Mitochondrial Inhibition by Sodium Azide Induces Assembly of eIF2α Phosphorylation-Independent Stress Granules in Mammalian Cells

Mitochondrial stress is involved in many pathological conditions and triggers the integrated stress response (ISR). The ISR is initiated by phosphorylation of the eukaryotic translation initiation factor (eIF) 2α and results in global inhibition of protein synthesis, while the production of specific proteins important for the stress response and recovery is favored. The stalled translation preinitiation complexes phase-separate together with local RNA binding proteins into cytoplasmic stress granules (SG), which are important for regulation of cell signaling and survival under stress conditions. Here we found that mitochondrial inhibition by sodium azide (NaN3) in mammalian cells leads to translational inhibition and formation of SGs, as previously shown in yeast. Although mammalian NaN3-induced SGs are very small, they still contain the canonical SG proteins Caprin 1, eIF4A, eIF4E, eIF4G and eIF3B. Similar to FCCP and oligomycine, other mitochodrial stressors that cause SG formation, NaN3-induced SGs are formed by an eIF2α phosphorylation-independent mechanisms. Finally, we discovered that as shown for arsenite (ASN), but unlike FCCP or heatshock stress, Thioredoxin 1 (Trx1) is required for formation of NaN3-induced SGs.


Introduction
Unicellular organisms and cells in multicellular organisms encounter many external and internal agents, some of which may induce stress. Stress represents a condition that negatively affects cellular growth and may lead to pathological conditions or cell death. All types of stress lead to the damaging of lipids, nucleic acids and proteins in cells [1]. Therefore, during evolution, it was essential for cells to develop different strategies to sense stress, transmit information and elicit an appropriate adaptive response. Adaptation to sudden changes in the cellular environment is mostly achieved through changes in gene expression, which are necessary to survive exposure to different environmental stresses and re-establish cellular homeostasis [2,3]. The regulation of translation has an advantage over transcriptional control, since drawing on pre-existing mRNA allows for faster changes in cellular protein concentrations [2]. The integrated stress response (ISR) is an evolutionarily conserved, protective molecular signaling mechanism that enables cells to cope with stress and maintain homeostasis [4]. The key players in the ISR are four kinases (GCN2, HRI, PERK, PKR), each for sensing a different type of stress, with the common substrate eukaryotic translation initiation factor 2 alpha (eIF2α) [4,5]. The phosphorylation of eIF2α on serine (S)51 converts the factor from a substrate to a competitive inhibitor of its guanidine exchange factor eIF2B [2,6]. As a result, the impaired nucleotide exchange blocks protein synthesis by preventing the recycling of inactive eIF2α-GDP to functional eIF2α-GTP, which is required for the delivery of the initiator Met-tRNAi to the 40S ribosomal subunit during translation initiation [2,6]. A consequence of stalled translation initiation is the formation of cytoplasmic stress granules (SGs) through a phase-separation process [7][8][9][10]. These granules are multimolecular aggregates of stalled translation pre-initiation complexes, small ribosomal subunits and locally present RNA-binding proteins (RBPs) [10,11]. The biological function of SGs is not yet fully understood, but they are proposed to be sites of mRNA storage and triage that allow a fast reinitiation of translation when stress is resolved [12]. Besides RBPs, SGs also contain other proteins that do not bind RNA directly, but which are recruited to SGs via protein-protein interactions [13][14][15]. The proteome of SGs shows variations according to stress type and level of stress [13,14,16], and the functions of SGs might differ accordingly. Since both pro-apoptotic and anti-apoptotic factors have been found within SGs, they may be able to determine the cellular fate under stress conditions [17][18][19]. In addition, SGs sequester other signaling proteins and enzymes and may thus serve as general signaling platforms that aid in the coordination of stress and immune signaling pathways [7]. As SGs are stress-specific, their characterization might help in obtaining a better understanding of the cellular response to hazardous stress or pathological states.
The mechanisms and consequences of SG formation upon mitochondrial inhibition are rather complex, and conflicting findings have been reported in the literature. As mitochondria are essential organelles, and mitochondrial dysfunction is a driver of pathogenesis, it is important to untangle the underlying processes of the SG response to mitochondrial stress. Oligomycine and mitochondrial uncouplers such as Carbonyl cyanide-ptrifluoromethoxyphenylhydrazone (FCCP) and Carbonyl cyanide m-chlorophenyl hydrazone (CCCP) were shown to activate HRI kinase [20,21]. However, Kedersha et al. observed that FCCP and oligomycine induced SGs independently of HRI activation and eIF2α phosphorylation [22]. Later, we confirmed these findings, showing that HRI is not important for SGs formed upon FCCP treatment [23]. Mitochondrial complex I inhibition was not investigated in terms of the SG formation, but HRI kinase is thought to mediate a stress response upon exposure to rotenone, which inhibits the mitochondrial complex I [21]. Malonate, a competitive inhibitor of the mitochondrial electron transport chain complex II, was found to induce eIF2α-independent, non-canonical SGs through a mechanism involving the translation inhibitor 4E binding protein (BP) 1 [24]. Antimycin A, an inhibitor of complex III and inducer of oxidative stress, was found to lead to an increase in eIF2α phosphorylation and a decrease in mTOR activity [25]. However, complex III inhibition has not been explored in respect to SG formation.
Carbon monoxide (CO), which inhibits mitochondrial complex IV, triggers HRI kinase activation [26]. Another study showed that CO also activated PERK, which was needed for SG formation upon exposure to CO [27]. Sodium azide (NaN 3 ) is another inhibitor of mitochondrial complex IV [28,29]. Its effect has been mainly investigated in yeast, with a SG response differing considerably from mammals [14,30,31]. In these studies, NaN 3 strongly inhibited translation and induced SGs in an unknown, but eIF2α phosphorylation-independent, manner [14,31]. To our knowledge, there is only one study to date about the NaN 3 -induced stress response in mammalian cells, showing the formation of NaN 3 -induced SGs in rat renal tubular cells [32]. In accordance with the reports from yeast, the authors of this study found eIF2α to be phosphorylated. However, it remains unclear whether the observed SG induction in response to NaN 3 was mediated via eIF2α phosphorylation.
In the study presented here, we further characterize SG formation upon mitochondrial inhibition that is caused by NaN 3 treatment in different mammalian cell types, thereby expanding our knowledge concerning: SG composition; SG kinetics in relation to translational inhibition and signaling; and the implications of eIF2α phosphorylation and other factors for the formation of SGs.

