Peak‐Chew, Steven W. Wingett, Anne Bertolotti,
Tópico(s)Phenothiazines and Benzothiazines Synthesis and Activities
ResumoArticle25 April 2022Open Access Transparent process Cellular responses to halofuginone reveal a vulnerability of the GCN2 branch of the integrated stress response Aleksandra P Pitera Aleksandra P Pitera orcid.org/0000-0001-7004-1615 MRC Laboratory of Molecular Biology, Cambridge, UK Contribution: Conceptualization, Formal analysis, Validation, Investigation, Visualization, Writing - review & editing Search for more papers by this author Maria Szaruga Maria Szaruga orcid.org/0000-0003-1673-1855 MRC Laboratory of Molecular Biology, Cambridge, UK Contribution: Formal analysis, Validation, Investigation, Visualization, Writing - review & editing Search for more papers by this author Sew-Yeu Peak-Chew Sew-Yeu Peak-Chew orcid.org/0000-0002-7602-6384 MRC Laboratory of Molecular Biology, Cambridge, UK Contribution: Investigation, Writing - review & editing Search for more papers by this author Steven W Wingett Steven W Wingett orcid.org/0000-0002-2343-0711 MRC Laboratory of Molecular Biology, Cambridge, UK Contribution: Software, Formal analysis, Writing - review & editing Search for more papers by this author Anne Bertolotti Corresponding Author Anne Bertolotti [email protected] orcid.org/0000-0002-9185-0558 MRC Laboratory of Molecular Biology, Cambridge, UK Contribution: Conceptualization, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review & editing Search for more papers by this author Aleksandra P Pitera Aleksandra P Pitera orcid.org/0000-0001-7004-1615 MRC Laboratory of Molecular Biology, Cambridge, UK Contribution: Conceptualization, Formal analysis, Validation, Investigation, Visualization, Writing - review & editing Search for more papers by this author Maria Szaruga Maria Szaruga orcid.org/0000-0003-1673-1855 MRC Laboratory of Molecular Biology, Cambridge, UK Contribution: Formal analysis, Validation, Investigation, Visualization, Writing - review & editing Search for more papers by this author Sew-Yeu Peak-Chew Sew-Yeu Peak-Chew orcid.org/0000-0002-7602-6384 MRC Laboratory of Molecular Biology, Cambridge, UK Contribution: Investigation, Writing - review & editing Search for more papers by this author Steven W Wingett Steven W Wingett orcid.org/0000-0002-2343-0711 MRC Laboratory of Molecular Biology, Cambridge, UK Contribution: Software, Formal analysis, Writing - review & editing Search for more papers by this author Anne Bertolotti Corresponding Author Anne Bertolotti [email protected] orcid.org/0000-0002-9185-0558 MRC Laboratory of Molecular Biology, Cambridge, UK Contribution: Conceptualization, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review & editing Search for more papers by this author Author Information Aleksandra P Pitera1, Maria Szaruga1, Sew-Yeu Peak-Chew1, Steven W Wingett1 and Anne Bertolotti *,1 1MRC Laboratory of Molecular Biology, Cambridge, UK *Corresponding author. Tel: +44 1223 267054; E-mail: [email protected] The EMBO Journal (2022)41:e109985https://doi.org/10.15252/embj.2021109985 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions Figures & Info Abstract Halofuginone (HF) is a phase 2 clinical compound that inhibits the glutamyl-prolyl-tRNA synthetase (EPRS) thereby inducing the integrated stress response (ISR). Here, we report that halofuginone indeed triggers the predicted canonical ISR adaptations, consisting of attenuation of protein synthesis and gene expression reprogramming. However, the former is surprisingly atypical and occurs to a similar magnitude in wild-type cells, cells lacking GCN2 and those incapable of phosphorylating eIF2α. Proline supplementation rescues the observed HF-induced changes indicating that they result from inhibition of EPRS. The failure of the GCN2-to-eIF2α pathway to elicit a measurable protective attenuation of translation initiation allows translation elongation defects to prevail upon HF treatment. Exploiting this vulnerability of the ISR, we show that cancer cells with increased proline dependency are more sensitive to halofuginone. This work reveals that the consequences of EPRS inhibition are more complex than anticipated and provides novel insights