Introduction

Ribosomal and translational integrity is essential for cellular and organismal homeostasis. Due to the continuous nature of protein translation, the process is highly sensitive to environmental perturbations. Translational impairment causes ribosomes to slow down, leading to ribosome stalling and collisions, and ribosomes may serve as cellular stress sensors in this capacity [1].

The stress- and mitogen-activated protein kinases (MAPKs) p38 and JNK are activated upon a range of cellular stressors, including reactive oxygen species (ROS), UV radiation, mechanical perturbation, and inflammatory mediators. Downstream signaling from p38 and JNK impacts inflammatory signaling, stress adaptation, and metabolic regulation, and directly affects cell fate decisions such as terminal cell cycle arrest, cell death, and differentiation [2,3,4,5]. MAPKs are activated through a three-tiered signal transduction cascade, in which a stress stimulus induces the activating phosphorylation of a MAP kinase kinase kinase (MAP3K). These phosphorylate and activate MAP kinase kinases (MAP2K) which phosphorylate and activate the MAPKs [3, 4, 6]. MAP3Ks constitute a large and diverse kinase family with at least 21 human members with non-overlap** activation mechanisms, of which our knowledge is still limited. One MAP3K that has recently risen to prominence is the ZAKα kinase. Through its binding to the ribosome via two C-terminal domains, ZAKα functions as a sensor of translational impairment [7, 8] and is activated upon stalling and/or collisions of ribosomes [9, 10]. Sensing of such structures induces the autophosphorylation and activation of ZAKα through poorly understood mechanisms and subsequent signaling through the downstream MAPKs, p38, and JNK. Collectively, this pathway is known as the ribotoxic stress response (RSR) [11].

Stalled and collided ribosomes constitute signals for additional surveillance and rescue mechanisms. GCN2, one of the four eIF2α kinases, is also activated by these aberrations to induce integrated stress response (ISR) signaling that enforces transcription and preferential translation of stress-relieving mRNAs [12, 13]. Collided ribosomes must be resolved to avoid cellular proteotoxicity from the accumulation of incomplete polypeptides [1, 14]. This reaction is mediated by the ribosome-associated quality control (RQC) pathway, in which ubiquitination of the ribosomal proteins RPS10 and RPS20 induces splitting of the ribosomal subunits, decay of the mRNA template, and proteasomal degradation of the nascent polypeptide [15]. The key upstream factor for the ubiquitination of collided ribosomes is the E3 ubiquitin ligase ZNF598, which is recruited to the interface between two collided ribosomes [16, 17]. Similar elaborate mechanisms exist for sensing and resolution of individually stalled ribosomes. These depend on the sensing complexes PELOTA-HBS1 [18] and the recently discovered yeast complex Fap1-Smu2-Fpr1 [19].

Most studies that interrogate responses to collided ribosomes are based on cell lines harboring engineered mRNA templates or treated with small-molecule ribosome inhibitors (e.g., anisomycin, cycloheximide) and ribotoxic enzymes (e.g., Shiga toxin, ricin) with unclear physiological relevance [7, 9, 11, 12, 20]. Thus, we are lacking insight into the nature of physiologically as well as pathologically relevant sources of endogenous ribosomal collision. One potential endogenous source of translational stress is nitric oxide (NO), which has been described to inhibit protein translation across different mammalian cell lines [21]. NO is a gaseous paracrine and autocrine signaling molecule with a variety of functions in the body, including its role as a vasodilator in the cardiovascular system, its function as a retrograde neurotransmitter, and its antimicrobial, cytotoxic, and proinflammatory effects in immunity [22,23,24,25,26]. The impact of NO is highly dependent on its concentration, ranging from low levels (nM) in the cardiovascular system to high, pathological levels (μM) as those produced by macrophages and neutrophils upon infection [22, 24, 27, 28]. NO production is mediated by NO synthases (NOS), which catalyze the production of NO from L-arginine and oxygen [25, 29]. There are three NOS isoforms: the endothelial (eNOS), the neuronal (nNOS), and the inducible (iNOS) [25, 29]. Of these, it is primarily the immunological stimulus-induced expression of iNOS that is linked to the production of high, cytotoxic concentrations of NO [22, 28]. NO is a free radical and a highly reactive molecule. The antimicrobial function of NO is non-target specific, damaging macromolecules such as DNA, lipids, and proteins [23, 28, 30]. Such damage has mainly been linked to the production of reactive nitrogen species (RNS), which are generated when NO reacts with other radicals including reactive oxygen species (ROS) in cells [27, 30].

