Phenotypic Mutation 'atchoum' (pdf version)
Alleleatchoum
Mutation Type splice site (6 bp from exon)
Chromosome2
Coordinate118,231,134 bp (GRCm39)
Base Change T ⇒ C (forward strand)
Gene Eif2ak4
Gene Name eukaryotic translation initiation factor 2 alpha kinase 4
Synonym(s) GCN2
Chromosomal Location 118,219,099-118,305,715 bp (+) (GRCm39)
MGI Phenotype FUNCTION: [Summary is not available for the mouse gene. This summary is for the human ortholog.] This gene encodes a member of a family of kinases that phosphorylate the alpha subunit of eukaryotic translation initiation factor-2 (EIF2), resulting in the downregulaton of protein synthesis. The encoded protein responds to amino acid deprivation by binding uncharged transfer RNAs. It may also be activated by glucose deprivation and viral infection. Mutations in this gene have been found in individuals suffering from autosomal recessive pulmonary venoocclusive-disease-2. [provided by RefSeq, Mar 2014]
PHENOTYPE: Homozygotes for a null allele have altered feeding behavior, synaptic plasticity and dendritic cell function. Homozygotes for another null allele show enhanced muscle loss and morbidity after amino acid deprivation. Homozygotes for an ENU-induced allele show higher susceptibility to viral infection. [provided by MGI curators]
Accession Number

NCBI RefSeq: NM_013719 (isoform 1), NM_001177806 (isoform 2); MGI: 1353427

MappedYes 
Amino Acid Change
Institutional SourceBeutler Lab
Gene Model not available
AlphaFold no structure available at present
SMART Domains Protein: ENSMUSP00000005233
Gene: ENSMUSG00000005102

DomainStartEndE-ValueType
low complexity region 2 13 N/A INTRINSIC
RWD 25 137 3.42e-38 SMART
coiled coil region 146 205 N/A INTRINSIC
Pfam:Pkinase 323 538 4.6e-27 PFAM
Pfam:Pkinase_Tyr 326 535 5.5e-18 PFAM
Pfam:Pkinase 589 663 1.7e-11 PFAM
Pfam:Pkinase_Tyr 589 663 1.2e-5 PFAM
low complexity region 728 738 N/A INTRINSIC
Pfam:Pkinase 781 1000 2.6e-38 PFAM
Pfam:Pkinase_Tyr 786 998 1.8e-18 PFAM
Pfam:tRNA-synt_His 1054 1380 5.7e-18 PFAM
Pfam:HGTP_anticodon2 1392 1647 5.8e-17 PFAM
Predicted Effect probably benign
SMART Domains Protein: ENSMUSP00000099586
Gene: ENSMUSG00000005102

DomainStartEndE-ValueType
coiled coil region 34 93 N/A INTRINSIC
Pfam:Pkinase 211 426 1.6e-22 PFAM
Pfam:Pkinase_Tyr 215 423 6.8e-18 PFAM
Pfam:Pkinase_Tyr 477 551 1.2e-5 PFAM
Pfam:Pkinase 477 552 3.9e-11 PFAM
low complexity region 616 626 N/A INTRINSIC
Pfam:Pkinase 647 888 9.4e-42 PFAM
Pfam:Pkinase_Tyr 672 886 1.4e-19 PFAM
Pfam:tRNA-synt_His 941 1268 4.8e-19 PFAM
Pfam:HGTP_anticodon2 1280 1535 1e-16 PFAM
Predicted Effect probably benign
SMART Domains Protein: ENSMUSP00000106496
Gene: ENSMUSG00000005102

DomainStartEndE-ValueType
coiled coil region 25 84 N/A INTRINSIC
Pfam:Pkinase 202 417 3.8e-22 PFAM
Pfam:Pkinase_Tyr 206 414 1.6e-17 PFAM
Pfam:Pkinase_Tyr 468 542 2.8e-5 PFAM
Pfam:Pkinase 468 543 9.1e-11 PFAM
low complexity region 607 617 N/A INTRINSIC
Pfam:Pkinase 638 879 2.2e-41 PFAM
Pfam:Pkinase_Tyr 663 877 3.3e-19 PFAM
Pfam:tRNA-synt_His 932 1259 1.1e-18 PFAM
Pfam:HGTP_anticodon2 1271 1526 2.2e-16 PFAM
Predicted Effect probably benign
SMART Domains Protein: ENSMUSP00000106498
Gene: ENSMUSG00000005102

DomainStartEndE-ValueType
Pfam:RWD 8 56 6.4e-8 PFAM
coiled coil region 68 127 N/A INTRINSIC
Pfam:Pkinase 245 460 1.1e-22 PFAM
Pfam:Pkinase_Tyr 247 457 4.2e-18 PFAM
Pfam:Pkinase_Tyr 511 585 7.8e-6 PFAM
Pfam:Pkinase 511 586 2.5e-11 PFAM
low complexity region 650 660 N/A INTRINSIC
Pfam:Pkinase 681 922 6.2e-42 PFAM
Pfam:Pkinase_Tyr 706 920 9.3e-20 PFAM
Pfam:tRNA-synt_His 975 1302 3.8e-19 PFAM
Pfam:HGTP_anticodon2 1314 1569 5.4e-17 PFAM
Predicted Effect probably benign
SMART Domains Protein: ENSMUSP00000106501
Gene: ENSMUSG00000005102