Mitochondrial Inhibition by NaN 3 Is a Potent Trigger for SG Formation in Mammalian Cells
The cellular SG response to mitochondrial stress caused by the inhibition of mitochondrial complex IV has not been investigated in mammalian systems in great detail. To better understand the effect of SG assembly upon mitochondrial stress, we treated U2OS cells with NaN 3 and monitored the SG formation by an immunofluorescence (IF) analysis of G3BP Stress Granule Assembly Factor 1 (G3BP1), a well-known SG marker [33,34]. The treatment conditions were the same as in yeast studies (a concentration of 0.5% v/v corresponding to 76 mM NaN 3 , using a low-glucose medium during treatment) [14] and always compared to control conditions where cells were kept in a low-glucose medium for the same amount of time but without the addition of NaN 3 . We performed MitoTracker staining (CM-H 2 TMRos) and confirmed that our treatment conditions led to a severe inhibition of mitochondrial activity ( Figure S1). As a consequence, NaN 3 effectively induced SG assembly in U2OS cells, HeLa cells and MEFs ( Figure 1). About 86% of U2OS and 66% of MEFs formed SGs after 2 h of NaN 3 treatment. The strongest response was seen in HeLa cells, where almost all cells formed SGs. Surprisingly, the observed SGs displayed an unusually small size in all tested cell lines. In the case of HeLa cells, NaN 3 treatment led to the formation of SGs with an average diameter of 0.19 µm (median value 0.15 µm), which is 15 times less than the average diameter of arsenite (ASN)-induced SGs (2.85 µm, median value 2.2 µm, p < 0.000001) ( Figure S2A,B). Treatment in a high-glucose medium reduced the number of SGs by approximately fourfold, implying that ATP depletion contributed to SG formation ( Figure S2C).   Next, we explored the composition of NaN 3 -induced SGs. In addition to the presence of well-known SG markers such as G3BP1, fluorescence in situ hybridization with an oligo-(dT) probe confirmed the presence of mRNA within the observed aggregates ( Figure 2A). Furthermore, cycloheximide (CHX) treatment, which interferes with polysome disassembly, prevented the formation of the granular structures ( Figure 2B,C), a feature that is typical for SGs. The immunostaining of NaN 3 -treated cells for eIF3B, a marker for canonical SGs [16,35], confirmed the co-localization of eIF3B with G3BP1 ( Figure 2D), indicating that NaN 3 induced canonical SGs by composition. By staining against eIF3B alone, we excluded a possible "bleed-through" artifact ( Figure S3A). Similar to ASN-induced SGs [36,37]

