Run,Ribosome,Run: From Compromised Translation to Human Health

Prevention of translation initiation on problematic mRNA templates

One of the early responders to collided ribosomes is EDF1, which recruits GIGYF2 that together with 4EHP represses translation locally to prevent reinitiation on problematic mRNAs (Hickey et al., 2020; Sinha et al., 2020). Recruitment of EDF1 is dependent on RACK1, a protein component of the 40S subunit, highlighting the central placement of RACK1 on the interface of collided ribosomes.

In parallel, activation of the ISR via the GCN2 kinase reduces the risk of further ribosome collision by globally inhibiting new rounds of mRNA translation initiation (Darnell et al., 2018; Stoneley et al., 2022; Wu et al., 2020). Here, GCN2 phosphorylates eIF2α, which in turn inhibits eIF2B. This is prohibitory to initiation of cap-dependent translation but does not impair a subset of transcripts that are required for ISR-induced translational reprogramming, such as the transcription factor ATF4 (Wek et al., 2006).

Disassembly of stalled and collided ribosomes

Another sensor of ribosome collision is the E3 ubiquitin ligase ZNF598 (Juszkiewicz et al., 2020a), which ubiquitinates RPS20/uS10 and RPS10/eS10. These ubiquitination events facilitate the splitting of the leading 80S ribosome by the ASC-1 complex composed of ASCC1–3 and TRIP4 (Juszkiewicz and Hegde, 2017; Juszkiewicz et al., 2020b; Stoneley et al., 2022). Despite ZNF598 playing an important role in recognizing collided ribosomes, very few RQC proteins directly depend on ZNF598 for recruitment to disomes (Sinha et al., 2020), highlighting a gap in our understanding of the role(s) of 40S subunit ubiquitination.

Once the leading ribosome is removed, trailing ribosomes can resume elongation and only become targets of degradation if they themselves stall and cause a collision (Juszkiewicz et al., 2020b). Stalled ribosomes with empty decoding centers are not disassembled by the ASC-1 complex but can be recognized by the HBS1L–Pelota complex. HBS1L hydrolyzes GTP and dissociates from the ribosome, allowing Pelota to bind and recruit the ATPase ABCE1, which then separates the ribosomal subunits (Juszkiewicz et al., 2020b).

ZNF598 is not the only E3 ligase recognizing stuck ribosomes. Mono-ubiquitination by the E3 ligase RNF10 on the 40S proteins RPS2/uS5 and RPS3/uS3 has been observed on blocked initiating ribosomes as well as on stalled elongating ribosomes (Garshott et al., 2021; Garzia et al., 2021), contrary to ZNF598, which only recognizes collided ribosomes. Stalled preinitiation complexes are ubiquitinated by RNF10, leading to 40S degradation, while ribosomes already engaged in translation undergo resolution, and the nascent chain is degraded by the proteasome (Garshott et al., 2021).

While the functional consequences of ubiquitination of the different 40S proteins by ZNF598 and RNF10 are still rather unclear, the ubiquitin code appears to be involved in (i) distinguishing physiologically paused from pathologically stalled ribosomes, (ii) disassembling compromised ribosomes, and (iii) acting as a signal for lysosomal or proteasomal degradation of the 40S subunit (An and Harper, 2020; Garshott et al., 2021; Meyer et al., 2020).

Degradation of the nascent peptide chain

After disassembly of the terminally stalled ribosome, the RQC system ensures degradation of the incomplete protein product. The obstructed 60S subunit is recognized by NEMF that recruits the E3 ubiquitin ligase LTN1, which ubiquitinates the lysine residues on the nascent chain, thereby marking it for proteasomal degradation. NEMF can facilitate ubiquitin-mediated proteolysis by adding C-terminal alanine and threonine (CAT) tails to stalled polypeptides, pushing the chain out of the exit tunnel and exposing lysine residues otherwise buried inside the ribosome (reviewed in Filbeck et al., 2022).

This process is evolutionarily conserved, and the bacterial NEMF homolog RqcH supports Ltn1/Listerin in degrading aberrant nascent polypeptides by adding C-terminal polyalanine tails (Crowe-McAuliffe, 2021; Filbeck et al., 2021; Lytvynenko et al., 2019). Tail composition differs between organisms, with mammalian C-terminal tails mainly being composed of alanine (Udagawa et al., 2021). In the absence of LTN1 or lysine exposure, the alanine tails are recognized as a degron by Pirh2 or CRL2^KLHDC10 E3 ubiquitin ligases, ensuring nascent chain ubiquitination and degradation (Thrun et al., 2021).

