Thomas E. Dever, PhD, Head, Section on Protein Biosynthesis
Byung-Sik Shin, PhD, Staff Scientist
Pankaj Alone, PhD, Postdoctoral Fellow
Madhusudan Dey, PhD, Postdoctoral Fellow
Jeanne M. Fringer, PhD, Postdoctoral Fellow
Preeti Saini, PhD, Postdoctoral Fellow
Eun Joo Seo, PhD, Postdoctoral Fellow
Chune Cao, Biological Laboratory Technician
Joo-Ran Kim, BS, Special Volunteer

We are studying the mechanism and regulation of protein synthesis, focusing on GTPases and protein kinases that control this fundamental cellular process. We are using molecular-genetic and biochemical studies to dissect the structure-function properties of the translation initiation factor eIF2, a GTPase that binds methionyl-tRNA to the ribosome. We are also studying how eIF2 function is regulated by phosphorylation. Focusing on the kinases that phosphorylate eIF2alpha, we recently showed that a common dimer configuration is required for activation of the antiviral kinase PKR, the ER stress-responsive kinase PERK, and the endogenous yeast kinase GCN2. We are also studying viral regulation of PKR by poxvirus inhibitors of the kinase. The translation factor eIF5B is a second GTPase that catalyzes ribosomal subunit joining in the final step of translation initiation. Our recent mutational and suppressor analyses combined with biochemical studies provided new insights into the structure-function properties of this universally conserved factor and defined the critical role of GTP hydrolysis in releasing the factor from the ribosome following subunit joining.

Structure-function analysis of eIF2gamma, the GTP-binding subunit of the eIF2 complex

Alone, Dever; in collaboration with Lorsch

The translation initiation factor eIF2 is composed of three polypeptide chains that assemble to form a stable complex. The gamma subunit of eIF2 contains a consensus GTP-binding (G) domain, and the factor must bind to GTP to form a stable eIF2•GTP•Met-tRNA ternary complex. The GTPase-activating protein (GAP) eIF5 promotes GTP hydrolysis by eIF2, and the guanine-nucleotide exchange factor (GEF) eIF2B is responsible for exchanging GTP for GDP on eIF2, enabling the factor to function in additional rounds of translation initiation. Curiously, previous studies revealed that these two effectors of eIF2 function interact with eIF2beta but not with the GTP-binding eIF2gamma subunit. To identify proteins that interact with eIF2gamma, we expressed full-length eIF2gamma as well as its individual domains as GST fusion proteins in yeast cells. GST pull-down experiments revealed that eIF2alpha, eIF2beta, eIF5, and eIF2B interacted with full-length eIF2gamma, whereas eIF5 and eIF2B, but not eIF2alpha or eIF2beta, bound to the eIF2gamma G domain. Importantly, these interactions were mapped to the catalytically critical N-terminus of eIF5 and C-terminal domain of eIF2Bepsilon. Thus, these critical regulators of eIF2 function make direct contacts with the G domain of eIF2gamma, consistent with their roles in promoting GTP hydrolysis and GTP-GDP exchange on eIF2 (Alone and Dever, 2006).

In addition to binding the Met-tRNA to the ribosome, eIF2 plays an important role in start codon recognition. Mutation of the conserved Switch 1 element in the G domain of eIF2gamma impairs yeast cell growth and enables translation to start at a non-canonical UUG codon (sui- phenotype). Biochemically, the eIF2gamma-N135D mutation impaired Met-tRNA binding, consistent with the idea that premature or unregulated release of Met-tRNA from eIF2 enables translation to initiate at non-AUG codons. However, characterization of several intragenic suppressors of the eIF2gamma-N135D mutation revealed a more complex connection between eIF2 function and start codon selection. Whereas an A208V mutation in eIF2gamma suppresses both the Met-tRNA binding defect and the sui- phenotype associated with the N135D mutation, an A382V mutation fully suppresses the Met-tRNA binding defect but not the sui- phenotype. This genetic separation of Met-tRNA binding affinity and non-AUG codon recognition suggests important roles for other factors in start codon recognition. Along these lines, overexpression of eIF1 suppresses the sui- phenotype in both the eIF2gamma-N135D mutant and A382V suppressor strain, consistent with other genetic and biochemical data indicating a critical role for eIF1 in monitoring start codon recognition.

