Review Article 0022-2836/� 2024 The Author by/4.0/). GCN2 in Viral Defence and the Subversive Tactics Employed by Viruses Victoria J. Gibbs 1, Yu H. Lin 1, Aditi A. Ghuge 1, Reuben A. Anderson 1, Anja H. Schiemann 1, Layla Conaglen 1, Bianca J. M. Sansom2, Richard C. da Silva 2,3, and Evelyn Sattlegger 1,2,4,⇑ 1 - School of Food Technology and Natural Sciences, Massey University, Palmerston North, New Zealand 2 - School of Natural Sciences, Massey University, Auckland, New Zealand 3 - Genome Biology and Epigenetics, Department of Biology, Utrecht University, Utrecht, the Netherlands 4 - Maurice Wilkins Centre for Molecular BioDiscovery, Palmerston North, New Zealand Correspondence to Evelyn Sattlegger: e.sattlegger@massey.ac.nz (E. Sattlegger) https://doi.org/10.1016/j.jmb.2024.168594 Edited by Eric O. Freed Abstract The recent SARS-CoV-2 pandemic and associated COVID19 disease illustrates the important role of viral defence mechanisms in ensuring survival and recovery of the host or patient. Viruses absolutely depend on the host’s protein synthesis machinery to replicate, meaning that impeding translation is a powerful way to counteract viruses. One major approach used by cells to obstruct protein synthesis is to phospho- rylate the alpha subunit of eukaryotic translation initiation factor 2 (eIF2a). Mammals possess four differ- ent eIF2a-kinases: PKR, HRI, PEK/PERK, and GCN2. While PKR is currently considered the principal eIF2a-kinase involved in viral defence, the other eIF2a-kinases have also been found to play significant roles. Unsurprisingly, viruses have developed mechanisms to counteract the actions of eIF2a-kinases, or even to exploit them to their benefit. While some of these virulence factors are specific to one eIF2a-kinase, such as GCN2, others target all eIF2a-kinases. This review critically evaluates the current knowledge of viral mechanisms targeting the eIF2a-kinase GCN2. A detailed and in-depth understanding of the molecular mechanisms by which viruses evade host defence mechanisms will help to inform the development of powerful anti-viral measures. � 2024 The Author(s). Published by Elsevier Ltd. This is an open access article under the CCBY license (http://creativecom- mons.org/licenses/by/4.0/). Introduction All viruses completely rely on the host cell’s translation machinery to produce viral proteins and replicate. Hence, shutting down � or drastically reducing � protein synthesis is an effective mechanism for combating viral infection. In extreme cases where the infected cell cannot resume synthesis of even its most fundamental proteins, the cell might initiate a controlled cell death; this would be a form of altruistic cellular suicide for the greater good of the organism. (s). Published by Elsevier Ltd.This is an open ac Eliminating the viral threat before the cell succumbs to the stress is key to preventing such drastic measures. Host cells have developed various strategies to alter the rate of protein synthesis. One powerful mechanism involves a family of protein kinases that phosphorylate the a subunit of the eukaryotic translation Initiation Factor 2 (eIF2a)1–3). Mam- malian cells harbour four eIF2a-kinases.1,4 These are General Control Non-derepressible 2 (GCN2, found in virtually all eukaryotes), encoded by the gene Eukaryotic translation Initiation Factor 2 Alpha cessarticle under theCCBY license (http://creativecommons.org/licenses/ Journal of Molecular Biology 436 (2024) 168594 mailto:e.sattlegger@massey.ac.nz https://doi.org/10.1016/j.jmb.2024.168594 http://creativecommons.org/licenses/by/4.0/ http://creativecommons.org/licenses/by/4.0/ http://creativecommons.org/licenses/by/4.0/ http://creativecommons.org/licenses/by/4.0/ https://doi.org/10.1016/j.jmb.2024.168594 V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 Kinase 4 (EIF2AK4 in mammals; Protein Kinase R (PKR, EIF2AK2, found in vertebrates); Haem-Reg- ulated Inhibitor of translation (HRI, EIF2AK1, found in some fungi and animals); and PKR-like Endo- plasmic Reticulum Kinase (PEK/PERK, EIF2AK3, found in animals).5 While each member of this fam- ily responds to a specific set of cues � which in large part are stress factors � they all have in com- mon the eIF2a Protein Kinase (PK)-domain. This domain becomes enzymatically active upon detect- ing a stimulating cue, leading to the phosphorylation of a specific amino acid in eIF2a (Ser-51 in the yeast Saccharomyces cerevisiae and mammals). This triggers a series of responses that ultimately enables the cell to counteract the initial insult. Apart from the PK-domain, each member of the eIF2a- kinase family harbours their own unique domains. These distinct domains allow each eIF2a-kinase to detect their specific activating cues. Despite responding to different stimuli, these eIF2a- kinases are all part of a signal transduction pathway that converges at eIF2a phosphorylation; therefore, this response system has been called the Inte- grated Stress Response (ISR) (Figure 1). Each of the eIF2a-kinases are best known for responding to specific stresses. GCN2, the focus Figure 1. Schematic diagrams of the four mammalian eI harbour four distinct eIF2a-kinases, namely GCN2, PKR, PE eIF2a Protein Kinase (PK)-domain, but also contain unique unique domains are required for the detection of distinctive phosphorylation of the eIF2a protein kinase (not shown in F subunit. eIF2a phosphorylation elicits a dual response. On other hand, translation of specific mRNAs, notably that of increased ATF4 protein levels lead to a shift in the cell’s gen Despite responding to different stimuli, eIF2a-kinases are p eIF2a; therefore, this response system has been called the 2 of this review, was first discovered to help cells cope with starvation for nutrients such as amino acids, and to this point this remains its best studied function. HRI, on the other hand, is named after its role in adjusting globin synthesis to the availability of haem in red blood cells. The characteristic role of PERK is to adjust the rate of protein synthesis to that of protein folding in the endoplasmic reticulum. Whilst all three of these kinases have been linked to combatting viral infection, it is PKR that is mainly recognised for this role. PKR has been found to be required for combating a large array of viruses.6 PKR can be activated in many ways in response to viral infection, such as by viral double-stranded RNA (dsRNA, its canonical activator molecule). Activation can also occur via conditions likely to result from viral infection, such as ER stress and cytokines resulting from an immune response (for a review, see 7). Therefore, it comes as no surprise that PKR has been consid- ered the main eIF2a-kinase involved in viral defence. Thus far, research on eIF2a-kinases and their role in viral defence primarily centres around PKR, and several reviews on this topic have been published, e.g.6,8,9 However, PKR does not con- F2a-kinases and the ISR signalling pathway. Mammals K/PERK, and HRI. All eIF2a-kinases share a conserved domains. For more detail on these domains see.181 The activating cues by each eIF2a-kinase, leading to auto- igure), followed by the phosphorylation of eIF2 at its a one hand global protein synthesis is decreased. On the ATF4, is increased. ATF4 is a transcription factor, and e expression profile to help cells combat the initial insult. art of a signal transduction pathway that converges on Integrated Stress Response (ISR). V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 tribute to the defence against all viruses. For exam- ple, the anti-viral response to influenza and Vac- cinia Virus (VV) is not observably affected in mice lacking PKR.10 Instead, findings suggest that another yet to be identified eIF2a-kinase triggers the response against these viruses.10 Critically, a specific virus does not necessarily stimulate all eIF2a-kinases. Instead, the type of infecting virus determines which eIF2a-kinase(s) will become acti- vated and mount an anti-viral response. Hence, it appears that eIF2a-kinases collectively contribute to viral defence. This underscores the need to gain a full picture of the roles each eIF2a-kinase plays in viral defence. For this reason, this review focusses solely on GCN2, the molecular processes by which it becomes activated during viral infection, and the mechanisms employed by viruses to counteract GCN2. The review commences by presenting a few examples of the importance of GCN2 in viral defence (see I), followed by an overview of the signalling pathways governing GCN2 (see II). This will then allow the reader to delve into the mechanisms by which GCN2 recognises and antagonises viral infections (see III), and to fully appreciate the ways by which viruses counteract GCN2 (see IV). A selection of experiments is presented to showcase how specific findings led to the discovery of the underlying molecular mechanisms. While this review intends to give a comprehensive evaluation of the knowledge acquired to date on this topic, it does not intend to be exhaustive, and we apologise to the authors whose work is not cited. I) Relevance of GCN2 in Viral Defence Studies continue to demonstrate the importance of GCN2 in the defence against viruses. For instance, GCN2 promotes host cell survival following infection with certain viruses. When infected with Sindbis Virus (SV), immortalised Mouse Embryonic Fibroblasts (MEFs) lacking GCN2 (GCN2�/�) display a higher mortality rate than MEFs containing GCN2 (GCN2+/+).11 Simi- larly, mice infected with Mouse Cytomegalovirus (MCMV) show a 20% lower survival rate when they lack functional GCN2, as compared to mice con- taining functional GCN2.12 This suggests that GCN2 is relevant for the host cells to overcome viral infection. In line with this, published studies suggest that GCN2 can hamper the replication of certain viruses. In one such example, in Henrietta Lacks (HeLa) P4 cells, the knockdown of GCN2 via small inhibitory RNA silencing (siRNA) leads to an almost 2-fold increase in infectivity to Human Immunodeficiency Virus 1 (HIV-1).13 Furthermore, as one would expect, overexpression of functional 3 GCN2 in NIH 3T3 cells (a MEF cell line) severely reduces the replication of SV, whereas this is not observed in cells overexpressing catalytically inac- tive GCN2.11 Although these observations were derived from cultured cell lines, they seem to trans- late to an entire organism as well. For example, in studies using SV and living mice, it was found that GCN2 is important for viral defence during early stages of infection. In particular, it was found that three to four days post-nasal infection, GCN2-/- mice have significantly higher SV titres in their brains as compared to wildtype animals, while at 5 days post-infection the viral titre in GCN2-/- and wildtype brain are similar.11 Together this shows that GCN2 can dampen viral replication, although its relevance may be influenced by the stage of infection. In line with the fact that GCN2 controls protein synthesis, one would expect that GCN2 blunts viral replication by severely reducing the rate of global protein synthesis. Accordingly, total protein synthesis in HeLa P4 cells is markedly reduced in response to early HIV-1 infection, and this is not observed in cells with reduced GCN2 abundance. This is in line with the idea that GCN2 is required for dampening protein synthesis in response to viral infection.13 Given that viruses depend on the host’s translation machinery, one would expect that GCN2-mediated suppression of translation also impedes translation of viral proteins. If this is the case, then cells lacking functional GCN2 should show increased synthesis of viral proteins. This has in fact been observed. For instance, SV- infected GCN2�/� MEFs produce significantly more viral proteins as compared to wildtype MEFs, and this correlates with higher mortality of GCN2-/- MEFs.11 Similar findings were obtained for Vesicular Stomatitis Virus (VSV).11 In NIH 3T3 cells, overexpression of GCN2 reduces the amount of SV or VSV proteins produced, but overexpres- sion of catalytically inactive GCN2 does not elicit this effect, demonstrating that GCN2 activity is required to dampen translation of viral proteins.11 Finally, since GCN2 needs to be activated to suppress protein synthesis, it