Evidence for regulation of protein synthesis at the elongation step by CDK1/cyclin B phosphorylation

Abstract

Eukaryotic elongation factor 1 (eEF-1) contains the guanine nucleotide exchange factor eEF-1B that loads the G protein eEF-1A with GTP after each cycle of elongation during protein synthesis. Two features of eEF-1B have not yet been elucidated: (i) the presence of the unique valyl-tRNA synthetase; (ii) the significance of target sites for the cell cycle protein kinase CDK1/cyclin B. The roles of these two features were addressed by elongation measurements in vitro using cell-free extracts. A poly(GUA) template RNA was generated to support both poly(valine) and poly(serine) synthesis and poly(phenylalanine) synthesis was driven by a poly(uridylic acid) template. Elongation rates were in the order phenylalanine > valine > serine. Addition of CDK1/cyclin B decreased the elongation rate for valine whereas the rate for serine and phenylalanine elongation was increased. This effect was correlated with phosphorylation of the eEF-1δ and eEF-1γ subunits of eEF-1B. Our results demonstrate specific regulation of elongation by CDK1/cyclin B phosphorylation.

INTRODUCTION

Changes in the protein synthetic machinery are important in the regulation of gene expression in eukaryotes (1). Compared to initiation, the elongation phase of protein synthesis requires only a small number of factors, namely eukaryotic elongation factors 1 and 2 (eEF-1 and eEF-2) (2). eEF-1 is composed of two elements, eEF-1A and eEF-1B. The G protein eEF-1A (formerly eEF-1α) is responsible for binding of aminoacyl-tRNA to the ribosome; the complex eEF-1B exchanges GDP for GTP on eEF-1A (2–5). On the basis of its structure and phosphorylation by several protein kinases, eEF-1B has been postulated to play a regulatory role(s) (6,7). The structure of eEF-1B has been analysed (7–12). Higher eukaryotic eEF-1B, from animal sources, contains two different guanine nucleotide exchange proteins, eEF-1β and eEF-1δ, a putative anchoring protein eEF-1γ and the unique valyl-tRNA synthetase responsible for attachment of valine to its cognate tRNA (7–12). The eEF-1B components are all substrates for different protein kinases (7,13). The cell cycle protein kinase CDK1/cyclin B (14) phosphorylates eEF-1B in vivo during meiotic maturation of Xenopus oocytes (15–17) in a manner correlated with changes in protein synthesis (18). Furthermore, phosphorylation of eEF-1B persists after fertilisation (19), along with a further increase in protein synthesis (20). eEF-1B contains three different phosphoacceptor sites for CDK1/cyclin B: one in the eEF-1γ component and two on each of the eEF-1δ isoforms present in Xenopus oocytes (21).

The functions of components of the protein synthetic machinery have been successfully characterised using cell-free extracts (22). However, two of the features of eEF-1B, the presence of the unique valyl-tRNA synthetase and the presence of phosphorylation sites for CDK1/cyclin B, have not been related to any biological function. Experiments performed using reconstituted systems revealed no difference between phosphorylated and dephosphorylated forms of eEF-1B from Xenopus oocytes (7,23). Lysates from rabbit reticulocytes (2) readily perform translation from exogenous template RNAs and mammalian eEF-1B was shown to be phosphorylated by CDK1/cyclin B on both the eEF-1γ and eEF-1δ subunits (7). Based on the hypothesis that phosphorylation of eEF-1B could influence valyl-tRNA synthetase associated with the eEF-1B complex, we designed a template RNA for the analysis of poly(valine) synthesis compared to poly(serine) synthesis. The template RNA was used in lysates adapted for elongation determination in the absence of factor-directed initiation.

Using this cell-free system we demonstrate a new regulation of protein synthesis elongation by CDK1/cyclin B phosphorylation.

MATERIALS AND METHODS

Production of the poly(GUA) template RNA

The oligonucleotide 5′-CCGGCGGAATTCTAG(GTA)₃₇TAGGGATCCGGCCGC-3′ containing EcoRI and BamHI restriction sites was synthesised by Eurogentec. The oligonucleotide was amplified by PCR at 48°C using the upstream and downstream primers 5′-CCGGCGGAATTCTAGGTA-3′ and 5′-GCGGCCGGATCCCTATA-3′. The PCR product was separated on a 2% agarose gel, excised and purified using a QiaQuick gel extraction kit (Qiagen). The purified DNA fragment was digested with EcoRI and BamHI (New England Biolabs) and inserted into the pBluescriptII KS^– phagemid vector (Stratagene). Epicurian Coli XL1-Blue supercompetent bacteria (Stratagene) were transformed with the plasmid, positive clones selected and the plasmid purified with a Flexiprep kit (Pharmacia). A T3 sequencing kit (Pharmacia) was used to select the clones containing the correct constructs. The plasmid was linearised with NotI (New England Biolabs) and used as a template for transcription using a Megascript T3 kit (Ambion). The product, referred to as poly(GUA) template RNA, was quantified by absorbance at 260 nm.