A few studies have linked NO-induced stress to the activation of p38 [31,32,33,34] and JNK [34,35,36] in mammalian cells, however, the physiological mechanisms and relevance of these responses are unclear. Moreover, NO has been shown to both suppress translation and activate the ISR [21]. Here we show that high levels of NO inhibit translation in vivo and in vitro. NO causes ribosome collision and activation of the RSR, the ISR, and the RQC, the three major systems that respond to these structures. Our results indicate that NO, at least at high concentrations, is a physiologically relevant inducer of ribosome collisions and ribosomal surveillance pathways.

Results

Nitric oxide induction in cells leads to ZAK-dependent activation of p38 and JNK

By treating U2OS cells with the synthetic, short-lived NO donor DEA NONOate [44, 45], while amino acid deprivation has been shown to mainly cause stalling and/or pausing of individual ribosomes [10]. In both cases, RSR signaling from these structures has been shown to impact physiologically important adaptation responses. Conversely, mice with deficiencies in the RQC pathway succumb to early-onset neurodegeneration, suggesting that especially neuronal cells are challenged by ribosome collision from unknown sources on a stochastic basis [46, 47]. In the case of NO, we do not at present understand the physiological importance of ribosomal stress responses, and this will be an important avenue of future study. One possibility is that cytotoxic NO production achieved by immune cells such as macrophages, neutrophils, and T cells [24, 28] is sufficiently high to cause ribosome collision in the producing cells. In this context, ISR signaling may increase the tolerance of the immune cells that produce NO in response to an acute infection. In addition, the RSR signaling pathway has the potential to enforce immunological signaling and cytokine production in activated and NO-producing immune cells [4]. The tumor microenvironment is another biological niche densely populated with NO-producing cells of both tumor and immune system origin, and NO levels can reach micromolar levels in this milieu [48, 49]. In the context of tumors, NO has both toxic and protective properties [50], and ribosomal surveillance mechanisms such as the RSR, ISR, and RQC pathways are likely to be involved in mediating some of these effects.

Fig. 5: Nitric oxide activates ribosomal surveillance mechanisms.
figure 5

Nitric oxide (NO) induces ribosome collision and stalling to activate the three major collision surveillance systems. RQC ribosome-associated quality control, ISR integrated stress response, RSR ribotoxic stress response.

Full size image

Previous studies have shown that NO both suppresses protein translation [21] and induces the activation of the two stress-activated MAPKs, p38 and JNK [31,32,33,34,35,36]. Our findings highlight that ZAKα, the known MAP3K sensor of ribotoxic stress and transducer of the RSR [7, 9, 11], connects these two phenomena, at least in human cell lines (Fig. 2b, c). ZAKα can be activated both by stalling and collision of ribosomes, with the latter being the more robust activation platform [7, 9, 10]. Treatment of cells with DEA NONOate was accompanied by an increase in MNase-resistant polysome peaks (Fig. 4c, right) and activation of the collision sensor ZNF598 (Fig. 4b; Fig. S2e), which indicate the formation of ribosome collisions. It is entirely possible that only a small part of stalled ribosomes progress to collision, and at present, we do not have insight into the ratio of stalled vs. collided ribosomes. We also corroborated previous findings [21] showing that NO activates the ISR, indicated by the phosphorylation of eIF2α (Fig. 2b, c). Here we report that this reaction is mainly mediated by the ribosome-associated eIF2α kinase GCN2 (Fig. 4a). Similar to ZAKα, but opposite to ZNF598, GCN2 activation can occur in response to several translational problems, including stalling and collision of ribosomes [9, 51,52,53,54,55].

