A. Abbruzzese, Dipartimento di Biochimica e Biofisica, Seconda Università di Napoli, Via Costantinopoli, 16 80138 Naples, Italy. Fax: + 39 81 5665863, E-mail: Alberto.Abbruzzese@unina2.it
In the past years, the attention of scientists has focused mainly on the study of the genetic information and alterations that regulate eukaryotic cell proliferation and that lead to neoplastic transformation. All therapeutic strategies against cancer are, to date, directed at DNA either with cytotoxic drugs or gene therapy. Little or no interest has been aroused by protein synthesis mechanisms. However, an increasing body of data is emerging about the involvement of translational processes and factors in control of cell proliferation, indicating that protein synthesis can be an additional target for anticancer strategies. In this paper we review the novel insights on the biochemical and molecular events leading to protein biosynthesis and we describe their involvement in cell proliferation and tumorigenesis. A possible mechanistic explanation is given by the interactions that occur between protein synthesis machinery and the proliferative signal transduction pathways and that are therefore suitable targets for indirect modulation of protein synthesis. We briefly describe the molecular tools used to block protein synthesis and the attempts made at increasing their efficacy. Finally, we propose a new multimodal strategy against cancer based on the simultaneous intervention on protein synthesis and signal transduction.
In the past, protein synthesis has not been considered to be fundamental in the control of cell proliferation. However, data are emerging on the involvement of this process in cell growth and tumorigenesis. Protein biosynthesis is a central process in all living cells. It is one of the last steps in the transmission of genetic information stored in DNA on the basis of which proteins are produced to maintain the specific biological function of a given cell. Protein synthesis takes place on ribosomal particles where the genetic information transcribed into mRNA is translated into protein. The process of protein synthesis on the ribosome consists of three phases: initiation, elongation and termination.
The initiation phase
In prokaryotic protein biosynthesis the initiation phase is controlled by a small number of IFs (IFs), IF-1, IF-2 and IF-3. Among these, IF-2 is the most likely to form a ternary complex with GTP and initiator . All three factors are involved in assembling the initiation complex of IFs, initiator , the ribosomal subunits and mRNA. Eukaryotic initiation is much more complex and involves a large number of eukaryotic IFs [1–3]. The 80 S ribosomes dissociate, and 40 S subunits are captured for initiation by binding eIF-1A and eIF-3; the size of the latter causes the particle to sediment at 43S. Initiator tRNA binds, in the form of a ternary complex with eIF-2 and GTP, to produce the 43S preinitiation complex [4,5]. The 43 S preinitiation complex binds to mRNA at the 5′ terminal mGTP cap structure, and then migrates along the mRNA towards the AUG initiation codon [6,7]. The initial binding involves the factors eIF-4E, eIF-4G and eIF-4A, which assemble at the 5′-end of mRNA, thus creating the conditions that allow the melting of intramolecular secondary structures within the mRNA that would otherwise prevent the binding of the 43 S preinitiation complex [1,3,8–10]. The term 48 S preinitiation complex is frequently used, and refers to the 43S–globin–mRNA complex formed in the reticulocyte lysate . When the 43S preinitiation complex stops at the initiation codon, the GTP molecule introduced as part of the eIF-2 complex is hydrolysed to GDP, and this gives energy for the ejection of the IFs bound to the 40 S ribosomal subunit . The IF eIF-5 is involved in this process which is likely to accelerate the hydrolysis of GTP . The release of these factors allows the association of a native 60 S ribosomal subunit, to reconstitute an 80 S ribosome at the initiation codon positioned to commence the elongation stage of translation. The continuity of initiation events requires the recycling of IF molecules. eIF-2 is released as a binary complex with GDP and requires a guanine nucleotide exchange factor, eIF-2B, to catalyse the regeneration of the eIF-2*GTP complex [12,13].
The elongation phase
The elongation phase has been the most extensively studied phase in prokaryotes, both functionally and structurally, over the last 30 years . This phase is controlled by three elongation factors (EFs). EF-Tu (EF-1 in eukaryotes) forms a ternary complex with GTP and amino-acylated tRNAs, it protects the amino-acid ester bond against hydrolysis and carries the amino-acylated tRNA to the ribosomal A-site for the decoding of mRNA by codon–anticodon interactions . When correct codon–anticodon recognition occurs, GTP hydrolysis on EF-Tu is stimulated by the ribosome and EF-Tu:GDP is released [16,17]. However, when compared with other G-proteins EF-Tu has a very low intrinsic GTPase . It works as a time-delayed molecular switch, and the GTPase activity is thus highly stimulated when the ternary complex interacts with ribosome. The nucleotide exchange factor EF-Ts (EF-1 in eukaryotes) converts EF-Tu:GDP into active EF-Tu:GTP. After a proofreading step, the amino-acylated-tRNA is brought into contact with the peptidyl-tRNA in the ribosomal P-site, where peptide bond formation is catalysed adding one amino acid to the growing peptide . The last EF, EF-G (EF-2 in eukaryotes), together with GTP controls the translocation of tRNAs and mRNA on the ribosome . It has been demonstrated that the EF-G*GTP complex mimics the ternary complex of EF-Tu*GTP with tRNA . Therefore, it is likely to make an ‘imprint’ on the ribosome for the ternary complex [21,22]. In this way the site it leaves will become fit for the ternary complex. Translocation will then take place by a conformational change in EF-G so that its ‘tRNA mimicking part’ will occupy the anticodon region of the A-site . Recent results  suggest that this conformational change and translocation are subsequent to the GTP hydrolysis. Another interesting finding is that the main contacts between the ribosome and EF-Tu are between domain 1 and the 50 S subunit and between domain 2 and the 30 S subunit. This is interesting, not only because these two domains are common between EF-Tu and EF-G, but also because similar domains are present in all the translational G-proteins . The implication is that all of them will bind to the ribosome in the same way, and that the GTPase activity will be stimulated by the same mechanism. The latter could explain for the late onset of GTPase activity in the translational G-proteins.