into ISR signaling, as well as a molecular framework to guide the targeted development of halofuginone as a therapeutic. SYNOPSIS Treatment with Halofuginone (HF), an inhibitor of the glutamyl-prolyl-tRNA synthetase, activates the integrated stress response (ISR) leading to both gene expression reprogramming and translational attenuation. However, the decrease of translation induced by HF is atypical as it occurs independently of GCN2 and eIF2α phosphorylation. ISR-related gene expression changes induced by HF depend on GCN2 and eIF2α phosphorylation. HF-induced translation attenuation is independent of GCN2 and eIF2α phosphorylation. HF induces translation elongation defect. Proline-dependent cancer cell lines are more sensitive to HF-induced cytotoxicity. Introduction Halofuginone (HF), a derivative of the natural product febrifugine extracted from the hydrangea Dichroa febrifuga, has been used for centuries in Chinese medicine to treat malaria (Pines & Spector, 2015). Halofuginone was synthesized to alleviate the toxicity of febrifugine and has been widely utilized in veterinary practice for more than two decades to treat parasites in poultry and cattle (Daugschies et al, 1998). HF also exhibits a wide range of experimental and clinical activities, including antifibrotic properties, inhibition of angiogenesis and metastasis (Pines & Spector, 2015). HF binds and inhibits the EPRS with low nanomolar potency and as a result induces the ISR by activating GCN2 (Sundrud et al, 2009; Keller et al, 2012). However, recent data revealed some non-ISR activities associated with anti-inflammatory properties (Kim et al, 2020). The ISR is a vital homeostatic pathway elicited by phosphorylation of the translation initiation factor eIF2α on serine 51 to allow cells to adapt and survive changes in their environment. ISR signaling triggers a coordinated response consisting of attenuation of translation initiation and reprogramming gene expression (Sonenberg & Hinnebusch, 2009; Wek, 2018). In humans, this response is orchestrated by four eIF2α kinases that sense different signals and two phosphatases. The kinase GCN2 is activated by amino acid shortage, PKR by double-stranded RNA during viral infections, HRI by heme deficiency and PERK (or PEK) by accumulation of misfolded proteins in the endoplasmic reticulum (Sonenberg & Hinnebusch, 2009; Wek, 2018). Two eIF2α phosphatases reverse the activity of the four eIF2α kinases. They are split enzymes composed of a common catalytic subunit, protein phosphatase 1 (PP1), bound to one of two specific substrate receptors: stress-inducible PPP1R15A (R15A), or the constitutive PPP1R15B (R15B) (Bertolotti, 2018). Unlike PP1 in isolation, the holoenzymes R15A-PP1 and R15B-PP1 are selective for their substrate (Harding et al, 2009; Carrara et al, 2017; Bertolotti, 2018). The antagonistic actions of the four eIF2α kinases and the two phosphatases tune the phosphorylation levels of eIF2α to the cellular needs and conditions. Because of its central role in controlling cell and organismal survival, the ISR has emerged as a prime target for new pharmacological manipulations (Costa-Mattioli & Walter, 2020; Luh & Bertolotti, 2020), particularly for approaches aimed at restoring proteostasis (Balch et al, 2008). Compounds that either prolong (Guanabenz (Tsaytler et al, 2011), Sephin1 (Das et al, 2015), Raphin1 (Krzyzosiak et al, 2018)), or block (PERK inhibitors (Atkins et al, 2013), ISRIB (Sidrauski et al, 2013)) eIF2α phosphorylation or its downstream signaling have been identified (Costa-Mattioli & Walter, 2020; Luh & Bertolotti, 2020). Guanabenz is an approved drug, initially developed as an α2-adrenergic agonist to treat hypertension (Holmes et al, 1983). It has recently shown efficacy in a phase 2 clinical trial in amyotrophic lateral sclerosis (Bella et al, 2021), 10 years after its activity as a proteostatic compound was revealed (Tsaytler et al, 2011). Like Guanabenz, Sephin1, a non-adrenergic guanabenz derivative (Tsaytler et al, 2011; Das et al, 2015; Krzyzosiak et al, 2018), as well as Raphin1, prolong eIF2α phosphorylation and protect cells and mice from protein misfolding stress and associated diseases (Luh & Bertolotti, 2020). Sephin1 