is reasonable to expect that viral infection would trigger the activation of GCN2. GCN2 activation can be easily scored by assessing its auto-phosphorylation (a key step in GCN2 activation), and subsequent phosphorylation of its substrate, eIF2a. For example, HeLa P4 cells infected with HIV-1 show increased GCN2 auto-phosphorylation as compared to uninfected cells,13 indicating that GCN2 senses HIV-1 infection. Furthermore, infec- tion of MEFs with Murine Norovirus (MNV) leads to increased eIF2a phosphorylation levels, but this is not the case in GCN2-/- MEFs.14 The fact that the presence of GCN2 in MEFs is required to V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 increase eIF2a phosphorylation demonstrates that GCN2 is the sole eIF2a-kinase relevant for detect- ing MNV infection. Compounds inhibiting the catalytic activity of GCN2 have been used to test whether GCN2 is required for viral defence. For example, the GCN2 inhibitor A-92 hampers the increase in eIF2a phosphorylation in MNV-infected MEFs in a dose- dependent manner.14 Furthermore, the GCN2- inhibiting compound SP600125 dampens nutrient deprivation-induced HIV-1 transcription in J-Lat A1 cells (a Jurkat cell-based model of latent HIV infec- tion), and HIV-1 reactivation in U1 cells (a subclone of HIV-1-infected U937 promonocytic cells).15 These findings imply the relevance of GCN2 in viral defence, though additional experiments may be required to validate that these observations were not due to these compounds also inhibiting other kinases, in particular other eIF2a-kinases. Compounds activating GCN2 have also been used to test the antiviral potential of GCN2 activation. For instance, the compound MG132 leads to enhanced eIF2a phosphorylation by inhibiting the proteasome – a cellular complex that degrades proteins that are damaged or no longer needed.16 Enhanced eIF2a phosphorylation not only correlates with a reduction in total protein syn- thesis, but also a reduction in VSV protein synthesis in infected MEFs. This phenomenon is not observed in GCN2-/- MEFs, where a significant reduction in eIF2a phosphorylation was also observed.16 This indicates that proteasomal inhibi- tion by MG132 stimulates GCN2-mediated eIF2a phosphorylation, and that GCN2 activation is required for the antiviral effect of MG132. This raises the intriguing possibility that pharmacological activation of GCN2 could be used as a means to reduce viral protein synthesis in the case of VSV. It will be interesting to investigate whether this mechanism of GCN2 activation is an effective mea- sure in whole organisms, and whether such mea- sures could be used to treat other viral infections. Together, these few examples clearly demonstrate the significant role that GCN2 plays in the cellular defence against viruses. Upon sensing viral infection, GCN2 effectively dampens mRNA translation, thereby hampering the production of viral proteins. Since protein production is crucial for viruses to replicate and infect other cells, this reduces the severity of infection and promotes survival of host cells. GCN2 could therefore act as an early responder to viral infection to mitigate its impact on the organism while other components of the cellular anti-viral response act to remove the threat. II) The GCN2 Protein Before discussing the links between GCN2 and viruses in molecular detail, it is necessary to first 4 introduce GCN2 and some of the additional components of the ISR. IIa) The domain structure of GCN2 GCN2 is a highly conserved protein found in virtually all eukaryotes.1 So far five domains have been identified (Figure 2). The N-terminal RWD domain (named for its presence in RING finger con- taining proteins, WD-repeat containing proteins, and yeast DEAD (DExD)-like helicases) is essential for direct binding to the effector protein GCN1. The RWD domain is followed by a charged region, and then a pseudokinase domain (YPK). The YPK domain bears homology to Protein Kinase (PK)- domains but it lacks some of the amino acids critical for enzymatic activity. Adjacent to the YPK domain is the PK-domain which phosphorylates eIF2a. C- terminal to the PK-domain is a domain with homol- ogy to Histidyl-tRNA Synthetases (HisRS-like domain) which is not enzymatically functional as an aminoacyl-tRNA synthetase. Together with the C-terminal Domain (CTD), the HisRS-like domain specifically binds uncharged tRNAs.17 The HisRS- like domain harbours conserved residues corre- sponding to those required for tRNA binding in aminoacyl-tRNA synthetases, such as the amino acids Tyr-Arg (position 1050–1051 in yeast Gcn2) within the m2 motif.18 Gel shift assays have shown that these amino acids are critical for binding to uncharged tRNAs.17 Finally, the CTD serves as a ribosomal binding site, and also constitutes the major site for GCN2 dimerisation.19–21 Since several viral proteins target the PK-domain of eIF2a-kinases (see IVa-b), it is useful to understand the architecture of this domain in more detail. As a typical member of Ser/Thr kinases, the PK-domain contains 12 subdomains (I-XII). Each of these are recognisable by containing distinctive patterns of conserved residues that are not interrupted by large amino acid insertions.22,23 Structurally, the PK-domain consists of the N (N- terminal) and C (C-terminal) lobe, with the catalytic cleft positioned between these.22–27 While the N- lobe (encompassing subdomains I-IV) mediates dimerisation of the PK-domain and plays a critical role in anchoring and orienting the nucleotide ATP, the C-lobe (VI-XI) is involved in both binding the substrate eIF2a and in catalysing the transfer of the ATP terminal phosphate group to eIF2a. Sub- domain V spans the two lobes, and functions as a hinge that can allow inter-lobe mobility. The C- lobe also contains the activation loop which encom- passes auto-phosphorylation sites critical for the PK-domain to adopt a catalytically active conforma- tion.28 Two separate features in the C-lobe facilitate binding of the substrate eIF2a, which are the aG helix and the P + 1 loop within the activation segment.25,29 GCN2 resides in the cell as a latent dimer, facilitated by autoinhibitory intramolecular interactions.19,24,30–32 Contributing to the latency Figure 2. Domain composition of GCN2 and the mechanisms proposed to activate GCN2. GCN2 is composed of several domains (from the N- to the C-terminus, image not to scale): The N-terminal RWD domain essential for direct interaction with the effector protein GCN1; the enzymatically inactive pseudokinase domain (YPK); the Protein Kinase (PK)-domain responsible for phosphorylating eIF2a; the domain with homology to Histidyl tRNA Synthetases (HisRS-like domain) containing the m2 motif responsible for binding uncharged tRNA; and the C- terminal Domain (CTD) harbouring the major dimerisation domain as well as the major site for ribosome binding. The HisRS-like domain, together with the CTD, has affinity to uncharged tRNAs. a-c. Several mechanisms have been proposed to stimulate GCN2: a. Amino acid starvation leads to the accumulation of uncharged tRNAs, which then bind to the HisRS-like domain; b. Under certain conditions, the ribosomal P-stalk can directly bind to and stimulate GCN2; c. Ribosome collision was proposed to promote GCN2 activation. d. GCN2 stimulation and concomitant GCN2 auto-phosphorylation e. leads to GCN2-mediated phosphorylation of eIF2a. V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 of GCN2, the rigidity of the hinge leads to a closed bi-lobal conformation, with the active site cleft being partially closed, and the ATP-binding pocket occluded. Furthermore, in the N-lobe the unfavour- able orientation of the aC helix prevents the proper positioning of ATP in the catalytic site.24,31–35 After recognising a stimulating signal, GCN2 undergoes conformational changes, relieving autoinhibitory intramolecular interactions and eliciting GCN2 auto-phosphorylation, allowing GCN2 to finally adopt a fully active conformation.19,24,27,28,30–35 GCN2 is then capable of phosphorylating its sub- strate eIF2ɑ. It is for this reason that GCN2 auto- phosphorylation is considered a reliable measure of GCN2 activation. IIb) Mechanism of GCN2 activation in uninfected cells Studies so far suggest that GCN2 protects host cells against viruses exclusively in its activated state. Therefore, it comes as no surprise that many viral mechanisms have arisen that target the process of GCN2 activation. For this reason, this chapter provides an overview of the current knowledge on how GCN2 is stimulated in ‘uninfected’ cells (Figure 2). GCN2 activation plays many roles in the cell aside from defending against viruses, such as its well-established role in 5 responding to amino acid starvation. Whilst much research has been done to better understand this activation pathway, the molecular mechanisms underlying the activation of GCN2 still remain elusive. Currently, studies suggest more than one mechanism of GCN2 activation. Since this topic is not the main focus of this review, the current working models for GCN2 activation are only briefly described, without outlining all scientific discoveries that have led to these models. Currently, two models have been proposed for GCN2 activation. The initially proposed � and still current � working model is based on the fact that under amino acid starvation the respective tRNAs cannot be aminoacylated, leading to the accumulation of uncharged tRNAs (Figure 2a). These are recognised by GCN2 as a direct starvation signal.17,36,37 The exact mechanism by which GCN2 detects these uncharged tRNAs is still unknown. It has been proposed that, when there is a shortage of an aminoacylated tRNA required for translation, a cognate but uncharged tRNA can enter the ribosomal Aminoacyl acceptor site (A- site) in its place. This uncharged tRNA is then detected by the HisRS-like domain of GCN2, lead- ing to GCN2 auto-phosphorylation (Figure 2d). As GCN2 auto-phosphorylation allows the PK-domain to adopt an active conformation (and thus is a mark of GCN2 activation), GCN2 may then phosphory- V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 late its substrate, eIF2a (Figure 2e). The protein GCN1 is required for GCN2 activation in vivo.38 GCN1 directly binds GCN2,39,40 and both proteins need to bind to ribosomes to allow for the activation of GCN2.20,21,39,41,42 The role of GCN1 may be to promote delivery of uncharged tRNAs to the A- site, to promote delivery of the uncharged tRNA to GCN2, and/or to position GCN2 on the ribosome in such a way that it can access the uncharged tRNA within the A-site.39 Further supporting the idea that uncharged tRNAs are an activating ligand for GCN2 was obtained by an independent study.43 Treatment of Human Embryonic Kidney (HEK)-293 T cells with halofuginone leads to the specific accumulation of uncharged tRNAPro, and this is associated with an increase in GCN2 auto-phosphorylation and eIF2a phosphorylation.43 Quantitative Reverse Transcrip- tase Polymerase Chain Reaction (qRT-PCR) con- ducted on GCN2-precipitates show a 3-fold enrichment of tRNAPro, as compared to cells not treated with halofuginone. Together with the fact that halofuginone treatment leads to a dramatic reduction in the abundance of aminoacylated tRNAPro, this supports the idea that the uncharged form of tRNAPro is detected by and bound to GCN2. The HisRS-like domain is involved in detect- ing the uncharged tRNAPro, as is suggested by the fact that the co-precipitation of tRNAPro with GCN2 is abolished by the m2 mutation in the HisRS-like domain.43 Research findings have led to the proposal of a second working model that suggests a more indirect involvement of tRNAs in the activation of GCN2. Instead, the ribosomal P-stalk is implicated in mediating GCN2 activation (Figure 2b).44–46 The P-stalk is involved in the cyclic recruitment of elongation factors during each round of translation elongation.47–49 Given that the absence of a cog- nate aminoacylated tRNA causes a pause in trans- lation, this working model proposes that the resulting stalled ribosome does not recruit transla- tion factors to its ribosomal P-stalk. Consequently, the P-stalk is able to instead contact and activate GCN2. The uncharged tRNA and P-stalk activation models for GCN2 are not mutually