The termination phase
Termination of protein synthesis occurs on the ribosomes and requires the presence of a termination (stop) codon in the mRNA at the ribosomal A-site and the polypeptide chain release factor (RF) of class 1 (reviewed in [25–30]). Termination is controlled by RFs which have been extensively studied in recent years [29–31]. In prokaryotes, the stop codons are specifically recognized by RF-1 (that recognizes the UAG and UAA stop signals) and RF-2 (that recognizes the UGA and UAA stop signals), while a third non codon-specific factor, RF-3, in complex with GTP stimulates the activities of the other two factors and the release of the fully synthesized protein from the ribosome . Finally, RF-4 (ribosome release or recycling factor) displaces peptidyl-tRNA from the ribosome in a way related to its proposed function in removing deacylated tRNA during ribosome recycling . The termination reaction leads to hydrolysis of the peptidyl-tRNA, which generates a free nascent polypeptide chain and a free (uncharged) tRNA followed by the release of these components from the ribosome. In eukaryotes, translational termination is governed by a single RF which recognizes all three termination codons . When the structure of this factor had been established it was termed eRF-1 . In contrast with earlier observations [36,37] where the termination reaction was accompanied by concomitant GTP hydrolysis, eRF-1 appeared to be able to catalyse in vitro peptidyl-tRNA hydrolysis at a high termination codon concentration in the absence of GTP and other factors [35–39]. This apparent controversy was resolved when a second RF, termed eRF-3, was identified and shown to be an eRF-1-dependent and ribosome-dependent GTPase [38,39]. It was shown that eRF-1 and eRF-3 bound to each other both in vitro and in vivo[39–42]. eRF-1*eRF-3 complex formation and eRF-3 GTPase activity were strongly interrelated as eRF-3 GTPase activity depended on the complex formation. eRF-3 alone, even capable of binding GTP, is virtually inactive in catalysing GTP hydrolysis on the ribosome . Recently, the mammalian homolog of eRF-3, the GTP-binding protein GSPT (for G1 to S phase transition), has been cloned [43,44]. Its C-terminal binds EF-1A while its N-terminal interacts with the polyadenilate binding protein (PABP) . GSPT inhibits the multimerization of PABP with poly(A) indicating that GSPT/eRF-3 may play an additional role in the regulation of mRNA stability .
The overall process of translation termination highly resembles the elongation process except for the fact that a stop codon, instead of a sense codon, is decoded at the A-site of the ribosome . A molecular mimicry has been found between the EF-Tu ternary complex and the RF-1/2-RF-3-GTP complex . The ribosome plays an active part during the translation of both sense and stop codons . The ribosomal A-site decoding centre at the bottom of the intersubunit cavity is opened upon starvation for amino-acylated-tRNA . Ribosomes that reach a stop codon probably undergo a similar conformational change to expose the A-site with the stop codon for RF interaction. Termination/readthrough at a stop codon is sensitive to the flanking codons . The nonsense codon context effect suggests that the RF complex or the competing EF-Tu ternary complex with the near-cognate tRNA, or both, are sensitive to the neighbouring codon(s) either directly or indirectly .
Moreover, the mRNA sequence immediately downstream of the stop codon is important for termination efficiency. Conversely, an A as the first base downstream of the stop codon produces inefficient termination. Thus, termination is ruled by a signal of four bases instead of the three that constitute the termination codon itself . In addition to the P-site tRNA that seems to interact with RF, the corresponding last amino acid in the nascent peptide also influences termination efficiency. Thus, a correlation can be found between a high propensity of this amino acid to participate in ordered structures (αL helix, β strand) and efficient termination at UGAA .
Recently, ribosome recycling factor or RF-4 has been described in prokaryotes. It unassembles the post-termination complex formed by ribosomes, mRNA and tRNA [34,50–52]. The biochemical processes of protein synthesis are described in Fig. 1.
Protein synthesis and cell proliferation
There is presently a growing body of evidence which suggests the involvement of protein synthesis and translational factors in the regulation of cell proliferation and transformation.
The regulation and overexpression of polypeptide chain IFs for cell transformation and malignancy
Over the past few years, studies aimed at understanding the regulation and function of protein synthesis IFs. Initially, study of eIF-4E culminated in the unexpected finding that a moderate overexpression of this factor results in dramatic phenotypic changes, including rapid proliferation and malignant transformation [54–59]. Conversely, the tumorigenic properties of cancer cells can be strongly inhibited by antisense RNA against IF, or overexpression of the inhibitory proteins, such as 4E-BPs also called phosphorylated heat and acid stable proteins I–III (PHAS I–III) [54–60].
When the eIF-4E is overexpressed in mammalian cells in vitro oncogenic transformation occurs [54–56,58,59]. Furthermore, it has been reported that c-myc expression stimulates cell growth by increasing expression of eIF-4E  and overexpression of eIF-4G also caused malignant transformation in NIH3T3 cells .