Activation and substrate recognition by the eIF2alpha protein kinases PKR and GCN2

Dey, Cao, Dever; in collaboration with Sicheri

Phosphorylation of eIF2alpha is a common mechanism for downregulating protein synthesis under stress conditions. Four kinases phosphorylate eIF2alpha on Ser-51 under different cellular stress conditions. GCN2 responds to amino acid limitation, HRI to heme deprivation, PERK to ER stress, and PKR to viral infection. Consistent with their common activity in phosphorylating eIF2alpha on Ser-51, the kinases show strong sequence similarity in their kinase domains. Phosphorylation of eIF2alpha converts eIF2 from a substrate to an inhibitor of its guanine-nucleotide exchange factor eIF2B. Such inhibition of eIF2B impairs general translation, slowing the growth of yeast cells and paradoxically enhancing the translation of the GCN4 mRNA required for yeast cells to grow under conditions of amino acid starvation. We used structural, molecular, and biochemical studies to define how the eIF2alpha kinases recognize their substrate. In collaboration with Frank Sicheri, we obtained the X-ray structure of eIF2alpha bound to the catalytic domain of PKR (Dar et al., 2005). The PKR kinase domain resembles typical eukaryotic protein kinases, with the active site located between a smaller N-terminal lobe and larger C-terminal lobe. In the crystal structure, the PKR kinase domain dimerizes in a back-to-back orientation mediated by conserved residues in the N-terminal lobe. One molecule of eIF2alpha is bound to each PKR kinase domain protomer, with the conserved eIF2alpha "KGYID" motif interacting with helix alphaG in the kinase domain C-terminal lobe and the Ser-51 phosphorylation site positioned near the PKR active site. Given that all four eIF2alpha kinases share the PKR residues mediating kinase domain dimerization and eIF2alpha recognition, we propose that the four kinases dimerize and recognize eIF2alpha in a similar manner.

A hallmark of the PKR dimer interface is the presence of reciprocal salt-bridge interactions between R262 of one PKR protomer and D266 of the other protomer. Charge-reversal mutations of R262 to D or D266 to R, designed to disrupt the salt-bridge interaction, abolished PKR function in yeast cells and in vitro. However, combining the two mutations into the same PKR allele, which would restore the salt-bridge interaction with opposite polarity, rescued PKR function both in vivo and in vitro. Likewise, the corresponding single mutations designed to disrupt putative salt-bridge interactions in GCN2 and PERK abolished kinase activity. More important, the double mutations in GCN2 and PERK, which would restore the putative salt-bridge interactions, restored the kinases' function both in vivo and in vitro. We conclude that the back-to-back dimer orientation observed in the PKR crystal structure is critical for the activity of PKR, GCN2, and PERK and that PKR structure represents the active state of the eIF2alpha kinase domain.

Binding of double-stranded RNA to the regulatory domains in the N-terminal half of PKR is thought to promote dimerization and activation of the kinase. We screened for mutants that activate PKR in the absence of its regulatory domain. The mutations enhanced PKR kinase domain dimerization and mapped to the dimerization interface on the N-terminal lobe of kinase domain (Dey et al., 2005a). Sequence comparisons of all four eIF2alpha kinases and other unrelated kinases identified 17 residues preferentially conserved among the eIF2alpha kinases. We systematically mutated these conserved residues in PKR and screened for mutants that impaired PKR toxicity in yeast. Mutation of conserved residues on the kinase domain dimerization interface blocked PKR autophosphorylation and eIF2alpha phosphorylation while mutation of the Thr-446 autophosphorylation site in the kinase domain activation segment impaired eIF2alpha phosphorylation and viral pseudosubstrate binding. Finally, mutation of residues in helix alphaG prevented phosphorylation of eIF2alpha but not phosphorylation of the nonspecific substrate histone, consistent with the docking of eIF2alpha on helix alphaG in the PKR•eIF2alpha co-crystal structure. Based on our studies, we propose an ordered mechanism of PKR activation in which catalytic domain dimerization triggers Thr-446 autophosphorylation, which in turn is required for specific eIF2alpha substrate recognition (Dey et al., 2005a).