Employing a partitioned in vitro translation assay, we observed that soluble cytoplasmic translation factors, and to a lesser degree ribosomes, were exquisitely sensitive to the effects of NO release from DEA NONOate (Fig. 3c). Conversely, pre-treatment of the mRNA did not impair translation of the template. This suggests that NO, despite being a free radical, does not damage mRNA or rRNA to an extent that blocks translation and that proteins are the target for nitrosylation damage in this context. NO primarily reacts with cysteines in a process known as S-nitrosylation, which can both serve roles in cellular signaling [27, 56, 57] and constitute damaging oxidative modifications. Potential protein targets in the cytoplasmic and ribosomal fractions are initiation and elongation factors as well as ribosomal proteins. Finally, tRNAs in the cytoplasmic fraction may also be relevant targets. Future work is required to elucidate the translation-associated molecular targets of NO modification with the potential to disturb ribosomal function and activate ribosomal surveillance.

In conclusion, our work highlights NO as a reactive molecule with the potential to disturb key cellular processes such as protein translation, induce physiologically relevant ribosomal collision, and evoke ribosomal stress responses.

Materials and methods

Cell culture and reagents

Human osteosarcoma cells (U2OS - ATCC), human malignant cervical epithelial cells (HeLa - ATCC), and immortalized human embryonic kidney (HEK293 - ATCC) cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, Biowest) supplemented with 10% fetal bovine serum (FBS, Biowest), L-glutamine, 1% penicillin and streptomycin. Human near haploid cells (HAP1) were cultured in Iscove’s Modified Dulbecco’s Medium (IMDM) GlutaMAX™ Supplement (31980022, Thermo Fisher Scientific) supplemented with 10% FBS, 1% penicillin, and streptomycin. All cells were cultured at 37 °C in a humidified 5% CO2 cell incubator. Experiments were conducted when the cells reached 70–80% confluency. Cell lines were not recently authenticated but were tested for mycoplasma contamination on a quarterly basis.

U2OS ZAK KO cells, U2OS ZNF598 KO cells, and HAP1 ZAK KO cells were previously described [7, 38]. Plasmids for CRISPR/Cas9 mediated generation of U2OS ASK1 KO, U2OS ASK1/ZAK KO, HAP1 ASK1 KO, and HAP1 ASK1/ZAK KO cells were cloned using the plasmid pX459 (Addgene #62988). Briefly, gRNA oligos were ordered as complementary sequences, mixed in a 1:1 ratio, and annealed. pX459 was digested with BbsI, and the gRNA was introduced using a standard ligation reaction. The following gRNA sequences were used for ASK1 deletion: ASK1-sgRNA-1: 5′-ATTGTAAAAGCGGTCCAGCACGG and ASK1-sg-RNA-3: 5′-GTGATCAACGAAGCGAGCCAAGG. For U2OS ASK1/ZAK KO cells and HAP1 ASK1/ZAK KO cells, either U2OS ZAK KO or HAP1 ZAK KO cells were transfected with pX459-sgASK1. All constructs were verified by sequencing. Stable cell lines expressing wildtype or truncated versions of ZAKα and ZAKβ were previously described [7].

Chemicals and inhibitors used in this study were: Diethylamine NONOate sodium salt hydrate (DEA NONOate) (Sigma-Aldrich, #D184, 750 μM, 1 h), S-Nitrosoglutathione (GSNO) (Sigma-Aldrich, #N4148, 2 mM, 1–24 h), doxycycline (Sigma-Aldrich, D3347, 0.13 μg/ml, overnight), anisomycin (Sigma-Aldrich, A9789, 1 μg/ml, 1 h), emetine (Sigma-Aldrich, E2375, 1.8 μM, 15 min), puromycin (BioNordika, 13884, 10 μg/ml, 10 min), N-Acetyl-L-cysteine (Sigma-Aldrich, #A7250, 10 mM, 1 h pre-treatment), ZAK inhibitor (2 μM, 30 min pre-treatment, a gift from **aoyun Lu (**an University, China) [58]), GCN2 inhibitor A-92 (Axon medchem, #2720, 1 μM, 30 min pre-treatment), PERK inhibitor GSK2606414 (SelleckChem, #S7307, 1 μM, 30 min pre-treatment), ASK1 inhibitor Selonsertib, (Selleckchem, #S8292, 2 μM, 30 min pre-treatment) and L-arginine (Sigma-Aldrich, #A8094, 10 mM, 20 h), ISRIB (Sigma-Aldrich, #SML0843-5MG, 200 nM, 30 min pre-treatment), Earle’s Balanced Salts (EBSS) (Sigma-Aldrich, #E3024, 6 h).