The involvement of eIF-4E in carcinogenesis has been demonstrated also in vivo. In fact, eIF-4E is overexpressed not only in all head and neck squamous cell cancers but also in some dysplastic lesions [54,62,63]. Moreover, strong expression of eIF-4E and 2α was found in the germinal centres of reactive follicles of non-Hodgkin's lymphomas  and in active growing lymphocytes . eIF-4E expression is also high in colon tumours, and this increase is paralleled, in most but not all cases, by an increase in cyclin D1 levels . These findings suggest that eIF-4E may represent a downstream target of the APC/β-catenin/Tcf-4 pathway, and it is likely to be involved in colon tumorigenesis . Besides IF overexpression, another mechanism capable of inducing cell transformation is the disregulation of IF activity. In fact, mitogenic stimulation of protein synthesis is accompanied by an increase in eIF-4E phosphorylation that enhances its activity . On the other hand, the effects induced on protein synthesis by differentiation are less known. When P19 embryonal carcinoma cells are treated with the differentiating agent retinoic acid, protein synthesis increased during the first hour . However, the phosphorylation state, as well as the turnover of phosphate on eIF-4E, remained unchanged. Increased phosphorylation of the translational repressor protein 4E-BP1 was found in a Rous sarcoma virus-transformed cell line . This was accompanied by its dissociation from the complex with IF eIF-4E . More data seem to confirm the involvement of eIF-4E in the tumour invasiveness and angiogenic process. In fact, indirect evidence exists about the interaction between eIF-4E and the ras-dependent transduction pathway as ras-transformed rat fibroblast cells with reduced eIF-4E had delayed and reduced invasiveness and decreased experimental metastasis . Reduced eIF-4E levels correlates with decreased expression of the metastasis-associated 92-kDa collagenase type IV and of the CD44 adhesion molecule and with decreased levels of the putative metastasis-suppressor protein nm23 . Moreover, eIF-4E facilitates the synthesis of powerful tumour angiogenic factors by selectively enhancing their translation . In vivo evidence has demonstrated a marked increase in eIF-4E in breast invasive ductal carcinoma and in islets of viable cells in the centre of poorly vascularized ductal carcinoma in situ. Expression of eIF-4E was increased by hypoxia and, presumably, in hypoxic areas of these lesions. Clonal expansion of cancer cells, permanently over-expressing eIF-4E, is likely to give them a critical advantage to survive after hypoxia and marks the transition to the vascular phase of cancer progression . In the last years other IFs have been correlated with tumorigenesis, such as eIF-2α and eIF-3-p40 [57,64,70,71]. Moreover, their involvement in tumour cell survival and protection from cytotoxic agents has been shown. In fact, eIF-4G is specifically degraded by caspases during apoptosis in lymphoma cell lines, suggesting its involvement in the protection of tumour cells from programmed cell death . eIF-2α phosphorylation is induced in growth arrested cells by nitric oxide suggesting that nonspecific inhibition of protein synthesis may be a generalized response of cells exposed to high levels of nitric oxide and that inhibition of protein synthesis may contribute to many of the described cytostatic actions of nitric oxide .
The actions of the eIF-2 α-specific protein kinase PKR in the control of cell growth and apoptosis
A translational inhibitor was initially isolated from unstimulated Friend leukaemia cells. The inhibitor is a heat-labile, sulfhydryl reagent-insensitive protein . It inhibits protein synthesis in the peptide chain initiation phase by preventing IF-dependent binding of methionyl-tRNA to 40 S ribosomal subunits . The inhibitor contains protein kinase activity which phosphorylates the eIF-2α on Ser51 and inactivates it [74,75]. Later it became clear that eIF-2α kinase is the double-stranded RNA-dependent protein kinase (dsRNA-PK) thought to be a key mediator of the antiviral and antiproliferative effects of interferons (IFNs) and called INF-induced double-stranded RNA-activated kinase (PKR) [75,76]. PKR is a 68-kDa protein in human cells with serine/threonine kinase activity . Thr446 and Thr451 in the PKR activation loop are required in vivo and in vitro for high level kinase activity . In fact, mutation of either residue to Ala impaired translational control by PKR or autophosphorylation and eIF-2α phosphorylation by PKR in vitro. Studies of the physiological function of the kinase suggest that it participates in cell growth and differentiation by regulating protein synthesis. Further data suggest that PKR has additional substrates, and that the kinase may also regulate gene transcription and signal transduction pathways . In fact, PKR may have a tumour suppressor function under normal conditions [78,79]. Expression of a functionally defective mutant of human PKR in NIH3T3 cells resulted in malignant transformation, suggesting that PKR may function as a suppressor of cell proliferation and tumorigenesis through phosphorylation of eIF-2α. NIH3T3 cells transformed by a dominant negative PKR mutant have a dramatically reduced eIF-2α phosphorylation. Furthermore, expression of a mutant form of eIF-2α, which cannot be phosphorylated on Ser51 also caused malignant transformation of NIH3T3 cells . Moreover, studies of human malignancies and tumour cell lines suggest that, in general, patients bearing tumours with a higher PKR content have a more favourable prognosis . In human breast carcinoma cells, dysregulation of PKR may be associated with the establishment or maintenance of the transformed state . In fact, breast tumour cell lines express higher levels of active PKR as compared with nontransformed breast cells suggesting that cancer cell extracts contain a transdominant inhibitor of PKR . A cellular inhibitor of PKR (referred to as p58) has been purified and characterized recently [80,81]. Subsequent cloning and sequencing have revealed that p58 is a member of the tetratricopeptide repeat family of proteins . The stable transfection of p58 in NIH3T3 cells resulted in a transformed phenotype and reduced murine PKR activity and endogenous levels of eIF-2α phosphorylation. Therefore, p58 functions as an oncogene through the downregulation of PKR and subsequent deregulation of protein synthesis even if its expression in tumour cell lines is still to be investigated. The central role of eIF-2α phosphorylation in the regulation of protein synthesis and cell proliferation is proved by the discovery of the phosphorylative loop that involves the cellular glycoprotein p67 . This protein protects eIF-2α from phosphorylation and is, in turn, regulated by an enzyme (p67 deglycosylase) that, when active, removes the carbohydrate moieties from p67, inactivates it and, subsequently, protein synthesis is inhibited. It has recently been found that the phosphorylation of eIF-2α increases at the G2/M phase of cell cycle, and p67 is deglycosylated at the same time. In addition, the level and the activity of p67 deglycosylase also increase at the G2/M boundary of cycling U2-OS cells .
EIF-5A and hypusine
Eukaryotic translation IF-5A (eIF-5A; 18 kDa) is highly conserved from yeast to mammalian cells [83,84]. The human cDNA that encodes the eIF-5A precursor has been cloned and sequenced . eIF-5A precursor [(ec-eIF-5A(lys)] is the only cellular protein known to contain a specific lysine residue which is transformed into the unique amino acid hypusine [Nε-(4-amino-2-hydroxybutyl)-lysine]. This amino acid is formed by a series of post-translational reactions starting with the transfer of the butylamine moiety from spermidine to the ε-amino group of one of the lysine residues in the eIF-5A precursor protein, thus forming peptide-bound deoxyhypusine . This intermediate is not accumulated in cells but it is immediately hydroxylated at C-2 of the incoming 4-aminobutyl moiety to form hypusine [87–89]. Hypusine plays a key role in the regulation of eIF-5A function because eIF-5A precursors, which do not contain hypusine, have little, if any, activity . Moreover, the Lys50→Arg variant is unable to stimulate methionyl-puromycin synthesis in vitro[90,91] and is inactive in vivo. These data suggest that hypusine synthesis is required for the biological activity of the protein and for interaction with the ribosome.