In addition, CAT-tailed peptides are more prone to aggregation, which may be a contingency plan to sequester aberrant protein products and trigger a process called aggrephagy (autophagy-mediated degradation of protein aggregates) (Lamark and Johansen, 2012). This must be tightly regulated as excess aggregation can lead to cytotoxicity and disrupts neuronal morphogenesis (Udagawa et al., 2021). Once the incomplete protein has been released from the tRNA by ANKZF1, VCP delivers it to the proteasome for degradation, and the 60S subunit is free to be recycled (Verma et al., 2013; Yip et al., 2019).

Signaling from stalled and collided ribosomes

In contrast to local-acting ribosome splitting and subsequent RQC, the ISR and the RSR are two pathways with a broader reach: the ISR induces translation reprogramming to adapt or restore cellular homeostasis, and the RSR is a mitogen-activated protein kinase cascade directing multiple biological processes. The RSR is initiated by the mitogen-activated protein kinase kinase kinase ZAKα in response to stalled or collided ribosomes (Vind et al., 2020b; Wu et al., 2020) leading to activation of the mitogen-activated protein kinases p38 and JNK, which are both master regulators of many cellular functions including proliferation, inflammation, apoptosis, and differentiation (Wagner and Nebreda, 2009).

A possible interplay between ISR and RSR has recently become apparent under certain conditions (Snieckute et al., 2022; Stoneley et al., 2022; Wu et al., 2020). Indirect impairment of translation, such as arginine deprivation, causes ribosomal pausing and GCN2 activation (Darnell et al., 2018). Under these conditions, inhibition of GCN2 (or downstream steps of the ISR) promotes disome formation, likely resulting from a failure to block new initiation on the affected mRNA template (Darnell et al., 2018; Stoneley et al., 2022). In addition, translation inhibition by the antiviral protein Viperin leads to GCN2 activation in a ZAK-dependent manner (Hsu et al., 2022).

The presence of both stalled and collided ribosomes is sensed by specific ribosome-binding proteins. These factors initiate the activation of pathways for separation of the ribosomal subunits, degradation of the nascent polypeptide, local inhibition of new translation initiation and cell fate signaling (Fig. 3). In the next sections, we will discuss the signaling pathways in response to translational stress on a cellular level.

OPEN IN VIEWER

FIG. 3.

Pathways responding to different flavors of types of ribosome stalling and collision. Key factors that initiate the relevant pathways are indicated. RQC, ribosome-associated quality control; iRQC, initiation RQC.

Viral signaling

There seems to be a connection between ribosome collisions and viral infection as depletion of collision disassembling (ZNF598 and ASCC3) or scaffolding factors (RACK1) results in interferon-stimulated gene (ISG) expression (DiGiuseppe et al., 2018; Long et al., 2014; Wan et al., 2021), which inhibits viral replication (Fig. 4). Unresolved ribosome collisions are most likely sensed by the cGAS–STING pathway, triggering the innate immune system (Wan et al., 2021).

OPEN IN VIEWER

FIG. 4.

Links between ISG induction and ribosome collision. cGAS-mediated ISG expression can be induced by (i) ribosome collision, (ii) depletion of RQC factors, or (iii) virus infection-induced interferon signaling. Virus infection also leads to the phosphorylation of RACK1 with a positive impact on the translation of viral proteins. The ISG-encoded protein Viperin causes ribosome collisions that in turn lead to activation of the ISR sensor GCN2 and the RSR sensor ZAKα. ISG, interferon-stimulated gene.

In support of this, depletion of either NEMF (a downstream RQC factor), or the main splitting factors for stalled ribosomes, HBS1L and Pelota, does not seem to affect ISG expression (Wan et al., 2021), suggesting that ribosome collision itself constitutes the relevant signal. This mechanism appears potentially conserved in evolution, since the bacterial STING/cGAS system is also a key player in phage defense (Morehouse et al., 2020). In addition, a recent study showed that ISGs induce Viperin, a virus-inhibitory enzyme. The product of Viperin induces ribosome collisions that are sensed by GCN2 and ZAKα, reducing overall translation, which interferes with the replication of the RNA-virus Zika (Hsu et al., 2022).