has passed through a favorable phase 1 clinical trial (https://clinicaltrials.gov/ct2/show/NCT03610334). The development of PERK inhibitors has stopped due to on-target pancreatic toxicity (Atkins et al, 2013). ISRIB has shown benefit in mouse models of diverse diseases (Costa-Mattioli & Walter, 2020) and the development of its derivatives is evaluated for vanishing white matter disease (Wong et al, 2019). Because HF has progressed to phase 2 clinical trials for the treatment of HIV-Related Kaposi's Sarcoma and Duchenne muscular dystrophy treatment (www.clinicaltrials.gov), it is a potentially attractive compound to explore the benefit of ISR modulation both experimentally and clinically. Here, we dissected the mechanism of action of HF and revealed that although HF induces the two canonical ISR adaptations consisting of attenuation of bulk protein synthesis and gene expression reprogramming, this response is surprisingly atypical because translation attenuation occurs independently of GCN2 and eIF2α phosphorylation. We found that these changes following HF treatment are all rescued upon proline addition, demonstrating that they result from EPRS inhibition. The knowledge that the observed activities of HF are all on-target provides the molecular basis to select specific disease conditions for optimal responsiveness to the compound, as exemplified here with the increased sensitivity of proline-dependent cancer cells to HF treatment. Results Atypical ISR induction by HF To comprehensively characterize the activities of HF, we started by conducting a dose-response treatment of HeLa cells and monitored induction of key ISR markers. As anticipated (Sundrud et al, 2009; Keller et al, 2012; Misra et al, 2021), HF induced an ISR response manifested by increased phosphorylation of eIF2α and increased levels of ATF4, R15A (Fig 1A and B). Surprisingly, ATF4 and R15A were induced from 12.5 nM to 312.5 nM but no longer with higher concentrations of HF. Under these conditions, eIF2α phosphorylation continued to increase (Fig 1A and B) either because of the loss of R15A expression or because of persistent kinase activation or both. In contrast to HF, tunicamycin, which activates the PERK branch of the ISR, induced phosphorylation of eIF2α and increased ATF4 and R15A levels in a dose-dependent manner (Fig 1C). The blunted ISR at high concentrations of HF was also present in other studies (Sundrud et al, 2009; Keller et al, 2012; Misra et al, 2021), but the mechanism underlying this phenomenon remains unknown, a knowledge gap which motivated this investigation. We next examined whether the lack of induction of ATF4 and R15A seen at concentrations of HF above 312.5 nM was a complete loss or a kinetic delay. Thus, we conducted time course experiments at 62.5 nM, a concentration of HF leading to typical ISR induction, as well as 312.5 nM, a concentration where ISR induction was dampened (Fig 1A). Prolonged treatment with 312.5 nM HF did not rescue the high induction of ATF4 observed at 62.5 nM (Fig 1D). As previously reported (Misra et al, 2021), mTORC1 signaling was elevated following HF, as manifested by an increased phosphorylation of the ribosomal S6 kinase (Fig EV1). This is because mTORC1 is activated due to an increased abundance of amino acids (Misra et al, 2021), possibly as consequence of ISR induction. Figure 1. HF induces an atypical blunted ISR Representative immunoblots of the indicated proteins in lysates from HeLa cells treated with indicated concentrations of halofuginone (HF) for 5 h. Quantification of ATF4 and P-eIF2α from experiments as in (A). Data are mean ± SD (n = 3 biological replicates). *P ≤ 0.0414, **P ≤ 0.0019, ****P < 0.0001, as determined by one-way ANOVA with Dunnett's multiple comparison test. Similar to (A), but with HeLa cells treated with indicated concentrations of tunicamycin for 5 h. Similar to (A), but using lysates from HeLa cells treated with 62.5 nM and 312.5 nM HF for indicated times. Relative abundance of the indicated mRNAs detected by qPCR in lysates from HeLa cells treated with indicated concentrations of HF for 5 h. Data are mean ± SD (n = 3 biological replicates). *P ≤ 0.0388, **P ≤ 0.0046, ***P ≤ 0.0008, as determined by