Kinetics of NaN 3 -Induced SG Formation Coincide with Translation Inhibition and Stress Signaling
SG formation is typically a downstream consequence of translation inhibition, frequently mediated via the phosphorylation of eIF2α [7]. To monitor the NaN 3 -induced stress-response kinetics, we measured SG assembly in MEF and U2OS cells over time in parallel to changes in translation and the activation of stress signaling pathways. Puromycin incorporation assays showed that translation was strongly suppressed already after 15 min of NaN 3 treatment in both human and murine cell lines ( Figure 3A,B). In contrast to the early and immediate inhibition of translation, the formation of SGs increased progressively over time in both U2OS cells and MEFs ( Figure 3C,D). Whereas only 8% of the U2OS cells and 15% of the MEFs contained SGs after 15 min of NaN 3 treatment, the proportion rose to 86% (U2OS) and 65% (MEFs) after 2 h. Along with the formation of SGs and translation inhibition, a Western blot (WB) analysis of extracts from NaN 3 -treated cells revealed multiple changes in the stress-response signaling pathways ( Figure 4A,B). First, we observed increased eIF2α phosphorylation upon NaN 3 treatment. In U2OS, eIF2α phosphorylation was detected after 15 min, and the phosphorylation levels remained the same during the exposure period of 2 h. Similarly, in MEFs, eIF2α phosphorylation was also detected after 15 min, but the levels gradually increased during the course of 2 h. Second, AMPK, which responds to reduced cellular energy levels, was found to be phosphorylated at T172, reflecting the activation of the kinase. The phosphorylation of AMPK had occurred already after 15 min exposure to NaN 3 , and it progressively increased in U2OS cells. Third, 4EBP1 was found to be dephosphorylated, which occurred through the inhibition of mTOR signaling and converted 4EBP1 into an inhibitor of cap-dependent translation [38].

NaN3-Induced SGs Are Assembled Independently of eIF2α Phosphorylation
The phosphorylation of eIF2α plays a central role in the induction of SGs under acute stress conditions [9,39]. Since we observed elevated eIF2α phosphorylation following exposure to NaN3, we wanted to investigate the importance of eIF2α phosphorylation for NaN3-induced SGs. To this end, we utilized eIF2α-S51A MEFs carrying a phospho-deficient mutation in both alleles of the endogenous eIF2α gene [40], which are unable to suppress translation and to trigger SG assembly under different conditions, such as ASNinduced oxidative or ER-stress [41,42]. As expected, eIF2α phosphorylation was not detected in S51A MEFs ( Figure S4A), but, surprisingly, both wildtype (wt) and S51A MEFs formed SGs upon exposure to NaN3 ( Figure 5A,B). Likewise, puromycin incorporation assays showed that NaN3, unlike ASN, inhibited translation to a similar level in both WT and S51A MEFs ( Figure 5C). To validate the observation that eIF2α phosphorylation is not necessary for the observed SG response, we used knockout (ko) MEF cell lines, each lacking one of the four eIF2α kinases. In agreement with the results from the S51A MEFs, the PKR −/− , PERK −/− , GCN2 −/− and HRI −/− MEFs were all able to form SGs in response to NaN3 treatment ( Figure S4B).