The importance of Viperins is further underlined by their evolutionary conservation, as prokaryotic Viperins protect bacteria against phage infection (Bernheim et al., 2021). In parallel to ISGs induction via RQC, the RQC contributes to efficient MHC-1 presentation possibly facilitating early viral detection (Trentini et al., 2020).

Many RNA viruses induce 40S remodeling via their complex internal ribosome entry sites to favor translation of the viral RNA (reviewed in Johnson et al., 2017). In a similar manner, phosphorylation in the RACK1 loop broadens the ribosome's translation capacity by supporting eIF4A-independent translation initiation (Rollins et al., 2021). This noncanonical mode of translation can be used by DNA viruses such as poxvirus to ensure optimal translation of its viral proteome (Rollins et al., 2021).

This process depends on both RACK1 (Park and Walsh, 2022) and ZNF598 and its RPS20/uS10 ubiquitination (DiGiuseppe et al., 2018), as genetic deletion of ZNF598 or mutation of RPS20/uS10 preventing its ubiquitination results in global reduction of viral protein synthesis of vaccinia virus (Sundaramoorthy et al., 2021). In addition to RQC, viral pathogens activate and regulate the ISR to facilitate their replication (recently reviewed in Wu et al., 2022).

Inflammatory signaling

Besides a role in viral propagation, ribosomal stress responses have been implicated in inflammatory responses. Activation of the RSR leads to upregulation of proinflammatory cytokines such as interleukin (IL)-8 (Jandhyala et al., 2008; Vind et al., 2020b), and to NLRP1 inflammasome activation and subsequent IL-1β secretion in response to UV-B in primary human keratinocytes (Robinson et al., 2022). Considering the possible crosstalk between ZAKα and GCN2 (Hsu et al., 2022; Wu et al., 2020), it is interesting to note that activation of GCN2 in macrophages promotes the expression of anti-inflammatory IL-10 and TGF-β (Ravishankar et al., 2015), and inhibits the production of the proinflammatory cytokine IL-1β (Battu et al., 2018).

Several reports have also proposed a function for GCN2 in T-helper cell proliferation and differentiation (Brenk et al., 2009; Munn et al., 2005; Sundrud et al., 2009). RSR and ISR may thus serve opposing functions in the context of immune signaling as GCN2 generally has an anti-inflammatory role, whereas ZAKα is involved in proinflammatory cytokine production.

Proliferation, autophagy, and senescence

Immune signaling is not the only cellular process affected by ribosome impairment. The different ribosome-surveillance pathways can also trigger changes in the cell cycle, and induce autophagy, senescence, or transcriptional reprogramming depending on the type of cell and translational problem.

It was recently shown that ASCC3 helps distinguish between resolvable collisions, such as the ones caused by anisomycin, and unresolvable ribosome collisions caused by UV-B or 4-nitroquinoline 1-oxide. In ASCC3-depleted cells, anisomycin treatment caused a ZAKα/p38-mediated G2 arrest due to ribosome collision, whereas unresolvable ribosome collision incurred after UV-B resulted in a more extended cell cycle arrest (Stoneley et al., 2022). This suggests that mechanisms exist that respond dynamically and that can modulate cell fate depending on the severity and persistence of the initial perturbation.

Activation of the ISR via GCN2 leads to upregulation of many autophagy genes in response to short-term leucine deprivation both in cultured cells and in mouse liver (Maurin et al., 2022). Accordingly, GCN2 deletion results in reduced autophagy and stress granule formation upon induction by Halofuginone (Battu et al., 2018), while d-serine-induced senescence and G2/M cell cycle arrest in kidney tubular cells are reduced upon GCN2 knockdown (Okada et al., 2017).

While protein synthesis and the activity of the nutrient-sensing mammalian target of rapamycin (mTOR) kinase are important at the onset of senescence, translation is globally reduced in fully senescent cells except for senescence-associated secretory phenotype (SASP) factors. Interestingly, mTOR supports senescence-associated expression of MAPKAPK2, the effector kinase downstream of ZAKα and p38. MAPKAPK2 is a negative regulator of the RNA-binding protein ZFP36L1 that mediates degradation of SASP-encoding mRNAs. This mechanism thus allows for mTOR-dependent translation of SASP factors in senescence (Anerillas et al., 2020; Herranz et al., 2015).

In skin keratinocytes, UV stress can also cause senescence. UV damage activates GCN2, which in turn phosphorylates eIF2α, leading to translational inhibition in senescent skin cells (Payea et al., 2021; Wu et al., 2020).