Cell-free elongation assay

Rabbit reticulocyte lysate (containing an ATP regenerating system, yeast and calf liver tRNAs and hemin) (Retic Lysate IVT™; Ambion) was used according to the manufacturer’s instructions. The incubation mix supplied in the kit was replaced by a mix devoid of added cold amino acids and supplemented with 5 mM MgCl₂. The high concentration of MgCl₂ allows non-specific binding of the RNA template to ribosomes, resulting in elongation of polypeptides in the absence of specific initiation (24). [³H]phenylalanine (55 Ci/mmol), [³H]valine (23–27 Ci/mmol) and [³H]serine (26–34 Ci/mmol) were purchased from Amersham. Assays were performed with 5 µCi labelled amino acid corresponding, respectively, to 7.5, 15.5–18.5 and 12–16 µM for phenylalanine, valine and serine. Optimum efficiency of translation from poly(uridylic acid) (Sigma) as RNA template was obtained using poly(uridylic acid) at a concentration of 2.2 µg/µl incubation. The poly(GUA) template RNA synthesised as indicated above was used at 3 µg/incubation. Assays were performed using 8 µl of lysate extract in a total volume of 12 µl and incubated for 1 h at 30°C. Incubations were stopped by dilution (1:2) and duplicate aliquots were subjected to TCA precipitation on Whatman filters (22). Radioactivity on the filters was determined by liquid scintillation spectroscopy.

Phosphorylation conditions

Purified CDK1/cyclin B was obtained from Dr Laurent Meijer (Roscoff, France). The specific activity of the enzyme was 5.7 pmol phosphate incorporated in histone H1 per min under standard conditions (25). One microlitre of enzyme solution was added to the 12 µl lysate assays just prior to elongation determination. The inhibitor of CDK1/cyclin B, roscovitine (26), was a gift from Dr Laurent Meijer (Roscoff, France) and was added at 3 mM to the lysates alone or together with CDK1/cyclin B. Denatured CDK1/cyclin B was obtained by boiling the enzyme for 1 min at 100°C.

The activity of purified CDK1/cyclin B in the lysate was measured as follows. The 12 µl translation lysate assays were incubated for 1 h at 30°C in the presence of 20 µCi [γ-³²P]ATP, 1 µl of purified enzyme and 12 µg H1 histone HIII-S-S (Sigma). Control incubations were performed without exogenous kinase and/or without exogenous substrate. Reactions were stopped by addition of electrophoresis sample buffer (27). Phosphorylated proteins were analysed by autoradiography on Kodak X-Omat film after resolution by 12% SDS–PAGE. Quantification of phosphorylation of histone H1 was performed by excision and Cerenkov counting of the histone H1 band.

Phosphorylation of eEF-1B by CDK1/cyclin B in the lysates was performed as follows. The 36 µl translation lysate assays were incubated for 1 h at 30°C in the presence of 20 µCi [γ-³²P]ATP and 2 µl of purified enzyme. The reaction was stopped by dilution in 1 ml of immunoprecipitation medium [50 mM Tris–HCl pH 7.4, 1% Nonidet P-40, 1% bovine serum albumin, 50 mM NaF, 10 mM pyrophosphate, 100 µM sodium orthovanadate, 10 mM β-glycerophosphate, 1 mM 4-(2-aminoethyl)-benzenesulfonylfluoride hydrochloride]. The eEF-1B complex was then immunoprecipitated using antibody to anti-human eEF-1β, a gift from Dr G. Janssen (Leiden, The Netherlands). The antigen–antibody complex was isolated on protein A–Sepharose beads, resolved by 12% SDS–PAGE and transferred to nitrocellulose membranes (28).

RESULTS AND DISCUSSION

Elongation from template RNAs

Reticulocyte lysates were used for translation of mRNA templates. Under our experimental conditions accumulation of poly(phenylalanine) readily occurred from the poly(uridylic acid) template (Fig. 1, left). Since the mRNA template does not contain an AUG initiation codon, incorporation of radioactivity corresponds to elongation in the absence of specific initiation. Elongation was linear with time for up to 2 h and resulted in incorporation of 33% of the amino acid amount in the reaction mixture, which compares favourably with elongation rates obtained in reconstituted systems (29). Synthesis of poly(valine) and poly(serine) was analysed under the same experimental conditions using poly(GUA) as template RNA. Elongation rates for both amino acids (Fig. 1, middle and right) were linear with time in the first hour. The relative efficiency of elongation was in the order phenylalanine > valine > serine. Increasing the amount of the poly(GUA) template (9, 27 and 81 µg) did not increase elongation and, moreover, elongation of both poly(valine) and poly(serine) was inhibited at higher template concentrations. Elongation of poly(serine) was always lower than elongation of poly(valine) and had higher background values in the absence of template RNAs (see Fig. 1, compare valine and serine elongation).