exclusive. There is evidence that they come into play differentially, depending on whether the stress signal is starvation-dependent or starvation- independent.50 Nevertheless, a recent study in yeast has shown that Gcn2 activation requires Gcn1, Gcn2-ribosome association, and the Gcn2 HisRS-like domain for detecting either type of acti- vating cue.50 A model was proposed in which the mechanism of GCN2 activation depends on the activating cue. Under starvation conditions, uncharged tRNA binds to the HisRS-like domain to stimulate GCN2. In contrast, during starvation- independent ribosome stalling, P-stalk proteins instead interact with the HisRS-like domain to stim- 6 ulate GCN2.50 An independent study in mammalian cells led to similar conclusions, in that GCN2 is acti- vated by at least two mechanisms, some of which necessitate binding of uncharged tRNA to GCN2, while others require ribosome stalling/collisions.43 This study also found that the GCN2 HisRS-like domain is required for both mechanisms of GCN2 activation.43 In addition to adjusting protein synthesis in response to external stresses such as starvation, it has been proposed that GCN2-activation also serves as a mechanism to decrease the likelihood of ribosome collisions (Figure 2c).51,52 Supporting this idea, GCN2 can become activated by ribosomal stalling in a manner that seems unrelated to the presence of uncharged tRNA mole- cules.13,39,41,45,46,48–51 An elongating ribosome that experiences stalling can form a roadblock, into which a succeeding elongating ribosome collides, resulting in a disome.52 GCN2 activation, which involves GCN2 auto-phosphorylation (Figure 2d), and subsequent eIF2a phosphorylation (Figure 2e) leads to a reduction in global protein synthesis, resulting in a ‘lighter’ load of mRNAs with elongating ribosomes and thereby reducing the probability of further ribosome collisions. GCN1 has been found to associate with disomes, raising the possibility that disome-bound GCN1 elicits GCN2 activation.51,53,54 Supporting this idea, studies suggest that an increase in transient colli- sions leads to GCN2 activation (likely in conjunction with GCN1).55,56 Hence, the formation of disomes may be implicated in GCN2 activation, though it remains to be validated experimentally whether GCN2 also binds to disomes for its activation. Ribosome stalling and the resulting collisions are considered to be the key cellular indicators of aberrant translation, which also triggers the activation of the Ribosome Quality Control (RQC) system. The RQC mechanism ensures that stalled ribosomes are resolved and recycled to re-enter translation, a process that involves ubiquitination of ribosomal proteins.56 Research suggests that the GCN2/ISR and the RQC pathways are linked,54,57 and studies in mammals have led to a model where the persistence of disomes determi- nes the type of cellular response.51,56 Under unstressed conditions, short-lived disomes are resolved via dedicated pathways such as RQC. Severe collisions would stimulate the ribotoxic stress response pathway, leading to cell cycle arrest or apoptosis. Under ‘intermediate’ stress on the other hand, GCN2-mediated eIF2a phosphory- lation decreases the rate of translation initiation to reduce the ribosomal load on mRNAs and thus the likelihood of further ribosome collisions.51 The evidence so far strongly points to a coordinated interplay between the GCN2/ISR and RQC regula- tory pathways in sensing unresolved ribosome stalling/collision in eukaryotes.51,56 This interplay is of significance to viral replication as well, given V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 that viruses place a heavy burden on the host’s translation machinery to produce large quantities of viral proteins. Due to the remarkable conservation of GCN2 and its signalling pathway, the budding yeast Saccharomyces cerevisiae was instrumental in uncovering the basic biological function of Gcn2, how Gcn2 becomes activated and how it is regulated. The knowledge gained in this most amenable eukaryotic model organism led to a surge in research focused on GCN2 in higher eukaryotes, particularly in relation to health and disease in humans and mammalian models. Some recent studies suggest that the human activation pathway may contain additional subtleties, but this area warrants further investigation. Therefore, the exact mechanisms for GCN2 activation under different stress conditions remain to be further investigated in both yeast and human systems. IIc) The function of eIF2a eIF2a is one of the three subunits of eIF2, which is essential for facilitating the initiation of protein synthesis.58 eIF2 forms a ternary complex with GTP and initiator methionyl tRNA (Met-tRNAi Met) to aid the ribosome in detecting the translation start codon. After subsequent GTP hydrolysis, eIF2 is released in its GDP-bound state. The GDP then needs to be replaced by GTP, which is mediated by the guanine nucleotide exchange factor called eukaryotic translation Initiation Factor 2B (eIF2B). eIF2-GTP can then bind Met-tRNAi Met again to form a ternary complex capable of promoting the next round of translation initiation. Hence, a perpetual rate of GDP-GTP exchange ensures plentiful abun- dance of ternary complexes and thus supports a rapid rate of protein synthesis. Phosphorylation of Ser-51 in the alpha subunit of eIF2 by the eIF2a- kinases (Figure 1) turns eIF2 from a substrate into an inhibitor of eIF2B. The concomitant decrease in the rate of GDP-GTP exchange leads to reduced levels of ternary complex. This can affect translation in two ways. On one hand, low abundance of ternary complex leads to the increased translation of specific mRNAs (Figure 1). This regulation is mediated by specific unique upstream Open Reading Frames (uORFs) present in these mRNAs that impair translation of the main ORF under non-stressed conditions (reviewed in 3). However, low availability of ternary complexes allows ribosomes to overcome these inhibitory uORFs, leading to enhanced translation of the main ORF. These ORFs code for transcription factors, such as General control non- derepressable 4 (Gcn4) in yeast or Activating Transcription Factor 4 (ATF4) in mammals.3 Thus, increased levels of phosphorylated eIF2a (eIF2a-P) essentially lead to an increased abun- dance of these transcription factors, thereby 7 altering the expression profile of a large array of genes and allowing the cell to adjust to and over- come the initial insult.1,3 The increase in eIF2a-P levels, or the elevated translation of ATF4, has also been found to lead to enhanced translation of a selection of mRNAs containing an Internal Ribosome Entry Site (IRES).59,60 IRESs can be found in stress-responsive transcripts, such as for the high affinity cationic amino acid transporter 1 (cat-1),59 which would enhance amino acid uptake to rectify cellular shortage of amino acids. Although various molecular mechanisms have been proposed for IRES-mediated translation ini- tiation, it seems that all of these utilize translation factors other than the traditional ternary com- plex.61 The exact mechanisms by which an increase in eIF2a-P levels enhances IRES- mediated translation initiation are still being deci- phered.61 Non-AUG translation start sites provide an additional layer of translation regulation.62 Some of these start sites are not affected by the ISR response, or are even upregulated specifi- cally under ISR conditions. Concurrent with the increased translation of specific mRNAs containing unique uORFs, IRES elements, or non-AUG translation start sites, low availability of ternary complex leads to a reduction in global translation (Figure 1). The degree by which global protein synthesis is dampened depends on the extent of eIF2a phosphorylation. While strong eIF2a phosphorylation can halt translation, weak eIF2a phosphorylation may not affect global protein synthesis but still lead to increased translation of specific mRNAs.58 At inter- mediate levels, the amount of eIF2a-P could deter- mine the rate of global protein synthesis. When it comes to viral infection, the impairment of protein synthesis would be desired to prevent viral replication. III) GCN2 Stimulation by Viral Infections To be employed in viral defence, GCN2 must first detect the presence of viral infection, which in turn leads to its stimulation. The mechanism(s) of GCN2 activation upon viral infection is currently not fully understood. Nevertheless, it appears that GCN2 can be activated via different mechanisms depending on the type of virus (Table 1, Figure 3). Here, an overview of the current understanding of virally-induced GCN2 activation is provided. Generally, the virally-induced activation of GCN2 and the subsequent increase in eIF2a phosphorylation leads to the attenuation of translation, thereby diminishing the synthesis of viral proteins (Figure 3f). The last section of this chapter (see IIId) introduces alternative mechanisms of GCN2-mediated viral defence that do not rely on eIF2a phosphorylation. V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 IIIa) Viral infection eliciting or mimicking amino acid starvation In the host cell, some viruses may indirectly trigger amino acid shortage or elicit the accumulation of uncharged tRNAs, leading to GCN2 activation. Yellow Fever Virus (YFV) is one such virus that appears to lead to a depletion of amino acids (Figure 3a). For example, infection of human monocyte-derived Dendritic Cells (hmDCs) with live attenuated Yellow Fever strain 17D (YF-17D) vaccine leads to the depletion of free Arg as assessed by Liquid Chromatography-Mass Spectrometry (LC-MS).63 Exposing YF-17D to heat or UV-irradiation renders it biologically inactive. This inactivated form fails to trigger a decrease in amino acid levels, indicating the need for live virus � and thus a physiological activity by the virus � in this process. Notably, YF-17D-mediated amino acid depletion correlates with an increase in phosphory- lation of GCN2 and eIF2a.63 A similar phenomenon was found in mouse Bone Marrow–derived Den- dritic Cells (BMDCs), where YF-17D infection leads to enhanced eIF2a-P levels, but biologically inactive YF17D does not (GCN2 phosphorylation was not investigated). The mechanism causing a decrease in the amino acid pool by YF-17D remains to be uncovered. One possibility is that the increased translation of viral proteins utilises the intracellular pool of amino acids faster than they can be replen- ished, leading to the observed amino acid starvation. HIV is also known to cause dysregulation of plasma amino acid levels, including a marked decrease in many of these amino acids (e.g. 64,65). Whilst the mechanisms by which HIV might alter amino acid levels warrant a dedicated review in themselves, it is possible that HIV also activates GCN2 through this elicited depletion in amino acids (see IIb, Figure 3a). A significant reduction in sev- eral plasma amino acids has also been found after infection with Sandfly Fever Virus (SFV),66 Murine Hepatitis Virus 3 (MHV-3),67 and murine Influenza A Virus (IAV),68 suggesting that amino acid deple- tion could be a more widespread route of GCN2 activation by viruses. The reduction in serum amino acids may lead to depletion of these amino acids intracellularly where it can be sensed by GCN2. Further experiments are required to determine whether infected patients exhibiting low plasma levels of amino acids indeed show intracellular amino acid depletion and GCN2 activation. The difference in codon usage between virus and host genes may lead to increased levels of uncharged tRNAs (Figure 3b). Given that viruses rely on the translation machinery of the host, one would anticipate the codon usage of viral genes to have evolved to match that of its host. This resemblance would promote the efficient use of the host’s tRNA pool, and the rapid aminoacylation of uncharged tRNAs by the 8 matched pool of tRNA synthetases, leading to abundant translation of viral proteins. However, several viruses have been reported to exhibit codon usage patterns that diverge from those of their hosts.69 In these cases, viral protein produc- tion is limited by the available pool of rare aminoacy- lated tRNAs. As the cellular pool of aminoacyl tRNA synthetases may not be able to charge