DomainStartEndE-ValueType
low complexity region 2 13 N/A INTRINSIC
RWD 25 137 3.42e-38 SMART
coiled coil region 146 205 N/A INTRINSIC
Pfam:Pkinase 323 538 2.6e-23 PFAM
Pfam:Pkinase_Tyr 327 535 1.1e-18 PFAM
Predicted Effect probably benign
Meta Mutation Damage Score Not available question?
Is this an essential gene? Non Essential (E-score: 0.000) question?
Phenotypic Category Autosomal Recessive
Candidate Explorer Status loading ...
Single pedigree
Linkage Analysis Data
Penetrance unknown, <100% 
Alleles Listed at MGI

All alleles(7) : Targeted, knock-out(2) Targeted, other(1) Gene trapped(4)
 

Lab Alleles
AlleleSourceChrCoordTypePredicted EffectPPH Score
IGL00233:Eif2ak4 APN 2 118294536 missense probably damaging 1.00
IGL00806:Eif2ak4 APN 2 118271647 missense probably benign 0.08
IGL01343:Eif2ak4 APN 2 118252570 missense probably benign 0.00
IGL01796:Eif2ak4 APN 2 118276785 missense probably benign 0.10
IGL02263:Eif2ak4 APN 2 118292259 missense probably benign 0.00
IGL02391:Eif2ak4 APN 2 118251272 missense probably benign 0.19
IGL02516:Eif2ak4 APN 2 118266735 missense probably damaging 1.00
IGL02603:Eif2ak4 APN 2 118280807 missense probably damaging 1.00
IGL02731:Eif2ak4 APN 2 118219295 missense probably benign
IGL02928:Eif2ak4 APN 2 118303168 critical splice donor site probably null
IGL02947:Eif2ak4 APN 2 118261514 missense probably benign 0.00
IGL03191:Eif2ak4 APN 2 118252693 missense probably damaging 1.00
IGL03202:Eif2ak4 APN 2 118231101 missense probably damaging 1.00
IGL03235:Eif2ak4 APN 2 118273621 missense probably damaging 1.00
IGL03375:Eif2ak4 APN 2 118252799 missense probably benign 0.08
absurdum UTSW 2 118251291 nonsense probably null
Ad UTSW 2 118266722 missense probably damaging 1.00
reductio UTSW 2 118266639 splice site probably null
PIT4520001:Eif2ak4 UTSW 2 118292808 missense probably damaging 1.00
R0023:Eif2ak4 UTSW 2 118293202 missense probably damaging 1.00
R0358:Eif2ak4 UTSW 2 118294410 splice site probably null
R0482:Eif2ak4 UTSW 2 118292828 missense probably damaging 1.00
R0505:Eif2ak4 UTSW 2 118261517 missense probably benign 0.01
R0523:Eif2ak4 UTSW 2 118272577 critical splice donor site probably null
R0578:Eif2ak4 UTSW 2 118305472 splice site probably benign
R0615:Eif2ak4 UTSW 2 118266666 missense probably damaging 1.00
R1300:Eif2ak4 UTSW 2 118294464 missense possibly damaging 0.79
R1531:Eif2ak4 UTSW 2 118273691 missense probably damaging 1.00
R1777:Eif2ak4 UTSW 2 118261320 missense probably damaging 0.98
R1866:Eif2ak4 UTSW 2 118303142 missense probably damaging 1.00
R1932:Eif2ak4 UTSW 2 118278967 missense probably damaging 1.00
R1977:Eif2ak4 UTSW 2 118292238 nonsense probably null
R2011:Eif2ak4 UTSW 2 118261428 missense probably damaging 1.00
R2046:Eif2ak4 UTSW 2 118281889 splice site probably benign
R2122:Eif2ak4 UTSW 2 118286274 missense probably damaging 1.00
R2125:Eif2ak4 UTSW 2 118252604 missense probably benign 0.02
R2126:Eif2ak4 UTSW 2 118252604 missense probably benign 0.02
R2193:Eif2ak4 UTSW 2 118252747 missense probably benign 0.12
R2259:Eif2ak4 UTSW 2 118286264 missense probably damaging 0.97
R2513:Eif2ak4 UTSW 2 118257064 missense probably damaging 1.00
R3798:Eif2ak4 UTSW 2 118304564 missense probably damaging 1.00
R3898:Eif2ak4 UTSW 2 118261404 missense probably damaging 1.00
R3900:Eif2ak4 UTSW 2 118305510 missense probably damaging 1.00
R4375:Eif2ak4 UTSW 2 118258405 missense probably damaging 1.00
R4423:Eif2ak4 UTSW 2 118269547 missense probably benign 0.01
R4589:Eif2ak4 UTSW 2 118247819 missense probably damaging 1.00
R4734:Eif2ak4 UTSW 2 118252568 missense probably damaging 1.00
R5173:Eif2ak4 UTSW 2 118238841 missense probably damaging 1.00
R5367:Eif2ak4 UTSW 2 118266639 splice site probably null
R5471:Eif2ak4 UTSW 2 118304613 missense probably benign 0.02
R5528:Eif2ak4 UTSW 2 118258419 missense probably damaging 1.00
R5634:Eif2ak4 UTSW 2 118292792 missense probably damaging 1.00
R5726:Eif2ak4 UTSW 2 118273613 missense probably damaging 1.00
R5756:Eif2ak4 UTSW 2 118293221 missense possibly damaging 0.95
R5779:Eif2ak4 UTSW 2 118243444 missense possibly damaging 0.85
R5807:Eif2ak4 UTSW 2 118219332 missense probably benign
R6045:Eif2ak4 UTSW 2 118219296 nonsense probably null
R6187:Eif2ak4 UTSW 2 118287638 missense probably damaging 0.98
R6193:Eif2ak4 UTSW 2 118231081 start gained probably benign
R6468:Eif2ak4 UTSW 2 118266722 missense probably damaging 1.00
R6555:Eif2ak4 UTSW 2 118258350 missense probably damaging 0.96
R6616:Eif2ak4 UTSW 2 118285326 nonsense probably null
R6737:Eif2ak4 UTSW 2 118292749 frame shift probably null
R6956:Eif2ak4 UTSW 2 118252748 missense probably damaging 0.96
R7075:Eif2ak4 UTSW 2 118251291 nonsense probably null
R7109:Eif2ak4 UTSW 2 118235532 missense probably damaging 1.00
R7228:Eif2ak4 UTSW 2 118287638 missense probably damaging 0.98
R7441:Eif2ak4 UTSW 2 118302377 missense probably benign 0.01
R7555:Eif2ak4 UTSW 2 118247764 missense possibly damaging 0.64
R7567:Eif2ak4 UTSW 2 118280795 missense probably benign
R8004:Eif2ak4 UTSW 2 118247775 missense possibly damaging 0.64
R8063:Eif2ak4 UTSW 2 118241382 missense possibly damaging 0.94
R8092:Eif2ak4 UTSW 2 118272513 missense probably damaging 1.00
R8195:Eif2ak4 UTSW 2 118280819 missense possibly damaging 0.50
R8306:Eif2ak4 UTSW 2 118287656 missense possibly damaging 0.68
R8470:Eif2ak4 UTSW 2 118293207 missense probably damaging 0.98
R8671:Eif2ak4 UTSW 2 118252667 missense possibly damaging 0.88
R8693:Eif2ak4 UTSW 2 118262718 missense probably damaging 0.98
R8714:Eif2ak4 UTSW 2 118292765 missense possibly damaging 0.89
R8744:Eif2ak4 UTSW 2 118261474 nonsense probably null
R8813:Eif2ak4 UTSW 2 118278806 missense probably damaging 1.00
R8917:Eif2ak4 UTSW 2 118287617 missense probably damaging 1.00
R8924:Eif2ak4 UTSW 2 118258513 missense probably damaging 1.00
R9177:Eif2ak4 UTSW 2 118271701 critical splice donor site probably null
R9189:Eif2ak4 UTSW 2 118258393 missense probably damaging 1.00
R9231:Eif2ak4 UTSW 2 118271662 missense probably benign 0.00
R9268:Eif2ak4 UTSW 2 118271701 critical splice donor site probably null
R9321:Eif2ak4 UTSW 2 118292798 missense possibly damaging 0.93
R9512:Eif2ak4 UTSW 2 118293196 missense probably damaging 1.00
R9569:Eif2ak4 UTSW 2 118251316 missense probably benign 0.00
R9658:Eif2ak4 UTSW 2 118269511 missense probably damaging 1.00
R9748:Eif2ak4 UTSW 2 118247730 missense probably benign 0.01
R9757:Eif2ak4 UTSW 2 118269398 missense probably benign 0.02
R9766:Eif2ak4 UTSW 2 118261313 nonsense probably null
X0061:Eif2ak4 UTSW 2 118298657 missense probably damaging 1.00
Mode of Inheritance Autosomal Recessive
Local Stock Live Mice, Sperm
MMRRC Submission 036115-MU
Last Updated 2016-05-13 3:09 PM by Stephen Lyon
Record Created unknown
Record Posted 2012-01-11
Other Mutations in This Stock Stock #: A9681 Run Code:
Validation Efficiency: 88/103