More recently it has been found that eIF-5A undergoes a post-translational modification catalysed by transglutaminase .
eIF-5A promotes the formation of the first peptide bond during the initial stage of protein synthesis . The actual in vivo function of eIF-5A, however, is still only partially known. A series of observations suggests that eIF-5A plays a role in cell growth and differentiation. In fact, the ec-eIF-5A(lys) modification is correlated with cell proliferation [93–96] and is vital for Saccharomyces cerevisiae growth . Moreover, agents that block the lys/hyp transformation [84,98–100] inhibit the growth of mammalian cells  inducing reversible arrest at the G1/S boundary of the cell cycle [102–104]. In fact, d,l-α-difluoromethylornithine depresses spermidine and, as a consequence, hypusine formation and it produces a G1/S block in 9L brain tumour cells . In hydralazine-treated cloned mouse T cells, growth arrest in late G1 and inhibition of deoxyhypusyl hydroxylation occur at the same time. In Chinese hamster ovary cells, 2-(4-hydroxy-toluene-3-yl)-4,5dihydro-5-carboxythiazole causes both inhibition of hypusine biosynthesis and reversible cell cycle arrest in late G1 and suppresses proliferation of human T lymphocytes in vitro at the G1/S boundary . Hypusine synthesis increases after mitogen treatment of human peripheral blood lymphocytes . Moreover, both polyamine and hypusine levels (together with the enzymes that regulate their metabolism) [88,89,95,97,105] are correlated to normal and malignant growth. More recently a correlation has been found between the polyamine-dependent modification of eIF-5A and the triggering of apoptosis in tumour cells . In fact, excess putrescine accumulation in hepatoma tissue culture DH23A/b cells induces apoptosis and suppresses the formation of hypusine-containing eIF-5A . Moreover, DH23A cells overexpressing ornithine decarboxylase accumulate putrescine with a consequent increase in apoptosis . The latter effect induces an inhibition of the post-translational modification of eIF-5A that is also blocked by diaminohepatne, another strong inducer of apoptosis in DH23A cells . eIF-5A is found in the cytoplasm in two pools: one free and another bound to endoplasmic reticulum (probably its proper site of action) . However, it is also been recently reported that eIF-5A can accumulate at nuclear pore-associated intranuclear filaments in mammalian cells . Moreover, the factor interacts with the general nuclear export receptor CRM1 and is transported from the nucleus to the cytoplasm . These findings open a new scenario in which eIF-5A may also function as nucleocytoplasmic shuttle protein of mRNAs eventually correlated with cell proliferation.
The involvement of the EF-1 family in oncogenesis
Wide evidence suggests the involvement of the EFs at the onset of oncogenesis. Altered expression of translation EF-1α, a core component of protein synthesis, and closely related sequences have been linked to transformed phenotypes in several independent studies in diverse systems [110,111]. EF-1α is also a microtubule-severing protein  whose up-regulation by p53 may be a cause of cell death. A dominant acting oncogene, prostate tumour-inducing gene 1 (PTI-1) has provided further evidence of the link between EF-1α and cell transformation. PTI-1 appears to be a hybrid molecule with components derived from both prokaryotic and eukaryotic origins. The predicted protein coding moiety represents an EF-1α molecule, with a truncated N-terminal to amino-acid residue 68 and six additional point mutations. This coding sequence is linked to a 5′ untranslated region (UTR) that shows a very strong homology to ribosomal RNA derived from Mycoplasma hyopneumoniae. Expression studies have confirmed the oncogenic nature of the molecule. A broad spectrum of tumour-derived cell lines, from varied tissue sources and blood samples from patients with prostate carcinoma, were positive for expression of PTI-1, while corresponding normal tissues or blood samples were negative. Moreover, blocking PTI-1 expression with antisense PTI-1 results in reversion of transformed phenotype in PTI-1-expressing cells. Therefore, it has been proposed that PTI-1 represents a new class of oncogene whose transforming capacity probably arises through mechanisms including: (a) protein translational infidelity, resulting in the synthesis of mutant polypeptides due to the loss of proofreading function during peptide chain elongation; (b) its association with and alteration of the cytoskeleton; (c) impinging on one particular or several different signal transduction pathways through its properties as a G protein . Furthermore, overexpression of EF-1α mRNA has been correlated with increased metastatic potential in mammary adenocarcinoma probably due to its interaction with the actin cytoskeleton, an effector in metastasis . It has been shown that EF-1α protein is overexpressed in metastatic compared with nonmetastatic cells and whole tumours. Similarly to other EF-1α, both types of tumour EF-1α bind to F-actin, but EF-1α from metastatic cells has a reduced affinity for actin. Following stimulation with epidermal growth factor (EGF), there is a parallel increase in the amount of F-actin and EF-1α associated with the cytoskeleton. Therefore, it has been proposed that a weakened association of EF-1α with actin may be related to the metastatic process via an altered organization of the actin cytoskeleton and the differential translation of mRNAs associated with the cytoskeleton . Also EF-1γ is often overexpressed in gastrointestinal cancers, and a cDNA clone, covering part of the C-terminal domain of human EF-1δ, has been isolated from mammary cancer cells by subtractive hybridization [114–116].
The Ca2+/Calmodulin-EF-2 loop and the tumorigenesis
eEF-2 is the target for a very specific Ca2+/calmodulin-dependent eEF-2 kinase . Phosphorylation of eEF-2 makes it inactive in translation, which suggests that protein synthesis can be regulated by Ca2+ through eEF-2 phosphorylation . Recent data demonstrate that eEF-2 phosphorylation can be involved in cell cycle regulation and other processes where changes of intracellular Ca2+ concentration induce a new physiological state of a cell. The main role of eEF-2 phosphorylation in these processes is the temporary inhibition of the overall translation in response to a transient increase in the Ca2+ concentrations in the cytoplasm .