Interestingly, ribosomes also act as sensors of translation stress in intestinal stem cells, where cellular changes mediated by ZAKα allow cells to adapt to the changing environment and adopt a fetal-like identity (Silva et al., 2022).

Cell death

Persistent ribosome stalling and collision can overwhelm the cellular stress response pathways and lead to cell death. Caspase-dependent apoptosis can be induced by the ZAKα–JNK pathway as seen after prolonged treatment of cells with doxorubicin (Sauter et al., 2010), anisomycin (Wu et al., 2020), or high doses of UV-B (Vin et al., 2014). Apoptosis can also be triggered by CAT-tailed nascent chain aggregation, which results from excessive expression and accumulation of nonstop protein products (Udagawa et al., 2021).

CAT-tail–mediated apoptosis is likely distinct from the above, as ZAKα binds directly to ribosomes for activation and CAT-tail–mediated apoptosis is only initiated after ribosome splitting. ZAKα is also associated with pyroptosis, a type of cell death characterized by caspase-1 activation and IL-1β secretion. Unlike classic apoptosis, which seems to be induced via JNK, pyroptosis is linked to the p38 prong of the RSR (Robinson et al., 2022).

Although ISR signaling mainly operates to transiently inhibit cap-dependent translation, some stress conditions may induce p-eIF2α/ATF4–mediated apoptosis (Ishimura et al., 2016). The contribution of the RSR, CATylation, and ISR pathways to apoptotic signaling varies among cell lines and stress agents (Stoneley et al., 2022; Udagawa et al., 2021), suggesting cell-type and tissue-specific susceptibility to impaired translation and proteotoxicity.

Neurodegenerative disorders

Dysregulation of ribosome-surveillance pathways is tightly coupled to neurodegenerative disorders. Both in mice and in humans, mutation of ribosome disassembling factors, RQC factors, and ISR factors has been implicated in neurological dysfunction. Interestingly, the RSR sensor ZAKα is not expressed in the brain (Liu et al., 2000), indicating that it does not play a role in neuronal homeostasis. The importance of functional RQC for neuronal health is evident from the embryonic lethality of LTN1 knockout (KO) or RING domain-mutated mice (Chu et al., 2009), and in line with this, knockdown of LTN1 (but not NEMF, ZNF598, or ASCC3) impaired neurite outgrowth in vitro (Udagawa et al., 2021).

Moreover, while a hypomorphic LTN1 allele bypasses embryonic lethality in mice, they suffer from an amyotrophic lateral sclerosis (ALS)-like disease (Chu et al., 2009). LTN1 and ZNF598 were also shown to suppress atypical translation of the pathogenic poly(GR) protein from the C9ORF72 locus, a genetic contributor to the pathogenesis of ALS and frontotemporal dementia (Park et al., 2021). In addition, Park et al. (2021) observed that RQC function is impaired in C9-ALS patient-derived neurons.

Mutations in RQC factor NEMF have been associated with juvenile neuromuscular disease (Martin et al., 2020) and inherited neurological disorders (Ahmed et al., 2021). Mechanistically, these mutations have been linked to NEMFs CAT-tailing ability, which normally supports protein degradation via RQC. However, if the RQC fails to remove CAT-tailed products, they will accumulate and aggregate, which is detrimental to proteostasis (Howard and Frost, 2021). Indeed, there is a correlation between the mutation-associated degree of impairment of CAT tailing and phenotype severity in mice (Martin et al., 2020).

Aberrant CAT tailing has also been implicated in the development of Alzheimer's disease, where translation of amyloid precursor protein APP.C99 leads to ribosome stalling and collision and CAT tailing of APP.C99 peptides (Rimal et al., 2021). These CAT-tailed APP.C99 products can act as a seed for amyloid-β aggregation, which is the main component of amyloid plaques.

Consistent with the idea that unresolved stalling and collision may contribute to Alzheimer's disease pathogenesis, brain tissue from Alzheimer's disease patients also shows an enrichment of ZNF598, NEMF, and ANKZF1 at the core of the amyloid plaque (Rimal et al., 2021). Finally, mice that are heterozygous for the ZNF598 interactor GIGYF2 show neurodegeneration and motor dysfunction in adulthood (Giovannone et al., 2009).