one-way ANOVA with Dunnett's multiple comparison test. Data information: Representative results of at least three independent experiments are shown in each panel. Download figure Download PowerPoint Click here to expand this figure. Figure EV1. mTORC1 activation upon HF treatment Representative immunoblots of indicated proteins in lysates from HeLa cells treated for 5 h with indicated concentrations of HF or 200 nM Rapamycin for 3 h. Download figure Download PowerPoint We next monitored changes in abundance of ISR mRNA targets (Wek, 2018). As previously observed with other ISR inducers (Dey et al, 2010; Schneider et al, 2020; Misra et al, 2021), HF caused a dose-dependent increase of Atf4 and R15a transcripts (Fig 1E). Expression of asparagine synthetase (ASNS) is controlled transcriptionally by ATF4 (Wek, 2018). HF increased accumulation of Asns transcripts at 62.5 and 312.5 nM (Fig 1E) but this did not occur at higher concentrations (Fig 1E). A similar pattern was observed for Pycr1, another ATF4 target (Nilsson et al, 2014) encoding a proline biosynthetic enzyme (Fig 1E). Thus, the dose-response of Asns and Pycr1 mRNA to HF precisely mirrored that of its transcriptional factor ATF4, peaking at 62.5 nM (Fig 1A and E). These detailed time course and dose-response analyses reveal that HF induces an atypical ISR, as this response is blunted downstream of eIF2α phosphorylation at high concentrations of the compound. ISR-dependent HF activities Because HF was recently reported to display some ISR-independent activities (Kim et al, 2020), we next examined if the activities observed here were ISR-related or off-target. eIF2α phosphorylation in response to various stresses is completely abolished in cells lacking all four eIF2α kinases (4KO; Taniuchi et al, 2016). However, signal-specific induction of eIF2α phosphorylation is restored upon reintroduction of the cognate kinase in the 4KO cells (Taniuchi et al, 2016). This provides a robust single eIF2α kinase system to examine ISR sensing. We monitored induction of eIF2α phosphorylation by HF in wild-type mouse embryonic fibroblasts (MEFs), 4KO cells, and in 4KO cells complemented by each of the four human eIF2α kinases (Taniuchi et al, 2016). HF induced eIF2α phosphorylation in WT cells but not in the 4KO cells and this was rescued upon complementation with GCN2 in the 4KO cells, but no other eIF2α kinases (Fig 2A). This demonstrates that eIF2α phosphorylation by HF is entirely dependent on GCN2 in mouse cells. To validate these findings, we knocked down GCN2 in HeLa cells (Fig 2B). siRNA targeting Gcn2 effectively eliminated GCN2 and as a consequence, cells were unable to increase eIF2α phosphorylation upon HF treatment (Fig 2B). In the 4KO cells treated up to 39 μM of HF, there was no detectable increase in eIF2α phosphorylation, in contrast to the 4KO cells complemented by GCN2 (Fig 2C). These results reveal that induction of eIF2α phosphorylation by HF is entirely dependent on GCN2. Figure 2. GCN2 mediates eIF2α phosphorylation by HF Representative immunoblots of indicated proteins in lysates from mouse embryonic fibroblasts (MEFs) of indicated genotype after treatment with 12.5 nM or 200 nM HF for 5 h. Similar to (A), but using HeLa cells untreated or treated with GCN2 siRNA 48 h before treatment with indicated concentrations of HF for 5 h. Similar to (A), but with high HF concentrations. Data information: Representative results of at least three independent experiments are shown in each panel. Download figure Download PowerPoint We next used quantitative proteomics with tandem mass tag mass spectrometry as an unbiased approach to investigate the molecular basis for the unusual disconnect between high eIF2α phosphorylation and ATF4 induction at high HF concentrations. Discrete changes were observed in HeLa cells treated with 12.5 nM of HF for 5 h, with only 12 proteins found increased by the treatment (P ≤ 0.05, fold change ≥ 1.5), ATF4 showing the highest induction (Fig 3A and Dataset EV1). This validated the approach and demonstrated the selectivity of HF as an ISR inducer. Induction of ATF4 increased 3-fold after 5 h treatment with 12.5 nM of HF, and only 1.5-fold with 312.5 nM HF in these quantitative