Figure 1.

Elongation of amino acids from template RNAs. Poly(phenylalanine) was determined using poly(uridylic acid) as template and poly(valine) and poly(serine) synthesis were determined using poly(GUA) as template. Reactions were initiated by radioactive amino acid addition in a total volume of 110 µl at 30°C. Aliquots of 10 µl were taken at the indicated times for the determination of elongation in c.p.m.

In different independent 1 h assays 31.2 ± 0.8 pmol phenylalanine (n = 16) was elongated under the standard conditions, whereas the rate for valine was 3.5 ± 0.1 pmol (n = 21) and for serine barely detectable over the background at 1.2 ± 0.1 pmol (n = 17).

Since the assay conditions were optimised to perform elongation by random attachment of ribosomes to RNAs in the absence of factor-directed initiation, the results reflect specific elongation rates for the three amino acids. Such differential efficiencies of translation from different synthetic mRNA templates have been observed previously for poly(phenylalanine) synthesis compared to others (30,31). A higher efficiency of poly(phenylalanine) synthesis could be related to the ability of a ribosome to enter a correct reading frame by attachment to any of the three nucleotides of a UUU codon.

Effect of CDK1/cyclin B on elongation

The effect of phosphorylation by the protein kinase CDK1/cyclin B was first analysed on poly(phenylalanine) synthesis from a poly(uridylic acid) template. In independent experiments the translation rate was found to increase to 230 ± 20% (n = 4) when CDK1/cyclin B was added to the extracts. Using poly(GUA) as template the synthesis of poly(serine) was also increased to 150 ± 20% (n = 7) in the presence of CDK1/cyclin B kinase. Therefore, the protein kinase is able to increase elongation rates of two unrelated codons. In contrast, using the poly(GUA) template, synthesis of poly(valine) was reduced significantly, by 25 ± 10% (n = 4), on inclusion of CDK1/cyclin B in the incubations. Elongation of valine was decreased in parallel with the increase in serine elongation. Although there was variability in the amplitude of the effect among the different experiments, as judged by the large standard error, the differential effect of CDK1/cyclin B was significant and always displayed an increase in poly(valine) synthesis and a decrease in poly(serine) [and poly(phenylalanine)] synthesis.

The effect of CDK1/cyclin B was further characterised. In the experiment shown in Figure 2, CDK1/cyclin B induced an increase in poly(serine) and poly(phenylalanine) synthesis to 218 and 191%, respectively, and a decrease in poly(valine) synthesis to 87%. Roscovitine, a specific inhibitor of the enzyme, totally prevented the effects of CDK1/cyclin B on elongation (Fig. 2) and boiled enzyme had no effect on the elongation rates (Fig. 2). Thus, the effect of CDK1/cyclin B was related to its kinase activity.

Figure 2.

Effect of CDK1/cyclin B on elongation. Elongation rates were determined as indicated in Materials and Methods for a 1 h incubation at 30°C. Columns represent elongation in the presence of the indicated component, expressed as percent of the corresponding control. Bars represent the standard errors of four determinations in the same experiment.

It was important to assess the activity of CDK1/cyclin B under the incubation conditions required in the translation assays with the lysates. Under standard conditions CDK1/cyclin B readily phosphorylated histone H1, as judged by the observed labelling (Fig. 3). Activity of the kinase was also assayed at a concentration of 1 mM ATP, a condition favouring stoichiometric phosphorylation of the substrate (Fig. 3). Under these conditions CDK1/cyclin B activity towards histone H1 was 104 pmol phosphate transferred in 1 h. Phosphorylation was then assayed under the conditions of the lysate incubations. The labelling of endogenous substrates or added histone H1 substrate by endogenous protein kinases present in the lysates was poorly detectable on the gels (Fig. 3), but with addition of CDK1/cyclin B, histone H1 labelling was clearly observed (Fig. 3). The activity of the kinase was calculated to be 2.5 pmol/h, assuming an ATP concentration of 1 mM. This activity was likely underestimated due to the ATP regenerating system, which lowers the specific activity of the ATP pool, and to the presence of high levels of potential endogenous competitive substrates. Thus, CDK1/cyclin B is readily active under the experimental conditions of the lysates.