Biallelic variants of ASCC3 were recently shown to underlie a neuromuscular syndrome in 11 individuals (Nair et al., 2021), and a homozygous mutation has been reported in 1 family with intellectual disability (Najmabadi et al., 2011). However, it is not yet known if these patients' disabilities are linked to the established functions of ASCC3 in the DNA damage response or ribosome splitting.

While a full KO is embryonically lethal, a hypomorphic mutation of the ribosome rescue factor HBS1L in mice and humans leads to developmental cerebellar defects (O'Connell et al., 2019; Terrey et al., 2021). Embryonic fibroblasts from KO mice show specific pause sites that are found mostly in protein coding sequences. Loss of HBS1L also leads to activation of mTOR signaling, although this might be an indirect effect of the defective quality control pathway (Terrey et al., 2021).

In humans, mutations in GTPBP2, a binding partner of Pelota, are linked to the development of Jaberi-Elahi syndrome. This newly classified neurodegenerative disorder is characterized by dystonia, motor and sensory neuropathy, ataxia, and cognitive dysfunction (Jaberi et al., 2016). In support of this, mice with GTPBP2 mutations coupled with a central nervous system-specific tRNA mutation suffer from age-dependent neurodegeneration, and exhibit increased ribosome stalling in neuronal tissue (Ishimura et al., 2014). The increased stalling activates GCN2, and like RQC, the ISR appears neuroprotective, since loss of GCN2 in the GTPBP2-mutated mice causes increased ataxia, tremors, and locomotor deficits (Ishimura et al., 2016).

While ISR and the GCN2 kinase mostly have protective effects, some data indicate a role for the phospho-eIF2α/ATF4 axis in neuronal cell death, suggesting a differentiated response depending on the intensity and duration of stress conditions (Han et al., 2013; Lange et al., 2008). This is also seen in mice where dominant mutations in at least six aminoacyl-tRNA synthetases cause peripheral neuropathy. These mutations activate the ISR via GCN2, and inhibition or genetic deletion of GCN2 improved the pathophysiology of the mice (Spaulding et al., 2021).

Chronic activation of ISR in the mutant mice suggests an increased level of stalled and collided ribosomes, and it is tempting to speculate that this overwhelms the ribosome-surveillance pathways consequently leading to increased neurodegeneration.

Other neurodegenerative disorders have been linked to ribosome stalling and RQC problems as well: In a Drosophila model of Parkinson's disease, a link between mitochondrial RQC and mitochondrial dysfunction was suggested (Wu et al., 2019), while in Huntington's disease CAG repeats in the Htt gene negatively impact translation and cause ribosome collisions in a cellular as well as a mouse model (Eshraghi et al., 2021).

RQC failure results in stalled and collided ribosomes that cannot be rescued. This leads to aberrant and aggregated protein products, accumulation of unresolved 60S subunits, and organelle dysfunction. All of these contribute to proteotoxicity and proteome imbalance, which are well-known hallmarks of neurodegenerative disorders. Interestingly, the viral protein Nsp1 was recently shown to offer neuroprotection due to its ability to manipulate the RQC pathway in Drosophila models of Alzheimer's disease, Parkinson's disease, and ALS (Wang et al., 2022).

Immune system and inflammation

Viral infections are accompanied by a substantial reprogramming of the translation machinery, in general suppressing translation of host factors and favoring extensive translation of viral proteins. Such changes may be accompanied by deregulation of the translation process itself and lead to ribosome collision. Early-acting RQC factors suppress innate immune signaling as they allow for disassembly of collided ribosomes. Such structures would otherwise be sensed as danger signals by the cGAS system resulting in ISG induction (Wan et al., 2021).

In general, ISGs function to inhibit viral replication, and all the ribosome-surveillance pathways have been implicated in viral propagation and immune signaling in a cellular context. Future work in animal models will be required to elucidate the biological importance of these stress responses in viral infections.

At an organismal level, both RSR and ISR have been associated with inflammatory signaling in various disease models. The RSR and its upstream kinase ZAKα trigger production of the proinflammatory cytokine IL-8 in response to ricin and Shiga toxins, and ZAK inhibitors might alleviate the illnesses and pathogenesis caused by ricin and Shiga toxin-producing bacteria (Jandhyala et al., 2008).

