Crystal structures of GCN2 protein kinase C-terminal domains suggest regulatory differences in yeast and mammals

The HisRS-like domain of GCN2 is a pseudoenzyme that can bind uncharged tRNA

GCN2 is a stress response kinase that phosphorylates the translation initiation factor eIF2α to inhibit general protein synthesis when activated by uncharged tRNA and stalled ribosomes. The presence of a HisRS-like domain in GCN2, normally associated with tRNA aminoacylation, led to the hypothesis that eIF2α kinase activity is regulated by the direct binding of this domain to uncharged tRNA. Here we solved the structure of the HisRS-like domain in the context of full-length GCN2 by cryoEM. Structure and function analysis shows the HisRS-like domain of GCN2 has lost histidine and ATP binding but retains tRNA binding abilities. Hydrogen deuterium exchange mass spectrometry, site-directed mutagenesis and computational docking experiments support a tRNA binding model that is partially shifted from that employed by bona fide HisRS enzymes. These results demonstrate that the HisRS-like domain of GCN2 is a pseudoenzyme and advance our understanding of GCN2 regulation and function.

Cryo-EM Structure of HRSL Domain Reveals Activating Crossed Helices at the Core of GCN2

GCN2 is a conserved receptor kinase activating the Integrated Stress Response (ISR) in eukaryotic cells. The ISR kinases detect accumulation of stress molecules and reprogram translation from basal tasks to preferred production of cytoprotective proteins. GCN2 stands out evolutionarily among all protein kinases due to the presence of a h istidyl t R NA s ynthetase-like (HRSL) domain, which arises only in GCN2 and is located next to the kinase domain. How HRSL contributes to GCN2 signaling remains unknown. Here we report a 3.2 Å cryo-EM structure of HRSL from thermotolerant yeast Kluyveromyces marxianus . This structure shows a constitutive symmetrical homodimer featuring a compact helical-bundle structure at the junction between HRSL and kinase domains, in the core of the receptor. Mutagenesis demonstrates that this junction structure activates GCN2 and indicates that our cryo-EM structure captures the active signaling state of HRSL. Based on these results, we put forward a GCN2 regulation mechanism, where HRSL drives the formation of activated kinase dimers. Remaining domains of GCN2 have the opposite role and in the absence of stress they help keep GCN2 basally inactive. This autoinhibitory activity is relieved upon stress ligand binding. We propose that the opposing action of HRSL and additional GCN2 domains thus yields a regulated ISR receptor.

Significance statement

Regulation of protein synthesis (translation) is a central mechanism by which eukaryotic cells adapt to stressful conditions. In starving cells, this translational adaptation is achieved via the receptor kinase GCN2, which stays inactive under normal conditions, but is switched on under stress. The molecular mechanism of GCN2 switching is not well understood due to the presence of a structurally and biochemically uncharacterized h istidyl t R NA s ynthetase-like domain (HRSL) at the core of GCN2. Here we use single-particle cryo-EM and biochemistry to elucidate the structure and function of HRSL. We identify a structure at the kinase/HRSL interface, which forms crossed helices and helps position GCN2 kinase domains for activation. These data clarify the molecular mechanism of GCN2 regulation.

Differential requirements for P stalk components in activating yeast protein kinase Gcn2 by stalled ribosomes during stress

The General Amino Acid Control is a conserved response to amino acid starvation involving activation of protein kinase Gcn2, which phosphorylates eukaryotic initiation factor 2 (eIF2α) with attendant inhibition of global protein synthesis and increased translation of yeast transcriptional activator GCN4. Gcn2 can be activated by either amino acid starvation or conditions that stall elongating ribosomes without reducing aminoacylation of tRNA, but it is unclear whether distinct molecular mechanisms operate in these two circumstances. We identified three regimes that activate Gcn2 in yeast cells by starvation-independent (SI) ribosome-stalling: treatment with tigecycline, eliminating the sole gene encoding tRNAArgUCC, and depletion of translation termination factor eRF1. We further demonstrated requirements for the tRNA- and ribosome-binding domains of Gcn2, the positive effector proteins Gcn1/Gcn20, and the tethering of at least one of two distinct P1/P2 heterodimers to the uL10 subunit of the ribosomal P stalk, for detectable activation by SI-ribosome stalling. Remarkably, no tethered P1/P2 proteins were required for strong Gcn2 activation elicited by starvation for histidine or branched-chain amino acids isoleucine/valine. These results indicate that Gcn2 activation has different requirements for the P stalk depending on how ribosomes are stalled. We propose that accumulation of deacylated tRNAs in amino acid-starved cells can functionally substitute for the P stalk in binding to the histidyl-tRNA synthetase-like domain of Gcn2 for eIF2α kinase activation by ribosomes stalled with A sites devoid of the eEF1A∙GTP∙aminoacyl-tRNA ternary complex.