Calmodulin-dependent protein kinase III (CaM kinase III, EF-2 kinase) is a unique member of the Ca2+/CaM-dependent protein kinase family . Recent cloning and sequencing of CaM kinase III revealed that this enzyme represents a new superfamily of protein kinases . The activity of CaM kinase III is selectively activated in proliferating cells, while the inhibition of the kinase blocked cells in G0/G1–S and decreased viability [119–122].
The specific activity of CaM kinase III in human breast cancer cell lines was equal to or greater than the one seen in a variety of cell lines with similar rates of proliferation [119–122]. The specific activity of CaM kinase III was markedly increased in human breast tumour specimens compared with that of normal adjacent breast tissue . Moreover, the combination of insulin-like growth factor I and EGF stimulated cell proliferation and activated CaM kinase III and the inhibition of the enzyme activity blocked cell proliferation induced by growth factors . Moreover, CaM kinase III was increased in S-phase more than in other phases of the cell cycle. The activity of Ca2+/CaM-dependent protein kinase III is controlled by breast cancer mitogens and appears to be constitutively activated in human breast cancer . A basal level of EF-2 phosphorylation exists in proliferating glioma cells that is markedly diminished or absent in normal glial tissue and is due to the activity of CaM kinase III . Serum addition to nonproliferating rat glial cells induces cell growth and restores CaM kinase III activity . Conversely, 8-Br-cAMP, a well known differentiating agent, stimulates dephosphorylation of eEF-2, promoting protein synthesis that leads to neuronal differentiation .
In conclusion, the involvement of EFs in cell growth regulation can be explained with the following hypothesis. Temporary changes of translation may trigger the transition of a cell from one physiologic state into another because of the disappearance of short-lived repressors and thus the activation of expression of new genes.
A description of the involvement of translational factors in cell proliferation and transformation is summarized in Table 1.
Table 1. Evidence of protein synthesis involvement in tumorigenesis.
Transformation of cells in vitro and specific induction of c-myc
Overexpression in lymphomas, head and neck and colon carcinomas
Involvement in invasiveness, metastatization and expression of VEGF and FGF-2
Transformation of cells in vitro and overexpression in colon carcinomas
Transformation of cells in vitro
Degraded by caspases during apoptosis
Hyperphosphorylation in transformed cells
Phosphorylation in NO-induced growth arrested cells
Its kinase (PKR) is a tumour suppressor gene
Cancers with higher PKR have a more favourable prognosis
The transdominant inhibitor of PKR (p58) functions as an oncogene
p67 protects eIF-2A from phosphorylation and is deglycosylated and inactivated by an enzyme
involved in transformation
Mitogens activate protein synthesis by eIF-4E phosphorylation
Hypusine formation in eIF-5A
Blockade of hypusine formation inhibits tumour cell proliferation
Inhibition of hypusine formation correlates with apoptosis
Overexpression in transformed cells
Homology with the prostate oncogene PTI-1
Overexpression in metastatic cells
Overexpression in gastrointestinal cancers
Overexpression in cancer cells
Its kinase, CaM Kinase III, is specifically activated in growing cells
EGF and IGF-I activate CaM Kinase III
8-Br-cAMP, a differentiating agent, dephosphorylates EF-2
Signal transduction pathways and protein synthesis
Interactions between signal transduction and protein synthesis components
Data are emerging on the interactions between the signal transduction pathways that control cell proliferation and the components of the protein synthesis machinery. In fact, the ribosomal S6 kinase (S6K) and several EFs and IFs are targets of signal transduction pathways . The ribosomal S6 protein is phosphorylated at five Ser residues . The major activity responsible for phosphorylation of ribosomal protein S6 is p70 S6K , which is activated by hierarchical phospholylation of seven Ser/Thr sites. It has been reported that multipotential S6K is activated and becomes membrane associated in response to insulin . Moreover, p90 S6K is activated in response to growth factors by forming complexes with extracellular signal regulated kinase 1 and 2 (ERK-1 and -2) and then by undergoing phosphorylation . Proteins homologous to p90 S6K are Mnk-1 and 2 and they were identified as binding partners and substrates of ERK-1 and -2 . They are activated by phosphorylation at Thr197 and Thr202 both by a mitogen-activated pathway via ERK-1 or -2 and by a stress-activated pathway via p38 mitogen-activated protein kinase (MAPKp38) . Mnk-1 and -2, in turn, phosphorylate the mRNA cap-binding protein at Ser209 in vivo[128–131], resulting in an increase of its affinity for caps . Interestingly, Mnk-1 also binds the C-terminus of eIF-4G, creating a steric link between this factor and signal transduction pathway [129–133]. The effect of S6 phosphorylation on protein synthesis is unclear. However, there is a correlation between the activation of p70 S6K and the translation of mRNAs containing a 5′ TOP tract , and the treatment of 80 S ribosomes with multipotential S6 kinase doubles the elongation rate . A recent study involves S6 phosphorylation in the interaction between 40 S subunits and mRNA . However, the multipotential S6K also has components of protein synthesis different from S6 as substrates; the multipotential S6K is also able to phosphorylate, at unknown sites, eIF-4G as part of the eIF-4F complex . The effect of the phosphorylation is the stimulation of in vitro protein synthesis by the complex eIF-4G–eIF-4F and of the binding of mRNA to the 43 S initiation complex. Another substrate of the multipotential S6K is eIF-1 that is further phosphorylated also by protein kinase C . These phosphorylations result in a stimulation of elongation and GDP/GTP exchange, respectively. Also eIF-2α can be phosphorylated in response to normal signal transduction pathways, e.g. those regulated by amino acids  and IL-3 . Phosphorylation of eIF-2α on Ser51 by several different kinases causes the formation of a stable complex with eIF-2B, thereby reducing the concentration of active eIF-2B. Growth factors and their signalling can interfere with the activity and phosphorylation status of eIF-4E. In fact, the treatment of leukemic cells with granulocyte monocyte colony stimulating factor (GM-CSF) stimulates significantly ERK-1/2 activity and the phosphorylation of both eIF-4E and PHAS . Increased PHAS phosphorylation induces eIF-4E activation. These data suggest that GM-CSF exerts part of its growth-promoting effects through the activation of ERK-1/2 and enhancement of eIF-4E and PHAS phosphorylation and dissociation . For a description of these interactions see also Fig. 2.