cognate tRNAs at the rate at which they are utilised in trans- lation, this could also lead to a build-up of uncharged tRNAs and subsequent GCN2 activa- tion. In this scenario, the moderated synthesis of viral proteins would not overburden the host’s trans- lation machinery, enabling survival of its host, albeit at the cost of a reduced rate of viral replication.69 Whilst there are differing views on whether these differences in codon usage emerged as a conse- quence of its impact on the host’s translation machinery, or simply due to the virus’ inherent nucleotide preference in its genetic material,69,70 this debate is beyond the scope of our review. In one example, while the total cellular tRNA composition remains unchanged, infection by VV or IAV leads to dramatic changes in the population of polysome-associated tRNAs. This strongly correlates with viral codon usage, suggesting the existence of localised tRNA pools tailored for efficient viral translation.71 As the host cell is not well-equipped to aminoacylate the rare tRNAs, their aminoacylation may not occur as quickly, and this may serve as the activating starvation signal for GCN2 (see IIb, Figure 3b). In the case of HIV-1, the codon usage of its early genes is similar to that of highly expressed host genes, whilst the codon usage of its late genes differs largely from that of its host.72,73 Interestingly, tRNA microarray studies revealed that HIV-1 pack- ages a variety of tRNAs, suggesting that the pack- aged tRNAs aid in changing the tRNA pool composition in the infected host to accommodate the virus’s codon usage and improve the translation efficiency of the late viral RNAs.73,74 This selective alteration of the tRNA pool would favour the transla- tion of HIV-1 genes over that of the host genes.73 However, the biased composition of the aminoacyl tRNA synthetases in the host cell may not be well- suited to efficiently charge the altered tRNA pool, leading to a build-up of uncharged tRNAs and thus GCN2 activation (Figure 3b). It remains to be ascer- tained whether a delay in tRNA aminoacylation can cause GCN2 activation, and this warrants further investigation. IIIb) GCN2 detecting viral RNA Another mechanism of GCN2 activation is by viral RNA binding to the HisRS-like domain of GCN2 (Figure 3c). Analogous to uncharged tRNAs binding to the HisRS-like domain, viral RNA can bind to this domain and lead to the stimulation of the GCN2 protein kinase (PK) domain. For example, in in vitro kinase assays, SV RNA is Table 1 Overview of viruses for which GCN2 activation has been reported, and the mechanism of GCN2 stimulation. The hosts known to be infected as well as the diseases / symptoms caused are indicated. The content of the table is primarily sorted by the mechanism of GCN2 activation, followed by the genetic material used by the virus to generate mRNA (Baltimore classification167), and then grouped into their specific viral family. For more detail see text. Proposed mechanism of GCN2 activation See section in review Baltimore class Viral family Virus (abbreviation)* Host Disease / Symptoms / Relevance Source** Viral infection leads to amino acid depletion (Figure 3a) IIIa + ssRNA Flaviviridae Yellow fever virus (YFV) Humans Abrupt onset of fever, chills, malaise, headache, back pain nausea and dizziness. Can lead to haemorrhagic manifestations and kidney failure in severe cases. b IIIa � ssRNA Bunyaviridae Sandfly fever virus (SFV) Humans Fever, headache, fatigue, joint, muscle and abdominal pain. 168 IIIa + ssRNA Coronaviridae Murine hepatitis virus 3 (MHV-3) Mice Causes hepatitis 169 IIIa � ssRNA OrthomyxoviridaeInfluenza A virus (IAV) Humans, pigs, horses Causes influenza, symptoms include sudden onset of fever, headache, body aches, fatigue and a dry cough. 170 IIIa + ssRNA- RT Retroviridae Human immunodeficiency virus 1 (HIV-1) Humans Can cause severe diseases such as AIDS, secondary infections, and lymphomas. Acute phase symptoms include fever, headache, rash, sore throat weight loss, diarrhoea b Viral infection leads to accumulation of uncharged tRNAs (Figure 3b) IIIa + ssRNA- RT Retroviridae Human immunodeficiency virus 1 (HIV-1) Humans Can cause severe diseases such as AIDS, secondary infections, and lymphomas. Acute phase symptoms include fever, headache, rash, sore throat weight loss, diarrhoea b Viral RNA binds to GCN2 HisRS-like domain (Figure 3c) IIIb + ssRNA Picornaviridae Poliovirus Humans Causes poliomyelitis (polio). Symptoms include fever, fatigue, stiffness in the neck and in rare cases paralysis or death. b IIIb + ssRNA Togaviridae Sindbis virus (SV) Humans, birds Humans: Exanthema over trunk and limbs, joint symptoms. Sometimes nausea, general malaise, headache, and muscle pain. Can be asymptomatic. g IIIb + ssRNA- RT Retroviridae Human immunodeficiency virus 1 (HIV-1) Humans Can cause severe diseases such as AIDS, secondary infections, and lymphomas. Acute phase symptoms include fever, headache, rash, sore throat weight loss, diarrhoea b Viral infection enhances ribosome- stalling (Figure 3d) IIIc, IVg dsDNA Herpesviridae Epstein-Barr virus (EBV) Humans Can be asymptomatic. Possible symptoms include tiredness, fever, sore throat, headaches and body aches, swollen lymph nodes, swelling in the liver and/or spleen, rash, causing mononucleosis. b, f, i, h IVg dsDNA Herpesviridae Human Cytomegalovirus (HCMV) Humans Can be asymptomatic. Possible symptoms include fever, sore throat, fatigue, swollen glands. Occasionally mononucleosis or hepatitis. f, h IIIc dsDNA Poxviridae Vaccinia virus (VV) Humans, cattle Symptoms are usually very mild to non-existent in humans but may produce localised skin infection. j IIId � ssRNA Rhabdoviridae Vesicular stomatitis virus (VSV) Humans, livestock, wild deer, rodents Human: Flu-like symptoms that can lead to encephalitis Livestock: lesions in the mouth and other parts of the body (e.g. feet, ears, udder, ventral abdomen), secondary infections. f,e (continued on next page) V .J . G ib b s , Y .H . L in , A .A . G h u g e , e t a l. J o u rn a l o f M o le c u la r B io lo g y 4 3 6 (2 0 2 4 ) 1 6 8 5 9 4 9 T a b le 1 (c o n ti n u e d ) P ro p o s e d m e c h a n is m o f G C N 2 a c ti v a ti o n S e e s e c ti o n in re v ie w B a lt im o re c la s s V ir a l fa m il y V ir u s (a b b re v ia ti o n )* H o s t D is e a s e / S y m p to m s / R e le v a n c e S o u rc e ** G C N 2 e x h ib it s a n a n ti v ir a l e ff e c t, G C N 2 a c ti v a ti o n m e c h a n is m u n k n o w n I + s s R N A C a lic iv ir id a e M u ri n e N o ro v ir u s (M N V ) M ic e A ff e c ts im m u n o c o m p ro m is e d m ic e , c lin ic a l s ig n s in c lu d e w a s ti n g , d ia rr h o e a a n d d e a th . A ls o , m a y d e v e lo p h e p a ti ti s , p e ri to n it is a n d in te rs ti ti a l p n e u m o n ia . a IV a + s s R N A F la v iv ir id a e D e n g u e v ir u s 2 (D E N V -2 ) H u m a n s S y m p to m s c a n in c lu d e fe v e r, h e a d a c h e , re tr o o c u la r p a in , m y a lg ia , a rt h ra lg ia , e x a n th e m a , a n d p ro s tr a ti o n w it h o r w it h o u t h a e m o rr h a g e . C a n b e a s y m p to m a ti c . h II Id + s s R N A - R T R e tr o v ir id a e M u ri n e le u k a e m ia v ir u s (M L V ) M ic e , ra ts C a n c e r in m ic e , m a y in d u c e n e u ro d e g e n e ra ti v e d is o rd e rs a n d p a ra ly ti c d is e a s e s . 1 7 1 * N o te th a t th e re m a y b e m o re v ir u s e s th a n th o s e m e n ti o n e d th a t h a v e th e s a m e e ff e c t o n G C N 2 a c ti v a ti o n , fo r m o re s e e te x t. ** F o r s o u rc e o f in fo rm a ti o n , s e e re fe re n c e s (p u b lic a ti o n s ) o r b o tt o m o f T a b le 2 (i n te rn e t lin k s ). V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 10 sufficient to enhance GCN2 auto-phosphorylation as well as eIF2a phosphorylation.11 Two short non-contiguous sequences called GCN2-Activating RNA (GAR) in the SV RNA were found to be suffi- cient for stimulating GCN2. Given that denatured GAR has no effect, this suggests that its secondary structure is essential for stimulating GCN2. In vitro northwestern assays revealed that GAR can bind to GCN2, but not to GCN2 with a mutated m2 motif.11 Since the detection of uncharged-tRNAs also requires the m2 motif in the HisRS-like domain17 this suggests that the same parameters mediate SV RNA binding in the HisRS-like domain as those found for tRNA binding. It is possible that this bipartite GAR adopts a structure resembling that of a tRNA that allows its detection by the HisRS-like domain.11 HIV-1 RNA has also been found to activate GCN2 in vitro. As found above, the GCN2 m2 mutation prevents HIV-1 RNA from eliciting phosphorylation of GCN2 and eIF2a,75 supporting the idea that HIV-1 RNA also activates GCN2 by binding to the HisRS-like domain. It will be interesting to validate whether this mechanism of GCN2 activation also occurs in vivo. Furthermore, the HIV-1 RNA sequence activating GCN2 remains to be determined. It is possible that GCN2 could also detect viral RNA following poliovirus infection. Initially, it was reported that poliovirus does not elicit an increase in eIF2a phosphorylation,76 but three independent studies found that poliovirus leads to GCN2 activa- tion, or to an increase in eIF2a phosphoryla- tion.11,16,77 It was suggested that the initial finding could have been due to differences in experimental procedure.77 With the advancement of technology and improvements in experimental methods since the initial report, this may be the reason why the effect of poliovirus is now detectable. Infection of HeLa cells with poliovirus leads to an increase in eIF2a phosphorylation, which is not as strong in HeLa cells infected with a mutant poliovirus.77 This correlates with an accumulation of dsRNA in cells, which occurs faster in cells infected with wild-type poliovirus than in cells infected with themutant, sug- gesting that dsRNA is the activating ligand.77 The eIF2a-kinases activated in this scenario likely include GCN2, given that the poliovirus RNA gen- ome activates GCN2 in vitro,11 although these find- ings have not been published yet. It would be interesting to confirm whether GCN2 is indeed acti- vated in vivo, and whether this is mediated by polio- virus RNA. IIIc) Viral infection eliciting ribosome stalling To enable viral replication, viruses exploit the host’s translation machinery which can be heavily burdened by the viral load, leading to an increase in ribosome stalling and collisions (Figure 3d). Consistent with this idea, it appears that VV infection aggravates ribosome collisions due to Figure 3. Mechanisms of GCN2 activation by viruses and the downstream anti-viral effects. a-d. Viruses can activate GCN2 through several proposed mechanisms. a. Depletion of the host’s intracellular amino acid reserve by viral replication elicits the cell’s GCN2-mediated amino acid starvation response. b. Increase in the abundance of intracellular uncharged tRNAs are detected by the GCN2 HisRS-like domain. c. Viral RNA binds to the GCN2 HisRS-like domain, leading to GCN2 activation. d. Virally caused promotion of ribosome collisions leads to GCN2 activation. e-f. Activated GCN2 elicits several anti-viral activities. e. GCN2 can directly phosphorylate the Integrases (IN) of viruses belonging to the Retroviridae family. This interferes with viral integration into the host genome, ultimately resulting in reduced viral load and infectivity. f. The reduction in global protein synthesis may limit the synthesis of viral proteins and thus hamper viral replication. For more see text. V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 both the increased burden on the translation machinery and the virally-induced reduction of eIF2a-phosphorylation. Signifying that the translation machinery is burdened, VV propagation enhances ubiquitination of a ribosomal protein. This is indicative of an increased burden on the RQC pathway,78 which is required for resolving stalled/collided ribosomes (see IIb). Furthermore, ribosome collisions during infection are more severe when eIF2a is rendered unable to be phosphorylated.78 This strongly impli- cates one of the eIF2a-kinases in responding to and managing the increased translational burden caused by VV infection. GCN2 is the most