Activation of Gcn2 by small molecules designed to be inhibitors

The integrated stress response (ISR) is an important mechanism by which cells confer protection against environmental stresses. Central to the ISR is a collection of related protein kinases that monitor stress conditions, such as Gcn2 (EIF2AK4) that recognizes nutrient limitations, inducing phosphorylation of eukaryotic translation initiation factor 2 (eIF2). Gcn2 phosphorylation of eIF2 lowers bulk protein synthesis, conserving energy and nutrients, coincident with preferential translation of stress-adaptive gene transcripts, such as that encoding the Atf4 transcriptional regulator. While Gcn2 is central for cell protection to nutrient stress and its depletion in humans leads to pulmonary disorders, Gcn2 can also contribute to the progression of cancers and facilitate neurological disorders during chronic stress. Consequently, specific ATP-competitive inhibitors of Gcn2 protein kinase have been developed. In this study, we report that one such Gcn2 inhibitor, Gcn2iB, can activate Gcn2, and we probe the mechanism by which this activation occurs. Low concentrations of Gcn2iB increase Gcn2 phosphorylation of eIF2 and enhance Atf4 expression and activity. Of importance, Gcn2iB can activate Gcn2 mutants devoid of functional regulatory domains or with certain kinase domain substitutions derived from Gcn2-deficient human patients. Other ATP-competitive inhibitors can also activate Gcn2, although there are differences in their mechanisms of activation. These results provide a cautionary note about the pharmacodynamics of eIF2 kinase inhibitors in therapeutic applications. Compounds designed to be kinase inhibitors that instead directly activate Gcn2, even loss of function variants, may provide tools to alleviate deficiencies in Gcn2 and other regulators of the ISR.

Activation and execution of the hepatic integrated stress response by dietary essential amino acid deprivation is amino acid specific

Dietary removal of an essential amino acid (EAA) triggers the integrated stress response (ISR) in liver. Herein, we explored the mechanisms that activate the ISR and execute changes in transcription and translation according to the missing EAA. Wild‐type mice and mice lacking general control nonderepressible 2 (Gcn2) were fed an amino acid complete diet or a diet devoid of either leucine or sulfur amino acids (methionine and cysteine). Serum and liver leucine concentrations were significantly reduced within the first 6 h of feeding a diet lacking leucine, corresponding with modest, GCN2‐dependent increases in Atf4 mRNA translation and induction of selected ISR target genes (Fgf21, Slc7a5, Slc7a11). In contrast, dietary removal of the sulfur amino acids lowered serum methionine, but not intracellular methionine, and yet hepatic mRNA abundance of Atf4, Fgf21, Slc7a5, Slc7a11 substantially increased regardless of GCN2 status. Liver tRNA charging levels did not correlate with intracellular EAA concentrations or GCN2 status and remained similar to mice fed a complete diet. Furthermore, loss of Gcn2 increased the occurrence of ribosome collisions in liver and derepressed mechanistic target of rapamycin complex 1 signal transduction, but these changes did not influence execution of the ISR. We conclude that ISR activation is directed by intracellular EAA concentrations, but ISR execution is not. Furthermore, a diet devoid of sulfur amino acids does not require GCN2 for the ISR to execute changes to the transcriptome.

