Coeur d'Alene resort. Photograph courtesy of Tim Eberly. The Keystone Symposium on Translational Regulatory Mechanisms was held between 28 January and 2 February 2008, in Coeur d'Alene, Idaho, USA, and was organized by N. Sonenberg, A.G. Hinnebusch and J.D. Richter.©Tim Eberly
About 140 scientists gathered in snowy Coeur d'Alene, Idaho, USA, to attend the Keystone Symposium on Translational Regulatory Mechanisms. This meeting was unique in bringing together a diverse group of researchers studying the mechanism and structural biology of the translation apparatus, messenger RNA (mRNA) degradation and transport, microRNA (miRNA)‐mediated control of gene expression, and the translational regulatory processes underlying development, learning and memory, and human diseases.
The initiation of protein synthesis requires the delivery of both the methionyl‐initiator transfer RNA (Met‐tRNAiMet) and the mRNA to the ribosome. These processes are coordinated by translation‐initiation factors as summarized in Fig 1. Several talks provided new insights into the ordered assembly of the translating 80S ribosome. V. Ramakrishnan (Cambridge, UK) presented a structural analysis of the yeast 40S ribosomal subunit, and revealed that the binding of eIF1 and eIF1A induced the opening of the mRNA‐binding channel (Passmore et al, 2007), perhaps providing a rationale for the ordered assembly of the initiation complex. J. Doudna (Berkeley, CA, USA) showed that the eIF3j subunit competed with eIF1A for binding to the 40S subunit, and that eIF3j prevented high‐affinity mRNA binding in the absence of the ternary complex (Fraser et al, 2007). Therefore, eIF3j might act as a regulatory switch on the 40S complex to ensure the ordered recruitment of the ternary complex and mRNA to the 40S subunit.
After binding to the mRNA, the 43S ribosome complex scans for, and selects, an AUG start site as directed by, at least in part, codon–anticodon interactions between the tRNAiMet and the mRNA. In separate talks based on their mutational and biophysical studies of eIF1A, as well as examination of start‐codon selection, A. Hinnebusch (Bethesda, MD, USA) and J. Lorsch (Baltimore, MD, USA) proposed that the negatively charged eIF1A carboxyl terminus promotes the open scanning‐competent state of the 40S subunit and blocks initiation at non‐AUG codons, whereas the positively charged amino terminus of eIF1A has the opposite function and is required for formation of the scanning‐incompetent closed 43S conformation at start‐codon selection (Fekete et al, 2007). A. Hinnebusch went on to describe a set of ribosomal RNA (rRNA) mutations that impaired reinitiation at short, upstream open reading frames on GCN4 mRNA. These studies provide the first insights into the ribosomal determinants in eukaryotes that facilitate Met‐tRNAiMet binding. Once the 43S complex locates a start site, eIF1A helps to recruit eIF5B, to promote subunit joining (Acker et al, 2006; Fringer et al, 2007).
The secondary structure near the 5′ end of mRNA impairs translation by preventing binding or scanning of the 43S complex. Ribosome access to the mRNA is facilitated by the cap‐binding complex eIF4F. The eIF4F complex consists of the cap‐binding protein eIF4E, the RNA helicase eIF4A and the scaffolding protein eIF4G, and is thought to unwind secondary structures near the cap that prevent 43S‐complex binding. The helicase activity of eIF4A is promoted by eIF4B and/or eIF4H. M. Moore (Worcester, MA, USA) reported that the binding of either eIF4B or eIF4H to eIF4A is mutually exclusive. Moreover, she drew a parallel between the interaction of eIF4B or eIF4H with eIF4A and the related exon‐junction complex (EJC) in which the RNA‐binding protein MLN51 interacts with the helicase eIF4AIII. Interestingly, analysis of eIF4A/eIF4AIII chimaeras led to the identification of an element in eIF4A that is crucial for the interaction with eIF4B or eIF4H. In a short talk, A. Marintchev (Boston, MA, USA) presented structural and functional studies of the eIF4A·4G·4H complex (Marintchev & Wagner, 2005; Oberer et al, 2005).