Figure 3.

CDK1/cyclin B activity under the translation assay conditions. Autoradiography of ³²P-labelled proteins after incubation as indicated in Materials in Methods and resolution by SDS–PAGE. Each lane corresponds to the indicated incubation conditions. Left, molecular weight markers are indicated in kDa.

Two of the mammalian eEF-1B subunits, namely eEF-1γ and eEF-1δ, were reported to be very efficient substrates for CDK1/cyclin B under standard assay conditions (7). The experimental conditions for the determination of translation activities of the lysates utilised a high concentration of ATP and the use of an ATP regenerating system. Such conditions impede the analysis of protein phosphorylation since they lead to poor labelling of the phosphorylated proteins. To ascertain that eEF-1B subunits were effectively phosphorylated under our experimental conditions, immunoprecipitation of eEF-1B was performed using antibodies directed against human eEF-1B subunits. The antibody to human eEF-1β was efficient in immunoprecipitating the rabbit eEF-1B complex as judged from immunoblots of eEF-1β in the eEF-1B immunoprecipitate (Fig. 4, left). When immunoprecipitation was performed with lysates incubated with [³²P]ATP, radioactivity was detected in a 47 kDa protein migrating at the level expected for eEF-1γ. Radiolabel was also present to a lesser extent in a 35 kDa protein migrating at the level expected for eEF-1δ (Fig. 4, right). As a control, the p47 and p35 proteins were not radiolabelled in a lysate that did not contain CDK1/cyclin B (Fig. 4, right). We therefore conclude that eEF-1B is a substrate for CDK1/cyclin B under the experimental conditions of the translation assay in the lysates.

Figure 4.

Phosphorylation of eEF-1B under the translation assay conditions of the lysates. Western blot and corresponding autoradiography of the eEF-1B complex immunoprecipitated from the lysate after incubation with [³²P]ATP, as indicated in Materials and Methods. (Left) Western blot performed with anti-eEF-1β antibody from incubations performed without (–) or with (+) CDK1/cyclin B. (Right) Corresponding autoradiogram. Left, molecular weight markers are indicated in kDa.

The effect of CDK1/cyclin B on elongation was analysed. Our results show that the rate of elongation of poly(valine) compared to poly(serine) is changed concomitantly and in an opposite manner by CDK1/cyclin B. Both determinations were made under rigorously identical experimental conditions, including measurements from the same mRNA template. Among the components of the translation machinery, eEF-1 may discriminate between valine and serine elongation due to the presence of valyl-tRNA synthetase, and no other synthetases, in a pool of eEF-1B (7). Furthermore, eEF-1B is a substrate for CDK1/cyclin B on both its eEF-1δ and eEF-1γ subunits. The eEF-1B complex is present as two pools, one containing the eEF-1β and eEF-1γ subunits and the other containing, in addition, valyl-tRNA synthetase and eEF-1δ (7). The simplest interpretation of our results involves a phosphorylation-dependent general increase in the activity of eEF-1B, which in turn activates eEF-1A and increases the general elongation rate. The increase in eEF-1B would be related to phosphorylation of eEF-1γ by CDK1/cyclin B. In parallel, phosphorylation of the eEF-1δ subunit in the pool of eEF-1B containing valyl-tRNA synthetase would lead to inhibition of the enzyme and therefore to specific inhibition of poly(valine) synthesis. Our results provide experimental support for a new type of translational regulation at the level of elongation, which affects translation of the valine codon in an opposite manner to other codons.

Our results suggest potential regulation of translation of valine-rich proteins compared to other proteins depending on the activation state of the protein kinase CDK1/cyclin B. During cell division CDK1 is inactive during the G₁ and S phases in which the cells grow by increasing their mass and CDK1/cyclin B is activated when the cells reach the G₂/M border and have finished growing (32). Since cells contain abundant amounts of structural proteins, cell growth is associated with quantitatively more translation of structural than other proteins. Most structural proteins contain coiled-coil valine-rich motifs (33); therefore the synthesis of such proteins would be favoured at the level of elongation when CDK1 is inactive, during the growing phase of the cell cycle, and translation of such proteins would be reduced at the G₂/M transition, when CDK1 is activated.

Previously it was reported that CDK1/cyclin B phosphorylates eEF-1B during meiotic maturation of Xenopus oocytes and that phosphorylation of at least the eEF-1δ component remains after fertilisation (19). The present findings suggest that this phosphorylation could favour the synthesis of regulatory proteins during the rapid cleavage phase of early development, in which cells divide and increase in number without growth and therefore without the necessity of producing structural proteins.