proteomic analyses performed in HeLa cells (Fig 3A). ATF4 also increased in HF-treated wild-type mouse embryonic fibroblasts (eIF2αS/S) but not in eIF2αA/A cells that are incapable of phosphorylating eIF2α (Scheuner et al, 2001; Fig 3B and C). As in HeLa cells, ATF4 induction was lower in eIF2αS/S cells treated with 200 nM HF than in cells treated with 12.5 nM (Fig 3B). This demonstrates that HF induces ATF4 through the canonical GCN2-to-eIF2α signaling pathway but this induction is dampened at high concentrations of HF. Figure 3. Global quantitative proteomic analyses of cell response to HF Tandem mass tag mass spectrometry analyses performed in triplicate from HeLa cells treated with 12.5 or 312.5 nM HF for 5 h. Similar to (A), but using eIF2αS/S MEF cells treated with 12.5 or 200 nM HF for 5 h. Similar to (B), but using eIF2αA/A cells. Data information: Results are presented as pairwise comparisons. Data points in magenta represent collagen proteins. Vertical dashed lines indicate Log2(Fold change) = 0.58 and that corresponds to 1.5-fold change. Download figure Download PowerPoint Whilst only a few proteins increased in abundance upon treatment of HeLa or MEF cells with 12.5 nM HF, 312.5 nM HF decreased the levels of 188 proteins (P ≤ 0.05, fold change ≥ 1.5) in HeLa, 204 in eIF2αS/S and 195 in eIF2αA/A cells (Fig 3 and Dataset EV1). Gene ontology analyses were performed (Dataset EV2). Amongst the different categories, enrichments for genes involved in collagen fibril organization, including various types of collagens were observed (Dataset EV1). Decreased collagen expression could provide the molecular basis for HF antifibrotic activities (Pines & Spector, 2015). This suggested that some of the medically relevant activities of HF may be a consequence of the decreased abundance of downregulated proteins. We then searched for the underlying mechanism. High concentrations of HF decrease R15B Amongst the proteins downregulated at high HF concentration was R15B (Fig 4A). This was unexpected because R15B is resistant to the translation attenuation resulting from eIF2α phosphorylation (Andreev et al, 2015; Schneider et al, 2020). Because the decreased abundance of R15B (Fig 4A) could explain the unusually high increase in eIF2α phosphorylation observed at high HF concentrations (Fig 1A), we investigated it further. First, we confirmed that R15B was detectable up to 312.5 nM HF (Fig 4B). However, in cells treated for 5 h at higher concentrations than 312.5 nM HF, R15B was essentially depleted (Fig 4B). Surprisingly, this decrease was GCN2-independent and even exacerbated in GCN2-depleted cells (Fig 4C). Thus, we next examined whether the depletion of R15B observed at high concentrations of HF resulted from an increased degradation. As previously reported, R15B (Jousse et al, 2003) and ATF4 (Rutkowski et al, 2006) are unstable proteins. Both R15B and ATF4 are targeted for degradation by the cullin-RING E3 ligase β-TRCP (Lassot et al, 2001; Coyaud et al, 2015). The activity of such ligases is controlled by NEDD8-activating enzyme, which can be inhibited by the small-molecule inhibitor MLN4924 (Soucy et al, 2009). Both R15B and ATF4 proteins were stabilized in cells treated with MLN4924 (Fig EV2A–C), confirming that they are targeted for degradation by a cullin-RING ligase. The decreased abundance of ATF4 and R15B in cells treated with high concentrations of HF was not affected by treatment with MLN4924 (Fig 4D) indicating that it does not result from increased degradation. To confirm these observations, we treated cells with the proteasome inhibitor MG-132. Proteasome inhibition resulted in a marked accumulation of ATF4 and a slight increase in R15B in absence of HF (Fig 4E). However, MG-132 did not prevent the loss of ATF4 and R15B observed after treatment with high HF concentrations, above 312.5 nM (Fig 4E). Thus, the loss of ATF4 and R15B at high HF concentrations does not result from increased degradation. Figure 4. HF decreases R15B abundance independently of GCN2 and protein degradation Fold change of ATF4 and R15B after 5 h treatment with 12.5 or 312.5 nM HF relative to untreated cells. Data obtained from the quantitative proteomic analyses shown in Fig 