Western blotting and antibodies

For whole cell extracts, cells were lysed in EBC buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.5% NP-40) supplemented with protease and phosphatase inhibitors. Samples were mixed with Laemmli sample buffer and boiled for 5 min. Protein samples were resolved by SDS-PAGE and transferred to nitrocellulose membranes. Membranes were blocked in PBS-Tween + 5% milk before incubation with primary antibody overnight at 4 °C. Next, membranes were washed in PBS-tween and incubated with HRP Goat Anti-Rabbit or Goat Anti-Mouse IgG Antibody (H + L) for 1 h at room temperature. Lastly, membranes were washed in PBS-tween and visualized by chemiluminescence (Clarity Western ECL substrate, Bio-Rad) using the Bio-Rad Chemidoc imaging system. Antibodies used in this study were: anti-phospho-p38 (Cell Signaling, #9216), anti-p38 (Cell Signaling, #9212), anti-phospho-SAPK/JNK (Cell Signaling Technology, #9255), anti-ZAK (Proteintech, #14945-1-AP), anti-ZAKα (Bethyl #A301-993A), anti-p150 (BD biosciences, #610473), anti-phospho-GCN2 (Abcam, #ab75837), anti-phospho-eIF2α (Cell Signaling, #3398), anti-puromycin (Millipore, #MABE343), anti-EDF1 (Abcam, #ab174651), anti-RPS2 (Bethyl, #A303-794A), anti-RPS10 (Abcam, #ab151550), anti-ZNF598 (Abcam, #ab111698), ASK1 (Thermo Scientific, #702278), α-tubulin (Sigma, #T9026) and anti-HA (Santa Cruz, #sc-7392).

Cloning and transient transfection of iNOS

Full-length iNOS (hNOS2 gene) was subcloned from the pLIX403-hNOS2 plasmid (Addgene plasmid #110800) into a pcDNA4/TO/Strep-HA vector using NotI restriction sites. The following primer sequences were used. hNOS2-Fw: 5′-AAGCGGCCGCGCCTGTCCTTGGAAATTTCTG and hNOS2-Rv: 5′-AAGCGGCCGCTCAGAGCGCTGACATCTCCAG. Transient transfection of U2OS cells with pcDNA4/TO/Strep-HA-hNOS2 was carried out with Polyethylenimine (PEI, 3.1 ng/μl–4.4 ng/μl, 4 h) when cells were 40–50% confluent. Cells were treated and harvested the following day.

siRNA transfections

Reverse transfection of siRNAs was carried out using Lipofectamine™ RNAiMAX Transfection Reagent (Thermo Fisher Scientific, #13778) following the manufacturer’s instructions. siRNA sequences used in this study were: ZAKα: 5′-GGUGCCCAUUAAGUAUCAA(dTdT), ZAKβ: 5′-CAUGCAAGCCAAGCAGAAU(dTdT) and ZAKαβ: 5′-GCUUAAAGAACGAGAAAGA(dTdT).

Polysome profiling

After cells were exposed to various treatments, cytosolic lysates were prepared using 20 mM Hepes pH 7.5, 100 mM NaCl, 5 mM MgCl2, 100 μg/ml digitonin, 100 μg/ml cycloheximide, 1X protease inhibitor cocktail (Sigma, #P2714) and 200 U RiboLock RNase Inhibitor (Thermo Fisher Scientific, #EO0382). Extracts were passed 10 times through a 26 G needle and incubated on ice for 5 min prior to centrifugation at 17,000 g for 5 min at 4 °C. After adding calcium chloride to a final concentration of 1 mM, lysates were optionally digested with 500 U micrococcal nuclease (MNase) (New England Biolabs, #M0247) for 30 min at 22 °C. Digestion was terminated by adding 2 mM EGTA. Equivalent amounts of lysate (100–120 μg of undigested RNA or 150–180 μg of MNase-digested RNA) were resolved on 15–50% sucrose gradients. Gradients were centrifuged at 38,000 rpm in a Sorvall TH64.1 rotor for 2.5 h at 4 °C. The gradients were passed through a Biocomp density gradient fractionation system with continuous monitoring of the absorbance at 260 nm.