likely candidate, given that GCN2 is activated by condi- tions which promote ribosome collisions (see IIb).54 This highlights the possibility that GCN2 senses VV infection indirectly through the concomi- tant increase in ribosome collisions. In further support of the idea that viruses can trigger GCN2 activation by promoting ribosome collisions, Epstein-Barr Virus (EBV) activates GCN2 in conjunction with reducing the RQC response.79 EBV produces a large tegument pro- tein called BPLF1 which harbours a Ubiquitin 11 Deconjugase (vDUB) at its N-terminus. BPLF1 reduces the ubiquitination of the 40S ribosome usually associated with the RQC response,56 and this is dependent on its vDUB catalytic activ- ity.79 This suggests that BPLF1 counteracts the host’s RQC response and thus impairs the host’s ability to resolve stalled ribosomes. Intriguingly, the BPLF1-mediated impairment of the host’s RQC is associated with an increase in eIF2a phosphorylation levels, and this is dependent on GCN2 activity.79 Together, this suggests that the increased level of stalled ribosomes � result- ing from RQC inhibition � are sensed by GCN2, leading to subsequent GCN2 activation (Figure 3d). More studies are required to further investigate the potential interplay between viral infections, GCN2 activation, and the RQC. Whilst the mechanisms of GCN2 activation in VV and EBV infection remain to be uncovered, the findings so far, along with the indication that GCN2 can be activated by ribosome collisions (see IIb), suggest that a virally-induced increase in ribosome collisions could be another avenue for GCN2 activation by viruses (Figure 3d). V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 IIId) Anti-viral mechanisms involving GCN2 but not eIF2a phosphorylation GCN2 activation, and the subsequent eIF2a phosphorylation, result in impaired global protein synthesis, thereby hampering the translation of both host and viral proteins (Figure 3f). Since viruses are entirely dependent on the host’s translation machinery, its inaccessibility will inevitably suppress viral replication. Interestingly, in response to VSV and HIV infections, GCN2 has been observed to also elicit anti-viral effects independently of eIF2a phosphorylation. GCN2 counteracts VSV infection without the need for eIF2a phosphorylation. Importantly, VSV infected GCN2-/- MEFs produce more viral proteins than wildtype MEFs, indicating that GCN2 is relevant for the defence against VSV.80,16 Inter- estingly though, wildtype MEFs and MEFs contain- ing eIF2a-S51A (to render eIF2a non- phosphorylatable) show similar amounts of viral protein production. This would indicate that the phosphorylation of eIF2a by GCN2 may not be essential for the defence against VSV, or that GCN2 phosphorylates a substrate other than eIF2a that is relevant to VSV defence. Supporting this idea, GCN2 has been reported to phosphorylate proteins other than eIF2a.81 Further research is necessary to uncover the precise mechanism by which GCN2 supports the anti-viral defence against VSV. It also needs to be verified whether the GCN2 catalytic activity per se is required to convey these anti-viral effects. Another example centres around the fact that GCN2 can bind to and phosphorylate the Integrase enzyme (IN) of HIV-1 (Figure 3e).13,82 IN mediates the integration of retroviral DNA into the host genome, an essential step for retrovirus replication.83 Studies revealed that GCN2 phospho- rylates IN at two highly conserved positions � Ser- 24 and Ser-255 � with Ser-255 being the main phosphorylation site.82 IN S255A substitution (to render Ser-255 non-phosphorylatable) leads to an increase in infectivity as well as viral DNA integra- tion. Strikingly, the same effect was observed with wildtype IN but in host cells lacking GCN2. Together, this supports the idea that GCN2- mediated phosphorylation of IN hampers integra- tion efficiency.82 GCN2 can also phosphorylate IN produced by the retroviruses HIV-2, Murine Leu- kaemia Virus (MLV) and Avian Sarcoma Virus (ASV).82 Notably, ASV poses a significant threat to the poultry industry and economy, especially con- sidering that chickens are a major food source worldwide. Interestingly, GCN2 is not able to effi- ciently phosphorylate the IN of the retrovirus Proto- type Foamy Virus (PFV) which contains a Gly at the position equivalent to Ser-255.74 Thus, it is tempting to speculate that PFV has acquired this mutation to counteract this particular anti-viral mechanism of GCN2. More studies are warranted to further vali- date the IN as a GCN2 substrate and to uncover 12 the biological relevance of IN phosphorylation. Unravelling the exact mechanisms by which viral infections lead to GCN2 activation could provide valuable insights into viral pathogenesis, and poten- tially lead to the discovery of novel drugs that help enhance GCN2-mediated viral defence in patients. IV) Viral Mechanisms Counteracting GCN2 Thus far, the majority of the viral virulence factors targeting the ISR have been researched in relation to PKR, but some of these also have the potential to inhibit GCN2. In addition, viral strategies have been uncovered that specifically counteract GCN2 function in viral defence (Figure 4). In the interest of providing a comprehensive review, here we discuss any viral mechanism able to inhibit GCN2, whether they are specific to GCN2 or also target other eIF2a-kinases. Table 2 provides an overview of these mechanisms. IVa) Mimicking the eIF2a-kinase substrate eIF2a A pseudosubstrate imitates the real substrate of an enzyme but does not undergo the typical chemical reaction or process that the intended substrate would. Because the pseudosubstrate competes with the actual substrate for binding to the enzyme, it hinders the enzyme’s catalytic activity. Interestingly, viral proteins have been found that mimic eIF2a, and several of these act as bona fide eIF2a pseudosubstrates (Figure 4d). A critical characteristic of a pseudosubstrate is its ability to imitate the part of the substrate that typically binds to the enzyme. Studies suggest that eIF2a-kinases recognise a large surface region of eIF2a.29 This includes residues flanking the Ser-51 phosphorylation site and the KGYID motif located � 30 residues C-terminal of the Ser- 51 phosphorylation site (�20 �A distance to Ser- 51).29,84 These residues are part of the Oligonu- cleotide Binding (OB) fold domain (residues 1– 89/1–87 in yeast/human),85,86 and studies suggest that its 3-dimensional structural integrity is relevant for substrate-enzyme interaction.25 Considering that eIF2a-kinases share a common substrate, it is anticipated that eIF2a pseudosubstrates should hinder the activity of all eIF2a-kinases, including GCN2. VV expresses the eIF2a pseudosubstrate called K3L which inhibits all four eIF2a-kinases, including GCN2.87–90 The K3L protein mimics eIF2a in that it contains an OB-fold and a KGYID motif which is critical for kinase binding.86,91,92 As one would expect from a pseudosubstrate, studies have shown that K3L-mediated inhibition of PKR depends on residues that are conserved between K3L and eIF2a.84,92 For example, the S. cerevisiae model system was employed to gain more insight Figure 4. Overview of viral mechanisms counteracting or exploiting GCN2 and the Integrated Stress Response (ISR). Viruses have developed various mechanisms to counteract or even exploit GCN2 activation. For example: a. The Herpes Simplex Virus 1 (HSV-1) glycoprotein H (gH) sequesters GCN1, thereby preventing endogenous GCN1-GCN2 interactions and GCN2 activation. b. Integrase (IN) from the Retroviridae family of viruses binds to the pseudokinase domain (YPK). The significance of this interaction with the YPK is not yet fully understood. Potentially, IN-YPK binding could impede the interaction between the YPK and Protein Kinase (PK)-domain, which is essential for facilitating the stimulation of the GCN2 catalytic activity. However, this remains to be experimentally tested. c. Nuclear polyhedrosis virus Protein Kinase-like 2 (PK2) interacts with the GCN2 PK-domain in a way that hinders activation of the PK-domain. The N- and C-lobe of the PK-domain are depicted (green and yellow) to illustrate the binding of the PK2 N-terminal extension (pink line) to the N-lobe. The PK2 C-lobe-mimic domain (pink square) takes the place of the C-lobe, resulting in a hybrid PK-domain consisting of the eIF2a-kinase N-lobe and the catalytically-inactive PK2 C-lobe-mimic domain. d. Vaccinia Virus (VV) protein K3L acts as a pseudosubstrate that competes with eIF2a for GCN2 binding, thereby hampering eIF2a phosphorylation. e. Viruses such as Hepatitis C Virus (HCV) capitalise on the eIF2a-P-mediated reduction in global translation to enhance Internal Ribosome Entry Site (IRES)-mediated translation of viral RNA. f. Viruses such as Ebola Virus (EBOV) exploit elevated eIF2a-P levels to promote upstream Open Reading Frame (uORF)-dependent translation of a viral gene, while in Sindbis Virus (SV) non-AUG mediated translation is enhanced. g. Viruses such as Human Immunodeficiency Virus 1 (HIV-1) can take advantage of GCN2-mediated ISR activation by exploiting the concomitant increase in ATF4 levels for enhancing transcription of its own genetic material, the Long Terminal Repeats (LTRs). h. Non-structural protein S-segment (NSs) from the Phleboviruses Rift Valley Fever Virus (RVFV) selectively hinders the binding of eIF2B to phosphorylated eIF2a, but not to unphosphorylated eIF2a, thereby allowing unimpeded GDP-GTP exchange on eIF2a. i. Early protein 6 (E6) from Human Papilloma Virus (HPV) type 18 binds to the host proteins PP1c and GADD34 to enhance eIF2a-P dephosphorylation. j. Protein ICP34.5 from HIV-1 recruits the host’s phosphatase PP1c to promote dephosphorylation of eIF2a-P, thereby counteracting GCN2-mediated eIF2a phosphorylation. k. The HIV protease (HIVPro) removes a GCN2 N-terminal portion that contains the binding site for GCN1 which is essential for mediating GCN2 stimulation. Note that viruses other than those mentioned in this figure also utilise one or more of these mechanisms. For more see text. V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 into the mechanism of K3L mediated inhibition of PKR.92 This system uses a yeast strain lacking its only eIF2a-kinase, Gcn2, and instead human PKR is expressed from a galactose-inducible promotor. In yeast, PKR is hyperactive, leading to eIF2a hyper-phosphorylation and a concomitant severe reduction in translation.93 As a result, the insuffi- 13 cient production of proteins leads to a severe reduc- tion in growth rate. Thus, the growth rate in this system inversely correlates with the level of PKR activity. Using this system, PKR expression was shown to dampen yeast growth, whilst additional expression of K3L reverts this growth defect.92 This correlates with high eIF2a-P levels in strains Table 2 Viral proteins with the potential to inhibit or exploit GCN2 or the GCN2 pathway. Overview of viral mechanisms known to counteract / exploit GCN2 or the GCN2 pathway. The content of the table is primarily sorted by the mechanism of GCN2 activation, followed by the genetic material used by the virus to generate mRNA (Baltimore classification)167, and the viral family. For more information on the extent to which GCN2 has been shown to be involved thus far, please refer to the text. Proposed mechanism: for counteracting GCN2 See section in review Baltimore class Viral molecule Viral family Virus* (abbreviation) Host Disease / symptoms Source ** Sequesters GCN1 (Figure 4a) IVc dsDNA gH Herpesviridae Herpes simplex virus 1 (HSV-1) Humans Causes cold sores in and around the mouth, fever, and swollen lymph nodes. b Reduces levels of functional GCN2 Cleaves GCN2 (Figure 4k) IVd + ssRNA- RT HIVPro Retroviridae Human immunodeficiency virus 1 (HIV-1) Humans Can cause severe diseases such as AIDS, secondary infections, and lymphomas. Acute phase symptoms include fever, headache, rash, sore throat, weight loss, and diarrhoea b Human immunodeficiency virus 2 (HIV-2) Humans (more prevalent in West Africa) Similar to HIV-1 but less virulent, i.e. slower disease progression. 