R. Jackson (Cambridge, UK) discussed the issue of reinitiation after translation of a short versus long ORF in mammalian cell extracts. Reinitiation after a short ORF required the primary initiation event to use eIF4F (or at least the central domain of eIF4G), indicating that constituents of the primary 43S complex might be retained through translation of the short ORF and promote re‐binding of the ternary complex or other factors (Poyry et al, 2004). By contrast, reinitiation after a long ORF is rare. One exception is the bicistronic subgenomic region of feline calicivirus. The binding of eIF3 to the distal 87 nucleotides of the upstream pre‐VP1 ORF positively correlated with the efficiency of reinitiation at the downstream ORF (Poyry et al, 2007). It will be interesting to determine whether this eIF3 dependence is related to the function of the factor in 40S recruitment or its role in promoting ribosome recycling (Pisarev et al, 2007).
Translational control and disease models
Translational control in eukaryotes is centred on the eIF2‐dependent binding of Met‐tRNAiMet to the 40S subunit and the eIF4F‐dependent binding of the 43S complex to an mRNA. Phosphorylation of Ser 51 on eIF2α impairs the function of the guanine‐nucleotide exchange factor (GEF) and eIF2B, and blocks the conversion of inactive eIF2·GDP to functional eIF2·GTP. T. Dever (Bethesda, MD, USA) reported on the structural basis for substrate recognition by the antiviral eIF2α kinase, PKR (Dar et al, 2005; Dey et al, 2005). Interestingly, binding of the kinase apparently induces a conformational change in eIF2α to expose Ser 51 to the kinase active site. Identification of PKR mutants that were resistant to inhibition by a poxvirus pseudosubstrate revealed an overlap with kinase residues that had been under positive selection during PKR evolution, lending support to the idea that poxviruses have had an important role in driving PKR evolution.
D. Ron (New York, NY, USA) showed that eIF2α phosphorylation is also important in regulating intermediary metabolism in the liver of obese mice. Conditional liver‐specific expression of the eIF2α‐targeting GADD34 subunit of PP1 resulted in an anti‐diabetic phenotype: that is, reduced glycogen and fat levels, susceptibility to hypoglycaemia, enhanced glucose tolerance and diminished hepato‐steatosis in animals fed a high‐fat diet. Phosphorylation of eIF2α stimulates translation of hepatic transcription factors, such as C/EBPα and C/EBPβ, which might mediate the deleterious effects of nutrient excess. D. Cavener (University Park, PA, USA) reported that tissue‐specific knockout of the endoplasmic‐reticulum‐resident eIF2α kinase, double‐stranded RNA‐dependent PERK, in the exocrine pancreas caused the cells to atrophy without affecting glucose homeostasis. By contrast, endocrine pancreas‐specific knockout of PERK resulted in a diabetic phenotype due to a reduced proliferation rate of β‐cells (Iida et al, 2007; Zhang et al, 2006). Therefore, PERK function and eIF2α phosphorylation are crucial determinants of glucose control in mammals.
The eIF4F‐dependent binding of mRNA to the 43S complex is regulated by eIF4E‐binding proteins (4E‐BPs) that disrupt the eIF4F complex by competing with eIF4G for binding to eIF4E. M. Costa‐Mattioli (Montreal, QC, Canada) showed that cells and mice that are devoid of both 4E‐BP1 and 4E‐BP2 display a marked resistance to virus infections due to increased translation of IRF7 mRNA and upregulation of type‐I‐interferon production (Colina et al, 2008). These findings are consistent with the idea that mRNAs burdened with secondary structure near the cap require greater amounts of eIF4F and its associated helicase activity for their translation. I. Mohr (New York, NY, USA) further demonstrated the importance of eIF4F‐complex formation by describing the various strategies that viruses use to enhance eIF4F activity. Whereas HSV‐1 encodes a chaperone that promotes eIF4G association with eIF4E, cytomegalovirus apparently drives eIF4F assembly by markedly increasing the levels of eIF4E, eIF4G and the poly(A)‐binding protein (PABP), which directly bind to eIF4G. By contrast, poxvirus infection causes the destruction of 4E‐BP, and the sequestration and concentration of eIF4F components near discrete cytosolic centres of viral replication (Walsh et al, 2008). R. Schneider (New York, NY, USA) reported that eIF4G is overexpressed in inflammatory breast cancer, which led him to assess the impact of depleting eIF4G in cells. Surprisingly, small‐interfering RNA (siRNA) targeting the eIF4GI isoform had little impact on general protein synthesis; however, importantly, subsets of mRNAs containing many upstream ORFs or of low abundance were excluded from polysomes (Ramirez‐Valle et al, 2008). These findings further support the idea that, based on their structural features, mRNAs require different levels of eIF4F for their translation. G. Wagner (Boston, MA, USA) described the identification of small molecules that inhibit formation of either the eIF4E·eIF4G or the eIF4A·eIF4G complexes. Interestingly, the eIF4E·eIF4G complex inhibitor, 4EGI‐1 (Moerke et al, 2007), binds to eIF4E but does not disrupt the binding of 4E‐BP1; instead, it stabilizes the eIF4E·4E‐BP complex. As 4EGI‐1 inhibits the proliferation of some cancer cells, these findings raise the possibility that eIF4F assembly might be a promising target for cancer and antiviral drugs.