Effects of the modification of translational factors on the species of translated mRNAs
On the basis of these data, one can assume that elevated protein synthesis may mark a critical transition in cancer progression. Establishing a greater protein synthesis output may be a necessary step for cancer cells in order to support their rapid proliferation [54,124]. However, analysis of cells transformed by several components of the translational process revealed that the synthesis of only a few proteins was greatly enhanced, while synthesis of most of them was slightly increased. One possible explanation for this is the following: translational factors cause these effects by increasing the translational efficiency of several specific oncogene transcripts, leading to the overexpression of their products. The feasibility of this hypothesis was confirmed experimentally by identifying the interaction between signalling and translational components and by finding several growth regulating proteins, such as growth factors and proto-oncogenes, that are specifically upregulated . In fact, signalling pathways do not stimulate translation of all mRNAs equally and messenger RNAs differ widely in translational efficiency. Factors contributing to low efficiency of translation include a highly structured 5′ UTR, the presence of upstream AUGs, and poor sequence context for the initiating AUG . Many proteins involved in cell growth and cell cycle progression are translated by mRNAs with low efficiency of translation . These mRNAs are poorly translated in quiescent cells but preferably recruited to ribosomes after a mitogenic signal [141–143]. Overexpression of eIF-4E in cultured cells preferentially stimulates translation of a number of mRNAs with high 5′ UTR secondary structure that are involved in cell growth and division . Expression of cell cycle-dependent proteins like c-myc  and cyclin D  requires mTOR, a Ser-Thr kinase effector of protein kinase B. The inhibition of mTOR by rapamycin prolongs the G1 phase in both T cells  and yeast . These results suggest that pathways which activate the unwinding machinery, i.e. by phosphorylation of eIF-4E, eIF-4G, disproportionately stimulate translation of growth-regulated mRNAs with high 5′ UTR secondary structure predominantly involved in cell proliferation. 5′ TOP mRNAs are also differentially regulated in response to extracellular signals . These encode many translational components, including ribosomal proteins, eEF-1α, eEF-2, and PABP. 5′ TOP mRNAs are recruited to polysomes in a growth-dependent fashion that is selectively inhibited by rapamycin [149,150]. These data support the hypothesis that protein synthesis machinery can be regulated by signal transduction effectors. However, this finding alone does not distinguish between signalling through PHAS and signalling through p70 S6K, because the pathway bifurcates downstream of mTOR (Fig. 2) . However, ectopic expression of a dominant interfering p70 S6K blocks both activation of p70 S6K and 5′ TOP mRNA translation, indicating a direct role for 70 S6K . Therefore, mitogen-induced protein synthesis is not only a general phenomenon involving the production and duplication of the components necessary for cell division. Growth factor-induced signalling also interacts specifically with translational factors in order to induce expression of proteins responsible for the progression in the cell cycle. In fact, the insulin-stimulated pathway that leads to general protein synthesis and to growth-regulated protein synthesis in 32D cells can be dissected with rapamycin . Insulin-stimulated total protein synthesis is only 10% inhibited by rapamycin and actin synthesis is not affected at all, but insulin-stimulated c-myc synthesis is inhibited completely. Also, expression of constitutively active protein kinase Cz in the absence of insulin response substance-1, which bypasses the mTOR pathway, allows insulin-stimulated general protein synthesis but not c-myc synthesis . This suggests that the insulin signal bifurcates at some point, one pathway stimulating general protein synthesis and the other stimulating growth-regulated protein synthesis, accounting for 10% of the total translational process, through eIF-4E and S6. This example demonstrates that signalling can differentially regulate the synthesis both of intracellular general proteins and of growth inducers specifically interacting with different translational factors that can represent useful targets for the inhibition of tumour cell proliferation.
Protein synthesis as a target of anticancer agents
Several biotechnological agents of natural origin can interfere and block protein synthesis. They are called toxins for their cytotoxic action on eukaryotic cells.
Bacterial and vegetal toxins
The most frequently used bacterial toxins are Pseudomonas exotoxin A (PE) [153,154] and diphteria toxin (DT) [155,156]. These toxins are synthesized as single polypeptide chains with domains devoted to binding, translocation and toxicity . DT is processed after synthesis in its mature form, made up of two disulfide-bonded chains [155,156]. The B chain carries both a cell binding region (domain I that probably binds the galactose receptor) and a translocation region (domain II) that facilitates the insertion of the A chain (domain III) through the membrane of an acidic intracellular compartment in the cytosol [158–160]. In fact, the N-terminal region of the B fragment contains many hydrophobic residues, the presence of which suggests that the B fragment forms a pore which allows the translocation of the A fragment . The A chain kills the cell by catalysing a modification of EF-2 that prevents its participation in protein synthesis. In detail, EF-2 is irreversibly ADP-ribosylated at a modified histidine residue, diphtamide, located at position 715 . PE kills cells by a mechanism similar to that of DT except for the fact that PE remains as a single polypeptide chain after synthesis although proteolysis of the molecule to release the enzymatically active domain in the cytosol is necessary [162,163]. The residues at the carboxyl end of PE are Arg-Glu-Asp-Leu-Lys . Deletion of Lys613 has no effect, but removal of Leu612 or additional amino acids greatly reduces cytotoxicity because the substitution makes the toxin unable to reach the cytosol . As REDL resembles the sequence KDEL, which functions to retain newly synthesized proteins in the endoplasmic reticulum, PE ending in KDEL was made and found to be fully active (somewhat more active than PE itself) . Therefore, after binding, PE enters the cell by endocytosis and is exposed to an acidic pH within an early endocytic vesicle (endosome) that unfolds the toxin and activates a cellular protease [167,168]. The protease cleaves PE between Arg279 and Gly280 . The 37-kDa fragment that is produced by proteolysis and reduction of the disulfide bond (Cys265–287) is transferred by membrane fusion from endocytic vescicle to the trans-golgi and from there to a compartment that returns KDEL-tagged proteins to the endoplasmic reticulum . Once in the endoplasmic reticulum, the 37-kDa fragment may use pre-existing protein transport complexes or pores to gain access to the cytosol. Domain Ia is not necessary for the translocation process.
