Insulin rapidly activates protein synthesis by activating components of the translational machinery including eIFs (eukaryotic initiation factors) and eEFs (eukaryotic elongation factors). In the long term, insulin also increases the cellular content of ribosomes to augment the capacity for protein synthesis. The rapid activation of protein synthesis by insulin is mediated primarily through phosphoinositide 3-kinase. This involves the activation of PKB (protein kinase B). In one case, PKB acts to phosphorylate and inactivate glycogen synthase kinase 3, which in turn phosphorylates and inhibits eIF2B. Insulin elicits the dephosphorylation and activation of eIF2B. Since eIF2B is required for recycling of eIF2, a factor required for all cytoplasmic translation initiation events, this will contribute to overall activation of protein synthesis. PKB also phosphorylates the TSC1 (tuberous sclerosis complex 1)–TSC2 complex to relieve its inhibitory action on the mTOR (mammalian target of rapamycin). Inhibition of mTOR by rapamycin markedly impairs insulin-activated protein synthesis. mTOR controls translation initiation and elongation. The cap-binding factor eIF4E can be sequestered in inactive complexes by 4E-BP1 (eIF4E-binding protein 1). Insulin elicits phosphorylation of 4E-BP1 and its release from eIF4E, allowing eIF4E to form initiation factor complexes. Insulin induces dephosphorylation and activation of eEF2 to accelerate elongation. Both effects are blocked by rapamycin. Insulin inactivates eEF2 kinase by increasing its phosphorylation at several mTOR-regulated sites. Insulin also stimulates synthesis of ribosomal proteins by promoting recruitment of their mRNAs into polyribosomes. This is inhibited by rapamycin. Several key questions remain about, for example, the mechanisms by which mTOR controls 4E-BP1 and eEF2 kinase and the control of ribosomal protein translation.
Insulin can stimulate protein synthesis in many types of cells and tissues, including human and rat muscles, and many types of cells in culture [1,2]. This involves two major kinds of effects: the rapid activation of existing components of the translational apparatus and the longer term increase in the capacity of the cell or tissue for protein synthesis, which includes an increase in ribosome number. Studies in this laboratory have focused primarily on the short-term regulation of proteins involved in controlling mRNA translation. These are termed translation factors. They are extrinsic to the ribosome but may bind transiently to it to promote individual steps of the mRNA translation process.
Protein synthesis (mRNA translation) is conventionally divided into three stages: initiation, elongation and termination. Both initiation and elongation can be controlled by insulin. Initiation is the stage where the mRNA and the initiator tRNAmMet (methionine tRNA) are recruited to the ribosome, initially to the small 40 S subunit. The start codon is then located by a process known as ‘scanning’ during which it is recognized by the anticodon of tRNAmMet. This culminates in assembly of the complete 80 S ribosome at the start codon, ready to proceed into elongation where the new polypeptide is assembled.
Insulin activates eIF2B (eukaryotic initiation factor 2)
tRNAmMet is brought to the ribosome by eIF2, a heterotrimeric GTP-binding protein. The GTP is hydrolysed upon successful engagement of tRNAmMet with the start codon. eIF2–GDP is recycled back to the active eIF2–GTP complex by eIF2B (Figure 1). This GDP/GTP exchange process is a key regulatory step for translation initiation under many conditions: for example, it is inhibited by phosphorylation of eIF2's α-subunit by any one of four mammalian eIF2 kinases, which are activated under specific stress conditions . The resulting inhibition of eIF2B impairs the recycling of eIF2 and thus inhibits overall translation initiation.
Regulation by insulin of eIF2B and recruitment of tRNAmMet to the 40 S subunit
Conversely, eIF2B is activated by insulin . This involves a change in the phosphorylation of its largest ϵ-subunit, which contains the catalytic domain. We discovered that insulin inactivated a kinase that phosphorylates eIF2Bϵ, which was subsequently identified as GSK3 (glycogen synthase kinase 3) [4,5]. GSK3 was already implicated in the control of another major anabolic process (glycogen deposition) by insulin, but little was known about its regulation. Phosphorylation of eIF2B by GSK3 occurs at a single serine residue, immediately N-terminal to the catalytic region, and this impairs its GDP/GTP exchange activity . The insulin-induced inactivation of GSK3 leads to the dephosphorylation of eIF2B, contributing to its activation. Subsequent studies in other laboratories revealed that insulin switches off GSK3 via signalling through PI3K (phosphoinositide 3-kinase), via PKB (protein kinase B or Akt), which phosphorylates both isoforms of GSK3 directly .