mRNA localization and translational control in development
The localization of mRNAs to achieve concentrations of specific proteins in different regions of a cell seems to be a common regulatory mechanism in development and neuronal signalling. For this localization to be effective, translation should be suppressed until after the mRNA is transported to its final destination. In his keynote address, R. Singer (Bronx, NY, USA) presented two cases in which 60S‐subunit joining is inhibited until the mRNA is transported. In mammalian fibroblasts and neurons, binding of the protein ZBP1 to elements in the 3′ untranslated region (UTR) of the β‐actin mRNA inhibits translation during sorting of the mRNA. Localized phosphorylation of ZBP1 by SRC kinase relieved this translational repression (Huttelmaier et al, 2005). In an analogous fashion, binding of Puf6 to the 3′ UTR of the yeast ASH1 mRNA represses translation and promotes localization of the mRNA to the daughter bud. Interestingly, the association of Puf6 with the subunit‐joining factor eIF5B was required for proper regulation of ASH1 mRNA in vivo, and phosphorylation of Puf6 by casein kinase II in the daughter cell relieved the translational inhibition (Deng et al, 2008).
Several talks reported on the identification of mRNA targets of transport factors. J. Casolari (Stanford, CA, USA) reported on his systematic screen of mRNAs associated with the cytoskeletal motor proteins in yeast. Interestingly, only a subset of the actin‐based motors, and none of the microtubule motor proteins, was associated with mRNAs. J. Darnell (New York, NY, USA) described a crosslinking and immunoprecipitation methodology (CLIP) that she has been using to identify polysome‐associated in vivo mRNA targets of the fragile X mental retardation protein (FMRP). J. Carson (Farmington, CT, USA) reported that, in neurons, several mRNAs are combined into granules for transport along microtubules to the synapse where individual mRNAs are then translated (Gao et al, 2008).
The cytoplasmic polyadenylation element‐binding protein (CPEB) was initially found to govern cytoplasmic polyadenylation of mRNAs and to have an essential role in mitotic cycling in Xenopus egg extracts. So, it was a surprise when CPEB‐knockout mice were found to be viable. J. Richter (Worcester, MA, USA) reported that mouse embryonic fibroblasts (MEFs) lacking CPEB bypass senescence, at least in part, through derepression of c‐Myc expression (Groisman et al, 2006). It will be interesting to learn whether the c‐Myc mRNA is a direct target of CPEB, and how loss of CPEB increases c‐Myc expression.
Translational control in learning and memory
A crucial requirement for the conversion of labile short‐term memory into consolidated long‐term memory is new protein synthesis. Similar to memory, activity‐dependent modulation of synaptic strength (long‐term potentiation; LTP)—a key cellular mechanism by which information is stored—has two temporally distinct phases: early LTP, which depends on the modification of pre‐existing proteins; and late LTP (L‐LTP), which requires new protein synthesis. If new protein synthesis is the rate‐limiting step that is necessary to strengthen existing synaptic connections between neurons, what are the signalling pathways involved in these processes?