The most frequently used plant toxins are ricin and abrin. Both are synthesized as single polypeptide chains and are processed post-translationally to yield the mature toxin, which is composed of two disulfide bonded subunits, A and B . The B chain is a galactose-specific lectin that binds a ubiquitous animal cell receptor. The B subunit at the C-terminus has two galactose-binding domains composed of three ancient galactose-binding units, of which only one is active. After binding, the toxin enters the cell by endocytosis and the A chain is translocated across the membrane of an intracellular vescicle into the cytosol, a process facilitated by B chain [170,171]. The process of translocation of A subunit into the cytosol is poorly understood. The A chain is then released from the B chain by reduction and kills the cell by catalitically removing an adenine residue at position 4324 from the ribosomal 60 S subunit, which is necessary for the binding to EF-2 [172,173]. Proteins analogous to the A subunit of ricin and abrin are gelonin and saporin that are single chain ribosome inactivating proteins [174–176]. They are evolutionarily related to the A chain of ricin and inactivate ribosome in an identical fashion. They lack B chain and, consequently, the nonspecific binding.
The tumour associated antigens: peptide growth factor receptors
It is not feasible to use ‘pure’ toxins in anticancer approaches as they bind aspecifically to all eukaryotic cells. Therefore, it is necessary to gain specificity for neoplastic tissues. This has been achieved by targeting tumour surface molecules called tumour associated antigens (TAA) which are proteic, glycoproteic or sphingolypidic structures expressed at higher density on the surface of tumour cells. The bullets generally used to recognize TAAs are mAbs or their fragments. In order to reduce their immunogenicity in vivo, chimeric or humanized mAb have been designed, produced and used in clinical trials. Alternatively, ligands of peptide growth factor receptors (PGF-R) can be satisfactorily used to target TAA. In fact, PGF-Rs can really be considered TAAs. Alterations of growth factor signal transduction are a common finding in cancer cells. Moreover PGFs are involved in the control of tumour cell proliferation in vitro[177,178] and the study of PGF-R expression on tumour biopsies indicates that the levels of PGF-Rs are much higher on tumours than on the normal tissues from which they have been derived [179–182].
Biotechnological products and immunotoxins
Biotechnological products, called immunotoxins, have been constructed in order to link toxins and bullets. An immunotoxin is a cytotoxic agent that consists of a cell binding moiety and a toxic moiety linked via a chemical cross-linker, natural peptide, or disulfide bond. Genetically engineered immunotoxins are single polypeptides in which the toxin and the binding domain of an antitumour mAb are fused by DNA recombination. An additional genetic construct for the selective targeting of tumour cells is made by linking the toxin to a peptide growth factor, or a peptidic hormone that specifically binds receptors at the tumour cell surface. The toxic moiety can be an intact toxin, its toxic A chain, a naturally occurring single chain ribosome inactivating protein, or recombinant toxins .
Increase in cytotoxic effects of agents raised against protein synthesis machinery
Immunotoxins and related agents have already been used in preclinical and clinical studies, but their activity has been somehow limited . The limitations of this approach are related to the vehicles used to deliver the toxin to the tumour cell surface and by the tumour cell biology itself (host immune response, moiety size, antigen density) . In fact, it has been demonstrated that the efficiency of in vivo targeting of human tumour cells is correlated to the number of membrane antigen sites expressed by the cells [185,186]. Therefore, it was suggested that a more efficient tumour cell targeting could be accomplished by the use of agents able to increase the expression of antigens at the cancer cell surface [186–188]. Another reason of failure for such anticancer approaches is the tumour cell homeostasis linked to the cancer biology. In fact, several mechanisms of escape to immunotoxins can start in a tumour cell with the consequent activation of compensatory or alternative mechanisms of survival. All toxins are raised against EF-2, however EF-2 does not represent the only translational factor involved in the control of tumour cell proliferation. Therefore, the concomitant intervention on the other translational factors directly or through the modulation of the signalling components that control protein synthesis machinery, can constitute an additional therapeutic approach.
Several attempts have been made in order to increase the activity and efficacy of biotechnological products raised against protein synthesis machinery.
As in the case of chemotherapy, the use of different agents prevents the onset of tumour cell resistance and allows the killing of the different clones that constitute the tumour cell population that is often differently sensitive to anticancer drugs. Therefore, immunotoxins have been used in combination with conventional drugs or biological agents, but with poor results [189–195]. An alternative approach is the use of immunotoxins raised against different antigens [196–198] or having different toxic moieties [199,200].
Increase in TAA expression
The increase in TAA expression can be a useful tool to overcome one of the major limitations of anticancer targeting as described above. INF-s have been shown to act as powerful modulators of antigenic expression on tumour cells and appear also capable of increasing the immunotargeting of tumour tissues in vivo[201–207]. In this light we have recently reported that INF-α upregulates the cell receptor for EGF and transferin (TRF) in human epidermoid cancer cells [96,208–210]. EGF-R expression is also increased by other antiproliferative agents such as retinoic acid, tumour necrosis factor-α, conventional anticancer drugs at a low concentration, cytosine arabinoside and 5-aza-2′deoxycytidine [211–220]. Although these studies showed that the binding of immunotoxins to target cells could be increased, they showed no inhibition of the onset of protective responses involving signalling and protein translation that can overcome the antiproliferative activity of the toxin.
Modalities of intervention on signalling and/or specific translational factors
As described above, the close interaction between protein synthesis and proliferative signal transduction pathways justifies the intervention on signalling in order to increase the inhibition of protein translation and vice versa.
Protein synthesis is one of the most complicated biochemical processes undertaken by the cell, requiring approximatley 150 different polypeptides and 70 different RNAs . Yet seven polypeptides (eIF-2α, eIF-2Bε, eIF-4E, eIF-4G, S6, eEF-1, and eEF-2) have been identified as targets for regulatory pathways to date . Multiple IFs and EFs were phosphorylated in response to a single extracellular signal [222,223]. This redundancy now seems more comprehensible as the modification of some factors affects the overall rate of translation whereas modification of others affects the spectrum of mRNAs translated. A better understanding of the pathways for the regulation of protein synthesis can open up novel approaches for cancer intervention , especially if pathways leading to growth-dependent protein synthesis can be selectively inhibited. In this respect, a widely studied signal transduction pathway is the ras-dependent mitogen activated protein kinase cascade which transduces signals from the growth factor receptor tyrosine kinases to the nucleus, thus initiating proliferation .