Poxvirus regulation of protein kinase PKR

Seo, Cao, Dever; in collaboration with Sicheri

The phosphorylation of eIF2alpha by PKR is one means used by mammalian cells to block viral replication. To subvert this host cell defense mechanism, viruses produce inhibitors of PKR. Several members of the poxvirus family express two types of PKR inhibitor: a pseudosubstrate inhibitor and a double-stranded RNA binding protein called E3L. The vaccinia virus K3L protein resembles the N-terminal third of eIF2alpha, with both proteins containing a beta-barrel fold of the OB-fold family. In previous studies, we showed that expression of the K3L protein in yeast prevented PKR inhibition of yeast cell growth. Moreover, we showed that the PKR-inhibitory activity of the K3L protein was dependent on residues conserved between the K3L protein and eIF2alpha, including the "KGYID" motif that is critical for recognition of eIF2alpha by the eIF2alpha protein kinases. More recently, we found that the related C3L protein from the smallpox (variola) virus is also able to block PKR toxicity in yeast. Thus, like the vaccinia virus, the smallpox virus uses a pseudosubstrate inhibitor to block PKR function.

To gain further insight into pseudosubstrate regulation of PKR, we screened for PKR mutants resistant to K3L inhibition. We isolated 12 PKR mutants that were toxic in yeast cells co-expressing the K3L protein. Interestingly, the 12 PKR mutations mapped to the C-terminal lobe of the kinase domain, consistent with the fact that the eIF2alpha binding site mapped to the same region of the kinase. Additional in vitro studies with the PKR-D486V mutant revealed decreased K3L binding. Moreover, the PKR mutant phosphorylated eIF2alpha with the same kinetics as the wild-type kinase; however, the mutant kinase was less sensitive to inhibition by K3L in vitro. Interestingly, the PKR mutant was not resistant to the viral E3L inhibitor. Thus, the D486V mutation appears specifically to desensitize PKR to inhibition by the viral pseudosubstrate K3L protein. We propose that the PKR mutations, either directly or indirectly, alter points of contact between PKR and its pseudosubstrate such that the mutations have a greater impact on pseudosubstrate inhibition than on substrate phosphorylation and thus confer resistance to pseudosubstrate inhibition.

In related studies, we recently identified PKR mutants resistant to inhibition by the E3L protein. These mutations map to both the kinase domain and the double-stranded RNA (dsRNA) binding domain of the protein. Further characterization of the mutants may provide novel insights into the dsRNA binding properties of PKR, the mechanism of dsRNA activation of the kinase, and the mechanism by which E3L blocks PKR function.

Kazemi S, Papadopoulou S, Li S, Su Q, Wang S, Yoshimura A, Matlashewski G, Dever TE, Koromilas AE. Control of eukaryotic translation initiation factor 2alpha (eIF2alpha) phosphorylation by the human papillomavirus type 18 E6 oncoprotein: implications for eIF2alpha-dependent gene expression and cell death. Mol Cell Biol 2004;24:3415-29.

Structure-function analysis of universally conserved translational GTPase eIF5B/IF2

Shin, Fringer, Cao, Kim, Dever; in collaboration with Lorsch

In the final step of translation initiation, the large 60S ribosomal subunit joins the 40S subunit, which is already bound to an mRNA, to form an 80S ribosome competent for protein synthesis. We previously discovered the translation initiation factor eIF5B, an orthologue of the bacterial translation factor IF2, and showed that eIF5B catalyzes ribosomal subunit joining. The eIF5B binds to GTP and hydrolyzes GTP in the presence of 80S ribosomes. The X-ray structure of eIF5B revealed a chalice-shaped protein in which GTP binding to the G domain residing in the cup of the chalice triggers lever-type domain rearrangements through a long alpha-helix forming the stem of the chalice, resulting in significant movement of domain IV, the base of the chalice. Our current efforts aim at elucidating the structure-function properties of eIF5B and understanding the role of GTP binding and hydrolysis by this translation initiation factor.