Tri-partite in vitro translation assay

Isolation of ribosomes

For isolation of ribosomes, HeLa or HEK293 cells were cultured to 80% confluency and scraped off in ice-cold PBS. Cells were lysed in polysome lysis buffer (20 mM Tris-HCl, 150 mM KCl, 5 mM MgCl2, 0.5% NP40, 2 mM DTT, Roche EDTA-free Protease Inhibitor, and murine RNase inhibitor (New England Biolabs, #M0314)) and incubated while rotating for 10 min at 4 °C. Lysates were cleared by centrifugation at 14,000 g for 10 min and supernatants were recovered and supplemented with 25 µM Hemin (Sigma-Aldrich, #51280). Cleared lysates were loaded on top of a 500 µl 34% sucrose cushion in hypotonic buffer (10 mM HEPES, 10 mM KAc, 0.5 mM MgAc2) and centrifuged for 240,000 g, 2 h 15 min, 4 °C. Pellets containing ribosomes were washed thoroughly and resuspended in 0.2 X of loaded lysate volume in hypotonic ribosome buffer (20 mM HEPES, 10 mM NaCl, 25 mM KAc, 1.1 mM MgAc2).

Isolation of ribosome-free cytoplasm

For isolation of cytoplasm, lysates were prepared as described [59] with some modifications: HeLa or HEK293 cells were cultured to 80% confluency and detached by trypsinization followed by trypsin inactivation. Cells were pelleted by centrifugation at 500 g for 4 min, washed in PBS, and resuspended in equal volumes of hypotonic lysis buffer (10 mM HEPES, 10 mM KAc, 0.5 mM MgAc2, 5 mM DTT, Roche EDTA-free Protease Inhibitor) and incubated rotating 1 h, 4 °C. After incubation, lysates were pulled through a 27 G, 3/4-inch needle 8–10 times to lyse the cells, and debris was cleared by centrifugation at 14,000 g for 2 min. Cleared lysates were supplemented with 25 µM Hemin and centrifuged for 240,000 g, 2 h 15 min, 4 °C. Pellets containing ribosomes were discarded and supernatants were recovered as ribosome-free cytoplasmic lysate.

Reconstituted in vitro translation

Translation reactions were reconstituted by mixing one-part isolated ribosomes, five-parts cytoplasmic lysates, one-part in vitro transcribed luciferase mRNA (final conc. 1–10 ng/µl), and three-part translation buffer final concentrations of 1.6 mM HEPES, 10 mM creatine phosphate (Sigma Aldrich, #27920), 50 ng/µl creatine kinase (Sigma Aldrich, #10127566001), 10 µM spermidine (Sigma Aldrich, # S0266), 10 µM amino acids (Promega, # L4461), 65 mM KAc, 0.75 mM MgAc2 and murine RNase inhibitor A dual mRNA construct with a Renilla luciferase coding sequence under cap-dependent translation control and a Firefly luciferase coding sequence under control of cap-independent EMCV IRES. This construct was transcribed from a HpaI-linarized plasmid using HiScribe™ T7 ARCA mRNA Kit with tailing (New England Biolabs, #E2060) according to the manufacturer’s specifications. Translation reactions were incubated at 37 °C for 30 min and translational output was measured as luciferase activity detected with Dual-Glo® Luciferase Assay System (Promega, # E2940) according to manufacturer’s specifications. For treatment with NO donor, fractions were treated with 1 mM DEA NONOate for 10 min. For UV irradiation, fractions were irradiated with 500 J/m2 UV-B and immediately added to the in vitro translation reaction. As a positive control, anisomycin (1 μg/ml) was added to the already combined reaction mix. Based on previous experiences with the system, a sample size of n = 3 was chosen for experiments with HeLa cells and n = 2 for experiments with HEK293 cells.

Statistical analysis and data availability

The in vitro translation data was analyzed in Prism 9 using one-way ANOVA either comparing each condition to the mock or comparing the mean of each condition to every other condition with respectively Dunnett correction or Tukey correction for multiple comparisons. A significance level of 0.05 was used and p-values were presented as ns., non-significant; p > 0.05; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001. Results were presented using bar plots of the mean with error bars for the standard deviation (SD). Values from treated samples were normalized to the mean value of the mock. The variance was similar between groups that were statistically compared. Full and uncropped original western blots are available in the "original data file”.