Growth factors activating signals throughout phosphorylation or other modifications regulate EFs and affect protein synthesis, as demonstrated before. Several attempts have been made to inhibit this pathway whose components are individually capable of inducing malignant transformation when aberrantly controlled . The possibility to interfere at the level of growth factor receptor interaction, receptor activation or at protein target distal to the receptor such as ras and raf has been widely explored in preclinical study and interestingly, in recent years, in some clinical trials. For a description of the most common anticancer strategies raised against signal transduction pathways that, as described above, could represent a useful target for intervening on protein synthesis machinery see Table 2 and Fig. 2.
Table 2. Modalities of intervention on signal transduction pathways.
Inhibition of peptide growth
Anti-sense oligonucleotides raised against PGF-R such as EGF-R, ErbB2, ErbB3 and ErbB4
factor receptor (PGF-R)
Blocking MAbs to specific receptors, e.g. EGF-R, erb-B2
Inhibition of farnesyl-transferase and geranyl-geranyl transferase
Interfering genes for ras have been constructed such as RASN17
Inhibition of raf-1
Antisense oligonucleotide targeting c-Raf-1
Pharmacological inhibition of
Flavonoids inhibitors of MEK-1 (i.e. PD098059) or blockers of ERK-1/2 activity (i.e. 8 CL-cAMP)
PKC inhibitors such as the staurosporine analogue CGP 41
New multimodal therapeutic strategies based on protein synthesis inhibition via signal transduction
The cross-talk existing between signal transduction and protein synthesis and the availability of different molecular tools interfering with both processes suggest the design of new multimodal therapeutic strategies. Our proposal is summarized in Fig. 2. We have demonstrated that INF-α induces apoptosis on KB cells, but the upregulation of the EGF-R caused by INF-α appeared to mediate an antiapoptotic signal. In fact, INF-α and EGF had opposite effects on apoptosis induction, expression of heat shock proteins and activity of the terminal enzymes of the stress signalling pathway. In more detail, we have found that the activity of N-terminal Jun Kinase-1 (JNK-1) is increased in INF-α-treated cells whereas EGF reduces the activity and the expression of the phosphorylated isoforms of JNK-1 in these cells. Furthermore, the transfection of KB cells with JNK-1wt itself induces apoptosis that was potentiated by INF-α and antagonized by EGF suggesting a direct involvement of JNK-1 in the death mechanisms of INF-α. On the other hand, EGF antagonized apoptosis induced by INF-α either in JNK-1wt-transfected or parental cells. The latter effect suggests that the protection from apoptosis induced by EGF is likely to target JNK-1 as a signalling molecule . These effects are paralleled by modulation of hypusine synthesis. In fact, INF-α induces a strong reduction in hypusine synthesis that is restored when INF-α-treated KB cells are exposed to EGF for 12 h . As post-translational modifications of eIF-5 A are essential for its activity, by reducing hypusine synthesis, INF-α is likely to inhibit the function of the translational factor. Therefore, we have hypothesized that the use of a fusion protein that targets EGF-R (that is also upregulated by INF-α) and inhibits protein synthesis through a downstream step (EF-2) could be useful in increasing the antiproliferative activity of INF-α. We have found that INF-α increases (by about 27-fold) the effect of TP40, formed by the fusion of transforming growth factor α (a ligand of EGF-R) and the modified toxin of Pseudomonas aeruginosa. This effect could be attributed both to INF-α-induced EGF-R upregulation and to INF-α- and TP40-induced inhibition of multiple steps of protein synthesis (see Fig. 2). On the other hand, it has been reported that the cytotoxic effect induced by recombinant toxins could be, at least in part, due to the onset of programmed cell death . Therefore, it can be hypothesized that the intervention on the stress-dependent JNK pathway evoked by INF-α and on the translational factors eIF-5A and EF-2 (induced by the cytokine itself and by TP40, respectively) can produce a synergistic effect on the induction of apoptosis in cancer cells (see Fig. 2). Moreover, the selection of pharmacological inhibitor/s of eIF-5A and/or the use of plasmid/s encoding for dominant negatives of other translational factors could be an additional approach to the selective inhibition of cell proliferation. Therefore, eIF-5A, on the basis of its intrinsic biochemical properties, could represent a useful target in combined approaches for the inhibition of different translational factors.
It was even demonstrated that the transfection of dominant negatives of other translational factors can inhibit cell proliferation (Table 1). Several efforts are being made by the gene therapy and clinical studies on the gene transfer through the use of viruses. It would be interesting to investigate if the concomitant inhibition of different protein synthesis factors could give an advantage in terms of cell growth inhibition. As it has been demonstrated that protein kinase C-, ras- and cAMP-dependent pathways regulate the activity of several elongation and IFs  the combined inhibition of these factors through the modulation of signal transduction pathways should be investigated (see Fig. 2). Moreover, the availability of the dominant negatives for the different translational factors could be useful not only for the design of new therapeutic approaches, but also for the identification of their role in the regulation of cell growth by the different pathways. The data obtained could again be useful in the design of new cancer treatments.
As our understanding of mechanisms of malignant transformation and tumour proliferation improves, new molecular targets are continuously revealed, thus providing support to the hope of developing more effective and selective anticancer therapies. On this topic the academic as well as biotechnology company research programs have already produced several clinical grade products, some of which are being used in clinical studies. It is now reasonable to consider that protein synthesis targeting could be an important component of a multimodal approach which might involve cytotoxic compounds as well as immunostimulatory or antiproliferative cytokines. We think that the understanding of the specific interactions at the molecular level among these proliferative and translational pathways could readily increase the activity of combined antitumour approaches that represent the next generation in the fight against cancer.
The authors thank Dr G. Giuberti, Dr M. Marra, G. Granata and P. Sands for their help in preparation of the manuscript.