GeneSubstitutionChr/LocMutationPredicted EffectZygosity
Ablim1 A to G 19: 57,161,755 (GRCm39) probably null Homo
Gm9943 T to A 17: 16,235,254 (GRCm39) Het
Ints1 T to C 5: 139,755,894 (GRCm39) E538G possibly damaging Homo
Or52n1 T to C 7: 104,383,610 (GRCm39) probably benign Het
Pakap A to G 4: 57,855,358 (GRCm39) Q270R probably damaging Het
Pld1 C to A 3: 28,139,981 (GRCm39) H600Q probably benign Homo

Stock #: ZE80 Run Code:
Validation Efficiency: 82/103

GeneSubstitutionChr/LocMutationPredicted EffectZygosity
AW551984 T to C 9: 39,504,963 (GRCm39) probably null Homo
Dop1a T to C 9: 86,382,895 (GRCm39) V276A probably damaging Het
Tspan12 T to C 6: 21,795,608 (GRCm39) H194R probably benign Homo
Phenotypic Description
The atchoum phenotype was identified among ENU-mutagenized G3 mice in an ex vivo macrophage screen for control of viral infection. Macrophages from the index atchoum mouse were highly susceptible to infection by the adenoviral vector Ad5-F16-GFP (1), as indicated by an increased percentage of GFP+ cells relative to wild type 72 hours after infection (2). The percentage of infected cells was similar to that seen for the STAT1 mutant domino (2;3). However, type I IFN production after infection was completely normal, unlike that of domino macrophages (2)Atchoum macrophages may be slightly susceptible to MCMV (4) infection, while both influenza and Rift Valley Fever Virus (RVFV) are completely controlled (Figure 1) (2). The phenotype is even stronger when macrophages are pre-treated overnight with IFNγ.