172 Enhances GCN2 degradation, mechanism unknown IVd + ssRNA Unknown Coronaviridae Severe acute respiratory syndrome coronavirus (SARS- CoV) Humans. Occasionally also dogs, cats, mink, gorilla, bats, pangolin Causes severe acute respiratory syndrome (SARS), presents in humans as respiratory and flu-like symptoms b, c Hampers GCN2 catalytic activity Forms a non- functional kinase domain (Figure 4c) IVb dsDNA PK2 Baculoviridae Autographa californica multiple Nucleopolyhedrovirus (AcMNPV) Winged insects Interferes with insect development d Acts as pseudosubstrate (Figure 4d) IVa dsDNA vIF2a Iridoviridae Ranavirus Amphibians, fish Affects kidneys, spleen, lungs, and other tissues depending on the species. Great concern for wildlife conservation and aquaculture industry due to mass mortality events e dsDNA K3L and its orthologues*** Poxviridae Vaccinia virus (VV) Humans, cattle Symptoms are usually very mild to non-existent in humans, but may produce localised skin lesions. i Camelpox virus (CMLV) Camels Fever and local or generalized pox lesions on the skin, mouth and respiratory tracts e Myxoma virus (MYXV) Rabbits, hares Causes Myxomatosis which is lethal in European rabbits. Symptoms include skin lesions, respiratory issues, and death. e, l Swinepox virus (SPV) Pigs Causes swinepox, symptoms include mild fever, inappetence, and dullness. f V .J . G ib b s , Y .H . L in , A .A . G h u g e , e t a l. J o u rn a l o f M o le c u la r B io lo g y 4 3 6 (2 0 2 4 ) 1 6 8 5 9 4 1 4 Table 2 (continued) Proposed mechanism: for counteracting GCN2 See section in review Baltimore class Viral molecule Viral family Virus* (abbreviation) Host Disease / symptoms Source ** Variola virus (VARV) Humans Causes smallpox. Symptoms include lesions in the mucous membranes of the nose and mouth as well as face and extremities, fever, fatigue, severe back pain, abdominal pain, and vomiting. h + ssRNA- RT Integrase Retroviridae Avian sarcoma virus (ASV) Chickens and other birds Inappetence, weakness, diarrhoea, dehydration and emaciation. Can lead to tumour formation and decreased fertility. f Human immunodeficiency virus 1 (HIV-1) Humans Can cause severe diseases such as AIDS, secondary infections, and lymphomas. Acute phase symptoms include fever, headache, rash, sore throat, weight loss, and diarrhoea b Human immunodeficiency virus 2 (HIV-2) Humans (more prevalent in West Africa) Similar to HIV-1 but less virulent, i.e. slower disease progression. 172 Murine leukaemia virus (MLV) Mice, rats Cancer in mice, may induce neurodegenerative disorders and paralytic diseases. 171 + ssRNA E2 Flaviviridae Hepatitis C virus (HCV) Humans Causes hepatitis, hepatocellular carcinoma, and cirrhosis. h Counteracts eIF2a-P Attenuates eIF2a- P-mediated eIF2B inhibition. (Figure 4h) IVf + ssRNA AcP10 Coronaviridae Beluga whale Coronavirus (Bw- CoV) Beluga whale May cause pulmonary disease and liver disease. 173 + ssRNA AiVL Picornaviridae Aichi virus (AiV) Humans Causes gastroenteritis h - ssRNA NSs Phenuiviridae Rift valley fever virus (RVFV) Humans, ruminants Humans: ranges from mild flu-like illness to severe haemorrhagic fever Livestock: fever, listlessness, anorexia, abortion, high mortality rates in neonates. b, e Sandfly fever Sicilian virus (SFSV) Humans, possibly other animals Fever, headache, photophobia, malaise, myalgia, and retro-orbital pain. 174 Promotes PP1c- mediated eIF2a-P dephosphorylation, various mechanisms (Figure 4i-j) IVe dsDNA IE180 Herpesviridae Pseudorabies virus (PRV) Pigs Affects the central nervous, respiratory, and reproductive systems. f dsDNA E6 Papillomaviridae Human papilloma virus (HPV) Humans Can cause cervical, vulval, vaginal, and penial cancers and warts h + ssRNA Unknown Coronaviridae Infectious bronchitis virus (IBV) Chickens, peafowl, non- galliform birds Decreased vitality and appetite, nasal discharge, sneezing, coughing, gasping, declined egg production. f dsDNA ICP34.5, its orthologues, and similar proteins*** Ascoviridae Trichoplusia ni Ascovirus 2c (TNAV2c) Cabbage looper Trichoplusia ni (Hübner) Opaque yellow-white discoloration on the larval body, incomplete shedding of the molted cuticle, decreased feeding activity, and slower growth rate alongside prolonged larval lifespan. Can be lethal. 179 (continued on next page) V .J . G ib b s , Y .H . L in , A .A . G h u g e , e t a l. J o u rn a l o f M o le c u la r B io lo g y 4 3 6 (2 0 2 4 ) 1 6 8 5 9 4 1 5 Table 2 (continued) Proposed mechanism: for counteracting GCN2 See section in review Baltimore class Viral molecule Viral family Virus* (abbreviation) Host Disease / symptoms Source ** Loopers pose a threat to crops such as cabbage, broccoli, and cauliflower. Asfarviridae African swine fever virus (ASFV) Domestic pigs, wild boar High fever, haemorrhages in the reticuloendothelial system, high mortality rate. e Herpesviridae Herpes simplex virus 1 (HSV-1) Humans Causes cold sores in and around the mouth, fever, and swollen lymph nodes. b Herpes simplex virus 2 (HSV-2) Humans Causes genital herpes, symptoms include bumps, blisters, and ulcers around the genitals or anus. May be accompanied by fever, headache and swollen lymph nodes. b Macropodid herpesvirus (MaHV) Kangaroos, wallabies Rhinitis, conjunctivitis, pneumonia, cloacal ulceration, splenic, pulmonic, and hepatic necrosis. j Hytrosaviridae Glossina pallidipes salivary gland hypertrophy virus (GpSGHV) Tsetse fly Glossina pallidipe Hampers tsetse fly fertility and production of offspring Tsetse flies are the vector of the pathogenic African trypanosomes that cause human and animal sleeping sickness in sub-Saharan Africa 180 Iridoviridae Anopheles minimus Iridovirus (AMIV) Mosquito Anopheles minimus Causes cytopathic damage, leading to the reduction of body size, fecundity and longevity Mosquito is a major Southeast Asian malaria vector. Can pose a threat to humans as infected mosquitoes are a vector for malaria. 177,178 Poxviridae Canarypox virus (CNPV) Birds High mortality rate, has three forms: Cutaneous form: wart-like lesions typical on the face, beak and legs Diphtheritic form: lesions on the mucosa of the oral cavity and respiratory tract Septicaemic form: internal lesions affecting the respiratory and GI tracts f Amsacta moorei Entomopoxvirus “L” (AmEPV) Some Lepidoptera such as the salt marsh moth Estigmene acrea Colour changes of body, decreased feeding, very characteristic is the extremely extended longevity of infected insects. Salt marsh moth caterpillars pose a threat to agriculture in North America, since it feeds on e.g. cabbage, cotton, walnuts, apple, tobacco, pea, potato, clovers, and maize 175,176 Reduces eIF2a-P levels, mechanism unknown IVe dsDNA IE63 Herpesviridae Varicella zoster virus (VZV) Humans Causes varicella (chickenpox) and shingles, which presents as lesions concentrated on the chest and back. h IVa + ssRNA Unknown, potentially NS2A / NS4A Flaviviridae Dengue virus 2 (DENV-2) Humans Causes Dengue fever and Dengue haemorrhagic fever, symptoms can include fever, headache, retroocular pain, myalgia, arthralgia, exanthema, and prostration with or without haemorrhage. Can also be asymptomatic. h V .J . G ib b s , Y .H . L in , A .A . G h u g e , e t a l. J o u rn a l o f M o le c u la r B io lo g y 4 3 6 (2 0 2 4 ) 1 6 8 5 9 4 1 6 Table 2 (continued) Proposed mechanism: for exploiting GCN2 See section in review Baltimore class Viral molecule Viral family Virus* (abbreviation) Host Disease / symptoms Source ** Exploitation of the GCN2 / ATF4 axis Exploits the GCN2 / ATF4 axis to promote translation of viral EBNA1 protein, mechanism to be elucidated IVg dsDNA vDUB- containing proteins*** Herpesviridae Epstein-Barr virus (EBV) Humans Causes mononucleosis, also known as glandular fever. Symptoms include tiredness, fever, sore throat, headaches and body aches, swollen lymph nodes, swelling in the liver and/or spleen and rash. Can also be asymptomatic. b, f, i, h Human Cytomegalovirus (HCMV) Humans Can occasionally develop into mononucleosis or hepatitis. Possible symptoms include fever, sore throat, fatigue, swollen glands. Can also be asymptomatic. f, h Kaposi’s sarcoma- associated Herpesvirus (KSHV) Humans In patients with weakened immune systems, can lead to Kaposi sarcoma, primary effusion lymphoma, and plasma cell variant of multicentric Castleman disease. Can also be asymptomatic. h, k Exploits GCN2 / ATF4 axis to activate the transcription of long terminal repeats (LTR) (Figure 4g) IVg + ssRNA- RT Transactivator proteins*** Retroviridae Human immunodeficiency virus 1 (HIV-1) Humans Can cause severe diseases such as AIDS, secondary infections, and lymphomas. Acute phase symptoms include fever, headache, rash, sore throat weight loss, diarrhoea. b Human T cell leukaemia virus 1 (HTLV-1) Humans Can be asymptomatic or induce adult T-cell leukaemia and associated myelopathy / tropical spastic paraparesis, neurological diseases. b + ssRNA- RT Unknown Retroviridae Simian immunodeficiency virus (SIV) Non-human primates Mostly non-pathogenic but may develop AIDS-like symptoms if chronically infected. e Exploits eIF2a-P to promote uORF- dependent translation of viral genes (Figure 4f) IVg -ssRNA Unknown Filoviridae Ebola virus (EBOV) Humans, great apes, monkeys, antelopes, bats Causes Ebola, symptoms in humans include fever, weakness, muscle pain, impaired liver and kidney function, internal and external bleeding, death. c Exploits eIF2a-P to mediate enhanced non-AUG mediated translation (Figure 4f) IVg + ssRNA Unknown Togaviridae Sindbis virus (SV) Humans, birds Exanthema over trunk and limbs, joint symptoms. Sometimes nausea, general malaise, headache, and muscle pain. Can be asymptomatic. g Exploits eIF2a-P to promote IRES- dependent translation of viral genes (Figure 4e) IVg + ssRNA Unknown Flaviviridae Hepatitis C virus (HCV) Humans Causes hepatitis, hepatocellular carcinoma, and cirrhosis. h * Note that there may be additional viruses beyond those listed above that share the same mechanism, for more see text. ** For source of information, see references (publications) or below (internet links). V .J . G ib b s , Y .H . L in , A .A . G h u g e , e t a l. J o u rn a l o f M o le c u la r B io lo g y 4 3 6 (2 0 2 4 ) 1 6 8 5 9 4 1 7 *** These viral molecules have been classified by group. For the individual names of the molecules in each species listed, please refer to text. Internet sources: a. Charles River website: https://www.criver.com/products-services/research-models-services/research-animal-diagnostics/infectious-agent-technical-info/murine-norovirus-mnv?region = 3701. b. World Health Organization (WHO) mainly addresses human viral disease symptoms: https://www.who.int/. c. ViralZone has information on genome, host range and short notes on diseases: https://viralzone.expasy.org/. d. Autographa californica: https://www.sciencedirect.com/topics/biochemistry-genetics-and-molecular-biology/autographa-californica. e. World Organization for Animal Health (WOAH): https://www.woah.org/en/what-we-do/animal-health-and-welfare/disease-data-collection/. f. MSD manual for diseases in animals: https://www.msdvetmanual.com/generalized-conditions/vesicular-stomatitis-in-large-animals/vesicular-stomatitis-in-large-animals/?autoredirectid = 22922. g. European Centre for Disease Prevention and Control (ECDC), details on disease symptoms: https://www.ecdc.europa.eu/en. h. Centers for Disease Control and Prevention (CDC), details on disease symptoms: https://www.cdc.gov/. i. Medscape, site commonly used by medical practitioners to learn about infections and diseases: https://emedicine.medscape.com/. j. Prevalence and Clinical Significance of Herpesvirus Infection in Populations of Australian Marsupials: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4519311/. k. Mayo Clinic: https://www.mayoclinic.org/ . l. WOAH Terrestrial Manual 2021: https://www.woah.org/fileadmin/Home/fr/Health_standards/tahm/3.07.01_MYXO.pdf. V .J . G ib b s , Y .H . L in , A .A . G h u g e , e t a l. J o u rn a l o f M o le c u la r B io lo g y 4 3 6 (2 0 2 4 ) 1 6 8 5 9 4 1 8 https://www.criver.com/products-services/research-models-services/research-animal-diagnostics/infectious-agent-technical-info/murine-norovirus-mnv?region https://www.who.int/ https://viralzone.expasy.org/ https://www.sciencedirect.com/topics/biochemistry-genetics-and-molecular-biology/autographa-californica https://www.woah.org/en/what-we-do/animal-health-and-welfare/disease-data-collection/ https://www.msdvetmanual.com/generalized-conditions/vesicular-stomatitis-in-large-animals/vesicular-stomatitis-in-large-animals/?autoredirectid https://www.ecdc.europa.eu/en https://www.cdc.gov/ https://emedicine.medscape.com/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4519311/ https://www.mayoclinic.org/ https://www.woah.org/fileadmin/Home/fr/Health_standards/tahm/3.07.01_MYXO.pdf V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 expressing PKR, and reduced eIF2a-P levels in strains expressing both PKR and K3L. This demon- strates that K3L can effectively inhibit PKR activity in vivo. Using this heterologous yeast assay sys- tem, a single amino acid substitution, H47R, was found to render K3L even more potent in reverting the growth defect, which correlates with a further reduction in eIF2a-P levels.92 This suggests that this K3L variant is more potent in PKR inhibition, likely because the eIF2a mimicry has been improved. Supporting this idea, the H47R substitu- tion increases the homology of K3L to eIF2a close to the Ser-51 phosphorylation site. Conversely, decreasing the eIF2amimicry of K3L should reduce or abolish K3L’s potency in inhibiting PKR. In fact, single amino acid substitutions in the KGYID motif conserved between K3L and eIF2a proteins, such as Y76A, fail to revert the growth defect associated with PKR expression, and this correlates with the inability to reduce eIF2a-P levels. Together, these studies support the idea that K3L mimics the sub- strate eIF2a. Consistent with the idea that K3L is a bona fide pseudosubstrate, pulldown assays revealed that K3L binds to PKR. The H47R substi- tution enhances this interaction, while the Y76A substitution has the opposite effect. Similarly to PKR, constitutively active yeast Gcn2 also leads to a growth defect in yeast, and K3L reverts this growth defect.92 The fact that K3L inhibits Gcn2 in in vitro kinase assays, binds to the Gcn2 kinase domain, and reduces eIF2a-P levels in yeast cells,88 supports the idea that K3L inhibits GCN2 via the same mechanism as found for PKR (Figure 4d). K3L orthologs have also been found in viruses belonging to the poxvirus family, such as C8L from swinepox virus,94 C3L from Variola Virus (VARV, causes smallpox) and CMLV032 from camel pox virus.95 These orthologues contain the KGYID motif as found for K3L, and studies suggest that these proteins exert similar inhibitory effects on eIF2a-kinases as K3L.94,95 The PKR PK-domain shows substantial sequence variability across species.96 This reduces the ability of viral K3L-type inhibitors to bind to PKR from species outside their natural hosts.96–98 Because of this, host organisms are likely to have a degree of innate resistance to poxvirus strains from other species, reducing the risk of severe zoo- notic disease. Curiously, this sequence variability in the PK-domain is not seen to that extent in GCN2,96 which would suggest that GCN2 may be more sus- ceptible to these pseudosubstrates irrespective of the natural host. Curiously, not all poxviruses contain K3L orthologues with a KGYID motif. One example is the Myxoma Virus (MYXV) which causes myxomatosis (a lethal disease in rabbits). This virus was released in Australia in the 1950s in an attempt to control the rabbit population.99 Its K3L orthologue, called M156R, has been shown to be 19 a competitive inhibitor of PKR.100 Curiously, M156R is also an efficient substrate of PKR in vitro even though it does not contain a Ser at a position equivalent to eIF2a Ser-51.100 It is not known yet whether M156R phosphorylation has any biological relevance in terms of M156 function. M156R contains the motif YVD instead of KGYID, while still being a structural mimic of eIF2a,100 sug- gesting that not every residue within the KGYID motif is critical for PK-domain docking. In fact, struc- tural analysis of PKR suggests that the Tyr and Asp in YVD are critical determinants for binding to the PK-domain.100 This is consistent with a prior discov- ery, where an Ala-substitution of Lys, Tyr, or Asp within the KGYIDmotif hampers K3L’s ability to inhi- bit PKR, and the Ala-substitution of Tyr also impairs K3L-PKR interaction.84 Additionally, the naturally occurring M156R-L98P mutation, which is proximal to the YVD motif, leads to a loss of function M156R protein unable to inhibit rabbit PKR.101 This substi- tution for Pro likely disrupts the structural features of the YVD motif necessary for kinase docking, highlighting the importance of the amino acids sur- rounding the KGYID motif to adopt the correct con- formation for PKR binding. The pox-like Ranaviruses are pathogens of lower vertebrates such as fish, amphibia and reptiles. Their human-induced spread and emergence has become a great concern for wildlife conservation and aquaculture industry, given that they can cause mass mortality events.102 Ranaviruses pro- duce a pseudosubstrate, the viral mimic of eIF2a (vIF2a), which acts in a similar fashion to K3L.103 As found for K3L, vIF2a has high sequence homol- ogy to the N-terminus of eIF2a, but in the KGYID motif the amino acid Ile is not absolutely conserved. This would suggest that the Ile in this motif is not critical for vIF2a function in inhibiting PKR,103,104 in agreement with the findings in M156R and K3L where not Ile, but the two neighbouring amino acids are critical for function.92,100 As has been mentioned previously, HIV-1 IN is a substrate of GCN2 (see IIId). In in vitro assays, GCN2 can phosphorylate IN produced by various Orthoretrovirinae including HIV-1, HIV-2, MLV, and ASV,82 supporting the idea that IN is a GCN2 substrate. Interestingly, IN contains an SGYIE motif, reminiscent of the KGYID motif in eIF2a, as well as a Ser residue downstream at a position that approximately corresponds to that of eIF2a Ser- 51.13,82 This may suggest that IN functions as a pseudosubstrate. Supporting this idea, in vitro kinase assays show that phosphorylation of IN decreases with increasing amounts of eIF2a added.82 Curiously however, IN phosphorylation occurs at Ser-24 and Ser-255 in vitro and in cellulo but not at Ser-57, which would be the equivalent to eIF2a Ser-51.82 It would be interesting to repeat the in vitro kinase assay with constant eIF2a levels but increasing amounts of IN in order to test whether IN can hamper eIF2a phosphorylation, which would V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 further support the idea that IN acts as a true pseudosubstrate. The Hepatitis C Virus (HCV) Envelope protein 2 (E2) enables the virus to evade PKR-mediated anti-viral activities.105 Unlike the previously men- tioned K3L orthologs, E2 lacks the KGYID motif, at least according to the HCV genome polyprotein sequence (Accession number Q9WMX2, Gibbs & Sattlegger unpublished). Instead, it was found that in most HCV isolates, E2 shows sequence homol- ogy to portions of both eIF2a and PKR. The homol- ogy to eIF2a includes Ser-51 (Ser-280 in E2) and residues located immediately upstream to eIF2a Ser-51 (amino acids 277–279 in E2 and 48–50 in eIF2a), while the homology to PKR includes again the Ser-280 residue in E2 plus downstream amino acid sequences (amino acids 280–287 in E2 and 83–90 in PKR) (Figure 5).105 The respective PKR segment is located in the RNA-binding domain of PKR, and within a region found to harbour an auto-phosphorylation cluster (PKR amino acids 81–107, contains six auto-phosphorylation sites).105,106 It was suggested that this PKR-eIF2a Phosphorylation Homology Domain (PePHD) is important for PKR binding and inhibition, given that deletion of a portion of E2 that also removes the PePHD significantly reduces E2-PKR interaction in in vitro co-precipitation assays.105 Furthermore, in vitro kinase assays have demonstrated that E2 inhibits PKR auto-phosphorylation. Similarly, PKR auto-phosphorylation is hampered by E2 in vivo, but not by an E2 variant lacking a specific region that includes PePHD.105 Repeating these assays with E2 only lacking the PePHD would provide definitive evidence that this motif is critical for PKR binding and inhibition. Co-precipitation assays suggest that E2 also binds PERK, and it has been shown that this interaction is mediated by the PK-domain.107 Simi- Figure 5. Viral proteins share sequence identity with protein 2 (E2) harbours the PKR-eIF2a Phosphorylation Hom with both, eIF2a and PKR. The homology with eIF2a covers 280 in E2), and the immediate N-terminal three residues (pink homology with PKR includes again the Ser-280 residu encompasses the immediate downstream residues located i 287 in E2, residues 84–90 in PKR). This region is known to six auto-phosphorylation sites (PKR amino acids 81–107). A the PePHD sequence in E2 against all Dengue Virus (DENV) sequence, accession NC_001474.2). Regions in the DENV were found that show some amino acids conservation to tho numbers of the respective DENV proteins are indicated in t 20 larly to PKR, E2 inhibits PERK in vitro, and in vivo it prevents the translational repression associated with PERK overexpression.107 As with PKR, E2 is less effective in counteracting PERK-mediated translational repression when a portion of E2 is missing, which also includes the removal of PePHD. Similar observations were made with an E2 variant that contains amino acid substitutions in the PePHD, reducing the homology to eIF2a (SELS substituted by GQQH).107 This supports the idea that it is in fact the PePHD that is responsi- ble for counteracting PERK. It will be interesting to investigate whether this substitution also hampers E20s ability to bind PERK and hamper PERK activa- tion at a molecular level. Nevertheless, given that E2 can inhibit PERK as well as PKR, and that E2 binds to the PK-domain, it is likely that E2 can inhibit all eIF2a-kinases includingGCN2. Additional exper- iments are needed to verify whether E2 acts as a true pseudosubstrate, or alternatively inhibits eIF2a-kinase activity by a different mechanism. The Flaviviridae Dengue Virus 2 (DENV-2) has also been found to inhibit eIF2a phosphorylation.108 While DENV-2 infection of 2fTGH (a human fibrosarcoma cell line) prevents PERK-mediated eIF2a phosphorylation in response to ER stress, it can also hamper GCN2 auto- phosphorylation and eIF2a phosphorylation in response to Leu starvation.108 This suggests that DENV-2 contains a molecule capable of inhibiting GCN2, which is intriguing considering GCN20s involvement in immune signalling during DENV-2 infection.109 It was suggested that DENV-2 may produce an inhibitor akin to that in HCV, given their phylogenetic link.108 If that is true, then the DENV genome should encode a protein equivalent to E2 which is identifiable by its PePHD domain. A protein Basic Local Alignment Search Tool (BLAST) search query against all DENV protein sequences was con- eIF2a and PKR. The Hepatitis C Virus (HCV) Envelope ology Domain (PePHD) that shares sequence homology the eIF2a phosphorylation site Ser-51 (burgundy, Ser- , residues 48–50 in eIF2a, residues 277–279 in E2). The e in E2 (burgundy, Ser-83 in PKR). Additionally, it n the RNA-binding domain of PKR (cyan, residues 281– harbour an auto-phosphorylation cluster which contains n NCBI protein BLAST search was conducted querying proteins (sequences derived from the DENV polyprotein Non-structural proteins 2A and 4A (NS2A and NS4A) se encompassing PePHD, as indicated. The accession he figure. V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 ducted, and the Non-structural proteins 2A and 4A (NS2A and NS4A) were found to contain sequence similarity to the PePHD of E2 (Figure 5; Gibbs & Sattlegger, unpublished). The NS2A protein har- bours two nearby sequences resembling the PKR- like part of the PePHD but lacks the part with resem- blance to eIF2a. The NS4A protein contains a region with similarity to both the eIF2a-like and PKR-like regions of PePHD, but with two notable differences. Firstly, NS4A lacks the Ser-51 equiva- lent. Secondly, there is a small insertion in the PKR-like section of this sequence. Hence, judging from this sequence analysis, it is not apparent that DENV antagonises GCN2 in a similar fashion as E2 from HCV. This underscores the need for further experiments to unveil how DENV inhibits eIF2a phosphorylation. One function of the Transactivator of transcription (Tat) protein