Review: Emerging roles of the signaling network of the protein kinase GCN2 in the plant stress response

The pan-eukaryotic protein kinase GCN2 (General Control Nonderepressible2) regulates the translation of mRNAs in response to external and metabolic conditions. Although GCN2 and its substrate, translation initiation factor 2 (eIF2) α, and several partner proteins are substantially conserved in plants, this kinase has assumed novel functions in plants, including in innate immunity and retrograde signaling between the chloroplast and cytosol. How exactly some of the biochemical paradigms of the GCN2 system have diverged in the green plant lineage is only partially resolved. Specifically, conflicting data underscore and cast doubt on whether GCN2 regulates amino acid biosynthesis; also whether phosphorylation of eIF2α can in fact repress global translation or activate mRNA specific translation via upstream open reading frames; and whether GCN2 is controlled in vivo by the level of uncharged tRNA. This review examines the status of research on the eIF2α kinase, GCN2, its function in the response to xenobiotics, pathogens, and abiotic stress conditions, and its rather tenuous role in the translational control of mRNAs.

Crystal structures reveal a novel dimer of the RWD domain of human general control nonderepressible 2

General control nonderepressible 2 (GCN2) is a serine/threonine protein kinase, detecting a variety of stresses including amino acid starvation, reactive oxygen species, etc. in eukaryotic cells. Activation of GCN2 requires the interaction of the N-terminal RWD domain with the upstream GCN1 protein and the dimerization by the kinase domain. In this study, we determined two crystal structures of the RWD domain of human GCN2 in two different crystal packing modes. These two different crystal structures reveal a same dimer of the RWD domain, which has not been reported in previous studies. We further confirmed this novel dimer interaction in solution using gel filtration experiments, and in human embryonic kidney (HEK) 293 cells using bimolecular fluorescence complementation (BiFC) and co-immunoprecipitation (co-IP) assays. Together, this study discovers a potential protein-protein interface on the RWD domain of human GCN2, and suggests a possible regulation between the interaction of GCN1 and the formation of GCN2 dimer.

The structure of human GCN2 reveals a parallel, back-to-back kinase dimer with a plastic DFG activation loop motif

When activated by amino acid starvation, the stress sensing protein kinase GCN2 phosphorylates the eukaryotic initiation factor 2 alpha, inhibiting translation to conserve energy and facilitate cell survival. Amino acid starvation, particularly of tryptophan and arginine, affects immune tolerance by suppressing differentiation and proliferation of T-cells via activation of GCN2 kinase. In addition, the GCN2 pathway mediates cancer survival directly within the context of metabolic stress. Here, we report the first crystal structures of the human GCN2 kinase domain (KD) in complex with two inhibitors of different size, shape, and chemical scaffold. Three novel activation loop conformations representative of different activation states of the kinase are described. In addition, a novel dimerization organization for GCN2 is observed. This arrangement is consistent with the hypothesis that the GCN2 KD forms an antiparallel inactive dimer until uncharged tRNA binds to it and triggers conformational changes that shift the equilibrium to the active parallel dimer.

The PERK Inhibitor GSK2606414 Enhances Reovirus Infection in Head and Neck Squamous Cell Carcinoma via an ATF4-Dependent Mechanism

Reovirus type 3 Dearing (reovirus) is a tumor-selective oncolytic virus currently under evaluation in clinical trials. Here, we report that the therapeutic efficacy of reovirus in head and neck squamous cell cancer can be enhanced by targeting the unfolded protein response (UPR) kinase, protein kinase R (PKR)-like endoplasmic reticulum kinase (PERK). PERK inhibition by GSK2606414 increased reovirus efficacy in both 2D and 3D models in vitro, while perturbing the normal host cell response to reovirus-induced endoplasmic reticulum (ER) stress. UPR reporter constructs were used for live-cell 3D spheroid imaging. Profiling of eIF2a-ATF4, IRE1a-XBP1, and ATF6 pathway activity revealed a context-dependent increase in eIF2a-ATF4 signaling due to GSK2606414. GSK2606414 blocked eIF2a-ATF4 signaling because of the canonical ER stress agent thapsigargin. In the context of reovirus infection, GSK2606414 induced eIF2a-ATF4 signaling. Knockdown of eIF2a kinases PERK, GCN2, and PKR revealed eIF2a-ATF4 reporter activity was dependent on either PERK or GCN2. Knockdown of ATF4 abrogated the GSK2606414-induced increase in reovirus protein levels, confirming eIF2a-ATF signaling as key to the observed phenotype. Our work identifies a novel approach to enhance the efficacy and replication of reovirus in a therapeutic setting.