The function of the protein encoded by atchoum was assayed by exposure of Eif2ak4atc/atc mouse embryonic fibroblasts with ultraviolet (UV) radiation, a known activator of GCN2 (5;6).  In the atchoum fibroblasts an increase in eIF2α phosphorylation was not observed upon UVB exposure; a drastic increase in phosphorylation of eIF2α is observed at 40 hours post-viral infection in normal fibroblasts (2).  This indicates that the atchoum mutation results in a complete loss of GCN2 function (2).  Injection of atchoum mice with MCMV led to the deaths of several on or before nine days post-infection, indicating that viral susceptibility stems from a defect in the innate immune response (2).  The atchoum mice were tested for defects in the adaptive immune response; they mounted a normal IgG response to stimulation with β-galactosidase and a normal IgM response to NP-Ficoll (2)

Macrophages from the index atchoum mouse were also found to have a defective TNF-α response to the TLR7 ligand R-848 in the TLR Signaling Screen. However, testing of offspring from a cross of the index male and his sister demonstrated that the defective response to R-848 and susceptibility to adenovirus are unrelated, and caused by independently segregating mutations. The TLR signaling phenotype is now called rsq3.

Nature of Mutation

The atchoum mutation was mapped by bulk segregation analysis (BSA) of F2 backcross offspring using C57BL/10J as the mapping strain (n=40 with mutant phenotype, 9 with normal phenotype). The mutation showed strongest linkage with marker B10SNPS0032 at position 103735349 bp on Chromosome 2 (synthetic LOD=6.0) (2). Whole genome SOLiD sequencing of an atchoum homozygote was carried out: no Chromosome 2 mutations were validated among successful nucleotide calls covered 3 or more times. However, among 327 “N pattern” nucleotides (those covered 1 or 2 times) on Chromosome 2 (validation efficiency 281/327), a single mutation was identified at position 118226389 bp, 14.5 Mb from the marker with peak linkage (2). The mutation corresponds to a T to C transition at position 12038 of the genomic DNA sequence of Eif2ak4 (NC_000068), encoding GCN2. The mutation affects the sixth nucleotide of Eif2ak4 intron 2:

 
	      <--exon 2  intron 2-->
12016 CCTACATACCCAGATGT  GTGAGTACCTGTAT
80    -P--T--Y--P--D--V…
 

The mutated nucleotide is indicated in red; the splice donor sequence is shown in blue.  cDNA sequencing of four major Gcn2 transcripts in atchoum mice revealed that the mutation invariably results in skipping of exon 2, in some cases along with skipping of other exons (2); splicing from exon 1 to exon 3, exon 4, exon 5, or exon 6  was observed in the four sequenced transcripts.  The most abundant transcript lacked exons 2, 3, and 4; this would result in an in-frame deletion of 123 amino acids.

Illustration of Mutations in
Gene & Protein
Protein Prediction

Four kinases, GCN2, PKR, PERK, and HRI, regulate protein translation in response to distinct environmental stresses (7). They do so by phosphorylating serine 51 of the translation initiation factor eIF2α, leading to inhibition of translation initiation and thereby, protein synthesis (see Background). Initial cloning of mouse GCN2 revealed the presence of three isoforms (α, β, and γ) that differed in their N-terminal sequences and were derived by alternative splicing/transcription start of the GCN2 transcript (8;9). GCN2β is the longest isoform (1648 aa) and is the form discussed here unless otherwise specified. Amino acids 1-278 of GCN2β are missing from GCN2α, resulting in a protein 1370 aa in length. The first 86 amino acids of GCN2β are replaced with eight unique residues in GCN2γ, resulting in a protein 1570 aa in length. Six protein coding Eif2ak4 transcripts are annotated in the Vega database, encoding proteins from 347 to 1648 amino acids in length. Mouse and human GCN2 are 91% identical in sequence.

 
Figure 2. Domain structure of GCN2 and nature of the atchoum mutation. The atchoum mutation affects the sixth nucleotide of Eif2ak4 intron 2.  See text for description of RWD, kinase, HisRS-like, and RB/DD domains.  A degenerate kinase-like domain (ψKinase) that lacks critical residues for ATP binding and catalysis is located N-terminal to the kinase domain and may regulate authentic kinase activity.  The ψKinase domain is conserved in yeast, flies, and mice.

The kinase activity of GCN2 is regulated by several domains that serve to couple eIF2α phosphorylation to amino acid or serum starvation (Figure 2). Near the C terminus of GCN2 is a histidyl-tRNA synthetase (HisRS)-like domain to which uncharged tRNA binds, activating GCN2 (10-13). The HisRS domain binds to various uncharged tRNAs with similar affinities, discriminating only against their aminoacylated forms (14). Two segments within the N- and C-terminal halves of the HisRS domain (HisRS-N and HisRS-C) have been shown to mediate dimerization of GCN2 (15). Both HisRS-N and HisRS-C are required for GCN2 function in vivo, but neither is essential for dimerization of the full length protein, likely because of the presence of other dimerization domains in GCN2 (within the kinase domain and RB/DD; see below). It appears that dimerization is critical to the function of GCN2, and some evidence suggests that dimerization of the HisRS domain is required for tRNA binding (15). In addition, HisRS-N physically interacts with the kinase domain, and select mutations in HisRS-N abolished kinase function without impairing tRNA binding, suggesting that the HisRS and kinase domains participate in stable intramolecular interactions that, upon binding of tRNA, stimulate kinase activity through a conformational change propagated from HisRS to kinase domain (15).