These studies delineated a complete signalling pathway from the insulin receptor via PKB to the activation of eIF2B. It serves to promote the activation of at least two major anabolic processes, glycogen synthesis and mRNA translation. Since eIF2B is required for all translation initiation events, this mechanism promotes overall protein synthesis.
Insulin controls the cap-binding complex via mTOR (mammalian target of rapamycin) signalling
Recruitment of the mRNA to the ribosome involves a set of protein factors. eIF4E binds the 5′-cap structure of the mRNA and interacts with a large multidomain protein, eIF4G, which can bind a number of other translation factors as well as the PABP [poly(A)-binding protein] and the eIF4E kinases, the Mnks . One of the factors that binds to eIF4G is eIF3, a multisubunit protein that interacts with the 40 S subunit.
eIF4G binds to eIF4E through a short motif that is also found in other binding partners for eIF4E. These include the 4E-BPs (eIF4E-binding proteins), which are small heat- and acid-stable phosphoproteins. The best understood of the three 4E-BPs in mammals, 4E-BP1, was first described by Denton and colleagues in the early 1980s as ‘22K’ , a protein whose phosphorylation was enhanced by insulin in a PI3K-dependent manner . As PHAS-I (phosphorylated heat- and acid-stable protein regulated by insulin), it was cloned by Lawrence and colleagues  and was shown to be a 4E-BP by Sonenberg and co-workers . By binding to eIF4E, 4E-BPs prevent eIF4E from interacting with eIF4G and forming productive initiation complexes. 4E-BPs thus impair the recruitment of ribosomes to normal cap-dependent mRNAs.
4E-BP1 undergoes phosphorylation at several sites, and there appears to be a complex interplay between them . Overall, the phosphorylation of 4E-BP1 leads to its release from eIF4E, allowing eIF4E to bind eIF4G and form initiation complexes. Among the protein partners for eIF4G is eIF4A, an RNA helicase, which together with eIF4G and eIF4E forms the eIF4F . eIF4A is thought to unwind regions of complementary secondary structure in the 5′-UTR (5′-untranslated region) which otherwise impede scanning. Formation of eIF4F complexes may be especially important for efficient translation of those mRNAs that have substantial 5′-secondary structure.
The phosphorylation of several sites in 4E–BP1 is controlled via the mTOR. mTOR is a protein kinase whose catalytic domain resembles lipid kinases such as PI3K. Some functions of mTOR are blocked by rapamycin, which also impairs the phosphorylation of 4E-BP1 and blocks its release from eIF4E.
mTOR signalling is activated by hormones such as insulin and by amino acids, notably leucine (reviewed in [13,14]) (Figure 2). Our recent results show that amino acids and insulin regulate different sets of phosphorylation sites in 4E-BP1 and that the insulin-modulated ones are sensitive to rapamycin, but the amino acid-modulated ones are largely insensitive . This implies that amino acids and insulin modulate downstream mTOR signalling in distinct ways. It remains unclear, in particular, how mTOR exerts its rapamycin-insensitive, but amino-acid-dependent, effects on 4E-BP1. Phosphorylation of the N-terminal sites in 4E-BP1 (Thr37/Thr46 in the human protein) depends on a short N-terminal motif RAIP (Arg-Ala-Ile-Pro) [12,15]. It is these sites that are regulated by amino acids in a rapamycin-insensitive manner. The sites immediately C-terminal to the eIF4E-binding site (Ser65, Thr70) are sensitive to rapamycin, are stimulated by insulin, and depend on prior phosphorylation of the N-terminal sites and on a second motif at the extreme C-terminus. This TOS (target of rapamycin signalling) motif (but not the RAIP feature) binds the mTOR partner raptor (see e.g. [16,17]), and may recruit mTOR–raptor complexes to 4E-BP1 to phosphorylate it at certain sites.
mTOR signalling controls several components of the translational machinery
Upstream control of mTOR by insulin
Insulin regulates mTOR signalling through the TSC1 (tuberous sclerosis complex 1)–TSC2 protein complex: this normally acts as a brake on mTOR signalling. Mutations in TSC1 or TSC2 that impair their function or expression therefore lead to hyperactivation of mTOR and this causes excessive cell growth, giving rise to the large cell tumours that are seen in patients with such mutations. The resulting condition is termed ‘tuberous sclerosis’. TSC1–TSC2 acts as a GAP (GTPase-activating protein) complex for the small G-protein Rheb, which in turn binds to, and in its GTP-liganded form activates, mTOR . TSC2 is phosphorylated by PKB in response to insulin and this appears to impair its GAP activity, although it is not known how [19,20]. Amino acids regulate mTOR by a different mechanism, which is independent of TSC1–TSC2  (Figure 2). A key question is therefore how do amino acids turn on mTOR?