NaN 3 -Induced SGs Are Assembled Independently of eIF2α Phosphorylation
The phosphorylation of eIF2α plays a central role in the induction of SGs under acute stress conditions [9,39]. Since we observed elevated eIF2α phosphorylation following exposure to NaN 3 , we wanted to investigate the importance of eIF2α phosphorylation for NaN 3 -induced SGs. To this end, we utilized eIF2α-S51A MEFs carrying a phospho-deficient mutation in both alleles of the endogenous eIF2α gene [40], which are unable to suppress translation and to trigger SG assembly under different conditions, such as ASN-induced oxidative or ER-stress [41,42]. As expected, eIF2α phosphorylation was not detected in S51A MEFs ( Figure S4A), but, surprisingly, both wildtype (wt) and S51A MEFs formed SGs upon exposure to NaN 3 ( Figure 5A,B). Likewise, puromycin incorporation assays showed that NaN 3 , unlike ASN, inhibited translation to a similar level in both WT and S51A MEFs ( Figure 5C). To validate the observation that eIF2α phosphorylation is not necessary for the observed SG response, we used knockout (ko) MEF cell lines, each lacking one of the four eIF2α kinases. In agreement with the results from the S51A MEFs, the PKR −/− , PERK −/− , GCN2 −/− and HRI −/− MEFs were all able to form SGs in response to NaN 3 treatment ( Figure S4B).

Trx1 Depletion Prevents Formation of NaN3-Induced SGs
In addition to translation inhibition acting as a main trigger for SG formation, a growing number of RBPs and enzymes have been identified to regulate the assembly, disassembly and dynamics of SGs. Recently, we identified Trx1, a major antagonist of cellular protein oxidation, as a novel component of ASN-induced SGs that is involved in SG assembly [23]. Trx1 depletion prevented the formation of SGs upon ASN-induced oxidative stress, but not upon heat shock or mitochondrial inhibition by FCCP [23]. To test whether Trx1 might play a role in the formation of NaN3-induced SGs, we performed a knockdown of Trx1 and monitored the SG formation upon exposure to NaN3. HeLa cells were used because siRNA transfection is very efficient when used in this system. For comparison, cells were also treated with ASN and subjected to HRI knockdown, which is known to prevent SG assembly upon ASN treatment. Trx1 knockdown efficiently reduced Trx1 expression levels and, as expected, prevented the formation of ASN-induced SGs ( Figure  6A-C). To our surprise, the knockdown of Trx1 also prevented SG formation upon NaN3 exposure. While 96% of cells transfected with control non-targeting siRNA formed SGs, only 22% of Trx1-depleted cells formed SGs under NaN3 stress conditions (p < 0.0001).
Finally, we investigated whether Trx1 localizes to NaN3-induced SGs by performing an IF analysis. However, we did not observe any co-localization of Trx1 and G3BP1 in SGs ( Figure 6D), indicating that Trx1 might act on the proteins that are important for SG formation outside of SGs, and thereby affect the assembly of SGs under NaN3-induced mitochondrial stress.