Towards a model of GCN2 activation

Cells must be able to sense and adapt to their surroundings to thrive in a dynamic environment. Key to adapting to a low nutrient environment is the Integrated Stress Response (ISR), a short-lived pathway that allows cells to either regain cellular homeostasis or facilitate apoptosis during periods of stress. Central to the ISR is the protein kinase General Control Non-depressible 2 (GCN2), which is responsible for sensing starvation. Upon amino acid deficiency, GCN2 is activated and initiates the ISR by phosphorylating the translation initiation factor eIF2α, stalling protein translation, and activating the transcription factor ATF4, which in turn up-regulates autophagy and biosynthesis pathways. A key outstanding question is how GCN2 is activated from an autoinhibited state. Until recently, a model of activation focussed on the increase of deacylated tRNA associated with amino acid starvation, with deacylated tRNA binding directly to GCN2 and releasing autoinhibition. However, in vivo experiments have pointed towards an alternative, deacylated-tRNA-independent mechanism of activation. Here, we review the various factors that may facilitate GCN2 activation, including recent research showing that the P-stalk complex, a ribosome-associated heteropentameric protein complex, is a potent activator of GCN2.

Activation of GCN2 by the ribosomal P-stalk

General control nonderepressible 2 (GCN2) phosphorylates eIF2α, regulating translation in response to nutritional stress. Here, we show that although tRNA stimulates purified, recombinant human GCN2 in vitro, mammalian ribosomes are even more potent GCN2 activators. Hydrogen/deuterium exchange–mass spectrometry (HDX-MS) showed GCN2 interacting with domain II of the uL10 P-stalk protein. The P-stalk is a uL10/P1₂/P2₂ pentameric complex that is part of the ribosomal GTPase-associated center. Recombinant human P-stalk greatly stimulates GCN2. Both domain II of uL10 and the C-terminal tails of P1 and P2 are necessary for maximal GCN2 activation. On actively translating ribosomes, the C-terminal tails of P1 and P2 are sequestered by elongation factors, suggesting P-stalk availability could link translational stress to GCN2 activation.

Cells dynamically adjust their protein translation profile to maintain homeostasis in changing environments. During nutrient stress, the kinase general control nonderepressible 2 (GCN2) phosphorylates translation initiation factor eIF2α, initiating the integrated stress response (ISR). To examine the mechanism of GCN2 activation, we have reconstituted this process in vitro, using purified components. We find that recombinant human GCN2 is potently stimulated by ribosomes and, to a lesser extent, by tRNA. Hydrogen/deuterium exchange–mass spectrometry (HDX-MS) mapped GCN2–ribosome interactions to domain II of the uL10 subunit of the ribosomal P-stalk. Using recombinant, purified P-stalk, we showed that this domain of uL10 is the principal component of binding to GCN2; however, the conserved 14-residue C-terminal tails (CTTs) in the P1 and P2 P-stalk proteins are also essential for GCN2 activation. The HisRS-like and kinase domains of GCN2 show conformational changes upon binding recombinant P-stalk complex. Given that the ribosomal P-stalk stimulates the GTPase activity of elongation factors during translation, we propose that the P-stalk could link GCN2 activation to translational stress, leading to initiation of ISR.