The C-terminal 156 amino acids of GCN2 form the RB/DD domain, which contains ribosome binding (16) and homodimerization activities (17). The domain is characterized by three hydrophobic segments and a predicted amphipathic helix (18). RB/DD is required for GCN2 function in vivo (19), being necessary to induce eIF2α phosphorylation in response to amino acid limitation (18). In addition to binding the translation machinery, the isolated RB/DD can bind to double-stranded RNA in vitro, suggesting that interaction with rRNA mediates ribosome targeting (16). Homodimerization via the RB/DD and kinase domain is thought to permit trans-autophosphorylation of Thr882 and Thr887 in the activation loop, activating GCN2 kinase activity. The RB/DD also participates in direct physical contacts with the kinase domain that are thought to inhibit kinase activity (15;17). Binding of uncharged tRNA to the HisRS domain may stabilize a stimulatory conformation in which the C-terminal portion of GCN2 dissociates from the kinase domain and also becomes more accessible to tRNA. Another proposed role for the extreme carboxyl terminus of GCN2 is to facilitate binding of the HisRS domain to uncharged tRNA (10).

 
Figure 3. NMR structure of GCN2 RWD domain. The triple β-turn is shown in blue. UCSF Chimera structure is based on PDB 1UKX, Nameki et al, Protein Sci. 13, 2089-100 (2004). Click on the 3D structure to view it rotate.
Activation of GCN2 requires binding to GCN1, which forms a stable complex with the ATP-binding cassette protein GCN20 (20;21). GCN2 uses the RWD domain located at its N terminus to interact with GCN1 (20;22;23). The RWD domain is named for several classes of protein in which it has been found: RING finger-containing, WD repeat-containing, and yeast DEAD-like helicase proteins (24). An NMR structure of the GCN2 RWD domain has been reported, revealing an α + β sandwich fold and a unique, stable loop with sequence PPTYPDVVP comprising a triple β-turn involved in hydrogen bonds (Figure 3, PDB 1UKX) (25). The triple β-turn (YPXXXP motif) characteristic of RWD domains causes the peptide backbone to change its trajectory by 180° three times within the loop between strands β3 and β4. The domain is not known to have catalytic activity, but it shares significant structural homology with UBC9, an E2 ubiquitin-conjugating enzyme, and with Mms2, a yeast UEV protein (25). At least two RWD-containing proteins have been shown to promote ubiquitination (26;27) or sumoylation (28) dependent on the presence of the RWD domain. However, GCN2 involvement in ubiquitination has not been demonstrated. Interestingly, while the GCN2β isoform contains the complete RWD domain, GCN2γ and GCN2α would lack either the N-terminal portion or the entire domain, respectively, suggesting that they are regulated in a GCN1-independent manner.
 
Figure 4. Crystal structure of GCN2 kinase domain. UCSF Chimera structures are based on PDB 1ZYC, Padyana et al, J Biol Chem. 280, 29289-99 (2005). Click on the 3D structure to view it rotate.

The serine/threonine kinase domain of GCN2 is highly similar to those of PKR, PERK, and HRI. Analysis of the X-ray crystal structure of the isolated GCN2 kinase domain (insert deleted) demonstrated a typical, bi-lobed kinase fold comprised of N- and C-lobes linked by a flexible hinge and the ATP binding cleft positioned between the two lobes (Figure 4; PDB 1ZYC) (see iconoclast, a mutation in the Src family kinase Lck) (29). The GCN2 kinase domain crystallized as a homodimer. The smaller N-lobe is preceded by a short α-helix and contains a five-stranded β sheet with two α-helices (αB and αC) linking β3 and β4. In the apo form of the structure, the position of the important αC helix is shifted outward, as typically seen in the inactive conformation of kinases. In this position, the αC helix is unable to coordinate ATP within the catalytic cleft. The kinase domains of each of the eIF2α kinase domains is divided by a large insert; in GCN2 the insert is 139 amino acids in length and occurs in the N-lobe, between sheets β4 and β5. The larger C-lobe of the kinase contains amino acids implicated in catalysis, activation, and substrate recognition; it is predominantly α-helical.

Comparison of the structures of apo and ATP-bound GCN2 revealed that partial closure of the active site cleft in the apo form restricts ATP binding by restraining the conformation of the hinge region and strongly coupling the two kinase lobes (29). In contrast, a tRNA-independent constitutively active GCN2 point mutant (R794G) (30) had a relatively open conformation resembling ATP-bound GCN2, and lacked multiple interactions that rigidify the hinge of GCN2. In GCN2R794G, Asn793, which normally blocks ATP entry into the active site cleft, was displaced away from the active site permitting ATP binding. Thus, it was proposed that increased flexibility of the hinge region induced by uncharged tRNA binding to the HisRS domain would permit productive binding of ATP to the active site and autophosphorylation of Thr887 and Thr882 in the activation loop, thereby activating GCN2. 3D modeling of a dimeric kinase-HisRS segment supports the idea that the hinge interacts directly with HisRS, such that structural changes in HisRS upon tRNA binding could propagate to the hinge region of the kinase domain. In addition to modulating the flexibility of the kinase domain hinge, tRNA binding to the HisRS domain relieves a network of hydrophobic interactions that constrains rotation of helix αC into a stimulatory conformation for kinase activity (31).

The atchoum mutation occurs in intron 2. The GCN2β isoform would be affected by the mutation; neither GCN2α nor GCN2γ would be affected.

Expression/Localization

GCN2 mRNA was detected in all tissues examined, including heart, brain, spleen, lung, liver, skeletal muscle, kidney, and testis, with the highest expression in liver and brain (8;9). GCN2 isoforms were differentially expressed among tissues.

The atchoum mutation results in no detectable protein expression even though splicing from exon1 to exon 4, 5 or 6 would maintain the correct translational reading fram, suggesting that the mutation leads to an unstable protein that is subsequently degraded (2).