3A. Representative immunoblots of indicated proteins in lysates from HeLa cells after 5 h treatment with indicated concentrations of HF. Similar to (B), but using HeLa cells untreated or treated with GCN2 siRNA 48 h before treatment with indicated concentrations of HF. Representative immunoblots of lysates from HeLa cells treated with indicated concentrations of HF for 5 h with or without the Nedd8-activating enzyme inhibitor MLN4924 (1 μM). Representative immunoblots of lysates from HeLa cells treated with indicated concentrations of HF for 5 h with or without proteasome inhibitor MG-132 (10 μM). Data information: Representative results of at least three independent experiments are shown in each panel. Download figure Download PowerPoint Click here to expand this figure. Figure EV2. The Nedd8-activating enzyme inhibitor MLN4924 stabilizes ATF4 and R15B Representative immunoblots of the indicated proteins in lysates from HeLa cells treated with cycloheximide (20 μg/ml) with or without the inhibitor of the NEDD8-activating enzyme MLN4924 (1 μM) for indicated times. Quantification of R15B and ATF4 in cells treated with MLN4924 (1 μM) for 120 min. Data are mean ± SEM (n = 3 biological replicates). *P = 0.0369, as determined by unpaired t-test. Quantification of experiments such as the ones shown in (A). t1/2 values were calculated with results from 3 independent experiments. Data are mean ± SEM (n = 3 biological replicates). Download figure Download PowerPoint Translational changes upon HF treatment To gain further mechanistic insights into HF response, we next focused on translation. We first took advantage of a published ribosome profiling dataset (Misra et al, 2021) and performed metagene analyses to characterize the translational response to HF at a global level. The distribution of ribosomes on mRNAs at a genome-wide level in control cells appeared as expected (Vorontsov et al, 2021), with a gradient of footprint density toward the 3' end attesting high translation activity (Fig 5A). A low ribosome density after stop codons revealed efficient translation termination in all conditions (Fig 5A). HF treatment caused a striking genome-wide redistribution of ribosomes, with an increased density at the beginning of the ORFs and a decreased density toward the 3' end (Fig 5A). This difference was confirmed by a polarity score analysis (Fig 5B) and may be explained by the fact that HF causes ribosome pausing at proline codons (Misra et al, 2021), as this would result in fewer ribosomes reaching the end of the transcript. Following treatment with HF, translation of some mRNAs decreased, others increased, whilst the majority remained unchanged (Misra et al, 2021). We next performed metagene analyses of ribosome footprints in the differentially regulated groups of transcripts. The increased 5' end occupancy on the ORFs observed at the global level after HF was not seen on transcripts that were preferentially translated upon HF (Fig 5C). The translationally repressed mRNAs displayed an increased ribosome density in their 5' UTR (Fig 5D). This has been seen at a genome-wide levels in yeast exposed to amino acid starvation (Ingolia et al, 2009; Schuller et al, 2017) and is consistent with the notion that uORF translation in 5' UTR often represses translation of the main ORFs (Hinnebusch et al, 2016). We then focused on Atf4 translation. High ribosome occupancy was observed on the uORFs but not on the main ORF of Atf4 in untreated cells (Fig 5E). HF (100 nM) increased occupancy of ribosomes on the main ORF (Fig 5E), in agreement with the increased expression of the protein observed here (Fig 1A). Figure 5. Global analyses of translational changes upon HF treatment Metagene analysis plot showing average ribosome occupancy from all genes aligned at start codons in MEF cells untreated (UT, 3 replicates in blue) or treated with 100 nM HF (3 replicates in red) for 6 h. The vertical dashed lines, from left to right, separate the length-normalized transcripts into the (i) 5′UTR, (ii) CDS and (iii) 3′UTR. All genes were included (8,712–7,889 transcripts depending on the sample). Polarity scores distribution for all genes. Negative values correspond to footprint enrichment at the 5′ end of a CDS. Metagene analysis plot for genes identified with increased translation efficiency following HF treatment (180–167 transcripts). Similar to (C), but for genes with repressed translational efficiency following HF treatment. (335–293 transcripts). Plot representing ribosome density across the Atf4 mRNA in untreated cells (UT1-3, in blue) and treated with HF (HF1-3, in red). Black lines represent introns, intermediate-sized blocks denote UTRs, and thick blocks represent CDS. uORF1 and uORF2, split across an intron, are represented. Note that the 6 tracks representing ribosome density on Atf4 mRNA have been scaled such that the maximum value in each track has a fixed height. Data information: 5A–E were analyzed from (Misra et al, 2021). Download figure Download PowerPoint From these diverse analyses, the genome-wide shift of ribosome density toward the 5' end of coding regions observed at a global level is the most intriguing because it is reminiscent to the redistribution caused in yeast upon depletion of a translation elongation factor (Schuller et al, 2017). This suggests that HF causes an elongation defect. This was unexpected because the GCN2-dependent eIF2α phosphorylation observed upon HF treatment (Figs 1 and 2) is expected to cause attenuation of translation initiation. HF decreases translation in a GCN2- and eIF2α-independent manner We next measured translation in cells treated with various concentrations of HF. Treatment of HeLa cells with 12.5, 62.5, and 312.5 nM HF decreased translation by ~20, 40, and 85%, respectively (Fig 6A). For comparison, the ISR-dependent translation attenuation induced by tunicamycin was ~30%, in contrast to the general translation inhibitor cycloheximide that blocked translation completely (Fig 6B). This suggested that the 40 and 85% decrease in translation at high HF concentrations may be ISR-independent. Thus, we next analyzed the consequence of HF treatment in the 4KO cells completely defective in ISR sensing. Surprisingly, a dose-dependent translation attenuation was detected in the 4KO cells at 12.5 and 200 nM HF (Fig 6C). Quantification of four replicates revealed that translation attenuation in the 4KO or 4KO+GCN2 cells was not statistically significant (Figs 6C and EV3). This was unexpected. Thus, we performed similar experiments in eIF2αA/A cells. The dose-dependent decrease in translation upon HF treatment was not significantly different in eIF2αS/S and eIF2αA/A cells (Figs 6D and EV3). In contrast, translation attenuation induced by tunicamycin was completely abolished in the eIF2αA/A cells (Fig 6E), as expected (Scheuner et al, 2001). This confirms our ability to detect ISR-dependent translational changes and reveals that the translation attenuation upon HF treatment is atypically ISR-independent in MEFs, requiring neither GCN2 nor eIF2α phosphorylation. This conclusion was corroborated in human cells, with translation attenuation upon HF treatment being resistant to GCN2 knockdown (Fig 6F). Figure 6. Translation attenuation in response to tRNA synthetase inhibitors is independent of the ISR Newly synthesized proteins pulse-labeled with 35S-methionine for 10 min in HeLa cells pre-treated with indicated compounds for 2.5 h, except (E) (Bottom) Coomassie-stained gel. Newly synthesized proteins in HeLa cells treated with 2.5 μg/ml Tm or 50 μg/ml cycloheximide (CHX). Newly synthesized proteins in 4KO or 4KO+GCN2 MEF cells treated with 12.5 or 200 nM HF. Newly synthesized proteins in eIF2αS/S and eIF2αA/A MEF cells treated with 12.5 or 200 nM HF. Newly synthesized proteins in eIF2αS/S and eIF2αA/A MEF cells treated with indicated concentrations of Tm for 2.5 or 5 h. Newly synthesized proteins in HeLa cells untreated or treated with GCN2 siRNA 48 h before treatment with indicated concentrations of HF. Newly synthesized proteins in HeLa cells treated with indicated concentrations of HF with or without proline supplementation (10 mM). Representative immunoblots of indicated proteins in lysates from HeLa cells after treatment with indicated concentrations of HF with or without proline supplementation (10 mM). Newly synthesized proteins in HeLa cells untreated or treated with GCN2 siRNA 48 h
Referência(s)