GCN2 phosphorylates HIV-1 integrase and decreases HIV-1 replication by limiting viral integration

GCN2 is a serine/threonine kinase involved in cellular stress response related to amino acid starvation. Previously, we showed that GCN2 interacts with HIV-1 integrase and is activated during HIV-1 infection. Herein, we identified HIV-1 integrase as a previously unknown substrate of GCN2 in vitro with a major site of phosphorylation at residue S255 located in the C-terminal domain of HIV-1 integrase. The underlying mechanism was investigated and it appeared that the integrase active site was required in order for GCN2 to target the integrase residue S255. Moreover, various integrases from other retroviruses (e.g. MLV, ASV) were also recognized as a substrate by GCN2. In cells, HIV-1 lentiviral particles harboring mutation at integrase position 255 were affected in their replication. Preventing phosphorylation resulted in an increase in infectivity that correlated with an increase in viral DNA integration. Infectivity of MLV was also higher in cells knocked-out for GCN2 suggesting a conserved mechanism to control viral replication. Altogether, our data suggest that GCN2 may constitute a general guardian of genome stability by regulating foreign DNA integration and as such be part of the antiviral armamentarium of the cell.

EIF2AK4 mutation as "second hit" in hereditary pulmonary arterial hypertension

Background

Mutations in the eukaryotic translation initiation factor 2α kinase 4 (EIF2AK4) gene have recently been identified in recessively inherited veno-occlusive disease. In this study we assessed if EIF2AK4 mutations occur also in a family with autosomal dominantly inherited pulmonary arterial hypertension (HPAH) and incomplete penetrance of bone morphogenic protein receptor 2 (BMPR2) mutations.

Methods

Clinical examinations in a family with 10 members included physical examination, electrocardiogram, (stress)-echocardiography and lung function. Manifest PAH was confirmed by right heart catheterisation in three affected subjects. Genetic analysis was performed using a new PAH-specific gene panel analysis with next generation sequencing of all known PAH and further candidate genes. Identified variants were confirmed by Sanger sequencing.

Results

All living family members with manifest HPAH carried two pathogenic heterozygous mutations: a frame shift mutation in the BMPR2 gene and a novel splice site mutation in the EIF2AK4 gene. Two family members who carried the BMPR2 mutation only did not develop manifest HPAH.

Conclusions

This is the first study suggesting that EIF2AK4 can also contribute to autosomal dominantly inherited HPAH. Up to now it has only been identified in a recessive form of HPAH. Only those family members with a co-occurrence of two mutations developed manifest HPAH. Thus, the EIF2AK4 and BMRPR2 mutations support the “second hit” hypothesis explaining the variable penetrance of HPAH in this family. Hence, the assessment of all known PAH genes in families with a known mutation might assist in predictions about the clinical manifestation in so far non-affected mutation carriers.

Activation of GCN2 kinase by ribosome stalling links translation elongation with translation initiation

Ribosome stalling during translation has recently been shown to cause neurodegeneration, yet the signaling pathways triggered by stalled elongation complexes are unknown. To investigate these pathways we analyzed the brain of C57BL/6J-Gtpbp2^nmf205-/- mice in which neuronal elongation complexes are stalled at AGA codons due to deficiencies in a tRNA^Arg_UCU tRNA and GTPBP2, a mammalian ribosome rescue factor. Increased levels of phosphorylation of eIF2α (Ser51) were detected prior to neurodegeneration in these mice and transcriptome analysis demonstrated activation of ATF4, a key transcription factor in the integrated stress response (ISR) pathway. Genetic experiments showed that this pathway was activated by the eIF2α kinase, GCN2, in an apparent deacylated tRNA-independent fashion. Further we found that the ISR attenuates neurodegeneration in C57BL/6J-Gtpbp2^nmf205-/- mice, underscoring the importance of cellular and stress context on the outcome of activation of this pathway. These results demonstrate the critical interplay between translation elongation and initiation in regulating neuron survival during cellular stress.

DOI:http://dx.doi.org/10.7554/eLife.14295.001

Information stored in DNA is used to make proteins in a two-step process. First, the DNA is copied to make molecules of messenger ribonucleic acid (or messenger RNA for short). Next, machines called ribosomes use the messenger RNAs as templates to assemble chains of amino acids – the building blocks of proteins – in a process called translation. Another type of RNA molecule called transfer RNA carries each amino acid to the ribosomes. If a specific transfer RNA is not available for translation at the right time, the ribosome might stall as it moves along the messenger RNA. At this point, the ribosome needs to be restarted or it will fall off the mRNA without finishing the protein.