The eIF2α protein regulates two fundamental processes that are essential for the generation of long‐lasting memories. Phosphorylation of eIF2α inhibits de novo protein synthesis but it also stimulates the synthesis of a specific protein, ATF4 (also known as CREB2), which is a well‐known memory repressor that blocks CREB‐mediated gene expression. N. Sonenberg (Montreal, QC, Canada) reported that deletion of the eIF2α kinase GCN2 or genetic reduction of eIF2α phosphorylation—both of which are conditions that lead to impaired expression of ATF4—reduced the threshold for induction of LTP, and led to enhanced learning and memory (Costa‐Mattioli et al, 2005, 2007). By contrast, induction of eIF2α phosphorylation with a small molecule not only leads to an inhibition of protein synthesis and increased expression of ATF4, but also has the opposite depressant effects on lasting synaptic changes and memory. These data indicate that eIF2α phosphorylation could represent a type of molecular switch that contributes to enduring long‐lasting memories.
E. Klann (New York, NY, USA) underscored the role of the mTOR signalling pathway, which is a known modulator of translation, in long‐lasting synaptic plasticity and memory. Mice with genetic deletion of either of the mTOR substrates 4E‐BP2 or S6K1—the phosphorylation of which is involved in translation initiation—displayed alterations in L‐LTP and impaired long‐term contextual fear memory. Moreover, genetic deletions of negative modulators of mTOR resulted in mice with enhanced L‐LTP and long‐term contextual fear memory, and behaviours reminiscent of autism spectrum disorders. J.‐C. Lacaille (Montreal, QC, Canada) reported that knockdown of Staufen1—an RNA‐binding protein that functions in mRNA transport, localization, translational control and decay—in hippocampal pyramidal cells blocked L‐LTP, decreased the frequency and amplitude of spontaneous excitatory synaptic activity, and reverted dendritic spines to phenotypically immature shapes (Lebeau et al, 2008).
Coupling of translation to mRNA stability
The surveillance mechanism by which eukaryotic cells remove potentially deleterious mRNAs harbouring a premature termination codon (PTC) is known as nonsense‐mediated decay (NMD). Two models of NMD have been proposed. In the first model, which has been best studied in yeast, aberrant termination—defined as the failure of ribosomes to terminate near the normal 3′ UTR and poly(A) tail—triggers NMD. Consistently, recruitment of PABP to the PTC blocks NMD (Amrani et al, 2006). In the second model, which has been best studied in mammals, the recognition of proteins (such as an EJC) deposited downstream from the PTC targets an mRNA for NMD. In normal mRNAs, these factors are displaced by the scanning ribosome during a pioneer round of translation, whereas their retention in aberrant mRNAs precipitates degradation (Isken & Maquat, 2007). At least three core factors, UPF1, UPF2 and UPF3, are required to execute NMD. UPF1 interacts with translation‐release factors eRF1 and eRF3; in addition, UPF1 triggers decapping and subsequent mRNA decay through recruitment of DCP1 and/or DCP2.
Studying mammalian NMD, L. Maquat (Rochester, NY, USA) showed that UPF1 phosphorylation suppresses translation initiation through binding to eIF3. This interaction prevents the eIF3‐dependent formation of 80S·mRNA initiation complexes, thereby promoting the degradation of PTC‐containing mRNAs (Isken et al, 2008). She also reported that increased cellular levels of the splicing factor SF2/ASF shifted NMD of the exported mRNA from the cytoplasm to a nuclear‐associated form, and proposed that the pioneer round of translation governs the efficiency of subsequent rounds (Sato et al, 2008). In a similar vein, J. Blenis (Boston, MA, USA) described a new function of the mTOR pathway in regulating the translation of spliced mRNAs by way of the EJC. Activated mTOR substrate p70‐S6K is recruited to the EJC of newly synthesized mRNAs by SKAR during the pioneer round of translation. This leads to the enhanced translation of intron‐containing mRNAs (Ma et al, 2008). A. Jacobson (Worcester, MA, USA) reported that ∼50% of the Upf1‐associated mRNAs in yeast might not be NMD substrates (as defined by twofold increases in NMD‐deficient cells); he proposed that these mRNAs might rely on the translation‐termination function of Upf1 (Johansson et al, 2007). Consistently, depletion of Upf1 in yeast led to defective reuse of ribosomes after termination. O. Muhlemann (Bern, Switzerland) reported that extending the 3′ UTR in human cells destabilized an mRNA in a manner that was dependent on UPF1 and could be rescued by recruitment of PABP close to the termination codon (Eberle et al, 2008). These studies imply that the aberrant termination, or ‘faux UTR’, model of NMD is relevant in human cells.