All G domains contain two structural elements referred to as Switch 1 (Sw1) and Switch 2 (Sw2) that undergo marked conformational changes upon GTP binding and hydrolysis. Conserved residues in Sw1 and Sw2 directly interact with the guanine nucleotide and play important roles in GTP binding and hydrolysis. Mutation of a conserved Thr in Sw1 of yeast eIF5B severely impaired yeast cell growth and eliminated eIF5B GTPase and translational stimulatory properties. However, the mutation did not impair GTP binding or ribosomal subunit joining. An intragenic suppressor of the eIF5B Sw1 mutant restored yeast cell growth and eIF5B translational stimulatory activities but not eIF5B GTPase activity. Thus, the eIF5B suppressor mutant uncoupled eIF5B GTPase and translational stimulatory activities. Additional studies revealed that the suppressor mutation lowered eIF5B ribosome binding affinity. We propose that a GTP-regulated switch governs eIF5B ribosome binding affinity. In the presence of GTP, eIF5B binds to the ribosome and, following GTP hydrolysis, is released (Shin et al., Cell 2002;111:1015). In further studies, we mutated a conserved Gly residue in Sw2 that is thought to be important for Sw2's structural flexibility. Substitution of Ala for the conserved Gly in eIF5B impaired yeast cell growth and the guanine-nucleotide binding, GTPase, and ribosomal subunit-joining activities of eIF5B. We identified two intragenic suppressor mutations: one mapping to Sw1 and the other extrinsic to the G domain. Interestingly, the residues mutated in both suppressors interact with Sw2 in the eIF5B crystal structure. The suppressor mutations restored the eIF5B nucleotide binding, GTPase, and subunit-joining activities. We propose that the Gly mutation in Sw2 distorts the geometry of the eIF5B active site, impairing nucleotide binding and hydrolysis and associated eIF5B domain movements. The suppressor mutations may indirectly reposition Sw2 and restore a favorable active site geometry, enabling the factor to bind to and hydrolyze GTP and to couple eIF5B domain movements with guanine-nucleotide binding.

While in vitro studies have clearly revealed eIF5B's function as catalyzing subunit joining, in vivo evidence has been lacking. By using an eIF5B degron allele, which permits rapid depletion of eIF5B in the cell, we discovered an increase in 48S complexes, which are 40S ribosomes bound to mRNA awaiting 60S subunit joining. This in vivo observation supports the role of eIF5B in catalyzing subunit joining. The C-terminus of eIF5B is known to interact with factor eIF1A. Recently, we found that overexpression of eIF5B rescues the growth defect in strains expressing a C-terminal mutant form of eIF1A. Moreover, the eIF1A mutant suppressed the toxic effects of a catalytically defective form of eIF5B. These genetic results are consistent with the notion that the eIF5B•eIF1A interaction promotes eIF5B recruitment to the ribosome. The results also support biochemical data showing that the eIF5B•eIF1A interaction is required for efficient ribosomal subunit joining.

We propose the following model for the role of eIF5B in translation initiation: following scanning of the 40S ribosomal complex to an AUG codon on an mRNA, eIF2 hydrolyzes GTP, causing the release of eIF2 and perhaps other factors; eIF1A remains bound to the 40S subunit and facilitates binding of eIF5B•GTP; binding of eIF5B•GTP stabilizes Met-tRNA binding to the 40S complex and promotes joining of the 60S ribosomal subunit; formation of the 80S ribosome triggers GTP hydrolysis by eIF5B; and the factors eIF5B and eIF1A are released, enabling the 80S ribosome to begin translation elongation.

COLLABORATORS

Jon R. Lorsch, PhD, The Johns Hopkins University, Baltimore, MD
Frank Sicheri, PhD, Samuel Lunenfeld Research Institute, Mount Sinai Hospital, and University of Toronto, Toronto, Canada

For further information, contact devert@mail.nih.gov.