Trx1 Depletion Prevents Formation of NaN 3 -Induced SGs
In addition to translation inhibition acting as a main trigger for SG formation, a growing number of RBPs and enzymes have been identified to regulate the assembly, disassembly and dynamics of SGs. Recently, we identified Trx1, a major antagonist of cellular protein oxidation, as a novel component of ASN-induced SGs that is involved in SG assembly [23]. Trx1 depletion prevented the formation of SGs upon ASN-induced oxidative stress, but not upon heat shock or mitochondrial inhibition by FCCP [23]. To test whether Trx1 might play a role in the formation of NaN 3 -induced SGs, we performed a knockdown of Trx1 and monitored the SG formation upon exposure to NaN 3 . HeLa cells were used because siRNA transfection is very efficient when used in this system. For comparison, cells were also treated with ASN and subjected to HRI knockdown, which is known to prevent SG assembly upon ASN treatment. Trx1 knockdown efficiently reduced Trx1 expression levels and, as expected, prevented the formation of ASN-induced SGs ( Figure 6A-C). To our surprise, the knockdown of Trx1 also prevented SG formation upon NaN 3 exposure. While 96% of cells transfected with control non-targeting siRNA formed SGs, only 22% of Trx1-depleted cells formed SGs under NaN 3 stress conditions (p < 0.0001).
Finally, we investigated whether Trx1 localizes to NaN 3 -induced SGs by performing an IF analysis. However, we did not observe any co-localization of Trx1 and G3BP1 in SGs ( Figure 6D), indicating that Trx1 might act on the proteins that are important for SG formation outside of SGs, and thereby affect the assembly of SGs under NaN 3 -induced mitochondrial stress.

Discussion
Depending on the cause of mitochondrial stress, the pathways that are activated and therefore, the implications for the cell, can differ. Mitochondrial stress, caused by NaN3