In 2014, a group of researchers reported that certain types of brain cells are very sensitive to ribosome stalling, and tend to die if translation does not continue. A protein called GTPBP2 was shown to play an important role in restarting stalled ribosomes in these cells. Here, Ishimura, Nagy et al. – including some of the researchers from the earlier work – investigated the molecular pathways that ribosome stalling triggers in brain cells using mutant mice that lacked the GTPBP2 protein. The experiments show that ribosome stalling activates an enzyme known as GCN2, which was already known to sense other types of malfunctions in cellular processes.

Ishimura, Nagy et al. also show that GCN2 triggers stress responses in the cells by activating a communication system called the ATF4 pathway. This pathway protects the cells from damage, and its absence results in more rapid cell deterioration and death. The next challenges are to understand the exact mechanism by which GCN2 senses stalled ribosomes, and to find out how ribosome stalling causes the death of brain cells.

DOI:http://dx.doi.org/10.7554/eLife.14295.002

Interaction between the tRNA-binding and C-terminal domains of Yeast Gcn2 regulates kinase activity in vivo

The stress-activated protein kinase Gcn2 regulates protein synthesis by phosphorylation of translation initiation factor eIF2α. Gcn2 is activated in amino acid-deprived cells by binding of uncharged tRNA to the regulatory domain related to histidyl-tRNA synthetase, but the molecular mechanism of activation is unclear. We used a genetic approach to identify a key regulatory surface in Gcn2 that is proximal to the predicted active site of the HisRS domain and likely remodeled by tRNA binding. Mutations leading to amino acid substitutions on this surface were identified that activate Gcn2 at low levels of tRNA binding (Gcd- phenotype), while other substitutions block kinase activation (Gcn- phenotype), in some cases without altering tRNA binding by Gcn2 in vitro. Remarkably, the Gcn- substitutions increase affinity of the HisRS domain for the C-terminal domain (CTD), previously implicated as a kinase autoinhibitory segment, in a manner dampened by HisRS domain Gcd- substitutions and by amino acid starvation in vivo. Moreover, tRNA specifically antagonizes HisRS/CTD association in vitro. These findings support a model wherein HisRS-CTD interaction facilitates the autoinhibitory function of the CTD in nonstarvation conditions, with tRNA binding eliciting kinase activation by weakening HisRS-CTD association with attendant disruption of the autoinhibitory KD-CTD interaction.

Involvement of general control nonderepressible kinase 2 in cancer cell apoptosis by posttranslational mechanisms

GCN2 exerts its proapoptotic function in cancer cell death by posttranslational mechanisms. Modulation of GCN2 expression can be used for molecular targeted cancer therapy and drug development. Na⁺,K⁺-ATPase ligands are the first identified small-molecule drugs that can trigger cancer cell death by modulating GCN2 signaling.

General control nonderepressible kinase 2 (GCN2) is a promising target for cancer therapy. However, the role of GCN2 in cancer cell survival or death is elusive; further, small molecules targeting GCN2 signaling are not available. By using a GCN2 level-based drug screening assay, we found that GCN2 protein level critically determined the sensitivity of the cancer cells toward Na⁺,K⁺-ATPase ligand–induced apoptosis both in vitro and in vivo, and this effect was largely dependent on C/EBP homologous protein (CHOP) induction. Further analysis revealed that GCN2 is a short-lived protein. In A549 lung carcinoma cells, cellular β-arrestin1/2 associated with GCN2 and maintained the GCN2 protein level at a low level by recruiting the E3 ligase NEDD4L and facilitating consequent proteasomal degradation. However, Na⁺,K⁺-ATPase ligand treatment triggered the phosphorylation of GCN2 at threonine 899, which increased the GCN2 protein level by disrupting the formation of GCN2–β-arrestin–NEDD4L ternary complex. The enhanced GCN2 level, in turn, aggravated Na⁺,K⁺-ATPase ligand–induced cancer cell apoptosis. Our findings reveal that GCN2 can exert its proapoptotic function in cancer cell death by posttranslational mechanisms. Moreover, Na⁺,K⁺-ATPase ligands emerge as the first identified small-molecule drugs that can trigger cancer cell death by modulating GCN2 signaling.