Background

One way that gene expression is regulated is through control of translation of mRNAs into proteins. Translational control is typically more rapid than transcriptional control because it exerts its effects on existing mRNAs, and can therefore meet the cell’s immediate needs to maintain homeostasis or adapt to stress in a changing environment. The process of translation can be divided into four stages: initiation, elongation, termination, and ribosome recycling. Most regulation occurs at the stage of initiation, in which the AUG start codon is identified and decoded within the P site of the ribosome by the methionyl tRNA specialized for initiation (Met-tRNAi) [reviewed in (32)].

 
Translation initiation
Figure 5. Translation initiation and regulation by eIF2α kinases. See text for details. Figure adapted from (32).

In eukaryotes, the start codon is generally identified by a scanning mechanism, where the small (40S) ribosomal subunit loaded with Met-tRNAi in a preinitiation complex (PIC) binds to the mRNA near the 5’ end and scans the 5’ untranslated region (UTR) for an AUG codon (Figure 5) (32). The mRNA itself is activated for PIC binding by the eukaryotic initiation factor 4F complex (eIF4F, composed of eIF4E, eIF4G, and eIF4A) binding to the 5’ cap, and poly A binding protein (PABP) binding to the poly(A) tail, circularizing the mRNA. The PIC is formed when eIF1, eIF1A, and eIF3 promote dissociation of the 80S ribosome into 60S and 40S subunits, and together with eIF5, ternary complex (composed of eIF2•GTP bound to Met-tRNAi), and the 40S subunit, assemble into the 43S PIC (33). This 43S PIC binds near the 5’ cap, facilitated by eIF3/eIF5 interactions with eIF4G/eIF4B, and scans the leader for complementarity to the anticodon of Met-tRNAi in an ATP-dependent reaction. Once the start codon has been identified, hydrolysis of GTP in the ternary complex leads to release of eIF2•GDP, Pi, and other eIFs from the 40S platform (34). Another factor, GTP-bound eIF5B, catalyzes the joining of the 60S ribosomal subunit to the 40S subunit, and subsequent release of eIF5B•GDP and eIF1A yields the final 80S initiation complex, ready to accept the appropriate aminoacyl-tRNA into the A site and synthesize the first peptide bond (35-38).

Cells use several mechanisms to regulate the efficiency of translation initiation. For example, RNA structures that impede the ability of ribosomes to interact with the 5′UTR in single-stranded form, or to scan the 5′UTR, reduce the efficiency of initiation. Decoy AUG codons in the 5′UTR can also slow a ribosome in finding the correct start codon. mRNA activation can be inhibited by inactivating components of the eIF4F complex. Some viral RNAs as well as eukaryotic mRNAs bypass the scanning process and the requirement for one or more eIFs by using specialized sequence elements called internal ribosome entry sites (IRESs) to recruit the 40S subunit directly to the start codon (39). Another prominent mechanism for blocking translation initiation is to reduce the activities of eIFs, such as eIF2, that stimulate Met-tRNAi recruitment to the 40S subunit.

 
Translational control by eIF2 phosphorylation

eIF2 is a heterotrimeric G protein consisting of α, β, and γ subunits that binds charged Met-tRNAi in a GTP-dependent manner and delivers it, in the context of the ternary complex, to the PIC (Figure 5). The eIF2 GTPase cycle begins when eIF2 is first activated by binding GTP, and the resulting eIF2•GTP interacts with Met-tRNAi to form a ternary complex, which is transferred to a 40S ribosomal subunit along with other eIFs to form the PIC. The assembled 43S PIC scans the mRNA, and upon AUG codon recognition, the factor eIF5 interacts with the ternary complex and functions as a GTPase activating protein (GAP) for eIF2•GTP, stimulating hydrolysis of GTP by eIF2 (35;40;41). eIF2•GDP, Pi, and other initiation factors are released, and the 60S ribosomal subunit joins to form the elongation competent 80S initiation complex. The inactive eIF2•GDP is then recycled by exchange of GTP for GDP, catalyzed by the guanine nucleotide exchange factor (GEF) eIF2B. A complex of eIF5/eIF2 antagonizes guanine nucleotide exchange by eIF2B (42;43).

Whereas the β and γ subunits of eIF2 mediate interactions with guanine nucleotide, Met-tRNAi, eIF2B, and eIF5 that permit ternary complex formation, GTP hydrolysis, and GTP/GDP exchange (44-51), the α subunit regulates these activities.  Phosphorylation of the α subunit at a conserved serine residue (S51) converts eIF2•GDP into a competitive inhibitor of eIF2B (52;53), resulting in functional sequestration of the limiting amounts of eIF2B present in the cell, and thereby decreasing levels of eIF2•GTP, ternary complex formation, and global protein synthesis. Regulation of eIF2 in this manner is conserved across species, and is controlled by four eIF2 kinases that respond to distinct environmental stresses: PERK is activated by misfolded proteins in the ER (ER stress); PKR by double-stranded RNA in virus-infected cells; HRI by heme deprivation, and oxidative and heat stresses in erythroid tissues; and GCN2 by amino acid deprivation, UV irradiation, and proteasome inhibition.