Recently, ZAKα was also shown to be responsible for inflammasome induction after UV-B irradiation of human primary keratinocytes. Interestingly, ZAKα phosphorylates a linker region in the immune sensor NLRP1, which is present in humans but not in mice. This suggests differences in the inflammatory response to UV damage, which should be considered for potential future treatments of inflammatory disorders (Robinson et al., 2022).

In humans, the highest expression of GCN2 is found in the colon, and these levels are increased in patients with ulcerative colitis and Crohn's disease compared with healthy controls (Ravindran et al., 2016). GCN2 KO mice do not present with obvious gut deficiencies, suggesting that GCN2 is not required for differentiation or proliferation of intestinal cells. However, GCN2 KO mice are more sensitive to dextran sodium sulfate-induced colitis and exhibit greater weight loss, inflammation, T_H17 responses, and colon shortening compared with wild-type littermates (Ravindran et al., 2016). These exacerbated effects were suggested to be caused by reduced autophagy in antigen presenting cells, leading to increased production of reactive oxygen species (ROS) and inflammasome activation (Ravindran et al., 2016).

As GCN2 inhibition in amino acid starved cells leads to increased levels of ribosome collisions (Stoneley et al., 2022) and the RSR initiates proinflammatory signaling and inflammation upon collisions, it would be interesting to examine a possible contribution of the RSR in the observed GCN2 KO phenotype (Jandhyala et al., 2008; Vind et al., 2020b). Also, in an experimental autoimmune encephalomyelitis mouse model mimicking multiple sclerosis, deletion of GCN2 was characterized by increased central nervous system inflammation, presence of T_H17 cells, and increased IL-17 and interferon-γ (Orsini et al., 2014).

In accordance with this, a GCN2 agonist protected mice from nephritic kidney damage by inducing autophagy and suppressing proinflammatory cytokine production (Chaudhary et al., 2015).

Metabolic dysregulation

The ISR-inducing kinase GCN2 has been shown to contribute to various aspects of metabolic regulation upon dietary challenges. GCN2 KO mice are hypersensitive to dietary leucine starvation correlating with an inability to phosphorylate eIF2α and induce the ISR in the liver (Anthony et al., 2004). Upon prolonged amino acid deprivation, mice normally repress lipid synthesis in the liver, improve hepatic insulin sensitivity, and mobilize adipose tissue lipid deposits. However, leucine-starved GCN2 KO mice increase expression of lipogenesis genes resulting in liver steatosis (Guo and Cavener, 2007), cannot suppress diet-induced mTOR activity, and do not improve insulin sensitivity (Xiao et al., 2011).

Moreover, leucine deprivation causes a rapid weight reduction, lower adiposity, and browning of white adipose tissue. Similarly, in mouse amygdalar PKC-δ neurons, GCN2 deletion blocked leucine deprivation-induced white adipose tissue browning and PKC-δ neuron activation, which could be rescued by overexpression of the downstream ISR-factor ATF4 (Yuan et al., 2020). We have recently shown that, unlike GCN2 KO mice, ZAK KO mice present with more browning of white adipose tissue, reduced steatosis, and improved glucose tolerance (Snieckute et al., 2022). These opposing phenotypes seen in GCN2 and ZAK KO mice in response to leucine-deficient diet underline a likely interplay between the ISR and RSR pathways.

A single nucleotide polymorphism in the GCN2 gene has been linked to type 2 diabetes mellitus in genome-wide association studies of the Japanese population (Kou et al., 2013; Yasuda et al., 2008), underscoring the role of GCN2 in metabolic regulation. In a follow-up study, carriers of the GCN2 risk allele showed reduced insulin secretion but no change in insulin sensitivity, when compared with controls (Kanno et al., 2020). When fed a high-fat diet, whole-body GCN2 KO and pancreatic β cell-specific GCN2 KO mice exhibited reduced glucose tolerance, insulin secretion, and reduced pancreatic β cell mass (Kanno et al., 2020).

A likely explanation is that WT mice increase proinsulin translation in pancreatic β cells when on a high-fat diet, potentially depleting the amino acid pool, leading to ribosome stalling and GCN2 activation. This leads to the induction of ATF4 and subsequent mTOR inhibition, which is required to maintain healthy pancreatic β cell mass. Thus, without GCN2, mice experience exacerbated metabolic decline (Kanno et al., 2020).