Discussion
Depending on the cause of mitochondrial stress, the pathways that are activated and therefore, the implications for the cell, can differ. Mitochondrial stress, caused by NaN 3 treatment, which inhibits complex IV, induced SG formation at early time points in both human and murine cell lines (Figure 1).
The observed stress response is similar to that in yeast, where NaN 3 strongly induces SG assembly [14]. However, while the SGs in yeast are of a similar size to those observed in other SG-inducers such as ASN [14], NaN 3 -induced SGs in mammalian cells are much smaller than typical SGs. In HeLa cells, the average diameter of SGs formed upon NaN 3 treatment was 15 times smaller compared to SGs induced by ASN exposure (Figure S2). Typical SGs are characterized by their growth during the stress response, which involves the fusion of small granules with each other or with P-bodies, which are condensates containing RNA-processing enzymes [8]. This fusion process happens in a microtubule-dependent manner and requires ATP [43]. Since NaN 3 inhibits mitochondrial complex IV and F0F1 ATPase, leading to ATP depletion [29,[44][45][46], this could explain why NaN 3 -induced SGs are much smaller in size. SGs observed upon treatment with FCCP or oligomycine [22], however, are not that small, indicating that other factors also influence the size of SGs.
The immunofluorescence microscopy of the U2OS cells revealed the composition of NaN 3 -induced SGs. They contain mRNA in addition to two well-known SG markers, G3BP1 and Caprin1 [37], indicating that the granules are bona fide SGs (Figure 2). Consistent with the presence of mRNA, CHX treatment, which inhibits the disassembly of polysomes and thereby the accumulation of stalled translation initiation complexes, prevented NaN 3induced SG assembly ( Figure 2). The SGs were also positive for eIF3B, another marker for canonical SGs in mammals (Figure 2), which is consistent with eIF3B stainings of NaN 3treated rat renal tubular cells [32]. Like in yeast [14], translation initiation factors eIF4A, eIF4G and eIF4E were recruited to mammalian NaN 3 -induced SGs (Figure 2), indicating the sequestration of stalled translation pre-initiation factors.
In line with the detection of SGs within mammalian cells upon NaN 3 treatment, we observed a massive inhibition of global protein synthesis preceding SG formation. Whereas translation inhibition occurred as early as 15 min after NaN 3 exposure, SG formation increased gradually over time as a consequence of translation inhibition (Figure 3). In parallel to the early suppression of translation, multiple signaling cascades were affected upon NaN 3 treatment, including the activation of AMPK kinase, mTOR inhibition and eIF2α phosphorylation (Figure 4). The phosphorylation of eIF2α is a key event in the ISR and is responsible for the formation of SGs under most types of stress [5]. In the case of NaN 3 exposure, the S51A MEFs expressing non-phosphorylatable eIF2α showed that eIF2α phosphorylation was not required for SG formation ( Figure 5 and Figure S4). This is similar to NaN 3 -induced SGs in yeast, which also form independently of eIF2α phosphorylation [14]. It is further in line with other mitochondrial stressors, such as the uncoupling agent FCCP or oligomycine, which also induce phospho-eIF2α-independent SGs in mammalian cells [22]. Our observation was additionally strengthened by the fact that the ko MEFs lacking the individual eIF2α kinases were still able to form SGs upon NaN 3 exposure ( Figure S4). Noticeably, these data show that PERK is not important for NaN3-induced SGs, although PERK has been reported to be activated in response to NaN 3 treatment [32]. Hence, alternative and/or parallel signaling pathways seem to control translation inhibition and SG assembly upon NaN 3 treatment in an eIF2α-phosphorylation independent manner. As eIF2α phosphorylation is not required, translation suppression and SG formation are also probably independent from the ISR in general. Interestingly, the phosphorylation of AMPK and inhibition of mTOR signaling were shown to contribute to translation arrest and SG assembly under conditions of cold shock [47]. Since mitochondrial dysfunction by NaN 3 treatment leads to a similar drop in ATP levels and AMPK phosphorylation, it would be interesting to further explore the role of AMPK signaling in translation inhibition and SG assembly.
Finally, we discovered that the depletion of Trx1 prevented the formation of SGs under mitochondrial stress caused by NaN 3 exposure ( Figure 6). This is similar for ASN-induced SGs, but different from FCCP-induced SGs, where Trx1 depletion had no effect on SG formation [23]. Unlike in ASN-induced SGs, Trx1 was absent from SGs formed upon NaN 3 treatment [23]. Therefore, Trx1 activity is most likely needed for the proper function of another SG assembly factor upon complex IV inhibition by NaN 3 . Out of all thioredoxin paralogs, only Thioredoxin 2 (Trx2) is present in mitochondria, where it provides protection against high ROS levels [48]. However, there is evidence highlighting the importance of Trx1 for mitochondrial homeostasis and health [49,50]. As we identified Trx1 to be essential for SG assembly upon NaN 3 treatment, and since SGs are thought to regulate cell survival [9,23], our results might provide a clue to the previous results showing that Trx1 is important for survival upon mitochondrial damage [50]. Given the connection between mitochondrial homeostasis and redox signaling, and considering that mitochondrial stress is associated with many pathological conditions [51], it might be interesting and useful to further explore the connections between mitochondrial stress, SG formation and the Trx1 system.

Conlusion
Within different mammalian cell types, we demonstrated that mitochondrial stress caused by NaN 3 triggered translation inhibition and the formation of tiny SGs with a similar composition as in yeast. NaN 3 -induced SGs in mammals are induced in an eIF2α phosphorylation-independent manner, which is similar to yeast and to SGs formed by other types of mitochondrial inhibitors such as FCCP and oligomycine. Our identification of Trx1 as a factor required in mammalian cells for SG formation upon NaN 3 treatment highlights a connection between redox signaling, mitochondrial stress and SG formation.

Immunofluorescence Analysis
Cells were seeded onto glass coverslips one day before drug treatment. Cells were fixed and permeabilized with ice-cold methanol for 3 min, washed with PBS and blocked with 3% BSA in PBS for 1 h at room temperature. Alternatively, cells were fixed with 4% paraformaldehyde in PBS for 15 min followed by 10 min post-permeabilization with 0.5% Triton-X in PBS.
Coverslips were then incubated with the appropriate primary antibodies diluted in PBS containing 0.1% NaN 3 (PBS-A) for 1 h rocking at room temperature or overnight at 4 • C. Cells were washed with PBS three times for 5 min before Cy2-or Cy3-coupled secondary antibodies (Jackson ImmunoResearch, West Grove, PA, USA) that were diluted 1:1000 in PBS and Hoechst dye (1:10,000, Sigma) were added for 1 h at room temperature. After washing again three times for 5 min with PBS, coverslips were mounted onto glass slides using FluoroG mounting medium (Thermo Fisher, Waltham, MA, USA). Microscopy was performed on a Nikon Ti-E microscope equipped with scMOS, a Leica DM 5000 Microscope with an Andor CCD camera; 40× dry objective was used for each type of microscope. Images were analyzed using Fiji software.