of HIV-1 is likely that of a pseudosubstrate.110,111 PKR has been shown to bind Tat in vitro and in vivo, and activated PKR was found to phosphorylate Tat.110,111 In vitro, increasing amounts of Tat clearly leads to reduced levels of PKR-mediated eIF2a phosphorylation in a dose-responsive manner.111 When doing the reverse experiment using increasing amounts of eIF2a, a dose-dependent decrease in PKR- mediated Tat phosphorylation was not as pro- nounced.111 Nevertheless, these findings support the idea that Tat competes with eIF2a for PKR bind- ing, as one would expect from a pseudosubstrate. Further supporting this idea, the region in Tat required for PKR binding (amino acids 40–58) shows homology to the area in eIF2a surrounding Ser-51.112 Given that PKR-Tat interaction requires a portion in Tat that encompasses the RNA- binding region,111 and that PKR binds dsRNA, one cannot exclude the possibility that the PKR-Tat interaction is mediated by RNA. The authors reported (as data not shown) that treatment of either PKR or Tat with RNase prior to interaction assays did not affect the interaction.111 These findings would agree with the idea that PKR-Tat interaction is not bridged by RNA. However, one cannot exclude the possibility that after an RNase digest, some RNA has remained that can potentially bridge a protein–protein interaction. Conclusive evidence could be provided by using Tat with point mutations that specifically abolish RNA binding. Since Tat acts as a pseudosubstrate, it is reasonable to assume that Tat may be able to inhibit eIF2a-kinases other than PKR, such as GCN2. However, this needs to be tested experimentally. IVb) Mimicry of the protein kinase C-lobe A viral protein has been reported that mimics a portion of the PK-domain of eIF2a-kinases and can likely inhibit any eIF2a-kinase including GCN2. Best studied is the imitation Protein Kinase 2 (PK2) from baculovirus Autographa californica Multiple Nucleopolyhedrovirus (AcMNPV), which 21 is highly pathogenic to a variety of larval lepidopteran insects (insects with scaly wings such as butterflies and moths). PK2 contains an N-terminal extension that is unique to PK2, but otherwise bears closest homology to the C-lobe of the highly conserved eIF2a-kinase PK-domain itself.113,114 The so-called C-lobe-mimic domain shares homology to C-lobe subdomains VI – XI, just that several of these subdomains vary from the con- sensus sequence typical for Ser/Thr kinases.23,113– 115 Together with the fact that the phosphorylation loop and potential auto-phosphorylation sites are lacking,116 this strongly suggests that the C-lobe- mimic domain is catalytically inactive. Heterologous yeast assays have revealed that PK2 functions as an eIF2a-kinase inhibitor. In this assay, the growth defect of yeast caused by the hyperactivity of overexpressed PKR is alleviated when PK2 is co-expressed.116 This is consistent with PK2 also reversing the hyperphosphorylation of eIF2a associated with hyperactive PKR. Similar findings were obtained for yeast cells expressing hyperactive yeast Gcn2, suggesting that PK2 can also inhibit Gcn2. In agreement with the idea that PK2 inhibits eIF2a phosphorylation, infected insect SF9 cells (a clonal isolate of Spodoptera frugiperda Sf21 cells) show reduced eIF2a-P levels as com- pared to uninfected cells, while this is not found in SF9 cells infected with a PK2-deficient virus.116 The C-lobe-mimic domain contains an aG helix that corresponds to the one in the C-lobe of the eIF2a-kinases’ PK-domain known to mediate eIF2a binding.113,114,25 Surprisingly though, PK2 does not bind to eIF2a. Instead, it binds to PKR, as determined for example by in vitro pulldown assays using purified proteins, or by in vivo co- immunoprecipitation assays using yeast expressing PK2 as well as PKR.115,116 The PKR PK-domain N-lobe, but not the C-lobe, directly interacts with PK2, as determined via in vitro pulldown and yeast 2-hybrid assays.115,116 Curiously, the 22 amino acid long N-terminal exten- sion of PK2 is necessary, but not sufficient, for bind- ing to the PK-domain of PKR in vitro, as determined by pulldown assays. This suggests that the PK2 C- lobe-mimic domain has a supporting role in PKR binding. Within the N-terminal extension, a single F18A substitution abolishes the interaction between PK2 and the PKR kinase domain. This correlates with the inability to inhibit PKR, underscoring the importance of the PK2 N-terminal extension for PKR inhibition.115 Together with results from addi- tional studies, this led to the so-called lobe- swapping model115 (Figure 4c), in which the PK2 N-terminal extension mediates high-affinity binding to the eIF2a-kinase N-lobe. The PK2 C-lobe- mimic domain may displace and take on the place of the eIF2a-kinase domain C-lobe. This would result in a hybrid protein kinase domain consisting of the eIF2a-kinase N-lobe and the catalytically inactive PK2 C-lobe-mimic domain, rendering the V.J. Gibbs, Y.H. Lin, A.A. Ghuge, et al. Journal of Molecular Biology 436 (2024) 168594 kinase unable to bind ATP and phosphorylate eIF2a. It has been shown that PK2 also hampers HRI, suggesting that PK2 can inhibit any eIF2a- kinase.115 PK2-devoid baculovirus is less able to replicate in Bombyx mori insect cells, and this correlates with higher eIF2a-P levels as compared to cells infected with wildtype baculovirus. Furthermore, PK2-devoid baculovirus takes 12 hrs longer to kill host larval insects. Together, these findings agree with the idea that PK2 is critical for the virus to counteract the host’s viral defence.115 PK2-devoid baculovirus regains the ability for budded virus pro- duction in cells where HRI is knocked down, but not in cells where GCN2 or PERK is knocked down, suggesting that HRI is the major eIF2a-kinase that counteracts AcMNPV infection.115 In agreement with this idea, PK2 shares the highest sequence homology with HRI, in particular insect HRI.115 It was suggested that PK2 evolved from an insect HRI-like kinase. It is possible that within the HRI PK-domain there are residues specific to HRI but not to other eIF2a-kinases. These residues, while not critical for the kinase’s enzymatic activity per se, may aid and enhance the interaction with PK2, thereby rendering PK2 more potent in inhibiting the eIF2a-kinase HRI. While these findings suggest that GCN2 may not be the primary target of PK2, one could imagine that the inhibitory effect of PK2 on other eIF2a-kinases is part of a welcome side effect in hampering host defence. IVc) Sequestering GCN1, a protein required for GCN2 activation The Herpes Simplex Virus 1 (HSV-1) expresses a protein called glycoprotein H (gH) which has been shown to hamper the host’s ISR response.117 Specifically, in Vero kidney cell lines, overexpres- sion of gH is sufficient to reduce eIF2a-P levels, demonstrating that gH can hamper eIF2a phospho- rylation in the absence of any other viral proteins or molecules.117 MEFs infected with HSV-1 show increased eIF2a-P levels, while this is hampered in MEFs knocked-down for GCN1. This suggests that GCN1 is critical for the defence against HSV- 1. Accordingly, given that GCN1 is the effector pro- tein for GCN2 and no other eIF2a-kinase, one would conclude that GCN2 phosphorylates eIF2a in response to HSV-1 infection. Interestingly, GCN1 co-immunoprecipates with transiently expressed gH in tandem affinity purification of 293 T cell extracts.117 Similarly, gH expressed in 293 T or Human Epithelioma 2 (HEp-2) cells co- immuniprecipitates GCN1, suggesting that GCN1 and gH reside in the same protein complex. According to immunofluorescence microscopy, GCN1 is mainly found in the cytoplasm of mock- infected HEp-2 carcinoma cell lines, while in HSV- 1 infected cells GCN1 predominantly colocalises with gH in the nuclear rim.117 This aligns with the notion that GCN1 and gH interact, and that this 22 interaction relocates GCN1 to a different area of the cell.117 Collectively, these findings suggest that gH binds and sequesters GCN1 away from GCN2, thereby preventing GCN2 activation (Figure 4a). It remains to be investigated whether the gH-GCN1 interaction is direct or mediated by a third molecule, and whether gH directly blocks the GCN1-GCN2 interaction. IVd) Reducing GCN2 protein levels Proteolytic cleavage of an enzyme is a way of permanently quashing its activity. In MT-2 human T-lymphocyte cell lines infected with HIV-1, it was found that GCN2 protein levels are dramatically reduced.75 The decrease in GCN2 levels is miti- gated in the presence of saquinavir, a specific inhi- bitor of the HIV-1 protease (HIV-1Pro), suggesting that GCN2 is subject to proteolytic cleavage. In fact, HIV proteases HIV-1Pro and HIV-2Pro cleave both human and mouse GCN2 in vitro at one specific site.75 The amino acid sequence of the cleavage site is 560YVETVIP566 for human GCN2 and 559- YIETVIP565 for mouse GCN2 respectively, with the cleavage site being between the first and sec- ond position.75 Such amino acid sequences are not shared in the other eIF2a kinases. Proteolysis removes the N-terminal end of GCN2 which encom- passes the binding site for its effector protein GCN1.39 Since the GCN1-GCN2 interaction is essential for GCN2 activation, this suggests that the removal of the GCN1-binding site by HIVPro is a way of hampering GCN2 activation (Figure 4k). Demonstrating that HIV-1-mediated protease cleavage indeed renders GCN2 inactive, incubation of GCN2 with HIV-1Pro prior to an in vitro kinase assay abolishes eIF2a phosphorylation.75 Interestingly, when conducting multiple sequence alignments of GCN2 proteins from a selection of organisms, we found that the Y[I/V]ETVIP cleavage sequence is not conserved throughout all eukaryotes (Figure 6). Within the organisms tested, the sequence of the protease site is almost completely conserved in placental mammals, while the first position of the site is only conserved in placental and pouched mammals. It appears that the conservation of the protease site decreases as a species becomes more distantly related to placental mammals. It remains to be determined whether any of the altered cleavage sequences can still be recognised by an HIV protease. In Drosophila melanogaster (representative of insects), Arabidopsis thaliana (representative of plants), and S. cerevisiae (representative of fungi), no sequence could be found that resembles the protease site (data not shown). Together, these findings suggest the possibility that the Y[I/V]ETVIP protease site has arisen progressively during evolution. This prompts the question of why this site has emerged � and still exists � in placental mammals, given the resulting Figure 6. The HIV protease recognition site is conserved in GCN2 of placental mammals. a. Cladogram depicting the relationships among the indicated species. b. Multiple sequence alignment of GCN2 from different species as indicated. Only the portion of the alignment is shown that encompasses the HIV protease site in mouse and human, and immediate upstream and downstream amino acid sequences. For each sequence the positional number of the last shown residue is indicated. The protease recognition site is indicated below the multiple sequence alignment. The GCN2 sequences and accession numbers, from top to bottom are: West African Lungfish (Protopterus annectens, XP_043929767.1), yellowfin tuna (Thunnus albacares, XP_044230996.1), zebrafish (Danio rerio, XP_017209968.2), African clawed frog (Xenopus laevis, XP_041429235.1), common toad (Bufo bufo, XP_040267430.1), Chicken (red junglefowl; Gallus gallus, XP_040527909.1), kiwi bird (Okarito brown kiwi; Apteryx rowi, XP_025946508.1), wild canary (Serinus canaria, XP_018763042.3), tiger rattlesnake (Crotalus tigris, XP_039198591.1), saltwater crocodile (Crocodylus porosus, XP_019406887.1), duck-billed platypus (Ornithor- hynchus anatinus, XP_028933853.1), short-billed echidna (Tachyglossus aculeatus, XP_038597019.1), koala bear (Phascolarctos cinereus, XP_020834905.1), opossum (grey short-tailed opossum; Monodelphis domestica, XP_007480090.1), Tasmanian devil (Sarcophilus harrisii, XP_031807935.1), human (Homo sapiens, NP_001013725.2), mouse (mouse isoform 1; Mus musculus, NP_038747.2), golden hamster (Mesocricetus auratus, XP_040601966.1), domestic dog (Canis lupus familiaris, XP_0056