The ribosomal branch of the ISR is important for metabolic adaption during under and overfed conditions, likely due to diet-induced ribosome stalling/collisions as seen in cell culture systems (Darnell et al., 2018). Intriguingly, GCN2 is also activated in brain cells in response to amino acid deficiency and imbalanced nutrition, linking the ISR to behavioral adaption to malnutrition and starvation (Hao et al., 2005; Maurin et al., 2022).

The RSR appears to contribute to metabolic adaption in response to translation stress as well. In intestinal organoids and mouse models, ribosomes and ZAKα can act as sensors for nutrient availability by reprogramming intestinal stem cell identity to a more fetal-like phenotype. This is followed by the metabolic switch from oxidative phosphorylation to glycolysis. This change is not mediated through the typical ZAKα-p38/JNK axis, but instead involves an alternative ZAKα–SRC–YAP signaling pathway (Silva et al., 2022).

Cancer

Tumor microenvironments are often characterized by hypoxia, oxidative stress, and nutrient starvation—stress conditions that lead to ISR-mediated global translational inhibition that cancer cells need to cope with (Leprivier et al., 2015). To adapt, cancer cells undergo translational reprogramming, where eIF2α favors translation of specific transcripts. The ISR target gene ATF4 is often upregulated in cancer, and has been shown to increase migration of breast cancer cells and metastasis of squamous cell carcinoma (Nagelkerke et al., 2013; Zhu et al., 2014).

In some colorectal cancers, the proto-oncogene MYC mediates an increase in global protein synthesis depleting the cellular amino acid and nucleotide pools, and ultimately causing cell death. However, depletion of energy resources activates the GCN2–eIF2α axis that suppresses translation, and thus preserves resources and prevents apoptosis. Inhibition of GCN2 in tumor organoids suppresses tumor growth and has the potential in targeted therapy of tumors with deregulated MYC expression (Schmidt et al., 2019).

The RSR is often upregulated in cancers, too. Transcriptome analysis comparing cancer tissue with adjacent healthy tissue showed that ZAKα expression is highly upregulated in gastric, colorectal, urinary, and breast cancer (Adler et al., 2014; Liu et al., 2000; Rey et al., 2016). In addition, overexpression of ZAKα negatively correlates with cancer survival, especially in breast cancer patients (Li et al., 2018; Tang et al., 2019).

Intriguingly, the risk of developing cutaneous squamous cell carcinoma, a type of UV-induced skin cancer, is increased in melanoma patients treated with the BRAF inhibitor sorafenib. Sorafenib displays a strong off-target effect against ZAK and has been reported to suppress UV-induced apoptosis in keratinocytes in a ZAK-dependent manner (Vin et al., 2014).

In support of this, UV-B–induced ribosome collisions are resistant to the ASC-1 complex, and unresolved stalled ribosomes are recognized by the RSR inducing a G2 cell cycle arrest (Stoneley et al., 2022). ZAKα also mediates UV-B–induced pyroptosis in human skin keratinocytes by activating the NLRP1 inflammasome (Robinson et al., 2022). These connections offer a potential explanation for the skin cancer-promoting effects of sorafenib and related BRAF inhibitors.

Aging

Reducing protein synthesis promotes longevity. A classic example is attenuation of the mTOR pathway, which extends life span in yeast, nematodes, fruit flies, and mice (Bjedov and Rallis, 2020; Papadopoli et al., 2019). While it was originally believed that calorie restriction was necessary for life span extension, it is now clear that limiting essential amino acids is sufficient to extend life span in many organisms. Like mTOR, GCN2 responds to amino acid restriction, although indirectly, and mediates global translation inhibition and induction of stress response genes. Together with mTOR signaling, GCN2 and the ISR thus directly impact organismal longevity, fitness, and stress resistance (Gallinetti et al., 2013; Rousakis et al., 2013).

Aging increases the level of ribosomal pausing, resulting in more ribosome collisions in yeast and worms (Stein et al., 2022). Consequently, the RQC is overwhelmed, causing polypeptides to accumulate and aggregate, which contributes to aging-associated proteome imbalance (Stein et al., 2022). These connections have yet to be observed in mammals, but it has been shown that translation elongation rates are decreased by ∼20% in livers from old mice (Gerashchenko et al., 2021). This might be due to exacerbated ribosome pausing, considering that liver is the tissue with the highest elongation rate (Gerashchenko et al., 2021).

Further links between translation fidelity and aging were provided by a recent study showing premature aging phenotypes in mice harboring a heterozygous mutation of the ribosomal protein RPS9/uS4 (Shcherbakov, 2022).