 

GCN2

Amino acid limitation results in an increase in uncharged tRNAs, which in turn activates GCN2. Paradoxically, phosphorylation of eIF2α reduces translation initiation generally while at the same time inducing translation of particular mRNAs through specialized mechanisms that enable the cell to turn on stress-specific response pathways. GCN2 was in fact first identified as an inducer of GCN4, a transcriptional activator of amino acid biosynthesis in yeast (54-56). Four short open reading frames (upstream ORFs, uORFs) in the GCN4 mRNA leader that support ribosomal reinitiation act as inhibitors of GCN4 translation under conditions of amino acid sufficiency (i.e. when eIF2•GTP is plentiful) (57;58). After translating the 5’ uORF (uORF1), post-termination 40S subunits rapidly re-acquire the activated, GTP-bound ternary complex, form a new PIC, and resume scanning. They re-initiate translation downstream at uORF2, uORF3, or uORF4; scanning does not resume after termination at these uORFs. Under conditions of amino acid starvation leading to GCN2 activation, eIF2α phosphorylation, and downregulation of eIF2•GTP, the rate of reassembly of the PIC after translation of uORF1 is slowed, thus delaying re-initiation of translation. The consequence is that start codons of uORFs 2-4 are bypassed by the ribosome, which instead re-initiates at the GCN4 start codon. Thus, eIF2α phosphorylation by GCN2 in amino acid-starved yeast cells translationally induces GCN4, which in turn transcriptionally activates amino acid biosynthesis.

In mammals as in yeast, GCN2 is required for adaptation to amino acid limitation, with the exception that non-essential amino acids (i.e. those that are synthesized by mammals) are also regulated in an eIF2 phosphorylation-independent manner in mammals. Although mice deficient in GCN2 and freely fed a normal diet developed and reproduced as well as wild type mice, prenatal and neonatal lethality were significantly increased in Gcn2–/– mice whose mothers were reared on leucine-, tryptophan-, or glycine-deficient diets during gestation (59). Cells from Gcn2–/– mice failed to induce phosphorylation of eIF2α and repression of eIF2B activity during histidine or leucine deprivation. Gcn2–/– mice maintained on a leucine-free diet for six days also lost more body mass than wild type mice, and 40% died or required euthanasia after 3-5 days (60). In the liver, in response to six days of leucine deprivation, wild type mice decreased protein synthesis concomitant with increased phoshphorylation of eIF2, whereas Gcn2–/– mice showed reduced eIF2 phosphorylation and levels of protein synthesis similar to those of mice fed complete diets. As a result, liver weights were preserved while skeletal muscle mass was sacrificially reduced in Gcn2–/– mice fed a leucine-free diet. GCN2 is also required for lipid homeostasis in the liver in response to amino acid deprivation: Gcn2–/– mice fed leucine-free diets for seven days developed steatosis and failed to repress lipogenic genes and the activity of fatty acid synthase (FAS) in the liver, as well as failed to mobilize lipid stores in adipose tissue (61). Finally, GCN2 participates in the sensing of nutritional imbalance that guides food selection for survival. Cells in the anterior piriform cortex of the brain are sensitive to essential amino acid deficiency and signal aversion to otherwise nutritious foods lacking deficient amino acids. GCN2 deficiency impaired this aversive response and phosphorylation of eIF2α in the anterior piriform cortex in response to an imbalanced diet (62-64).

GCN2 is though to act in the brain to inhibit long term memory formation through activation of ATF4. Although local protein synthesis at individual synapses is required for sustained increases in synaptic transmission, excessive translation is deleterious to memory formation. Thus, Gcn2–/– mice exhibited enhanced memory under a weak training protocol relative to wild type mice (65), as did eIF2αSer51Ala/+ heterozygous knock-in mice (66). The mechanism of enhanced memory formation is reduced translation of ATF4 as a result of decreased eIF2α phosphorylation. ATF4 inhibits CREB-mediated gene expression, which is critical for long term enhancement of synaptic strength (67). How GCN2 activity is regulated in response to synaptic stimulation is unknown.

In mammalian cells, upregulated translation of the bZIP transcription factor ATF4 is induced by GCN2 phosphorylation of eIF2α following a similar mechanism as used in control of yeast GCN4 translation upon amino acid starvation (68-71). Two uORFs in the 5’ leader of the ATF4 mRNA act to inhibit its translation during amino acid sufficiency, but promote translation during amino acid deficiency. Elevated levels of ATF4 lead to induction of transcription factors including ATF3 and CHOP, proteins that activate additional stress response genes, and GADD34, a regulatory subunit of an eIF2α phosphatase that provides feedback control of GCN2 activity [(70;72;73); reviewed in (74)].

GCN2 phosphorylation of eIF2 is reportedly required for activation of NF-κB in response to either amino acid starvation or UV irradiation of MEFs cultured in vitro (6;75). However, the activation of NF-κB in response to eIF2α phosphorylation was not dependent on IKK phosphorylation and proteasome-mediated degradation of IκB. Instead, GCN2 activation leads to decreased translation of IκB as part of the general inhibition of translation, contributing to the release of NF-κB from IκB and subsequent NF-κB translocation to the nucleus. GCN2 may downregulate caspase activity through NF-κB in order to protect against UV irradiation-induced apoptosis. GCN2 also mediates translation inhibition in response to UV irradiation (5). GCN2 is also activated and induces eIF2 phosphorylation in MEFs in response to treatment with proteasome inhibitors (76). NF-κB is not induced by proteasome inhibition. However, in this case abrogation of eIF2 phosphorylation blocked caspase activation and apoptosis.

GCN2 during viral infection

Viruses depend on the host translational machinery to synthesize their proteins. One mechanism mammalian cells use to counter viruses is to downregulate protein translation in infected cells through phosphorylation of eIF2α. The eIF2 kinase PKR is specialized to sense viral infection, being induced by interferon and activated mainly by dsRNA, which is synthesized in virus-infected cells as a byproduct of viral replication or transcription. Phosphorylation of eIF2α by PKR leads to inhibition of host and viral mRNA translation; PKR can also trigger apoptosis of virally infected cells [reviewed in (77)]. Consequently, viruses have developed a variety of strategies to gain access to the host translational machinery and avoid the deleterious effects of PKR activation. These include proteins that interfere with PKR activation, sequester dsRNA, inhibit PKR dimerization, act as PKR pseudosubstrates, activate antagonist phosphatases, and degrade PKR. Even in the face of a general inhibition of protein synthesis through downregulation of translation initiation, viruses can selectively promote translation of their own mRNAs through the use of IRESs and RNA structures that correctly position ribosomes without the function of eIF2.