In addition to 4E-BP1, other proteins involved in the translational machinery are also controlled by mTOR. These include the S6Ks (S6 kinases), which phosphorylate multiple sites in the C-terminus of the 40 S ribosomal protein S6  (Figure 2). The S6Ks are activated by insulin and other agents that turn on protein synthesis, and in many types of cells require that the cellular medium contains amino acids [13,23]. The S6Ks were thought to regulate the translation of the set of mRNAs in mammalian cells that encode the ribosomal proteins: these contain a 5′-TOP (5′-terminal oligopyrimidine tract) which impairs their translation under serum-starved conditions. Serum stimulation enhances their translation in a rapamycin-sensitive manner. However, recent results from animals in which both S6K genes have been inactivated reveal that the S6Ks are not required for the normal control of 5′-TOP mRNA translation. Thus additional and so far unknown targets for mTOR signalling must be involved in modulating 5′-TOP mRNA translation (Figure 2). The ability of agents such as insulin to activate ribosomal protein biosynthesis probably contributes to the long-term effects of insulin on cellular ribosome content referred to at the beginning of the present paper. S6Ks actually play a role in the down-regulation of insulin signalling, by phosphorylating IRS1 (insulin receptor substrate 1) [24,25].
Translation elongation in mammalian cells requires two eEFs (eukaryotic elongation factors), eEF1 and eEF2. eEF2 mediates the translocation of the ribosome by three nucleotides along the mRNA after addition of each new amino acid. Phosphorylation of eEF2 on Thr56 impairs its ability to bind ribosomes, thus inactivating it . Insulin induces the rapid dephosphorylation of eEF2, leading to accelerated rates of elongation as measured by transit-time measurements . Both effects are blocked by rapamycin, implying a key role for mTOR. Insulin also causes a decrease in the activity of eEF2 kinase, the specific Ca2+/CaM (calmodulin)-dependent enzyme that phosphorylates eEF2. This effect is also prevented by rapamycin. This raises the issue of how mTOR regulates eEF2 kinase.
eEF2 kinase is a phosphoprotein and mTOR signalling affects at least three different sites in eEF2 kinase. We showed that Ser366 is a target for S6K1 and that phosphorylation here inhibits the activity of eEF2 kinase, especially at low [Ca2+] . Phosphorylation of Ser359 is also regulated in an mTOR-dependent manner  and phosphorylation of this site also impairs eEF2 kinase activity. Most recently, we identified Ser78 as a novel mTOR-dependent phosphorylation site in eEF2 kinase . It lies immediately adjacent to the CaM-binding domain of eEF2 kinase and its phosphorylation impairs CaM binding, thus decreasing eEF2 kinase activity. The identities of the protein kinases acting at the latter two mTOR-modulated sites are so far unknown. The amino acid input to mTOR signalling makes excellent sense in terms of the control of elongation: when amino acid supply is limited, mTOR signalling is impaired and phosphorylation of these inhibitory sites in eEF2 kinase will be decreased. This will result in increased eEF2 kinase activity and eEF2 phosphorylation, thereby slowing the rate of elongation and the consumption of amino acids. Conversely, when amino acid supply improves, eEF2 can be re-activated and elongation can accelerate.
Insulin thus activates multiple steps in mRNA translation, including (i) tRNAmMet and mRNA recruitment to the ribosome, (ii) the translation of specific mRNAs (the 5′-TOP mRNAs) and (iii) the elongation process. The relative quantitative contributions that each effect makes to the overall short- and long-term stimulation of protein synthesis remain to be established. The effects of insulin and other stimuli on protein synthesis rates in immortalized cell lines are often rather modest compared with the much larger effects seen in primary cells, e.g. cardiomyocytes . Primary cells may therefore be a more useful system in which to address these questions, as well as probably yielding data that are more physiologically relevant.
Insulin, Calcium and the Control of Mammalian Metabolism: Focused Meeting to honour the retirement of Professor Dick Denton FRS, held at Willis Hall, University of Bristol, U.K., 22–23 September 2005. Organized and edited by A. Halestrap, G. Rutter and J.M. Tavaré (Bristol, U.K.).
eukaryotic elongation factor
eukaryotic initiation factor
eIF4E-binding protein 1
glycogen synthase kinase 3
mammalian target of rapamycin
protein kinase B
5′-terminal oligopyrimidine tract
tuberous sclerosis complex
Work on insulin control of mRNA translation in my laboratory has been funded by the Medical Research Council (U.K.), the Biotechnology and Biological Sciences Research Council (U.K.), the British Heart Foundation and the Wellcome Trust. It is currently supported by the Heart and Stroke Foundation of Canada and the University of British Columbia.