Fluorescent In Situ Hybridization
Mammalian cells were seeded onto glass coverslips and treated with NaN 3 or cultivated under control conditions. Cells were washed with 1x PBS, fixed for 15 min with 4% PFA and cell membranes were permeabilized for 10 min using 0.5% Triton-X in PBS. The coverslips were washed in 2× SSC and incubated in a 1:1000 dilution of Alexa555-coupled oligo(dT)50 probe (100 pmol/µL, Invitrogen) in hybridization buffer (1 mg/mL yeast RNA, 20% formamide, 2 mg/mL BSA, 0.1 g/mL dextrane sulphate, 1× SSC) for 1 h at room temperature. Coverslips were sequentially washed with 2× SCC, 0.2× SCC and 1× PBS. After that, the immunostaining against G3BP was performed as stated above. The DNA was visualized using Hoechst dye at a dilution of 1:10,000. Coverslips were mounted onto microscope slides using FluoroG mounting medium (Thermo Fisher).

Puromycin Incorporation Assay
Prior to lysis, cells were treated with 10 µg/mL puromycin (Gibco, Life Technologies, Waltham, MA, USA) for 5 min at 37 • C, washed three times with PBS and collected in protein lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 15 mM MgCl 2 , 1% Triton X-100; freshly supplemented with EDTA-free protease inhibitor cocktail (Roche), 1 mM Na 3 VO 4 and 50 mM NaF). Total protein concentrations were determined by Bradford assay. Equal amounts of cell lysates were separated by SDS-PAGE and analyzed with WB using anti-puromycin antibody (Millipore MABE343) for the visualization of puromycinylated polypeptides.

Western Blot Analysis
Cells were washed with PBS and collected using a cell scraper in a protein lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 15 mM MgCl 2 , 1% Triton X-100; freshly supplemented with EDTA-free protease inhibitor cocktail (Roche), 1 mM Na 3 VO 4 and 50 mM NaF). After incubation for 25 min on ice, extracts were centrifuged (10,000g) in order to separate the nuclear and cytoplasmic proteins. Cytoplasmic extracts were resolved on Tris-glycine polyacrylamide (5-20% gradient or 12%) gels and transferred onto nitrocellulose membranes with a 0.2 µm pore size (Peqlab, Erlangen, Germany). Membranes were blocked in 3% BSA in PBS-A. Primary antibodies were diluted in PBS-A; membranes were incubated overnight at 4 • C and washed three times in a 150 mM NaCl solution containing 50 mM Tris [pH 7.5] and 1% Tween-20. After washing, membranes were incubated in Horseradish peroxidase-coupled secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA, USA) diluted in PBS. Five additional washes were performed before using Western Lightning enhanced chemiluminescence substrate (PerkinElmer, Waltham, MA, USA) for detection.

MitoTracker TM Staining
HeLa cells were seeded on coverslips, then treated with NaN 3 for 1h. In the last 45 min of treatment, MitoTracker TM Orange CM-H 2 TMRos (Molecular Probes) was added to the culture media at a final concentration of 0.5 µM (the concentration level was chosen according to the manufacturer's manual). Finally, the cells were fixed and prepared for microscopy as already described (4.3. Immunofluorescence analysis).

Statistical Analysis
Statistical data analysis and graph generation was done using R. Within an R environment, besides base-r, the ggpubr package was used to generate graphs. The statistical significance was calculated by performing Student's t-test. All experiments have been performed at least three times.