Although PKR plays a major role in defense against viruses, other eIF2α kinases may have redundant functions in antagonizing viral infection. Knock-in mice or cells expressing a kinase-deficient PKR protein show normal antiviral responses to influenza, Sindbis, vaccinia, and encephalomyocarditis viruses, and normal levels of phosphorylated eIF2α under non-infected conditions (78-80). PERK and GCN2 have been implicated in antiviral responses to Sindbis virus, vesicular stomatitis virus (81;82) and herpes simplex virus 1 (HSV1) (83) through their phosphorylation of eIF2α. Upon HSV1 infection, PERK may be activated by accumulation of viral proteins in the ER (83), whereas GCN2 is directly activated in vitro by binding of the Sindbis virus genomic RNA (two noncontiguous regions) to the histidyl-tRNA synthetase-like domain (81). GCN2-deficient mice display increased susceptibility to Sindbis virus replication relative to wild type mice. However, one report suggests that phosphorylation of substrates other than eIF2α may be important for GCN2- and PERK-mediated antiviral responses (84). Vaccinia virus K3L protein acts as a pseudosubstrate of GCN2 to inhibit GCN2 phosphorylation of eIF2α, supporting the idea that GCN2 contributes to the eIF2-dependent antiviral response (85).

Putative Mechanism

Adenovirus contains a double-stranded DNA genome and infects cells of epithelial and lymphocytic origin. During the late phase of infection following DNA replication (after 18 hours post-infection (?)), adenovirus uses several mechanisms to promote viral protein synthesis while inhibiting cellular protein synthesis. One of these, targeted towards prevention of eIF2α phosphorylation, is the expression of the virus-associated RNA molecule I (VAI RNA), a noncoding, 160-nucleotide RNA transcribed by RNA polymerase III and highly expressed during the late phase of infection (86;87). VAI RNA directly binds PKR and prevents its activation (88). Through an unknown mechanism involving the ubiquitin ligase activity of the associated complex, the adenovirus E1B 55-kilodalton (E1B-55K) and E4 open reading frame 6 (E4orf6) proteins also help to prevent activation of PKR during the late phase of infection (89).

Activation of GCN2 in response to adenovirus infection has not been reported; neither have mechanisms utilized by adenovirus to counteract GCN2 been reported. Direct binding of adenoviral dsRNA species, such as VAI RNA, may directly activate GCN2 in a manner analogous to Sindbis virus activation of GCN2. The mechanisms by which GCN2 contributes to the antiviral response to adenovirus, and possibly MCMV, are unknown. 

It is proposed that viral infection may restrict amino acid availability and that the pool fo free amino acids within the cell is depleted during MCMV and adenovirus infection.  GCN2 would subsequently downregulate further translation which would affect both cellular and viral proteins (2)

Primers Primers cannot be located by automatic search.
Genotyping

Atchoum genotyping is performed by amplifying the region containing the mutation using PCR, followed by sequencing of the amplified region to detect the single nucleotide change.

Primers

atchoum (F): 5’-AATTGGCTGGGACGGTGTCAAG -3’

atchoum (R): 5’- GGAAGCACTTTAAATGCTCGCCAC -3’

PCR program

1) 95°C             2:00

2) 95°C             0:30

3) 56°C             0:30

4) 72°C             1:00

5) repeat steps (2-4) 29X

6) 72°C             7:00

7)  4°C              ∞

Primers for sequencing

atchoum_seq(F): 5’- GCTGTAACTTAGTGTACAGGGAC -3’

atchoum_seq(R): 5’- CATCGGTAGTGAGCACTCTACAG -3’

The following sequence of 581 nucleotides (from Genbank genomic region NC_000068.6 for linear genomic sequence of Eif2ak4, sense strand) is amplified:

       aattggctg ggacggtgtc aagcagctgt aacttagtgt acagggacag gtggggagaa

11881 ggaaaaaccc tacgagatgt gttctttttc tcttttaggt gagagagcct cctgaaatca

11941 acttagttct ttaccctcag ggcctagctg gtgaagaagt atacgtgcaa gtggaactga

12001 gggttaaatg cccacctaca tacccagatg tgtgagtacc tgtataagaa gcttccgtgt

12061 catttcgctt ttctgttttg acagcatagg aatattacag ttgggagcat ttcctctttc

12121 cttcctttgt ctgccaatgt aaagtttcat ttcccatcac ctatatctca gtaacactat

12181 atccgcccta gactttttgc aaagcagcct ttaagtctgc atatagctgt agagtgctca

12241 ctaccgatgc taatattcac tcattgaccg gatgctagta tttactcatt gaccaccttt

12301 ttgcttttgc aactttcatg cttatttaat tctttgtaat ctagagagaa ggaaggtcta

12361 tacatatagg ataaagatgt ggcgagcatt taaagtgctt cc

Primer binding sites are underlined; sequencing primer binding sites are highlighted in gray; the mutated T is indicated in red.

References
Science Writers Eva Marie Y. Moresco
Illustrators Diantha La Vine
AuthorsCeline Eidenschenk, Sungyong Won, Bruce Beutler