Mechanisms of microRNA‐mediated gene silencing
MicroRNAs are a family of small RNAs (∼21 nucleotides in length) that regulate gene expression post‐transcriptionally in different phyla. The molecular mechanism by which miRNAs silence the expression of their target mRNAs remains controversial. Several mechanisms have been reported, namely inhibition of translation initiation, inhibition of translation elongation, ribosome drop‐off, co‐translational protein degradation and mRNA degradation.
To investigate how miRNAs repress gene expression, M. Hentze (Heidelberg, Germany) used a cell‐free translation system from Drosophila melanogaster embryos. In his keynote address, he demonstrated that miR2 represses the translation, but does not affect the mRNA stability, of a reporter mRNA containing six miR2‐binding sites. Sucrose‐gradient analysis showed that miR2 not only inhibits formation of the 80S ribosome complex, but also induces the formation of dense micro‐ribonucleoproteins, which have been called ‘pseudopolysomes’, despite the apparent absence of ribosomes. Consistent with the idea that miRNAs inhibit translation initiation through the 5′ cap (m7G) structure, an uncapped mRNA bearing an ApppG 5′‐end (in place of m7GpppG) was resistant to miR2‐mediated translational repression (Thermann & Hentze, 2007). E. Izaurralde (Tuebingen, Germany) revisited the miRNA‐mediated translational silencing mechanism and challenged current models. Interestingly, she reported that in Drosophila S2 cells, it is not the interaction between the Argonaute 1 (Ago1) protein and the mRNA cap that is essential for miRNA‐mediated translational repression or mRNA degradation, but rather the interaction between Ago1 and the P‐body component, GW182. She further showed that miRNA‐mediated repression of translation is eIF6‐independent (Eulalio et al, 2008). J. Belasco (New York, NY, USA) described the differences among the four mammalian Ago proteins in terms of their influence on translation and on the specificity of RNA interference. Additionally, P. Sarnow (Stanford, CA, USA) reported the liver‐specific positive regulation of HCV replication by miR‐122. Interestingly, depletion of certain P‐body markers resulted in a loss of HCV proteins. Furthermore, infection of liver cells with infectious HCV resulted in the loss of P bodies. It will be interesting to determine whether HCV gene expression is modulated by the release of general miRNA‐mediated translational repression, the absence of P bodies, or the lack of certain P‐body marker proteins.
By bringing together a broad range of topics, such as ribosome structure, mRNA localization and stability, and miRNA function, as well as the roles of translation in learning and memory, diabetes, obesity and virus infections, this symposium provided a comprehensive and up‐to‐date compendium of the translational control field. The meeting highlighted the considerable progress that has been made in deciphering the translational regulatory mechanisms governing a range of biological processes, and it is clear that new and exciting findings will continue to emerge as we delve deeper into the mechanism and control of protein synthesis.
We are grateful to J. Richter, A. Hinnebusch and N. Sonenberg for organizing this highly stimulating meeting, and we apologize to those whose data could not be discussed here owing to space limitations.
See Glossary for abbreviations used in this article.
- asymmetric synthesis of HO1
- activating transcription factor 4
- CCAAT/enhancer‐binding protein
- decapping protein
- eukaryotic initiation factor
- eukaryotic release factor
- growth arrest and DNA‐damage‐inducible protein 34
- general control non‐derepressible
- glycine‐tryptophan protein of 182 KDa
- hepatitis C virus
- herpes simplex virus 1
- interferon‐regulatory factor 7
- metastatic lymph node 51
- mammalian target of rapamycin
- poly(A)‐binding protein
- protein kinase R‐like endoplasmic reticulum kinase
- protein kinase R
- phosphatase 1
- S6 kinase 1
- splicing factor 2
- S6K Aly/Ref‐like target
- major capsid protein of calicivirus
- zipcode‐binding protein 1
- Copyright © 2008 European Molecular Biology Organization